- Permanent Link:
- Agreement Reflexes of Emerging Optionality in Heritage Speaker Spanish
- Pascual Cabo, Diego
- Place of Publication:
- [Gainesville, Fla.]
- University of Florida
- Publication Date:
- Physical Description:
- 1 online resource (222 p.)
- Doctorate ( Ph.D.)
- Degree Grantor:
- University of Florida
- Degree Disciplines:
- Romance Languages
Spanish and Portuguese Studies
- Committee Chair:
- Rothman, Jason Leonardo
- Committee Members:
- De Prada Perez, Ana
Lord-Ward, Gillian E
Potsdam, Eric H
- Graduation Date:
- Subjects / Keywords:
- Adults ( jstor )
Child psychology ( jstor )
Cultural preservation ( jstor )
Language ( jstor )
Language acquisition ( jstor )
Linguistic agreement ( jstor )
Linguistics ( jstor )
Sentences ( jstor )
Syntax ( jstor )
Verbs ( jstor )
Spanish and Portuguese Studies -- Dissertations, Academic -- UF
acquisition -- bilingualism -- heritage
- bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Romance Languages thesis, Ph.D.
- This study contributes to current trends of heritage speaker (HS) acquisition research by examining the syntax of psych-predicates in HS Spanish. Broadly defined, psych-predicates communicate states of emotions (e.g., to love) and have traditionally been categorized as belonging to one of three classes: class I-temere â€˜to fearâ€™, class II-preoccupareâ€˜to worryâ€™, and class III-piacere â€˜to likeâ€™. In addition to the notorious structural opacity of class III psych-verbs, a large body of literature has documented them as being problematic for Spanish HSs. Considering this, I propose a novel analysis that aims to explain the patterns observed; i.e. class III psych-verbs â€”those that only have an unaccusative syntax in monolingual grammarsâ€”have been reanalyzed as class II psych-verbs â€”which have available both an agentive and unaccusative syntax. In other words, there is a simplification of the Spanish system of psych-predicates (in the sense of reducing three classes to two). As a result of this adjustment, Spanish HSs should be able to project an optional agentive syntax for gustar-like verbs (a use deemed ungrammatical by monolingual speakers) which is akin to other verbs like asustar â€˜to scareâ€™ or molestar â€˜to botherâ€™ (typical class II psych-predicates). To test this prediction, a total of 114 completed a battery of tests that examined the participantsâ€™ knowledge and use of (un)grammatical items in relation to the syntactic and morpho-syntactic reflexes that should obtain if the hypothesized analysis is on the right track. For example, I predict that Spanish HSs will (variably) accept passive constructions with gustar-like. The data presented reveal trends that are consistent with this prediction Furthermore, the results seem to indicate that a particular feature of change in 1st generation immigrant input providers, the loss of dative â€œaâ€ marking, seems to be what gives rise to the change in syntax by the HS generation. The data, as well as the pairing of the groups which include child and adult aggregates, further contribute to the current debate in formal HS acquisition regarding the sources of variability. ( en )
- General Note:
- In the series University of Florida Digital Collections.
- General Note:
- Includes vita.
- 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.
- Thesis (Ph.D.)--University of Florida, 2013.
- Adviser: Rothman, Jason Leonardo.
- Statement of Responsibility:
- by Diego Pascual Cabo.
- Source Institution:
- Rights Management:
- Copyright Pascual Cabo, Diego. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
- LD1780 2013 ( lcc )
This item has the following downloads:
1 AGREEMENT REFLEXES OF EMERGING OPTIONALITY IN HERITAGE SPEAKER SPANISH By DIEGO PASCUAL CABO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013
2 2013 Diego Pascual Cabo
3 Para vosotros, Teo y Pau, porque lo sois todo.
4 ACKNOWLEDGEMENTS In no way could this project have been completed with out the help of many individuals. First, and foremost, I am grateful to all of the people who participated in my study Without their involvement, none of this would have been possible. I am the most indebted to Jason Rothman, my dissertation adviso r and friend whose energy, passion, and knowledge have been instrumental in my training. I hope my work makes you proud. My debt is also to Ana de Prada Prez I cannot thank her enough for all that she has invested in me. I am also grateful for Gillian Lord an d her valuable guidance and support I am also appreciative of Eric Potsdam who is a great role model for anyone entering academia. His professionalism, in and out of the classroom, sets him apart. Working closely with all four of them has been an absolut e honor and privilege In addition to the members of my dissertation committee, I am also grateful for the assistance given throughout the entire process by many people in different capacities: Jessi Aaron, Mark Amengual, Theresa Antes, Susana Braylan M anuel Burgos, Antonio Cardentey, Wind Cowles, t a Cuca, Alex Cuza, Marisel Cuza Laura Domnguez, Hans Duque, Anna Mara Escobar, Vernica Gonzlez, Edith Kaan, Monica Kelly, Kelly Lowther Pereira, Shannen Mirarchi, Silvina Montrul Ximena Moors Israel Or tet Luis Ortiz Lpez, Melissa Perurena, Diana Pineda, Manuela Pinto, Graeme Porte, Ana Roca, Mike Sharwood Smith, Guillaume Thierry Elena Valenzuela, Julio Villa Garca, Stefanie Wulff Their contributions did not go unnoticed. Special thanks go to Mike Iverson His help with the stati sti cal analys i s was invaluable. While at the University of Florida, I have been fortunate enough to share experiences with an amazing cohort of people I want them to know that I valu e their
5 friendship and support: Felipe A maro Jennifer Cabrelli Amaro Claudia Costagliola, Maria Fionda, Jake Firestine, Kate and Matt Fredericks Katherine and Chad Honea Tiffany Judy Heather Kaiser Delano Lamy Anne Lingwall Mike Odio Diana Pedraza Luisa Quiroga, Whitman Suarez and Val erie and Don Trujillo among many others. I will forever cherish the good times we spent together. My thanks and appreciations also go to those who helped me with my classes: Susana Braylan Claudia Costagliola Meagan Day Antonio Gil, Becky Halloran, Su Ar Lee Dmaris Mayans Ramn Antonio Sajid Lpez Clara Sotelo and Whitman Surez Next, I would like to acknowledge my family and friends in Spain. Their love and support have always been unconditional Thanks for not giving up on me and for instilling i n me the drive an d desire to accomplish my dreams ; lo cons egu I hope my grandparents would be proud. Likewise, always supportive of my choices even though it meant moving far away. Last but not least, I wan t to express my most sincere appreciation to mi esposotona y mis dos nenes I just hope that one day I can repay for all the sacrifices that they have had to make.
6 TABLE OF CONTENTS Page ACKNOWLEDGEMENTS ................................ ................................ ............................... 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 10 LIST OF ABBREVIATIONS ................................ ................................ ........................... 12 ABSTRACT ................................ ................................ ................................ ................... 13 CHAPTER 1 INTRODUCTORY REMARKS ................................ ................................ ................ 15 1.1 General Introduction ................................ ................................ ......................... 15 1.2 Research Problem ................................ ................................ ............................ 20 1.3 Research Proposal ................................ ................................ ........................... 23 1.3.1 Research Questions ................................ ................................ .............. 23 1.3.2 Hypotheses ................................ ................................ ............................ 24 1.4 Dissertation Structure ................................ ................................ ....................... 24 2 BACKGROUND INFORMATION ................................ ................................ ............ 26 2.1 Introduction ................................ ................................ ................................ ....... 26 2.2 Definition of HL and HS ................................ ................................ .................... 26 2.3 Importance of Studying HS Bilingualism ................................ ........................... 31 2.4 Heritage Grammars as Linguistic Systems ................................ ....................... 35 2.4.1 Attrition ................................ ................................ ................................ .. 35 2.4.2 Incomplete Acquisition ................................ ................................ ........... 39 2.4.3 Missing Input Competence Divergence (MICD) ................................ .... 43 2.5 Conclusion ................................ ................................ ................................ ........ 46 3 SYNTAX OF REVERSE PSYCHOLOGICAL PREDICATES ................................ .. 47 3.1 Introduction ................................ ................................ ................................ ....... 47 3.2 Description of the Property ................................ ................................ ............... 47 3.3 Review of Theoretical Analyses ................................ ................................ ........ 52 3.3.1 Belletti and R izzi (1988) ................................ ................................ ......... 52 3.3.2 Grimshaw (1990) ................................ ................................ ................... 54 3.3.3 Montrul (1995) ................................ ................................ ....................... 56 3.3.4 Landau (2010) ................................ ................................ ....................... 58 3.4 Acquisitional Studies ................................ ................................ ......................... 65 3.4.1 L1 Acquisition ................................ ................................ ........................ 65 3.4.2 L2 Acquisition ................................ ................................ ........................ 68
7 3.4.3 HS Acquisition ................................ ................................ ....................... 69 3.5 Conclusion ................................ ................................ ................................ ........ 76 4 EMPIRICAL STUDY: RESEARCH QUESTIONS AND METHODOLOGY .............. 78 4.1 Introduction ................................ ................................ ................................ ....... 78 4.2 Research Questions & Hypotheses ................................ ................................ .. 78 4.2.1 Research que stions ................................ ................................ ............... 78 4.2.2 Hypotheses ................................ ................................ ............................ 79 4.3 Methodology ................................ ................................ ................................ ..... 84 4.3.1 Participants ................................ ................................ ............................ 84 4.3.2 Materials ................................ ................................ ................................ 90 22.214.171.124 Language Proficiency test ................................ ......................... 90 126.96.36.199 Socio linguistic Background Questionnaire ............................... 91 4. 3.2.3 Test 1: Scalar Grammaticality Judgment Task (GJT 1) ............ 92 188.8.131.52 Test 2: Scalar Grammaticality Judgment Task (GJT 2) ............ 95 184.108.40.206 Test 3: Picture Elicited production task. ................................ .... 99 4.4 Procedure for Data Collection ................................ ................................ ......... 105 4.5 Procedure for Data Analysis ................................ ................................ ........... 106 4.6 Pilot Study ................................ ................................ ................................ ....... 107 4.7 Conclusion ................................ ................................ ................................ ...... 108 5 RESULTS 109 5.1 Introduction ................................ ................................ ................................ ..... 109 5.2 Grammaticality Judgment Task 1 (GJT1) ................................ ....................... 111 5.2.1 Distracter Items ................................ ................................ .................... 112 5.2.2 Transitive Predicates in Passive Constructions ................................ ... 117 5.2.3 Unaccusative Predicates in Passive Constructions ............................. 120 5.2.4 Class III Psych predicates in Passive Constructions ........................... 121 5.2.5 Class II Psych predicates in Passive Constructions ............................ 127 5.2.6 General Discussion of GJT1 ................................ ................................ 128 5.3 Grammaticality Judgment Task 2 (GJT2) ................................ ....................... 135 5.3.1 Filler Tokens ................................ ................................ ........................ 137 5.3.2 Distracter Tokens ................................ ................................ ................. 139 5.3.3 Critical Tokens ................................ ................................ ..................... 143 220.127.116.11 Canonical use of gustar like verbs. ................................ ......... 143 18.104.22.168 : ................................ ................. 145 22.214.171.124 Clitic omission ................................ ................................ ......... 149 126.96.36.199 Case agreement innovations ................................ .................. 153 188.8.131.52 ................................ ............ 156 5.3.4 General Discussion of GJT2 ................................ ................................ 160 5.4 Elicited Productio n Task ................................ ................................ ................. 163 5.4.1 Verb Agreement ................................ ................................ ................... 165 5.4.2 ................................ ................................ .................. 170 5.4.3 Clitic Agreement ................................ ................................ .................. 173 5.4.4 General Discussion of Elicited Production Task ................................ .. 174
8 5.5 Summary of Empirical Data ................................ ................................ ............ 175 6 DISCUSSION AND CONCLUSION ................................ ................................ ...... 177 6.1 Introduction ................................ ................................ ................................ ..... 177 6.2 Macro analysis ................................ ................................ ................................ 177 6.3 Limitations of the Study ................................ ................................ ................... 186 6.4 Conclusion and Sugges tions for Future Studies ................................ ............. 188 APPENDIX A LANGUAGE PROFICIENCY TEST ................................ ................................ ...... 193 B SOCIOLINGUISTIC BACKGROUND QUESTIONNAIRE ................................ ..... 197 C PARTICIPANT CONSENT FORM ................................ ................................ ........ 199 D GRAMMATICALITY JUDGMENT TASK 1 ................................ ............................ 200 E GRAMMATICALITY JUDGMENT TASK 2 ................................ ............................ 202 F OMISSION OF DATIVE CLITIC (HS ADV) ................................ ........................... 207 G OMISSION OF DATIVE CLITIC (HS INT) ................................ ............................ 208 LIST OF REFERENCES ................................ ................................ ............................. 209 BIOGR APHICAL SKETCH ................................ ................................ .......................... 222
9 LIST OF TABLES Table Page 3 1 verbs ................................ ................................ ... 56 4 1 Monolingual adult informants ................................ ................................ .............. 86 4 2 Monolingual child informants ................................ ................................ .............. 87 4 3 Heritage child informants. ................................ ................................ ................... 87 4 4 Summary of informant groups ................................ ................................ ............ 89 5 5 Accuracy means of class III psych verbs. ................................ ......................... 168
10 LIST OF FIGURES Figure Page 3 1 Syntax of Class I Psych Predicates ................................ ................................ ... 53 3 2 Syntax of Class II and III Psych Predicates ................................ ........................ 53 3 3 Clitic doubling in Spanish ................................ ................................ ................... 57 3 4 Dative with RPPs ................................ ................................ ...... 58 3 5 Stative Psych Predicates ................................ ................................ .................... 59 3 6 Eventive Psych Predicates ................................ ................................ ................. 61 3 7 Eventive Psych Predicates: LF ................................ ................................ ........... 63 3 8 Stative Psych Predicates: LF ................................ ................................ .............. 64 4 1 Singular _singular ................................ ....................... 101 4 2 Plural _singular ................................ ........................... 101 4 3 Singular _plural ................................ ........................... 102 4 4 Plural _plural ................................ ............................... 103 4 5 Distracter item ................................ ................................ ................................ .. 103 5 1 Group means o f distracter tokens. ................................ ................................ .... 113 5 2 Group means of passives with transitive verbs. ................................ ................ 117 5 3 Group means of passives with unaccusative verbs. ................................ ......... 120 5 4 Group means of passives with class III psych predicates. ............................... 122 5 5 Group mea ns of passives with class II psych predicates. ................................ 127 5 6 Group means of all four critical conditions among adult groups. ...................... 129 5 7 Group means of all four critical conditions among child groups. ....................... 132 5 8 Group means of distracter tokens. ................................ ................................ .... 137 5 9 Group means of transitive distracter tokens. ................................ .................... 140 5 10 Group means of unaccusat ive pseudo distracter tokens ................................ .. 14 0
11 5 11 Group means of unergative pseudo distracter tokens. ................................ ..... 141 5 12 Group mea ns of canonical class III psych verb tokens. ................................ .... 144 5 13 ................................ ................. 147 5 14 Group means of prescriptively grammatical class III psych verb tokens. ......... 150 5 15 Group means for agre ement innovation with class III psych verbs. .................. 153 5 16 Group means for agreement innovation and clitic omission. ............................ 155 5 17 Group means of prescriptively grammatical class III psych verb tokens. ......... 157 5 18 Group means of grammatical innovation. ................................ ......................... 158 5 19 Group means of percentage accuracy for overall verb agreement. .................. 166 5 20 ................................ .. 170
12 L IST O F ABBREVIATIONS AA 1 st Generation i mmigrants (attriters) ADV Advanced BP Brazilian P ortuguese EP European Portuguese HC Heritage c hildren HL Heritage l anguage HS Heritage s peaker IH Interface Hypothesis INT Intermediate L1 First l anguage L2 Second l anguage L2er Second l anguage l earner MA Monolingual a dults MC Monolingual c hildren MICD Missing Input Competence Divergence RPP Reverse p sychological p redicate UG Universal Grammar US Unit ed States
13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy AGREEMENT REFLEXES OF EMERGING OPTIONALITY IN HERITAGE SPE AKER SPANISH By Diego Pascual Cabo May 2013 Chair: Jason Rothman Major: Romance Languages This study contributes to current trends of heritage speaker (HS) acquisition research by examining the syntax of psych predicates in HS Spanish. Broadly defined psych predicates communicate states of emotions ( e.g., to love ) and have traditionally been categorized as belonging to one of three classes : class I temere preoccupare piacere In addition to the n otorious structural opacity of class III psych verbs, a large body of literature has documented them as being problematic for Spanish HSs. Considering this I propose a novel analysis that aims to explain the patterns observed; i.e. class III psych ve rbs t hose that only have an unaccusative syntax in mo nolingual grammars have been reanalyzed as class II psych verbs which have available both an agentive and unaccusative syntax In other words, there is a simplification of the Spanish system of psych predic ates ( in the sense of reducing three classes to two). As a result of this adjustment Spanish HSs should be able to project an optional agentive syntax for gustar like verbs (a use deemed ungrammatical by monolingual speakers) which is akin
14 to other verbs like asustar typical class II psych predicates ). To test this prediction, a total of 114 completed a battery of tests that examined c and morphosyntactic reflexes that should obtain if the hypothesized analysis is on the right track. For example, I predict that Spanish HSs will (variably) accept passive constructions with gustar like. The data presented reveal trends that are consiste nt with this prediction Furthermore, t he results seem to indicate that a particular feature of change in 1 st gives rise to the change in syntax by the HS generation. Th e data, as well as the pairing of the groups which include child and adult aggregates further contribute to the current debate in formal HS acquisition regarding the sources of variability.
15 CHAPTER 1 INTRODUC TORY REMARKS 1.1 General Introduction Language is a complex human specific trait. Despite its complexity successful attainment of a mature linguistic system is manifested at a very early age ( e.g., Hyams 1986; Guasti 2002; Synder 2007 for review from the generative perspective ) Moreover, linguistic deve lopment seems to follow a delimited path universally; that is, irrespective of the language children reach the same milestones at rough ly the same times independent of socioeconomic status, overall intelligence and social context (at least in true linguis tic terms). This observation raises important questions concerning the nature of our linguistic knowledge and the possible domain specificity of language. A reasonable explanation is to hypothesize that humans are genetically predisposed to acqui re langua ge. That is, all human s are biologically endowed with an innate language faculty that constraints our linguistic development enabling (if not obliging) us to converge on fully developed linguistic systems This is the perspective put forward by the Chomsk yan approach to linguistics, also referred to as Universal Grammar (henceforth UG) or generative grammar, the theoretical framework under which this dissertation is carried out ( e.g., Chomsky 200 4 for discussion of this theory from its genesis to its then current state) From the generative point of view, the language faculty is unders tood as an independent modular subsystem of the mind This mental system of abstract computation interacts with other cognitive systems, but is argued to be fully encapsulat ed ( e.g., Fodor 1984 ). That is, although our linguistic endowment can use input from other cognitive systems and even provide relevant outputs for other cognitive
16 domains, it exists as its own entity of modular specific design This language faculty repres ents the true initial state of language acquisition for each individual and includes a set of unive rsal principles and parameters (however one envisions this to be). Principles are truisms of L anguage, that is, they are fundamental and i mmutable rules, so to speak, that constrain the forms all language s can take Parameters are also universal constraints on competence formation. These, however embody the locus of observable crosslinguistic variation via their enumeration of restricted choices for the reali zation of any given property. Parameters are set via the interaction of the particular grammatical facts available in primary linguistic input and the possibilities, or settings, enumerated within UG. Principles, on the other hand, having no options, simpl y restrict grammatical formation in every language equally, although evidence of such restrictions or the 1 The most direct evidence in favor of the existence of UG comes fro m the so called logical problem of language acq uisition ( e.g ., Schwartz and Sprouse 2000 inter alia e.g., Chomsky 1965 1972, 1986 ). This line of argumentation involves the acquisition of complex and subtle linguistic knowl edge ( e.g., semantic entailments that fall out from syntactic structure see Sprouse 2005 for how ) that cannot restricted and e.g., Chomsky 1972:27). In other words, the primary linguistic data available to any individual are underdetermined and are therefore not sufficient to 1 For example, subjacency is no less applicable to Chinese than to any other language, but since it is a universal constraint on movement a nd Chinese lacks movement, then there is limited or no evidence to its instantiation in the grammar because it is simply superfluous given the particular facts of that language.
17 give rise in a one to one mapping sense to the linguistic knowledge that characterizes mature adult grammars Consequently, if the acquisition of these subtleties cannot be reasonably accounted for through experiences with input then it is logical to assume that such knowledge is derived from the innate language faculty. Another important piece of evidence in favor of UG is the existence of a critical period for the development of grammar ( e.g., Penfield & Roberts 1959; Lenneberg, 1967; Curtiss 1977) As mentioned above, the language is primari ly accomplished in the first years of life ( e.g., Hyams 1986; Guasti 2002; Synder 2007 for reviews) During that time, i nteraction with other language users is paramount since it provides the raw materials needed to activate the growth of the lin guistic system Failure to be sufficiently exposed to language before the critical period argued to be from birth to the off set of puberty ( e.g.,Lenneberg 1967 ), is hypothesized to be deterministic in terms of e.g., Curt iss 1977 for discussion of which aspects of language might be more or less subject to this timeframe). Since it is relatively uncontroversial to claim that there is a critical period for the acquisition of a first language (L1), it is only natural to ponde r the consequences of such claims for cases of language acquisition other than monolingual L1 in childhood. For example, many have sought to address the extent to which the Critical Period Hypothesis (CPH) extends to the case of adult second language acqui sition (L2) as well as other subsequent cases of language acquisition (Ln) after normal L1 acquisition has taken place. Within generative SLA, this has translated into the oft investigated question of adult UG accessibility (e.g., White 2003; Meisel 2011). There is, to date, no consensus on whether or not adults who have acquired an L1 in a
18 extension to adult language acquisition in UG accessibility terms (e.g ., Birdsong 1 999; Rothman 2008 ) On the other hand, a nother logical consequence of a strict interpretation of the CPH could be an expectation that all naturalistic language acquisition in early childhood should result in expected levels of ultimate attainment. However, formal linguistic work on heritage speaker bilingualism ( Section 2.1) has consistently shown over the past two decades that such a conclusion is not totally sustainable ( e.g., Montrul 2008 and references therein). This dissertation deals precisely with th is latter group of individuals, namely heritage speaker (HS) bilinguals, in an attempt to examine their grammatical competence and explain, to the extent that their knowledge differs from expected outcomes, how this can be so, given that they are indeed na turalistic child language acquirers. In light of this, it is logical to ask the following two questions: first, does early naturalistic language acquisition always result in convergent monolingual like grammars? Secondly, assuming a naturalistic and early acquisition of the heritage language (henceforth HL) what i s the effect of sustained b ut asymmetrical bilingualism for the linguistic knowledge of either language? These questions currently guide research in the emergent field of formal heritage languag e acquisition to which this dissertation will contribute Bilingual contexts and bilingualism in general, tend to create conditions that give rise to linguistic outcomes that differ from those expected of monolingual speakers of the same language(s). These distinctive outcomes, which from the UG perspective are thought to lar gely obtain as a result of some sort of incomplete acquisition ( e.g., Montrul 2008 inter alia ; also Section 184.108.40.206) are characteristic of grammars.
19 Briefly defined, a HS is a b ilingual individual that grows up speaking a minority language at home, which is different from the majori ty societal language (e.g., Montrul 2008, 2011; Polinsky & Kagan 2007; Rothman 2009). Although the HL is most often the sole chronological L1 or a sim ultaneous L1 (as in 2L1 acquisition) with time the societal majority language usually becomes the HS's dominant linguistic system. As a result, the home language becomes relegated to increasingly more restricted domains of use. The shift to the dominant l anguage most typically takes place in early childhood before ultimate attainment in all domains of the HL grammar occurs. This sh ift has been hypothesized to be at least one primary source of HS outcome differences. The current debate in formal HS acquisi tion regarding the abovementioned source incomplete acquisition ( e.g., Montrul 2008) attrition ( e.g., Polinsky 2006, 2011), and missing input competence divergence ( e.g., Rothman 2007; Pires & Rothman 2009 ) Determining the extent to which HL grammars obtain the way they do as a result of either one or a combination of these factors is at the core of current HL acquisition research. Although no one denies that all of the aforement ioned causes, among others, cannot simultaneously co exist to form the most adequate explanation of HS grammatical uniqueness it is reasonable to ponder the extent to which some of them might be more deterministic than others in a domain by domain sense. One of the motivations for this study therefore, was to conduct experimental research with relevant morpho syntactic properties and across several generations in an attempt to tease apart which of the three factors mentioned above (or a combination the reof) are explanatorily deterministic for the domain of morpho syntax under
20 investigation With this in mind, t he results of this research will provide for greater understanding of HS Spanish grammatical knowledge. The data will be relevant to formal ling uistic theoretical approaches to L anguage and its acquisition by adding to our understanding of the mental (cognitive) representation of human language in a specific subcase of bilingualism The data will also be relevant to contact linguistic theories by reporting on the structural consequences of the bilingual byproducts of language in asymmetrical contact. Finally, the study will shed some light on the role of diachronic change and its link to attrition as well as its compounded effect as subsequent gene ratio ns are provided input by speakers who themselves are undergoing attrition 1.2 Research Problem HSs differ from monolingual children, bilinguals in a societ ally balanced bilingual context, and classroom second language learners (henceforth L2ers) in seve ral ways which include, but are not limited to, the quantity and quality of input, access to formal education in the HL and the social distribution of the HL ( e.g., Beaudrie & Fairclough 2012; Montrul 2008, 2012; Rothman 2007 2009 ; Valds 2005 among othe rs ). For example, while adult L2 learners are usually primarily exposed to input from formal instructed environments, HSs typically receive little to no education in the HL (at least through high school ages). Nonetheless, HSs seem to have some competence and performance advantages over L2ers at low to intermediate levels ( e.g., Montrul 2004; Montrul & Rodrguez Louro 2006). These advantages however, tend to dissipate at more advanced levels (Montrul 2005, 2010) which is to say that the differences in com petence seem to lessen as increased proficiency in the L2 rises Regardless of the proficiency level and property examined, the benefits of examining HSs are manifold
21 Such research can offer insightful contributions and can build bridges between formal la nguage acquisition linguistic theor y diachronic change, sociolinguistics, psycholinguistics, and cognitive science among others ( e.g., Benmamoun, Montrul & Polinsky 2010; Domnguez 2009; Montrul 2008, 2011; Pires & Rothman 2009 ; Rothman 2007, 2009 among others). The growing body of HL acquisition studies has shown that naturalistic acquisition in early childhood can result in significant differences in linguistic outcomes which is an interesting fact especially within the UG paradigm How can naturalis tic acquisition in young childhood result in such differences from expected norms? Without ignoring the determinism of the HS sociolinguistic environment, theorizing about why this is so at the cognitive linguistic level rests at the core of formal HL stu dies. The aim of this study is therefore to better understand, in linguistic and cognitive terms, why some structure s might be m ore susceptible to HS difference from the monolingual model and, more specifically, what linguistic and cognitive variables migh t combine beyond the obvi ous sociolinguistic asymmetries to best explain the competence patterns observed. As is true of all cases of bilingualism, HS acquisition of course cannot be fully described or explained without understanding and then bringing toge ther the linguistic and sociolinguistic aspects involved. However, before such enterprise can be meaningfully undertaken it is worthwhile, especially in a nascent field such as this one, to promote investigations that narrowly address crucial questions fr om each side separately and exhaustively (without ignoring deterministic aspects from the complementary side that naturally come to bear). The objective of this dissertation then is to address primarily the cognitive linguistic side of the issue by describ ing and explaining linguistic competence,
22 and providing a methodology that teases apart, to the extent possible, the s ource(s) of linguistic divergence. Even if I am able to (i) describe and explain the mental competence of these HSs for the structure of i nquiry and (ii) map the source of the difference ( e.g., incomplete acquisition or attrition), I will not be able to fully answer why this has occurred for this domain beyond what is linguistic (in its purest sense). This study examines these questions and goals through the analysis of a somewhat understudied structure relevant to this approach, that is, the argument structure and semantic mappings of r everse p sychological p redicates (i.e. c lass III p sych p redicates (Belletti & Rizzi 1988; Parodi Lewin 1991 ); gustar encantar hapter 3) 2 Examining HS competence with gustar type verbs is of special interest for HS studies as they embody a possible domain of vulnerability for at least two interrelated reasons. First they present an aty pical mapping of the arguments to syntactic positions in light of the canonical predicate mapping of Spanish ( e.g., Gutirrez Bravo 2007 ; Parodi Lewin 1991). In turn, these unique properties render these predicates especially problematic for second languag e learners ( e.g., Montrul 1997, 1998 ; Toth 2003), monolingual attriters ( e.g., Cazzoli Goeta & Young Scholten 2011), and adult HSs ( e.g., Dvorak 1983 ; Dvorak & Kirschner 1982 ; De Prada Prez & Pascual y Cabo 2011 ; Toribio & Nye 2006 ). Interestingly, howeve r, whereas L2 learn ers have difficulty with gustar type psych verbs through development, they do achieve monolingual like competence at higher levels of proficiency (de Prada Prez & Pascual y Cabo 2 Gustar type verbs have also been described as experiential or benefactive (F illmore 1968; Cook 1972) due to the fact that the subject does not function as an agent, but rather functions in an experiential or benefactive capacity (Dvorak & Kirschner 1983)
23 forthcoming) This is somewhat contrary to what has been s hown for Spanish HSs (de Prada Prez & Pascual y Cabo 2011) an interesting comparison to which I return in greater detail in subsequent chapters Second, studies on monolingual children have shown that these verbs are not used in an adult like fashion unt il approximately the age of 6;0 in monolingual Spanish ( e.g., Torrens, Escobar & W r exler 2006; but Gmez Soler 2011, 2012), an age at which most HS have already shifted to En glish dominance. As will be argued in subsequent chapters these issues couple tog ether to create an ideal situation for structural reanalysis in HS acquisition. The focus of this study is therefore to further explore HS Spanish grammars and structural outcomes in the domain of class III psych verbs and the properties associated with t hem. Building on previous work (syntactic theory, L1 acquisition, L2 acquisition, and HS acquisition) this study begins by offering a syntactic hypothesis regarding the underly ing competence of class III psych verbs in Spanish as a HL This hypothesis als o allows for testable empirical predictions beyond what has been shown previously ( Section 3.3 ). 1.3 Research Proposal 1.3.1 Research Questions Before providing further details regarding the theoretical background information on the acquisition and syntax of class I II psychological predicates it is necessary to contextualize this study by stating the specific research questions that guide it: 1) Are class III psych verbs in HS Spanish undergoing a reanalysis of their argument structure? If so, in what ways and why? 2) To what extent are attrition, incomplete acquisition and/or input delimited differences at least expla natory as sources of HS differences and to what extent can these three factors be teased apart?
24 1.3.2 Hypothese s Based on previous research that suggests class III psych verbs to be problematic for this bilingual population ( e.g., Toribio & Nye 2006; de Prada Prez & Pascual y Cabo 2011) I predict that Spanish HSs will show innovations for class III psych predicates and the properties associated to them ( e.g., dati ve a marking, clitic agreement, theme verb agreement). More specifically, I predict that i n HS Spanish, class III psych verbs have been reanalyzed projecting an optional agentive reading instead of the prescribed stative reading (Belletti & Rizzi 1988; Pe setsky 1995) 1.4 Dissertation S tructure This dissertation seeks to weigh in on questions that have dominated the formal linguistic study of HS acquisition in recent years, relating mainly to determining with descriptive and explanatory adequacies what areas of linguistic competence are most susceptible to HS divergence and what linguistic factors contribute to these outcomes. With these general goals in mind, I intend to show that there is a simplification of the argument structure regarding this particular g roup of verbs in Spanish that is shifting towards a reduction favored by the majority language, En glish. As will be described in C hapter 4, this prediction is easily falsifiable given the methodology I propose. If correct, this hypothesis will be able to a ccount for the HS variation attested in previous literature. The remaining chapters of this dissertation will unfold the previous argument in more detail. Chapter 2 provides a comprehensive theoretical discussion of issues related to the specific nature of heritage speaker bilingualism. Chapter 3 provides the necessary theoretical background in terms of the syntactic properties related to psychological predicates in general and class III psych verbs particularly. Before turning to the specifics of the prese nt study in Chapter 4 ( e.g., research questions, hypotheses,
25 description of participants, materials, etc), a review of the primary findings from most relevant research in the different domains of acquisition (L1, L2, HS) is provided. A detailed discussion of the results in terms of the proposed research questions and hypotheses is presented in C hapter 5. To conclude in C hapter 6, and after acknowledging some theoretical and methodological limitations, the main findings are summarized and future lines of inv estigation relevant to the field of HS bilingualism are proposed
26 CHAPTER 2 BACKGROUND INFORMATION 2.1 Introduction This chapter provides a comprehensive theoretical discussion of heritage grammars by surveying and critiquing the principal models that have been propose d to explain the linguistic outcomes that characterize heritage speaker competence namely incomplete acquisition (e.g., Montrul 2008), L1 attrition (e.g., Polinsky 2011) and m issing i nput c ompetence d ivergence (e.g., Rothman 2007; Pires & Rothman 2009). Prio r to this discussion, however, it is necessary to operationalize the terms heritage speaker (HS) and heritage language (HL) ( e.g., Polinsky & Kagan 2007; Rothman 2008 for more details ) an issue which I address next. 2.2 Definition of HL and HS To be clea r from the outset, a HL is a language that is spoken at home and, crucially, is different from the dominant language spoken by the larger society (e.g. Montrul 2011, Rothman 2009 Valds 2005 ). In the context of the United States (US) Spanish is unquesti onably the most represented of the heritage languages (HL) studied. T he increasing interest in studying Spanish HSs in the US over the past two decades is of no surprise since this population is on the rise both in number and in socioeconomic importance ( e .g., Beaudrie & Fairclough 2012; Field 2011) For example, consider the recent growth of the Hispanic population, which grew from 35.3 million in the year 2000 to 50.5 million in the year 2010 ( e.g., the 2010 U S Census for more information on this issue ) But Spanish is of course not the only HL studied nor are the issues pertaining to or pursued in such research unique to Spanish HS bilingualism ( e.g., Field 2011 ; Potowski 2010 ). For example, recent work on HSs has been conducted on a variety of
27 heritage languages including Portuguese ( e.g., Pires & Rothman 2009 ; Rothman 2007 ), Russian ( e.g., Polinsky 2006, 2008a, 2008b, 2011) or Hindi ( e.g., Montrul, Bhatt, & Bhatia 2012). In spite of the current interest in the study of HS bilingualism ( e.g., Benmamoun Montrul & Polinsky 2010; Montrul 2008, 2010; Rothman 2009 ) it seems reasonable to consider its formal linguistic study as an emergent discipline in its initial stages (especially if compared to the fields of first language (L1) and other instances of bi lingual acquisition s (e.g., second language (L2) in adulthood, childhood simultaneous bilingualism (2L1) ) In light of this and considering the increasing relevance of this field, one important theoretical issue that stands out as a current source of cons iderable debate is the one concerning the definition of the HS per se ( e.g., Beaudrie & Fairclough 2012 and references therein for more on this issue). This is of ut most importance here since, until there is one clear definition agreed upon by all (within each specific subfield) the studies conducted and the results obtained will not be truly comparable. In the past years, w e have seen how the concept of HS has been used with different names to mean different things ( e.g., Polinsky & Kagan 2007 ; Ro thman 2 008 for more on this issue). To give a few examples, HSs have also been called Ethnic community speakers ( Montrul 2008), background speakers (Montrul 2008), or semi speakers (Dorian 1989). Although the general assumptions and presuppositions regarding HSs do not change from definition to definition for example, a HS is usually a descendant of immigrants who speaks a language at home that is different from the
28 l anguage spoken outside the home more specific ones do 1 These differences tend to correspond to the specific goals in the research agendas of individual scholars, paradigms or disciplines (e.g., anthropological, socio economic, political) Within the field of linguistics, Polinsky & Kagan (2007) make a distinction between a broad and a narrow definit ion for HSs In the broad sense, an individual i s considered a HS as long as sh e has strong cultural connections to the HL Because t his view does not necessarily take into considerat ion the actual linguistic knowledge of the individuals, its adoption woul d imply that one would have to label passive bilinguals or (non verbal) heritage language learners as HS s even if they can not produce language in the HL. More drastic would be the case of the latter: individuals who have stro ng cultural ties to a historic family language, perhaps as close as one generation before them, but have no linguistic competence whatsoever beyond some knowledge of culturally significant lexical items (presumably because despite strong cultural ties they have not received (sufficient) early naturalistic exposure to the HL). Since the term HS includes the word speaker it might seem a bit confusing to use such a label for individuals who seemingly lack the ability to productively use the language. Therefore such (non verbal) individua ls, under a formal linguistic approach, would not be included as HSs. This is not to say that the above definition is wrong or 1 I am not excluding the possibility that young bilinguals brought up in a H L environment who themselves are not ethnically part of the HL group can be treated as HSs in the context of studies seeking to gauge linguistic mental representation. In other words, should we preclude a young English Spanish bilingual who grows up in a p redominantly Hispanic community and is functionally incorporated into Spanish in her daily life or someone who is raised with a HL caregiver other than her genetic parents to a similar way as those who are ethnically Hispanic and who likely suffers the sam e fate of reduced input to Spanish over time as she enters school? The answer to this of course depends on whether or not you take ethnic identity to be an inclusion/exclusion factor. However, linguistically speaking this person will share many of the sam e characteristics that define HSs with ethnic originals in the HL. She is a naturalistically acquired bilingual of a minority language in a majority language context who has some competence in the non dominant language.
29 misguided; in fact, the study of such individuals and consideration of these criteria bring much to bear on other pertinent quest ions that fall outside the remit of formal linguistic investigations. In the case of passiv e bilinguals, who are also not speakers by definition, there clearly is (some) underlying linguistic competence for the HL since they are, in principle, able to und erstand to some extent the HL. Concerned with the underlying competence of HS grammars, the question of how and why passive bilinguals do not readily produce the HL is outside the scope of this dissertation, as it likely relates to processing factors such as inhibition thresholds and access to grammatical structure in real time. And so, from a formal linguistic perspective it is at least debatably reasonable to include passive bilinguals in the larger cohort of HSs. Again, b ecause the goal of this disse rtation is to explain why the constitution of HS grammars take the shape they do, the broad definition offered above (Polinsky & Kagan 2007) is rendered inappropriate for this project. Admittedly, for studies focusing on other worthy questions that fit nic ely into the larger enterprise of HL studies such as those on HS/HL linguistic identity the broad definition is exactly what needs to be adopted (e.g., Valds 1995, 2003 200 5 ; Val ds, Fishman, Chvez & Prez 2006 ; Wiley & Valds 2000 ) The narrow definit ion offered by Polinsky and Kagan (2007), however, is more appropriate for the present study as it takes into account actual linguistic abilities as inclusion/exclusion criteria to classify individuals as HS s In the strict sense of the word, it is not eno ugh for an individual to grow up in an environment that is surrounded by a (home) culture and a language that differs from the societal one to be considered a heritage speaker. Crucially, a HS must have some (at least minimal) communicative
30 capacity in the HL. definition, which in turn is the one I adopt here: A language qualifies as a heritage language if it is a language spoken at home or otherwise readily available to young children, and crucia lly this language is not a dominant language of the larger (national) society. Like the acquisition of a primary language in monolingual situations and the acquisition of two or more languages in situations of societal bilingualism/multilingualism, the her itage language is acquired on the basis of an interaction with naturalistic input and whatever in born linguistic mechanisms are at play in any instance of child language acquisition ( Rothman 2009: 156). More recently, Rothman and Iverson (2010) reiterated that HSs must be members of a naturalistic bi/multilingual environment, in which there is societal imbalance of the languages involved. Such asymmetry, combined with the age of exposure to the tic development (e.g., Montrul 2008) In terms of age 2 a HS is considered ( a ) a simultaneous bilingual if the individual is exposed to both languages from birth or (b) a sequential bilingual (a child L2er of the majority language) when the societal langua ge is introduced after the speaker has been developing as a monolingual in the HL (typically receiving significant input at roughly school age, so 4 5) Whether the HS is a simultaneous bilingual or a child L2er of the societal language, varying levels of dominance in the societal language are often the linguistic outcome (e.g., Montrul 2008; Pires & R othman 2009; Rothman 2009; Vald s 1995, 1997, 2000, 2003, 2005) In sum, for the purpose of this study, I conclude that, in order for an individual to be consi dered a HS, she must be able to sufficiently speak a home/minority language in addition to the language spoken by the mainstream society where she lives. Crucially, 2 For a detailed discussion on the different variables that weigh in on the issue of describing bilingual
31 acquisition of the HL must be the result of exposure to naturalistic input at an early age, and in a specific sociolinguistic environment, as discussed above. In this sense, it is fair to say that, by definition, all HSs are bilinguals, but not all bilinguals are HSs. This latter claim alone makes HSs as a group different from other bilinguals a case especially worthy of scientific inquiry. 2.3 Importance of Studying HS Bilingualism As Benmamoun et al. (2010) claim and many others have pointed out the st udy of HS bilingualism has not received the same degree of attention as L1 or L2 acquisition in the theoretical linguistics literature, at least until relatively recently. Nonetheless, there are numerous benefits to examining HSs, including contributions to formal acquisition and linguistic theories by gaining a better general understanding of the mental (cognitive) representation of human language, to contact linguistics by reporting on the structural consequences of contact bilingualism, to diachronic linguistic change and its link to attrition as well as its compounded effect as subsequent gener ations are provided input by attrited speakers, and to sociolinguistics by focusing on a particular community (e.g., Benmamoun et al. 2010; Montrul 2008, 2011; Rothman 2007, 2009; Pires & Rothman 2009; among others). Because HSs are neither monolingual no r second language speakers, they thus fall between two groups, and the study of their bilingualism provides us with important information regarding the nature of linguistic knowledge as it develops under reduced input conditions (in the HL) in comparison t o monolingual grammars. As described, the amounting literature dealing with HL acquisition shows that naturalistic acquisition in childhood can indeed result in different paths and linguistic outcomes to a much greater degree than what can be gleaned by lo oking at L1
32 acquisition alone or even child bilingualism in societal bi/multilingual environments That is, HSs more often than not end up displaying some linguistic forms that differ from the monolingual norms ( e.g., Montrul 2008) a point of comparative fallacy to which I return in greater detail in Section 2.4.2 Trying to understand the the oretical implications that result from these differences l ie s at the core o f current formal HL studies For example, one important area of i nquiry to which this rese arch aims to contribute has been dedicated to understand ing so called grammatical vulnerability ( Section 2.4). In other words, one specific goal of this research is grammars and to discern between cases of a ttrition ( Sec tion 2.4.1) and incomplete a cquisition ( Section 2.4.2 ) as well as to study the role that differences in the HL input linguistic quality may have Section 2.4.2). This, as we will see is still a very much in progress effort The study of HS bilingualism also allows us to make important claims regarding the (in)stability of the adult HL and even of grammar in general, and the role age of exposure can play in linguistic development and ultimate attainment The advancement of knowledge in such areas in turn, has significant implications for some long assumed debates regarding the existence of so called critical p erio ds (Lennebergh 1967) as they pertain strictly or even partially to neurobiological maturation ( e.g., Montrul 20 08 ). Briefly stated, because HS grammars can diverge from monolingual grammars to the same and sometimes even more drastic extent as typical L2 grammars despite the fact that acquisition of the HL grammar takes place in early childhood then this allows u s to question the role that age of acquisition alone confers on determining linguistic outcomes in non native speakers. Although HS grammars are also different from L2
33 grammars, they do share some similarities, primarily in their differences from monoling ual grammars (e.g., Montrul 2011), a fact that should not be ignored when making the claim that L1 and L2 acquisitions are fundamentally different in terms of accessibility to language specific mechanisms. What the similarities between HS and L2 grammars tell us is that, for these domains (and perhaps others), different from monolingual norms does not have to be a direct byproduct of age and the subsequent loss of neurological plasticity as is argued by theories that maintain critical period effects for a dult acquisition (e. g., D eKeyser 2000 ; Long 2005 ; Hyltenstam & Abrahamsson 2000 ). In addition to re examining age as the most deterministic factor for the succe ss or the failure in reaching linguistic ultimate attainment, the study of HS bilingualism also forces us to examine more closely the nature and role of the input assumed to be available to the speaker, which by all accounts of acquisition is needed and delimits develop ment and ultimate attainment As will be seen next differences between monolingua l speakers and HSs are thought to obtain as a product of some sort of incomplete acquisition ( e.g., Montrul 2008 for a detailed discussion on this issue) but, is such a claim justified? Are these two groups really comparable in these terms? In other word s, we must examine whether HSs and monolingual speakers receive the same quality of input, where quality simply refers to the linguistic composition of the input, devoid of any evaluative judgment. If not, this fact alone already forces us to question the fallacy of such comparison in the first place, a point to which I return in Section 2.4.2.
34 Because it is assumed that the main providers of HL input are in most cases displaced native speakers (who were first exposed to an L2 at an advanced age), we also need to consider the study of L1 attrition In other words, we need to accept the possibility that HS input is affected by L1 attrition and begin t o incorporate this possibility into our research design so as to move beyond acknowledgment to actual account ing for it (Pascual y Cabo & Rothman 2012). I t has been proposed that L1 attrition may increase the v ariability present in the input which in turn a ffect s the perceived (in)stability of certain (vulnerable) domains. In fact, d omains of grammar that have su ffered apparent L1 attrition are more likely potential candidates to undergo some kind of diachronic linguistic change in a compounded sense over time (e.g., Pires & Rothman 2009) Therefore, th e study of HS bilingualism can also inform us about the role o f universal linguistic mechanisms that drive linguistic reanalysis and restructuring under reduced and/or qualitatively different input conditions (Domnguez 2009 ; Rothman 2007; Pires & Rothman 2009 ), and the role of age and cognitive deve lopment in langua ge acquisition, maintenance and loss, among others (e.g. Montrul 2011 ; Pires & Rothman 2009; Rothman 2009 ). Although not directly related to the present dissertation project, the study of HS bilingualism also has some practical classroom applications, sin ce a better understanding of the nature of their linguistic competence would allow us to best handle the specific needs of HSs as (re)acquirers of the HL ( Montrul 2011; Polinsky & Kagan 2007; Rothman 2009; Valds 2005). HL education is definitely an import ant issue to be desire to relearn, maintain or expand their knowledge of the language for both personal
35 As it currently s tands, m any of these HSs attend classes at the (post)secondary level and end up being (mis)placed in language (e.g., Valds 2005 and references therein). This approach can be found in many institutions across the U.S., but because heritage language education in its own right is still in the process of building itself and developing at least from the point of view of fulfilling their specific needs ( e.g., literacy and regist er awareness) the issue of HS education continues to increase in popularity (e.g., Brinton, Kagan & Bauckus 2008; Carre i ra & Potowski 2011; Kondo Brown 2006; Potowski 2005, 2010; Potowski & Cameron 2007; Valds, Fishman, Chvez & Prez 2006 as cited in Mon trul 2008). 2.4 Heritage Grammars as Linguistic Systems As discussed, o ne of the attested outcomes of HS bilingualism involves the competence and/or use of linguistic properties in a way that differs from that of monolingual speakers of the heritage language d ialect implicated. From the generative point of view, this distinctive outcome is thought to be a product of either incomplete acquisition (e.g., Montrul 2008), L1 attrition (e.g., Polinsky 2011), or input delimited differences (e.g., Pires & Rothman 2009) What follows is a broad review of these s ources of linguistic divergence 2.4.1 Attrition often used as a synonym of general widely examined from different perspectives and theoretica l approaches (e.g., Schmid, K pke Keijzer & Weilemar 2004 ; Kpke Schmid, Kejzer & Doster 2007; inter alia ) in reference to the linguistic erosion of phonological ( e.g., Ventureyra, Pallier, & Yoo 2004), morphosyntactic ( e.g., Hlavac 2003; Montrul 2002; Myers Scotton 2007; Sorace
36 200 2, 2004 ) lexical ( e .g., Schmid & Kpke 2008), semantic (e.g., Seliger 1991), or pragmatic (e.g., Schmid & Duddeldorp 2010) properties. Although comparable, to one relationship since the former refers to a state of partial language loss while the latter entails an all or nothing dichotomy (Schmid 2011 ). In other words, in the same way there is a continuum in which one can place a bilingual speaker in terms of proficiency, from low proficiency (quasi passi ve) bilinguals to h igh ly a dvanced (near native) bilinguals, there are also different levels of attrition. These different levels may materialize in different degrees of language loss and/or processing consequences ( e.g., lexical retrieval, morphological si mplification, etc ) (e.g., Schmid 2011) Some have argued that there can be multiple outcomes as a result of L1 a ttrition in the context of HL acquisition in HSs specifically, which are generally referred to under the umbrella term incomplete acquisition (e.g., Polinsky 1994, Montrul 2008 ) a topic that will be further developed in the next subsection. Such view is less concerned with the developmental differences that giv e rise to the distinct outcomes since it is true that both L1 attrition and true in complete acquisition (i.e., arrested development) do converge in similar, if not the same, outcomes in adulthood when HSs are tested. However, as will be shown here, there are undesirable consequences to collapsing these two concepts, both theoretically and empirically. C rucially, for a particular property to be considered to have undergone attrition at least two conditions must be met : 1) there must be evidence indicating that the specific property examined had been acquired by the speaker prior to its ero sion and 2) there must also be evidence indicating that, after its acquisition, the speaker was able to establish a target like use of the property for a relatively long period of time that is, the property in question reached a state of stability
37 Accord ing to these two stipulations, L1 attriters are those individuals who, after having truly acquired property X and having been able to maintain property X in a stable target like condition for a relatively long period of time, experience d a linguistic chan ge such that the stability of property X becomes destabilized to various degrees of possible loss. Such a change typical among migrants ( e.g., Hispanics in the US) tends to alter the communicative needs of an individual creating therefore an ideal locus for cross linguistic influence (e.g., Silva Corvaln 1994; Paradis & Genesee 1996) which, in turn, may surface in the form of the reanalysis of some linguistic properties ( i.e. L1 properties that may become more similar to the structures represented in th e L2). Various theoretical approaches have been proposed to try to explain attrition effects in individuals across properties and across languages. For example, earlier studies on attrition adopted different versions of the Regression Hypohtesis, a theory originally put forward by Jakobson (1941) based not so much on linguistic grounds per se but o n the overall cognitive nature of memory (e.g., Schmid 2002). As K pke & Schmid put it, linguistic regression was conceptualized both in terms of chronological e learned best or is most often used/reinforc :16). More recently, other work within cognitive paradigms followed this line of rea s Activation Threshold Hypothesis (2004) claims that (i) the properties that are used the most, will remain more activated; and (ii) those properties that are used less commonly are more prone to being inhibited and, therefore, more prone to undergo attrition. Data from different studies showed that the
38 equilibrium between activation/inhibition and its relationship with frequency of use seems to play a determining role in regards to attrition (e.g., Kpke 2004 ; Grel 2004 ). T he I nterface Hypothesis (h enceforth IH) has also contributed to L1 attrition theorizing ( e.g., Tsimpli, Sorace, Heycock & Filiaci 2004; Sorace 2011) Briefly defined, the IH is based on the general idea that having more than one grammar represented in the mind is inevitably associated with important cognitive costs related to the allocation of finite cognitive resources These cognitive costs can potentially result in measurable (processing based) differences for bilinguals ( e.g., Sorace 2011 ). Sorace identifies the properties found at external interfaces (syntax discourse/p ragmatics) as more susceptible to vulnerability These properties are therefore more likely to be either lost earlier in the case of language attrition (e.g., Tsimply et al. 2004) or acquired later in the case of L2 acquisition (e.g., Sorace 2011, inter alia ) Core syntactic properties, on the other hand, are predicted to be less vulnerable to such linguistic erosion ( e.g., Sorace 2011 ). In the case of the Spanish language, two such properties that can be found at the syntax discourse interface are subject position across verb type s juxtaposed against information structure (e.g., in focus environments), and mood selection with complements of negated epistemic predicates (e.g, saber pe nsar creer Although cases of HS acquisition were not originally included in the predictions made by the IH, some have recently argued (e.g., Montrul & Polinsky 2011; Pascual y Cabo, Rothman, & Lingwall 2012) that the patterns of non convergence and residual optionality that characterize HSs should also be included in the scope of the IH predictions, whether or not these come to support the IH itself or prove particularly
39 useful in accounting for HS grammars. In other words, they argue that there is no principled reason following the tenets of the hypothesis itself for which HSs should be excluded. Sorace (2012) acknowledges the validity of these claims and states that : the IH can make predictions for subsequent stages of attrition, as long as the differences between on the right track, and in fact embodies one of the goals of this dissertation. As described, l a nguage attrition is related to incomplete a cquisition in the sense that there is a clear process of language loss. There are however important differences between both processes in the ways explained above. Nonetheless, the us e of the term incomplete a cquisition has been used freely, which has created some controversy within the field of HS bilingualism (e.g., Pascual y Cabo & Rothman 2012) To contextualize the debate, next I operationalize the term and expand on the current s tate of affairs 2.4.2 Incomplete Acquisitio n The term incomplete acquisition is most often found in relation to the observable linguistic vulnerability that characterizes most, if not all, HSs in their use of the HL Although this concept is not new, it is in its current form (Montrul 2008) arguably the most frequently Schachter (1990), following Bley Vroman ( 1990 ) and others proposed the Incompleteness H ypothesis as a way to describe the so called deficiency in the (2008) current definition of the term incomplete acquisition reads as follows:
40 In my view, incomplete acquisition and L1 attrition are specif ic cases of language loss across generations What I broadly refer to as incomplete acquisition (for lack of a better term), is a mature linguistic state, the outcome of language acquisition that is not complete or [of] attrition in childhood. Incomplete acquisition occurs in childhood, when some specific properties of the language do not have a chance to reach age appropriate levels of proficiency after intense exposure to the L2 begins ... Although L1 attrition can also occur in childhood, I consider att rition as the loss of a given property of the language after property y was mastered with native speaker level accuracy and remained stable for a while, as in adults (Montrul, 2008 :21 ). As it currently stands, the use of term i ncomplete a cquisition to d escribe language loss among HSs is problematic at different levels. First we have seen that this approach is too broad and it does not appropriately account for the distinction that needs to be made between those linguistic properties that are part of the primary linguistic data and those that are not. I n other words, i t seems reasonable to claim that certain linguistic properties can only be acquired as long as the triggers that give rise to them are part of the input that HSs receive. If these properties are not part of the primary linguistic data, th en it would be very difficult for HSs to acquire them. In light of this, P ires and Rothman specific in terms of the role of the input HSs receive, put forward a more fine grained distinction ( Section 2.4.3) Second, abovementioned HS outcomes are also somewhat controversial Although she claims that the idea of nd not as a value judgment (2008:7), the actual choice of words inevitably carries a negative effect that is associated with the figure of the HS ( e.g., Pascual y Cabo & Rothman 2012) In this sense, the dichotomy that results between the i dea of complete vs. incomplete does not seem to do justice in terms of accurately reflecting a process in which a language has been naturally acquired in the exact same way that monolinguals do. From this it follows
41 that convergence on what one can logically converge on g iven the actual input they receive (and crucially not measured against the input that monolinguals receive) should always be considered complete. In that sense, the term incomplete i s imprecise if not inaccurate. Several studies have convincingly shown t Differential Object Marking 3 (DOM henceforth) in HS Spanish Although this property has been known to represent a learnability probl em for L2 learners (Bowles & Montrul 2009; Guijarro Fuentes & Marinis 2007), acquisition studies with monolingual children have revealed that DOM is acquired rather early ( e.g., Rodriguez Mondoedo 2008): by age 3, children reveal a near ceiling level accu racy (98.38%). Because HSs are exposed to Spanish from birth, one would expect early bilinguals to be equally successful in acquiring DOM (Montrul 2011 b ) reexamination of the oral narrative task performed by monolingual and b ilingual children suggest otherwise: while the monolingual children were 95% accurate, this level of precision decreases substantially among sequential bilinguals (62.9%) a nd even more drastically among simultaneous bilinguals (32.8%). These non target like results among 3 refers to the particular marking of Spanish direct objects when th ey are [+ HUMAN ] and [+ SPECIFIC ayer visit *(a) mis amigos ayer visit (*a) unos amigos ayer visit (*a) Atarfe (yesterday I visitied (*to) Atarfe). In object (my friends) is [+ HUMAN ] and [+ SPECIFIC ]. In the case of (ii), the sentence is ungrammatical with HUMAN ] but [ SPECIFIC ]. In the case o f (iii), the sentence is HUMAN ] and [ SPECIFIC human animate objects when they are interpreted as specific and individual (as with pets). I refer t he reader to see Zagona 2002:12 14 for examples and a more detailed discussion on this topic.
42 HS children have also been shown to persist into adulthood (Montrul 2004b; Montrul & Bowles 2009). t to be understood as a synonym of wild or rogue, since both complete and incomplete grammars are considered to be fully constrained by Universal Grammar ( Montrul 2008:167). In spite of this all encompassing view, L1 a ttrition and true incomplete acquisiti on are two different concept s and therefore should be understood and treated as such ( e.g., Pascual y Cabo & Rothman 2012; Polinsky 2011). As stated, L1 attrition refers to the loss of a property (or set of properties) that had been previously acquired and mastered by an individual. True i ncomplete a cquisition, on the other hand, refers to those instances in which acquisition of those same properties could have taken place given input that provides the linguistic information that should result in reflex X but for some reason features that would give rise to reflex X. That being said, teasing these two processes apart without access to longitudinal data is impossible, at least in any direct way. As Pascual y Cabo and Rothman (2012) put it: a posteriori and recall that HSs are tested in a mature state of knowledge as adults, the course of development. That is, there is no way to know for sure working backwards if something did not develop or if it was acquired and then eroded, the former being actual incomplete acquisition and the latter being attrition ( Pascual y Cabo and Rothman 2012:4) Even so, efforts are being made to develop methodologies that would allow us to more tangibly distinguish between the two processes. For example, in her study of relative clauses among Russian HSs in the US, Polinsky (2011) examin ed data from child and adult HSs and compared them to age matched co ntrol groups of monolingual
43 speakers of the sa words, both models ( incomplete acquisition a nd attrition ) make the following predictions : a. Incomplete acquisition: If a child and an adult deviate from the baseline in the same way, it can be assumed that the feature has not been acquired. b. Attrition: If a child performs as his or h er age matched baseline control but the adult does not, the feature can b e assumed to have been acquired but may have subsequently been lost or rea nalyzed. (Polinsky 2011: 306) Such predictions can advance the field of HS study by capturing more convincingly the developmental path that we are unable to observe directly. Nonetheless, the success of her approach is not without some level of compromise; being an indirect measure, one would have to assume that the child and adult participants in her study (or in any similar stu dy for that matter) grew up in identical ( or at least comparable ) language learning background s and received identical ( or at leas t exceedingly similar ) quality and quantity input in each of the languages involved. Of course any method that is not based on self reporting, is very difficult to carry out, but a detailed examination of the input received in the HL while controlling fo r all other relevant variables would be desirable if one is to make any claims regarding the linguistic competence of heritage speakers (Pires & Rothman 2009; Domnguez 2009). This, as I describe with more detail in the following subsection, is the basis of Pires and Input Competence Divergence proposal. 2.4.3 Missing Input Competence Divergence (MICD) Uncontroversially, input delimits acquisition outcomes ( e.g., Pires & Rothman 2009; Rothman 2007 ; Sorace 2004). As described, HSs compr ise a unique group of individuals with uniq ue linguistic characteristics. But u nlike monolinguals, HSs grow up in a bi/multilingual setting where the (imbalanced) societal distribution of the languages
44 involved, the age of exposure to the dominant language and crucially the delimited linguistic input they receive in the home language all play deterministic roles in the ir linguistic d evelopment. As stated in the previous subsection, t hough noticeably different from those of monolingual speakers, these outc omes cannot necessarily be described available. To be sure, a s Pires and Rothman (2009) note the use of the term ncomplete a cquisition is not entirely adequate sin c e the line that separates true i ncomplete a cquisition from child L1 Attrition and even from outcomes resulting from input differences is not clearly established. Again, t rue i ncomplete a cquisition entails that the learner fails to acquire grammatical pr operties that are sufficiently pr esent in the linguistic input s he receives. Once again, i n the absence of ample longitudinal data that provide a detailed description of such input, it would be extremely difficult for anybody to claim with certainty whethe r or not a specific property is or is not available in the input 4 According to Pires and Rothman (2009) if the property is not readily available in the input, then the process of acquisition cannot be referred to as incomplete. Although Montrul (2008) do es not explicitly make such a distinction, Pires and Rothman (2009) target like competence outcomes when the input clearly provides triggers for such convergence but for whatever To support their claim, Pires and Rothman (2009) compare and contrast acquisition of inflected infinitives 5 in U.S. HSs of European Portuguese (EP) and of 4 This is particularly relevant since HS acquisition is marked by individual differences. 5 Inflected infinitives carry overt person/number agreement yet lack ove rt specification for tense.
45 Brazilian Portuguese (BP). Crucially, while this property is commonly found in EP, its use has practically disappeared in colloquial BP. In fact, s peakers of BP are only exposed to it through formal education (Rothman 2007) which we know HSs do not usually receive. The implications of such groupings are clear: if inflected infinitives are prone to being a vulnerable property, then there should not be differences between both groups, that is, not only neither EP HSs nor BP HSs should use inflected infinitives both groups should display the same negative intuitions regarding its grammaticality. If, on the other hand, there are differences in their knowledge and use of inflected infinitives then i ncomplete a cquisition cannot be the only reason to explain such a difference. W ith this approach, Pires and Rothman (2009 ) claim that the source of variability in HSs competence cannot be explained solely by I ncomplete A cquisition because EP HSs consistently demonstrate robust knowledge of inflected infinitives, while BP HSs consistently lack such k nowledge (2009:23). Thes e results indicate that differences in the input must (Pires & Rothman 2009: 1) by HSs of Brazilian Portuguese Ta rget like deviancy therefore seems to obtain as a result of a compounded effect that necessarily includes differences in the quantity and quality of the input informants receive in the heritage language. Again, d grammars take place as a result of any one of these possibilities (L1 a ttrition, i ncomplete a cquisition, and m issing i nput c ompetence divergence ) is at the core of current HL acquisition research. As stated, a lthough all these three approaches entail diffe r it is extremely complicated to convincingly tease them apart. As of now, and in the
46 absence of extensive longitudinal data collected across different generations within the same community of speakers, this is practically im possible. As will be shown, b y conducting experimental research that adequately controls for the participants members hip in one specific community as well as the particular structures under study, and by testing across different generations within the s ame community, the current study is better position ed to document the cross generational effects that may contribute to these unique and outcomes of HS acquisition. 2.5 Conclusion In this chapter I have presented a broad theoretical discussion of heritage gr ammars, paying specific attention to both the definitions of the heritage language/heritage speaker and the different theoretical models that have been proposed to explain the linguistic outcomes that characterize heritage speaker competence namely incomp lete acquisition (e.g., Montrul 2008), L1 attrition (e.g., Polinsky 2011) and m issing i nput c ompetence d ivergence (e.g., Rothman 2007; Pires & Rothman 2009 ) Chapter 3 further describes the motivation for carrying out the present study by providing a deta iled description of ( r everse ) p sychological p redicates the linguistic propert y under study here, and a review of the existing literature associated with the acquisition of this property in Spanish.
47 CHAPTER 3 SYNTAX OF REVERSE PSYCHOLOGICAL PREDICATES 3.1 Introductio n The purpose of this chapter is threefold. The first goal is to provide a detailed description of the linguistic properties of focus in this dissertation, that is, those associated with the argument structure an d semantic mappings of Spanish r everse p sych ological p redicates ( or class III psych verbs ) as well as the ir English equivalency The second goal is to examine the most prevalent theoretical syntactic analyses of class III psych verbs The third and final goal is to review the existing literature ass ociated with the ir acquisition in Spanish as a first and subsequent languages This chapter will thus lay the foundation for t he empirical study that will be presented in C hapters 4 and 5 3.2 Description of the P roperty T he term psychological predicate refer s to a class of verbs whose lexical semantics denote a mental/emotional state or the psychological process that results in a mental/emotional state. Verbs such as to love in English or odiar (to hate) in Spanish are two representative examples of pysch v erbs One important feature that characterizes this group of verbs is that, unlike common agentive predicates that subcategorize for the thematic roles of an < agent > (i.e. the doer that performs the action) and a < theme > (i.e. the entity directly affected by the action) psych predicates subcategorize for two internal roles: an < experiencer > (i.e. the entity that receives or contains the mental/emotional state) and a < t heme > 1 (i.e. the entity that triggers such feeling). 1 This will be modified below following Pesetsky (1995).
48 In seminal work, Belletti and Rizzi (1988) proposed a three class distinction of p sychological predicates for Italian which also extend s to Spanish ( Parodi Lewin 1991 ). (1) Class I (temere to fear) Gianni teme questo (2) Class II (preoccupare to worry) Questo pr eoccupa Gianni This worries Gianni (3) Class III (piacere to like, to please) a. A Gianni piace questo b. Questo piace a Gianni As can be seen in example (1) above, c lass I psych verbs are gen erally treated as transitive verbs with a subject < experiencer > in nominative case and a < theme > in accusative 2 Some Spanish verbs that belong to this class are temer ( to fear ) or odiar ( to hate ) (4) Teo odia la lechuga Teo. NOM hate. 3.SG.PRES the lettuce. ACC Class II psych verbs contain a < theme > with nominative case that controls verb al agreement and a postverbal accusative < experiencer > Some Spanish verbs that belong to this class are molestar ( to bother ) or asustar ( t o scare) 3 Consider example (5) below: (5) Teo molesta a Pau Teo. NOM bother. 3SG.PRES to Pau 2 Both Belle t ti and Rizzi as well as Landau treat these verbs as regular transitive verbs, although Landau shows that they may also act as locatives in certain contexts (2 010: 11 15). 3 Interestingly, as I will explain in more detail in section 3. 3, class II verbs can be treated as either transitives or unaccusatives allowing for two different readings: Agentive vs. Stative.
49 Class III psych predicates are also known in the literature as r everse psych ological p redicates (RPP) due to the reverse mapping of the a rguments to the syntactic positions. That is, even if it may look like a SVO word order on the surface verbal agreement is controlled by a postverbal < theme > 4 This pref erred word order ( e.g., Gutirrez Bravo 2007 ), however, can vary and non canonical wor d orders arise for discourse pragmatic reasons This is related to the rigidity of the Nuclear Stress Rule and the fact that Spanish has no phonotactic recourses (differently than English) ( e.g., the work of Zubizarreta for a review) If a discourse envir onment makes salient the < theme > over the < experiencer > or vice versa, then the word order of the utterance will reflect this: the focalized element will move to sentence initial position (or the CP layer) where these types of syntax discourse inte rface e lements are realized (e.g., Zubizarreta 1998 among others ). Class III psych verbs are thus no different than other verbs in this respect and hence the apparent optionality of position. e.g., in topic and foci environments, as can be seen in examples 6a d b elow. (6) Class III gustar (Spanish) to like, to please) a. A Pau le gusta esto. To Pau Him ( DAT.CLI ) likes this. b. Le gusta esto. Him ( DAT.CLI ) likes this. c. Esto le gusta a Pau. This Him ( DAT.CLI ) pleases to Pau d. Esto le gusta. This Him ( DAT.CLI ) likes 4 Despite t heir object like function, dative < ex periencers > in Spanish have also been referred to in the literature as logical subjects ( e.g., Masullo 1992 ; Montrul 1995). Non nominative subjects are licensed in languages such as Icelandic or Hindi ( e.g., Bhaskararao & Subbarao 2004)
50 Spanish and Italian differ in this class syntactically in terms of the possibility of so called clitic doubling. In Italian, either the dative clitic or the dat ive PP is used, but never both. In Spanish, if the < experiencer > is specified (as in 6a & 6c above), then it must be doubled by a clitic (indirect object pronoun). There are few verbs in Spanish that are true class III psych verbs, that is, RPPs; however, these verbs like gustar (to like) or encantar (to love) are highly frequent These differences bring no bearing to the properties of interest in this dissertation. Crucially for the purpose of this study v erbs that belong to this group are considered stat ive, that is, they cannot be used agentively, do not project v P, and are thus incompatible with passive constructions as in (7): ( 7 ) La pelcula es gustada The movie be. 3 SG PRES like The abovementioned reverse argument structur e mapping becomes esp ecially evident when comparing r everse p sych predicates to other more common a gentive verb s such as leer (to read). For example, in (8 a) the < a gent > ( Laurie ) appears in sentence initial position and controls verbal agreement: hence, th e ungrammaticality of the plural form leen (they read) in (8b). Conversely, in (9 a) a postverbal thematic subject ( los poemas de amor love poems ) controls verbal agreement: hence, the ungrammaticality of the singular form gusta in (9 b) 5 (8) a. Laurie lee los poemas de amor Laurie 3 RD SG .read the poems of love b. *Laurie leen los poemas de amor Laurie 3 RD PL .read the poems of love love poems 5 But see Ortega S antos (2006)
51 (9) a. A Doug le gustan los poemas de amor To Doug DAT. CLI 3 RD PL. like the poems of love Doug likes love poems b. *A Doug le gusta los poemas de amor To Doug DAT. CLI 3 RD SG .like the poems of love Doug likes love poems Therefore, a class III psych verb contains a nominative < theme > that controls verbal agreement and a dative < experiencer > which, according to Landau ( Section 3.4), is inherently and universally case marked (2010:54). The focus of this study is limited to the study of class I II psych predicates; however, as will be seen, in the experiment s conducted other psych verb classes as well as non psych verbs are used for counter balances The linguistic notion of p sych predicates is universal; that is, depending on the semantic functi on of the verb, it is or it is not considered a p sychological verb 6 As such, the equivalent English verbs are psych predicates as well. The main difference between Spanish and English in this respect is syntactic. In other words, in English, being a n Subj ect Verb Object language 7 does not have dative < experiencer s > that could be mistaken for subjects (White, Brown, Bruhn Garavito, Chen, Hirakawa, and Montrul 1999: 173), the thematic (semantic) mappings are never (syntactically) reversed. That is, they are mapped on to canonical SVO word order, irrespective of the type of psych predicate ; t herefore, verbal agreement is always controlled by the preverbal argument 6 I mean universal in the sense of the universality of semantics and conceptual structure (e.g., Jackendoff 2002). 7 Here, I acknowledge and put aside some rare word orders in modern English that seemingly do not have a strict SVO order. For example, some remnants of V2 structures exist giving rise to non canonical word orders, for example with negative polarity as in Never did I know vs. *Never I did know.
52 (examples 5a and 5c) Any deviation to the SVO order, at least for these structures (footnote 13 below), results in an ungram matical sentence ( examples 10b and 10 d). (10 ) a. Drew likes dogs b. Drew like dogs c. They like chocolate d. *They likes chocolate 3.3 Review of T heoretical A nalyses R esearchers from different paradigms have proposed several typological, semantic and syntactic analyses for the observable phenomena associated with psych predicates ( e.g., Arad 1998; Belletti & R izzi 1988; Bouchard 1995; Grimshaw 1990; Landau 2010; Masullo 1992 ; Pesetsky 1995 among others ). In this sect ion, I present an overview of some of the most relevant analyses of psych predicates: (i) Belletti and Rizzi (1988); (ii) Grimshaw (1990); (iii) Montrul (1995) and (iv) Landau (2010) 3.3.1 Belletti and Rizzi ( 1988 ) As discussed, B elletti and Rizzi put forward a tripartite classification of psych ological verbs ( examples 1 3 above), trying to explain in a systematic way the apparently arbitrary mappings of the thematic roles to the syntactic positions. In their analysis, which included the deep structure surface structure division of then current syntactic theory they proposed two different structures : one for c lass I psych predicates and another for c lass II and III As can be seen on figure 3 1 below the structure B elletti and Rizzi propose for class I is a simple transitive structure in which the < experiencer > ( Gianni ) appears in the subject position and the < theme > ( questo ) is located inside the VP as complement of V ( teme )
53 Figure 3 1. Syntax of Class I Psych Predicates On t he other hand, the D structure they propose for classes II and III is a double object construction where (i) the subject position is empty and (ii) both the < theme > and the < experiencer > are projected as internal arguments. See figure 3 2 below. Figure 3 2. Syntax of Class II and III Psych Predicates According to B elletti and Rizzi while either the < t heme > or the < experiencer > can raise ( B elletti and Rizzi 1988:335) it is the case that the < experiencer > raises because it is structurally higher than the < theme > Regarding its landing site, Belletti and Rizzi
54 provide independent evidence that demonstrates that these elements move to subject position and not higher up as topics or left dislocated elements. In this respect, Belletti and Rizzi (1988) noted t he distinct properties of some elements that display most canonical subject properties except for agreement, but bear inherent case (Landau 2010:81). These elements are also known in the literature as quirky (subjects/datives). They noted that (i) while qu irky datives can be quantified NPs, left dislocated cannot and (ii) that left dislocated elements only appear marginally in embedded clauses. Quirky datives, on the other hand often occur in this contexts and do not constitute a barrier for Wh extraction. 3.3.2 Grimshaw ( 1990 ) In light of a less syntactic centric view, Grimshaw (1990) appealed to the interaction between the T hematic and the Aspectual H ierarchies as the cause o f the reverse mapping of ar guments observed for class III p sych predicates. According to Grimshaw each hierarchy establishes and defines relations of prominence betw een the participating arguments, in which the external argument is always the most prominent. The Thematic Hierarchy arranges the arguments with respect to their thematic promine nce, the agent being the most prominent argument. (11) Thematic hierarchy ( Agent [ Experiencer [Goal/Source/Location [ Theme ]]) The Aspectual Hierarchy adds aspectual properties to predicates and distinguishes between eventive and stative p redicates. As can b e seen in example (12) below, the c auser is the highest argument in the Aspectual Hierarchy ( 12 ) Aspectual hierarchy (Cause [other[ ]])
55 As discussed, Grimshaw attributes the inverse word order of class III psych verbs such as gustar (and o pposed to the direct word order revealed by regular agentive verbs) to the (mis)alignment that results from the interaction between these two hierarchies. To better illustrate this idea, consider two comparable constructions: one with an agentive predicate ( e.g., lavar class III psych verbs ( e.g., gustar as (1 3 and 1 4 ) below. (13 ) Agentive verbs: , Felipe lava los coches Felipe wash. 3 SG PRES the cars Felipe (14 ) Class III psych verbs: , A Felipe le gustan los coches To Felipe 3.sg.CLI like. 3 SG PRES the cars In the case of (13), the interaction between the two hierarchies is non problematic since the most prominent arguments in each of the hierarchies align. That is, the most prominent element in the Thematic Hierarchy (the < agent > of the action) just happens to be the most prom inent element in the Aspectual H ierarchy (the < causer > of the action). In other words, Fe lipe is both the < agent > responsible for washing the cars as well as the one responsible for the cars to be washed. In the case of class III psych verbs as in ( 14 ), a clear mismatch takes place between the most prominent element in the Aspectual Hierarchy ( i.e., the < c ause > of the mental/emotional state) and the most prom inent element in the Thematic Hierarchy ( i.e., the < e xperiencer > ). This mismatch according to Grimshaw, explains the reverse structure of this particular group of verbs. Table 3 1 below il (1994) analysis.
56 Table 3 1. : Gr verbs Agentive verbs Psych verbs Thematic hierarchy < Agent > < Theme > < Experiencer > < Theme > Aspectual hierarchy < Cause > < Other > < Cause > < Other > 3.3.3 Montrul (1995) Focusing specifically on Spanish and in line with other analyses ( e.g., Masullo 199 2), Montrul claims that dative < e xperiencers > are in subjects; they are dative subjects (non Her discussion is largely based on the properties associated with doubled dative ciltic s obligatory in Spanish if the full < experiencer > is to be spelled out in t he case of class III psych verbs According to her, clitics are doubled both with indirect objects (as in 15) and with dative < e xperiencers > (as in 16 ). Crucially, while the clitic is optional for non RPP objects, its use is obligatory for < e xperiencers > : (15) Mara (le) escribi a Katherine Mara.NOM (3SG .DAT) wrote to Katherine .DAT ( 16) La msica *(le) gusta a Chad The music.NOM *(3SG .DAT) likes to Chad .DAT The obligato ry presence of the clitic in (16 ), Montrul claims, is related to the clitic doubling nature of the Spanish language, a phenomenon more akin to subject doubling than to indirect object doubling (1995:184). In the case of the latter, clitics are considered the overt morphological manifestatio n of indirect object agreement on the verb. Clitic doubling is explained in terms of movement (to AgrIO) before or after the spell out stage Examples 17a and 17b below better illustrate this idea
57 (17) a. Dmaris le escribi a Osmer Dmaris Cli.3sg.DAT wrote to Osmer b. Dmaris escribi a Osmer Dmaris wrote to Osmer Dmaris wrote to Osmer In the case of ( 17 a), the indirect object (a Osmer spec of AgrIOP but it moves overtly in (17b), thus for cing deletion of the clitic (1995:185) A representation is given in figure 3 3 below. Figure 3 3 Clitic doubling in Spanish In the case of the former (dative < experiencers > with class III psych verbs ), clitics seem to be inherent (1995:185) and revea l a case of subject doubling. Since dative < e xperiencers > exhibit many subject like properties, Montrul proposes that dative < experiencers > are in fact dative subjects, and that the clitic, which is base generated in AgrS, functions as another morphologica l property of the subject. See figure 3 4 below.
58 Figure 3 4 Dative < experiencers > with RPPs 3.3.4 Landau ( 2010 ) Landau (2010) foc uses on the nature of (Object) as entities with special grammatical properties and adopts Pesetsky (1995) propos al to further divide < e xperiencers > into (i) s ubject < e xperiencers > and (ii) o bject < e xperiencers > Pesetsky argued that the thematic roles involved in each of the two subgroups are not identical, claiming the need for expan ding the Thematic H ierarchy to i nclude Causer & Target/Subject Matter (henceforth T/SM). While Subject take T/SM as the object, Object have a c auser as the subject. Landau builds on such previous research to provide a comprehensive, yet compact and accessibl e, analysis of the distinct properties associated with the syntax of p sych verbs which he refers to as p sych effects. Psych effects are syntactic properties
59 associated with < experiencers > in non main claim is that all < e x perienc ers > are mental locations in the sense that they are considered the recipients tha t contain the mental/emotional/p sychological state 8 expressed by a psych verb. This abstract n otion is best exemplified in (18 ) below, where X is the and Y i s the As shown, (18a) can be interpreted as (18 b). (18 ) a. X frightens Y b. X causes fear of X to come to be in Y. (from Jackendoff (199 0:330) as cited in Landau 2010) And so it follows from the previous argument that, if < e xperiencers > are locati ons and locations are int roduced by a preposition, then < e xperiencers > should also be introduced by a preposition. For the particular case of Spanish p sych predicates 9 such a claim is non problematic for those < e xperiencers > that appear with c lass II and III psych verbs because they are always headed by a preposition ( Section 2). This is illustrated in (19) below. As can be seen, since c lass III psych verbs are always stative, they do not project v P. The < t heme > in this case is considered a Target/Subject Matter when it appears with c lass III psych verbs (Pesetsky 1995) Figure 3 5. Stative Psych Predicates 8 Psych effects are syntac tic properties associated with in non a genti ve readings ( e.g., Spanish Class III verbs). For example, in Greek, clitic doubling of accusative objects is generally optional but it becomes ob ligatory when the object is an
60 when the < experiencer > is spelled out but its presence would be ungrammatical otherwise. o claim that, in such cases, the < e xperiencer > is in fact efers to as An additional issue arises with class II psych verbs since they, as Pesetsky (1995) also noted, can alternate in their readings as either stative or eventive verbs. This alternation is shown in examples (19a & b) below where both asustar (to scare) an d molestar (to bother) can be used with an agentive and a non agentive meaning 10 The two readings, however, are undistinguishable morphologically. (19 ) a. Diana asust a Whitman (intentionally) Diana .NOM scare. 3 SG PRES to. Whitman .ACC. Diana sc ared b Diana molest a Whitman ( intentionally) Diana .NOM bother. 3 SG PRES to Whitman .ACC. Diana bothered In their a Diana intentionally causes n the case of 19 Diana intentionally b others b). A s seen in (19 a and b), these sentences can be passivized showing that indeed they are clearly not stative in these environment, with these specific meanings. 10 In non leista varieties of Spanish, these different readings can be easily teased apart by substituting the < experiencer > with its corresponding third person singular/plural accusative ( lo/la/los/las ) or dative ( le/l es ) pronoun ( e. g., Parodi Lewin 1991, Parodi & Lujn 2000). Speakers of leista varieties of Spanish, on the other hand, cannot make such distinctions because the indirect pronoun le/les is used in place of the direct object pronoun( lo/la/los/las ) when the accusative object is [+human, + masculine] ( e.g., Par odi 1991, Parodi & Lujn 2000).
61 (20 ) a. Wh itman es asustado por Diana Whitman is scared by D b. Whitman es molestado por Diana is bothered by This claim has obvious syntactic consequences: if a c lass II psych predicate is taken to have a stative reading, then it is treate d as an unaccusat ive verb and, obviously, it cannot be used a gentively. When that is the case, stative unaccusative c lass II psych predicates align with c lass III psych verbs and follow the structure represented in figure 3 5 above. On the other hand, if a class II psych predicate is taken to have an agentive reading, then v P is projected. Figure 3 6 below represents the relevant constructions for class II psychological predicates with a gentive read ings. As a consequence of this a gentive read ing, the < t heme > argument is not a t arget/ s ubject m atter but a c auser ( e.g., Pesetsky 1995). This discussion, referring to the alternation between transitive and unaccusative structures that seemingly define c lass II type psych verbs, will prove crucial to the syntactic proposal I will make for the representation of class III psych verbs in Heritage Speaker Spanish. Figure 3 6 Eventive Psych Predicates
62 In structures ( 3 5 ) and ( 3 6 ) above, the null/overt preposition is responsible for the assignment of dative case (2010 :21). Following Belletti and Rizzi Landau proposes that such inherent case is q uirky and that in the case of Spanish class III psych verbs, dative < e xperiencers > end up in subject position 11 (2010:81) and not as a topic or dislocated element ( Belletti & Ri zzi 1988). Conversely, other languages require DPs with nominative case in [Spec,TP] at PF ( e.g., English) Consequently, Landau, in line with other analyses ( e.g., Montrul 1995; Masullo1992), concludes th at even if previous studies of p sych predicates hav e treated dative < e xperiencers > as objects because they do not trigger verbal agreement, they do behave like subjects in several ways. For example, preverbal datives participate in raising constructions just like normal nominative subjects do (Masullo 1992 ) as in (21) below. ( 21 ) A Adriana parece gustarle la m sica coral Adriana seems to like choral music (Masullo:1992:92) As a result subjects 12 (2010:86). Consequently, all non nom inative < e xperien cers > are q uir ky (2010:86 88 ) including the < e xperiencers > that appear with Spanish class III psych verbs The a ssociation established between < e xperiencer > location allows Landau to conclude that movement of the < e xperiencer > to subject position is a cas e of locative inversion. In a locative inversion configuration, a locative PP and a subject DP switch positions. An example of locative inversion is given in (2 2 ). (22) a. Cuatro personas viven en esta casa 11 A Quirky subject is an argument that displays most canonical subject properties (except for agreement), but bears inherent case (Landau 2010:81). 12 Landau comments that
63 b. En esta cas a viven cuatro personas According to Landau, this movement is made possible by the [loc] 13 feature t hat resides on the head of the < e xperiencer > (2010:118). In order to accommodate a landing site for this movement, and a ccount for a PF and an LF subject, Landau proposes the occurrence of multiple project ions of [Spec,TP], one for the < e xperiencer > and one for the < t heme > (in any order). In the particular case of Spanish class III psych verbs the < e xperiencer > argument mu st raise overtly to [Spec,TP], and because it already has inherent case, nominative case is blo cked (Landau 2010: 82 83). The < t heme > could raise covertly to the second [Spec,TP]. With that said, the (LF) final representations of p sych p redicates can be s e en in the structures below. Figure 3 7 repre sents class II psych predicates with an agentive readings. Figure 3 8 represents c lass III psych predicates, as well as c lass II psych predicates with stative readings. Figure 3 7. Eventive Psych Predicates: L F 13 According to Landau, all locative relations are encoded by the feature [loc] (2010:89).
64 Figure 3 8. Stative Psych Predicates: LF Alternatively, other languages ( e.g., English, French, Hebrew) may exhibit a mirror image process of what was previously described, that is, the nominative raises overtly to subject position and the would later move at LF to the second projection of [Spec,TP] 14 As discussed the uni que properties associated with ( r everse ) p sychological p redicates have long been a topic of debate resulting in various theoretical account s While most of th e analyses reviewed here share certain assumptions, in varying degrees, about class III psych verbs such as inherent case o r the subject qualities of the < experiencer > their specific approaches differ noticeably. This, however, does not mean that the anal yses are incompatible or wrong. Additionally, and to be clear, the 14 To account for the optionality in word order discussed in section (3.1) in Spanish, one woul d need to simply stipulate the is the one that raises overtly and is followed by the covert movement of the , and of course this option is motivated by pragmatic considerations. In other words, these facts present no problem for Landa
65 goal of this dissertation is not to show evidence in support (or opposition) of one analysis over another. Rather what is important is that clear acquisition predictions can be derived fro m the differences that exist between the syntactic representations of class II and class III psych verbs This will prove crucial to the hypothesis I advance in C hapter 4 I will now turn fr om the theoretical descriptions to reviewing the acquisition lite rature for class III psych verbs in Spanish. 3.4 Acquisitional Studies O ver the last decades, e xtensive research on the acquisition of p sych predicates has repeatedly shown that this particular domain is a challenging property to acquire. In fact, class III psych verbs represent a n important learnability problem for the language learner since she has to intuit and then acquire the associated non canonical mapping of thematic roles to syntactic positions. The following subsections review the most relevant stud ies from L1 acquisition ( Section 3 .5.1), L2 acquisition ( Section 3 .5.2), and HS acquisition ( Section 3 .5.3). As is to be expected, the sections become increasingly more detailed as a function of their applicability to the current project. That is, I brie fly review studies from the L1 and L2 literatures given their connectivity to the project in the sense of understanding the challenging nature of class III psych verbs acquisition in general and in other instances of Spanish acquisition. However, I delve into much greater detail in the studies involving HSs since this is the target group of the current research endeavor. 3.4.1 L1 Acquisition The majority of the studies that have examined knowledge and/or use of psychological predicates have targeted monolingual speakers with some sort of
66 pathological conditions such as aphasi as or Alzheimer (e.g Beretta & Campbell 2001; Manovilidou 200 8; Thompson & Lee 2009). Results from these and other related studies have consistently demonstrate d that these verbs are probl ematic for these populations. Comparable conclusions have been reported when testing monolingual children without any linguistic or cognitive impairment. Lord (1979) and Figueira (1984), for example, independently showed that monolingual children of Engli sh and Portuguese, respectively, produce a high rate of errors in terms of mapping theta roles to their actual syntactic structures. This does not seem to be different for Spanish monolingual children. Although in testing the Maturation Hypothesis 15 Gmez S oler (2011) reported that, in an analysis of spontaneous speech from five children from the CHILDES data base, children apparently start producing gustar constructions target like at an early age (approximately at age 1;10), in an experimental study that analyzed the responses of 35 children to two comprehension tasks, Torrens, Escobar and Wrexler (2006) however, have argued that it is not until much later (the age of 6;0 approximately) that Spanish monolingual children start having knowledge of this type of predicates with comparable adult like accuracy. More recently Gmez Soler (2012) used puppets to conduct a comprehension experiment (Truth value judgment task) with 35 Spanish monolingual children ages 3 4 years old. The experiment consisted of a total of 12 items (8 critical tokens and 4 distracters) that included sentences with and without A movement (i.e. movement of a DP into an argument position) with four different verbs ( gustar 15 The Maturation Hypothesis (Borer & Wexler 1987) claims that there is a maturation of certain structures within UG such that not everything is fully available from birth, for example, they claim that A chains mature resulting in poor child performance on properties involving A chains, e.g., passives ( e.g., Wexler 1990; but see Fox & Grodinsky 1998).
67 encantar faltar dar asco study, each child was presented with a statement about puppet 1 ( e.g., La fruta favorita de Mickey es la naranja pero l odia la manzana is oranges but art of the previous statement but in this case the information presented was not always true ( e.g., A Mickey le gusta la naranja A Mickey le gusta el pltano After hearing the second statement, the child wa s asked whether puppet 2 had said the truth or not and why. 16 did not show differences between the tokens with and without A movement. This was taken as evidence that children as young as 3 year old were able to handle the absence of the external argument (2012: 10). Their performance, however differed in terms of verb type, that is, they were more accurate with gustar (79%) and encantar (78%) than with faltar (52%) and dar as co (49%). The higher accuracy on the former verbs is explained in terms of input frequency and markedness effects. Although in general Gmez Soler (2011, 2012) finds high accuracy in gustar like constructions (both productio n and comprehension), it is not clear how productive of a construction it actually is, meaning one cannot preclude some uses based on unanalyzed chunking. It seems reasonable to believe that it is considerably later that children use these predicates prod uctively, closer to the claims of Torrens et al. (2006). Support for this claim comes from experimental evidence based on interpretation as 16 The performance of the 3 year olds (58% accurate) and 4 year olds (68% accurate) were not significantly different (p=0.078).
68 well that this is a later acquired property, certainly later than normal agentive type predicates which are fully pr oductive very early in Spanish, as in other languages. 3.4.2 L2 Acquisition Not surprisingly, this particular structure also poses challenges for second language learners (White et al. 1999), especially if their L1 lacks class III psych verbs In principle, the acquisition task is no different than that of the Spanish child; however, adult L2 acquisition is complicated by among other variables, the L1 structure that in a sense has to be abandoned when the L1 and L2 differ in addition to the target structure bei ng acquired. As a result of these challenges, similar developmental patterns of errors have been found among L2 learners from different L1 backgrounds. That is, L2 learners usually tend to adopt the < experiencer > to be the subject in cases in which it is an object. This is not totally surprising since this pattern of error is in accordance wit h the Thematic H ierarchy ( example 11 above) since < experien cers > are projected higher than < t hemes > Montrul (1997), for example, studied the acquisition of Spanish gustar by French and English natives and observed that learners from both languages revealed a preference towards such linear word order even though French, unlike English, has class III psych verbs Similar results were also obtained by Montru l (2000), who investigated the acquisition of Spanish by English and Turkish natives, and Quesada (2008), who examined three different proficiency levels of English natives learning Spanish. Their results largely support the idea that L2 learners of Spanis h tend to assume a linear sentence structure when dealing with this sort of predicates a problem that persists even after pedagogi cal interventions take place (Lpez Jimnez 2003; Rubio 2005). Nonetheless, Quesada (2008), in line with Montrul (1997), note s that while
69 L2 learn ers at the initial stages of acquisition tend to conceptualize object < experiencers > a s the structural subjects, w ith time, and crucially as proficiency increases, learners go through a stage of optionality in which < experiencers > appe ar to function both as subjec ts and objects. Regarding the use of the clitic pronoun with class III psych verbs previous research has provided contradictory results. While Toth (2003) reported low levels of accuracy, at the initial stages, Quesada (2008) reported accurate results in terms of clitic agreement. natives were asked to complete four production tasks: (i) a description of a short movie; (ii) a personal narrative; (iii) a personal description; and (iv) a description about future plans. In general, the results of this study revealed that both the experimental and the control groups have high accuracy means with respect to the use of the clitic with psych verbs. With regards to verb agreement, Quesad a notes that although with time, the experimental informants considerably, the learners still have difficulties producing target like constructions (2008:60) Interestingly, she also notes that as a group, the native speakers also prod uce some non target like verb agreement configurations with psychological predicates. This, as we will see next has also been reported in acquisition studies with HSs. 3.4.3 HS Acquisition Class III psych verbs competence in Spanish HSs, appears to be, from the few available studies not very different from what has been reported for English Spanish L2 learn ers : the tendency is also to produce ( e.g., Dvorak & Kirschner 1982; Silva Corvalan 1994) and accept ( e.g., de Prada Pere z & Pascual y Cabo 2011, forthcoming ; Toribio & Nye 2006) what a priori could be considered targe t divergent forms. For example, one of the first studies if not the first one to
70 empiric ally examine class III psych verbs in HS Spanish in the US tested the use of class III psych verbs among 26 Puerto Rican HSs in New York City. Informants were asked to complete an English to Spanish translation task. Only 14 out of 50 English sentences elici ted the use of class III psych verbs in Spanish, and these wer e divided in to four conditions: (i) singular object singular subject ( e.g., (ii) plural object plural subject ( e.g., plural subject ( e.g., he likes sports a lot); and (iv) plural object si ngular subject ( e.g., do you all like this book?). Putting aside the obvious priming effects that a translation study might have a rds having the preverbal < experiencer > control verbal agreement instead of the prescribed postverbal < theme > Furthermore, it a abovementioned innovation in the SV agreement), t le was taken as evidence that the HSs did not consider the preverbal < experiencer > as the subject but as the surface object. In regards to the clitic, Dvorak and Kirschner observed a tendency towards the use of the singula r dative marker regardless of what would be the expected agreement pattern, that is, the third person singular form le tended to appear invariably irrespective of whether the < experiencer > was singular (in le Toribio and Nye (2006) examine d production and comprehension of psych verb constructions among 24 Spanish HSs in the US Participants completed a background
71 qu estionnaire and two tasks: an elicited (written) production task and a scalar grammaticality judgment task. In the written task, informants were presented with a question that contained a psychological verb and the first few words of the answer They were asked to complete the sentence using a psychological v erb. See (22) below for an example (22 ) A quin le encanta la salsa puertorriquea? (mi ta) In re gards to the judgment task, informants were asked to provide a judgment on 64 critical tokens (and 16 distracter tokens) according to three criteria: me suena bien sounds fine to me me suena ms o menos bien no me suena bien The critical tokens were further d ivided into grammatical and ungrammatical tokens. The grammatical tokens included three types of items: initial (as in 23a) theme initial (animate (as in 23b) a nd inanimate (as in 23c)) and (clitic+)verb initial (as in 23d) ( 23 ) a. A los atletas les gusta el entrenador b. Las chicas le gustan al joven c. El motor antiguo les gusta a los mecn icos d. Le gustan las rosas amarillas a la anciana the o The remaining tokens targeted ungrammaticality in terms of verb agreement (24a) and clitic agreement (24b) as well as omission o f the dative marker (24c) and case marking of the (24d), as shown in examples below.
72 ( 24 ) a *Al cantante le gusta las admiradoras the Sing b. A la escritora les gustan las narraciones breves th e writer likes short c. *Las rosas amarillas le gustan la anciana t d. *Yo me gustan los mangos In addition to find ing strong evidence in support of Dvorak and invariable le Toribio and Nye (2006) also obtained results that revealed three interesting tendencies : (i) indeterminacy with respect to the mapping of the arguments to syntactic positions, although a preference towards a linear word order was clear (i.e. that the control s verb agreement) (ii) omission of the dative a marker, and (iii) that animate DPs are possibly taken to be the structural subject irrespective of actual grammatical function. In general, these tendencies indicate that Spanish HSs use strategies towards a more transparent mapping of class III psych verbs which result in a reduction of the word order possibilities existing in the monolingual varieties. Relatedly, Montrul and Bowles (2009) studied knowledge and use of the Differential Object Marker a (DOM) among US Spanish HSs and reported a generalized to examine whether absence of DOM could also be generalizable to other instances of inherent dative case, namely class III psych verbs To specifically examine this issue, 13 Spanish natives and 69 HSs completed an acceptability judgment task that targeted grammatical and ungrammatical tokens related to DOM (25a,b), indirect objects (26a,b,c,d), and gustar like verbs (27). Informants were instructed to use a 1 (non native like) to 5 (native li k e) scale to judge the tokens provided to them. Example items follow:
73 (25) a. Javi conoce *(a) mi h ermana animate direct object Javi b. Gerard visit (*a) la bibl ioteca inanimate direct object Gerard (26) a. Ernesto envi regalos *(a) su novia indirect object Ernesto b. Teresa le devolvi el dinero *(a) Israel dative clitic doubling mo ney back to Israel c. Gillian dio *(a) Mara el libro double object construction Gillian d. *(A) Susana le trajo el plato Isabel Clitic left dislocation To Susana, Isabel brought the plate (27) a. *(A) David le gusta Ana V b. David le gusta *(a) Ana V Their data confirmed that for these participants problems with dative marking omission seem to extend to dative in class III psych verbs This was in terpreted to be a result of reduced lingui stic input in childhood combined with influence from English. Following on this research, Montrul and Bowles (2010) explored the effects of instruction on dative case marking in heritage language learners. P articipants (45 HL learners and 12 Spanish native speakers) completed two tasks. The first one was a written translation task targeting (i) transitive verbs (with animate and inanimate objects), (ii) ditransitive verbs, and (iii) gustar type verbs. The sec ond task was a 1(totally incorrect) to 5(totally correct) scalar grammaticality judgment test that targeted (i) sentences containing the so called personal a (ii) sentences containing the preposition a with ditransitive verbs, and (iii) sentences cont aining psych verbs. To examine the role of instruction, two versions of each task were created ( e.g., pre and
74 post intervention). Each participant was asked to complete the pre intervention version of the experiment as well as the post intervention versio n. The intervention consisted of providing the learners with positive ( e.g., explanation and examples) and negative ( e.g., contrasts between the structures in both languages) evidence as well as practice exercises. The overall results revealed that instruc tion improved HS performance. This was taken as evidence that HL learners, unlike in the case of L1 acquisition, are sensitive to the effects of negative evidence and explicit feedback. Most recently, de Prada P rez and Pascual y Cabo (2011) tested basic subject verb agreement and clitic agr eement with two high frequency r everse p sych predicates ( gustar encantar 17 experimental participants as well as 10 Spanish natives completed a scalar grammaticality judgment task. This test consisted of a total of 72 items (24 critical tokens and 48 distracters) in which the informants were presented with a short paragraph, a question, and 4 answers The answers were distributed as follows: 1 grammatical sentence, 1 ungrammatical se ntence because of verb agreement, 1 ungrammatical sentence because of clitic agreement, and 1 ungrammatical sentence because of both verb and clitic agreement. Informants were tested on sentences with 3rd person sin gular 3rd person plural < theme > (as in 28 below) and sentences with 3rd person plural 3rd person singular < t heme > (as in 29 below). (28 ) A Katherine le gustan los kiwis To Katherine her DAT. CLI like 3 RD P L the kiwis kiwis 17 In addition to the abovementioned experimental task, participants completed a sociolinguistic answers to the DELE were considered as the basis for the group divi sion (AD V.=25; INT.= 23; LOW= 13)
75 (29 ) A mis padres les To my parents them. DAT. CLI like 3 RD SG Their overall responses revealed tw o important tendencies. First, HS participants demonstrated robust knowledge of clitic agreement; evidence that questions the notion of the invariable le (Dvorak & Kirshchner 1982; Toribio & Nye 2006). Second, H Ss revealed a strong preference for the use of the thi rd person singular verbal form, coined invariable gusta (de Prada Prez & Pascual y Cabo 2011) regardless of actual subject verb agreement. Consider example (30) below where the ( e.g., Dela no) is followed by the singular form of the verb < theme > is singular ( e.g., la chaqueta ) or plural ( e.g., las chaquetas ). While the former is prescriptively correct, the latter is not. (30 ) A Delano le gusta la chaqueta/las chaquetas To Delano DAT.CLI Like. 3 RD SG the jacket/jackets Most interestingly, these preferences were in some ways comparable to the group of Spanish native speakers (n=10 ) tested. The controls, while Spanish dominant, were L2 learners of English. This of course was an unexpected finding a priori yet significant and influential for the present study In other words, these results revealed the need to include, in addition t o a group of monolingual speakers, a group of 1 st generation immigrants since these individuals are the ones that provide the input to HSs. As was previously established, the study of L1 attrition relates directly to HS acquisition issues. That is, HSs ge nerally receive input from 1 st generation immigrants who in turn are experiencing L1 attrition. In this context, Cazzoli Goeta and Young
76 Scholten (2011) investigated knowledge and use of sentence initial non nominative constructions ( e.g., r everse p sycholo gical p redicates) by adult Spanish speaking immigrants to the UK. Their study included 24 informants whose ages ranged from 25 to 65. Length of residence in the UK (5 to 25 years) and the level of contact/interaction with the English language (high contact and low contact) were hypothesized to play a deterministic role in terms of linguistic patterns ( e.g., monolingual divergent/convergent patterns). In addition to the experimental group, 10 Spanish monolinguals (ages ranging from 15 to 70) were used for co mparative purposes. P articipants completed two experimental tasks: (i) a picture description task and (ii) an aural preference task. In regards to (i), participants were asked to describe 10 pictures representing different actions The goal of this task w as to elicit structures containing sentence initial non nominatives. The goal of (ii) was to obtain grammaticality judgments on 24 pa irs of sentences that targeted, among other propert ies, nominative and non nominative subjects. Overall results revealed a strong preference towards nominative over non nominative subjects. The data also showed a correlation between the acceptance rate of ungrammatical sentences and the amount of time spent in the UK as well as level of contact/interaction with the English lan guage. 3.5 Conclusion As I have shown in this chapter, class III psych predicates have been documented to be inherently challenging to acquire for first, second and heritage language learners due to the non canonical mapping of the arguments to syntactic posi tions. Furthermore, b ased on the structural differences that exist between the two languages involved ( e.g., English and Spanish) as well as the results obtained in previous research (Dvorak & Kirschner 1982; Montrul & Bowles 2009; Toribio & Nye 2006; de P rada Prez & Pascual
77 y Cabo 2011), I predict that Spanish HSs will show differences for class III psych verbs and the properties associated to them. Chapter 4 further describes the present study by providing the research questions that guide the present s tudy as well as a falsifiable set of hypotheses and predictions that could explain the patterns of use found in the literature Additionally, C hapter 4 includes a detailed description of the methodology specifically designed to obtain answers to the resear ch questions put forth
78 CHAPTER 4 EMPIRICAL STUDY : RESEARCH QUESTIONS AND METHODOLOGY 4.1 Introductio n The main goal of this chapter is to present a detailed account of the methodology employed in this study. The chapter begins with a review of the research questions (and corresponding hy potheses) and then discusses the subject populations, materials and test design. 4.2 Research Questions & Hypotheses 4.2.1 Research questions In C hapters 2 and 3 I reviewed the current debate in formal heritage speaker ( HS ) acquisition regardin g the source of vu incomplete acquisition ( e.g., Montrul 2008) L1 attrition ( e.g., Polinsky 2006, 2011) and missing input competence divergence ( e.g., Pires & Rothman 2009). Although determining t he extent to which heritage language ( HL ) grammars obtain as a result of either one or a combination of these possibilities is at the core of current HL acquisition research the number of studies that has attempted to do just this is scarce (e.g., Polinsky 2011 ; Cuza, Prez Tattam, Barajas, Miller & Sadowski 2013) To fill this gap in the literature, the present study examines the argument stru cture and semantic mappings of re verse p sychological p redicates a cross different generations of Spanish speakers in the US. With th e above in mind the research questions at the core of the present study are the following: 1) Are class III psychological predicates in HS Spanish undergoing a reanalysis of their argument structure? If so, in what ways and why?
79 2) To what extent are attrition incomplete acquisition and/or input delimited differences at least explanatory as sources of HS differences and to what extent can these three factors be teased apart? By conducting experimental research that focuses on the particular properties addresse d here and, methodologically, across several generations of speakers within a single Spanish dialect one has a better chance of teasing apart which factors or combination of factors are likely to be deterministic in the outcomes observed. 4.2.2 Hypotheses Base d on previous research ( e.g., Dvorak & Kirschner 1982; Montrul & Bowles 2009 ; de Prada Prez & Pascual y Cabo 2011 ; Toribio & Nye 2006), it is hypothesized that Spanish HSs will show differences (from the monolingual norms) for class III psych verbs and th e properties associated with them subject verb agreement) Since such differences are well documented already, I seek to go beyond further documentation and/or description of the differences alone, by offering fr om the outset a set of domain specific hypothes e s based on the syntax of class III psych verbs (i.e., Landau 2010; Montrul 1995; Pesetsky 1995) First, I predict that in HS Spanish, c lass III type psych verbs ( e.g., gustar those that only have available RPP structures in monolingual grammars, are either reanalyzed already or in a process of being reanalyzed as c lass II type psych verbs ( e.g., which have available both an RPP and an agentive syntax (as described in S ectio ns 3.2 and 3.3) In other words, there is a simplification of the Span ish system of psych predicates from three classes to two according to Belletti and shift towards a reduction favored by the majority language, English.
80 Recall from C hapter 3 that class III psych verbs are considered stative, that is, they cannot be used agentively, do not project v P, and are thus incompatible with passive constructions. Conversely, class II psych verbs can alternate from having an agentive and a stative reading and therefore they do have the possibility of appearing in passive constructions. Thus, i f on the right track, this hypothesis predicts that Spani sh HSs should (variably) accept canonical c lass III type verbs ( e.g., gu star in p assive constructions as an emerging reflex of the aforeme ntioned new/optional agentive reading. In other words, having an available agentive reading should allow this subgroup of verbs to appear in passivized constructions. This sort of constructions shoul d, a priori not be allowed in monolingual grammars because this shift is not hypothesized to be taking place in monolingual environments. That is, in monolingual verbs ( e.g., do not have an agenti ve sy ntactic structure available and therefore cannot be passivized. On the other hand, c lass II type verbs ( e.g., asustar available (even in monolingual environments) which allows them to be passivized To illustrate these differences, consider example (1a b) below. ( 1 ) a. La pizza es gustada por Jason the pizza is liked by b. Becky fue asustada por David As discussed, e xample 1b above should be grammatical for S panish monolingual speakers (as well as Spanish HSs) since asustar verb that can alternate from having an agentive or a non agentive reading. Example (1a), on the other hand, should be ungrammatical for Spanish monolingual speakers since
81 verbs which are characterized by only having a non agentive reading available However, if indeed class III verbs ( e.g., are reanalyzed as c lass II ( e.g., asustar environments, then HSs should be able to incorporate this optional structure in their grammars. Experiment I ( S ection 4 .3.2.3) is designed specifically to test this 1 Second, I also predict that one might observe cases of inva riable gusta ( S ection 3.5.3) with apparent su rface retention of the dative clitic (lack of verb agreement) as observed by de Prada Prez and Pascual y Cabo (2011). This could be accounted for by assuming that gustar assigned an agentive structure and that the apparent dative clitic is actually serving as preverbal agreement morphology that has nominal phi features (what postverbal agreement morphology typically has in Spanish). This would be compatible with surface productions of the following four types ( 2a d), all robustly attested in previous research ( e.g., Cazzoli Goeta & Young Scholten 2011; Dvorak & Kirschner 1982; de Prada Prez & Pascual y Cabo 2011; Silva Corvaln 1994; Toribio & Nye 2006) : ( 2 ) a. ( ) Yo me gusta la pizza I.NOM to me like. 3 RD .SG the pizza b. Me gusta la pizza to me like. 3 RD SG the pizza c. ( ) Yo me gusta las pizzas I.NOM to me like. 3 RD SG the pizzas d. ( ) Me gusta las p izzas. to me like .3 RD SG the pizzas 1 Some other ways to test this shift (stative>agentive) would be with the use of certain adverbs such as
82 extent that an agentive syntax is being projected in (a) (d), all are ungrammatical within monolingual Spanish even if the surface production is misleading in some cases. Since Spanish is a null subject language and HSs have been shown to retain this property ( e.g., Montrul 2008), example (2 b) produces a surface form that cannot be appreciated as different from mo nolingual Spanish assuming that the subject is dropped and because it happens to be that the < theme > object in this case (what should be the subject controller of agreement) is singular and thus renders opaque the underlying structure. Cases ( 2 a), ( 2 c) an d ( 2 d) are clearer Example ( 2a) is possible with an agentive syntax since in Spanish subjects can be overt and under th is proposal the clitic me is reanalyzed as the agreement morphology bearing the same features as the overt subject Example ( 2 c) is als o possible with an agentive syntax for the same yo that pizzas is plural is of no consequence s ince with an agentive syntax the controller of verbal agreement is the overt nominative subject This would actually represent what in the literature has been referred to as invariable gusta (de Prada Prez & Pascual y Cabo 2011). Example ( 2 d) is also poss ible for the same reasons as ( 2 c) but has a null subject 1 st person pronoun. One might also expect HSs to be less sensitive to the ungrammaticality of a fully conjugated paradigm of c lass III psych predicate with canonical Spanish agreement morphology ( e.g ., yo gusto t gustas l/ella gusta etc.) in experiments, mirroring their English equivalents, although this is less attested to in previous empirical work. Notwithstanding, some
83 studies do provide some evide nce of this very type of production in HS Spanish ( e.g., Silva Corvaln 1994) and L1 attrition ( e.g., Cazzoli Goeta & Young Scholten 2011) Crucially, what I do not expect HSs to accept under this account is an invariable gusta (examples (a) (d) above) with omission of the once dative clitic, which under this account has been reanalyzed as (obligatory) preverbal agreement morphology. For example, in the counter parts to (b) and (d), gusta la pizza and gusta las pizzas respectively the EPP has no way to be satisfied and the d erivations should crash. This approach also predicts that the dative marker a will be lost (as previously attested in e.g., Montrul & Bowles 2009; Toribio & Nye 2006) when the syntactic structure assigned is agentive precisely because the reanalyz ed subject needs to have nominative case. Third, because I argue that class III psych verbs shift to class II, this means that an agentive sy ntactic structure is not obligatory but rather that it acquires an optional reading As a result of this shift, o ne would expect to find variability whereby HSs sometimes produce and accept gustar/encantar used as true class III psych verbs (stative verbs) with the prescribed /canonical syntactic structure and sometimes favor the proposed innovation (hybrid stative/ag entive verbs) This would be akin to what Spanish monolingual s peakers do for verbs like asustar molestar to bother/annoy which can have either an RPP syntax or an agentive one. In this section I have presented the motivation and core r esearch questions that guide this study. Based on these questions, the study described in S ection 4.3 was designed to test the viability of the hypotheses proposed here.
84 4.3 Methodology 4.3.1 Participants Although I am interested in the question of HS linguistic co mpetence for a domain of Spanish that should not show HS dialectal differences I nonetheless designed the methodology to make use of maximally homogeneous groups. As such, because the Cuban community represents the largest Hispanic group in the state of F lorida (2010 US Census briefs) all 11 4 informants that participated in this study were exclusively Cuban 2 or of Cuban descent and, at the time of data collection, all resided in either Cuba (Havana and surrounding areas) or Florida (Gainesville & Miami), respectively. In the interest of disentangling the sources of divergence in HS grammars in this specific domain, I have adopt ed the methodological pairing of participant groups suggested in Polinsky (2011) As discussed, i n her study Polinsky analyzed the comprehension of relative clauses in child and adult HSs of Russian (and a baseline of child and adult Russian monolingual speakers) a group combination that provides an indirect way to separate incomplete acquisition from attrition. Like Polinsky (2011 ), I also collected data from adult and child HSs, as well as adult and child Spanish monolingual speakers. Recall that, a if a child and an adult deviate from the baseline in the same way, it can be assumed that the feature has not b child performs as his or her age matched baseline control but the adult does not, the feature can be assumed to have been acquired but may have subsequently been lost or 2 Although the history behind the immigration of the Cuban community is very interesting and important for studies that focus on the social and cultural aspects of this community it falls outside of the scope of this dissertation. For a detailed discussion on this topic (as well as other sociolinguistic issues), I refe r the reader to Alberts (2005), Boswell (2000), Buajasn Marrawi and Mndez Mndez (2005), Castro (1992a, 1992b), Castro and Roca (1990), De la Torre (2003), Lpez Morales (2003, 2009), Lynch (1999, 2000, 2009), Ortega (1998), Por cel (2006)
85 on (Polinsky 2011: 306). The results of th at study showed that both monolingual and HS child speakers revealed full mastery of relative clauses by age 6. the structure. Ba sed on these differences, Polinsky concluded that, for this particular domain, HS innovations emerge as a result of L1 attrition and not as a product of the fossilization of child language (i.e. incomplete acquisition). Notwithstanding its limitations, esp ecially as compared to a longitudinal study, comparing this combination of informant groups allow s us to indirectly tease apart the effects of incomplete acquisition and attrition in adult HS grammars more reliably. Thus, in trying to replicate (2011) methodology, I collected data from two control groups (child and adult monolingual speakers) and two experimental groups (child and adult HSs). Within the control groups, a total of 16 adult and 13 child monolingual speakers participated in this stu dy. Table 4 1 below provides a summary of the basic demographic information of these groups. Data collection from these two groups of speakers took place during the month of December, 2012 in different areas of Havana, Cuba (La Alt a habana, El Miramar, El V edado and La Habana Vieja). All adult monolingual speakers (4 male and 12 female) reported being born and raised in Cuba by Cuban parents. Of the 16 informants, only one reported having lived outside the country (Norway) for a period of approximately 2 mon ths. Additionally, besides three adult participants that reported having very basic knowledge of Russian, no other participants reported having knowledge of a foreign language, includin g English. In terms of education, 7 reported having completed (or being in the process of completing their college education), and 9 having completed secondary education.
86 Table 4 1.: Monolingual adult informants Place of birth Gender Age Education M1 Cuba F 29 College M2 Cuba F 71 High School M3 Cuba F 25 College M4 Cu ba F 68 College M5 Cuba F 63 College M6 Cuba F 60 High School M7 Cuba F 18 High School M8 Cuba F 18 High School M9 Cuba M 18 High School M10 Cuba F 33 College M11 Cuba M 25 College M12 Cuba F 49 High School M13 Cuba M 46 College M14 Cuba M 24 Hig h School M15 Cuba F 18 High School M16 Cuba F 30 High School Total 16 4 M/12 F Avg. 37;1 7 College/9 H.S. schoolsssssSchool SchoolScs In regards to the child participants, and d ue to the results obtained in a pilot study ( S ection 4.6 below) as well as the documented late acquisition of some of the properties that I am testing in this project ( e.g., Pierce 1992; Fox & Grodzinsky 1998; Borer & Wexler 1997), the monolingual (n=13) and HS children (n=18) g roups were limited to those who, at the time of dat a collection, were between the ages of 6;6 and 11;1 As can be seen in table 4 2, all monolingual children reported (i) being born and raised in Cuba from Cuban parents, (ii) never having left Cuba, and (iii) having very minimal knowledge of English. Wit h respect to the heritage speaker groups, a ll child informants w ere required to have been born and raised in the US from first generation immigrants Table 4 3 includes most necessary biographical information from HS children.
87 Table 4 2.: Monolingual chi ld informants Place of birth Gender Age Mother/Father Main Language MC 1 Cuba F 8 Cuban/Cuban Spanish MC 2 Cuba M 10 Cuban/Cuban Spanish MC 3 Cuba F 10 Cuban/Cuban Spanish MC 4 Cuba F 10 Cuban/Cuban Spanish MC 5 Cuba F 9 Cuban/Cuban Spanish MC 6 Cuba F 11 Cuban/Cuban Spanish MC 7 Cuba M 7 Cuban/Cuban Spanish MC 8 Cuba F 7 Cuban/Cuban Spanish MC 9 Cuba F 7 Cuban/Cuban Spanish MC 10 Cuba M 6 Cuban/Cuban Spanish MC 21 Cuba M 6 Cuban/Cuban Spanish MC 22 Cuba M 7 Cuban/Cuban Spanish MC 23 Cuba M 6 C uban/Cuban Spanish Total 13 6 M/7 F Avg. 8 Table 4 3.: Heritage child informants. Place of birth Gender Age Mother/Father Primary Language HC1 US F 8 Cuban/Cuban Spanish HC2 US F 8 Cuban/Cuban Spanish HC3 US M 11 Cuban /US Eng lish /Sp a n ish HC4 US M 9 Cuban/Cuban Spanish HC5 US F 9 Cuban/Cuban Spanish HC6 US M 11 Cuban/Cuban Spanish HC7 US M 9 Cuban/Cuban Spanish HC8 US M 7 Cuban/Cuban Spanish HC9 US F 8 Cuban/Cuban Sp anish /Eng lish HC10 US M 9 Cuban/Cuban Spanish HC11 US M 8 Cuban/C uban Spanish HC12 US M 8 Cuban/Cuban Spanish HC13 US M 6 Cuban/Cuban Spanish HC14 US F F 9 Cuban/Cuban Spanish HC15 US F 8 Cuban/Cuban Spanish HC16 US M 9 Cuban/Cuban Spanish HC17 US M 7 Cuban /US Eng lish /Sp a n ish HC18 US M 7 Cuban/Cuban Sp anis h /Eng lish Total 18 6 M/12 F Avg. 8.3
88 Most adult HSs were students attending Spanish classes at the University of Florida and Florida International University. Ranging in age from 18 to 24 years old (average 20;1), all adult HSs were US born or had arr ived in the U.S. before age 2 ;0 Crucially, as was stated above all informants in this group were also either Cuban or of Cuban descent, thus controlling for any possible effects regarding dialectal differences that may affect the final results The adult HS group (n= 50 participants) was further divided into advanced (n=21), intermediate (n= 24 ) and low 3 (n=4) proficiency groups according to their responses to a standardly used Spanish proficiency test. In an effort to make this project maximally comparabl e to other published works, was assessed with an adapted version of the DELE that has been used in a number of previous works (e.g., the work of Rothman) This instrument is described in detail in S ection 220.127.116.11 4 below. Although the monolingual data allow us to test the validity of the experimental design as a whole as well as the theoretical descriptions that appear in the literature, this group is, admittedly, not necessarily the optimal control group for our HSs (e.g Rothman & Iverson 2010). In fact, I argue that can be traced back to contact induced changes in first generation immigrant input providers, it is crucial for studies that seek to describe and explain HS competence to inc lude a first generation immigrant group as a more appropriate comparison group (Pascual y Cabo & Rothman 2012). Therefore, in addition to the monolingual data (children and adults) 3 Due to the small number of informant s who scored 29 or less in the proficiency test (n=4), the low HS group will not be included in this study. 4 Only the adult HS groups completed the Spanish proficiency test.
89 and to the experimental groups (HSs children & adults), I also collected da ta from a third group of adult bilinguals. This group includes a total of 1 6 individuals that were born and raised in Cuba as Spanish mon olingual speakers, coming to the US between the ages of 15 and 41 (average 22;8) and having resided in the U.S. for at least 10 years (average 24;1 years). This combination of participant groups improves on those examined in most HS studies since not only does it include 2 monolingual control groups (adult & children), it also adds crucial information regarding the source (s) of input that HSs receive with an additional group of 1 st generation immigrants of comparable ages and socioeconomic status, and that speak the same regional variety (as suggested by Montrul 2012:113). For that reason, I believe that this procedure can further tease apart (albeit still indirectly) attrition from true incomplete acquisition and even input delimited differences as the primary source of divergence in this domain. To sum up this section, in this study I collected data from a total of 11 4 i nformants 5 that were divided in 5 groups: (i) adult heritage speakers, (ii) child heritage speakers, (iii) adult monolingual speakers, (iv) child monolingual speakers, (v) adult L1 attriters. Table 4 4 below summarizes the five participant groups. Table 4 4.: Summary of informant groups Monolinguals Immigrants Heritage Speakers SPEAKERS Child n=13 ----n=18 Adults n=16 n=17 N=46 (21=ADV./25=INT.) Total (n=114) n= 29 n=17 n=64 5 Recall that low proficiency HSs (n=4) were not included in this study. Thus, o nly 110 participants were included in the current study.
90 4.3.2 Materials 18.104.22.168 Language Proficiency test As mentioned the in measured by means of a standardized language test 6 (i.e. Diploma de Espaol como Lengua Extranjera or DELE). This revised version of the test consists of two different sections: In the first section, participants we re asked to provide answers to 30 multiple choice questions that target general lexical proficiency as well as grammatical competency. Each token consists of a short sentence that contains a blank and four possible choices. These choices include lexical items and different verbal forms. An example is provided in (3) below The reader is referred to Appendix A for the full instrument (3) Aqu est tu caf, Juanito. No te quemes, que est muy_________. Here you have your coffe Juanito. Watch out, don very___________. ) a. dulce b. amargo c. agrio d. caliente In the second section of the test, informants are asked to select the most appropriate answers to a contextualized cloze test. In other words, informants are presented with a text containing 20 blank spaces and they are instructed to select the correct item for each space. For this activity, participants have to pick one of 3 possible choices which include, but are not limited to, preposi tions or verb tenses. 6 include self evaluation (e.g ., Flege, MacKay, & Piske 2002 ) speech rate (e.g. Polinsky 2008; Polinsky & Kagan 2007), and sentence comp letion tasks (e.g. Lynch 2000 ; Silva Corvaln 1994). For a detailed discussion on this topic, I refer the reader to Lynch (2012).
91 ( 4 ) Hoy se inaugura en Palma de Mallorca la Fundacin Pilar y Joan Mir, en el mismo lugar en donde el artista vivi sus ltimos treinta y cinco aos. El sueo de Joan Mir se ha ____________.(Today, the Pilar & Joan Mir foundation w ill be inaugarated in the same place where the artista lived the last 35 years of his live. His dream has been_________. a. cumplido b. completado c. terminado The sum of both sections adds up to a total of 50 points T he cutoffs for placement for each proficiency level are typically the following: Advanced= 40 or higher, Intermediate =30 39, Low= 29 or lower. This version of the DELE has been widely used in Spanish L2/HS acquisition generative studies in the last decade ( e.g. Cuza & Frank 2011; de Prada Prez & Pascual y Cabo 2011, 2012; Montrul 2000, 2002, 2005, 2009; Pascual y Cabo, Rothman, & Lingwall 2012; White, Valenzuela, Koz lowska MacGregor, & Leung, 2004 ). Although I am aware that using such a formal task may not the best option to test linguistic proficiency especially considering that HSs have been exposed primarily to an informal variety of Spanish (and have low if any literacy skills in the language). B ecause I want these data to be maximally comparable to other projects, this option was chosen. 22.214.171.124 Socio linguistic Background Questionnaire HSs can differ from one another in terms of age of acquisition of both the HL and the societal language, domains of use of the HL, HL proficiency, and HL literacy among oth er variables (e.g. Montrul 2012; Polinsky & Kagan 2007 ; Rothman 2009 ). T o also control for these variables, a comprehensive sociolinguistic background questionnaire was administered ( Appendix B ). This questionnaire collect ed general information including place of birth, places and length of residence, family members occupation
92 history, education background, and knowledge of other languages. It also inquire d English and Spanish as well as day to day l anguage use (with whom and in what contexts). Keeping in mind that the remit of this dissertation is one of examining HS mental linguistic representation, these data are used when appropriate to supplement and provide more precision to my analyses based o n the traditional way of testing grammatical competence given the traditions of the paradigm to which this dissertation bel ongs. A priori, the working hypothesis is that as time in the US and prolonged contact with English increase across the generations, more evidence of linguistic change in the domains of RPPs will emerge. Such a correlation, or lack thereof, will become significant for HS theorizing that increasingly understands the need of mapping the input to which particular HS generations are exposed to ( Sorace 2004 2012 ; Rothman 2007, 2009; Montrul 2008, 2011 ; inter alia ). 126.96.36.199 Test 1: Scalar Grammaticality Judgment Task (GJT 1 ) Participants (adults and children) were trained on and then asked to use a 1 (completely unnatural) to 4 (completely natural) Likert scale to rate a total of 40 sentences were not completely sure, did not know, or did not hear the sentence properly. The reader is referred to Appendix D for the ful l instrument Of the 40 sentences, half of them were critical items and the other half served as distracters In light of the prediction put forth above the critical items test acceptance/rejection with respect to four types of constructions: (i) passivi zed transitive constructions (as in 5) (ii) passivized unaccusative constructions (as in 6), (iii) passivized RPP constructions (as in 7), and (iv) passivized psych verb constructions (as in 8)
9 3 ( 5 ) Transitive sample token Esa carta fue escrita por Carol Carol ( 6 ) Unaccusative sample token Hugo es llegado a la casa Hugo ( 7 ) RPP sample token (Class III psych verbs ) (*) La pizza es gustada por Fernan do Fernando ( 8 ) PP simple token (Class II psych verbs ) Pau fue asustado por Jorge Jorge If this prediction is on the right track we should see evidence of c lass III psych verbs shifting to resemble c la ss II verbs. This would be evidenced by HSs converg ing with those of the control groups in the case of ( 5 ) since agentive verbs can appear in both active and passive voice constructions and ( 6 ) since both Spanish and English do not allow pas sivization with unaccusitives. Conversely, I expect to find differences within the different HS groups (e.g., child vs. adults if attrition is the source) and/or with respect to the control group judgments in the case of ( 7 ) since c lass III psych verbs ( i .e., stative predicates) cannot be passivized, unless they acquire an (optional) agentive syntax. That is, if on the right track, I expect HSs to variably accept gustar like verbs in passive constructions as a reflex of the new available agentive reading. The remaining 20 distracter tokens were divided in three different groups: (i) 10 tokens tested acceptance of grammatical sentences (as in 9 ), (ii) 5 tokens tested ungrammatical sentences due to noun adjective agreement (as in 10 ), and (iii) 5 tokens that tested ungrammatical sentences due to subject verb agreement (as in 11 )
94 ( 9 ) Nosotros vivimos en un apartamento ( 10 ) El carro blanca fue caro FEM car MASC ( 11 ) Ella hago la tarea todos los das 1 ST SG. Homework every day. As reviewed in Chapter 2, HSs do not usually receive education in the HL. To avoid any of these possible literacy effects (obvious for children and differentially applicable for HSs), tokens were presented in a video modality. The video lasted 8 minutes 16 seconds; the first minute provided instructions in Spanish and 3 practice items. The remaining 7 minutes included a total of 40 sentences. Between one sentence and the next, there was an 8 second pause. To ensure co nsistency across testing sessions ( e.g., controlling prosody), all subjects w ere asked to provide the judgments as they hear d the tokens from recorded videos of one male speaker born and raised in Cuba At the time of the recording, this speaker was 32 yea rs old and was starting his first semester as a graduate student at the University of Florida where he was pursuing a degree in Spanish literature. All adult participants were asked to give written judgments. Child participants reported their intuitions or ally. They were first distinction was used to choose between a 1 and a 2 (in the case that the judgment was negative) or between a 3 and a 4 (in the case that the judgment was positive). (1996) methodological suggestions: before the start of the test, children were presented with different words and sentence types ( e.g., declaratives, questions) and were trained
95 to focus on form, not on content. Throughout this training period, the children received as much practice and feedback as was deemed necessary It was responses consistently indicated that s/he understood the task at hand, that we began the data collection process. 188.8.131.52 Test 2: Scalar Grammaticality Judgment Task (GJT 2 ) To better explain the results obtained in the previous ex periment, all adult participants completed a second GJT. The main goal of this task was to examine the properties associated with class III psych verbs in relation with the hypothesis put forth in S ection 4.2 .2, that is, that the dative clitic is actually serving as preverbal agreement morphology (what postverbal agreement morphology typically has in Spanish). This task included a total of 137 sentences. These sentences were divided into critical items (n=60) and distracter /filler items (n=77 ). The critical items are further divided equally among five conditions (5x12=60) that aim to test acceptance or rejection with respect to the proposals I spelled out in prediction (2) and (3) above: (i) the canonical use of class III psych verbs (ii) the omission of th e dative marker a (iii) the omission of the dative clitic le (iv) use of gustar grammatical construction that I propose with a nominative subject + clitic ( reanalyzed as agreement morphology) + invar iable gusta As was the case for GJT1, in this task all tokens were also presented in a video recording. The same person as in experiment 1 was used for this recording The recording lasted 19 minutes and 7 seconds. The first 1.53 minutes were used as ins tructions and included a total of 4 practice items. Between one sentence and the next, there was a 5 second pause. After listening to each token, participants we re asked to provide a 1 (completely unnatural) to 4 (completely natural) grammaticality judgmen t
96 Participants were asked to circle the option in the answer sheet provided. As was also the case for GJT 1 completely sure, did not know, or did not hear the sentence properly. What follo ws are representative examples for each of these 5 conditions. 1. Canonical use of class III psych verbs : According to my predictions, HSs should still be able to accept canonical uses of RPPs since a stative reading is still available to class II type ve rbs. (12) A Mara le gusta/an la /l as cucarachas To Mara DAT.CLI Like. 3 RD SG/PL the cockroach / e s cockroach / e Of the 12 tokens that make up this condition, 6 appear with a 3 rd person singular < experiencer > and 6 w ith a 3 rd person plural 2. Omission of Dative marker a : Prediction (3) anticipates that HSs will accept gustar type verbs in constructions that lack the dative marker a as long as the gets reanalyzed with nominative case. As suc h, I expect that they will reject gustar type sentences that lack the dative marker but fail to be reanalyzable (the ) as nominative: mi me gusta t te gusta (13) m me gusta la tarta/las tartas Me DAT.CLI Like. 3 RD SG the cake/the cakes Only 1 st and 2 nd person singular pronouns ( m & t respectively) are tested under this condition since with the a dative marker construction the morpholog ical distinction from nominative disappears at the surface forms ( (a)l/ella, (a)nosotros, (a) ellos/ustedes ). Of the 12 tokens that make up this condition, 6 appear with m ( 1 ST SG ) as the and 6 with t ( 2 ND SG )
97 3. Clitic omission: As desc ribed in the second prediction, I do not expect HSs to accept an invariable gusta form with omission of the once dative clitic since, under this proposal, the clitic serves as preverbal agreement morphology in such environments. Answers to this condition w ill test this assertion. (14) A Mara gusta la playa To Mara Like. 3 RD SG the beach Of the 12 tokens that make up this condition, 6 appear with a 3 rd person singular < experiencer > and 6 with a 3 rd person plural These tokens were combined with singular and plural in equal numbers. 4. Case agreement innovations: If the absence of the dative marker a prompts the reanalysis of the sentence initial argument as nominative I could expect HSs to start accepting tokens suc h as the ones illustrated in (15 a b). In other words, gustar, like other class II psych verbs, could have a fully inflected paradigm used agentively. (15) a. (*) Yo encanto la playa I. N OM love.1 ST SG the beach b. (*) Nosotros gustamos la playa We. NOM love. 1 ST PL the beach Of the 12 tokens that make up this c ondition, 6 appear with a 3 rd person singular < theme > and 6 with a 3 rd person plural < theme > Prediction (2) anticipates that in class III psych verb constructions the sentence initial argument gets reanalyzed as nom inative and appears with an invariable form of gusta (as observed by de Prada Prez & Pascual y Cabo 2011). As described, this would be possible because the once dative clitic has become preverbal agreement morphology (Montrul 1995) Of the 12
98 tokens that make up this condition, 6 appear with a 3 rd person singular and 6 with a 3 rd person plural < theme > (1 6 ) ( ) Yo me gusta la pizza/las pizzas I. NOM DAT.CLI like .3 RD SG. the pizza/the pizzas The remaining 7 7 tokens are divided among distracters ( sentences that contain problems with verbal agreement across transitive (n=1 2 ), unergative (n=1 2 ), and unaccusative (n=1 2 ) verbs (as in 19a c respectively)) and two types of fillers : (i ) grammat ical sentences (n=3 0 ) as in 17 and (ii) sentences that contain problems with adjectival agreement (n=11 ) as in 18 With the distracter (un) grammatical items in relation to the findings from de Prada Prez & Pascual y Cabo (2011). In other words, I aim to test whether the use of an invariable 3rd person singular form has extended to other different types of verbs ( e.g., transitive, unergative, unaccusative) or whether this is exclusive of cla ss III psych predicates Within each I test six different pronominal subjects (informal) 7 (17 19) below are representative examples of the distracter tokens used in this instrument ( 1 7 ) Yo bailo salsa con mis amigos (I dance salsa with my friends) 7 In Spanish, the formal version of the second person singular (i.e. usted) is conjugated as a third person singular. Within this condition, all informants should find this type of token acceptable.
99 ( 18 ) Yo bebemos caf todos los das (*I drink ( 1 ST PL ) cofee everyday) (19) a. 1 st person singular : Yo llega a casa muy tarde (*I arrived ( 3 RD SG ) home very late) b. 2 nd person sin gular (informal) : T sale todas las noches (*You go out ( 3 RD SG ) every night) c.2 nd person singular (formal) : Usted sale todas las noches (you go out ( 3 RD SG ) every night) d. 3 rd person singular : l entra en las fiestas con facilidad (He gets into ( 3 RD SG ) parties without any problems e. 1 st person plural : Nosotros aparece por la tarde (*We appear ( 3 RD SG ) in the evening) f. 2 nd person plural : Ellos desaparece del trabajo (*They dissappear ( 3 RD SG ) from work) Again, b ecause of the complex nature o f this task and the attention span of children, this task was only performed with the adult participants. 184.108.40.206 Test 3: P icture Elicited production task The goal of this task was to elicit production of gustar like verbs in different contexts and forms. The e xperiment included a total of 18 tokens, 6 of which served as distracters and 12 as experimental items Each item consist ed of four consecutive slides projected in a power point presentation. On the first slide, informants read a short description that con textualizes the specific sets of pictures that will appear on the three following slide s. At the same time that they read this short paragraph, and to avoid any literacy issues, participants were also presented with an audio file in which a narrator also r ead out loud and with normal intonation the same paragraph The narrator, who was the same Cuban male who performed the recordings for Tasks 1 and 2, was instructed to read the paragraph with normal intonation and at a normal speech. On the
100 second slide, t he informant simultaneously saw three pictures, which from left to right represented (i) one (or two) of the characters introduced to the informant during the (iii) t he object that triggers such emotion. The third slide served as a linking element between the second and the fourth slide as the y either saw the word y pero (but). The fourth slide reflected the opposite idea represented in the second slide. Tha t is, if character A liked object X in slide 2, character B disliked object X in slide 4. The 12 critical items were equally divided in 4 conditions that alternated singular and plural themes, always in third person, with singular and plural < experiencers > also always in third person. Tokens in each condition were designed to contain < themes > and < experiencers > differing from each other in regards to the number. That is, : (i) 3 rd person singular with 3 rd person singular ; (ii) 3 rd per son singular with 3 rd person plural ; (iii) 3 rd person plural with 3 rd person singular ; and (iv) 3 rd person plural with 3 rd person plural . Four representative examples (one for each conditio n) are given in 20a d below. In figure 4 1, the token belongs to condition (i) described above, that is 3 rd person singular with 3 rd person singular The narrator in th is token contextualizes the pictures by saying: Hoy todos van a jugar a la casa de Dilan. Su madre va a comprar helado. ( Today, His mother is going to buy ice cream. But
101 there is a problem because s he does not know which one to buy becau Informants were prompted to describe the following pictures : Figure 4 1. Singular < experiencer > _singular < theme > kes chocolate icecream but Dilan ) Figure 4 2 below belongs to condition (iii), that is 3 rd person plural with 3 rd person singular . The narrator of the following set of slides contextualizes the pictu res by saying: Hoy voy a invitar a los cuatro amigos a tomar caf en la cafeteria Versalles. Pero hay un problema porque yo s que (Today, I am going to treat the four friends to have coffee in the coffee shop Versailles B ut there is a problem because I k Informants were prompted to describe the following pictures : Figure 4 2. Plural singular < theme >
102 The target response was do (like coffee). Figure 4 3 below belongs to condition (iii), that is 3 rd person singular with 3 rd person plural . The narrator of the following set of slides contextualizes the pictures by saying: Carmen y Esteban estn hablando de insectos. (Carmen and Esteban are talking about insects. I Informants were prompted to describe the following pictures : Figure 4 3. Singular < experiencer>_plural t (like butterflies)). Figure 4 4 below belongs to condition (iii), that is 3 rd person plural with 3 rd person plural . The narrator of the following set of slides contextualizes the pictures by saying: Para comer, la madre de Dilan ha hecho hamburguesas. Hay un problema porque ella no sabe que mother has cooked some hamburguers.
103 Figure 4 4. Plural < experiencer > plural < theme > A Carmen y a Esteban le s gustan las hamburguesas, pero a Dilan y a Lilia n no le s gustan las hamburguesas Esteban like hamburguers, The distracter items mirrored the same structure from the critical items. The only difference between the distracter items and the criti cal tokens is that the former did not include emoticons that would tr i gger class III psych verb constructions while the latter did As can be seen, e xample (20 e ) below illustrates a distracter item The narrator of the following set of slides contextualize s the pictures by saying: Hoy Esteban y Carmen han ido al cine a ver la pelcula Casablanca Aparentemente han tenido (Today, Esteban and Carmen went to the movies to watch Casablanca Apparently, they have had very differ Figure 4 5. Distracter i tem
104 Esteban llor con la pelcula Casablanca, pero Carmen se aburri con la pelcula (Casablanca). Carmen was bored with the mov ie (Casablanca) ). Prior to the beginning of the experiment, a ll informants receive d training on the task and become familiarized with the 4 characters that appear throughout this experiment. Recall that, according to de Prada Prez and Pascual y Cabo (2011 ), it is not clear whether this preference towards an invariable gusta form is a byproduct of a phonological simplification in which /n/ gets weakened/elided in final position. If this were the case, one would expect similar outcomes with other words where /n/ is not functional morphology only. If not this would suggest that this observed pattern is not a phonological simplification, but rather one related to verbal morpholog y To test for the possibility of general /n/ reduction in coda position, the name s of the characters used all ended in unstressed /n/ (Dilan, Esteban, Lilian, and Carmen). The prediction is that if the use of an invariable 3 rd person singular verbal form ( e.g., invariable gusta) is the result of a phonological reduction/elision, this e lision would also be visible not only in verbal forms but in other parts of speech. Therefore, if my informants were to pronounce the /n/ in these names this would allow me to eliminate the possibility that an invariable use of a 3 rd gust a at least a general one that is not specific to verbal morphology. To make sure that participants become sufficiently familiarized with the names of the 4 characters, these appear in a total of 6 slides performing different activities. During the training period, each character is recognized by means of a sign that includes his/her full name ( e.g., Dilan, Esteban, Lilian, and Carmen) as well as their initial (D, E, L, and C respectively) displaced on
105 their chest. O nce the test begins, participants are only shown the initial on their chest. Furthermore, participants also have the opportunity to practice one set of slides before the start of the test. 4.4 Procedure for D ata C ollection As discussed, a total of 11 4 informa nts participated in this study 8 The process of data collection took place during the 2012 fall semester. All monolingual speakers, children and adults alike, were recruited through previously made contacts in the city of Havana, Cuba. Bilingual speakers w ere recruited from previously made contacts in the area as well as the snowballing technique in two different locations: Gainesville, FL and Miami, FL. Child participants were recruited from a (pre)school in the area. Prior to the beginning of the intervie w, each informant (or the parent/guardian of the child in the case of child informants) read and signed the consent form approved by the University of Florida IRB office ( Appendix C ). Testing, in the case of the adult informants (both monolingual and bilin gual), consisted of all three experiments described above. Adult informants were tested in the following order: (i) GJT 1, (ii) GJT 2, (iii) and elicited production task. The average testing period was 58 minutes. In the case of the child informants, testi ng consisted on only one experiment (GJT 1) described in Section 220.127.116.11. Though the length of the training periods varied depending on the age of the child, the average testing period was 21 minutes. 8 Recall from section 4.3.1 that I opted from eliminating the data from 4 Low HSs.
106 4.5 Procedure for Data A nalysis Once the informants or thei r parent(s)/guardian(s) signed the required consent form, they were assigned random participant numbers to respect their anonymity. At that moment, all personal information was completely detached from the documents pertaining to the experiment and the par ticipant number was exclusively used to identify the participants and their answers in the different sections of the experiment. The signed consent forms were kept secured at home office. The data gathered from the different background questionnaires and experiments were systematically coded and stored electronically using Microsoft Office Excel (2007). The sound files collected from the p roduction task were stored as individual MP3 files; one for each participant. These files were trans cribed by the investigator and subsequently checked for accuracy by two trained individuals. These individuals checked two different samples of approximately 10% each. In cases of disagreement, the token was transcribed again. A backup copy of all files as well as all tasks used within this project is stored on the t connected to any network. Upon comple tion of the coding process, and as will be described in C hapter 5 statistical tests were run between and w ithin groups. Chapter 5 will also discuss the results, which will be analyzed and compared against the claims of the theoretical descriptions present in the literature and against the hypothes e s and predications presented in Section 4.2 above. The experime will be examined to see whether their judgments fall within the confidence intervals obta ined from the control groups.
107 4.6 Pilot Study To make sure that the proposed methodologies were valid for the target groups and the goals of t his study, all experiments were successfully trialed with three HS children (ages 4;7, 4;9 and 8;6), ten adult HSs, and four 1 st generation Spanish speaking immigrants. Adult informants did not find any difficult ies when completing the tasks. In regards to the child HSs, it soon became evident that only the oldest participant (age 8;6) could provide appropriate grammatical judgments. Based on previous research with similar methodologies, and on what it is known from child L1 acquisition literature regarding passives in general, it was decided that only children ages 6;6 and above would be able to handle the proposed methodology. As a result, in the current study, I only collected data from child informants (HSs and monolinguals alike) that at the time of dat a collection were between the ages of 6;6 and 11 ;0 Moreover, this would en sure that monolingual children have also acquired the property at study target like results contrast with the responses obtained from the Spanish natives group suggesting this to be a fruitful ground for investigation. Interestingly, the data from the Spanish natives also reveal ed some level of variation, albeit much less than the HS group. Since these controls for the pilot were not monolin guals, but rather clearly Spanish dominant 1 st generation speakers in the US, it is reasonable to believe that such variation can have a compounding effect for HSs who receive input from 1 st generation immigrants of this kind ( e.g., Pascual y Cabo & Rothma n 2012; Rothman 2007 ; Sorace 2004 ). T his unexpected variation strongly indicate d that the 1 st generation group should be maintained as a formal control group. Nonetheless, as stated above, I have also
108 included a second control group made up by Cuban monoli ngual speakers residing in Cuba. 4.7 Conclusion In this chapter I have presented the motivation and core research questions that guide this study. Based on these questions, I proposed an explanatory analysis informed on findings from previous studies and base d on the syntax of class III psych verbs To test this analysis, I designed a methodology that includes 5 groups of Spanish speakers and 4 experiments. In C hapter 5 I present the results obtained from these experiments.
109 CHAPTER 5 RESULTS 5.1 Introduction In this chapt er I present the empirical findings of the experiments carried out with relevance to the research questions outlined in Section 4.2.1, repeated below for convenience. 1) Are class III psych verbs in HS Spanish undergoing a reanalysis of their argument struct ure? If so, in what ways and why? 2) To what extent are attrition, incomplete acquisition and/or input delimited differences at least partially explanatory as sources of HS differences? and to what extent can they be teased apart? With these questions in mind and adopting the pairing of participant groups suggested in Polinsky (2011) and beyond I designed a methodology that tested the abovementioned argument structure innovation while at the same time exam ining the sources of possible difference s grammars, namely incomplete acquisition ( e.g., Montrul 2008) L1 attrition (e.g. Polinsky 2006, 2011) and missing input competence divergence (e.g., Pires & Rothman 2009). Recall that the present methodology consisted on three experimental tasks, two of these tasks tested knowledge and use of class II psych verbs and the other one tested their production. These experiments were completed by a total of 114 informants distributed among 5 different groups: adult & child Spanish HSs, adult & child Spanish mon olingual speakers, and adult 1 st generation immigrants. As discussed, b y examining children and adult HSs as well as children and adult monolingual speakers all from the same dialect and with the same empirical methodology, I aim to document and understand the extent to which this grammatica l domain (i.e. class II I psych verbs) is possibly changing cross generationally in a language contact situation Recall that
110 f ollowing Polinsky (2011), the logic of such a comparison is the following: if a HS child and a n HS adult deviate from the baseline in the same way, one can assume that the feature has not been acquired, that being in her view in complete acquisition. However, i f a HS child performs as his or her age matched baseline control but the HS adult does no t, one can assume that the feature had been acquired but was later on lost or reanalyzed, that being attrition 1 Crucially by bri nging to the mix additional data from 1 st generation immigrants from the same dialect (i.e. primary HS input providers) I bel ieve that one is better able to map the influence of cross generational input modifications (combined with an effect of contact with English) to explain how HS knowledge is obtained in this domain in a progressively mor e English convergent and monolingual Spanish divergent manner. Methodologically, this crucial piece of the puzzle has been largely overlooked in previous research on HS bilingualism (but see Pires & Rothman 2009; Rothman 2007). I maintain here, as I have in the past elsewhere, that without co nsidering the effects that contact induced changes in first generation immigrants may have on second and subsequent generations of HSs, one cannot appropriately make any concluding remarks regarding the so called (in)complete nature of HS grammars (Pascual y Cabo & Rothman 2012). As discussed in C hapter 2, HSs d o not usually receive education in the HL and are therefore not assumed to have been exposed to forma l (or standard) varieties of Spanish A nd so, t o avoid any possible literacy effects (obvious esp ecially for children 1 Though this scenario can be hypothetically true, it is also possible that both children groups ma y reveal the same adult monolingual divergent judgments. In that case, these results would not be at all supportive of claims of attrition since we know that, at least the monolingual children, end up converging in judgments that are consistent with those found in the theoretical literature. This is a shortcoming in on the adult monolingual grammar for relative clauses in Ru ssian.
111 and differentially applicable for the adults HSs), tokens were presented in a video (oral) modality. Moreover, this methodological procedure guarantees that all participants were presented with the exact same presentational modality ac ross the different groups. The ensuing sets of data obtained were systematically coded and then submitted to statistical analyses using SPSS 17.0 Descriptive and inferential statistics ( i.e. repeated measures ANOVA as well as a binomial logistic regressio n with Bonferroni post hoc tests) were run between data sets and within group (counterbalanced) comparisons ( e.g., Norman 2010 for a detailed discussion on the appropriateness of such statistical measures) Such comparisons allowed me to compare the data a gainst the claims of the theoretical descriptions present in the literature and against the hypotheses I advanced in Section 4.2.2 What follows is the description of the results, experiment by experiment condition by condition, along with what these ma y reveal about the current status of class III psych verbs in HS Spanish. Additionally, and with the aim to better understand the source of HS differences for this particular property, the crucial comparisons to be made involve monolingual children with bi lingual children and monolingual adults with bilingual adults. A more detailed discussion of these results and what they bring to bear on larger questions as a whole is presented in C hapter 6. 5.2 Grammaticality Judgment Task 1 (GJT1) In this section I report the data obtained from GJT1, described in Section 18.104.22.168. The primary goal of this task is to test the prediction I put forward in Section 4.2.2 regarding the innovative use of the verb gustar The prediction was that class III psych verbs (e.g have been reanalyzed as class II psych verbs (e.g. in favor of this prediction would obtain
112 via an experimental demonstration of class III psych verbs taking an optional agentive syntax. As discussed, this sort of evidence was predicted to be compatible with the use of gustar in passive constructions. Recall that class III psych verbs or obligatory RPPs, are stative (unaccusative) and, just like other unaccusatives ( e.g., morir llegar incompatible with passive constructions Conversely, class II psych verbs have the option of be ing used statively or agentively and can therefore appear in passive constructions, indicating that when they are, they have an agentiv e syntax Again if on the right track, this hypothesis makes the following specific prediction: Spanish HSs should (variably) accept canonical class III type verbs in passive constructions as an emerging reflex of the aforementioned new/optional agentive reading. In other words, having an available agentive reading should allow this subgroup of verbs to app ear in passivized constructions to a degree of acceptability statistically different from monolinguals who should overwhelmingly reject this. To exami ne this hypothesis, p articipants were asked to use a 1 (completely unnatural) to 4 (completely natural) Likert scale to rate 20 critical tokens (distributed in 4 conditions that test structures relevant to the aforementioned prediction: passive constructio ns with transitive verbs (n=5), unaccusative verbs (n=5), class II psych verbs (n=5), and class III psych verbs (n=5)) and 20 distracter tokens (i.e. grammatical sentences (n=10) and ungrammatical sentences due to noun adjective agreement (n=5) and subject verb agreement (n=5 ) ) 5.2.1 Distract er I tems In regards to the di stracter items which I present first to show that all participants were able to handle the task itself, a three ( TYPE : Grammatical, *Verb agreement, *Adj. agreement) by six ( GROUP : Advanced Ad ult HS, Intermediate Adult HS Child HS,
113 Attriters, Monolingual C hildren, and Monolingual Adults) repeated measures ANOVA revealed a main effect for TYPE ( F( 2 96 ) = 716.47 p <.001 ), GROUP ( F( 5 97 ) = 9.277 p <.001 ) and a TYPE BY GROUP interaction ( F( 10 1 94 ) = 9.521 p<.0 01), The significant effect for type was expected because grammatical tokens should yield significantly higher ratings than the ungrammatical ones. Figure 5 1 below gives the average ratings for each of the distracter conditions for each group. Figure 5 1. Group means of distracter tokens. *ADJ= Adjective agreement; *Verb= Verb agreement; GRAMM=Grammatical ; HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigrants/Attriters; HC= Heritage Chil dren; MC= Monolingual Children; MA= Monolingual Adults As can be seen, this significant difference discussed above stems from the overall clear contrast that exists between an almost categorical acceptance of grammatical sentences and the tendency to reje ct ungrammatical ones. That said, results from Bonferroni multiple pairwise comparisons revealed that, for the groups tested, all three distracter conditions (including the two ungrammatical ones) differ at a
114 level of statistical significance (i.e., *ADJ v s. *Verb, p < .001; *ADJ vs. GRAMM, p< .001;*Verb vs. GRAMM, p<.001) Although such general trend is also visible in the HS children data, this group stands out from all others in that the ungrammatical sentences receive higher ratings (2.41 in the case o f *ADJ and 2.02 in the case of *VERB), while the grammatical sentences are judged significantly lower than the other groups (3.17). The results from Bonferroni post hoc tests indicate that within this group, these differences are statistically significant (i.e., *ADJ vs. *Verb, p < .001; *ADJ vs. GRAMM, p< .001;*Verb vs. GRAMM, p=.007). Similarly, the group of intermediate HSs also shows less categorical judgments for the ungrammatical sentences; especially those that correspond in the noun adjective agreem ent (2.19). The results from Bonferroni post hoc tests indicate that within this group, these differences are also statistically significant (i.e., *ADJ vs. *Verb, p < .001; *ADJ vs. GRAMM, p< .001;*Verb vs. GRAMM, p <.001). These results, I argue, should n ot be taken as a sign for target divergent performance for these two groups, since both of them do make a significant distinction between grammatical and ungrammatical tokens. Instead, the y reveal some indeterminacy which may possibly result from a number of explanations. First, one may contemplate the possibility of appealing to some sort of arrested development in the HC gr oup since their monolingual counterparts (matched in age and dialect) clearly outperform them for the properties examined. However, i f one accepts the (admittedly) hypothetical scenario in which, with time these HS children do eventually become an equivalent version of the HS adults tested herein or simply continue to progress even if still arriving at a different grammar from adults then one
115 would have to consider an alternative option since adult HSs do end up becoming more target convergent the ranges of the adult monolingual group for the properties tested herein incomplete acquisition cannot be argued t o be at play in this domain Instead, it can be reasoned that HS children may be experiencing a delay in the acquisition of the properties tested (i.e., noun adjective agreement) possibly as a result of the difficul ties associated with balancing the two grammars involved. That is, while Spanish overtly express such agreement, English does not and thus, it is possible that HS children may be engaging their Spanish lexicon on their English syntax, whereby noun adjectiv e agreement would need not be overtly expressed. Alternatively, it might not have anything to do wit h crosslinguistic interference as in the scenario just described, but be a result of a more generalizable delay stemming from bilingualism itself, that is, a temporal manifestation of having to allocate finite cognitive resources across the acquisition of multi ple system at the same time ( e.g., Bialystok 2009 for review). Another likely explanation can be found in the differential salient nature of the ungram matical tokens themselves. That is, while problems with noun adjective agreement in Spanish usually correspond to the absence or presence of a low perceptual salient form ( e.g., a mere vowel alternation in the case of gender (i.e. most verb agreement in Spanish tend to be more acoustically salient. Similar morphophonological issues pertaining to variable HS performance have been discussed in previous studies examining modality alternat ions ( e.g., Montrul 2009; Pasc ual y Cabo et al. 2012; Rothman, Pascual y Cabo, & Lingwall i n preparation) and differential object marking
116 ( e.g., Montrul & Bowles 2010) among comparable Spanish HS groups in the US. Moreover, the testing modality employed in this study would certainly lend support to this explanation since less salient forms tend to be easily concealed in oral speech. Up to this point, I have shown that tokens were generally on target, though somew hat indeterminate in the case of the HS children and Intermediate HSs. That said, all groups, including the HS children and the Intermediate HSs, were successful at making the necessary distinctions between grammatical and ungrammatical utterances. I take such an important distinction as evidence that all groups were able to handle the task itself, that both the experiment grammatical representations, at least for these p roperties, are more or less the same. Next, I proceed to discuss the results for the critical conditions which, for ease of presentation, I describe one by one. A more general discussion of these results is offered at the end of this section; the point at which I bring together the results from all groups under all conditions. Recall that the critical items in this experiment were distributed equally among four different types conditions. These conditions differed from one another in terms of the verb type ( i.e., transitive verbs unaccusative verbs class III psych verbs, and class II psych verbs). To test for statistically significant differences, a repeated measures ANOVA was run with the variables of GROUP and VERB TYPE The results of this ANOVA showed a main effect for VERB TYPE (F(3, 95) = 265.79, p <.001), GROUP (F(5,97) = 3.021, p<.014), as well as a high order interaction between GROUP and
117 VE R B TYPE (F(15 ,291) = 6.487, p <.001). In the next subsections, I further examine these r esults, condition by condition. 5.2.2 Transitive P redicates in P assive C onstructions To establish that informants were sensitive to the semantic and syntactic restrictions of the passive voice, they were asked to provide judgments to passive sentences with transitive verbs. Figure 5 2 below presents the average results across all groups for this condition ( Section 22.214.171.124) Figure 5 2 Group means of passives with transitive verbs. HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigr ants/Attriters; HC= Heritage Children; MC= Monolingual Children; MA= Monolingual Adults As expected all groups clearly accept passive constructions with transitive verbs. This is not surprising given that Spanish and English work the same in this regard. In spite of these strong acceptance rates, one cannot help but notice a gradual decline in the judgments of these items as they go from almost categorical ratings, in the case of the monolingual groups to more moderate ratings in the case of the bilingual groups. The results from a Bonferroni post hoc test indicate that the monolingual child group
118 already reveal target like accuracy with this construction as their responses do not differ statistically from those of the monolingual adults (p=1.0) or the fi rst generation immigrants (p=1.0) On the other hand, and in spite of being generally accepting, the responses from the heritage child group do differ statistically from those of the monolingual children (p=.002) as well as those of the monolingual adults (p<.001) and first generation immigrants (p=.004) In regards to the adult informants, it appears that all groups strongly accept the use of transitive verbs in passive constructions. This observation is confirmed statistically since no differences were found between the two adult HS groups (p =1.0), nor between the attriter group and any of the HS groups (AA vs. HS INT, p=.993 ; AA vs. HS ADV, p=1.0), nor between the monolingual adult group and the attriter group (p=1.0) or any of the HS groups (MA vs. HS INT, p=.125) (MA vs. HS ADV, p= .774). Given these results, and because HS children do show differences with respect to monolingual children (p=.002) and to monolingual adults (p<.001) one could be misguided into concluding that incomplete acquisition is responsible for such differences. But, if we compare the results obtained by HS children and HS adults one can see that the latter do eventually overcome the delays observed in the HC data. In other words, incomplete acquisition cannot be explanatory for this specific property in these specific groups either because as adults their monolingual counterparts (HS ADV vs. MA, p.774; HS INT vs. MA, p=.125) For the exact same reason, that is, because convergence into tar get like results takes place over time, attrition cannot be identified as the source of diff erences either. Instead, note that the mean responses of all adult bilingual groups (AA=3.55; HS ADV=3.38; HS
119 INT=3.27) are found in between those of the monolingua l adults (3.72) and the HC (2.94) and recall that such declining trend corresponds with significant p values when comparing the judgments of the HC group and those of monolingual adults (p < .001) and between the HC group and the mono lingual children group (p = .002 ), but crucially not between the HC group and the attriters group (p= .084 ), the advanced HS group (p=. 068 ) or the intermediate HS group (p = 87 ). Therefore, the results presented here are more in line with what the third alternative would predict ( i.e ., Missing input competence divergence). That said, because it was originally thought to describe differences in ultimate attainment, to better explain the results obtained in the present context, this hypothesis is in need of modification. Thus, in t he spirit of Rothman (2007) as well as Pires and Rothman (2009), differences in the input (qualitatively as well as quantitatively) may delay the time course of developmental sequencing and/or the path itself which then inevitably effects ultimate attainme nt. According to this view, the outcome of child heritage language acquisition is generally different from monolingual acquisition because the input does not provide the necessary primary linguistic data for ies of the mono lingual dialects being examined (Rothman 2007; Pires & Rothman 2009). Irrespective of the differences found across the groups, what is crucial here is that, so far, all groups strongly accept the grammatical constructions (i.e., passive sentences with tran sitive verbs as well as the grammatical sentences in the distracter tokens) and reject the ungrammatical ones. Although no conclusions can be drawn at this point, so far these results clearly indicate the design of the methodology and the procedures follo wed are valid for the populations tested. Left to be determined is the
120 restrictions of verbal passive sentences in Spanish are consistent with the theoretical description s found in the literature, an issue to which I turn next. 5.2.3 Unaccusative P redicates in P assive C onstructions As a counterbalance condition, and to further establish whether informants were sensitive to the semantic and syntactic restrictions of the passive voice, they were asked to provide judgments to passive sentences with unaccusative verbs. Figure 5 3 below presents the average results across all groups for condition 2 described in Section 126.96.36.199 Figure 5 3. Group means of passives with unaccusative verbs HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigrants/Attriters; HC= Heritage Children; MC= Monolingual Children; MA= Monolingual Adults As can be seen, all groups largely reject this sort of constr uctions. This was the expected outcome given that unaccusative verbs intransitive verbs that portray actions in which there is no agent involved are incompatible with the passive voice. Moreover, and differently from the results observed in the previous condition, bilingual
121 and monolingual children seem to behave homogenously as they both marginally reject passive constructions with unaccusative verbs (MC=1.75; HC=2.21). This observation is confirmed statistically as the corresponding p value does not rea ch s ignificant standards (p=.269). In regards to the adult groups, we find differences between advanced and intermediate HSs (p=.0002) but not between the monolinguals and the attriters (p=1.0). Moreover, the adult monolingual group differs statistically f rom the HS INT (p=.002) but not from the HS HS INT (p=.007) but not from the HS ADV (p=1.0). Further intragroup comparisons reveal that, in spite of this proficiency effect in adul t HSs, all groups without exception 2 make a statistically significant difference between passives with unaccusative predicates and passives with transitive predicates ( Section 5.2. 2 above). Such a distinction is crucial for the purposes of the present stud y as it provides unequivocal evidence that all groups are sensitive to the syntactic and semantic restrictions of passive sentences and that all have the ability to form argument chains which is required to move the object to subject position to p roperly a nalyze verbal passives. Having established such important transitive unaccusative distinction, it is now time to turn the focus to the fundamental issue of interest in this study, that is, the acceptance or rejection of class III psych verbs in passive con structions. 5.2.4 Class III P sych predicates in P assive C onstructions Recall that gustar is an unaccusative verb and, as such, should be incompatible with passive constructions. B ut b ecause I hypothesize d that, in HS grammars, gustar the prototypical class III psych verb has shifted to a class II psych verb which means 2 Results from Bonferr oni post hoc test : MA, p<.001; MC, p<.001; HC, p<.001; HS ADV, p<.001; HS INT, p<.001; AA, p <.001.
122 that it can take an optional agentive syntax it was predicted that HSs should (variably) accepted strict class III psych verbs the sentences in cluded in this condition were crucial to test the extent to which the aforementioned hypothesis was upheld To support this hypothesis, in addition to showing adequate behavior in the counterbalanced tokens, a point to which I will return at the end of thi s chapter, HSs would have to accept passive gustar significantly more than other unaccusatives to a degree that is different from monolingual adults who a priori should strongly reject this innovation. Figure 5 4 below presents the average results across all groups for condition 3 described in Section 188.8.131.52 Figure 5 4. Group means of passives with class III psych predicates HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigrants/Attriters; HC= Heritage Children; MC= Monolingual Children; MA= Monolingual Adults As can be seen, all adult groups seem to reject gustar in passive constructions. That said, this rejection is not observed to the same degree by all groups. For example, while m onolingual adults ( MA=1.10) and 1 st generation immigrants (AA=1.53)
123 categorically ADV=1.86; HS INT=2.21). Upon further investigation, the results from pairwise intergroup comparisons reveal statistically significant results that in turn provide important insights into the nature of these HS differences. For example, these comparisons indicate that while the responses from the two HS groups do not differ statistically from each other (p=1.0), t he responses from the mon olingual adult s do differ at a level of statistical significan ce from t hose of both HS groups (HS INT, p< .001; HS ADV p=.037)). This difference is essential for the purpose of this study as it is the first indication that class III psych predicates may ha ve p ossibl y undergo ne a reanalysis of their argument structure in the ways I hypothesized in Section 4.2.2. In other words these results indicate that adult HSs and adult Spanish monolingual speakers differ with respect to t heir intuitions of gustar patibility with passive constructions. Admittedly, this difference may not be as categorical as one would want it to be to indisputably confirm the abovementioned hypothesis; however, I argue that this is enough to show that these results are consistent wi th the proposed change in the direction that I hypothesized. This argument is based on the premise that the syntax of these HSs is dichotomous in nature and thus, it should either allow this sort of construction or reject it Because the adult HSs show dif ferences compared to the adult monolinguals in this respect, one can postulate that their syntax does in fact allow for this alternation. Additionally, intragroup comparisons reveal that the Advanced HSs make a statistically significant distinction between the conditions that tested acceptance of class III psych verbs and strict unaccusative verbs
124 predicates in passive constructions, then they should reject it more categorica lly, very much the same manner that they reject passives with unaccusatives. But why is it that their grammar would allow for this construction yet their acceptance rates are not as categorical as other grammatical constructions (e.g., passives with transi tive verbs)? The answer to this question may possibly correspond to a frequency effect. That is, even though these HSs have the two structures available, the agentive one is not as readily accessible as the stative one because it is not as frequent in the input, that is, these two structures are not in free variation. So far, we have seen how all groups categorically accept passives with transitive verbs ( Section 5.2. 2) and generally reject passives with unaccusatives ( Section 5.2. 3) In this section, I ha ve reported a few important differences in regards to the acceptance of class III psych verbs in passive constructions. These differences have been argued to result as a reflex of the emerging availability of the agentive reading in this group of verbs. A to conditions 2 and 3; that is, between unaccusatives and strict class III psych verbs This comparison is key since class III psych verbs are in fact unaccusatives and therefor e, both conditions should be considered equally ungrammatical. If they are not treated equally, then this would constitute further evidence that the change I hypothesized is only pertinent to gustar like verbs and does not extend to all unaccusative verbs. It is apparent, by taking a look at the two pertinent figures th at th ere are differences in the direction of the change I hypothesized. That is, in passive sentences, class III psych verbs seem to be accepted more favorably than class II psych verbs. If
125 these differences turned out to be statistical ly significant, then this would constitute additional evidence that class III psych verbs have in fact been reanalyzed. As expected, adult monolinguals and 1 st generation immigrants do not reveal statisticall y significant differences between these two conditions. For them, the use of class III psych verbs and unaccusatives in passive constructions is equally ungrammatical. Conversely, the two children groups do make a statistically significant difference betwe en them (MC, p<.001; HC, p=<.001). Because we know that these same children do make a distinction between transitive verbs and unaccusatives in the way one would expect ( Section 184.108.40.206), this second difference can be argued to indicate either changes in b oth children groups or perhaps simply a delay in their acquisition of class III psych verbs Because gustar like verbs have been reported in previous studies as difficult to acquire even among monolingual children ( Section 3.5.1), I believe that this secon d alternative is more plausible given that (i) adult monolinguals eventually end up converging with the descriptions found in the literature, and (ii) that adult HSs show development in the same direction. Having established that class III psych verbs ha grammars the question now becomes, which of the aforementioned sources of HS differences is responsible for this change? Recall that the current debate in formal HS acquisition regarding the source of vulnerability among HSs centered along the concepts of incomplete acquisition attrition and missing input competence divergence ( e.g., Montrul 2008; Polinsky 2006, 2011; Pires & Rothman 2009). This is a rather challenging question to answer since individual realities and experiences differ from one another in ways that cannot be fully controlled in a study of
126 this kind. Moreover, the methodological nature of this study (e.g., group pairings adopted) calls for an even more cautious argumentation since in the a bsence of longitudinal data, all we have is an approximate estimation that is provided by indirect measures. Given this operational necessity, we are to assume that the children examined herein will eventually become a somewhat corresponding version of the adult groups included here. That said by looking at the data as a whole, it is safe to assume that L1 attrition alone cannot be considered as the one source of HS grammar differences since both children groups perform worse than their adult counterparts In other words, both to what would be expected had L1 attrition been the cause of these differences. From an incomplete acquisition point of view, these results c an be interpreted in a similar way. That is, both HS children and monolingual children show what a priori could be described as incomplete acquisition since both groups alike seem to not be sensitive to the semantic and syntactic restrictions of class II I psych verbs However, if we are to assume that (i) the methodology employed in this experiment did not alter or influence their responses which in turn are thought to be representative of their linguistic mental representations, and (ii) that these child ren will become an equivalent version of the adults tested herein then incomplete acquisition alone cannot be responsible for these differences either. That is, as adults, both monolinguals and HSs reveal improvement with respect to the intuitions of th e children. If the child data obtained are due to true incomplete acquisition, then no developmental changes should surface from that moment on But because there is unequivocal development in both groups, an
127 analysis that points in the direction of a dela y in the acquisition of class III psych verbs should be favored. 5.2.5 Class II P sych predicates in P assive C onstructions The last condition of this experiment included passive sentences with class II psych verbs. These tokens were used to compare and contrast t knowledge of passive voice, especially with respect to (class II & class III) psych verbs. Figure 5 5 below presents the average results across all groups for this condition ( Section 220.127.116.11) Figure 5 5 Group means of passives with class II psych predicates HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigrants/Attriters; HC= Heritage Children; MC= Monolingual Children; MA= Monolingual Adults As was also the case with the condition that towards passives with transitive verbs ( Section 5.2.2 ), all groups tested herein clearly accept passive constructions with class II psych predicates ( asustar molestar tly grammatical construction given their hybrid
128 nature ( Section 4.2.2) which allows them to have both agentive or a stative readings available. Thus, when used in passives sentences, class II psych verbs must be unambiguously interpreted as having an agent ive reading. In spite of the diverse overall rating differences that range from 3.09 (in the case of the advanced HSs) to 3.77 (in the case of monolingual adults) Pairwise intergroup comparisons reveal that none of the groups statistically differ from eac h other for this specific condition. I take this to indicate that all groups, holistically speaking, have full fledged knowledge of the hybridity that char acterizes class II psych verbs. 5.2.6 General Discussion of GJT1 The main goal of GJT 1 was to examine the status of gustar in HSs Spanish as well as to test the hypothesis put forth in Section 4.2.2. Recall that the main difference between class II and III psych verb s lies in the hybrid nature of class I I which allows them to be passivized as opposed to class I II verbs which cannot. With this in mind, I predicted that HSs would be more likely to accept gustar in passive constructions than other unaccusatives, but not necessarily to the same degree as other class I I predicates M onolingual adults on the other h and, would reject this use since their syntactic representation for the lexical verb gustar only has a stative syntactic structure. To examine this prediction, informants were tested with four counterbalanced types of passive sentences : (i) passivized tran sitive constructions, (ii) passivized unaccusative constr uctions (iii) passivized co nstructions with gustar and (iv) passivized constructions with class II verbs In light of this, I expected that HSs would converge on their judgments to those of the cont rol groups in the case of transitive verbs, (i) & (iv) since transitive (strictly transitive verbs and class II psych verbs when they come as transitives) can appear in passive voice constructions, as well as in (ii) since both
129 Spanish and English do not a llow passivization with unaccusa tive predicates. Conversely, I expected to find differences within the HS groups (e.g., child vs. adults in a specific direction if attrition is the source) and/or with respect to the control group judgments in the case of (iii) since class III psych p redicates cannot be passivized unless they acquire an (optional) agentive syntax. To better observe the differences and similarities between the different groups, I present in figure 5 6 below the average results across all gro ups for all four critical condition s tested. Figure 5 6. Group means of all four critical conditions among adult groups TRANS= Passives with transitive verbs; UNACC= Passives with Unaccusative verbs; RPPs= Passives with reverse psych verbs; PPs= Passiv es with psych verbs ; HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigrants/Attriters; MA= Monolingual Adults Taken together, the data from GJT 1 show that while adult HSs informants generally exhibit monoli ngual like judgments the se also differ from the judgments of the two control groups in interesting ways. As I have discussed throughout the previous subsections, all groups generally demonstrate knowledge of the syntactic and semantic restrictions of the passive voice in Spanish since they tend to accept passive sentences with canonical transitive verbs and class II psych predicates (the only two grammatical
130 constructions) A t the same time, all groups are generally sensitive to the grammatical violations produced by the use of unaccusative predicates in passive constructions. That said, we observe differences in ratings of class III psych verbs in passive sentences. Upon further inspection, intragroup comparisons across all four conditions reveal e d that such performance differences correspond with statistical significant results that are consistent with changes in the argument structure of gustar in the ways that support the working hypothesis of this dissertation. To better describe these linguist ic outcomes, I will first review the overall results for each of the adult groups tested. The case of the monolinguals adults is very clear. Their judgments across the four critical conditions are unambiguously consistent with the descriptions found in the literature. That is, they strongly accept grammatical conditions (e.g., passives with transitive verbs, passives with class II psych verbs) and categorically reject ungrammatical ones (e.g., passives with unaccusative verbs, passives with RPPs, ungrammati cal distracter tokens). Like monolingual speakers, 1 st generation immigrants also show sensitivity to the (un)grammaticality of all four conditions. In fact, statistically, these two groups behave as one uniform group in all four conditions. Again, t his is not completely surprising since no changes are expected to take place at the level of the mental representations among the 1 st generation immigrants. What is expected however is (i) that 1 st generation immigrants will be less susceptible than monolingual speakers to ungrammaticality as it relates to the properties associated with strict class III psych predicates ( e.g., dative marker), and (ii) that in a production task, 1 st generation performance will be quantitatively and qualitatively differ ent than that of the monolingual adults. The idea is that it is precisely this hypothetical variation present in
131 the input provided by 1 st generation immigrants (whereby sometimes the property in question will be target like and some other times it will no t ) that plays a decisive role in changes to the syntactic representations in the formation of (monolingual different) HS grammars. In other words, because this sort of input will be variable, and because variability creates ambiguity for the linguistic par ser, HSs will be in a way forced to reanalyze aspects of the grammar in the most economical way; in the same sense as Lightfoot describes the role of children in diachronic change from a ge nerative perspective ( e .g., Lightfoot 1991, 1999 2010; Lightfoot & Westergaard 2007). As And as this internal language grows, new external langua ge(s) emerge, which in turn lead to growth of new internal representations. From this it follows that HSs differences do not necessarily obtain as a result of erosion at the level of mental representation among 1 st generation immigrants per se In fact, th eir mental representation may well be intact The claim is that continuous exposure to superficial performance innovations provided in the input from the immigrant generation contribute to changes in HS grammars. achronic linguistic changes, the HS regularizes the grammar in accord with the input she is exposed to from the perspective of what the input affords the parser. In this sense, emerging variation even if at the surface level for the provider can have gram matical consequences for the next generation. Certainly, class III psych verbs fall within this category as their inherent complex nature ( e.g. reverse word order), makes them perfect candidates for surfacing with performance errors among 1 st generation immigrants ( e.g., Cazzoli Goeta &
132 Young Scholten 2011) and potential innovation, in the case of the HSs (de Prada Prez & Pascual y Cabo 2011; Toribio & Nye 2006). To sum up to this point, a subgroup of adult HSs have shown crucial differences with respect to Spanish monolingual speakers in their judgment of critical tokens for GJT 1. These differences indicate that for these HSs class III psych verbs may allow for an agentive reading very much like class II psych verbs These results are consistent with the predictions I articulated in Section 4.2.2. This competence difference was hypothesized to emerge as a result of L1 attrition, incomplete acquisition and input delimited differences. In an attempt to disentangle the source of HS grammar differences, I now p resent in figure 5 7 the child data f or all four conditions examined. Figure 5 7. Group means of all four critical conditions among child groups TRANS= Passives with transitive verbs; UNACC= Passives with Unaccusative verbs; RPPs= Passives with r everse psych verbs; PPs= Passives with psych verbs ; HC= Heritage Children; MC= Monolingual Children T he child data, for the most part, show less defined intuitions concerning the (un)grammaticality of the different passive structures tested. This is speci fically true for
133 the child HS s whose responses, in spite of follow ing the general tendency observed by all groups are largely in the mid range of 2 to 3. Although as a group HC do not show categorical intuitions for either the grammatical or the ungrammat ical conditions, they do make a statistically significant distinction between passives with transitive an d unaccusative verbs (p < .001). As discussed, this distinction is important to show that they have general knowledge of the passive voice and its cons traints. That said, they do fail to make a distinction between transitive verbs and gustar (p = 1.0), as well as between strict transitive verbs and class II verbs (p = 1.0). This, along with the ir general predisposition to rate these conditions more posit ively than negatively, is of utmost importance because it reveals that HC are treating these three conditions as one. Crucially, p values do reach statistical significance when comparing passives with unnacusative verbs and passives with class III psych ve rbs with a clear preference towards the latter (p = 0.004). In regards to the monolingual children, they also seem to be sensitive to the semantic and syntactic restrictions of passive sentences since they make the distinction between the grammatical ( i.e ., passives with transitive verbs and class II psych verbs) and ungrammatical tokens (i.e., passives with unaccusative verbs). However, what was somewhat unexpected was their higher acceptance of class III psych verbs in passive constructions. Like the HC, monolinguals seem to be unsure about the ungrammaticality of this sort of constructions. What is certain is that as adults, these monolingual children are assumed to end up showing intuitions that converge with the descriptions found in the theoretical li terature. Therefore, one must reasonable conclude that these unexpected ratings must be attributed not to some sort of arrested development but to a
134 delay in the acquisition of class III psych verbs and all the properties associated with them. To conclude this section the data reported show that in comparison to the adult monolinguals (as well as the group of 1 st generation immigrants), adult HSs are generally more accepting of the presence of class III psych verbs in passive constructions. As an interim conclusion, this difference is consistent with the hypothesis I proposed in Section 4.2.2 as it supports the idea that this group of verbs may have undergone a reanalysis of their argument structure. Theoretically, such HS difference s have been purported t o obtain as a result of either attrition, incomplete acquisition, input delimited differences, or possibly a co mbination of all three options. Up to this point, the data available do not allow us to draw any definitive conclusions in this respect However we can tentatively discard some of these options For example, because as children, both monolinguals and HSs rate class III psych verbs equally, we cannot determine whether these differences in adulthood obtain as a result of true incomplete acquisition 3 or a delay in acquisition in childhood. The one option that we can discard is attrition in childhood since with time, both children groups tend to reveal improvement in a direction that is opposite to what L1 attrition would predict. That is, their jud gments become more target like with time (provided that we accept the methodological caveat that these HSs will one day become the equivalent version of the HS adults tested herein). Furthermore one could claim that because we see no statistically signifi cant differences between the ratings of the 3 Recall that the cross sectional analysis offered herein assumes that HC and MC will become their adult counterparts. This is a reasonable assumption to ma ke yet it is definitely not certain. In order to shed some light on this issue, one would have to carry out a longitudinal study.
135 first generation immigrants and those of the monolingual speakers, the ensuing differences in adult HS linguistic outcomes cannot be a result of input delimited differences from L1 attrition (i.e., MICD) Though this may seem like a somewhat reasonable assumption to make, I would like to argue that the opposite is true 4 : r ecall that what is being tested here is a syntactic reflex that follows from a particular mental representation (i.e., transitivity) That first generation immigrants show convergent judgments and therefore do not differ in such mental representation from their monolingual counterparts does not constitute counterevidence against this input delimited differences view since t he claim is not that int ernal changes take place in the grammars of the 1 st generation immigrants. In fact, w this perspective only makes explicit predictions regarding the occurrence of differences in the input provided by 1 st generation immigrants This does not necessarily imp ly that these differences are in fact representative of their mental representations Note however that these interim conclusions are highly speculative and should be taken as such until confirmed with by the remainder of the experiments included in this s tudy as well as other studies (preferably longitudinal). What is left to be determined therefore is whether the actual input that 1 st generation immigrants provide differs from that of the monolingual speakers. To this end, the subsequent experiments furth knowledge ( Section 5. 3 ) and use ( Section 5. 4 ) of class II psych predicates. 5.3 Grammaticality Judgment Task 2 (GJT2) In this section I report the data from the GJT2 (described in Section 18.104.22.168). The primary goal of this task is to provide a more fine grained analysis of the current status 4 This will be further supported in Section 5.3 when the production data is presented.
136 of class III psych verbs in HS Spanish with regard to the predictions I spelled out in 4.2.2. Recall that t he present task includes a total of 137 sentences. Of the m 60 were critical items an d the rest distracters (n=77). The critical items were further divided equally among five conditions (5x12=60): (i) the canonical (grammatical) use of class III psych verbs (ii) the omission of the dative marker a (iii) the omission of the dative c litic le (iv) *the use of gustar as a fully conjugated (agentive) verb, and (v) the proposed ( nominative subject + clitic (reanalyzed as agreement morphology) + invariable gusta ) Because of the more complex nature o f the task and the short attention span of children ( e.g., Crain & Thornton 1998 inter alia ) this experiment was only completed by the adult participants. To be clear, c hildren (HSs and natives alike) serve as comparative control groups to tease apart the possi ble sources of change, not as equally weighted experimental focus group s per se. The focus of this study is decisively on the adult HS grammar. Also r ecall that to avoid any possible interference effect lack of literacy skills in th e HL, all experimental tokens were presented in a video modality. The same person as in experiment 1 was used for this recording. After listening to each token, participants were asked to provide a 1 (completely unnatural) to 4 (completely natural) grammat icality judgment. As was also the case for GJT 1, I first present those results that correspond to the distracter items to show that all participants were able to handle the task itself.
137 5.3.1 Filler T okens As described in Section 22.214.171.124, this experiment inc luded a total of 77 distracter tokens. These were divided into two groups: fillers (n=41) and distracters (n=36). The fillers included 30 grammatical sentences and 11 ungrammatical sentences because of noun adjective agreement. The distracters, the remaini ng 36 tokens, were designed to test whether the invariable form of the third person singular reported by de Prada Prez and Pascual y Cabo (2011) for gustar had extended to other types of verbs (e.g., other unaccusatives, transitives, and unergatives). The refore, while the latter are clearly not fillers per se they also serve as distracters to the critical items described above. Figure 5 8 below shows the results obtained from all groups to the filler tokens. Figure 5 8. Group means of distracter tokens. *ADJ= Adjective agreement; GRAMM= GrammaticalHS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigrants/Attriters; MA= Monolingual Adults In this figure we can observe a clear asymmetry in the responses of the two filler grammatical tokens reach almost ceiling levels for all groups (yet we see a small decline
138 among the two HS groups), the ungrammatical tokens tested herein are largely rejected. Not surprisingly, this asymmetry corresponds to statistical significance across all four groups. The results from a two b y four repeated measures ANOVA with TYPE ( Grammatical, *Adj. agreement ) and GROUP ( Advanced HS, Intermediate HS, Attriters, a nd Monolingual Adults) as variables, revealed a main effect for TYPE (F( (1,70)= 595.026, p<.001) and a TYPE by GROUP interaction (F(3,70)=11.294, p<.001). Upon further inspection, the results from Bonferroni post hoc test s indicate that Intermediate esponses differ from those of the monolingual speakers and 1 st generation in both the ungrammatical (p=.003; p=.015 respectively) and the grammatical conditions (p=.002; p=.00 3 respectively). Notwithstanding the relevance of this statistical difference, th distinction even if less polarized. As was also the case for the ungrammatical distracter items that tested noun adjective agreement in the first experiment the informants seem to not be as sensitive to this ungrammaticality as one would expect. This, again, can be explained in terms of the low salient nature of the ungrammatical tokens themselves since noun adjective agreement in Spanish generally correspond to the absence or presence of a phonologically low perceptual form ( e.g., a mere vowel alternation in the case of gender above related morphophonological issues pertaining to HS performance have been examined in previous stu dies examining modality alternations (e.g. Montrul 2009; Pascual y Cabo et al. 2012; Rothman et al. In preparation) and differential object marking (e.g. Montrul & Bowles 2010) among comparable Spanish HS groups in the US. Additionally, the
139 testing modal ity employed in this study would certainly lend support to this explanation since less salient forms tend to be easily masked in oral speech. 5.3.2 Distracter T okens Recall that in addition to the true filler tokens, this experiment also included a total of 36 ( pseudo ) distracter tokens. These were divided equ ally among three conditions with sentences that contain a mismatch in subject verb agreement whereby an invariable 3 rd person singular is used across: (i) transitive verbs (n=12), (ii) unergative verbs (n=1 2), and (iii) (other) unaccusative verbs (n=12) As discussed, t he main goal of these ( pseudo ) sensitivity towards the (un) grammatical items in relation to the findings from de Prada Prez and Pascual y Cabo (2011). In other words, with these three conditions, I aim ed to test whether the use of an invariable 3 rd person singular form reported for gustar like verbs ha s extended to other different types of verbs (e.g. transitive, unergative, unaccusative ) or whe ther this innovation is exclusive of class III psych verbs. Figures 5 9, 5 10, and 5 11, below show the results obtained from all groups to these pseudo distracter tokens divided by verb type (transitive, unacussative, unergative) and subject prono un ( 1 st SG, 2 nd SG, & 3 rd SG; 1 st PL, & 3 rd PL) As can be seen in all three figures, and w ith one exception, all informant groups tend to reject these invariable forms for all three verb types and across all subject forms. The exception corresponds to the ratings of th e tokens that tested the only grammatical sentence type; namely the 3 rd person singular form 5 5 As described in the methodology sect ion, each one of these 3 conditions included 4 tokens that tested the abovementioned invariable verbal form with a subject form 3 rd person singular subject (l/ella= 2 tokens; usted= 2 tokens). Because these 4 tokens are grammatical, they served as counter balances to the remaining 8 ungrammatical tokens.
140 Figure 5 9. Group means of transitive distracter tokens. *ADJ= Adjective agreement; GRAMM= Grammatical HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigra nts/Attriters;; MA= Monolingual Adults Figure 5 10. Group means of unaccusative pseudo distracter tokens *ADJ= Adjective agreement; GRAMM= Grammatical HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigrant s/Attriters;; MA= Monolingual Adults
141 Figure 5 11. Group means of unergative pseudo distracter tokens. *ADJ= Adjective agreement; GRAMM= Grammatical ; HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigrants/A ttriters;; MA= Monolingual Adults To statistically confirm this observation, a repeated measures ANOVA was run with the variables of GROUP (Monolingual, 1 st generation immigrants, Advanced HSs, a nd Intermediate HSs), VERB TYPE (Unaccusative, Unergative, Transitive), and subject type (1 st SG, 2 nd SG, 3 rd SG, 1 st PL, 3 rd PL). The results show a main effect for TYPE (F(12,59)=77.383; p< .001), and a significant interaction between all three variabl es (F(36,822) =2.497; p < 0.001). Upon further inspection, the results from Bonferroni post hoc test s indicate that all groups behave more or less homogeneously with respect to these conditions since only monolinguals and advanced HSs differ significantly in regards to their judgments with transitive verbs (p =.023), and 1 st generation immigrants differ with intermediate HSs (p=0.29) in regards to their judgments with unaccusative verbs. Crucially, additional intragroup comparisons show that the judgments f or 3 rd person singular reached a level of statistical significance with respect to all of the other subject forms for all verb types (MA, p< .001; AA, p< .001; HS
142 ADV, p< .001; HS INT, p< .001). Though not so categorical, fairly high acceptance rates are a lso observed among the tokens that tested the 2 nd person singular subject form. This is not surprising given that in Spanish, the only difference between conjugating a verb in the 3 rd person singular form (present indicative) and a verb conjugated in the 2 nd person singular form (present indicative) resides in the presence/absence of a final /s/ Admittedly, this fact in of itself would probably render differences in the jud gments among speakers of different dialects of Spanish monolinguals However, recal l that to control for dialectal differences, all informants in this study are of Cuban origin and as is also the case in most Caribbean Spanish dialects, syllable and word final /s/ tend to be phonologically weakened, if not elided altogether in Cuban Spa nish ( e.g., Lpez Morales 2003 ). As an interim conclusion, fillers and distracter tokens in GJT 2 were generally on target. the intermediate HS group seems to show less defined judgments, especially towards the ungrammatical tokens. All in all, these results are also consistent with those obtained in the GJT 1. I take this to indicate that all groups understood the task at hand as demonstrated by their success in distinguish ing between grammatic al and ungrammatical utterances where expected; a necessary condition that validates the experiment and procedure of this methodology. Next, I proceed to discuss the results for the critical conditions which, for ease of presentation, I ag ain present one by one. A more general discussion of these results is offered at the end of this section; the point at which I bring together the results from all groups under all conditions.
143 5.3.3 Critical T okens The critical items in this experiment were dist ributed among five different conditions. Each condition consisted of the manipulation of a relevant grammatical property that tested the hypotheses put forth in C hapter 4. To test for statistically significant differences, a repeated measures ANOVA was run with the variables of Group (Monolingual; 1 st generation immigrants; Advanced HSs; and Intermediate HSs), and Type (Prescriptive use of gustar like verbs; Omission of dative marker a ; Omission of dative clitic; Case agreement innovation; New grammatical construction) The results of this ANOVA showed a main effect for Type (F(4, 67) = 201.93, p <.001), Group (F(3,70) = 5.139, p<.003), as well as a high order interaction between Group and Type (F(12 ,207) = 5.162, p <.001). In the next subsections, I furt her examine these results, condition by condition. 126.96.36.199 Canonical use of gustar like verbs Recall that my hypothesis ( Section 4.2.2 ) predict s that in HS Spanish, class III psych verbs are possib ly being reanalyzed as class II Such a change predicted that HSs would variabl y accept class III psych verbs in passive constructions. As we saw in Section 5.2.4, this prediction was partially borne out given the generalized differences found between HSs and monolingual adults. Crucially however, because in class II ps ych verbs, a stative reading (the only possibility for class III psych verbs) is still available the expectation is that HSs should still be able to accept class III psych verbs used in its canonical form Recall that the (monolingual) grammatical form of ten includes the presence of a dative marker as well as a dative clitic (pronoun), a preverbal < experiencer > and a postverbal < theme > that controls verbal agreement. Examples
144 (1a,b) below show the canonical use of class III psych verbs with a singular and a plural . ( 1 ) a. ( A Tania) le gusta el perro To Tania DAT.CLI Like. 3 RD SG the dog b. (A Tania) le gustan los perros To Tania DAT.CLI. Like. 3 RD SG/PL the dogs Tania As discussed, informants were asked to use a 1 (completely unnatural) to 4 (completely natural) Likert scale to rate a total of 12 prescriptively correct tokens. Figure 5 12 below presents the average results for this condition. Figure 5 12. Group mea ns of canonical class III psych verb tokens. HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigrants/Attriters;; MA= Monolingual Adults As expected, all informant groups strongly accept sentences that reveal the canonical use of class III psych verbs The results of a Bonferroni post hoc test further
145 corroborate the absence of any statistically significant differences between any of the four groups for this specific condition In isolation, these results are consistent with the hypothesis put forth in Section still allow this sort of constructions. 188.8.131.52 Omission of d ative marker a : Prediction II ( Section 4.2.2) anticipates that HSs will be able to accept gustar type verbs in constructions that lack the dative marker a provided that assuming their grammars however distinct from monolinguals are UG constrained the can be reanalyzed to have nominative case. As such, I expect ed that it should be easier for dou blet referential expressions (full DPs) in gustar constructions to be reanalyzed as nominative than it would be for 1 st and 2 nd person singular doublet pronouns which would be overtly marked as dative. This is true since in the c ase of the latter, only the morphological indication that the < experiencer > doublet is itself (i.e. the referential expression in isolation is otherwise is lacking for 1 st and 2 nd person singular doublets that the unique dative morphology that remains ( m and t as opposed to nominative yo and t ) would block the nominative reanalysis. Keep in mind that under the syntactic proposal of an emerging agentiv e syntax with class III psych predicates the (semantic subject) would become the actual syntactic subject when the agentive structure is realized and thus require overt nominative case agreement. The prediction is that this does not happen wi th 1 st and 2 nd is phonologically reduced is overtly marked as dative as opposed to being ambiguous as is the case with the other
146 grammatical persons pronominally and all referential expressions. I should acknowledge that it is possible to argue that m and t themselves have a status in HS grammars as both nominative and dative, which would be one way to account f or accepting these sentences I offer this proposal, however, to make testable predi ctions given the methodology of the present task. As will be seen the data to be presented warrant revisiting this and other possibilities, which I do later in the final analysis of the dissertation. Again, with this proposal in mind, it is predicted tha t all informant groups will reject gustar type sentences that lack the dative marker but are not easily reanalyzed (the ) as nominative: mi me gusta t te gusta 2 below illustr ates this a priori ungrammatical construction. (2) m me gusta la tarta/las tartas Me DAT.CLI Like.3 rd SG the cake/the cakes Recall that only 1 st and 2 nd person singular pronouns ( m & t respectively) were tested in this condition The logic behind this follows from the proposal reviewed above; making such reanalysis easier, for all other grammatical persons would render the morphological distinctio n between dative and nominative pronouns opaque at the surface level in the same way it does for referential expressions ( e.g., (a)l/ella, (a)nosotros, (a) ellos/ustedes ). Thus o f the 12 tokens that make up this condition, 6 appeared with m (1 st SG) as the < experiencer > and 6 with t (2 nd SG). Figure 5 13 below presents the average results across all groups for this condition.
147 Figure 5 1 3 Group means of HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Gene ration immigrants/Attriters; MA= Monolingual Adults Except for the monolingual group whose responses are generally consistent with the descriptions found in the literature, the rest of the informant groups do not seem to be as sensit ive to the ungrammaticality that results from the absence of the dative marker a Intergroup comparisons for this particular condition reveal that monolinguals differ in their judgments with respect to 1 st generation immigrants (p=.005) and intermediate HSs (p < .001), but not with respect to advanced HSs (p= .120). No other significant differences were observed between the groups in this condit ion. Crucially, however, within group comparisons reveal that with the exception of the intermediate HSs (p=.561 ), all groups do make a statistically significant distinction between the two conditions examined thu s far (i.e., canonical use of class III psych verbs vs. omission of dative marker: MA, p< .001; AA, p< .001; HS ADV, p <.001). But, in spite of these stat istical differences, the high ratings reported by most groups for this specific condition
148 seem to constitute evidence against the prediction put forth above. This is of course an unexpected outcome deserving of further consideration. One reasonable explan ation for these generalized higher acceptance rates corresponds to the low phonological saliency of the dative marker a In a sense, this is not that surprising given that previous studies have reported a generalized lack of instantiation of inherent cas e in the grammars of comparable Spanish HSs (e.g., Montrul & Bowles 2009). Moreover, and as was also suggested during the analysis of the distracter conditions for both GJT 1 and GJT 2, the methodological nature of the experiment used in the present study (audio visual) questions the impact of these results. In other words, because informants were presented with the tokens orally, and because, that all informants were unconsci ously filling in so as to make sense out of the tokens provided. So, if the low phonological salience of the dative ungrammatical items among the bilingual speakers, the question then becomes, why straightforwardly, the answer to this question can be found in the different (i.e., higher) processing loads associated with bilingualism. In othe r words, considering the limited cognitive resources available to all individuals, then it logically follows that bilinguals must have a higher cognitive processing load as they have the additional burden of having to manage two languages within one mind It is uncontroversial in the psycholinguistic literature to claim that all grammars are simultaneously activated whether or not one is contextually relevant for a given task at hand ( e.g., Bialystok
149 2009; Kroll & Dussias 2012 for review). Given this doub le activation a need for I nhibition emerges in the case of bi /multilinguals only, which diverts some of the finite cognitive resources otherwise engaged in linguistic processing. Obviously, monolinguals do not have to divert attentional resources for the purpose of inhibition, which might manifest in the case of bilinguals as compared to monolinguals as processing based differences in task performances of this kind (e.g., Kroll & De Groot 2005 for a review ) A ll the experiments included in th e present s tudy are off line measures and cannot directly comment on this real possibility On line meas ures such as self pace reading reaction times or eye tracking, for example, would be needed to make any final claims in this respect. A more general discussion of this issue is offered at the end of this section. 184.108.40.206 Clitic omission As described in the second prediction ( Section 4.2.2) I did not expect HSs to accept an invariable form of gusta like with omission of the clitic since, under this proposal, it is the pre sence of the once clitic itself reanalyzed as preverbal agreement morphology that licenses the invariable form ( e.g., Montrul 1995 for some suggestions in line with this ). Consider (3) below as an example of the tokens included in this condition. ( 3) A J en gusta la playa To Jen Like. 3 rd SG the beach Jen the sentences included in this condition tested this assertion. Figure 5 14 below presents the average results acr oss all four groups
150 Figure 5 1 4 Group means of prescriptively grammatical clas s III psych verb tokens.HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigrants/Attriters; MA= Monolingual Adults As can be s een, monolingual speakers and 1 st generation immigrants reveal judgments that are consistent with the descriptions found in the theoretical literature. That is, they strongly reject sentences that omit the obligatory dative clitic. Likewise, judgments also reveal an almost categorical rejection of this sort of tokens. This generalized rejection c orresponds to non statistical significant differences across these three groups for this specific condition. Conversely, the intermediate HS group a verages are slightly higher ( 1.88 ). As a consequence of this somewhat unexpected acceptance, we obtain a statistically significant difference between intermediate HSs and all other groups for this condition (HS INT vs. AA, p< .001; HS INT vs. MA, p<.001; H S INT vs. HS ADV, p< .001). As discussed, it was predicted that all informants would reject th e tokens included in th is condition, yet for different reasons depending on the groups. Monolinguals and 1 st generation immigrants were predicted to reject the o mission of the clitic since the ir
151 mental representations include the clitic a s an obligatory el ement with class III psych verbs In the case of the HS groups, it was predicted that they would reject this omission as well either because their grammars requi re obligatory clitics like for monolinguals or because the preverbal dative clitic had been reanalyzed as obligatory verbal morphology. B y looking only at this set of data in isolation, it is impossible to determine whether these HSs may have rejected the sentences in this condition because they also know that class III psych verbs require the dative clitic; or for some other reason But crucially, the point to be made here is that when we compare the intragroup results across the different conditions (espe cially against the two innovations predicted), one can observe that the monolinguals, and to a certain extent also the 1 st generation immigrants, do not have this as an option while the evidence seems to suggest that HSs do. But even if that is the case, it is not certain that this is exactly what the HSs are doing here. That said, what is certain is that by combining, comparing and contrasting all the results included in this task, that my analysis is an emerging possibility. In other allow for the canonical use of class III psych verbs in addition to these other proposed alternatives. Monolinguals on the other hand only find the canonical use acceptabl e and none of the rest of the crucial comparisons. Generally speaking, t hese results suggest that their grammar does not avail itself of the options in the HS grammars. Additionally, that Intermediate HSs tend to rate the sentences included in this condition slightly higher was perhaps an unexpected outcome since, under any circumstances, omission of the dative clitic is predicted to result in ungrammatical sentences. This finding inevitably makes us wonder about the nature of these ratings as
152 well as about the differences between the advanced and intermediate HSs. To fully answer this que stion, it is first necessary to take a closer look at the data, individual by individual. Recall that this group included a total of 24 HSs. Informants were included in this group if their proficiency scores ( Section 220.127.116.11) ranged from 30 to 39 points ( out of 50 possible). Of the 24 informants, 6 of them clearly accepted this condition with significantly contributed to the overall higher averages of the whole group as well as to the apparent general indeterminacy of the intermediate HS group 6 In addition to this, by looking at the individual data, it was evident that HSs in general and intermediate HSs in particular, revealed a higher degree of variation than the monol inguals as well as the 1 st generation immigrants not only within their respective groups, but also within the same individual. This, as I will discuss in more detail at the end of this chapter may be due to the different sociolinguistic realities in which these individuals have been brought up into. So far, I have shown that (i) all informant groups categorically accept the canonical use of class III psych verbs, (ii) that bilingual individuals in general do not notice the absence/presence of low phonologi possibly as a result of the higher cognitive processing load of managing two languages instead of one, and (iii) that all groups, including the HS intermediate (albeit less categorically), reject the ungrammatica l tokens that result from the absence of the dative clitic when using class III psych verbs. Next, I present the results for the condition that tested the use of gustar like in a fully inflected paradigm. 6 Additional evidence in favor of this analysis will come from the results of the elicited production task (section 5.3 ) since intermediate HS production data show similar variation.
153 18.104.22.168 Case agreement innovations If on the right track, a nd class III psych verbs are being reanalyzed as class II psych verbs in Spanish HS grammars, one could also expect HSs to (variably) accept tokens such as the ones illustrated in (4a b) below (4) a. (*) Yo encanto la playa I.NOM love.1s tSG.the beach b. (*) Nosotros gustamos la playa We.NOM love.1stPL the beach In other words, gustar like verbs, like all c lass II psych verbs, could also have a fully inflected paradigm used with an agentive reading. Figure 5 15 below presents the average results Section 22.214.171.124. Figure 5 15 Group mean s for agreement innovation with class III psych verbs. AA= 1 st Generation immigrants/Attriters; MA= Monolingual Adults; HS ADV= Advance Heritage Speakers; HS INT =Intermediate Heritage Speakers
154 As can be seen, monolinguals and 1 st generation immigrants se em to reject the tokens included in this condition while both HSs are more accepting of the abovementioned agreement innovation That monolinguals and 1 st generation immigrants categorically reject this sort of tokens is not surprising because changes were not predicted to obtain in their grammars. In fact, it seems that their underlying mental representations for gustar like verbs do not differ from each other since, with the throughout the two experiments conducted this far have been practically identical and always consistent with the descriptions found in the theoretical literature. What was somewhat unexpected, however, were the relatively low averages reported by the two HSs groups; specially the advanced one. These low ratings may seem to constitute counterevidence against the prediction advanced above, whereby class III psych verbs like class II could surface with a fully inflected paradigm Upon a closer look to the data, however, we can observe that both the intermediate HS group as well as the advance HS group differ in their judgments from all other groups at a level of statistical significance ( (INT HS vs. MA <.001, INT HS vs. AA p=.003 ) (ADV HS vs. MA =.004, ADV HS vs. AA p=.00)). And crucially, these two HS groups do not differ from each other (HS INT vs. HS ADV p=.096). Though, the polarized as one might expect they are consistent with the predictions put forth in Section 4.2.2, providin g therefore further support for the abovementioned hypothesis. Additionally, the within group comparisons reveal that only the HS groups make a statistically significant difference between the previous condition, which tested the *omission of the dative cl itic, and the current one, which tests the abov e mentioned
155 emerging structure (HS INT, p=.03; HS ADV, p =.001). This difference is imperative because it shows that for the HS grammars tested herein, but crucially not for the tence types are different, whereby one can be grammatical and the other one is undoubtedly ungrammatical. Figure 5 16 below shows these differences. Figure 5 1 6 Group means for agreement innovation and clitic omission HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigrants/Attriters; MA= Monolingual Adults Up until now, we have seen a trend whereby the intermediate HS group seems to consistently reveal judgments that are less categorical than those reveal ed by the other groups. That said, their ratings tend to be consistent with the descriptions found in the theoretical literature and/or with the predictions here anticipated. Admittedly, this was not the case in the previous condition in which the omission of the dative clitic was tested ( Section 5. 3 .3.3) Back t hen, results were not in line with what was expected since they revealed a s urprisingly high average (n=1.88 /4) considering that the dative clitic is always obligatory with cla ss III psych verbs. To
156 account for this unanticipated outcome, it was noted that a small subgroup of intermediate HSs was responsible for these higher than normal ratings. Since this was the case for the previous condition, it is reasonable, therefore, to assume that this im balance may also be affecting the average results for the current condition too. Upon further examination of the individual data in this condition, it was clear that there was some variation across these informants, yet not as pronounced as in the previous section. more uniform, as indicated by the smaller standard deviation (0.72 vs. 0.54). Th us, I take this average to be more or less representational of the grammars. Taken together, these results show that the two HS groups are more accepting than monolinguals and 1 st generation immigrants when it comes to class III psych verbs used with a fully inflected paradigm. Since this observation w as supported statistically across conditions and within the groups, it is reasonable to clai m that this emerging structure has a basis rooted in differences in representation at some level between HSs and the other two control groups, a point to which we w ill return in much greater detail. Left to be determine d is whether HSs accept the use of an invariable 3 rd person singular class III psych verb ( e.g., de Prada Prez & Pascual y Cabo 2011) in the presence of a nominalized < e xperiencer > 126.96.36.199 N construction: Recall that p rediction II ( Section 4.2.2) anticipate d that in class III psych verb constructions the sentence initial argument may get reanalyzed as nominative and may appear with an invariable form of gusta (as observed by de Prada Prez & Pascual y
157 Cabo 2011). As described, I believe this is possible because HSs are able to interpret dative clitic s as preverbal agreement morphology (Montrul 1995) as in (5). (5 ) ( ) Yo me gusta la pizza/las pizzas I. NOM DAT.CLI like. 3 RD SG the pizza/the pizzas Figure 5 1 7 below illustrates the average results acro ss all four adult groups for this condition ( Section 188.8.131.52 ) Figure 5 17. Group means of prescriptively grammatical class III p sych verb tokens.HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigrants/Attriters; MA= Monolingual Adults At first sight, it appears that only the monolinguals (MA=1.54), are sensitive to the ungrammaticali ty that results from the invariable use of a 3 rd person verbal form (AA=2.19; HS ADV=2. 3 8; HS HS INT=2.58). This is especially true for the intermediate HS group whose judgments are visibly higher than the rest (HS INT=2.58). In fact, a cross the group comp arisons reveal that monolinguals stand alone as the only group with target like ratings, as they differ in their judgments with respect to the other three
158 groups (MA vs. HS INT, p < .001; MA vs. HS ADV, p =.003; MA vs. AA, p = .003). All other comparisons resulted in non statistically significant differences. But t o better observe the status of the innovation suggested in this condition, consider figure 5 18 below in which the data are further divided by group (MA, AA, HS ADV, HS INT) and by (singu lar, plural). Figure 5 18. Group means of grammatical innovation HS ADV= Advance Heritage Speakers; HS INT=Intermediate Heritage Speakers; AA= 1 st Generation immigrants/Attriters; MA= Monolingual Adults Rec all that this condition includes 12 tokens, 6 of which appear with a 3 rd person singular and 6 with a 3 rd person plural . In reexamining the new variables, a repeated measures ANOVA with TYPE (Singular theme, Plural theme) and GROUP (Monolingual, 1 st Generation immigrant, Advance HSs, Intermediate HSs) as variables was run The results from this ANOV A showed a main effect for TYPE (F(1,70)=10.712; p=.002) and GROUP (F(3,70)=7.661; p<.001), but not for a TYPE by GROUP interaction (F(3,70)=1.075; p=.365). In regards to the main effect fo r GROUP Bonferroni post hoc tests attributed this significant effect to the comparison between the
159 monolinguals and intermediate HS (p<.001) monolinguals and advanced HSs (p =.003), as well as monolinguals with the 1 st generation immigrants (p = .003) A s mentioned, a ll other comparisons resulted in non statistically significant differences T hese results show two clear tendencies On the one hand we have the monolinguals whose judgments are consistent with the descriptions found in the theoretical liter ature. Note, however, that their target like judgments for this specific condition are not as categorical as they were for previous ones (e.g., greement innovation ( Section 184.108.40.206 ( Section 220.127.116.11 ) ). On the other hand, we see that the other three groups have a higher tolerance to these a priori ungrammatical sentences. Particularly interesting is the case of the 1 st generation monolinguals. T his outcome however, is not completely unexpected as similar results were reported by de Prada Prez and Pascual y Cabo (2011) In this study, the native control group highly rejected ungrammatical sentences due to verb agreement when the target response was gusta (3 rd person singular) Conversely, this was not the case when the target response was gustan (3 rd person plural). In such cases, they favored the a priori ungrammatical gusta ( 3 rd person singular) over the grammatical gustan (3 rd person plural) ( de Prada Prez & Pascual y Cabo 2011:115). At the time, the argument was that the variability found in the native control data (which crucially was gathered from 1 st generation immigrants and not monolinguals) was the locus of interlanguage influence 7 (201 1:118). Thus I maintain here that HS differences, at least for the individuals and properties examined herein, can be traced back to contact induced 7 Similar results ha ve also been reported in De Prada Prez Rodrguez Ricelli, Woodfine, and Rogers ( 201 2).
160 changes in first generation immigrants. Evidence to support this includes the fact that no significant dif ferences were found when comparing the ratings of the 1 st generation immigrants with those of either the advanced (p=1.0) or the intermediate HSs (p=.254). Conversely, the p value reaches a level of statistical significance when comparing 1 st generation im migrants with monolinguals (p= .003). 5.3.4 General Discussion of GJT2 The main goal of this experiment was to gain insight into the current status of class III psych verbs in HS Spanish while testing the predictions put forth in Section 4.2.2. These prediction s were guided primarily by previous findings in this domain ( e.g., de Prada Prez & Pascual y Cabo 2011) as well as by their syntactic structure ( e.g., Belletti & Rizzi 1988 ; Landau 2010; Montrul 1995 ). As described, i nformants were examined on five condi tions, each one testing a piece of the puzzle that, when brought together, would support or cast some questions on the abovementioned hypothesis. The first condition tested for acceptance or rejection of the canonical use of gustar like verbs. This was nec essary given that all informants were predicted to have access to this structure. The results obtained were in accordance with this prediction as all groups categorically accepted it Next, informants were tested on an ungrammatical set of sentences that i ncluded a class III psych verb but with omission of the obligatory 8 Contrary to expected trends, all groups with the exception of the monolinguals were slightly inclined to accept this condition. Though admittedly unexpected this outcom e was explained by primarily appealing to the higher cognitive processing loads for bilinguals The third condition tested another 8 Recall that the dative marker is only obligatory when the is spelled out.
161 set of ungrammatical sentences, this time due to omission of the obligatory clitic 9 The results obtained were also consisten t with the predictions advanced since a ll informants largely reject ed th ese tokens. The remaining conditions examined the two proposed innovations that logically result from the hypotheses advanced; namely a case agree ment innovation and a verbal agreement innovation. In regards to the first one, the data obtained are also consistent for the abovementioned hypothesis since the HS groups were more inclined to favor these case agreement innovations in comparison to other ungrammatical conditions (e.g., omissi on of the obligatory dative clitic). Monolinguals and 1 st generation immigrants, on the other hand, categorically rejected both conditions equally. Even more polarized ratings were observed in the case of the second innovation. While monolinguals strongly rejected it, all bilingual groups (AA included) favored the use of the invariable gusta more. Most importantly, intragroup pairwise comparisons revealed that neither the monolingual group nor the two HSs groups differed in their judgments with respect to these two innovative constructions. In other words, while for the monolinguals these two innovations are equally and categorically ungrammatical, the two HS groups variably accept them, lending therefore support for the abovementioned hypothesis. Though no t categorically, the overall findings from GJT 2 are, once again, consistent with the predictions I proposed in Section 4.2.2. Back then, I anticipated that one might observe cases of invariable gusta with apparent surface retention of the dative clitic ( a s observed by de Prada Prez & Pascual y Cabo (2011) ) This was thought to be possible by assuming that gustar 9 case that the is spelled out, the clitic is always obligatory.
162 that the apparent clitic is actually serving as preverbal agreement morphology (Montrul 1 995). T he results here presented have shown that while the monolingual control group is highly sensitive to the syntactic restrictions imposed on class III psych groups, irrespective of their proficiency, are not. In fact, the trends clearl y show how both groups of HSs tend to favor the two predicted innovations. So far, I have considered the data obtained from GJT 2 with respect to the 1 st research question. Next, I proceed to discuss the nature of the differences observed between the group s. As discussed, in the absence of longitudinal data, it is impossible to discern whether these MAs/ HSs differences are attributable to either attrition, incomplete acquisition, or input delimited differences. In this particular case, this complexity is f urther increased since none of the children groups that participated in GJT 1 completed this task T his precludes us from making the essential comparisons needed to allow us into observ ing ( albeit indirectly) what HS and monolingual development look like f or this particular domain. In turn, these comparisons are the only reliable indicator of when these changes begin. We do have, however, the data from 1 st generation immigrants. Though often overlooked in many previous HS studies (e.g., Cuza et al. 2013, Mo ntrul 2004, 2006 ; Polinsky 2011 ), this is another important piece of the puzzle because it allows us to observe whether these HS differences can be traced back to changes in 1 st generation immigrants If indeed we can observe these changes in the input pro viders, then we could safely assume that incomplete acquisition 10 alone cannot be the cause of the differences (Pascual y Cabo & Rothman 2012 for detailed discussion). This is what 10 For now, I maintain the possibility of so called incomplete grammar under a definition where it refers to arrested development of a given property. However, I will return to pr oblematize what incomplete acquisition actually is to the extent that the term makes little sense especially under a generative perspective of linguistic representation in Chapter 6.
163 we see in this task. But even if on the hypothetical scenario in which HSs a nd 1 st generation immigrants (along with monolinguals) differed significantly in their judgments, this would not constitute sufficient evidence nor an adequate enough argument to warrant incomplete acquisition again referring to arrested developmental out comes, as the source of HS differences. In such cases, additional evidence from production tasks would be needed to validate that grammatical judgments do in fact mirror performance. This is the point to which I turn next. 5.4 Elicited P roduction T ask Recall f rom Section 18.104.22.168 that in addition to the two grammaticality judgment tasks ( Section 5.2 and 5. 3 ) all adult informants were asked to complete an elicited production task The goal of this task was to prompt informants into using class III psych verbs i n different contexts and forms These data are important for two main reasons. On the one hand, production data can further function as supporting evidence, or not, for the results observed in the previous tasks. Additionally, because different generationa l groups were tested with the exact same methodology, any differences and/or similarities in their production data could also allow us to better understand the source of the HS differences observed Put simply, if HS differences at least partially, can b e traced back to changes already observed in the production of 1 st generation immigrants, then we can conclude that arrested development in the form of incomplete acquisition as used by Montrul and others cannot be the only possible explanation of differ ences Alternatively, if HSs, as adults, seem to not converge on an adult monolingual grammatical competence for property X, but rather show ultimate attainment for property X that equates to a pre adult like stage true of m onolingual childhood development and, additionally, cannot be attributed to possible innovations
164 from the input they receive in language contact environments either, then one could reasonably argue for a domain of true arrested development ( what Montrul 2008 refers to with the blanket term incomplete acquisition). However, showing this would not mean that true incomplete acquisition in the sense of Montrul is the only explanation. For example, one might also take (and ideally be able to show) the position advocated by Putman and Sn chez (In press) in which the course of development simply changes as a reflex of the unbalanced societal bilingual reality in which HSs find themselves Given the results obtained in the previous two experiments (i.e., HS variable acceptance of class III psych verbs in passive constructions as well as variable acceptance of so called invariable gusta in different environments), differences were expected to obtain between the MA group representational differences are theo rized to obtain (at least partially) as a result of their interpretation of (e.g., Lightfoot 1999, 2010; Lightfoot & Westergaard 2007 ) changes with respect to the monolinguals were also p redicted to obtain in the speech of 1 st generation immigrants. To support this claim, we should find some AA differences as well although such differences might be at levels between those of the MAs and the HSs. The experiment consisted of a total of 18 s ets of slides projected in a power point presentation, 6 of which served as distracters and the remaining 12 as experimental items. As discussed in C hapter 4, each critical token was designed to contain and in all four possible comb inations, that is: (i) 3 rd person singular with 3 rd person singular ; (ii) 3 rd person singular with 3 rd person plural ; (iii) 3 rd person plural with 3 rd person
165 singular ; and (iv) 3 rd person pl ural with 3 rd person plural . All 78 adult informants completed this experiment. From these recordings, a total of 2,602 utterances were produced. These were distributed in the following manner: Monolinguals, n=541; 1 st Generation immi grants, n=494; Advanced HSs, n=669; Intermediate HSs, n=898. These utterances were transcribed and coded according to the following six variables: Group (Monolinguals, 1 st Generation immigrants, Advanced HSs, and Intermediate HSs), Verb type (Class III psy ch verb, Other), Verb agreement (target like, non target like), Clitic agreement (target like, non target like), type (singular, plural), type (singular, plural). The results obtained from this experiment are presented in the next su bsections in a way intended to highlight the similarities/differences between the groups and the three major conditions examined, namely (i) verb agreement, (ii) clitic agreement, and (iii) 5.4.1 Verb A greement As was als o the case for the previous two experiments, the present data were submitted to statistical analysis using SPSS. That said, because of the different nature of the data collected (i.e., categorical binomial data), it was deemed more appropriate to use a Mix ed Model Binomial Logistic Regression instead of a repeated measures ANOVA The main advantage of using the linear Mixed Model Binomial Logistic Regression is that it can process uneven, categorical (binomial) data while considering within subject dependen ce. Turning now to the results figure 5 19 below includes the mean percentages of accuracy for canonical subject verb agreement divided by group (monolinguals, 1 st
166 generation immigrants, advanced HSs, intermediate HSs) and verb type (class III psych verbs other) Figure 5 19. Group means of percentage accuracy for overall verb agreement.Class III= Class III Psych verbs; Other = Non Class III Psych verbs ; MA= Monolingual Adults; AA= 1 st Generation immigrants/Attriters; HS ADV= Advance Heritage Speakers; H S INT=Intermediate Heritag e Speakers As can be seen, when it comes to producing subject verb agreement, overall accuracy means are noticeably elevated for all groups. This is especially visible for verbs other than class III psych predicates since all the groups perform at or close to ceiling levels (+90%). As expected, while these elevated means are somewhat maintained with respect to class III psych verbs among the monolinguals and 1 st generation immigrants, a decline in accuracy among the two HS groups can be observed. To statistically discriminate what otherwise seem homogeneous results, a binomial logistic regression was run with GROUP (MA, AA, HS ADV, HS INT), and VERB TYPE (class III psych verb, non class III psych verb) as variables. The results fro m this test
167 revealed that there is a main effect for both GROUP (F(3,83)=9.029, p<.001) and VERB TYPE (F(1,639)= 11.747, p=.001) but not for the GROUP by VERB TYPE interaction (F(3,765)=1.385, p=.246). To further examine these effects, within and across gr oup comparisons were made for each condition. In regards to verb type, all groups made a statistically significant distinction in the production of subject verb agreement between class III psych verbs and non class III psych verbs (MA, F(1,616)=4.158, p=. 042; AA, F(1,548)=5.023, p.025; HS ADV, F(1,416)=20.948, p<.001; HS INT, F(1, 2,035)=7.705, p=.006)). Post hoc Pairwise comparisons with Bonferroni adjustment reveal that the intermediate HS group seems to be different with respect to their production of s ubject verb agreement with non class III psych verbs (MA vs. HS INT, p<.001; AA vs. HS INT, p <.001; HS ADV vs. HS INT, p=.035). That said, the significance of these findings is unclear since all informant groups are at or above the 90% accuracy. In exami ning class III psych verbs, we observe a noteworthy decrease in accuracy in the production of canonical subject verb agreement across all four groups. Visibly, the differences in the accuracy means with respect to non class III psych verbs are greater in t he case of the HSs than in the case of MAs and AAs. This was the anticipated outcome given previous findings in related literature ( e.g., Dvorak & Kirschner 1982; Montrul & Bowles 2009; de Prada Prez & Pascual y Cabo 2011; Toribio & Nye 2006). It should b e noted that the results from post hoc Pairwise contrasts (with Bonferroni adjustment) show that the comparisons between monolingual speakers and first generation immigrants do not yield statistically significant results (MA vs. AA,
168 p=.608 ). By the same to ken, advanced and intermediate HSs do not differ from each other at a level of statistical significance either ( HS ADV vs. HS INT, p=.192). That said, in a cross examination of the control groups with the experimental groups we find that both HS groups di ffer significantly with respect to both the monolingual group (HS INT vs. MA, p< .001; HS ADV vs. MA, p< .001) and the 1 st generation immigrant group (HS INT vs. MA, p< .001; HS ADV vs. MA, p= .001). In general, HSs present more difficulties with respect to subject verb agreement with class III psych verbs than with non class III psych verbs. These differences reach a level of statistical significance with respect to the two control groups, which is consistent with the predictions stated in 4.2.2 as well a s with other previous findings ( e.g., Dvorak & Kirschner 1982; de Prada Prez & Pascual y Cabo 2011; Toribio & Nye 2006). But t o better understand the current status of class III psych verbs in HS Spanish, it is important to cross examine the different acc uracy means by group and agreement configuration. With this in mind, a summary of means of use in percentages across the four groups tested is given in Table 5 5 below. Table 5 5. : Accuracy means of class III psych verbs. SgExp/SgTheme SgExp/PlTheme Pl Exp/SgTheme PlExp/PlTheme /N % /N % /N % /N % Monolinguals 67/67 100 72/76 94.73 121/121 100 41/43 95.34 Immigrants 73/73 100 84/95 88.42 125/126 99.20 52/54 96.29 Adv. HSs 139/139 100 80/101 79.20 148/178 83.14 79/92 85.86 Int. HSs 183/188 97.34 105/145 72.41 161/214 75.23 110/127 86.61 SgExp/SgTheme=3 rd person singular with 3 rd person singular ; SgExp/PlTheme= 3 rd person singular with 3 rd person plural ; PlExp/SgTheme= 3 rd person plural w ith 3 rd person singular ; PlExp/PlTheme=3 rd person plural with 3 rd person plural Provided in each column are (i) the number of target like/produced tokens as well as (ii) the percentages of accuracy for each one of the four possible agreement configurations: singular with singular , singular
169 with plural , plural with singular , plural with plural . In examining the first and last columns fro m table 5 5 above (SgExp/SgTheme and PlExp/PlTheme respectively ), it becomes evident that none of the groups tested have difficulties with producing canonical subject verb agreement with class III psych verbs when the agreement configuration is a singular with a singular (with the target construction being gusta rd plural (with the target construction being gustan rd That said, the percentages are substantially lower for all groups 11 when there is a number mismatch between the and the , and especially when the latter is plural and the target construction is gustan rd 12 This, however, is not surprising given that the default verbal agreem ent form in Spanish seems to be that of 3 r d person singular ( e.g., Gr instead 2000, 2001 ; Villa Garca 2010 inter alia ). verb agreement patterns that differ from those of the monolinguals (and to a le sser extent from the 1 st generation immigrants). The evidence seem to suggest that t hese results are consistent with previous findings in regards to the preference towards the use of invariable gusta irrespective of actual verb agreement, as observed by de Prada Prez and Pascual y Cabo (2011). Far from being random, I would like to argue that these results are consistent with the predictions put forth in 4.2.2 and indicate that HSs may have an emerging syntactic optionality when it comes to class III psych verbs. In other words, while the canonical syntactic structure for class III psych verbs is still accessible (and 11 ther groups. 12 Note, however, that this is not the case for the monolingual speakers.
170 perhaps even preferred over the other options), HSs seem to have syntactic options that are not Left to be determined is whether similar tendencies are observed with respect to the other properties associated with class III psych verbs, namely the dative marker a and clitic agreement. 5.4.2 Dative M arker a Recall that, in the presence of a class III psych ver b in combination with a spelled utterance in principle becomes highly ungrammatical due to a case violation. Thus, to determine with accuracy the target only those utterances in which the was overtly expressed are included. Figure 5 20 below shows the accuracy ratings for each for each group. Figure 5 20 MA= Monolingual Adults; AA= 1 s t Generation immigrants/Attriters; HS ADV= Advance Heritage Speakers; HS INT= Intermediate Heritage Speakers As can be seen, the overall accuracy with each group from 86.69% (in the case of the monolingual speakers) to slightly
171 above chance level (in the case of the intermediate HSs). As was also the case with the previous conditions, the differences in the proportion of accuracy were submitted to statistical analysis using a binomial logistic regression with group ( MA, AA, HS ADV, HS INT) as a variable. The output of this test indicated that these observed differences ensued a statistically significant p value (F (3,44)= 5.947, p =.002). To further examine this effect and determine which of the group comparison(s) yi elded significant differences, post hoc Pairwise comparisons (with Bonferroni adjustment) were run between the groups. These tests revealed that p values reach statistical significance when comparing monolinguals with 1 st generation immigrants (p=.034), Ad vanced HSs (p< .001), and Intermediate HSs (p < .001). Similarly, Intermediate HSs were different from Advanced HSs (p =.0013) and 1 st generation immigrants (p <.001). The only group comparison that did not result in statistically significant p value was between Advanced HSs and 1 st generation immigrants (p=.876). In addition to these differences, another revealing piece of information is the lower than expected percentages revealed b y all four groups. T unexpected for the two HS groups, however, and to a certain degree not for the AA group either, since previous comparable studies have found this dative marker to be differentially realized (Montrul & Bowles 2009). It is somewhat surprising, however, in the case of the monolingu als. To explain these generalized low means it is necessary to take a second look at the methodology employed. Recall from Section 22.214.171.124 that all informants were prompted to produce sentences as they were being presented with a power point presentation Each slide provided informants with the necessary information to produce an utterance containing a class III psych verb (in the case of the critical
172 tokens). Crucially, the always appeared on the left of the screen priming the informant therefore to start the utterance without the obligatory dative marker. Often times, the informant realized midsentence that the output was the case of the HS groups, t in their production If these lower than expected accuracy means can be partially explained in terms of methodological all informants in the sam e way 13 ? Recall that similar results were reported in Section 126.96.36.199.2 whereby bilingual informants were not as sensitive to the ungrammaticality that resulted from the omission of dative marker a In fact, Intermediate HSs rated it above chance level (3 .12 out of 4). This, again, can be explained in terms of processing limitations in real time. Whereby bilinguals have to deal with the higher cognitive costs associated with managing two languages instead of one. All in all, these results are consistent w ith the predictions put forth in Section 4.2.2 whereby there is a simplification of the Spanish system of psych predicates from three taxonomy In fact, it seems logical that the absence of the dative triggers the abovementioned change. Because all groups, not only HSs, reveal lower than expected accuracy means in this domain, it seems unreasonable to claim that the 13 necessarily ungrammatical if there is a pause between the stated subject since the dative doublet marked with a pause and not choose to use the optional doublet in completely the utterance. Ideally, I would hav e coded for pauses to be able to see the extent to which this is true for all the groups, but this was not done and time does not permit going back to do this at this juncture. Suffice it to say that even in the case this were to prove explanatory, we wou ld still be left with the marked differences in performance across the groups that is worthy of explanation.
173 s must obtain as a result of incomplete acquisition. Instead, these results seem to suggest that what on the surface may seem like HSs differences can be attributed to input differences as suggested by Rothman (2007) and Pires and Rothman (2009) among oth ers. 5.4.3 Clitic A greement verbs is strictly dependent on whether the is overtly realized or not, the clitic must always be present and must always be in agreement wi th the full DP ; regardless of whether it is overtly realized or not. In its absence, the outcome is always ungrammatical. Recall from Section 188.8.131.52, that all of the agreement configurations used in this experiment elicited the use of a 3 r d person singular/plural in which cases the target clitic agreement should be As can be noted, the only difference between these two forms lies in the presence or absence of the word final / s/ which mark s singular and p lural number respectively. Because / s/ elision is a distinctive phonological characteristic of Cuban Spanish ( e.g., Lpez Morales 2003), and because all informants examined in this dissertation are of Cuban descent, it is therefore not possible within the methodological confines of this study to determine with precision the exact grammatica l nature of their production. In other words, the data with respect to clitic agreement are confounded because it is impossible to distinguish between cases of clitic a greement violations from cases of normal / s/ elision. As a result of this
174 class III psych verbs will not be discussed 14 With the above in mind, however, it is importan t to note that this dialectal phonological trait cannot explain the cases of clitic omission. In consonance with the descriptions found in the theoretical literature, neither of the two control groups omitted the clitic. Intermediate and advanced HSs, on t he other hand, produced class III psych verbs with omission of the clitic 28 times (out of 674 utterances) and 7 times (out of 510 utterances) respectively 15 Though highly ungrammatical if evaluated against the monolingual standard, this would be a grammat ical option in HS Spanish if indeed class III psych verbs have been reanalyzed as class II psych verbs. All in all, these results are also consistent with the predictions articulated in 4.2.2, since in HS Spanish gustar like verbs, just like asustar molestar option. 5.4.4 General D iscussion of E licited P roduction T ask like behavior with agreement with non class III psych verbs (HS Adv= 96.85 %; HS Int= 90%) contras t sharply with low accuracies in (HS Adv= 70.78 %; HS Int= 51.63%). When e xamining verbal agreement with class III psych verbs, it was observed that HSs behave mostly target like when both arguments are singul ar (HS Adv= 100 %; HS Int= 97.34%), but non target like when the < experiencer > < theme > combination is singular plural (HS Adv= 84.15 %; HS Int= 72.41%), plural singular (HS Adv= 80.33 %; HS Int= 14 Anecdotally, it should be noted that indeed most of what on the surface seem like clitic agreement violations were observed with plural . 15 See A ppendixes F (HS ADV) and G (HS INT) for the transcription of sentences with omission of clitic.
175 75.23%), or plural plural (HS Adv= 85.86 %; HS Int= 86.61%) As discussed, these were expected outcomes given the hypotheses and predictions detailed above. On the contrary, monolingual speakers revealed a more robust knowledge of all the properties associated with class III psych 86 .69% ), yet not categorically The 1 st This is especially true for the relevant conditions racy in production of class III psych verbs. In terms of verb agreement with verbs other than class III psych verbs the mean average was cate gorically on target (99.31%). With class III verbs, the average was slightly lower (95.98%) yet still mostly on ta rget. Additionally, t hey also revealed a robust knowledge of all the properties associated with class III psych verbs, yet not as strongly 76.83%). al group in ways that are consistent with the predictions stated in 4.2.2. That is, both groups of HSs show evidence of distinct subject verb agreement patterns that can be explained with the proposed shift of class III into class II psych verbs. Additiona lly, both groups of would be otherwise obligatory. Interestingly, the data from the 1 st generation immigrants also reveal some level of variation, even if much less than the HS groups. 5.5 Summary of E mpirical D ata In this chapter I presented the results obtained from three experimental tasks. The first task was designed to test the viability of the hypothesis proposed in Section 4.2.2 whereby in Spanish HS grammars, cla ss III psych verbs are being reanalyzed as
176 class II psych verbs. The results obtained are consistent with this prediction as HSs, but crucially not monolinguals, (variably) accept gustar type verbs in passive constructions; a syntactic reflex of the abovem entioned simplification. The second task mainly tested for acceptance of so called invariable gusta (as observed in de Prada Prez and Pascual y Cabo 2011) in different environments. This was hypothesized to be possible by assuming that class III psych ver bs were assigned an agentive structure and that the apparent dative clitic was taken to be preverbal agreement morphology. Consistent with the predictions articulated in 4.2.2, the results showed differences between the two HS groups and the monolingual gr oup in regards to the anticipated grammatical innovations. The third and final task was designed to further examine the production of gustar like verbs in different contexts and forms. The results revealed important differences between the experimental and control groups in regards to the patterns of use of class III psych verbs as well as the properties associated with them. In C hapter 6 I bring together the data obtained from all three tasks and I further discuss them in relation to the research question s I articulated in Section 4.2.1. Additionally, I will address the limitations and suggest directions for future research.
177 CHAPTER 6 DISCUSSION AND CONCLUSION 6.1 Introduction The purp ose of this chapter is twofold. First, it aims to bring together all the data reviewed in C hapter 5 in an effort to provide a macro analysis with respect to the status of class III psych verbs in HS Spanish The second goal is to dissect this macro analysis so as to address the research questions that guided this dissertation outli ned in Section 4.2.1 and repeated below for convenience. 1) Are class III psych verbs in HS Spanish undergoing a reanalysis of their argument structure? If so, in what ways and why? 2) To what extent are attrition, incomplete acquisition and/or input delimited d ifferences at least partially explanatory as sources of HS differences and to what extent can they be teased apart? With this in mind, the remaining sections of this chapter are organized as follows: Section 6.2 brings the data from the 3 experiments toget her to provide an overarching analysis. In Section 6.3, I revisit the research questions of the dissertation and provide answers to them based on the results obtained here In Section 6.4, I review some possible limitations of the present study as well as explore how such limitations can be addressed in future research. 6.2 Macro a nalysis In an attempt to provide an adequate explanatory analysis of the nature of the HS differences observed in this domain ( e.g., de Prada Prez & Pascual y Cabo 2011; Toribio & Nye 2006), I hypothesized that in HS Spanish, class III type psych verbs ( e.g., gustar have b een reanalyzed as class II ( e.g., Section 3.2 that, unlike class III psych verbs which are strictly unaccusative ve rbs (e.g.,
178 Belletti & Rizzi 1988) class II psych verbs can alternate from being interpreted with an agentive or a stative reading. As mentioned, t his semantic hybridity has syntactic consequences : if asustar class II psych pr edicate for that matter) is interpreted with a stative reading, then it is treated as an unaccusative verb and cannot be used agentively. In such cases, class II psych verbs have the same underlying syntactic structure as canonical class III psych verbs. O n the other hand with an agentive reading, class II psych verbs (i) can appear with a fully inflected verbal paradigm 1 thus projecting an external theta role, which, in turn, (ii) allows them to be passivized Consequently, from this it follows that, if g ustar and other class III psych verbs have in fact been reanalyzed in HS Spanish as class II psych verb s then they would also have the same options Therefore, a s a syntactic reflex of this emerging syntactic optionality, it was hypothesized that HSs woul d accept, or at least have a reduced threshold for rejection of these two innovative options. As discussed, experiment I a GJT, was designed specifically to test th e first of these prediction s The responses from the two adult control groups to this exper iment revealed trends that were generally consistent with the descriptions found in the theoretical literature That is, they were able to make precise distinctions between all relevant grammatical and ungrammatical conditions in similar ways. Recall that both the monolinguals and the 1 st generation immigrants categorically rejected passives with unaccusatives and class III psych verbs, but accepted passives with transitive verbs 1 To be clear, class III psych
179 including class II psych verbs. This tendency was therefore taken to indicate that their mental representations for class III psych verbs were identical and on target 2 Like the control groups, the adult HSs were also able to make a categorical distinction between grammatical 3 and ungrammatical 4 passive constructions (as well as wi th the distracter tokens). That said, they also showed a tendency to accept (though not categorically) passives with gustar like verbs, a use deemed decidedly ungrammatical for the two control groups. As discussed in Section 5.2 ates are not categorical for the relevant condition should not be taken as dubious evidence for the proposed reanalysis. Instead, it should be noted that their grammars allow for passivization of a specific unaccusative, even if marginally, which should be impossible unless they can optionally project a syntactic structure for this verb that would permit passivization. Crucially, if this were not one of the structures allowed in the HS grammars, then it would be rejected categorically, as was the case for t hem with respect to ungrammatical sentences due to subject verb agreement violations or, more directly related to this discussion, of the ungrammatical passives with other unaccusatives. With this in mind, these results are consistent with the hypothesis a rticulated in 4.2.2 whereby class III psych verbs had been reanalyzed as class II psych verbs. Further evidence in support of the abovementioned hypothesis comes from the results relevant to the second prediction; namely the use of class III psych verbs w ith a fully inflected verbal paradigm. In an examination of the production data, it was noted 2 The MA and AA responses to all relevant conditions across experiments also support this statem ent. 3 Passives with transitive verbs and class II psych verbs. 4 Passives with unaccusative verbs.
180 that the abovementioned innovation was in fact produced, yet only a handful of times and only by the intermediate HSs. That said with the data available it is n ot possible to conclude with certainty whether these utterances are instances of case agreement innovations or not because, c rucially, the production task only elicited the use of gustar like verbs with 3 rd person singular/plural subjects, in which cases t he target subject verb agreement forms would be gusta/gustan respectively and these forms would be the expected targets in both structures. Thus, it is not clear whether the forms elicited are a true reflection of case agreement innovation or not. That sai d, recall that all adult informants were also tested explicitly on this innovation in the experiment 2 ( GJT2 ) Examining the results obtained from this condition, it became evident that the two HS groups were more inclined than MAs and AAs to accept these innovations T heir respective ratings, however, were relatively low and did not reach the cut off value to determine their acceptance ( Section 184.108.40.206 ). But before claiming that this is evidence against the suggested hypothesis, recall that the proposal do es not imply that for the HSs, the agentive reading and/or the inn ovative structure for that matter is now additional alternative, an optionality that is precluded in monoling ual Spanish. As a result of this, it was predicted that one could find variability in the data, whereby HSs sometimes would favor gustar/encantar used as true stative class III psych verbs with the prescribed syntactic structure and some other times with t he proposed reflex innovation. Indeed, the above hypothesis does not make precise predictions in regards to which option will be favored more in actual use. That said, it was argued that frequency
181 differences are possibly deterministic for such HSs prefer ences. That is, the more frequent a structure is in the input/output, the more likely these HSs are to use it. Crucially, then, evidence in support of the hypothesis is found in HSs accepting and/or producing the reflexes that emerge in any proportion. In other words, the variability observed was, by virtue of the hypothesis, expected to be constrained in certain ways and this is exactly what was found. In fact, the data obtained from all three experiments seem to suggest the hypothesized scenario that al though HS grammars can opt for an agentive reading of gustar, as demonstrated from the results of GJT 1, that this would not necessarily be the preferred reading. First recall from Section 220.127.116.11 that, as part of the GJT2, all adult groups were tested sp ecifically on the acceptance/rejection of the class III canonical structure. The results obtained clearly showed that all adult groups accepted this condition at ceiling rates. This acceptance was counterbalanced with a strong rejection of the ungrammatica l distra cter items as well as rejection of the condition that tested clitic omission Also though production data revealed somewhat elevated rates with respect to the accurate production of the properties associated with the canonical use of class III psy ch verbs (i.e., subject verb agreement, clitic were also observed 5 With these results combined, the evidence reported thus far unambiguously suggests that while all adult groups (monolinguals, 1 st generation immigrants and heritage speakers alike) have access to the canonical class III psyc h verb structure, HSs reveal some competence differences with respect to the monolingual grammars 5 See Appendixes F and G.
182 As discussed, these differenc es are by and large consistent with the reanalysis hypothesized in C hapter 4. That said, such reanalysis cannot fully explain all documented cases of innovation in this domain, i.e., cases of invariable gusta with apparent surface retention of the dative c litic as observed by de Prada Prez and Pascual y Cabo (2011). Therefore, to explain the attested cases of invariable gusta it was predicted that HSs were assigning the aforementioned agentive structure to the verb and that the clitic was actually servin g as preverbal agreement morphology. This prediction was also adapted to empirical verification whereby HSs would not be expected to accept sentences with omission of the clitic since for them, it is now obligatory verbal morphology. The relevant results o btained from the grammaticality judgment task we re also consistent with this prediction as both HS groups largely reject sentences with omission of obligatory clitic; yet advanced HSs more. The HS production data are also consistent with this analysis sinc e, in spite of some anticipated clitic agreement violations, the clitic itself was only omitted 28 times by the intermediate HS group and 7 times by the advanced HS group. Crucially, whenever they omitted the clitic, they produced what on the surface seems like case agreement innovations (resembling more its English equivalent) 6 That said, these results can be interpreted in (at least) two possible ways: First, it should be noted that, while it is possible that this rejection is a reflection of the new ro le of the clitic (i.e., preverbal morphology), it is representations require the presence of the clitic with class III psych verbs just as is the case for AAs and M A s 6 See Appendixes F and G.
183 Second, it is also possible that what I am cla iming to be reanalyzed as verbal morphology when an agentive syntax is project ed can be, alternatively, an allomo r ph of a for example, would not only h ave the status of accusative and dative clitics as i n monolingual Spanish, but a lso nominative subject as well. In such a case, it would also have the same phi features as 7 This account, however, me gusto me gusta would be an agreement feature mismatch that the syntax should a priori reject. At present, the data at hand only have 3 rd person forms in which this cannot be teased out with unique predictions of morphological spel lout as in the cases of other grammatical persons as I have just shown. This is ultimately an empirical question that further research could pursue. In general, the results of this study have shown that the adult Spanish HSs tested can assign an additiona l agentive reading to class III psych verbs. As demonstrated by the data presented, this option is not available in the grammars of monolingual speakers of the same dialect. As discussed, the results obtained from the elicited production task provide furt her evidence of the potential for HS innovation in this domain. As anticipated, the monolingual control group showed little evidence of divergence from the expected 7 Me gusta el chocolate me gusta nal morphology as I suggested or if it were a clitic as in monolingual this should in principle be possible. Me siempre gusta X Yo siempre hablo con Hugo eak with Hugo).
184 target 8 case of the two grammaticality judgment tasks discussed above in which monolinguals and 1 st generation immigrants exhibited an almost homogeneous behavior, the results from the elicited production task revealed differences between these two groups. Interes tingly, these differences increased in regards to verb agreement with class III psych verbs and especially with regards to properties that were predicted to be problematic for HSs. A couple of questions emerged from this observation. First, what is the nature of these differences? And second, can these differences be related to the innovative linguistic outcomes observed in subsequent generations? In regards to the first question, i t was hypothesized that 1 st ge neration immigrants performed differently as a result of interference from the L2 and/or other linguistic processing deficits that are a byproduct of being bilingual. Testing between these options or knowing with certainty the extent to which they both con tribute, however, cannot be confirmed with the limited data available. In regards to the second question, these differences seem to support a view in which (qualitative and quantitative) input differences are deterministic in the process of HS acquisition ( e.g., Pires & Rothman 2009; Rothman 2007). As discussed, L1 attrition logically increases the variability present in the input HSs receive. Such variability is predicted to affect the perceived (in)stability of certain (vulnerable) domains (e.g., class I II psych predicates) which in turn are (re)analyzed in some way by subsequent 8 In terms of clitic agreement (Section 5.4.3), all informant groups, monolinguals included, revealed a relatively low average for cliti c agreement. This, however, was not completely unexpected as / s/ elision in word final position has been extensively documented in Cuban Spanish (e.g., Lpez Morales 2003).
185 generations of HSs (e.g., Lightfoot 2010 inter alia). From the trends observed, it seems that monolinguals and 1 st generation immigrants reveal very similar judgments in regards to both grammatical and ungrammatical conditions examined herein. That said, they do show some significant differences in the production tasks; more specifically with well be one of the key factors that trigger the reanalysis in the process of acquisition for subsequent generations of HSs since in its absence, the preverbal could be interpreted, and therefore reanalyzed, as a regular nominative subject. T his, however, is only one possible explanation. The exact nature of the cause needs further investigation. Because as a whole, the data reported are generally consistent with the predictions advanced in Section 4.2.2, it is argued that the observed HS inno vations are a reflection of differences at the level of mental representation with respect to the adult controls. In other words, when it comes to class III psych verbs, HSs and MAs have distinct underlying mental representations that can license different outputs. But to be clear, though different, these two mental representations can yield identical surface structures since both of them have access to the canonical structure of class III psych verbs. Though not always convergent with the descriptions foun d in the theoretical literature, it should be noted that these outcomes are systematic and still representative of UG constrained grammars. Therefore, in an attempt to break away from the trend observed in most previous HS acquisitional studies that have d escribed the end result of HSs as incomplete, it should be noted that these underlying differences are not to be taken as deficiencies of any kind, but rather as emerging optionalities. In fact, as will be
186 evident in the ensuing discussion, I argue that th is particular optionality does not obtain as a result of an acquisition process that falls short of being complete at all ; especially since using such a term to describe a bilingual individual who has acquired a the HL naturalistically and in early childho od is questionable given our theoretical paradigm (Pascual y Cabo & Rothman 2012) Instead, it is proposed that HSs have acquired a fully developed linguistic system that differs from monolingual grammars of the same dialect as a result, among other things of the nature of the input available to them (Rothman 2007; Pires & Rothman 2009) A gain, and a s anyone would expect, most if not all of the input available for these HSs is affected to some degree by L1 attrition which in turn produces output that can b e more ambiguous for the parser. Consequently, HSs have no other recourse but to converge on grammars where compared to the monolingual varieties there has been some (re)analysis ( e.g., Lightfoot 2010). 6.3 Limitations of the S tudy As discussed in the previ ous sections, the findings of this study are important in that they contribute to our understanding of how and why HS grammars, despite being acquired naturalistically and in early childhood, can obtain so differently from monolingual norms. Undoubtedly, t hese findings contribute in non trivial ways to the current literature on formal (HS) language acquisition, syntactic theory, contact linguistics, and diachronic theory. This contribution is however, not without limitations. For example, one important lim itation of this study is the group pairing used. Though reasonable to a point, this methodology assumes that all informants (children and adults alike) have the same language learning background (Polinsky 2011:307). This can obviously not be true for
187 all i nformants. After all, a distinctive characteristic of HSs is the variation that can be found not only within groups but also within the same individuals. This variation must be representative of the different sociolinguistic realities in which these indivi duals have been brought up. For example, and to name just a few, differences in the quantity and quality of HL input they receive, instruction in the HL (or lack of it thereof), and/or the stigma attached with the HL in the area where they grew up. A much better alternative, which no one denies but that has not been done to date for reasons of practicality, would consist of a longitudinal study whereby one could witness knowledge and use of any given property (or lack of it) from the beginning and through t he different stages of development. If attrition is truly at play th e n such a methodology would capture the change in real individuals across time as oppose d to simply reconstructing and reasoning change out in cross linguistic comparisons. The same compar ative fallacy that applies as a limitation to Polinsky (2011) thus applies herein a priori Differently, however, whatever shortcomings might inherently apply do not seem to be deterministic here since the data showed no indirect evidence of attrition in this domain by comparing the children to the adult HSs anyway. Relying primarily on previous findings (i.e., invariable gusta) to shape the course of the methodology within the dissertation was both a strength as well as a limitation. In other words, alth ough, I was able to propose a specific hypothesis that was subject to empirical verification, it soon became obvious that not all possible alternatives for the data obtained had been considered. For example, because the focus of this dissertation has been on 3 rd person singular/plural agreement issues, I was not able to discard all possible explanations. For example, as discussed in Section 5. 4 .3, due to
188 the phonological characteristics of Cuban Spanish (i.e. / s/ deletion), I was not able to determine wi th precision the exact nature of the clitic agreement violations in HS Spanish. Thus, future studies examining knowledge and use of this property should aim at identifying ways to include all possible verbal and clitic agreement configurations (as well as other limitations 9 ) in experimental designs. That said, in spite of these methodological drawbacks, I believe that the data here discussed illustrate general linguistic trends of a representative group for each one of these communities. This study therefo re contributes to current linguistic research of communities in language contact situations. 6.4 Conclusion and S uggestions for F uture S tudies As discussed throughout this study one of the now commonplace findings of HS bilingualism studies involves competenc e and/or performance differences as compared to monolingual speakers of the HL dialect implicated (e.g ., Montrul 2008 inter alia ). The actual source of said differences remains, however, largely unidentified. Discerning which linguistic properties are poss ibly more susceptible than others to HS innovati on and providing an explanator y analysis of how and why this might come to be is at the forefront of current HS bilingualism studies (e.g., Cuza et al. 2013; Montrul 2008; Pires & Rothman 2009; Polinsky 2011) This is precisely what I have attempted to do in this study in the domain of class III psych verbs in HS Spanish. The present project aimed to accomplish two overarching goals. The first goal was to describe and explain the morpho syntactic consequences of a Spanish English 9 Other limitations of this study include the absence of critical and counterbalan ce items to further tease apart or compare other alternatives. For example, informants should have been tested on the possibility tested on the cano nical use of class III psych verbs, informants should have also been examined on the canonical use of class II psych verbs. This would have been optimal to show that there are in fact no differences between class III and class II.
189 language contact environment in an examination of class III psychological predicates. This somewhat understudied property was deemed appropriate for this purpose due to the structural differences between the two languages involved and due to previous findings that have documented gustar type verbs to be challenging for L2 learners (e.g., Montrul 1997, 2000, 2001) as well as HSs (e.g., de Prada Prez & Pascual y Cabo 2011; Toribio & Nye 2006). To move us beyond further documentation and description alone, I provided a series of hypotheses that were informed by previous findings (e.g., de Prada Prez & Pascual y Cabo 2011) as well as by syntactic theory (e.g., Belletti & Rizzi 1988; Landau 2010; Montrul 1995; Pesetsky 1995). Simply put, i t was predicted that in Spanish HS grammars, class III psych verbs had been reanalyzed as class II psych verbs. With this in mind, the first question of this dissertation asked whether the HS differences documented in this domain could be accounted for und er the formal proposed reanalysis I hypothesized. As discussed in detail in the first part of the present have already been discussed The second overarching goal w as to understand at a deeper level what apparent incomplete acquisition is, which I view as a mere difference devoid of evaluative terminology, and wha t its source(s) might be in the morpho syntactic domain of class III psych verbs With this in mind, I combined various experimental groups of children and adult HS participants as well as age matched monolingual speakers from the same linguistic background. Additionally, in an attempt to control for the type of i nput that HSs receive, I included a group of adult 1 st generation immigrants from the same linguistic background. This inclusion allowed
190 me to map, albeit indirectly, the possible influence of cross generational input modifications that, combined with an e ffect of contact with English, explain how HS knowledge is obtained (in the domain of class III psych verbs). In light of this, the second question aimed to document the extent to which L1 attrition incomplete acquisition and/or input compete can explain the differences observed in HS grammars in this domain. Recall from the discussion of exp eriment one that as children, both HSs and MAs showed uncertainty with respect to the innovation proposed. In examining the projected devel opment of each of these groups, it was clear that, with time, both groups would end up revealing developmental changes in a direction that is opposite to what L1 attrition would predict. In addition to this, the data convincingly show that, as adults, bo th groups of HS do have access to the canonical structure of class III psych verbs in the same way that monolingual speakers do ( Section 18.104.22.168 for the results of the grammaticality judgment task and Section 5.4 for the results of the elicited production task). This, in and o f itself, renders the incomplete acquisition approach unable to explain the additional differences observed in this domain ( e.g., invariable gusta ). Left to be determined is whether these HS outcomes could result from the differences i n the input speakers receive from 1 st generation immigrants. Although the results obtained here seem to support such a view, it is not possible to conclude this based solely on the data available since there are many other factors that influence and/or are involved in the process of acquisition. Nonetheless the evidence suggests that this is in fact a possibility. In other words, although it is not possible to determine without further questions that changes in the input available to HSs are in fact the
191 de fining factors that determine/motivate innovations in HS grammars, this is the only approach standing after factoring in all the data. But to be clear, this is not to say that all innovations must directly come from the input. Direct influence from English in the HS generation, distinct developmental paths as discussed by Putnam and Snc hez (in press) as well as changes that obtain as a consequence of the higher cognitive costs of bilingualism are likely primary contributors to emerging options in HS gramma rs also. To conclude, this dissertation has helped us advance our understanding of monolingual adult & children, 1 st generation immigrants) and a wide array of properti es related to class III psych verbs ( e.g., verb agreement). The results presented have laid some groundwork for further studies on HS linguistic development and have raised awareness about the explanatory adequacy of the notion of incomplete acquisition as a theoretical construct within formal approaches to language acquisition. In turn, I have also attempted to reasonably generalize beyond the domain of grammar examined in an effort to increase our understanding of the nature of HS grammars, the acquisition and maintenance of multiple linguistic systems within the same individual over time and, consequently, what all of this can tell us about the general architecture of linguistic representation. Finally, along with overcoming the limitations stated above, this project did not address the question of how processing differences may affect the use of class III psych predicates. Certainly, an analysis of processing resources is one of the next logical steps for the curr ent line of research since it will allow us to provide a more in depth analysis of the acquisition of the properties associated with psychological
192 predicates Furthermore, another logical extension of this dissertation is the study of other argument struct ure overgeneralizations/innovations in HS grammars, such as the use of intransitive verbs (both unaccusative and unergative) in transitive configurations w ith lexical causative meaning. It will be interesting to see to what extent we will be able to apply the hypothesis offered here that difference in HS grammars is more reflective of emerging optionality alongside intact monolingual knowledge. Testing more domains of grammar with this same mindset will also enable us to uncover, to the extent possible, the upper limitations of treating HS grammars as such.
193 APPENDIX A LANGUAGE PROFICIENCY TEST Instructions: Each of the following sentences contains a blank space ___ indicating that a word or phrase has been omitted. From the four choices select the one which, wh en inserted in the space ___, best fits in with the meaning of the sentence as a whole. 1 Al or del accidente de su buen amigo, Paco se puso _____ a. alegre b. fatigado c. hambriento d. desconsolado 2 No puedo comprarlo porque me _____ dinero. a. fal ta b. dan c. presta d. regalan 3 Tuvo que guardar cama por estar __ a. enfermo b. vestido c. ocupado d. parado 4 Aqu est tu caf, Juanito. No te quemes, que est muy _____ a. dulce b. amargo c. agrio d. caliente 5 Al romper los anteojos, Juan se asust porque no poda ____ sin ellos. a. discurrir b. or c. ver d. entender 6 Â¡Pobrecita! Est resfriada y no puede _____ a. salir de casa b. recibir cartas c. respirar con pena d. leer las noticias 7 Era una noche oscura sin _____ a. estrellas b. camas c. lgrimas d. nubes 8 Cuando don Carlos sali de su casa, salud a un a migo suyo: Buenos das, ____ a. Qu va? b. Cmo es? c. Quin es? d. Qu tal? 9 Â¡Qu ruido haba con los gritos de los nios y el _____ de los perros! a. olor b. sueo c. hambre d. ladrar 10 Para saber la hora, don Juan mir el _____ a. calendario b. bolsillo c. estante d. despertador 11 Yo, que comprendo poco de mecnica, s que el auto no puede funcionar sin _____ a. permiso b. comer c. aceite d. bocina 12 Nos dijo mam que era hora de comer y por eso _____ a. fuimos a nadar b. tomamos asiento c. comenzamos a fumar d. nos acostamos pronto
194 13 Â¡Cuidado con ese cuchillo o vas a _____ el dedo! a. cortarte b. torcerte c. comerte d. quemarte 14 Tuvo tanto miedo d e caerse que se neg a _____ con nosotros. a. almorzar b. charlar c. cantar d. patinar 15 Abri la ventana y mir: en efecto, grandes lenguas de _____ salan llameando de las casas. a. zorros b. serpientes c. cuero d. fuego 16 Compr ejemplares de todos los diarios pero en vano. No hall _____ a. los diez centavos b. el peridico perdido c. la noticia que deseaba d. los ejemplos 17 Por varias semanas acudieron colegas del difunto profesor a _____ el dolor de la viuda. a. aliviar b. dulcificar c. embromar d. estorbar 18 Sus amigos pudieron haberlo salvado pero lo dejaron _____ a. ganar b. parecer c. perecer d. acabar 19 Al salir de la misa me senta tan caritativo que no pude menos que _____ a un pobre mendigo que haba all sentando. a. peg arle b. darle una limosna c. echar una mirada d. maldecir 20 Al lado de la Plaza de Armas haba dos limosneros pidiendo _____ a. pedazos b. paz c. monedas d. escopetas 21 Siempre maltratado por los nios, el perro no poda acostumbrarse a _____ de s us nuevos amos. a. las caricias b. los engaos c. las locuras d. los golpes 22 Dnde estar mi cartera? La dej aqu mismo hace poco y parece que el necio de mi hermano ha vuelto a _____ a. dejrmela b. deshacrmela c. escondrmela d. acabrmela 23 P ermaneci un gran rato abstrado, los ojos clavados en el fogn y el pensamiento _____ a. en el bolsillo b. en el fuego c. lleno de alboroto d. Dios sabe dnde 24 En vez de dirigir el trfico estabas charlando, as que t mismo _____ del choque. a. sabe s la gravedad b. eres testigo c. tuviste la culpa d. conociste a las vctimas
195 25 Posee esta tierra un clima tan propio para la agricultura como para _____ a. la construccin de trampas b. el fomento de motines c. el costo de vida d. la cra de reses 2 6 Aficionado leal de obras teatrales, Juan se entristeci al saber _____ del gran actor. a. del fallecimiento b. del xito c. de la buena suerte d. de la alabanza 27 Se reunieron a menudo para efectuar un tratado pero no pudieron _____ a. desavenirse b. echarlo a un lado c. rechazarlo d. llevarlo a cabo. 28 Se negaron a embarcarse porque tenan miedo de_____ a. los peces b. los naufragios c. los faros d. las playas 29 La mujer no aprob el cambio de domicilio pues no le gustaba _____ a. el calle jeo b. el puente c. esa estacin d. aquel barrio 30 Era el nico que tena algo que comer pero se neg a _____ a. hojearlo b. ponrselo c. conservarlo d. repartirlo CLOZE TEST Instructions: In the following text, some of the words have been replaced by spaces which are numbered from 1 to 20. First, read the complete text in order to understand it. Then reread it and choose, from the list of words on the answer sheet, the correct word for each space. Mark your answers by circling your choice on the ans wer sheet, not on the text. El sueo de Juan Mir Hoy se inaugura en Palma de Mallorca la Fundacin Pilar y Joan Mir, en el mismo lugar en donde el artista vivi sus ltimos treinta y cinco aos. El sueo de Joan Mir se ha ______ (1). Los fondos donados a la ciudad por el pintor y su esposa en 1981 permitieron que el sueo se ________ (2); ms tarde, en 1986, el Ayuntamiento de Palma de Mallorca decidi ________ (3) al arquitecto Rafael Moneo un edificio que _______ (4) a la vez como sede de la entidad y como museo moderno. El proyecto ha tenido que _______ (5) mltiples obstculos de carcter administrativo. Mir, coincidiendo ________ (6) los deseos de toda su familia, quiso que su obra no quedara expuesta en ampulosos panteones de arte o en ________ (7 ) de coleccionistas acaudalados; por ello, en 1981, cre la fundacin mallorquina. Y cuando estaba _________ (8) punto de morir, don terrrenos y edificios, as como las obras de arte que en ellos _________ (9). El edificio que ha construido Rafael Moneo s e enmarca en _________ (10) se denomina "Territorio Mir", espacio en el que se han ________ (11) de situar los distintos edificios que constituyen la herencia del pintor.
196 El acceso a los mismos quedar ______ (12) para evitar el deterioro de las obras. Po r otra parte, se ______ (13), en los talleres de grabado y litografa, cursos ______ (14) las distintas tcnicas de estampacin. Estos talleres tambin se cedern peridicamente a distintos artistas contemporneos, _______ (15) se busca que el "Territorio Mir" ______ (16) un centro vivo de creacin y difusin del arte a todos los ______ (17). La entrada costar 500 pesetas y las previsiones dadas a conocer ayer aspiran ________ (18) que el centro acoja a unos 150.000 visitantes al ao. Los responsables esp eran que la institucin funcione a _______ (19) rendimiento a principios de la _______ (20) semana, si bien el catlogo completo de las obras de la Fundacin Pilar y Joan Mir no estar listo hasta dentro de dos aos.
197 APPENDIX B SOCIOLINGUISTIC BACKGROUND QUESTIO NNAIRE BACKGROUND QUESTIONNAIRE Particip ant #: _______________ 1) Gender: M F 2) Age : 3) Languages that you speak: 4) Place of birth: 5) 6) Places where you have lived for 6 months or longer: 7) Age when you arrived in the US: 8) Education level (pick one) Elementary High School College Graduate School 9) What is your dominant language? _____ __________________________________ 10) Do you speak any other languages besides English and Spanish? 11) Have you (explicitly) stu died Spanish grammar at school/university? If so, at what age did you start?________________ 12) (approximately) How many years have you studied Spanish? _______ __________ 13) What exposure to Spanish, other than school, do you have? How many hours a week ? With what regularity? ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ 14) What language do you use to communicate with the following people: Father: Mother: Brother(s): Sister(s): Son/Daughter: Friend(s): Partner: Co worker(s): Relatives: 15) What language do you primarily use to communicate in the following places/activiti es: Home: School: Work: Religious services: Free time activities: Hospital: 16) Have you ever spent more than one month in a Spanish speaking country? Explain: ______________________________________________________________________ ______ ________________________________________________________________ ______________________________________________________________________
198 17) On a scale of 1 10, ten being the highest (native speaker level), what do you believe your level of Spanish is now? ____________ Based on what factors are you calculating this number? ______________________________________________________________________ ______________________________________________________________________ ____ ________________________________________ __________________________ 18) On a scale of 1 10, ten being the highest (native speaker level), what do you believe your level of English is now? ____________ Based on what factors are you calculating this number? _______________________________________ _______________________________ ______________________________________________________________________ ____ __________________________________________________________________ 19) What are your motivations for learning Spanish? (Leave blank if it does not app ly) ______________________________________________________________________ ______________________________________________________________________ ____ __________________________________________________________________ 20) What are your motivations for learni ng English? (Leave blank if it does not apply) ______________________________________________________________________ ______________________________________________________________________ ____ ________________________________________________________________ __
199 APPENDIX C PARTICIPANT CONSENT FORM Informed Consent Protocol Title: Spanish English heritage speaker bilingualism Please read this consent document carefully. 1. Purpose of the research study: The purpose of this study is to examine the speech of Spanish Engl ish bilingual speakers. 2. What you will be asked to do in the study: You will be asked to give information about your language history, to fill out a Spanish questionnaire and do an oral production task. 3. Time required: Approximately 1 hour and 30 minutes 4. R isks and Benefits: There are no risks associated with this research project. 5. Confidentiality: Your identity will be kept confidential to the extent provided by law. Your information will be assigned a code number. The list connecting your name to this num ber will be kept in a locked file in the Principal Investigator's office. When the study is completed and the data have been analyzed, the list will be destroyed. Your name will not be used in any report. 6. Voluntary participation: Your participation in this study is completely voluntary. There is no penalty for not participating. 7. Right to withdraw from the study: You have the right to withdraw from the study at any time without consequence. 8. Whom to contact if you have questions about the study: Diego Pascual y Cabo, Graduate Student, Department of Spanish and Portuguese Studies, 170 Dauer Hall, P.O. Box 117405, Gainesville, FL 32611 7405, phone 815 915 5371 Jason Rothman, Assistant Professor, Department of Spanish and Portuguese Studies, 170 Dauer Hall, P.O. Box 117405, Gainesville, FL 32611 7405, phone 352 392 2016 9. Whom to contact about your rights as a research participant in the study: IRB02 Office, Box 112250, University of Florida, Gainesville, FL 32611 2250; phone 392 0433. 10. Agreement: I have read the pr ocedure described above. I voluntarily give my consent to participate in the Linguistics Research in bilingualism. I have recei ved a copy of this description. Participant: _____________________________________ _______ Date: _________________ Principal Inves tigator: ____________________________________ D ate: _________________ I have read the procedure described above. I voluntarily give my consent for my child, ________ ______________________, to participate in this research. I have received a copy of this de scription. Parent/Guardian ________________________________________ Date: _________________ 2 nd Parent/Witness_______________________________________ Date: _________________
200 APPENDIX D GRAMMATICALITY JUDGMENT TASK 1 1) TRANSITIVE PASSIVES La casa fue diseada p or los arquitectos El libro fue escrito por unos escritores El libro fue ledo por los estudiantes La tarea fue completada por los alumnos El paquete fue enviado por mis amigas 2) UNACCUSATIVE PASSIVES La mesa fue llegada por los estudiantes El telfono fue entrado por las chicas El chico fue aparecido por mis primos El vestido fue desaparecido por las chicas El paquete fue salido por los estudiantes 3) RPPS PASSIVES (CLASS III) La pizza fue gustada por mis amigos El pastel fue gustado por los nios La pelcu la fue gustada por mis amigos El chocolate fue gustado por mis amigos El helado fue gustado por los nios 4) PP PASSIVES (CLASS II) La nia fue asustada por los perros El profesor fue asustado por los estudiantes El chico fue asustado por sus amigos Mis padr es fueron asustados por los ladrones Mi hermano fue asustado por los profesores 5) DISTRACTERS UNGRAMMATICAL (adjective agreement) La casa blanco es muy grande El carro blanca fue caro El caf americana estuvo caliente Las nias altos eran de Costa Rica Lo s nios guapas vivan en Colombia GRAMMATICAL Nosotros vivimos en un apartamento Mis amigos son altos La comida est caliente Los libros de aventuras son interesantes
201 Las pelculas muy largas son aburridas Por las maanas siempre bebo un caf con leche El los tienen mucho dinero El perro bonito es pequeo Las pelculas cmicas son divertidas El libro rojo es aburrido UNGRAMMATICAL ( verb agreement ) Nosotros voy al centro comercial los fines de semana Ellas tengo dos hermanos muy altos Ella hago la tarea tod os los das Nosotros hago la comida algunos das Yo compramos la comida en el supermercado
202 APPENDIX E GRAMMATICALITY JUDGMENT TASK 2 1) Prescriptive use of RPPs. SG A la chica le gusta el perro A la estudiante le gusta el libro A la nia le gusta la pizza A los chicos les gusta la pelcula A los estudiantes les gusta el pastel A los nios les gusta el gato PL A la nia le gustan los carros A mi amigo les gustan las matemticas A el nio le gustan las ciudades A las nias les gustan los restau rantes A los nios les gustan los juguetes A mis amigos les gustan lasbicicletas 2) Omission of dative marker a SG m me gusta la pasta m me gusta la pizza m me gusta el carro t te gusta la leche t te gusta la leche t te gusta el caf P L m me gustan loslibros m me gustan las pelculas m me gustan los perros t te encantan los zapatos t te gustan los dibujos t te gustan las gatas
203 3) Clitic Omission SG A la nia gusta la playa A la chica gusta el arroz A la estudiante gusta el pollo A las estudiantes gusta la fsica A los estudiantes gusta tu vestido A los nios gusta el helado PL A la nia gustan los carros A la estudiante gustan los libros A la mujer gustan los zapatos A los nios gustan los peces A las nias gustan las pelculas A los hombres gustan los deportivos 4) Case agreement innovation SG Yo gusto tu vestido T gustas el arroz La nia gusta la playa Usted gusta el pastel Nosotros gustamos el pollo Mis amigos gustan l a pelcula PL Yo gusto tus zapatos T gustas las empanadas La nia gusta los aros Usted gusta los libros Nosotros gustamos los pasteles Mis amigos gustan las pelculas 5) The new grammatical construction SG Yo me gusta la pizza T te gus ta la pelcula ella le gusta el libro
204 usted le gusta la historia Nosotros nos gusta la fiesta Yo me gusta la fiesta PL Yo me gusta las empanadas T te gusta las pelculas Ella le gusta los libros Usted le gusta los habanos Nosotros nos gusta las fiestas Yo me gusta las fiestas 6) Distracters Transitive Yo compra el peridico en la tienda Yo vende fruta en el mercado T compra el pan todos la panaderia T vende carne en el supermercado Ella compra el peridico en la libreria Ella vende pescad o en el mercado Usted escucha msica en la discoteca Usted compra comida en el supermercado Nosotros compra la comida en el supermercado Nosotros vende ropa en el centro comercial Ellos compra el peridico en el parque Ellos vende fruta en el mercado Unna cusatives A la casa llega yo Del trabajo sale yo En mi casa entra tu En el parque aparece tu En mi casa aparece usted Por el trabajo llega usted Por el bar aparece el En el trabajo desaparece el De la clase sale nosotros En los museos entra nosotros Al tr abajo llega ellos De la clase sale ellos Unergatives Yo sonre en mi casa
205 Yo canta en las fiestas T canta en el bar T corre en el parque Usted llora en su casa Usted corre en el parque l trabaja en la oficina l llora en la casa Nosotros canta en la fiesta Nosotros corre en la playa Ellos corre en el gimnasio Ellos trabaja en el bar **Adjectival agreement Los chicos guapa no vienen a la fiesta Las mujeres alto no viven aqu El nio pequea duerme tranquilamente El magnfico profesora hablar de su ex periencia Las estudiantes tranquilo no estn en clase hoy La nia guapo no me habla La mujer casado tiene 2 hijas El restaurante cara est lejos La mesa rojo est rota La silla amarillo est en tu habitacin Grammatically correct Yo limpio mi cocina por l a maana Yo viajo con mis amigos durante las vacaciones Yo voy al gimnasio por la tarde Yo quiero comprar una computadora nueva Yo bailo salsa con mis amigos Yo vivo en una casa bonita T llegas al trabajo 5 minutos tarde T cocinas muy bien T vienes de vacaciones conmigo T compras tu ropa en el centro comercial T tienes una hermana muy guapa T no necesitas una computadora nueva Ella lee revistas en su tiempo libre Ella hace deporte todos los das l escribe poemas de amor l tiene 3 computadoras Ella trabaja por las tardes Ella quiere comprarse unos zapatos nuevos
206 Nosotros nos levantamos temprano todos los das Nosotros comemos pasta casi todos los das Nosotros vemos la televisin toda las noches Nosotros vamos de fiesta con nuestros amigos Nosotros trabajamos en un supermercado Nosotros no queremos ir a la tienda Ellos son estudiantes universitarios Ellos van a Florida todos los aos Ellas no quieren venir con nosotros Ellos tienen peces en su casa Ellos viajan a Europa todos los aos Ellos comen p asta todos los das
207 APPENDIX F OMISSION OF DATIVE CLITIC (HS ADV) INFORMANT TRANSCRIPTION HA1 C armen y E steban encantas las zanahorias HA1 D ilan gusta ir en aviones HA27 E duardo no gusta viajar HA27 Dilan si gusta viajar HA34 D ilan y L ilian gustan superman HA34 pero C armen y E duardo solo gustan spiderman HA46 l gusta viajar
208 APPENDIX G OMISSION OF DATIVE CLITIC (HS INT) INFORMANT TRANSCRIPTION HA3 C armen y E steban encantas las zanahorias HA4 D ilan gusta ir en aviones HA4 E duardo no gusta viajar HA4 Dilan si gusta viajar HA7 D ilan y L ilian gustan superman HA7 pero C armen y E duardo solo gustan spiderman HA11 E duardo encanta helado HA11 C armen y E duardo encantan verduras HA11 pero dilan y lilian no gustan verduras HA18 C armen y E duardo gustan los dep ortes HA18 pero D ila n y L ilian no gustan los deportes HA21 C armen y Eduardo gustan el C aribe HA21 pero D ilan y L ilian no gustan el C aribe HA22 L ilian gusta a a comprar cosas HA22 y C armen gusta comprar comida HA22 Carmeny .. y E duardo gustan las hamburguesas HA22 y D i la n y L ilian no gustan las hamburguesas HA22 pero Carmen no gusta la s pizzas HA41 C armen no gusta pasta HA41 pero Lilian s gusta pasta HA41 D ilan y L ilian gustan sushi HA41 pero C armen y E duardo no gustan sushi HA41 C armen y E duardo no gustan caf HA41 pero D ilan y L iliansi gustan caf HA41 C armen gusta los insectos HA20 pero E duardo no gusta los insectos HA20 D ilan y L ilian gusta sushi HA20 pero E steban gusta a spiderman
209 LIST OF REFERENCES Alberts, H 2005 Changes in E thnic S olidarity in Cuban Miami. The Geographical Review 95, 231 248. Arad, M. 1998. VP Structur e and the Syntax lexicon Interface. Doctoral dissertation, University College London. Austin, J., M. Blume and L. Snchez. Forthcoming Morphosyntac tic Attrition in the L1 of Spanish English Bilingual Children Hispania Beaudrie S. M. and M. Fairclough. 2012 Spanish as a Heritage Language in the United States: The State of the Field. Georgetown: Georgetown University Press. Belletti, A. and L. Rizzi. 1987 Psych Theory. Natural Language and Linguistic Theory 6, (3), 291 352. Benmamoun, A. S. Montrul, and M. Polinsky. 2 010 White Paper: Prolegomena to Heritage Linguistics. Harvard University. Beretta, A., and C. Campbell. 2001. Psychological V e rbs and the D ouble dependency H ypothesis. Brain and Cognition 46, (1/2), 42 46 Bialystok, E. 2009. Bilingualism: The G ood, the B ad, and the I ndifferent. Bilingualism: Language and Cognition 12, 3 11. Birdsong, D. 1999 Second L anguage A cqu isition and the Critical Period Hypothesis. Mahwah, NJ: Erlbaum. Bley Vroman R 1990 The L ogical P roblem of F oreign Language L earning Linguistic Analysis: 20 : 3 49 Borer, H. and K. Wexler. 1987 The M aturation of S ynta x. In T.Roeper and E.Williams (e ds.), Pa rameter setting 123 172. Dordrecht: Reidel. Boswell, T 2000 Demographic Changes in Florida and their Importance for Effective educational policies and practices. In A. Roca ( e d.) Research on Spanish in the United States: Linguistic Issues and challenges Somerville, MA: Cascadilla. 406 431. Bouchard, D. 1995 The Semantics of Syntax University of Chicago Press: Chicago Bowles, M., Mo ntrul, S., 2009. Instructed L2 A cquisition of D ifferential O bject M arking in Spanish. In: Leow, R., Campos, H., Lardiere, D. ( e ds. ), Little Words. Their History, Phonology, Syntax, Semantics, Pragmatics and Acquisition, Georgetown University Round Table. Georgetown University Press,Georgetown, 110 199.
210 Brinton, D., O. Kagan, and S. Bauckus. 2008. Heritage Language Education: A New Field Emerging New York: Routledge. Buajasn Marr awi, J. and J. L. Mndez Mndez. 2005 La Repblica de Miami La Habana: Ciencias Sociales. Ca rreira, M. and Potowski, K. 2011 Commentary: Pedagogical I mplications of E xperimental SNS research. Invi ted article, special issue on Spanish linguistics. Heritage Language Journal, 8 (1), 134 151. Castro, M., M. and A. Roca. 1990 The O fficial English M ovement in Florida. In Adams K. & Brink, D. ( e ds.) Perspectives on official English: The campaign for Engli sh as the official language of the USA Berlin: Mouton de Gruyter 151 116. Castro, M.1992a The P olitics of L anguage in Miami. In Grenier, G. & Stepick III, A. ( e ds .) Miami now! Immigration, E thnicity, and S ocial C hange Gainesville, FL: Universit y Press of Florida 109 132. Castro, M. 1992b On the C urious Q uestion of L anguage in Miami. In Crawford, J. ( e d.) Language loyalties: A source book on the official English controversy Chicago: Chicago UP 178 186. Cazzoli G oeta, M. and M. Young Scholten. 2011 Yo G C hoice or S yntactic A ttrition?. In ( Eds. ) Potowski, K. and Rothman, J., Bilingual Youth: Spanish in English speaking Societies Joh n Benjamins: Philadelphia, PA. Chomsky, N. 1965. Aspects of the Theory of Syntax Cambridge, MA: MIT Pr ess. Chomsky, N. 1972. Problems of Knowledge and Freedom London, Fontana Chomsky, N. 1986. Knowledge of Language: Its Nature, Origin and Use New York, Praeger Chomsky,N. 2004. The Generative Enterprise Revisited: Discussions with Riny Huybregts, Henk van Riemsdijk, Naoki Fukui, and Mihoko Zushi, with a new foreword by Noam Chomsky Berlin: Mouton de Gruyter. Cook, W. A. 1972 A Case Grammar Matrix. Georgetown University Working Papers on Languages and Linguistics, 6. Washington DC: Georgetown University P ress. Crain, S. and R. Thornton. 1999 Investigations in Universal Grammar: A Guide to Experiments on the Acquisition of Syntax and Semantics Cambridge: MA: MIT Press. Curtiss, S. 1977. Genie: A P sycholinguistic S tudy of a M odern day "'Wild Child." New Yo rk: Academic Press.
211 Cuza, A. and J. Frank. 2011 Transfer Effects at the Syntax Semantics Interface: The Case of Double que Questions in Heritage Spanish. Heritage Language Journal (8) 1. Cuza, A., R. Prez Tattam, E. Barajas, L. Miller, and C. Sadowski. 2 013. The D evelopment of T ense and A spect M orphology in C hild and A dult H eritage Spanish: Implications for H eritage L anguage P edagogy In John Schwieter ( e d.), Innovative R esearch and P ractices in S econd L anguage A cquisition and B ilingualism. Amsterdam: Joh n Benjamins. DeKeyser, R. M. 2000. The R obustness of C ritical P eriod E ffects in S econd L anguage A cquisition. Studies in Second Language Acquisition 22 : 499 533. De la Torre, Miguel. 2003 La L ucha por Cuba. Religion and P olitics on the S treets of Miami. Berkeley: University of California Press. De Prada Prez, A. and D. Pascual y Cabo. 2011 Invariable G usta in the Spanish of Heritage Speakers in the US. Proceedings of the 11 th Generative Approaches to Second Language Acquisition (GASLA 2011) edited by Jul ia Herschensohn and Darren Tanner, 110 120. Somerville, MA: Cascadilla Proceedings Project. De Prada Prez, A. and D. Pascual y Cabo. 2012 Interface Heritage S peech across P roficiencies: Unaccusativity, F ocus, and S ubject P osition in Spanish. In Geeslin, K. and Daz Campos, M. ( E ds.), Proceedings of the Hispanic Linguistic Symposium (HLS). De Prada Prez, A. and D. Pascual y Cabo (Forthcoming). Input and Quirky Agreement: Comparing Heritage Speakers and L2ers. Proceeding of the Generative Approaches to La nguage Acquisition. De Prada Prez, A. and A. Rodrguez Ricelli, K. Woodfine, and S. Rogers. 201 2 Weakening in Subject Verb Number Agreement in Spanish in Contact. Poster session presented at the Hispanic Linguistics Symposium Gainesville, FL. Domnguez, L. 2009 Charting the Route of Bilingual Development: Contributions from Special Issue of the International Journal of Bilingualism: Understanding the Nature of Early Bilingualism: Romance Languages as Heritage Languag es 13 (3), 271 287. Dorian, N. C. 1989 Investigating Obsolescence: Studies in Language Contraction and Death Cambridge: Cambridge University Press. Ducar, C. 2008 Student Voices: The Missing Link in the Spanish Heritage Language Debate. Foreign Language Annals 41 (3), 678 700. Dvorak, T. and C. Kirschner. 1982 Mary Likes Fishes: Reverse Psychological Phenomena in New York Puerto Rican Spanish. Bilingual Review 9 (1), 59 65.
212 F ield, F. 2011. Bilingualism in the USA: The Case of the Chicano Latino Community Amsterdam & Philadelphia: John Benjamins, Figueira, R. 1984 On the D evelopment of the E xpression of C ausativity: A Syntactic H ypothesis. Journal of Child Language 11, 109 127. Flege, J., I. Mackay, and T. Piske. 2002. Assesing Bilingual Dominance. Applied Psycholinguistics 23, 567,598. Fillmore, C.J. 1968. The Case for Case. In E. Bach and R. Harms ( Eds. ), Universals in Linguistic Theory New York: Holt, Rinehart and Wins ton. Fodor, J. 1984. Observation R econsidered. Philosophy of Science, 51 : 23 43. Fox, D., and Y. Grodzinsky. 1998 iew from the B y phrase. Linguistic Inquiry 29, 311 332 Gmez Soler, I.2011. The L1 Acquisition of G ustar : Evidence ag ainst Maturation. In Selected Proceedings of the4th Conference on Generative Approaches to Language Acquisition North America (GALANA 2010) ed. Mihaela Pirvulescu et al., 51 65. Somerville, MA: Cascadilla Proceedings Project. Gmez Soler, I. 2012. Acquiri ng Spanish Psych Verbs: Maturation or Continuity? In Alia K. Biller, Esther Y. Chung, and Amelia E. Kimball (Eds.) Proceedings of the 36th a n nual Boston University Conference on Language Development (BUCLD 36). Guasti, M. T. 2004. Language Acquisition: T he Growth of Grammar MA: MIT Press Guijarro Fuentes, P., and T. Marinis. 2007 Acquiring the S yntax/semantic I nterface in L2 Spanish: T he P ersonal P Eds. ), Eurosla Yearbook (7). Amsterdam: John Benjamins. 67 87 Grel, A. 2002: Linguistic C haracteristics of S econd L anguage A cquisition and F irst L anguage A ttrition: Overt versus N ull P ronouns. Unpublished PhD dissertation, McGill University, Montreal, Canada. Gutirrez Bravo, R. 2007 Prominence Scales and Unmarked Word Or der in Spanish. Natural Language and Linguistic Theory (25) 2, 235 271. Grimshaw, J. 1990 Argument Structure Cambridge, MA: MIT Press. Grinstead, J. (2001). Wh Movement in Child Catalan. Issues in Applied Linguistics 12(1), 5 28. Grinstead, J. (2000). Constraints on the C omputational C omponent vs. G rammar in the L exicon: A D iscussion of Bates & Goodma n Journal of Child Language 27(3), 737 743.
213 Havlac, J. 2003 Second G eneration S peech, L exicon, C ode switching and M orpho syntax of Croatian English B ilinguals Frankfurt: Peter Lang. Horgan, D. 1978 The D evelopment of the F ull P assive. Journal of Chil d Language 5: 65 80 Hurtado, Luz M. 2002 El M antenimiento del E spaol en Miami Dade, Florida. Litterae 11: 148 170. Hyams, N. 1986. Language Acquisition and the Theory of Parameters Dordrecht: Reidel. Hyltenstam, K. and N Abrahamsson. 2000. Who C an B e come N ative like in a S econd L anguage? All, S ome, or N one? On the M aturational C onstra ints C ontroversy in S econd L anguage A cquisition. Studia Linguistica 54.150 66. Jacke ndoff, R. 1990 Semantic Structures Cambridge, MA: The MIT Press. Jakobson, R. 1941 Child Language, Aphasia, and Phonological Universals. The Hague: Mouton. Kondo Brown, K. 2006. Heritage L anguage D evelopment: Focus on East Asian I mmigrants St udies in Bilingualism Series 32 Amsterdam, Netherlands: John Benjamins Kpke, B 2004. Neurolinguistic A spects of A ttrition. Journal of Neurolinguistics,17 (1), 3 30. Kpke, B. and M. Schmid. 2004 Language Attrition: The Next Phase. In M. Schmid, B. Kopke, M. Keijzer, and Weilemar, L. ( Eds. ), First Langu age Attrition: Interdisciplinary Perspectives on Methodological Issues. Studies in Bilingualism 28, 1 43. Amsterdam: John Benjamins. Kpke, B., M. Sc hmid, Keijzer, and S. Dostert. 2007 Language Attrition: Theoretical Perspectives Studies in Bilingualism 33. Amsterdam: John Benjamins. Kroll, J. and P.E. Dussias. 2012. The C omprehension of W ords and S entences in T wo L anguages. In T. Bhatia & W. Ritchie (Eds.). The Handbook of Bilingualism and Multilingualism 2nd Edition. Malden, MA: Wiley Blackwell Publish ers. Landau, I. 2010. The Locative Syntax of Experiencers Cambridge, MA: MIT Press. Lenneberg EH .1967. Biological Foundations of Language New York: Wiley. Levin, B. 1993 English Verb Classes and Alternations: A Preliminary Investigation Chicago: Univer sity of Chicago Press.
214 Lightfoot, D. 1991. How to Set Parameters: Arguments from Language Change Cambridge, MA.: MIT Press. Lightfoot, D. 1999. The Development of Language: Acquisition, Change and Evolution Oxford: Blackwell. Lightfoot D. 2010. Language Acquisition and Language Change Wiley Interdisciplinary Reviews: Cognitive Science (1), 5: 6 77 684. Lightfoot, D. and M. Westergaard 2007. Language Acquisition and Language Change: Inter relationships. Language and Linguistics Compass 1:396 416. Long M. 2005. Problems with S upposed C ounter evidence to the C ritical P eriod H ypothesis. International Review of Applied Linguistics 43(4).287 317. Lpez Jimnez, M. D. 2003 Grammar I nstruction and the A cquisition of G ustar type V erbs by English speaking L e arners of Spanish. ELIA 4, 255 279. Lpez Morales, H. 2003 Los Cubanos de Miami; Lengua y Sociedad Miami: Ediciones Universal. Lpez Morales, H. 2009 Madrid: Instituto Cervantes, 2009. Print. Lord, C. 1979 G eneralizations R egarding C ause and T ransitivity. PRCLD 17, 81 89. Lynch, A. 1999 The Subjunctive in Miami Cuban Spanish: Bilingualism, Contact, and Language Variability. PhD diss., University of Minnesota. Lynch, Andrew. 2000. Spanish Speaking Miami in Sociolinguistic Perspective: Bilingualism, Recontact, and Language Maintenance among Cuban Origin Population. Rese arch on Spanish in the United States Ed. Ana Roca. Somerville MA: Cascadilla Press. 271 283. Lynch, A. 2009 Expression of C ultural S tanding in Miami: Cuban Spanish D iscourse about Fidel Castro and Cuba. Revista Internacional de Lingstica Iberoamericana 14, 21 48. Lynch, A. 2012. Key Concepts for Theorizing Spanish as a Heritage Language In S. Beaudrie & M. Fairclough (Eds.), Spanish as a Heritage Language in the United States: The State of the Field Washington DC: Georgetown University Press. MacWhinn ey, B. 2000. The CHILDES P roject: Tools for A nalyzing T alk. Third Edition. Mahwah, NJ: Lawrence Erlbaum Associates. Manouilidou, C. 2008 psychological predicates. Journal of Neurolinguistics 22,2. 167 186.
215 Masullo, P. J.1992. Incorporation and Case Theory in Spanish. A crosslinguistic perspective. Doctoral dissertation, University of Washington, Seattle. McDaniel, D., C. McKee, and C.H. Cairns. 1996 Methods for A ssessing C S yntax C ambridge, MA: MIT Press Meisel, J. 2011. First and Second Language Acquisition. Parallels and Differences Cambridge: Cambridge University Press. Montrul. S. 2000 Causative P sych V erbs in Spanish L2 A cquisition. In ( Eds. ) Ronald P. Leow and Cristina Sanz Spanish Applied Linguistics at the Turn of the Millennium 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, 3 9 68. Montrul, S. 2004a The Acquisition of Spanish. Morphosyntactic Development in Monolingual and Bilingual L1 Acquisition and in Adult L2 Acquisition. [Series on Language Acquisition and Language Disorders]. Amsterdam: John Benjamins. Montrul, S. 2004b Subject and O bject E xpression in Spanish H eritage S peakers: A C ase of M orpho syntactic C onvergence. Bilingualism: Language and Cognition 7, 1 18. Montrul, S. 2005 Second L anguage A cquisition and F irst L anguage L oss in A dult E arly B ilinguals: Exploring s ome D ifferences and S imilarities. Second Language Research 21, 3, 199 249. Montrul, S. 2006 On the Bilingual Competence of Spanish Heritage Speakers: Syntax, Lexical Semantics and Processing International Journal of Bilingualism, 10, 37 69. Montrul, S. 2 008 Incomplete Acquisition in Bilingualism. Re examining the Age Factor Amsterdam: John Benjamins. Montrul, S. 2009 Knowledge of Tense Aspect and Mood in Spanish Heritage Speakers. International Journal of Bilingualism, 13 (2), 239 269. Montrul, S. 201 0. Is Grammar Instruction Beneficial for Heritage Language Learners? Dative Case Marking in Spanish. Heritage Language Journal 7, (1), winter, 47 73. Montrul, S. 2011 a The Linguistic C ompetence of Heritage Speakers. Studies in Second Language Acquisition 33(2), 15 5 161. Montrul, S. 2011b. Multiple I nterfaces and I ncomplete A cquisition. Lingua 121, 591 604
216 Montrul, S. 2012 The Grammati cal Competence of Spanish Heritage Speakers. In S. Beaudrie & M. Fairclough (Eds.), Spanish as a Heritage Language in the United States: The State of the Field Washington DC: Georgetown University Press. Montrul, S. and M. Bowles. 2009. Back to Basics: In complete Knowledge of Differential Object Marking in Spanish Heritage Speakers. Bilingualism: Language and Cognition 12 (3), 363 383. Montrul, S. and M. Bowles. 2010. Is Grammar Instruction Beneficial for Heritage Language Learners? Dative Case Marking in Spanish. Heritage Language Journal 7(1), 47 73. Montrul, S., R Bhatt, and A. Bhatia 2012. Linguistic Approaches to Bilingualism 2:2, 141 176. Mon trul, S. and M. Polinsky. 2011 Why not H eritage S peakers? Linguistic Approaches to Bilingualism. 1(1). Montrul, S. and C. Rodrguez Louro. 2006. Beyond the S yntax of the Null Subject Parameter. A look at the discourse pragmatic distribution of null and overt subjects by L2 learners of Spanish. I n The Acquisition of Syntax in Romance Languages ed. V. Torrens and L. Escobar. Amsterdam: John Benjamins. Myers Scotton, C. 2007 The G rammatical P rofile of L1 S peakers on the S tairs of P otential L anguage S hift. In Kpke B, Schmid M S, Keijze M and Doste rt S ( Eds. ), Language Attrition: Theoretical perspectives (69 82). Amsterdam: John Benjamins. Norman, G. 2010 Likert Scales, Levels of Measurement Advances in health sciences education, 15 : 625 632. Ortega, L 1998 Cubanos e n Miami La Habana: Ciencias Sociales Ortega Santos, I. 2006. On Postverbal Subjects, PF and the Copy Theory: The Spanish Case. Proceedings of the Joint Meeting of the Hispanic Linguistics Symposium & The Conference on the Acquisition of Spanish and Portug uese as First and Second languages. Cascadilla University Press. 56 66. Pascual y Cabo. D., and J. Rothman. 2012. The (Il)logical Problem of Heritage Speaker Bilingualism and Incomplete Acquisition. Applied Linguistics 33 (4), 1 7. Pascual y Cabo, D., J. Rothman, and A. Lingwall. 2012 Applying the Interface Hypothesis to Heritage Speaker Acquisition: Evidence from Spanish Mood. In Alia K. Biller, Esther Y. Chung, and Amelia E. Kimball (Eds.) Proceedings of the 36th a n nual Boston University Conference on Language Development (BUCLD 36).
217 Paradis, M. 2004. A Neurolinguistic Theory of Bilingualism Amsterdam: John Benjamins. Paradis, J. and F., Genesee. 1996. Syntactic Acquisition in Bilingual Children: Autonomous or Interdependent? Studies in Second Langua ge Acquisition, 18 : 1 25. Parodi Lewin, C. 1991. Aspects in the S yntax of Spanish P sych verbs. Doctoral dissertation. University of Cali fornia, Los Angeles. Parodi, C, and M. Lujn. 2000. Aspect in Spanish psych verbs. In H. Campos (ed.), Hispanic Linguis tics at the Turn of the Century. Somerville, MA: Cascadilla Press, 210 221. Pavlenko, A. 2000 L2 Influence on L1 Late Bilingualism. Issues in Applied Linguistics 11 (2), 175 205. Penfield, W., and Roberts, L. (1959 ). Speech and Brain mechanisms. Princeto n University Press, Princeton, N.J. Pires, A. and J. Rothman. 2009 Disentangling S ources of I ncomplete A cquisition: An Explanation for C ompetence D ivergence across H eritage G rammar. International Journal of Bilingualism 13(2), 1 28. Pesetsky, D. 1995 Zer o Syntax: Experiencers and C ascades MIT Press: Cambridge Pierce, A. 1992 The A cquisition of P assive in Spanish and the Q uestion of A chain M aturation. Language Ac quisition 2:55 8 Polinsky, M.: 1994, Structural D imensions of F irst L anguage Loss. Chicago Linguistic Society. 30 th Regional Meeting. Parasession on Variation 257 276. Polinsky M. 2006 Incomplete Acquisition: American Russian Journal of Slavic Linguistics ,14(2), 191 262. Polinsky, M. 2008a Russian G ender under I ncomplete A cquisition. Herita ge Language Journal 6, 1. http://www.heritagelanguages.org/ Polinsky, M. 2008b Heritage L anguage N arratives. In D. Brinton, O. Kagan & S. Bauckus ( Eds. ). Heritage language education. A new field emerging, 149 164. New York: Routledge. Polinsky, M. 2011 Reanalysis in A dult H eritage L anguage: A case for attrition S tudies in Second Language Acquisition 33: 305 328. Polinsky M. and O. Kagan 2007 Heritage L W and in the C lassroom Language and Linguistics Compass 1: 368 395.
218 Porcel, J 2006 The Paradox of Spanish among Miami Cubans. Journal of Sociolinguistics 10, 1: 93 110. Potowski, K. 2005 Fundamentos de la E nse anza del E spaol a los H ablantes N ativos en los Estados Unidos Madrid, Spain: Arco/Libros. Potowski, K.2010 Language D iversity in the United States. Cambridge University Press. Potowski, K. and Came ron, R. ( Eds. ). 2007. Spanish in Contact: Educational, Li nguistic, and Social Inquiries. Selected proceedings from the 20 th Conference on Spanish in the United States and the 5 th Conference on Spanish i n Contact with Other Languages. New York: John Benjamins. Putnam, M. and L. S omplete about Incomplete Acquisition: A Prolegomenon to Modeling Heritage Language Grammars. To appear in Linguistic Approaches to Bilingualism Quesada, M. L. 2008. Yo G usto P asteles de C hocolate: De la T ransitividad hacia la I ntransitividad en la A dquisi cin de V erbos P sicolgicos en E spaol.. In, Rosa Maria Ortiz Ciscomani (ed.), Memoria del IX EILN. Hermosillo, Mxico: Editorial Universidad de Sonora, 55 72. Rod rguez Mondoedo, M. 2008 The A cquisition of D ifferential O bject M arking in Spanish. Probus 20, 111 145. Rothman, J. 2007 Heritage Speaker Competence Differences, Language Change and Input Type: Inflected Infinitives in Heritage Brazilian Portuguese. International Journal of Bilingualism, 11(4), 359 389. Rothman, J. 2009. Understanding the N atu re and O utcomes of E arly B ilingualism: Romance L anguages as H eritage languages. International Journal of Bilingualism 13(2), 155 165. Rothman, J. and M. Iverson. 2010. Independent Normative Assessments for Bi/Multilingualism, Where art thou? In Cruz Ferrei ra (ed.), Multilingual Norms. Frankfurt: Peter Lang. Roth man J., D. P ascual y Cabo & A. L ingwall ( In preparation ). The Interface Hypothesis and Heritage Speaker Bilingualism: Modality Alternations in HS Spanish. Rubio, F. 2001 Psychological V erbs in Spanish: Two T eaching A pproaches. Academic Exchange Quarterly. 136 141. Sanchez, L. 2004. Functional Convergence in the Tense, Evidentiality and Aspectual Systems of Quechua Bilingualism: Language and Cognition. 7(2), 147 162.
219 Schachte r J. 1990 On the Issue of C ompleteness in S econd L anguage A cquisition Second Language Research. 6 : 93 124 Schmid, M.S. 2002: First L anguage A ttrition, U se, and M aintenance: T he C ase of German Jews in Anglophone C ountries Amsterdam/Philadelphia : John Benjamins. Smid, M.S. and B. Kpke. 2008. L1 A ttrition and the M ental L exicon. I n: The bilingual mental lexicon: Interdisciplinary approaches de A. Pavlenko (Ed.). Clevedon: Multilingual Matters. Schmid, M.S., B. Kpke M. K eijzer and L. Weilemar ( 2004). First L anguag e Attrition : I nterdisciplinary P erspectives on M ethodological I ssues Amsterdam: John Benjamins. Schmid, M.S. and E. Dusseldorp. 2010. Quantitative A nalyses in a M ultivariate S tudy of L anguage A ttrition: The I mpact of E xtralinguistic F a ctors. Second Language Research, 26 (1),125 160. Schmid, M. 2011 Language Attrition Cambridge: Cambridge UP. Schwartz, B. D. and Sprouse, R. 2000 When S yntactic T heories Evolve:C onsequences for L2 acquisition R esearch. In J. Archibald., editor, Second Language Acquisition and Linguistic theory Oxford: Blackwell, 156 86. Seliger. H. 1991: Language A ttrition, R educed R edundan cy, and C reativity. In Seliger, H. and Vago, R.M., (eds.), First language attrition. Cambridge: Cambridge University Press, 227 40. Silva Corvaln, C. 1994 Language Contact and Change: Spanish in Los Angeles Oxford: Oxford University Press. Snyder, W. 2007 Child Language: The Parametric Approach Oxford, UK: Oxford University Press. Sorace, A 2000 Syntactic O ptionality in N o n native G rammars. Second Language Research, 16 (2), 93 102. Sorace, A. 2002. Differential Effects of Attrition in the L1 S yntax of N ear native L2 S peakers. In C. Howell, S. Fish, and T. Keith Lucas ( Eds. ), Proceedings of the 24 th Boston University Confere nce on Language Development, 714 725. Somerville, MA: Cascadilla Press. Sorace, A. 2004 Native L anguage A ttrition and D evelopmental I nstability at the S yntax D iscourse I nterface: Data, I nterpretations and M ethods. Bilingualism: Language and Cognition, 7 143 145.
220 Sorace, A. 2011. Pinning D own the C I B ilingualism. Linguistic Approaches to Bilingualism 1 (1), 1 33. Sorace, A. 2012. Pinning D own the C oncept of I nterface in B ilingual D evelopment: A R eply to P eer C ommentaries. Linguistic Approaches to Bilingualism 2(2), 209 217. Thompson, C. and M, Lee. 2009. Psych Verb Production and Comprehension in Jour nal of Neurolinguistics 22:. 354 369. Toribio, A. J. 2004 Spanish/English S peech P ractices: Bringing C haos to O rder. In Bilingualism and Language Pedagog y, J. Brutt Griffler and M. Varghese ( Eds. ), 41 62. Multilingual Matters Toribio, A. J. and C. Nye. 20 06. Restructuring of Reverse Psychological Predicates. New Perspectives on Romance Linguistics 263 277. John Benjamins: Amsterdam. Torrens, V ., L. Escobar, and K. Wrexler. 2006 The Acquisition of E xperiencers in Spanish L1 and the E xternal A rgument R e quirement H ypothesis. In Torrens, V. and Escobar L. ( Eds. ), The Acquisition of Syntax in Romance Languages 183 202. John Benjamins: Philadelphia, PA. Toth, P. D. 2003.Psych Verbs and Morphosyntactic Development in Instructed L2 Spanish. In S. Montrul & F Ordez (eds.), Linguistic Theory and Language Development in Hispanic Languages Somerville, MA: Cascadilla Press, 468 497. Tsimpli, I.M., A. Sor ace, C. Heycock and F. Filiaci. 2004 First L anguage A ttrition and S yntactic S ubjects: A S tudy of Greek and Italian N ear native S peakers of English. International Journal of Bilingualism, 8 257 277. Tsimpli, I.M.. 2007 First Language Attrition from a Minimalist Perspective: Interface Vulnerability and Processing Effects. In Kpke, B., Schmid, M., Keijzer, Dost ert, S. ( Eds. ), Language Attrition: Theoretical Perspectives Studies in Bilingualism 33, 83 98. Amsterdam: John Benjamins. Valds, G. 1995 Pedagogical and Theoretical Challenges. Modern Language Journal 79, 299 328. Valds, G. 2003 Expanding Definitions of Giftedness: Young I nterpreters of Immigrant Background. Mahwah, NJ: Erlbaum Valds, G. 2005 Bilingualism, Heritage Language Learners, and SLA Research: Opportunities Lost. The Modern language Journal, 89(3) 410 426.
221 Valds, G., J. Fishman, R. Chvez & W. Prez. 2006 Developing Minority Language Resources. The Case of Spanish in California Clevedon, UK: Multilingual Matters Valds, G., A. Lozano, R. Garca Moya. 1981 Spanish Language Progra m for Hispanic Minorities: Current Needs and Priorities. In Dexter Fisher (ed.), Minority Language and Literature. New York: Modern Language Association. Ventureyra, V., and C. Pallier. 2004. In S earch of the L ost L anguage: The C ase of A dopted Koreans in F rance. In M. Schmid, B. Kpke, M. Keijzer and L. Weilemar ( Eds. ), First language attrition: Interdisciplinary perspectives on methodological issues (207 221). Amsterdam: John Benjamins. Villa Garca J. 2010. To Agree or Not to Agree: Beyond Quintess entia lly Syntactic Agreement in Spanish. In Colina, S., A. Olarrea, & A. M. Carvalho (eds.), Selected papers from the 39th Linguistic Symposium on Romance Languages (LSRL) Amsterdam/Philadelphia: John Benjamins, 249 266. White L., C. Brown, J. Bruhn Garavito D. Chen, M. Hirakawa, and S. Montrul. 1999. Psych verbs in S econd L anguage A cquisition. In The Development of Second Language Grammars: A Generative approach E. Klein and G. Martohardjono ( Eds. ), 173 199, Amsterdam: John Benjamins. Wi ley, T. G. and G. V alds. 2000 Heritage Language Instruction in the United States (Special Issue). Bilingual Research Journal 24 (4). Zagona, K. 2002 The Syntax of Spanish New York: Cambridge University Press Zubizarreta, M. L. 1998. Prosody, Focus, and Word Ord e r Cambri dge, MA: MIT Press
222 BIOGRAPHICAL SKETCH A native of Madrid (Spain), Diego Pascual primary research interests lie in the areas of heritage speaker bilingualism and second language acquisition Diego work which is approached from a generative point of view has appeared in several scholarly journals (inter)national proceedings, and edited volumes After receiving a Bachelor of Arts in English p hilology from the University of Granada in Spain and a Master of Arts in Spanish from Northern Illinois University Diego completed his Ph.D. from the University of Florida in the Spring of 2013 Diego is currently an Assistant Professor at Texas Tech University.