University of Florida Thesis Or Dissertation Formatting

THE PROCESSING OF GENDER AGREEMENT IN MONOLINGUAL AND HERITAGE SPEAKERS OF SPANISH

DÁMARIS MAYANS RAMÓN

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

2018

Para mi madre, mi padre y el Arturo

ACKNOWLEDGMENTS

My deepest gratitude would first go to my advisor Dr. Valdés Kroff for his continuous help and for sharing his knowledge and passion for psycholinguistics with me. I also want to express my gratefulness to Dr. De Prada Pérez for her guidance and outstanding encouragement during my first years in the doctoral program. Besides my advisors, I would like to thank the rest of my committee members, Dr. Kaan and Dr. Blondeau, for their support throughout my dissertation work and for being such outstanding professors in their respective fields. My sincere thanks also go to Dr. Lord for being part of my committee when I needed her and for her complete support from the very beginning of the Ph.D. program.

I want to express my gratitude to Dr. Ana Margarita Haché, Rafaela Carrasco, Dr.

Francisco Cruz and Teresa Ureña from Centro de Desarrollo Curricular and the Spanish department from Pontificia Universidad Católica Madre y Maestra (PUCMM) for providing me with the help and the facilities to collect my data in the Dominican Republic. Les agradezco de corazón por el gran apoyo que me concedieron.

Special thanks to Dr. Syed Rahman who has helped me with the statistical analysis.

Thanks for your knowledge, patience and good heart. Also, to my fellow graduate student

Carolyn Bradley who patiently edited my dissertation and always had words of encouragement during this arduous process.

I do not have words to express how thankful I am to have shared this journey with Osmer

Balam, thanks for your unconditional support and incessant help with all my projects. I also want to deeply thank Francesc Morales for being always there for anything I needed, and Clara Sotelo and Kathy Navajas for their constant care and love.

My immense appreciation for those graduate fellows who substituted my classes when I needed it most. I cannot forget to thank Tania Fleming, the office manager at the Spanish and

Portuguese department, for her extraordinary work and just for being always willing to help with anything. Also, I want to thank my favorite teaching coordinators Victor Jordán, Su Ar Lee,

Susana Braylan and Clara Sotelo for being outstanding mentors. Besides, I want to thank

Jennifer Wooten for her guidance, especially during the job search process.

I want to acknowledge the effort and care that all Gators for Equal Opportunity family invested in our community project, in Alachua city and Gainesville. I want to thank everyone who gave their time and helped to make a difference: Jaime and Maria Eugenia Zelaya, Lyndsey

Wright, Zhen XU, Yiyan Zeng, Wanqi Liu, Shengrui Khan, Archangela Dias, Josefina Almanza,

Matthew Morse, Sally Leinthal-Rist, Emmalyn Deak, Katherine Barko-Alva, Dr. Wooten, Dr.

Coady and Dr. de Jong. Also, our amazing faculty advisors Kathy Navajas and Clara Sotelo.

Because my Cronhs´s disease has been part of my Ph.D. journey, I want to thank the remarkable individuals that were by my side in the worst times. I am immensely thankful for

Michael Claudio, Osmer Balam, Kathy Navajas, Elisabet Liminyana, Ana de Prada Pérez, Clara

Sotelo, Alberto Escudero and Jaime and Maria Eugenia Zelaya. Also, my fellow graduate students, co-workers, professor and friends who showed me all the love when I most needed it:

Nehemie Joseph, Filipe Da Silva, Gerlanda Valencia, Enrique Ibañez, Yanina Becco, Carlos

Bertoglio, Andrea Villa, Antonio Cardentey, Dr. Lord, Dr. Wind Cowles, Souad Kehder, Corina

Liu, Layla Mieses, Henry Robinson, and Ana Hellmuth. Furthermore, I need to thank Dr. Padilla and her medical team at the infirmary, and Dr. Glover, my GI doctor, who was always available for me. I also want to thank the Disability Resource Center at UF and Jim Faubel for the help and guidance.

I am immensely grateful for all the people who directly or indirectly made this possible, such as my dear friends in Gainesville, Jacksonville, Spain or any other parts of the world:

Alberto Escudero, Filipe Da Silva, Elisabet Liminyana, Jimena Ojeda, Mercedes Tejera, Caridad

Jiménez, Corina Liu, Leandro Santos, Anna Helmuth, Layla Mieses, Henry Robinson, Amy and

Ruben Laureano, Kristen Hughes, Anum Tariq, Lydia Burnao, Marisa Delgado, Paula Gonzalez,

Veronica Davila, Viviana Davila, Olga Ellis, Antonio Cardentey. Also, all my friends back in

Illinois, particularly, Marisol Garrido, Lauren Hetrovicz, Antonio Perez Nuñez and Karima

Belghiti who supported and have been supporting me in every step, not only academically but also personally.

I cannot be grateful enough for my dearest family, mis padres, Enrique Mayans Lacruz y

Ángeles Ramón Cascales, mi hermano, Arturo, mi tía, Nati, mis tíos Juanma y Manu y mis primos, Gabriel y Nereida. Mis abuelos Paco y Carmen. Mi familia postiza Margarita Espino,

Ignacio Balsalobre, Sandra y Tamara. Especialmente agradecer a mis padres y a mi hermano por lidiar conmigo en los peores momentos y también por apoyarme siempre y en cada decisión.

También, quiero agradecer de corazón a mi tío Juanma por siempre creer en mí.

Last but not less, I want to thank with all my heart to Kenneth Harris for being my beloved partner during all this process. Thanks for your support, love and patience during this

Ph.D. craziness and, more importantly, thanks for always making me laugh.

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 10

LIST OF FIGURES ...... 12

LIST OF ABBREVIATIONS ...... 13

ABSTRACT ...... 16

CHAPTER

1 INTRODUCTION ...... 18

1.1 Heritage Speakers and Heritage Languages ...... 18 1.2 Grammatical gender ...... 20 1.3 Overview of Research Questions and Hypothesis ...... 22 1.4 Dissertation structure ...... 26

2 LITERATURE REVIEW ...... 27

2.1 Heritage Speakers of Spanish ...... 27 2.3 Gender Assignment and Gender Agreement ...... 34 2.4 Gender in Psycholinguistic Models of Lexical Access and Word Production ...... 41 2.5 Studies on Grammatical Gender ...... 52 2.6 Gender Agreement with Noun Morphology ...... 52 2.6.1 Native Speakers ...... 52 2.6.2 Bilingual Speakers ...... 61 2.7 Gender Processing with Noun Frequency ...... 65 2.7.1 Native Speakers ...... 65 2.7.2 In Bilingual Speakers ...... 66 Frequency effects in retrieval processing: Frequency-Lag Hypothesis ...... 66

3 RESEARCH QUESTIONS AND METHODOLOGY ...... 71

3.1 Current Study and Research Questions ...... 71 3.1.1 Effect of Noun Morphology in Gender Access in Monolingual and HSs of Spanish ...... 71 3.1.2 Frequency Effects on Gender Access in Monolingual and HSs of Spanish ...... 76 3.2 Methods ...... 78 3.2.1 LEAP-Q and Language Background Questionnaire ...... 78 3.2.2 Standard Proficiency Measures: DELE, MELICET and MINT ...... 79 3.2.3 Vocabulary Test Experiment 2 ...... 80 3.2.4. Participants ...... 80

3.3 Materials and Design ...... 85 3.4. Procedure ...... 88

4 EXPERIMENT 1 RESULTS: PICTURE NAMING ...... 91

4.1 RT Analysis ...... 93 4.1.1 Results: Monolingual, D+ and D- Bilinguals ...... 93 4.1.2 Results: Monolingual Group ...... 96 4.1.3 Results: D+ and D- Bilinguals ...... 97 4.2 Accuracy Analysis ...... 99 4.2.1 Results: Monolinguals and D+ and D- Bilingual Group ...... 100 4.2.2 Results: Monolingual Group ...... 101 4.2.3 Results: D+ and D- bilingual group...... 102 4.3 Summary of Results Experiment 1 ...... 103

5 EXPERIMENT 2 RESULTS: PICTURE DESCRIPTION ...... 105

5.1 RT Analysis ...... 107 5.1.1 Results: Monolingual, D+ and D- Bilingual Group ...... 107 5.1.2 Monolingual Group ...... 109 5.1.3 Results: D+ and D- Bilingual Group ...... 111 5.2 Accuracy Analysis ...... 113 5.2.1 Results: Monolingual, D+ and D- Group ...... 113 5.2.2 Results: Monolingual Group ...... 115 5.2.3 Results. D+ and D- bilingual group ...... 116 5.3 Summary of Results Experiment 2 ...... 119

6 DISCUSSION AND CONCLUSION ...... 121

6.2 Other Results ...... 130 6.2.1 Gender assignment and gender agreement ...... 130 6.2.2 Masculine default in HSs of Spanish ...... 131 6.3 Conclusion ...... 132 6.4 Implications and Future Research ...... 134 6.5 Contributions of Findings ...... 136

APPENDIX

A MONOLINGUAL PARTICIPANTS ...... 138

B BILINGUAL PARTICIPANTS ...... 139

C EXPERIMENTAL SEQUENCE SAMPLE ...... 140

D CONSENT FORM IN SPANISH ...... 141

E MONOLINGUAL SPEAKERS’ LBQ ...... 142

F BILINGUAL SPEAKERS’ LBQ ...... 144

G MELICET (Adapted) ...... 147

H DELE (Adapted) ...... 151

I LIST 1 AND 2 FILLERS WITH LEXICAL FREQUENCY ...... 155

J VOCABULARY TEST EXPERIMENT 2/STIMULI 1 ...... 157

K VOCABULARY TEST EXPERIMENT 2/STIMULI 2 ...... 158

LIST OF REFERENCES ...... 159

BIOGRAPHICAL SKETCH ...... 169

LIST OF TABLES

Table page

2-1 Word endings in Spanish ...... 33

3-1 Mean averages and p-values for MSs and HSs (both groups) results in the LEAP, LBQ and DELE...... 84

3-2 Mean averages and p-values for the two bilingual groups’ results in the LEAP, LBQ, MINT and MELICET ...... 84

3-3 Transparent and opaque, high and low frequency, feminine and masculine nouns used as experimental stimuli in List 1...... 86

3-4 Transparent and opaque, high and low frequency, feminine and masculine nouns used as experimental stimuli in List 2...... 87

3-5 Standard deviation, mean averages and P values of two tailed t-test comparing list 1 and list 2 of experimental items ...... 88

3-6 Standard deviation, mean averages and P values of two tailed t-test comparing List 1 and List 2 of filler items ...... 88

4-1 Experiment 1 mean RTs in ms. and SE split by frequency and morphology ...... 94

4-2 Experiment 1 three-groups 2 x 2 x 3 repeated-measures ANOVA output ...... 95

4-3 Experiment 1 monolinguals 2 x 2 repeated-measures ANOVA output ...... 97

4-4 Experiment 1 bilingual group comparison 2 x 2 x 2 repeated-measures ANOVA output ...... 98

4-5 Experiment 1 D+ bilingual group 2 x 2 repeated-measures ANOVA output ...... 99

4-6 Experiment 1 D- bilingual group 2 x 2 repeated-measures ANOVA output ...... 99

4-7 Experiment 1 three-groups 2 x 2 x 3 repeated-measures ANOVA output ...... 101

4-8 Experiment 1 monolinguals 2 x 2 repeated-measures ANOVA output ...... 102

4-9 Experiment 1 bilingual group comparison 2 x 2 x 2 repeated-measures ANOVA output ...... 102

4-10 Experiment 1 D+ bilingual group 2 x 2 repeated-measures ANOVA output ...... 103

4-11 Experiment 1 D- bilingual group 2 x 2 repeated-measures ANOVA output ...... 103

5-1 Experiment 2 mean picture naming times and SE split by frequency and morphology ..108

5-2 Experiment 2 three-groups 2 x 2 x 3 repeated-measures ANOVA output ...... 109

5-3 Experiment 2 monolinguals 2 x 2 repeated-measures ANOVA output ...... 110

5-4 Experiment 2 monolinguals pairwise test ...... 110

5-5 Experiment 2 bilingual group comparison 2 x 2 x 2 repeated measures ANOVA output ...... 112

5-6 Experiment 2 D+ bilingual group 2 x 2 repeated-measures ANOVA output ...... 112

5-7 Experiment 2 D- bilingual group 2 x 2 repeated-measures ANOVA output ...... 113

5-8 Experiment 2 three-groups 2 x 2 x 3 repeated measures ANOVA output ...... 115

5-9 Experiment 2 monolinguals 2x 2 repeated measures ANOVA output ...... 116

5-10 Experiment 2 bilingual group comparison 2 x 2 x 2 repeated measures ANOVA output ...... 117

5-11 Experiment 2 D+ bilingual group 2 x 2 repeated-measures ANOVA output ...... 117

5-12 Experiment 2 D- bilingual group 2 x 2 repeated-measures ANOVA output ...... 117

5-13 Experiment 2 pairwise comparison for D+ bilinguals ...... 118

5-14 Experiment 2 pairwise comparison for D- bilinguals ...... 118

6-1 Summary of theories, predictions and findings ...... 122

LIST OF FIGURES

Figure page

2-1 Bilingual continuum...... 29

2-2 Stages in lexical retrieval ...... 42

2-3 Example of word access in the Weaver ++model of lexical retrieval ...... 45

2-4 Lexical access networks (A) representation of models that claim a level between conceptual and word form information (Weaver++) (B) models that proposed two levels of representation (IN) ...... 48

2-5 (A) Schematic model and (B) the processing of Spanish words mesa [table], silla [chair] and tigre [tiger] according to the Independent Network model (cascade view) ....50

4-1 Experiment 1 mean RTs and SE split by frequency and morphology ...... 95

4-2 Experiment 1 mean accuracy and SE split by frequency and morphology ...... 100

5-1 Experiment 2 mean RTs and SE split by frequency and morphology ...... 108

5-2 Experiment 2 mean accuracy and SE split by frequency and morphology ...... 114

LIST OF ABBREVIATIONS

ADJ Adjective

CJ Conjunction

D- Less dominant bilinguals

D+ More dominant bilinguals

DET Determiner

EPT Elicited production task

ERP Event related potentials

FEM Feminine

Freq. Frequency

Ges Generalized Eta-Squared

H High frequency noun

HL Heritage Language

HO High Frequency Opaque

HS1 1st generation Heritage Speakers

HS2 2nd generation Heritage Speakers

HSs Heritage Speakers

HT High Frequency Transparent Nouns

LBQ Language Background questionnaire

LF Low frequency nouns

L2 Second language

L2ers L2 learners

L2ers Second language learners

LH Left hemisphere

LO Low Frequency Opaque Nouns

LT Low Frequency Transparent Nouns

LVF Left visual field

MASC Masculine

Mono Monolinguals

Morph. Morphology

Morph. Morphology ms. Milliseconds

MSs Monolingual speakers

N Noun

O Opaque nouns

PART Participle

PL Plural

PNT Picture naming task

PP Preposition

RH Right hemisphere

RTs Reaction times

RVF Right visual field

SE Standard Error

SG Singular

SMs Simultaneous bilinguals

SQs Sequential bilinguals

ST Standard Error

T Transparent nouns

TOT Tip of The Tongue

V Verb

VHF Visual half field

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

THE PROCESSING OF GENDER AGREEMENT IN MONOLINGUAL AND HERITAGE SPEAKERS OF SPANISH By

Dámaris Mayans Ramón

December 2018

Chair: Jorge Valdés Kroff Cochair: Ana de Prada Pérez Major: Romance Languages

This dissertation examines the impact of lexical frequency and noun morphology on agreement processing in monolingual and heritage speakers of Spanish (HSs). To date, there are mixed findings on the role of noun morphology in gender agreement among monolinguals and HSs.

What remains inconclusive is whether noun morphological endings are employed as useful cues that facilitate gender agreement during speech production and comprehension. With regard to lexical frequency, previous research on the role of lexical frequency has provided evidence of frequency effects when accessing nouns, and this effect has proven to be more prominent in bilingual populations. The current research expands on the antecedent research on lexical access by examining this lexical effect in gender agreement processing in two bilingual populations that differ in relation to their exposure to Spanish input and in their language dominance in Spanish. Findings in the present dissertation contribute to the understanding of different processing mechanisms in two adult bilingual populations and to our knowledge of the overall role of noun lexical frequency and morphology in gender agreement processes in both monolingual and bilingual populations

We analyzed data from a total of 86 participants (34 monolingual speakers and 44 HSs), who completed picture naming task (PNT) and picture description task (PDT). Experiment 1 was a PNT where participants named in Spanish what appeared on the screen with the corresponding article

while speakers ‘reaction times (RTs) were recorded and measured. Experiment 2 was a PDT where participants described what appeared on the screen using the adjective claro “clear” with its corresponding gender morphological ending in Spanish. RTs were also recorded and measured in the

PDT. We predicted that exposure to Spanish would play an influential role in the access of gender, therefore, HSs with less exposure to Spanish would show the slowest naming times of all groups. In line with previous studies on monolingual speakers (Navarrete et al., 2007), findings revealed an overall effect of noun frequency among monolinguals in both agreement operations. Both bilingual participant groups had a greater frequency effect than the bilingual group. Furthermore, the less

Spanish-dominant group showed the slowest naming times, in line with the predictions of the

Frequency-Lag Hypothesis (Gollan, Slattery and Rayner, 2012; Emmory, Petrich and Gollan, 2012).

With regard to morphology, transparent morphology seemed to have a facilitatory effect, particularly among monolingual speakers. Within the bilingual group, we found no morphology effects on both experiments’ RTs analyses. There was, however, an effect of morphological endings in the accuracy rate of both experiments, specifically , a facilitatory effect of transparency, consonant with with other studies with HSs (Alarcón, 2011; Montrul, 2013; Montrul, 2008).The morphology and frequency main effect in Experiment 1 and the interaction between frequency and morphology found in

Experiment 2 strongly suggest that noun transparency has a facilitatory effect in the gender agreement process. This corroborates the predictions of the Two-Route Hypothesis proposed by

Gollan and Frost (2001) and previous psycholinguistic studies that have found a morphological and lexical route to gender access (Caffarra et al., 2014).

CHAPTER 1 INTRODUCTION

1.1 Heritage Speakers and Heritage Languages

Unlike other parts of the world, such as Africa or Asia, where being bilingual is the norm, in North America, around 35% of the population speaks more than one language. In the United

States the percentage is lower, around 18%-20% of the population (Grosjean, 2007). In the USA,

Spanish is the language with most minority speakers due to an increment of Hispanic immigrants in the last decade (from 35.3 million to 50.5 million, United States Census, 2010). For this reason, in the past decade the amount of research on Spanish in the United States, a heritage language (HL), has also increased (Field 2011; Beaudrie and Fairclough, 2012). According to

Guadalupe Valdés (2000), Heritage speakers (HSs) are ‘individuals raised in homes where a language other than English is spoken and who are to some degree bilingual in English and the heritage language’ (381).

There is a common perception that growing up bilingual or being bilingual means that the person can speak both languages with the same level of fluency, but this is far from being the reality. As stated by Grosjean (2012) “Bilinguals usually acquire and use their language for different purposes, in different domains of life, with different people. Different aspects of life often require different languages” (12). This is called the Complementary Principle (Grosjean,

2012)

HSs and HLs differ in many ways from their monolingual norm. One of the factors

(among others), studied in the present investigation, that accounts for this difference between adult bilingualism and the monolingual variety is the amount of exposure or input that speakers have from the HL. Yip and Matthew (2007) refer to this as the logical problem of bilingual acquisition. They pose that the problem of the input is both quantitative and qualitative.

According to Grosjean (2007), the more domains in which a bilingual uses the HL, the more frequency they use the HL and are exposed to the HL, which will lead to greater fluency in the bilingual. However, when the HL is only used in very few domains, it results in less frequency of use and exposure to the HL resulting in a less proficient bilingual. However, the domains in which HSs can be exposed to the HL tend to decrease according to the generation they are born in, bilingualism in their community, education and governmental support of the HL. With regard to the generation they are born in, bilinguals born in the US from immigrant parents (second generation HSs) tend to have more monolingual input from their parents and/or grandparents.

This situation might differ when bilinguals are born from other bilingual speakers of Spanish

(third generation HSs) since the shift of language use, from the HL to the societal language, might have occurred in this generation, thus reducing the number of domains or exposure to the

HL.

The study of HSs and HLs is of growing importance in the last decade because it provides evidence on how children acquire and develop their HL grammars from childhood to adulthood. More importantly, the study of HSs’ bilingualism provides us with important information on how language is processed when the input in both languages is reduced, in comparison to other bilingual contexts where both languages are equally functional in the society. Also, in general, it allows us to understand how humans acquire multiple languages

(Yip, 2007). Particularly, previous research studying HSs of Spanish has focused on providing evidence of the different outcomes of adult heritage languages. Most research has tried to account for whether some structures have or have not been acquired by HSs leading to some unfortunate labelling such as “Incomplete Acquisition” (Montrul 2008) to explain non- monolingual-like outcomes. The present research uses psycholinguistic methodology to help

determine the source of cognitive difficulty that might cause difficulty with grammatical gender processing in Heritage Speakers of Spanish. We are testing two different groups of bilingual HSs of Spanish born in the United States (whose parents or grandparents immigrated to the USA) who had different exposure to Spanish while growing up. With regard to the linguistic variables under investigation, we are studying lexical frequency and morphophonological gender cues in nouns through grammatical gender agreement in two agreement operations. Our investigation focuses particularly on grammatical gender agreement since it is present in every noun and every agreeing element (e.g., determiners and adjectives) must demonstrate gender agreement in

Spanish; thus, it is a very important linguistic feature for both comprehension and production.

1.2 Grammatical gender

Grammatical gender is a morphological property of nouns in many languages. In Spanish, nouns are classified as feminine or masculine and proper gender selection is important for the entire sentence. In some cases, gender can be intrinsic, that is, denoting the biological properties of the referent, but grammatical gender is mostly arbitrary, that is, the gender of the noun is not derived from the biological properties of the noun. The grammatical gender of Spanish is highly predicted by its noun morphology. Most feminine nouns in Spanish end in -a and most masculine nouns in Spanish end in -o. For this reason, it is considered that Spanish gender marking morphology is very transparent, in other words, the gender of the noun can be deducted from its ending. However, there are irregular nouns in Spanish, such as nouns ending in -a that are masculine (such as el mapa “the map”) and nouns ending in -o that are masculine (such as la mano, “the hand”). Furthermore, there are opaque nouns whose gender cannot be predicted from its sublexical properties or noun endings (e.g. nouns ending in -z, -e, -l, -n).

Agreement refers to a type of relation between syntactic elements that have some sort of dependency among each other, or as Steel (1978) states “the term agreement commonly refers to

some systematic covariance between a semantic or formal property of one element and a formal property of another. For example, adjectives may take some formal indication of the number and gender of the noun they modify.” (610).

Grammatical gender has an important communicative role. It helps to disambiguate anaphoric reference and other deictic markers, maintaining a cohesion and helping to process discourse (Urrutia, Domínguez and Álvarez, 2009). Although grammatical gender is an arbitrary grammatical (i.e., lacking semantic) feature, research has also demonstrated an anticipatory (i.e., facilitatory) effect in processing since it helps to predict upcoming nouns (i.e. the gender congruency effect: Van Berkum, 1996; Van Berkum, Brown, Zwitserlood, Kooijman and

Hagoort, 2005).

It is uncommon that native speakers of Spanish make gender agreement mismatches

(Eberhard et al., 2005); however, agreement mismatches are more common in the production of

Spanish HSs (Martinez-Gibson, 2011; Montrul and Potowski, 2007). Also, second language learners demonstrate persistent difficulty with gender agreement even at advanced proficiency levels (Carrol, 1989; Dewaele and Veronique, 2001; Montrul et al., 2008).

Most research on HL acquisition has tried to provide evidence of the different outcomes of adult heritage languages, that is, it has tried to account for whether some structures have or have not been acquired by HSs or whether the HL has been forgotten, a process also known as

“attrition” or “language loss” (Polinsky 2006,2011; Schmid, Köpke, Keijzer and Weilema,

2004). The motivation of the present study is to provide evidence about the factors that might affect the grammatical processing of gender agreement, linguistic factors such as lexical frequency and noun transparency (opaque and transparent nouns) and sociolinguistic or non- linguistic factors such as exposure to the HL. The present investigation would also provide

further evidence on the role of noun lexical frequency and noun morphology in monolinguals speakers of Spanish.

The following section will provide a brief outline of the present study research questions and hypothesis.

1.3 Overview of Research Questions and Hypothesis

The research questions addressed in the present investigation and their hypotheses are briefly summarized here:

RQ (1): What is the role of noun morphology in gender agreement processing?

Based on previous research, the impact of the morphological cues on gender retrieval and lexical access can have three outcomes.

a) Participants would be expected to retrieve gender faster with transparent nouns than

with opaque nouns, showing that gender is more easily retrieved with the help of

morphological cues. This will support the reliable cue hypothesis (Bates, Devescovi,

Hernandez, Pizzamiglio, 1996; Taft and Meunier, 1998).

b) The second possible outcome is that gender will be retrieved without the help of noun

morphological endings via a lexico-syntactic route. We would expect that gender

would be retrieved faster in high frequency words regardless of noun morphological

endings. These possible findings will support a lexico-syntactic access view in line

with other research that found that gender can be retrieved without the help of

morphophonological cues (Caramazza and Miozzo, 1997; Miozzo and Caramazza,

1997; Vigliocco, Antonini and Garrett,1997; Badecker, Miozzo and Zanuttini, 1995;

Costa, Miozzo and Caramazza, 1999; Gonon, Bruckert. and Michel, 1989; Kulke and

Blanken, 2001)

c) A third alternative outcome, which may be partially compatible with the two previous

views, is that the retrieval of gender features will depend on both morphological cues

and lexical information, assuming a dual route model, or as Gollan and Frost (2001)

named it, the Two-Route Hypothesis. If we assume that both morphological and

lexical cues play a role in retrieving gender, we will expect that the gender of high

frequency nouns will be easily accessed, but also, a second morphological route will

play a role in the retrieval of gender, therefore high frequency transparent nouns will

be retrieved faster than high frequency opaque nouns (as observed in Alarcón, 2009,

Resende and Mota, 2017) and low frequency transparent nouns will be retrieved

faster than low frequency opaque nouns, since the morphological cue will help in the

retrieval, corroborating the two-route hypothesis.

There is another possible outcome, but it will be discussed under another Dual Mechanism

Model that contemplates two different mechanisms to process irregular and regular nouns in RQ

(2).

RQ (2) Are the morphology effects similar to other dual-route models?

Based on the view that participants use two separate mental mechanisms to compute regular and irregular nouns -transparent and opaque- (Pinker, 1991, 1999; Bybee, 1985, 1995;

Jackendoff, 2002) we will expect that participants will be faster with high frequency opaque nouns than high frequency transparent nouns because the former, not having regular endings, will be stored in memory and not computed, therefore, opaque high frequency nouns will have an advantage. With low frequency nouns, we will observe the opposite pattern, low frequency opaque nouns would need to be look up in memory, while low frequency transparent nouns will be benefited by being computed.

In terms of accuracy, we will expect that participants will be more accurate with high frequency opaque nouns than high frequency transparent nouns. Also, they will be more accurate with low frequency transparent than low frequency opaque (Birdsong & Flege, 2001). Note that the Two-Route Hypothesis (Gollan and Frost, 2001) denotes retrieving both the morphophonological and lexico-syntactic (gender) information in transparent nouns while with opaque nouns speakers will only make use of a single lexical route. The Dual Mechanism Model assumes that speakers retrieved regular nouns (transparent) applying a rule-based mechanism, while irregular nouns (opaque) are stored in memory.

RQ (3): Do different linguistic populations differ in the use of noun morphology when accessing gender?

a) There is no research using online techniques (measure speakers responses via RTs,

brain activity or eye movement, a.o.) that studies noun morphology in grammatical

gender processing in monolinguals and two different groups of HSs of Spanish. Our

predictions are therefore based on previously mentioned studies on the role of

morphology in monolingual speakers (further review in section 2.6.1) and other off-

line studies on oral production that compared monolinguals, L2 learners and HSs´

accuracy responses (Montrul et al., 2013; Montrul et al., 2008; Alarcón, 2011). On

the one hand, most studies comparing these populations do not report the results of

the monolingual group since they focus on the differences between HSs and L2

learners (Montrul, 2013; Montrul, 2008; Alarcón, 2011). According to off-line studies

that found accuracy effects of morphological ending in HSs (Montrul et al, 2013;

Montrul et al, 2008; Alarcón, 2011), we hypothesize that HSs in our experiment will

behave similarly and we will see effects of noun morphology in both accuracy and

RTs results.

b) Another possible outcome is based on the studies and view that support a separate

cognitive mechanism to process regular and irregular nouns (in our investigation

opaque and transparent nouns). Under this view, the Dual Mechanism Model, we will

see that high frequency will play a role with the irregular forms, in our study opaque

nouns, but not transparent nouns. Low frequency will benefit the regular morphology,

-o and -a endings, since these forms will be used without the need to be looked up.

RQ (4): Does noun frequency impact the retrieval of gender features in different linguistic populations?

We hypothesize that, similar to previous research with monolingual speakers (Navarrete,

Basagni, Alario, and Costa, 2006), we will find a frequency effect in the retrieval of gender features in the two agreement operations under investigation, determiner-noun and noun- adjective agreement.

Studies comparing bilingual and monolingual speakers, such as the one proposing the

Frequency-Lag Hypothesis (Gollan, Slattery and Rayner, 2012; Emmory, Petrich and Gollan,

2012), hypothesizes that noun frequency will show a greater effect on gender retrieval in bilinguals than monolinguals. Especially less Spanish-dominant bilinguals will show the largest difference between high frequency and low frequency nouns.

In order to study these variables, we carried out two experiments. The first experiment was a picture naming task (PNT) where participants had to name in Spanish what appeared on the screen using the correct determiner “el” or “la” (the). Experiment 2 consisted of a picture description task (PDT) where participants had to describe what appeared on the screen using the

adjective “claro” or “claro” (claro) agreeing with the noun referent. The stimuli consisted of low frequency and high frequency Spanish nouns with different noun morphology (transparent and opaque). First, we manipulated the noun morphological endings to study what is the role of noun morphological endings in two agreement operations, determiner-noun agreement and noun- adjective agreement, in monolingual speakers of Spanish and HSs. Second, we manipulate the noun lexical frequency, high frequency and low frequency nouns, to observe the frequency effects in monolinguals and HSs.

Following, a summary of how the dissertation will be organized.

1.4 Dissertation structure

The present investigation seeks to answer the questions outlined in the previous section and provide further evidence on the different outcomes of adult bilingualism in Heritage

Language Acquisition (HLA) where the exposure to the HL varies considerably among speakers.

Furthermore, it will also contribute to the scarce literature on monolinguals and HSs of Spanish and the role of morphosyntactic cues (or morphological endings) using psycholinguistic methodology.

The rest of this dissertation is organized in the following manner. Chapter 2 will further explain the notion of HS and HL as well as an in-depth explanation of gender agreement in

Spanish. It will also introduce previous research on lexical frequency and noun morphology in monolingual and bilingual speakers (particularly HSs of Spanish). This chapter will also provide information on the current models of lexical access in oral production and the hypothesis under investigation. In chapter 3, the research questions and hypotheses are provided followed by the research methodology. In chapter 4, we will provide the results for Experiment 1 and, subsequently, chapter 5 will introduce the Experiment 2 results. Finally, chapter 6 will present the discussion and conclusions of the findings.

CHAPTER 2 LITERATURE REVIEW

This chapter provides a descriptive and theoretical overview of grammatical gender in

Spanish, the focus of the present dissertation. Since there are different types of bilingual speakers, it is necessary to start the chapter with a description of the bilingual population that will participate in this study, Spanish-English bilinguals who are heritage speakers of Spanish. In section 2.2, I provide information on gender in Spanish. In section 2.3, I explain the differences between gender assignment and gender agreement and, in section 2.4, I provide an overview on gender under psycholinguistic models of lexical access, primarily, (i) the Weaver ++ model and

(ii) the Independent Network model (IN) and their predictions for the current investigation.

Subsequently, in section 2.5, I present the studies carried out on gender agreement. In section

2.6, I present research done on agreement with a focus on noun morphology (using offline and online methods) and section 2.7, research done on noun frequency and lexical frequency effects modulated by bilingual dominance. This section will also explain the hypotheses that are under investigation in this study. Namely, for the morphophonological effect: the reliable cue hypothesis (Bates, Devescovi, Hernandez and Pizzamiglio 1996; Taft and Meunier, 1998) , the lexico syntactic access view (Badecker, Miozzo and Zanuttini,, 1995; Miozzo and Caramazza,

1997; Vigliocco, Antonini, and Garrett, 1997), and the Two-Route Hypothesis (Gollan and Frost,

2001; Caffarra, Janssen and Barber, 2010) and, for the lexical frequency effect on Bilinguals

Heritage Speakers of Spanish: the Frequency-Lag Hypothesis (Gollan, Montoya, Cera, and

Sandoval, 2008).

2.1 Heritage Speakers of Spanish

The present study will recruit monolingual speakers of Spanish (MSs) from the

Dominican Republic and second and third generation Spanish-English heritage speakers (HSs) in

the US (as defined by Silva-Corvalán, 1994, see below for further explanation) with differing levels of exposure to their heritage language. The term heritage language refers to the languages spoken by recent immigrants and their children. From a social perspective, and specifically in the

United States, Spanish is primarily a heritage and minority language while English is spoken by the wider speech community (Montrul, 2004). In the United States, the term heritage language refers to all non-English languages, including those spoken by Native Americans. Heritage

Speakers (HSs) are bilinguals who grew up in families which speak a different language than the dominant language of the mainstream society they live in (Valdes, 2005). In the few years before they begin school, HSs usually receive much, if not exclusive, input of the heritage language at home, from their parents (if they are newly arrived in the States), closer relatives, and parents’ friends and acquaintances. After starting school, HSs become more fully exposed to the dominant language of the mainstream society, therefore, their exposure to their languages changes, resulting in shifting language dominance to the second, exposed language (Valdes,

2005; Montrul, 2012). According to Montrul (2012), HSs can be divided between simultaneous bilinguals (SMs) or sequential bilinguals (SQs). SMs are bilingual children who grew up being exposed to both the heritage language and the majority language and thus acquired both languages at the same time. Meanwhile, SQs are bilingual children who were mainly exposed to the heritage language and acquired it first, then started school and acquired the majority language.

For the present investigation, I recruited young adult SM and SQ Spanish-English bilinguals born in the United States that exhibit different degrees of language dominance: (1) a group of bilinguals who are highly exposed to and use the heritage language frequently on a daily basis, i.e. a more Spanish-dominant bilingual group and (2) a group of bilinguals who are

not highly exposed to and less frequently use the heritage language on a daily basis, i.e. a less

Spanish-dominant bilingual group. The difference in language dominance in Spanish represents frequency of use (percentage of usage of their languages on a daily basis) of the bilingual speaker’s languages. Since lexical frequency will be manipulated in the present experimental study, effects of bilingual language dominance will be expected when retrieving words that are more or less frequent. (See below for more details on frequency effects.)

Researchers who study bilingualism suggest that there are many different types of bilinguals and that bilingual abilities fall along a continuum (Valdes, 2005). This continuum is represented in Figure 2-1 below where upper and lower-case letters represent competence in the bilingual speaker's languages, as opposed to an idealized and balanced bilingualism where both languages would be represented with capitalized letters representing equal proficiency in both languages.

Figure 2-1. Bilingual continuum, adapted from Valdes (2014: 5)

Bilingualism is best seen as a dynamic condition, since over a lifespan, a bilingual’s language dominance can vary immensely, depending on background experiences and schooling.

Regarding the generation of HSs, according to Silva-Corvalán (1994), first generation

HSs are those who immigrated to the United States sometime during their childhood, while

second generation will be their children, and third generation will be their grandchildren.

However, as mentioned above, HSs will vary among their own generation and personal experience. Some immigrants will remain functionally monolingual in their first language throughout their lives, while others will acquire some English, and still others will become highly proficient bilinguals who will remain more dominant in the first language. By the second and third generation, most members of the immigrant community will have acquired English quite well. The majority of these individuals will likely be English dominant, or at a minimum, they will prefer English in their daily interactions. Many, nevertheless, will continue to function in two languages to communicate with members of the first generation. Finally, in general, by the fourth generation, most individuals from an immigrant background will have become monolingual English speakers in the context of the USA.

HSs of Spanish are not a homogenous group since they differ in their age of arrival to the

United States, the early acquisition of both of their languages, differing amounts of exposure while they were growing up, how much they use these languages during their lives, and so on.

The study of their bilingualism provides us with important information on how language is processed when the input in both languages is reduced, in comparison to other bilingual contexts where both languages are equally functional in the society (diglossic/non-diglossic context) or monolingual situations. Another important reason to study HSs, specifically those from second and third generation, like the participants in the present investigation, is the fact that, in the

United States, approximately 64% of the total Spanish heritage population is USA born (USA.

Census Bureau, 2010, as cited in Cuza and Perez-Tattam, 2015), so HSs represent the type of bilingualism of the majority of the Hispanic population in the USA.

Lastly, Pascual y Cabo (2013) does not consider heritage speakers those individuals who, having been born and grown up in a heritage speaker environment, lack any linguistic competence in the heritage language, i.e., speakers who have a more passive knowledge of

Spanish. However, since these speakers are unable to effectively “speak” the language, it is unsuitable to label them as heritage “speakers” as the term suggests. For this reason, the present study recruited only HSs who were linguistically able to effectively communicate in Spanish.

2.2. Grammatical Gender in Spanish

Grammatical gender (henceforth “gender”) is an inherent morpho-syntactic feature that arbitrarily classifies nouns in many languages (e.g., German, Dutch, French, Portuguese, Greek,

Arabic, etc.), among them, Spanish (Carroll, 1989). Additionally, Spanish has richer noun inflectional morphology than English.

In the Determiner Phrase1 (DP), that is, noun phrases headed by a determiner (Abney,

1987), English only marks for number. In Spanish, DPs often mark for gender and number (but not always, e.g. su-his/her/their casa-house), as examples (1) and (2) illustrate.

(1) El gato gordo The-DET.MASC.SG. cat-N.MASC.SG. fat-ADJ.MASC.SG. “The fat cat” (2) Los gatos gordos The-DET.MASC.PL. cat-N.MASC.PL. fat-ADJ.MASC.PL. “The fat cats”

According to generative syntax, there are two different types of gender specification, one semantic and one grammatical (Antón-Méndez, Nicol and Garrett, 2002). Grammatical gender, unlike semantic gender, is a lexico-syntactic property of nouns (Biran and Friedman, 2012) that

1The determiner + noun construction may go by different names such as noun phrase or determiner phrase across linguistic frameworks. This dissertation uses terminology from generative Chomskyan syntax, and will use Determiner Phrase (DP) to refer to this construction as introduced by generative syntactician, Abney (1987).

is determined in an arbitrary way; that is, the referent itself does not inherently determine its gender (Montrul, 2004). The majority of nouns in Spanish have no relation between their grammatical gender and any concept of feminine or masculine properties in the real world (e.g., la mesa-N.FEM “the table”, Harris, 1991). The class of semantic gender nouns includes a small set of animate nouns that have masculine and feminine forms corresponding to their biological gender:

(3) Dueñ-o Owner-N.MASC.SG Dueñ-a Owner-N.FEM.SG

However, there are some exceptions, such as el/la víctima (“victim”) or el/la atleta

(“athlete”) whose overt morphology is invariant but which demonstrate gender agreement on determiners and adjectives (Antón-Méndez, Nicol, and Garret, 2002). In some languages, such as

German, the gender of the noun is difficult to predict only from its phonological form or overt morphology, that is, its noun ending. Spanish has a binary system in which one of the two genders is assigned to all nouns (Lloret and Viaplana, 1998; and Alarcos, 1999). Unlike German,

Spanish noun morphology is mainly transparent for gender as most masculine nouns end in –o and most feminine nouns end in –a (Green, 1988), representing 68.15% of all nouns (O´Rourke and Van Petten, 2011). Most analyses assume that these nouns are morphologically marked for gender and are called overt gender nouns or transparent nouns. The rest of the nouns have ambiguous morphology and are called non-overt gender nouns or opaque nouns (Alarcón, 2011).

However, this last statement is debatable, since as we see in Table 2-1 that illustrates the distribution of gender across different word endings, there are other noun endings (apart from –a and –o) with a high correlation with feminine, e.g., –ión, -dad, or with masculine, e.g., -r.

Table 2-1. Word endings in Spanish Ending type count feminine % masculine% o 3206 0.2 99.8 a 3157 95.3 4.7 ión 1017 98.3 1.7 e 674 16.6 83.4 r 488 1.2 98.8 d 488 97.1 2.9 n 488 3.9 96.1 l 220 8.2 91.8 s 176 35.8 64.2 z 102 75.5 24.5 t 25 0.0 100.0 y 20 20.0 80.0 é 15 6.7 93.3 í 13 0.0 100.0 c 12 0.0 100.0 i 11 9.1 90.9 g 11 0.0 100.0 m 10 0.0 100.0 x 8 0.0 100.0 á 3 33.3 66.7 p 3 0.0 100.0 j 3 33.3 66.7 h 2 0.0 100.0 f 2 0.0 100.0 ú 2 0.0 100.0 ó 2 0.0 100.0 b 2 0.0 100.0 w 1 0.0 100.0 u 1 0.0 100.0 all nouns 10014 47.8 52.2 (Adapted from Ralph, Sage, Green, Berthier, Martinez Cuitin, Torralva, & Patterson et al., 2010)

The above table comes from an analysis of the LEXESP Spanish database (LEXESP -

Léxico informatizado del español, Sebastián et al., 2000) carried out by Ralph et al. (2010). This database has 5 million words from different texts such as narratives, scientific papers, essays, press, seminars, sporting press, etc. This database provides a variety of words and their frequency of occurrence. They exclusively selected from the database the nouns (approximately

20,000 types). As seen in the table, across the whole set of nouns in Spanish, masculine and

feminine nouns occur almost equally with a slightly larger proportion of masculine nouns (52.2% versus 47.8%). Another fact to point out is that the majority of the noun endings tend to favor one gender; that is, there is no morphological ending that equally favors masculine and feminine gender (Ralph et al., 2010). For example, Teschner and Russell (1984) conducted a statistical analysis showing that nouns with the morphological ending –ad were feminine 99.7% of the time. Therefore, they concluded that words like ciudad were transparent nouns because the noun gender could be predicted from its ending.

Another characteristic of Spanish is that determiners and adjectives must agree with the noun in local and non-adjacent agreement. In Spanish, words within a noun phrase as well as any adjectives or participles outside of the noun phrase that modify the head noun must carry gender features. Many adjectives (inside or outside the DP) are inflected with -o and -a for masculine and feminine gender respectively. Examples below illustrate agreement with adjacent and nonadjacent dependencies. Example (4) represents adjacent agreement inside the DP and the data in (5) illustrate non-adjacent agreement.

(4) Adjacent

Una casa grande cuesta mucho en Santiago

A-DET.FEM.SG. house-N.FEM.SG. big-ADJ.FEM.SG. costs-V.SG.3P. a lot-ADV. in-PREP Santiago-N. “A big house costs a lot in Santiago”

(5) Non-adjacent

La casa es bastante vieja y necesita reparaciones The-DET.FEM.SG. house-N.FEM.SG. is quite-ADV. old-ADJ.FEM.SG. and needs-V.SG.3P repairs-N.MASC.PL. “The house is quite old and needs lots of repairs”

2.3 Gender Assignment and Gender Agreement

White, Valenzuela, Kozlowska–Macgregor and Leung, (2004) and Valenzuela, Faure,

Ramírez-Trujillo, Barski, Pangtay, and Diez (2012) mention two operations of agreement, one within the DP called “concord,” and other operation involving verbs which refers to agreement.

Even though adjectives in copula constructions outside of the DP must agree with the head noun

(e.g., la ciudad es ruidosa “the city is noisy”), this operation is considered to be part of the verbal domain, thus, an agreement operation. In the present chapter, we will not further discuss these two different operations, as the present investigation focuses on concord, that is, agreement operations within the DP, i.e. determiner-noun and noun-adjective, known in the literature as gender assignment and gender agreement respectively (Audring, 2008). However, given that studies of grammatical gender have found differences between determiner-noun assignment and noun-adjective agreement within and outside the DP, in this section, we will mention some of these studies in monolinguals and bilinguals and also provide the rationale to use these two agreement operations in the present investigation.

There are key differences between gender assignment and gender agreement (despite both being agreement operations). Gender assignment occurs between the determiner - e.g., definite article el, la “the” or indefinite article un, una “a” - and the noun. Gender agreement, on the other hand, occurs between the noun and other lexical items such as adjectives - e.g., viejo/a

“old,” nuevo/a “new,” pequeño/a “small.” The two examples below exemplify gender assignment (6) and gender agreement (7):

(6) El coche The-DET.MASC.SG. car-N.MASC.SG. “The car” (7) Coche nuevo Car-N.MASC.SG new-ADJ.MASC.SG. “The new car”

Gender assignment is seen as an inherent lexical property of Spanish nouns (Audring, 2008;

Cuza and Perez-Tattam, 2015). Gender assignment as carried out through cues from determiners with overt/transparent morphology that occur in prenominal agreement in Spanish is thought to help the acquisition of nouns (Audring, 2008). According to Correa (2001), the gender of the

noun is identified from analyzing the determiner phrase (that is, determine+noun or determiner+adj+noun) and the processing of the agreement within that phrase. O’ Rourke and

Van Petten (2011) also argued that the determiner-noun agreement has a different status than other agreement sequences because the number of determiners with masculine or feminine overt morphology is smaller, e.g. el/la (the), un/una (a/an), este/esta, (this) otro/otra (other) (with the exception of possessives which are not inflected for gender). Therefore, they suggest that it is easier to determine the gender of a noun in a determiner-noun sequence because these two elements are combined very frequently (unlike postnominal adjective agreement where there is an endless number of noun-adjective combinations). Also, unlike adjectives, determiners always occur before the noun and they are mostly obligatory in the noun phrase.

As mentioned by Audring (2008), some studies on lexical gender are not concerned with noun-adjective agreement since they focus on the overt morphological manifestation of the assigned representation in the noun (e.g., article, noun morphological endings). However, studies have reported that successful gender assignment as marked by determiner-noun agreement does not guarantee successful agreement with other elements (Montrul and Potowski, 2007; Montrul et al., 2008; Martinez-Gibson, 2011). The following section provides details about studies and research done on determiner-noun assignment and noun-adjective agreement (within and outside the DP) in monolinguals and bilingual HSs. It is worth mentioning that most of the bilingual studies in the following section will include studies from the representational view; that is, these investigations examine whether gender is instantiated in the bilingual lexicon.

Studies on gender agreement have found a difference between native and non-native performance. Some scholars, such as Clahsen (1998) and Hawking and Chan (1997), have argued that the difference stems from a representational problem in second language learners.

These scholars have proposed that speakers lack certain functional features not present in their first language (e.g., Failed Functional Features Hypothesis). However, other scholars, such as

Prevost and White (2000) and Schwartz and Sprouse (1997), suggest that the problem lays in the mapping of uninterpretable features. Even though the present dissertation will not address this debate, the following section will summarize this literature so as to provide a complete picture of monolingual and bilingual (specifically HSs) competence with lexical agreement.

Previous work has shown that monolingual children learning Spanish demonstrate complete acquisition of gender by the age of 4 (Lleó, 1998; López Ornat, 1997). Noun-adjective gender agreement in Spanish appears during the child’s two-word stage, approximately at around

18-24 months. In the bilingual acquisition of languages in which one language has gender agreement and the other does not, previous research has shown that gender assignment

(determiner-noun) is mastered earlier than noun-adjective agreement (Montrul and Potowski,

2007). Research in oral production with second language (L2) learners of Spanish has shown that noun-adjective agreement is a source of persistent difficulty for L2ers (Finneman, 1992;

Fernandez Garcia, 1999) in comparison to determiner-noun assignment.

The following studies testing agreement in HSs provide some evidence of an overall difference between monolingual and HSs’ performance on gender agreement and also provide supporting evidence for the idea that gender assignment and gender agreement are independent phenomena (Cuza and Perez-Tattam, 2015; Montrul and Potowski, 2007; Montrul et al., 2008;

Martinez-Gibson, 2011; but see Montrul, Davidson, de la Fuente and Foote, 2013 for opposing results).

Montrul and Potowski (2007) tested a group of 29 monolingual children from Mexico, 39 heritage speakers divided in two groups, simultaneous bilingual (SMs) children (exposed to both

languages at birth) and sequential bilingual (SQs) children (exposed to Spanish first and later to

English), and 22 L2ers (children acquiring Spanish in a dual immersion school in Chicago) to see whether agreement is affected by the bilinguals’ background. They used an oral narrative task that consisted of retelling a story in the past and an elicited production task. The monolingual group was more accurate than SQs, SMs, and the L2ers, and the HSs were in turn statistically more accurate than L2ers. Performance on masculine agreement was better than performance on feminine agreement (e.g., *La-DET.FEM.SG casa-N.FEM.SG rojo-ADJ.FEM.SG. “The red house”).

Bilinguals’ error rate was approximately 5% with determiners (in the oral narrative task) and more than 30% with adjectives (much higher for the L2 group). Within the bilingual group, researchers reported that the results of the two tasks were slightly different. In the oral narrative task, SMs and SQs did not differ from each other. In the elicited production task that specifically assessed noun-adjective agreement, SQs showed more target-like performance and the SMs did not differ from the L2ers. Martinez-Gibson (2011) showed similar results in her research. She tested 16 Spanish-English HSs born in the USA, whose parents immigrated to the USA (HS1),

14 Spanish-English heritage bilinguals, born in the USA whose grandparents immigrated to the

USA (HS2), and 14 L2 Spanish learners. They performed an elicited production task and a 25- minute interview. She obtained a total of 287 tokens for adjective agreement. Gender mismatch was mostly a feminine noun with a masculine adjective (67.5% of total errors). L2 learners produced the most gender mismatch occurrences, and HS2 produced more gender mismatches than HS1. In a recent study using off-line tasks, Montrul, Foote and Perpiñán (2008) compared

HSs and L2ers. Participants were tested in an untimed sentence completion task (pencil and paper) and an elicited production task in which they heard an ongoing conversation and had to select among different pictures the right referent to complete the sentence. The production task

consisted of hearing a conversation, then the following screen presented an incomplete sentence and three different pictures (referents). Participants read the sentence on the screen (e.g., No quiero llevar las de ese color “I don’t want to wear the ones-FEM.PL.of that color”) and chose the correct referent among different pictures. Participants found more difficulties agreeing with opaque nouns as revealed by error rates. They extended the masculine form instead of the feminine (i.e., a masculine default2 strategy). They performed better with determiner-noun assignment than with noun-adjective agreement. The L2 group outperformed the HS group in the written test, but oral results corroborated that HSs performed better than L2ers. They attributed this difference to the fact that L2ers have a different level of metalinguistic awareness (classroom taught metalinguistic knowledge) that they bring to the task. Alarcón (2011) used a similar methodology than the one used by Montrul et al. (2008), however, in Alarcón´s oral task, participants had to orally produce the target object with the corresponding article and a descriptive adjective “veo un choche rojo” (I see a red car). Similarly, HSs were more accurate with determiner than with adjectives, but L2 learners were similarly accurate in both domains.

Both HSs and L2 learners were more accurate with transparent than opaque nouns.

Montrul, Davidson, de la Fuente, and Foote (2013) studied gender agreement in HSs and

L2ers. Their study investigated whether the amount and type of exposure (more aural for HSs and more metalinguistic for L2ers) provide advantages to HSs over L2ers with gender processing. They also tested gender morphological endings in determiner-noun assignment and noun-adjective agreement. Participants were 29 Spanish-English HSs and 37 L1 English-L2 learners (L2ers) of Spanish. They used two picture naming tasks and an online elicited

2 Masculine default refers to the speaker’s tendency to use the masculine form as the default form, thus overextending the masculine to all gender agreement instances (Finnemann, 1992; De Garavito and White, 2002).

production task, measuring accuracy and reaction times of participants’ response. Results showed that the HSs were overall more accurate with gender than the L2 group. HSs and L2ers almost performed at ceiling level with canonical noun endings (-o and -a), but were less accurate with non-transparent nouns. There was also a main effect for gender as both groups were more accurate with masculine nouns. L2ers were more accurate on the gender of determiners than on gender agreement with adjectives, especially with feminine nouns. HSs revealed that they were equally accurate with gender agreement on both determiners and adjectives. Native speakers scored at ceiling with gender agreement. RTs analysis only revealed information on the RTs differences between monolingual, HSs and L2 learners. Accuracy results were in line with previous studies that found that agreement on the adjective is acquired later in L2 learners (De

Garavito and White, 2002). Contrary to previous studies, however, they did not find any difference between gender assignment and gender agreement in HSs (Alarcón, 2011, Martinez-

Gibson, 2010; Montrul and Potowski, 2007). They do not provide further explanation about the different results obtained in other studies with regard to these two different agreement operations but these results can be possibly explained by methodological differences: oral picture description task (Alarcón, 2011) and oral production (Gibson, 2010; Montrul and Potowski,

2007) versus elicited production task using diminutives (Montrul, Davidson, De la Fuente &

Foote, 2013).

More recently, Cuza and Perez-Tattam (2015) examined gender assignment and gender agreement in 19 Spanish monolinguals and 34 Spanish-English bilingual children. They administered a picture naming task in which children had to name the picture with a color adjective. Results showed that the mean correct response of HSs’ was lower than that of the monolinguals. The researchers found that HSs tended to overuse the masculine form, especially

in noun-adjective agreement, thus providing further evidence in support of the view that these operations are independent.

Because determiner-noun assignment occurs more frequent that any other agreement operation and the almost obligatory nature of determiner in the noun phrase, the present investigation also includes noun-adjective agreement, a post nominal agreement between the adjective and it anaphoric referent, the noun. It also is worth noting that none of these above- mentioned results measure the difference in processing costs between these two agreement operations which could provide further information on the actual difficulties of these two agreement processes.

2.4 Gender in Psycholinguistic Models of Lexical Access and Word Production

In the following section, I introduce two main psycholinguistic models (Independent

Network Model and Weaver ++) with the aim of illustrating how different linguistic information is processed in speech production and the predictions these models assume for the current study.

Lexical retrieval models describe the different stages in the word production process and the possible difficulties that a speaker might experience (Biran and Friedmann, 2012). Figure 2-2 is a simple example of word retrieval. The example begins with a first stage in which conceptual information takes form. Subsequently, this representation activates the semantic representation

(meaning) that is stored in the lexicon. After the semantic representation is formed, the phonological form (also referred to as word form) is activated. This stage will contain metrical and segmental information such as number of syllables, length of the word or stress, number of phonemes, consonants, etc. When this activation takes place, it is transferred to the phonological output buffer that is the stage responsible for the morphological composition of the word (Dotan and Friedmann, 2007, 2010; Kohn and Melvold, 2000) and contains the phonological form until the word is uttered (e.g., Biran and Friedmann, 2005; Butterworth, 1989, 1992; Dell, 1986, 1988;

Garrett, 1976, 1992; Kempen and Huijbers, 1983; Levelt, 1989,1992; Patterson and Shewell,

1987). Figure 2-2 is a simple illustration of the word retrieval process. Gender information or syntactic features are not specified at any level in this model; however, this figure serves as a baseline to provide details on the main literature and debates on the models of lexical retrieval.

Figure 2-2. Stages in lexical retrieval (Taken from Biran and Friedmann, 2011:5)

Models of lexical access agree on the assumption that there are two stages of noun retrieval, assuming a dual stage in noun retrieval (Caramazza, 1997; Miozzo and Caramazza,

1997; Roelof, 1992; Garrett, 1975,1980). Studies have provided substantial evidence that there are different planning spans when producing a word, particularly, the planning span in longer when selecting the lexico-syntactic and lexico-semantic features of the word than when they select the phonological form (Ferreira and Swets, 2002; Jescheinak, Schriefers and Hansch,

2003; Meyer, 1996; Smith and Wheeldon, 1999). Furthermore, it has been shown that the semantic and syntactic features of a noun become available before the properties of its

phonological form (for semantic processing: Indefrey and Levelt, 2004; Jescheniak, Schriefers,

Garrett and Friederici, 2002; Schmitt, Münte and Kutas, 2000; for syntactic processing: Van

Turennout, Hagoort and Brown,1998).

Despite the consensus on this dual access on lexical retrieval, there is disagreement on (1) the relationship between the lexical elements involve in noun production such as syntactic, semantic and morphophonological elements and (2) when these properties are activated or , in other words the time-course of the activation of these units.” (Meyer and Belke, 2007)

With regard to the relationship among these lexical units (semantics, syntax and morphophonology), according to Levelt, Roelofs and Meyer (1999), when lemma information, which is the noun’s underlying representation, is accessed, its syntax “becomes accessible for forming a syntactic environment that is appropriate for the word” (Biran and Friedmann,

2011:5). This information includes, per some researchers, among other things, the gender information that will determine the inflection of the phrasal elements that agree with it. On the other hand, the lexeme is the lexical representation of the word in its phonological output (Biran and Friedmann, 2011). It is widely accepted that the lemma (underlying representations) and the lexeme (phonological form) are stored separately (Caramazza, 1997; Garrett, 1976; Levelt et al.,

1998; Roelofs, 1992). However, what is controversial is the assumption by some researchers that the lemma includes semantic as well as syntactic information (Kempen and Hoenkamp, 1987;

Kempen and Huijbers, 1983; Levelt, 1989, 1992). Other researchers assume that the lemma has only syntactic information, separate from semantic information (Bock and Levelt, 1994; Levelt et al., 1998; Roelofs, 1992). With respect to gender, researchers have been investigating whether gender features are in the lemma (Levelt, 1989; Roelof, 1992, Levelt, Roelof and Meyer, 1999) or, as suggested by others, gender is adjoined to the lexeme (Caramazza, 1997).

Further extending on the debate on the relationship of the lexical units during lexical access, there is one view, supported by the group of researchers who are proponents of the

Weaver model of lexical access, that poses that lexeme features (morphophonological information) only becomes available after syntactic information has been selected (Levelt

1989,1992; Levelt et al.,1992,1999; see also Dell and collaborators: Dell, 1986, 1988; Dell,

Burger and Svec, 1997). This view is also called by Caramazza (1997) the “syntactic mediation hypothesis” (SM) (182). Another view, supported by the Independent Network model

(Caramazza, 1997; Caramazza and Miozzo,1997), assumes that the activation of the word lexeme features is not syntactically mediated. They assume that the lexical semantic node activates the lexical syntactic node and lexeme node in parallel. Below, these influential models of language production will be described, providing further evidence against and for the predictions these models undertake.

In the Weaver ++ model, Roelofs (1992) identifies four stages, with the first consisting of the identification of the concept, followed by a second stage in which the retrieval of the lemma takes place, a third stage in which the phonological form is encoded, and a fourth stage in which the word is articulated (Roelofs, 1992:114). Below an illustration of this model.

Figure 2-3. Example of word access in the Weaver ++model of lexical retrieval (adapted from Roelof, 1992:116).

Figure 2-3 above illustrates Weaver++ model (Levelt, Roelofs and Meyer, 1999), which assumes that “recognizing a word, whether spoken or written, involves accessing its syntactic potential, that is, the perceptual equivalent of the lemma and, therefore, activation of the corresponding lemma-level node” (Levelt, Roelofs and Meyer, 1999: 7). Therefore, lexical access always involves the retrieval of lemma features. This assumption has been discredited by a number of research, specifically in gender retrieval. Caramazza and Miozzo (1999), proponents of the Independent Network Model(further explained below), presented a model that did not postulate a distinction between the lemma and the lexeme. They based this assumption on the basis of evidence found in their research using a paradigm that taps into the Tip of the Tongue

phenomenon. The Tip of the Tongue (TOT) phenomenon can be explained as the momentary stage in which speakers cannot retrieve a word from memory or the word is weakly activated and speakers cannot recall it (Bock and Meyer, 1992). In their study, they found that the correct retrieval of morphosyntactic features (i.e., gender) did not correlate with the correct retrieval of morphophonological features (i.e., the initial phoneme of the word). They interpreted these results suggesting that the availability of lexeme features was not mediated by the prior activation of the syntactic node. Other evidence supporting this claim has been presented with brain damaged subjects. Brain damaged participants in Miceli and Caramazza (1988) made syntactic errors despite being able to access information about the phonological form of the noun

(for more studies with brain damage patients, see Cubelli and Perizzi, 1996; Semenza, Mondini and Cappelletti, 1995; for more studies with healthy subjects, see Starreveld and La Heij, 2004).

Another claim of the Weaver ++ model is that the gender congruency effect3 “should only be obtained when agreement must be computed” (Levelt, Roelofs and Meyer, 1999: 14) since only then would the gender features be selected. This effect is explained by the Weaver ++ model by suggesting that gender nodes will only be activated when the lemma is selected when it is necessary for its use in an agreement structure. This assumption has been studied via the picture-word interference paradigm, widely used to investigate the dynamics of when lexical units are activated (Cubelli, Lotto, Paolieri, Girelli and Job, 2005). In the picture-word interference paradigm, participants are presented with stimuli that consist of different components. Participants should respond to the main stimulus and ignore the distractor (visual or aural stimulus). The main goal of manipulating the linguistic features of the main stimulus and

3 By gender congruency effect, the researchers refer to those studies that focus on grammatical gender with the picture-word interference paradigm manipulating the grammatical gender of the distractor word in order to determine whether the gender of the distractor word affects the retrieval of the target stimulus.

the distractor/s is to gain insight into the different stages of lexical access. If the distractor’s lexical features slow down the participant’s target response, this is called a congruency effect (or inhibitory effects). On the other hand, if the distractor’s lexical features speed up the participant’s target response this is called a facilitatory effect. Based on the Weaver ++ model predictions, participants will show congruency effects when they are asked to name noun phrases

(e.g., determiner-noun) but not in bare noun production. This prediction was supported by studies that did not find any effect in bare noun production (La Heij, Mak, Sander and

Willeboordse, 1998). However, contrary to the Weaver ++ model, a study conducted by Cubelli et al. (2005), and a more recent study on bare noun production by Paolieri, Lotto, Morales, Bajo,

Cubelli and Job (2010), found congruency effects in bare noun production concluding that the activation of a noun’s gender feature is mandatory since it occurs in bare noun production.

Similar to the Weaver++ model (Roelofs, 1992, Levelt et al.,1999), the Independent

Network model (Caramazza 1997; Caramazza and Miozzo, 1999) differentiates between a lexico-semantic and lexico-syntactic level and the phonological level proposing three different levels representing syntactic, semantic and morphophonological information. It also assumes that gender features are not accessed in bare noun production based on the assumption that these abstract features are not necessary outside of syntactic contexts. However, this model does not assume the existence of a lemma level as a compulsory step between the lexico-semantic and phonological levels. This model assumes that the semantic representation can activate the lexeme or morphophonological form directly. In Caramazza’s model, there is a lexico-semantic system that represents “word meanings as sets of semantic properties, features or predicates”

(Caramazza, 1997: 194) and a lexico-syntactic system that represents “features such as grammatical category, gender, auxiliary type, tense” (Caramazza,1997: 194). Within the lexico-

syntactic network there are subnetworks for different grammatical features such as grammatical category, gender, number, phrasal structure, etc. Caramazza’s model does not have independent representations of lexemes such as phonological and orthographic lexemes. What is involved in word production is the activation of the lexico-semantic representation that later spreads to the lexico-syntactic system and to the phonological or orthographic representation. The spread of activation from the lexico-semantic representation to the lexico-syntactic and lexeme networks is independent and simultaneous. Below, Figure 2-4 exemplifies the two contrasting views, one that postulates a level between semantic or conceptual representation and lexeme (such as

WEAVER++) and another view that postulates one less stage of representation (as in IN).

Figure 2-4. Lexical access networks (A) representation of models that claim a level between conceptual and word form information (Weaver++) (B) models that proposed two levels of representation (IN). (Adapted from Caramazza and Miozzo, 1997: 313)

Research mentioned above (Caramazza and Miozzo, 1999; Miceli and Caramazza, 1988;

Cubelli and Perizzi, 1996; Semenza, Mondini and Cappelletti, 1995; Starreveld and La Heij,

2004) also represents evidence in favor of the Independent Network model. It provides evidence that the access of word form information is not mediated by previous access of its syntactic lemma node. Thus, the IN model assumes a different relationship between the lexical units involved than the one proposed by the proponents of the Weaver ++ model where word form information is only activated after the semantic and syntactic units are activated.

Another ongoing debate is the time course of the activation of lexical units (Meyer and

Belka, 2007). According to this debate, models are classified as cascaded or serial stage models.

The models described above also assume different time spans during the stages of word access.

The Weaver ++ model is an example of a serial stage model where the access of information during word production is described as separate and sequential phases. This type of model assumes that word form information would only become available after the previous node (or stage) has been fully activated. The Independent Network model is considered a cascaded model, where the activation of the semantic or conceptual node is enough to activate word form activation. Proponents of this model, assume that, even though the information flow is considered to be done in processing stages, these stages sometimes overlap. Figure 2-5 below is an extended version of the IN model. We can clearly observe in detail the parallel connections and the fluctuation of lexical syntactic and word/form information.

Figure 2-5. (A) Schematic model and (B) the processing of Spanish words mesa [table], silla [chair] and tigre [tiger] according to the Independent Network model (cascade view).(Adapted from Caramazza and Miozzo, 1999: 31)

In summary, we can state that these two models illustrate how lexico-syntactic information is retrieved in word form production and offer different predictions on where syntactic information is stored. On the one hand, the Weaver ++ (Roelef 1992, Levelt, Roelef and Meyer, 1999) predicts that morphophonological features are only accessed after syntactic information becomes available, thus postulating syntactic mediation between the conceptual and lexeme nodes. Also, it postulates that the information or activation of a lexical node is done in fixed stages. On the other hand, the Independent Network model (Caramazza and Miozzo, 1997) proposes three different networks, but this model assumes that semantic representations can activate word representations directly, without assuming an intermediate lemma node. Important for our current investigation, the IN proposes that once the conceptual node is activated lexico- syntactic information, including gender, and lexico-phonological nodes become available (such as gender morphological endings).

Unlike the Weaver ++ model that “consists on a set of discrete stages that are completed in a specific temporal order” (Belke and Helga, 2007:481), the cascade model represented by the

IN model postulates that “lexemes activate and allow selection of their associated syntactic features” (Caramazza and Miozzo, 1997: 340). This dissertation studies the role of noun morphology and frequency in gender agreement operations in two different linguistic populations. That is, we investigate whether the morphophonological features in the lexeme interact with syntactic features during lexical access. Therefore, if the current investigation is consistent with the predictions postulated by the IN model, then we should find a possible interaction between noun morphology (transparent v opaque) and gender access. The following section reviews research that provides evidence on this postulation and also on the role of frequency in two linguistic populations.

2.5 Studies on Grammatical Gender

Psycholinguistic research has focused on the cognitive difficulty that agreement may present during sentence processing for monolingual, bilingual, and second language speakers.

This section will provide an overview of the literature and methodology used on how monolingual and bilingual speakers use gender information, specifically how gender access is impacted by noun morphology (transparent versus opaque), lexical frequency, and how gender access differs in speakers that vary in language dominance (+/- Spanish-dominant). Additionally,

I will review the theoretical frameworks in which this research is contextualized.

The following studies will provide a comprehensive account of the role of gender morphological endings and noun frequency in gender access in Spanish in monolingual speakers and heritage speakers.

2.6 Gender Agreement with Noun Morphology

2.6.1 Native Speakers

An ongoing issue in gendered languages like Spanish is whether noun morphology affects gender retrieval. Spanish is overall considered a transparent language with respect to gender morphology and the type of agreeing elements that encode gender (e.g. articles, adjectives). First, one of the competing views on the effect of gender morphology considers form-based cues (i.e. overt gender markings, such as -a or -o in Spanish, on articles, determiners, suffixes) the most reliable type of cue for gender retrieval. Following the terminology used in

Gollan and Frost (2001), this view will be called the “Reliable Cue Hypothesis”, and it is supported by studies that have found strong effects of gender-marking cues in determiners and nouns (Bates et al., 1996; Taft and Meunier, 1998). These studies use a gender decision task in which participants indicated the gender of target words. Second, some studies have given rise to

a view in which gender can be retrieved without the help of overt morphology at the form level4.

These studies, mostly using the Tip of the Tongue paradigm (TOT), examine the relation between grammatical gender and lexical access. These studies have been carried out in different languages (e.g. Italian, French, Spanish, and German) with healthy participants (Caramazza and

Miozzo, 1997; Miozzo and Caramazza, 1997; Vigliocco et al., 1997) and with participants with language impairments such as anomia (Badecker et al., 1995; Costa et al., 1999; Gonon et al.,

1989; Kulke and Blanken, 2001). Results from both populations have provided evidence that speakers can retrieve gender even when they do not access phonological information (or when they have a deficit, in the case of participants with anomia). Under this assumption, participants could access the lexico-syntactic information of gender, e.g. grammatical gender knowledge, although they had incomplete or deficient phonological information. These results provide a strong argument in favor of the view that gender information is accessed separately from morphophonological form. We will call this view the “lexico-syntactic access view”. In conclusion, these studies using TOT provide evidence that gender can be retrieved accurately for both transparent and opaque nouns; that is, the gender of the noun can be retrieved independently of its phonological form (Badecker et al., 1995; Miozzo and Caramazza, 1997; Vigliocco et al.,

1997). The authors of these studies suggest that gender retrieval can be done without relying on gender-correlated morphemes, meaning that there is no difference in gender retrieval between transparent and opaque nouns. A third alternative, called the Two-Route Hypothesis (Gollan and

Frost, 2001), is represented by studies that acknowledge the presence of two routes to retrieve

4 This view may not be fully incompatible with the former one (reliable cue hypothesis) since word form morphology could still be a strong cue for gender retrieval but not the most reliable one. The two-route hypothesis suggests that lexico-syntactic information could be retrieved through a lexical and morphological word-form route and proposes a double access in which abstract lexical information and morphological form can be accessed and used in different stages of language production.

gender, one purely syntactic and another morphophonological, both present in the lexical access process. Gollan and Frost (2001) studied gender retrieval in Hebrew (a language without articles), using a gender decision task and a timed grammaticality judgement task (GJT). In

Experiment 1, which consisted of naming the gender of the noun that appears on the screen (a noun appeared on the screen and participants had to say the gender of the noun), the researchers found strong effects (showing faster naming time) of regularity (masculine and feminine morphologically marked nouns), replicating prior work using the same task (Bates et al., 1996;

Taft and Meunier, 1998). Regular nouns provided the fastest RTs and the fewest errors, indicating that gender-marked cues impact access to gender. However, in Experiment 2, the researchers tested retrieval of grammatical gender in what they consider a more syntactic context

(e.g. noun –adjective agreement versus a noun on the screen from Experiment 1) using a timed

GJT. In this experiment, noun-adjective sequences appeared on the screen and participants had to judge the grammaticality of the sequence via keyboard. Results on RTs and accuracy indicated that participants had the same difficulty making grammaticality judgements on regular nouns as they did on irregular nouns. In both the RT and error analysis of grammatical pairs, the researchers did not encounter a significant main effect of regularity. In other words, there was no difference between regular and irregular nouns. However, there was a significant effect of regularity in the pairs that were judged ungrammatical, in both the RT and error analysis. In other words, when the researchers analyzed the ungrammatical pairs, they found an effect on regularity. The difference found in the experiment (lack of an effect of regularity on grammatical responses) was interpreted as suggesting that “the detection of correct gender agreement was far less affected by gender-marking” (Gollan & Frost, 2001: 642). To explain the contradictory results on ungrammatical responses, the researchers suggested a two-route hypothesis, in which

the first route “derives gender from its correlation with gender marking at the level of form”

(Gollan and Frost, 2001: 644). Yet, they suggested that a second route (a lexical one) is necessary because access to gender was done as easily and accurately for gender marked and unmarked forms of grammatical pairs--in other words, without any effect of gender morphological endings. The researchers interpret these results as indicating that speakers of gendered languages have an abstract representation of gender. These results are also in line with the proponents of the IN network (Caramazza, 1999) since it implies a closer link and interaction between lexico-syntactic and lexeme nodes.

A recent study on gender processing in Spanish by Caffarra, Jassen and Barber (2014) provided more evidence of the existence of two routes in the brain to access gender. The researchers conducted an ERP study using the visual half field (VHF) paradigm while participants performed a grammatical judgement task in which they assessed the grammaticality of the determiner-noun pairs that were displayed on the computer screen. The VHF paradigm is used to observe how the two hemispheres contribute to processing (Hellige, 1983). Previous studies on VHF have found that both left and right hemisphere (LH and RH, respectively) are engaged in grammatical operations, however, the LH is more specialized in identifying morphological information (Koenig, Wetzel and Caramazza, 1992) and uses this sublexical information in syntactic operations (Zaidel, 1983). Therefore, researchers utilize this paradigm to observe the role of morphological cues in each hemispheric computation. To test this effect in the experiment, the determiner was visually presented in the middle of the screen while the subsequent noun was visually presented on the right or left of the VHF. The Left Visual Field

(LVF) is associated with RH computation, and the Right Visual Field (RVF) is associated with

LH computation (for more lexico-semantic studies using VHF see Federmeier, Mai and Kutas,

2005; Federmeier, Wlokto and Meyer, 2008 and for syntactic VHF studies, Kemmer, Coulson and Kutas, 2014). Following the predictions of the of Two-Route Hypothesis (Gollan and Frost,

2001), Caffara et al. (2014) expected a first route where lexical gender information is retrieved as an abstract feature, which should be found (via RTs and ERPs) in both visual fields, and a second route that will use the gender morphological endings (e.g. -o or -a in Spanish). This second route should be most apparent in LH processing and therefore more evident in the RVF.

Behavioral results showed that participants were faster and more accurate with the agreement than disagreement condition. Also, they were more accurate with transparent than opaque nouns.

Participants were also faster and more accurate for stimuli presented on the right visual field.

Results on the ERPs also showed effects of accuracy and transparency in the two time windows that were analyzed. However, they did not find any interaction between morphological endings and agreement in the first window arguing that “the initial computation of grammatical agreement does not rely on distributional information conveyed by the word ending” (Caffarra et al., 2014: 9). However, the second window showed an interaction between morphology and transparency and they interpreted this finding as an index of the activation/use of word morphology via a second route to analyze agreement mismatches. With regard to the specific hemispheric computation, they reported the interaction of the visual field with agreement and transparency. They observed an effect of morphology for the stimuli presented in the RVF, while the LVF only showed effects with the agreement factor. The researchers interpreted these results as the first ERP evidence for the presence of two routes to process gender agreement (Gollan and

Frost, 2001). They interpreted their ERP results as evidence that the lexical route recovers gender abstractly while morphological endings are used at different stages of gender processing.

Another recent behavioral study (Resence and Mota, 2017) using a timed gender selection task found some transparency effects in both conditions (same as Experiment 1 and 2 from the present investigation but presented at the same time in one experiment) but not in the low and high frequency conditions. Investigators manipulated word frequency and word morphology, opaque and transparent to test monolingual speakers of Brazilian Portuguese in two tasks (determiner+noun and noun+adjective agreement). Participants had to do the correct agreement by pressing a key on the keyboard that represent masculine or feminine agreement.

Results found faster naming times in high frequency transparent nouns (against the Dual

Mechanism Model further explained in section 2.6.1.1) and also faster naming times for low frequency transparent nouns in adjective noun agreement. It is worth noting that this study includes both opaque and irregular nouns, that is, regular morphological endings that are attached to words of the opposite grammatical gender, e.g. la mano (the hand).

Grammatical gender processing has been examined extensively using a picture-word interference paradigm (Rosinski, Golinkoff and Kukish, 1975). As briefly mentioned above, this paradigm consists in a naming task in which objects/pictures/words appear on the screen along with a distractor object/picture/word. These distractors can be semantically or phonologically related or not (e.g. picture of a cat with a picture of a dog as an example of semantically related distractors) or they may be manipulated to exhibit similar or different agreement features (e.g., gender features: picture of a feminine noun with a picture of a masculine noun as an example of syntactic unrelated distractor). Studies that focus on grammatical gender with the picture-word interference paradigm manipulate the grammatical gender of the distractor word to determine whether there is a gender congruency effect; that is, whether the grammatical gender of the distractor matches the target noun and affects lexical access of the target item. Cubelli et al.

(2005) used this paradigm to investigate whether grammatical gender was also activated in the production of bare nouns (against the predictions of the Weaver++ and IN that postulated no effects of noun congruency in bare noun production) and corroborate the lack of gender congruency in the production of noun phrases in Romance languages. They administered four experiments to Italian monolinguals. Experiment 1 included only bare noun production,

Experiment 2 included additionally the production of determiners in noun phrases, and

Experiment 3 replicated the first experiment using a new set of stimuli. Experiment 4 tested bare noun production again, but this time it did so while manipulating noun morphology (i.e., transparent or opaque). Results of Experiments 1 and 3 showed slower naming times when the target and distractor noun share the same grammatical gender. Results of Experiment 2 with full

NPs showed no gender congruency effect, that is, speakers did not slow down when presented with a distractor with the same gender (see following paragraph for more details on this issue). In the Experiment 4, in which the researchers manipulate noun morphology, the effects of gender congruency were also found, thus replicating the results of Experiments 1 and 3. However, experimenters did not provide further details on the role of the morphological endings in this last experiment.

Unlike studies that have found gender congruency effects in the production of NPs in

Dutch (Van Berkum, 1997), Croatian (Costa, Kovacic, Franck, and Caramazza, 2003) and

German (Schiller and Caramazza, 2002; Schriefers and Teruel, 2000), studies testing Romance languages, such as the one reviewed in the previous paragraph (i.e., Experiment 2), have not found congruency effects so far (Spanish and Catalan: Miozzo and Caramazza, 2000; Costa et al., 1999; Italian: Miozzo and Caramazza, 1999; French, Alario and Caramazza, 2002). The variability observed in the retrieval of determiners in these two different language families has

been claimed to be due to the different type of information that has to be accessed when a determiner is needed (Costa, Alario, and Sebastián-Gallés, 2007). In languages such as German and Dutch the phonological environment (or phonological form) does not play a role in the selection of the determiner, however, in Romance languages it does. For this reason, Germanic languages are thought to be “early selection” languages, because the determiner can be selected earlier in the process without relying on the phonological context. Romance languages are considered “late selection” languages since word form information also plays a role. For instance, in Italian, the masculine article (the) has two forms il or lo, whose selection depends on word form information (specifically the initial phoneme of the following noun) (Caramazza,

Miozzo, Costa, Schiller and Alario, 2001). Therefore, based on these differences across languages, some researchers (e.g., Caramazza et al., 2001; Miozzo and Caramazza, 1999) hypothesized that “for those languages in which the determiner form can be selected on the basis of only semantic and grammatical information, the gender congruency effect will be present, while for those languages in which the selection of the determiner form need to wait until the related phonological information is retrieved, the effect will be absent” (Costa, Alario, and

Sebastián-Gallés, 2007: 536).

In contrast with the evidence that gender information can be retrieved without the help of morphological cues, these findings imply a closer connection between lexico-syntactic and word form, therefore an evidence that gender grammatical cues are used during the agreement process.

This prediction is consistent with the models that assumes a fluctuation of the information between the lexeme and lexico syntactic node (Caramazza, 1999; Caramazza and Miozzo, 1997).

In conclusion, findings on how gender features are retrieved and whether gender-cues override other lexico-syntactic information are not conclusive. Also, the different methodologies

presented above--with some being more metalinguistic in nature, such as grammaticality judgement tasks or gender decision tasks--might have compelled the speakers to use strategies other than those that would typically be recruited in gender processing. Therefore, the aim of the present study is to provide more research on the impact of noun morphology by employing a more implicit task that does not require speakers to use their metalinguistic knowledge of gender.

Additionally, we will expand results to a population of HSs of Spanish.

Frequency and morphology interaction: The Dual Mechanism Model

There has been an intensive research on the different mechanism the brain utilizes to access regular and irregular forms in the language, such as the case of irregular pasts in English

(Pinker, 1991, 1999; Ullman, 2001; Bybee, 1985, 1995; Jackendoff, 2002). One of the views proposes the processing occurs in two linguist processes, one rule-based and the other relies on memory. For this reason, frequency will not play a role with high frequency regular nouns because these regular nouns will be computed and used a rule application system. However, for high frequency irregular nouns, since they are stored in memory as a whole, they will have an advantage in naming times over the regular nouns. Opposite to the pattern we observe with high frequency, low frequency will affect the retrieval of low frequency irregular nouns since they will have to be looked up in the mental lexicon, while low frequency regular nouns will show an advantage since they have to be computed without accessing memory. Under this view, regular nouns (also called complex words) will not be subject to frequency effects. Another view

(Bybee, 1985, 1995; Mcwhinney, 2001), claims that there is only a single mechanism based on procedural memory to process both regular and irregular form in the language. Therefore, frequency will play a role in both regular and irregular forms.

Similar to regular morphology, transparent nouns consist on complex nouns, with -o or a- endings, that are assigned to a grammatical gender. Similar to irregular nouns, with opaque nouns morphophonological cues are not predicted from the ending, therefore, these cues do not convey any particular gender. However, there is not much research that compares the processing patterns of regular and irregular morphology with the patterns of transparent and opaque morphology in languages with grammatical gender. There is only one study (mentioned above) that tested grammatical gender in Spanish with transparent and opaque nouns (Resence and

Mota, 2017). This study did not find evidence in favor of the Dual Mechanism Model to process transparent and opaque nouns since they found frequency effects in high frequency transparent nouns.

2.6.2 Bilingual Speakers

Another unresolved question is how noun morphology affects the retrieval of gender features in different populations. Research on bilinguals, specifically second language learners, has indicated that learners are more accurate in gender assignment and agreement with transparent nouns than with opaque nouns.

Montrul et al. (2008) administered an oral picture description task that included both transparent and opaque nouns and found that L2 learners produced a larger number of errors with

Spanish nouns that were opaque, particularly when those nouns were feminine. Alarcón (2011) replicated Montrul et al. (2008), and the pattern of results was the same. Alarcón (2011) examined the role of transparent gender morphology in the production of agreement by L2

Spanish learners and HSs, using a written production task in which participants were asked to choose the correct definite article to accompany a noun and then produce an adjective to modify the noun in a meaningful context. Learners were more accurate in producing agreement with transparent nouns than with opaque ones. Montrul et al. (2013) also tested two types of

bilinguals, second language learners and HSs of Spanish. In this study, participants were required to produce nouns in their diminutive form in Spanish (e.g. un pez ‘a fish’, diminutive: un pececito ‘a little fish’; una cruz ‘a cross’, diminutive: una crucecita ‘a little cross’). They also found that agreement on both the determiners and the adjectives accompanying the nouns was less accurate when the nouns were opaque, as well as when the nouns were feminine.

Psycholinguistic research on agreement has also widely researched attraction errors.

These attraction errors occur when there is an intervening noun between the head noun and the agreement element. These types of errors have been investigated in English mainly with number agreement (Bock & Cutting, 1992; Bock & Eberhard, 1993; Bock & Miller, 1991). Number attraction errors occur when the second element of an NP has different number from the subject noun, eliciting number errors in the verb (e.g., 3rd person singular subject/plural verb: The family of cats *are safe). The verb seems to be attracted by the number features of the locally intervening noun instead of the head noun, with plural providing a stronger attraction than singular. Gender attraction, as illustrated in example (8), occurs similarly to number attraction when there are two nouns in the NP and a post verbal adjective is incorrectly marked for gender with the preceding noun rather than the head noun.

(8) El pato en la laguna está *escondida The-DET.MASC.SG duck-N.MASC.SG. in-PREP the-DET.MASC.SG pond-N.MASC.SG is hidden-PART.FEM.SG. “The duck in the pond is hidden”

Attraction errors have been studied using Bock and Miller’s (1991) sentence completion task and, as its name indicates, participants are presented with a noun phrase with two nouns, and then participants are instructed to complete the sentence. This experimental paradigm has been conducted with monolinguals and bilinguals in comprehension and production tasks

(Eberhard,1997; Gennari and MacDonald, 2009). Although fewer studies have been done with

grammatical gender, some studies have provided evidence for gender attraction in Spanish

(Alarcón, 2009-further explained below) and in other languages (Viglioco and Franck, 1999).

This paradigm allows researchers to manipulate the characteristic of the noun phrase

(head noun and distractor, example 11 and 12). Alarcón (2009) investigated linguistic variables that influence agreement in Spanish monolingual speakers and L2 learners of Spanish. The linguistic variables were gender (masculine or feminine), head noun morphology (overt or non- overt, previously mentioned as transparent or opaque), noun class (biological or arbitrary gender) and gender congruency (matched and mismatched). L2 learners were divided into three groups according to their proficiency level as measured by a university placement test. Alarcón used an online sentence completion task. Sentences consisted of a head noun and a second noun (the attractor) in a copula construction with an adjective that matched and mismatched and which the participant needed to select.

(9) El amigo de la dueña … está cansado/cansada. The-DET.MASC.SG. friend-N.MASC.SG. of-PREP. the-DET.FEM.SG. owner-N.FEM.SG. is-V. cansado/cansada-ADJ.MASC/FEM.SG. “The owners´s friend is tired”

Head noun Attractor Adjective selection

The goal of manipulating the two nouns (variables explained above) in the noun phrase was to observe the linguistic variables that enhance native speakers’ and L2 learners’ vulnerability to agreement errors. Participants read the first sentence with the head noun and the attractor and the copulative verb that appeared on the screen and the following screen presented the two inflected adjectives; then, participants were asked to select one. Reaction times were measured from the moment participants pressed the “continue” key to the moment one of the adjectives was selected. As in previous studies, results showed that all participants were faster with animate head nouns (Sagarra and Herschensohn, 2009). L2 learners, but not native

speakers, regardless of proficiency level, were faster with animate attractors. Regarding gender morphology of the head noun, she found that native speakers but not L2 learners were faster when the head noun had overt morphology. As she mentioned, providing evidence that overt morphology is a reliable cue for native speakers. The gender of the head noun was only predictive of faster naming times for the more advance L2 group, having faster naming times with feminine nouns. The gender of the attractor was not predictive of faster naming times for any proficiency level. Unexpectedly, gender mismatch conditions (between head and attractor) produced faster naming times in intermediate and advanced L2 groups.

Montrul, Davidson, De La Fuente and Foote (2014) conducted three spoken word recognition tasks with intermediate and advanced L2 learners of Spanish and found that participants were less accurate and slower with nouns that were not transparently marked for gender. Similarly, results from a noun repetition task conducted by Foote (2014) indicated that advanced L2 learners were better able to make use of gender cues on determiners during spoken word recognition in Spanish with transparent nouns than with opaque nouns. However, using an online task, Foote (2015) studied how noun morphological endings impact gender agreement processes. She compared a population of native speakers of Spanish and advanced L2 learners.

She carried out a sentence fragment completion task in which a computer presented a preamble, el castillo del pueblo… (“the castle of the village…”), and adjectives in both the masculine and feminine forms. The participants’ task was to repeat the preamble and complete it using the adjective provided. She found that native speakers of Spanish were not affected by the form- based cues present in the noun. However, despite being an online study, results reported are based on the accuracy analysis.

Most of the research presented above uses offline methods, some of them metalinguistic in nature eliciting the speaker’s metalinguistic knowledge of the language, to study how different bilingual populations, particularly HSs and L2 learners, perform on gender agreement with different noun morphology. The present study will compare RTs and accuracy rates of two bilingual population (HSs of Spanish) which differ in language dominance and a group of monolingual speakers to observe the effect of noun morphological endings in gender agreement.

2.7 Gender Processing with Noun Frequency

2.7.1 Native Speakers

There is no extensive research on the role of noun frequency in the retrieval of gender features in agreement operations in Spanish. Navarrete, Basagni, Alario, and Costa (2006) used two production experiments to test access to gender with different noun frequencies in monolingual Spanish speakers. Experiment 1 was a determiner picture naming task with the following structure: demonstrative + copula + adjective, este/a es nuevo/a (“this-DET.MASC/FEM

(one) is new-ADJ.MASC/FEM”) with high frequency (HF) v. low frequency (LF) nouns. Response latencies were measured from picture presentations on the screen. Results indicated that the naming latency of the target sentence was affected by the frequency of the noun that is, low frequency nouns elicit slower responses. Experiment 2 was a gender decision task in which participants had to decide the gender of a picture by pressing a button box. Results from

Experiment 2 found frequency effects similar to those found in Experiment 1. These findings contradicted the predictions of a lack of a word frequency effect in those tasks in which the speaker is asked to provide information about the lexico-syntactic features of the noun but not the word itself (lexeme). Therefore, the researchers interpreted these results as evidence that frequency effects appear in tasks in which only the lexico-syntactic information of the agreement source is necessary without lexeme retrieval.

2.7.2 In Bilingual Speakers

Frequency effects in retrieval processing: Frequency-Lag Hypothesis

Gollan, Montoya, Cera, and Sandoval (2008) proposed the “Weaker Links Hypothesis,” more recently renamed as the “Frequency-Lag Hypothesis” (Gollan, Slattery, Goldenberg, Van

Assche, Duyck, Rayner, 2011; Emmory, Petrich and Gollan, 2012) which poses that bilinguals are different from monolinguals on production tasks because they divide frequency-of-use between two languages. Bilinguals necessarily speak each language less often than monolinguals do, which results in frequency effects (slower and faster naming times) in both languages and particularly in their non-dominant language. The researchers based this account on production and comprehension models (even though they only tested their model in production) that explain frequency effects by assuming that lexical representations store basic levels of activation with increased use (i.e., exemplar-based models). According to Gahl and Yu (2006: 213), exemplar- based models suggest that “mental representations consist of memory traces of specific tokens.”

This concept is different from other models (e.g. Underspecification Theory, Chomsky,

1993,1995) that try to develop extremely simple, non-redundant representations as a way to economize “their conception of the lexicon and its grammatical interfaces” (Gahl and Yu, 2013:

213). Therefore, due to these levels of activation promoted by increased use, the hypothesis predicts that monolinguals will exhibit the highest activation levels, and among bilinguals, the activation level will be higher in the speaker’s dominant language than non-dominant language.

In other words, the hypothesis predicts faster/slower noun retrieval according to language dominance in bilinguals and that monolinguals will exhibit the fastest retrieval since their level of activation will be highest. Thus, bilinguals will have to share frequency of use of specific concepts across their two languages whereas monolinguals only make use of one language. To illustrate with a simple example, if a bilingual speaker produces or comprehends a certain

concept one hundred times during a year, a certain percentage of those times will be in either language, thus leading to reduced frequency in either language. In comparison, monolinguals would say or comprehend the same concept in one single language the full 100 times. Those researchers tested the frequency-lag hypothesis in a production task in which they predicted that frequency effects (as seen by faster or slower naming times) between high and low frequency words, would be stronger in low frequency words in all speakers. Thus, the researchers predicted a disadvantage in the performance of bilinguals relative to monolinguals and stronger frequency effects within the bilingual group for a bilingual’s non-dominant language versus their dominant language. Of interest for the current study is Experiment 1, which tested 57 English monolinguals and 73 Spanish-English bilinguals (57 bilingual speakers dominant in English and

16 bilingual speakers dominant in Spanish) using a picture naming task that included high and low frequency nouns in English. Bilinguals showed overall slower naming time differences than monolinguals. The bilingual cost was greater for low-frequency words than for high-frequency words (even for bilingual speakers dominant in English). There was an interaction between language dominance and frequency category such that bilinguals showed larger frequency effects in the non-dominant language than in the dominant language. The results of Experiment 1 provided direct support for the Weaker Links (Frequency-Lag) Hypothesis by replicating other findings that had shown that bilinguals name pictures more slowly than monolinguals (Gollan,

Montoya, Cera and Sandoval, 2008) and by showing that the difference between groups is more pronounced for low-frequency words. That is, bilinguals showed larger frequency effects in picture naming because reduced language use affects low-frequency nouns more than high- frequency nouns. Another study has corroborated the effects of frequency on other syntactic structures (Runnqvist, Gollan, Costa and Ferreira, 2013). In that study, researchers tested passive

and active structures and pre- and post-modified possessive NP in two groups of bilinguals

(Spanish-English and Chinese-English). Findings replicated previous results under the same hypothesis with noun frequency; that is, more frequent syntactic structures were faster to produce than less frequent syntactic structures (Gollan, Montoya, Cera and Sandoval, 2008; Gollan,

Slattery and Rayner, 2012; Emmory, Petrich and Gollan, 2012).

Using a different bilingual population, Emmorey, Petrich and Gollan (2012) tested 28

ASL deaf signers and 40 hearing ASL-English bilingual signers (i.e., bimodal bilinguals) and 21 monolingual English speakers in a PNT. As predicted by the frequency-lag hypothesis, bimodal bilinguals were slower, less accurate, and exhibited a larger frequency effect when naming pictures in ASL as compared with naming pictures in English (their dominant language) and as compared with deaf signers. However, contrary to previous results, when naming in English, bimodal bilinguals did not differ from English monolinguals; in other words, bimodal bilinguals showed no frequency effect in English presenting similar results as monolingual speakers. The results contrast with prior results with bilingual groups where both languages (dominant and non-dominant) showed frequency effects. They suggested that these findings were the result of

ASL and English used at the same time (code-blending) by bimodal bilinguals. Also, many ASL signs are accompanied by their English sound articulation or “mouthing.” Instead, unimodal bilinguals cannot use their two languages at the same time but must switch between them.

As presented above, research on bilinguals and frequency effects has focused on the lexico-semantic properties of nouns (Gollan, Montoya, Fennema-Notestine and Morris, 2005) or syntactic structures (Runnqvist, Gollan, Costa and Ferreira, 2013). However, frequency effects on the retrieval of lexico-syntactic features in agreement operations, which appear to be a source of errors for many bilinguals, have only been studied in monolinguals (Navarrete et al, 2006).

The proposed research project would test gender agreement between determiner-noun and noun- adjective manipulating noun frequency, but also its overt morphological representation: transparent and opaque morphology in bilingual populations that experience reduced exposure in

Spanish (i.e. HSs). Moreover, I propose that frequency effects in bilinguals with different language dominance and language experiences will predict different outcomes with regard to access to gender, i.e., with lemma retrieval. This hypothesis is an extension of this original weaker links/frequency-lag proposal based on lexical retrieval.

The effects of gender morphological cues on gender have been examined before in monolinguals. A first wave of research, mainly using gender decision tasks (Bates et al., 1995;

Taft and Meunier, 1998), indicated that morphological cues are a strong predictor of successful agreement. A second wave of research, mainly TOT studies (Caramazza and Miozzo, 1997;

Miozzo and Caramazza, 1997; Vigliocco et al., 1997), concluded that gender can be retrieved without the help of morphological endings. Finally, a third wave of research, using gender decision tasks (Gollan and Frost, 2001) and grammatical judgments with ERPs (Caffarra et al,

2010), proposed two routes to access gender. The reason for these differing results might have been the metalinguistic nature of the tasks employed in prior research in which participants are asked directly about the gender of a noun or the grammaticality of the sentence. Therefore, these types of tasks might recruit different processing strategies than those normally used when accessing gender in real time. For this reason, the present study utilizes a less metalinguistic task which entails asking participants to name the stimulus (with no explicit reference to gender) that appears on the screen while participants’ reaction times are recorded.

Furthermore, the effect of gender morphological cues in bilingual gender processing has been understudied (Alarcón, 2011; Montrul et al, 2013: Foote, 2015). Most of the research has

been done on the effect of gender morphology with L2 learners finding effects specifically with opaque nouns (Alarcón, 2009; Sagarra and Herschensohn, 2009; Montrul et al, 2014; Foote,

2014). Therefore, the present investigation will help fill a critical gap in our understanding of gender retrieval in heritage speakers of Spanish.

The present investigation uses a picture naming task (PNT) similar to the one used in

Gollan et al (2008) in one of the researchers’ first tests of the Frequency-Lag hypothesis.

However, unlike in that study, the focus of this investigation is lexico-syntactic features in the noun, specifically gender features, so we will test lexical retrieval through agreement processes, determiner-noun and noun-adjective agreement.

CHAPTER 3 RESEARCH QUESTIONS AND METHODOLOGY

As presented in the previous chapter, research on the role of noun morphological endings in the retrieval of gender by different populations has been inconclusive. Meanwhile, no extensive research has been conducted on the role of noun frequency in the retrieval of gender features in Spanish monolinguals or bilinguals. This dissertation aims to fill this gap. The main goal of this chapter is to first present the research questions and predictions of the present investigation and then provide a description of the methodology employed in this study. In section 3.2, this chapter will introduce the study, the research questions, and the corresponding hypotheses. Then, in section 3.3., this chapter will present the methods section starting with a profile of the participants followed by a description of the materials used in the experiment and the experimental procedure.

3.1 Current Study and Research Questions

In the previous chapter, I reviewed the studies that have provided evidence for the role of morphology and frequency in the retrieval of lexico-syntactic gender features in monolingual and bilingual HSs of Spanish. Below, I will introduce the present research on the effect of overt morphology and frequency in accessing the lexico-syntactic features of nouns in monolingual speakers of Spanish and Spanish-English bilingual speakers during a Picture Naming Task

(PNT) and picture description task (PDT). I will elaborate on these effects and further outline the research questions and predictions for the present research.

3.1.1 Effect of Noun Morphology in Gender Access in Monolingual and HSs of Spanish

Regarding noun morphology, research focused on monolingual speakers applying different methodologies has shown inconclusive results regarding native speakers’ use of morphological endings in retrieving gender. Three main views predominate on the subject. The

first view, known as the ‘reliable cue hypothesis,’ considers noun morphology as the most reliable cue to access gender; this view has been supported by research that found that morphological endings helped speakers to retrieve gender (Bates et al., 1996; Taft and Meunier,

1998) (See section 2.7.1 for a review of these studies). The second view, which we will call the

‘lexico-syntactic cue hypothesis,’ mainly supported by tip-of-the-tongue (TOT) studies, does not consider morphological endings important in the retrieval of gender. This view has emerged in light of research which found that participants could retrieve gender regardless of noun morphology (Badecker et al, 1995; Miozzo and Caramazza, 1997; Vigliocco et al., 1997). In other words, noun morphological endings were not predictive of more accurate gender access or faster responses (See section 2.7.1. for a review of these research). The third view, the ‘Two-

Routes Hypothesis,’ proposed by Gollan and Frost (2001), conceives of two routes to access gender, one lexical and one based on noun morphological endings or phonological form. Gollan and Frost found that noun morphological endings were important in the detection of agreement mismatches, yet morphological endings did not play a role in detecting the gender of grammatical pairs. Therefore, Gollan and Frost interpreted their results as revealing speakers’ ability to use both morphological endings and abstract lexical information as routes to access gender.

A fourth outcome of the hypothesis outlined above is that transparent and opaque morphology are going to go in line with studies and view that support a dual mechanism to process regular and irregular forms (Dual Mechanism Model) (Pinker, 1992,1999). This is our second research question where we investigated whether there are separate cognitive mechanisms to process irregular and regular forms. It is important not to mistake this mechanism with the Two-Route Hypothesis proposed above (Gollan and Frost, 2001) to process regular and

irregular forms, or in our study, transparent or opaque morphology. Just to clarify, the Dual

Route Hypothesis (Gollan and Frost, 2001) denotes retrieving both the morphophonological and lexico-syntactic (gender) information in transparent nouns while, with opaque nouns, speakers will only make use of a single lexical route. The Dual Mechanism Model assumes that speakers retrieved regular nouns (transparent) applying a rule-based mechanism, while irregular nouns

(opaque) are stored in memory. Therefore, this model assumes two separate cognitive mechanism for irregular and regular forms (opaque and transparent nouns in our experiments).

According to this view, there will be no effect of frequency for the transparent nouns, because they are computed, but frequency effects will be seen in the irregular forms, specifically, in HF opaque nouns since this will be retrieved fasted than HF transparent. With low frequency nouns, low frequency transparent will be named faster than low frequency opaque nouns because transparent noun, having regular morphology, will be computed using a rule application.

Regarding the role of noun morphology in heritage speakers (HSs) of Spanish, it has been mainly studied using off-line methods, such as picture naming task (Martínez-Gibson, 2011) or an elicited production task (Montrul et al., 2013; Alarcón, 2009,2011), among other studies and methodologies employed (outlined in section 2.6.2), but there is no research that reports participants RTs and investigates whether HSs of Spanish might differ from monolingual speakers in the use of morphological cues when accessing gender. Therefore, one of the goals of the present dissertation was to investigate the role of noun morphology in monolingual speakers and HSs of Spanish with different linguistic dominance in Spanish. My main contribution is to use an on-line methodology to differentiate within these groups of heritage speakers with different linguistic dominance because no study has cross frequency and morphology in Spanish, providing evidence on the different views of HSs and gender processing.

To investigate this, as well as the impact of noun frequency, we carried out a PNT and

PDT, in which participants had to name and describe respectively different pictures with different morphological endings. Noun frequencies were also manipulated since frequency effects have been reported in naming tasks (more detailed information on the effects of frequency will be provided below). Consequently, in order to determine how different populations of Spanish speakers make use of noun morphological endings when retrieving gender, a set of high and low frequency nouns with opaque and transparent noun endings were selected to see whether (a) speakers make use of morphological cues when accessing gender or

(b) speakers rely on noun frequency (regardless of the morphological ending of the noun) providing more evidence of an abstract lexical route to access gender. Based on this, one of the research questions of the present investigation was whether morphological endings affect gender retrieval.

Based on previous research, the impact of the morphological cues on gender retrieval and lexical access can have three outcomes. Following previous studies that support the reliable cue hypothesis (see Section 2.7.1; Bates et al., 1996; Taft and Meunier, 1998), which posits that morphology is the most reliable cue to access gender, participants would be expected to retrieve gender faster with transparent nouns than with opaque nouns, showing that gender is more easily retrieved with the help of morphological cues.

The second possible outcome is that gender will be retrieved without the help of noun morphological endings via the lexico-syntactic route. In this case, following previous studies that found no effect of gender cues in nouns (see Section 2.7.1; Badecker et al, 1995; Miozzo and

Caramazza, 1997; Vigliocco et al., 1997), we would expect that gender would be retrieved faster in high frequency words regardless of noun morphological endings.

A final alternative outcome, which may be partially compatible with the two previous views, is that the retrieval of gender features will depend on both morphological cues and lexical information, assuming a Two-Route Hypothesis (see Section 2.7.1, Gollan & Frost, 2001). If we assume that both morphological and lexical cues play a role in retrieving gender, we will expect that the gender of high frequency transparent and low frequency transparent nouns will be retrieved faster than high frequency and low frequency opaque nouns since a second morphological route will play a role in the retrieval of gender within their lexical range, and HF and LF transparent nouns will be retrieved faster than HF and LF opaque nouns, corroborating the two-route hypothesis.

Our third research question in the present investigation was to compare monolingual speakers’ performance with the bilingual group and observe whether HSs with different language dominance differ in the use of noun morphological endings in gender retrieval. Since there is no online research that compares these three groups, our predictions are based on previously mentioned online studies (see section 2.6.1) with monolinguals that found morphophonological effects. For the HSs group, since there are no studies that measure RTs with different noun morphology or studies that compared two different HSs population we based our predictions on previous off-line studies that found morphology effects in the accuracy results of the HSs (oral production: Alarcón, 2011; Montrul, 2013; Montrul, 2008). Therefore, we hypothesize that morphological endings (transparency) will be predictive of higher accuracy rates but also, morphology effects will be seen in faster reaction times in both HSs groups. Dominance plays a role (as it plays in studies of lexical frequency-reviewed below) and we might find no effect of morphological endings for the less dominant bilinguals as reported in other research investigating other linguistic populations, such a L2 learners, a linguistic population with less

exposure in Spanish, therefore with less linguistic dominance in Spanish (Alarcón, 2009). Based on everything is published , based on the difference reported between monolinguals and HSs

(Gollan & Silverberg, 2001; Gollan et al., 2002; Gollan et al., 2005; Gollan, Montoya, Cera, &

Sandoval, 2008; Sandoval, Gollan, Ferreira, & Salmon, 2010), more dominant Spanish should pattern more with monolingual speakers and it will be reasonable to assume that less Spanish dominant bilingual might behave with L2.

3.1.2 Frequency Effects on Gender Access in Monolingual and HSs of Spanish

Noun frequency was manipulated primarily to test whether morphological cues are used more prevalently than frequency information stored in the noun lexico-syntactic information when retrieving gender. However, due to the limited research on noun frequency and the retrieval of lexico-syntactic information with noun frequency, the present investigation will also investigate the role of noun frequency in monolingual and bilingual speakers when accessing gender (our research question 4). We used the same tasks, a PNT and a PDT, to investigate the effects of noun frequency in the retrieval of gender features.

The effects of noun frequency in gender access with monolingual speakers of Spanish have been tested only by Navarrete, Basagni, Alario & Costa (2007) in a gender decision task and picture description task using a sequence of determiner + copula + adjective (new/old).

Results in this experiment corroborated frequency effects when accessing gender (See section

2.8.2. for a review of the study). We hypothesize that we will provide more evidence for the role of noun frequency on the retrieval of gender features, expanding the results found in Navarrete,

Basagni, Alario & Costa (2007). In other words, we hypothesize faster response times for higher frequency (HF) than lower frequency (LF) words in all speaker groups in this study.

This research questions is connected to the impact of frequency in different linguistic populations, especially in bilinguals with different language dominance, a topic that has recently

been the focus of many studies and hypotheses. According to the Frequency-Lag hypothesis and its antecedent, the Weaker Links Hypothesis (Gollan & Silverberg, 2001; Gollan et al., 2002;

Gollan et al., 2005; Gollan, Montoya, Cera, & Sandoval, 2008; Sandoval, Gollan, Ferreira, &

Salmon, 2010), noun frequency effects show a greater magnitude difference in bilinguals than monolinguals. In turn, less dominant bilinguals show the highest magnitude frequency effects between HF and LF nouns. Additionally, monolinguals have the highest activation with HF words. Based on this hypothesis, we wanted to investigate whether noun frequency plays a role in the retrieval of gender features in bilingual speakers with different linguistic dominance.

We hypothesize that, similar to previous research with monolingual speakers (Navarrete,

Basagni, Alario & Costa, 2007) and previous research on lexical access with monolingual and bilingual speakers (Gollan & Silverberg, 2001; Gollan et al., 2002; Gollan et al., 2005; Gollan,

Montoya, Cera, & Sandoval, 2008; Sandoval, Gollan, Ferreira, & Salmon, 2010), we will find a frequency effect in the retrieval of gender features. Monolinguals will show the fastest naming times with HF and LF words and the smallest magnitude difference between HF and LF words.

Among the bilingual groups, both bilingual groups will show the greater difference with LF nouns. Within the bilingual groups, the bilingual group with highest dominance will show the fastest naming times with HF and LF nouns and consequently the smaller difference between HF and LF nouns. In other words, the Spanish-dominant bilingual group will show greater activation of HF words than the less Spanish-dominant bilingual group, also leading to faster agreement responses than the less dominant group. The less Spanish-dominant bilingual group will show the least activation with HF nouns, thus the slowest response times. This impact and difference will be more visible with LF words where the group will show the smallest activation levels showing slower response times than the other groups.

The research questions addressed in the present investigation are again summarized here:

RQ (1): What is the role of noun morphology in gender agreement processing?

RQ (2) Are the morphology effects similar to other dual-route models?

RQ (3): Do different linguistic populations differ in the use of noun morphology when accessing gender?

RQ (4): Does noun frequency impact the retrieval of gender features in different linguistic populations?

3.2 Methods

3.2.1 LEAP-Q and Language Background Questionnaire

Participants complete the Language Experience and Proficiency Questionnaire (LEAP-Q;

Marian, Blumenfeld, & Kaushanskaya, 2007) in Spanish. This questionnaire elicits self-reported information about bilingual linguistic background such as age of acquisition or history of past and present language exposure of the bilingual in different contexts but also provides information about the bilingual’s self-reported language proficiency and dominance. It is considered a reliable questionnaire, as its developers corroborated its validity with a large sample pool of bilingual participants comparing the questionnaire with other standardized proficiency measures, and it was shown to be predictive of bilingual linguistic performance in both languages (Kaushanskaya and Yoo, 2011). Apart from the LEAP-Q, participants also fill a short language background questionnaire (LBQ) (Appendix E and F) that was designed by the experimenter to elicit other information that the LEAP-Q did not prompt such as Spanish and

English usage, schooling in Spanish and English, time spent in a Spanish and/or English country, parents and grandparents’ place of birth, and writing proficiency.

3.2.2 Standard Proficiency Measures: DELE, MELICET and MINT

To assess proficiency in Spanish, monolinguals and bilinguals completed the DELE

(Diploma de Español como Lengua extranjera) (Appendix H), a standardized Spanish grammar exam usually administered to L2 speakers of Spanish. Among other versions of DELE, the version that was used in the experiment is an adapted test that consists of three sections, all multiple choice, with a total of 50 items that test grammar and vocabulary. The results of this test can be seen in Table 3-3 in the participants’ section.

We measure English proficiency in the bilingual population by administering an adapted version of the standardized Michigan English Language Institute College English Test

(MELICET) (Appendix G). The version that was used in the experiment consists of two parts, all multiple choice, with a total of 50 items that test grammar and vocabulary. The results of this test can be seen in Tables 3-1 and 3-2 in the participants’ section.

Also, to determine Spanish dominance, we administered the Multilingual Naming Task

(MINT) (Gollan, Weissberger, Runnqvist, Montoya and Cera, 2012) to bilinguals. This test is similar to the Boston Naming Task (BNT; Kaplan, Goodglass and Weintraub, 1983), but it was designed for bilingual speakers. As mentioned by the developers of the MINT, the BNT underestimates bilinguals’ ability in Spanish. The Multilingual Naming Task has been proven to be a better evaluation of bilingual performance than the BNT; it avoids cognates, maximizes proficiency in language-specific knowledge, and diminishes influence from the non-target language (Gollan et al, 2012).

The MINT consists of an off-line PNT presented on a computer using a Power Point presentation that consists of 68 pictures. Bilingual speakers are asked to name in Spanish the images in the pictures as fast as possible. As instructed, the experimenter would provide

participants up to 6 seconds to name each image, and then proceed to show the next picture. The results of this test can be seen in Tables 3-1 and 3-2 in the participants’ section.

3.2.3 Vocabulary Test Experiment 2

This vocabulary test consists of providing the names with the corresponding article of the experimental pictures that appear in Experiment 2 (Appendix J and K). The vocabulary test was carried out because Experiment 2 was a Picture Description Task, and participants did not provide the name of the item that they saw on the screen, instead they just describe whether the picture was clear or blurry (see Appendix C). Therefore, participants needed to provide the name and article of the experimental pictures to verify that they were familiar with the name of the picture, and hence, its grammatical gender. The results of this vocabulary test were considered during that data coding to discard the noun trials where the speakers did not know the target referent or provide the incorrect article.

3.2.4 Participants

A total of 34 monolingual Spanish-speaking undergraduate students from Pontificia

Universidad Católica Madre y Maestra (PUCMM) in Santiago de los Caballeros, Dominican

Republic (14 men, mean age= 20.8 years) voluntarily took part in the study. Monolingual participants were recruited via university professors from the Department of Spanish and the

Department of Curriculum Design at PUCMM. A total of 44 Spanish-English bilingual speakers from the University of Florida (17 men, mean age=19.8 years) participated voluntarily or participated for class credit in the present investigation. Monolingual and bilingual speakers were not significantly different in mean age (p=0.06). We use the Multilingual Naming Test (Gollan,

Weissberger, Runnqvist, Montoya and Cera, 2012) to divide the bilingual group in two linguistic groups: a more dominant Spanish-speaking bilingual group (+dominant) and less dominant

Spanish-speaking bilingual group (-dominant) (MINT scores are shown in table 3-4 and the

MINT test is further explained below). Bilingual participants were recruited with the help of university professors, graduate teaching assistants and Hispanic student associations at the

University of Florida. Both groups were recruited by means of an email announcement or in- class recruitment stating the conditions to participate in the study.

According to the information provided in the language background questionnaires (a detailed explanation of all experimental tests and measures is given in the following section), all monolingual participants were born in the Dominican Republic and have lived there for their entire life. All monolingual participants were pursuing their Bachelor degree at PUCMM and one of them was completing their Master’s degree. Bilingual participants were all born in the United

States. Following Silva-Corvalan’s criteria for classifying generation in HSs, twenty-seven bilingual participants were 2nd generation HSs since they were born in the US but not their parents. Seventeen bilingual participants were 3rd generation HSs since they were born in the US as well as their parents. The group of bilingual speakers determined as more Spanish dominant

(+dominant) (determined by Multilingual Naming Test, explained below) had nineteen 2nd generation HSs, three 3rd generation HSs and two participants where one parent was born in the

United States (We have included these participants as 2nd generation because they were born in the United States and one of the parents was born in the United States). In the less dominant group (-dominant), there were eight 2nd generation HSs , ten 3rd generation HSs and two participants who had one parent that was born in the United States (i.e., 2nd generation). All bilingual speakers were completing their Bachelor’s degree at the University of Florida.

Appendix A and B summarize monolingual and bilingual participants’ characteristics.

Below I will summarize participants’ characteristics subtracted from the Language

Background Questionnaire(LBQ), the LEAP-Q (Language Experience and Proficiency-

Questionnaire, Marian, Blumenfeld, and Kaushanskaya, 2007) and the results from the standard proficiency measures: DELE (Diploma de Español como Lengua Extranjera; Ministry of

Education, Sports and Culture, Spain), MELICET (Michigan English Language Institute College

English Test; English Language Institute. University of Michigan) and MINT (Multilingual

Language Test; Gollan et al., 2012). The MINT test was administered to divide our HSs group in two groups (+ dominant and -dominant) based on their dominance in Spanish. Table 3-1 and 3-2 provide overall participants’ average data and p-values.

All monolinguals self-reported to be dominant in Spanish, even when they self-reported to have some proficiency in English or another language. All bilingual participants reported to be dominant in English. Scores on the MINT task, the other measure to assess linguistic dominance, classified the bilingual participants in a more dominant (+dominant) and less dominant group (- dominant) (See Table 3-2 for MINT mean scores and p-values). In the LBQ, monolingual and bilingual participants assessed their frequency of usage of Spanish and English on a scale of 1 to

8 (where 1 was “everyday” and 8 was “never”). Self-rated Spanish usage was greater in the monolingual speaker group: speaking (=1.05), reading (=1.7) and writing (=1.02) than the bilingual speaker group (both groups): speaking (=2.1), reading (=3.6) and writing (=2.7). Within the bilingual group, (+dominant) bilingual speakers self-reported a high frequency of use of

Spanish: speaking (=1.58), reading (=2.8) and writing (=2.04) than the (-dominant) bilingual speakers group: speaking (=2.8), reading (=4.7) and writing (=3.55). In the LEAP questionnaire,

MSs self-reported their competence in Spanish and HSs self-reported their competence in

Spanish and English. On a scale from 1 to 10 (where 1 was “no competence at all” and 10 was

“perfect competence”) MSs self-reported a higher competence in Spanish: speaking (=9.2) comprehension (=9.2) and reading (=9.2) than the HSs group (both groups): speaking (=8.1)

comprehension (=8.7) and reading (=8.3). Within the HSs group, HSs (+dominant) self-assessed to have a high competence in Spanish: speaking (=8.7) comprehension (=9.5) and reading (=9.1), than the HSs (-dominant) group: speaking (=7.4) comprehension (=7.8) and reading (=7.4).

Monolingual and bilingual participants self-reported their exposure of Spanish in different linguistic contexts (family, friends, watching TV and listening to music, radio etc.). Monolingual mean rates were higher in all linguistic contexts: family (=9.96), friends (=9.5), watching TV

(8.02=) and listening to music, radio etc. (8.1=) than the HSs (both groups): family (8.4=), friends (=4.09), watching TV (=3.8) and listening to music, radio etc. (=4.6). Within the HSs group, HSs (+dominant) mean rates were higher in all linguistic contexts: family (=9.25), friends

(=5.6), watching TV (=5.2) and listening to music, radio etc. (=5.6), than HSs (-dominant): family (=7.4), friends (=2.2), watching TV (=2.15) and listening to music, radio etc. (=2.8).

Results on the DELE showed that MSs scored higher (=46.55), than HSs (both groups) (=24.45).

Among the HSs, HSs (+dominant) mean rates were higher (=25.3) than HSs (-dominant)

(=23.35), however, p-values showed no statistical significance between the bilingual group.

DELE results correlated with the self-report competence results (in comprehension, speaking and reading) of both MSs and HSs (both groups) and HSs (+dominant and -dominant; see Table 3-4 for mean rates and p-values). Results on The Multilingual Naming Task (MINT; Gollan et al.,

2012) showed that, among the HSs, 24 were more dominant (+dominant) in Spanish (based on a score higher than 34 in the MINT), than the other 20 HSs (-dominant) in Spanish (who scored lower than 34). MELICET mean rates were slightly higher in the HSs (+dominant) group

(=44.41) than the HSs (-dominant) group (=44.55), however, p-values showed no statistical significance between the two groups. Below, Table 3-1 and 3-2 summarize the results of the tests mentioned.

Table 3-1. Mean averages and p-values for MSs and HSs (both groups) results in the LEAP, LBQ and DELE LBQ and LEAP MSs HSs p-value Age 20.8 19.8 0.06 Self-reported dominance Spanish English n/a Spa. frequency of use (writing) 1-8 scale* 1.02 2.7 <0.001*** Spa. frequency of use (speaking) 1-8 scale* 1.05 2.1 <0.001*** Spa. frequency of use(reading) 1-8 scale* 1.7 3.6 <0.001*** Only LEAP Spa. self-rated prof. (speaking) 1-10 scale 9.2 8.1 <0.001*** Spa. self-rated prof. (comprehen.) 1-10 scale 9.2 8.7 <0.001*** Spa. self-rated prof. (reading) 1-10 scale 9.2 8.3 <0.001 *** Exposure/usage to Spanish (family) 1-10 scale 9.96 8.4 <0.001*** Exposure/usage to Spa. (friends) 1-10 scale 9.5 4.9 <0.001*** Exposure to Spanish (watching TV) 1-10 scale 8.02 3.8 <0.001*** Exposure to Spa. (music/radio) 1-10 scale 8.1 4.6 <0.001*** Objective proficiency measures Prof. In Spa. (DELE) (out of 50) 46.55 24.45 <0.001*** *1=frequency,8=infrequency (see paragraph above for details on this test)

Table 3-2. Mean averages and p-values for the two bilingual groups’ results in the LEAP, LBQ, MINT and MELICET HSs HSs LBQ and LEAP (+dominant) (-dominant) p-value Age 19.9 19.75 0.463 Self-reported dominance English English Spa. frequency of use (writing) 1-8 scale* 2.04 3.55 <0.001*** Spa. frequency of use (speaking) 1-8 scale* 1.58 2.8 <0.001*** Spa. frequency of use(reading) 1-8 scale* 2.8 4.7 <0.001*** Eng. frequency of use (writing) 1-8 scale* 1.45 1.25 0.32 Eng. frequency of use (speaking) 1-8 scale* 1 1.1 0.16 Eng. frequency of use (reading) 1-8 scale* 1.54 1.4 0.54 Only LEAP Spa. self-rated prof. (speaking) 1-10 scale 8.7 7.4 <0.001*** Spa. self-rated prof. (comprehen.) 1-10 scale 9.5 7.8 <0.001*** Spa. self-rated prof. (reading) 1-10 scale 9.1 7.4 <0.001*** Eng. self-rated prof. (speaking) 1-10 9.7 9.8 0.49 Eng. self-rated prof. (comprehend.) 1-10 scale 9.9 9.95 1 Eng. Self-rated prof. (reading) 9.75 9.95 0.18 Exposure/usage to Spanish (family) 1-10 scale 9.25 7.4 0.01* Exposure/usage to Spa. (friends) 1-10 scale 5.6 2.2 <0.001*** Exposure to Spanish (watching TV) 1-10 scale 5.2 2.15 <0.001*** Exposure to Spa. (music/radio)1-10 scale 6.25 2.8 <0.001***

Table 3-2 Con’ mean averages and p-values for the two bilingual groups’ results in the LEAP, LBQ, MINT and MELICET HSs HSs LBQ and LEAP (+dominant) (-dominant) p-value Exposure to Eng. (family) 1-10 scale 1.91 5.55 <0.001*** Exposure to Eng. (friends) 1-10 scale 7.3 9.8 <0.001*** Exposure to Eng (watching TV) 1-10 scale 7.8 9.2 0.03* Exposure to Eng. (music/radio) 1-10 scale 7.75 9.15 0.02* Objective proficiency measures Prof. In Spa. (DELE) (out of 50) 25.3 23.35 0.07 Prof. In Eng. (MELICET) (out of 50) 44.41 44.55 0.62 Dominance in Spa. (MINT) (out of 68) 45.6 28.3 <0.001*** *1=frequency,8=infrequency (see paragraph above for details on this test)

Two sample t-tests were carried out to see whether above mean rates were statistically different from each other.

3.3 Materials and Design

The materials for the PNT and PDT consisted of color drawings depicting the stimuli selected for the experiment (see Table 3-3 and 3-4 for the lists of experimental stimuli). All experimental pictures were color pictures obtained from Google images with a resolution larger than 20 pixels per cm. Images were edited using picture editing software (Microsoft Paint 2010, version 1709) so that all pictures were no larger than 600 x 600 pixels and had the same width and height dimensions. In experiment 2, thirty-six pictures were blurred using an online photo editor (LunaPic photo editor, 2018). Only the filler pictures were blurred (See Appendix C for experiment sequence sample). To ensure that the pictures chosen for the experiments were representative of the thing they depicted, as well as to avoid dialectal variation, 10 speakers from different varieties (Cuba, Mexico, Venezuela, Puerto Rico and Colombia) participated in a pilot study to norm the picture names. They also checked for possible errors in the pictures or instructions.

The stimuli consisted in two experimental lists, with each list containing 44 experimental items, 88 fillers (Appendix I) and 10 practice pictures, resulting in a total of 142 pictures. Each list had different words and they were created to be administered in the two different experiments carried out in the present investigations. Cognates and opaque nouns with morphological endings predictive of a specific gender (e.g. -dad is 99.7% feminine) were excluded. The morphological endings of the stimuli used in this study consisted in endings, such as -e, -z or -j (see Tables 3-3 and 3-4 below for experimental stimuli) and varied across masculine and feminine nouns; therefore, gender is not predicted by these morphological endings in the experimental tasks (see Table 2-1 for morphological endings in Spanish and their percentage of usage with masculine and feminine nouns).Half of the experimental items were feminine (22) and half masculine (22), however, feminine and masculine words were not balanced across conditions due to lack of stimuli, particularly, feminine low frequency opaque nouns.

Table 3-3. Transparent and opaque, high and low frequency, feminine and masculine nouns used as experimental stimuli in List 1 HT English LT English HO English LO English toro (m) Bull bufanda(f) scarf flor (f) flower coliflor (f) cauliflower vestido(m) Dress pulpo (m) octopus puente(m) bridge buzón(m) mailbox arco (m) Arch fresa(f) strawberry juez (m) judge tenedor(m) fork cuello(m) Neck bota(f) boot león(m) lion cicatriz (f) scar playa(f) Beach flecha (f) arrow fuente (f) fountain guante (m) glove ojo(m) Eye trofeo (m) trophy red (f) net imán(m) magnet caja(f) Box huevo (m) egg sol (m) sun barril (m) barrel corona(f) crown hueso(m) bone corazón (m) heart botón (m) button bolsa(f) Bag vaso (m) glass noche (f) night miel (f) honey estrella(f) Star hombro(m) shoulder frente(f) forehead nariz (f) nose luna(f) Moon hoja(f) leaf muerte(f) death llave (f) key

Table 3-4. Transparent and opaque, high and low frequency, feminine and masculine nouns used as experimental stimuli in List 2 HT English LT English HO English LO English lluvia (f) Rain uña (f) nail traje (m) suit maiz(m) corn brazo(m) Arm búho(m) owl paquete(m) parcel pincel(m) brush caballo(m) horse pera(f) pear leche(f) milk nuez(f) nut castillo (m) castle corbata(f) tie tren(m) train jabón(m) soap mesa(f) table codo(m) elbow raíz(f) root lápiz(m) pencil banco(m) bank barba(f) beard árbol(m) tree pez(m) fish fuego(m) fire pierna(f) leg torre(f) tower serpiente(f) snake lengua(f) crown cuchillo(m) snife carne(f) meat nube(f) cloud puerta(f) door pluma(f) feather sangre(f) blood reloj(m) watch libro(m) book plato(m) plate cruz(f) cross nieve(f) snow casa(f) house casco(m) helmet pie(m) foot sal(f) salt

The lists were controlled for noun frequency, letter length, noun morphology, gender, imageability and concreteness (See Table 3-5 below). Noun frequency, imageability and concreteness were obtained from the ESPAL database (Duchon, Perea, Sebastián-Gallés, Martí,

Carreiras, 2013; retrieved from http://www.bcbl.eu/databases/espal/index.php). High Frequency nouns were those nouns that had a frequency of more than 30 per million occurrences, and low frequency nouns had fewer than 30 per million occurrences. The experimental stimuli in each list contained 11 high frequency opaque nouns, 11 low frequency opaque nouns, 11 high frequency transparent nouns, and 11 low frequency transparent nouns. Each participant saw a different list in each experiment. The order of the lists were administered to the participants arbitrarily. For instance, if participants saw List 1 in experiment 1, then List 2 was used for Experiment 2 and if

List 2 was administered in Experiment 1, then List 1 was used for Experiment 2.

Gender was also balanced through the experimental stimuli so that half of the experimental stimuli were feminine, and half were masculine nouns. The fillers were also controlled for frequency, morphology, gender, as well as imageability, concreteness and letter length. Since there were not enough high frequency opaque nouns to use as fillers, cognates and

morphological opaque nouns with endings predictive of a particular gender, e.g. –dad, -ión, were included. Overall, taking into account the experimental stimuli and the fillers, half of the items were feminine and the other half masculine, half of the items were high frequency words and the other half low frequency words, and half of the stimuli were opaque nouns and the other half transparent nouns. Table 3-5 shows mean averages of picture nouns characteristic of the experimental stimuli. Two-tailed t-tests corroborated that there were not significant differences between the two lists.

Table 3-5. Standard deviation, mean averages and P values of two tailed t-test comparing list 1 and list 2 of experimental items Characteristics SD (mean) List 1 SD (mean) List 2 P-values Frequency (ESPAL) 78.9 (55.04) 72.39 (55.05) 0.98 # of Letters 1.32 (4.9) 1.2 (4.8) 0.69 Imageability 1.21 (6) 0.47 (6.1) 0.85 Concreteness 0.64 (4.9) 0.55 (4.8) 0.32

Table 3-6. Standard deviation, mean averages and P values of two tailed t-test comparing List 1 and List 2 of filler items

Characteristic SD (mean)List 1 SD (mean) List 2 P-values Frequency (ESPAL) 96.11 (48) 234.71 (48.8) 0.96 # of Letters 1.55 (5.6) 1.38 (5.6) 0.84 Imageability 0.69 (5.9) 0.56 (6) 0.15 Concreteness 0.61 (5.8) 0.66 (5.8) 0.59

3.4. Procedure

The two background questionnaires were administered first, and then the experimental tasks were administered. Pictures from both experimental tasks were presented using E-Prime

2.0 (SP1, Psychology Software Tools, Pittsburgh, PA), with the monolingual data presented on a computer laptop with Intel ® Core™ i7-4610m CPU 3.00 GHz processor and Windows 10 enterprise operating system. The laptop computer had a 15-inch monitor with a data resolution of

1600 x 900 pixels. The bilingual data were presented on a desktop computer with an Intel ®

Core™ i7-4770 CPU 3.40 GHz processor and Windows 7 operating system. The desktop computer had a 22 -inch LCD monitor with a data resolution of 1920x1080 pixels. Naming times were recorded using an Audio-Technica AT1200 microphone connected to a response box.

Correct and incorrect responses were verified manually. Each trial began with a 500 ms fixation cross (+) in the middle of the screen immediately followed by the picture. Participants started the experiment by pressing the space bar. The pictures appeared in color at the center of the screen.

In experiment 1, participants were instructed to name the pictures with the corresponding definite article el-DET.MASC.SG or la-DET.FEM.SG (the) as quickly as possible using one word. They were also instructed not to name the pictures using plurals or phrases such as la mesa de comer (“dining- room table”, literally “the table for eating”). Participants were also told that if they did not know or recognize the object/thing displayed on the screen, they should respond no sé (“I don’t know”) or remain silent until the next picture. Participants completed a 10-item practice set and were asked whether they had any questions or concerns at the end of the practice session. The first experiment lasted around 15-20 minutes.

Experiment 2 was a picture description task (PDT). Participants were presented with a picture stimulus on the screen and were instructed to use borroso/borrosa (“blurry-ADJ.MASC.SG

/blurry-ADJ.FEM.SG”) or claro/clara (“clear- ADJ.MASC.SG /clear- ADJ.FEM.SG”) to describe what appeared on the screen. For instance, if a blurry cereza (cherry-N.FEM.SG) appeared on the screen, participants had to respond using exclusively the adjective inflected for gender, borrosa. (blurry-

ADJ.FEM.SG). If a clear image of a flor (flower-N.FEM.SG) appeared on the screen, participants were instructed to respond clara (clear- ADJ.FEM.SG) The stimuli (words/pictures) used for both experiments were counter-balanced such no participant saw an item repeated for both experiments (See Appendix C for an example of two sequences introducing experimental

stimuli) Participants were also told that if they did not know or did not recognize the object displayed on the screen they could respond no sé (“I don’t know”) or wait for the picture to disappear from the screen. Similar to Experiment 1, participants completed a 10-item practice set prior to the main experiment and were asked whether they had any questions or concerns at the end of the practice session. The second experiment lasted around 15-20 minutes. Instructions were given in Spanish and the experimenter used Spanish throughout the whole session.

After completing the two experimental tasks, monolingual participants completed a

LBQ, DELE, LEAP-Q and a vocabulary test. HSs participants completed a bilingual LBQ,

DELE, MELICET, a vocabulary test and the MINT. The MINT was the test administered to evaluate language dominance in Spanish.

In this chapter, I introduced the study’s research questions, which were motivated by the conclusions from the previous literature review. The following chapter provides the analysis of the findings that resulted from the designed methodology presented above that tested one group of monolinguals and two groups of HSs of Spanish with different language dominance.

CHAPTER 4 EXPERIMENT 1 RESULTS: PICTURE NAMING

In this experiment, participants were instructed to name pictures with their corresponding definite article el-DET.MASC.SG. or la-DET.FEM.SG. “the” as quickly as possible using one word.

Participants were also told that if they did not know or recognize the object/thing displayed on the screen, they should respond no sé “I don’t know” or remain silent until the next picture.

Reaction time (RT) analyses were conducted on correct responses only. A response was accepted as correct if participants could name the picture that appeared on the screen with the correct article el or la before the picture disappeared from the screen (3000 ms.). Responses with un or una were also accepted. Correct responses after self-correction were not accepted for the

RT analysis. Answers that were not clear due to lack of audibility or unclear pronunciation were discarded.

Accuracy analyses were conducted on trials where participants named the noun with the

(correct/incorrect) article. A response was accepted as correct if participants could name the picture that appeared on the screen with the correct article el or la. Responses with un or una were also accepted. Accurate responses that were outside the RT frame (3000 ms.) or were not recorded by the program were also coded for the accuracy analysis. Correct responses after self- correction were accepted for the accuracy analysis. Answers that were not clear due to lack of audibility or unclear pronunciation were discarded.

The independent variables investigated in both analyses were Morphology and Frequency

(within-subject variables) and group (between-subject variable): Monolingual speakers, more

Spanish-dominant bilinguals (D+ bilinguals) and less Spanish-dominant bilinguals (D- bilinguals)-. We carried out a repeated measures ANOVA to obtain the statistical significance of

the data. Below, section 4.2. outlines the results for the RT analysis and section 4.3. outlines the results for the accuracy analysis.

Monolingual group data

From a total of 1495 possible trials, accuracy data consisted in 1275 data points. There were no inaccurate trials for monolinguals speakers, that is, the noun was always named with the corresponding article. Skipped trials or non-identifiable responses consisted in 178 data points

(12%) (these were excluded from the analysis). From a total of 1495 possible trials, RTs data consisted in 1243 data points. Data points outside the 3000 ms window, non-recorded responses, skipped trials, inaccurate or non-identifiable responses constituted 209 data points (14%). Target responses were equally distributed over the 4 conditions in experiment 1: High frequency transparent nouns: HT (314, 25%); high frequency opaque nouns: HO (328, 25%); low frequency transparent nouns: LT (322, 25%); and low frequency opaque nouns: LO (310, 25%).

Also, accurate responses consisted in 674 feminine tokens and 600 masculine tokens.

Bilingual group data

For D+ bilinguals, from a total of 1057 trials, accuracy data for this group consisted of

843 data points. Inaccurate tokens (determiner-noun mismatches) were 35 data points (3%) (32 feminine, 3 masculine; 2 HT, 5 HO, 10 LT and 18 LO tokens). Skipped trials or non-identifiable responses constituted 178 data points (16%) (those were not included in the analysis). From a total of 1057, RTs data consisted in 692 data points. Data points outside the 3000 ms. window, non-recorded responses, skipped trials, inaccurate or non-identifiable responses constituted 365 data points (34.5 %). Determiner-noun mismatch responses constituted 35 data points (4%).

Target responses were unequally distributed over the 4 conditions for the D+ bilingual group: HT

(210, 26%); HO (236, 29%); LT (202, 25%); and LO (159, 19%). Also, accurate responses consisted in 432 feminine tokens and 375 masculine tokens.

For D- bilinguals, from a total of 880 trials, accuracy data for this group consisted of 471 data points. Inaccurate trials (determiner-noun mismatches) consisted in 98 data points (88 feminine and 10 masculine tokens; 7 HT, 19 HO, 28 LT and 43 LO). Skipped trials or non- identifiable responses constituted 311 data points (35%) (those were excluded for the analysis).

From a total of 880, RTs data consisted in 398 data points. Data points outside the 3000 ms. window, non-recorded responses, skipped trials, inaccurate or non-identifiable responses constituted 481 data points (54 %). Target responses were unequally distributed over the 4 conditions for the D- bilingual group: HT (132, 16 %); HO (157, 20%); LT (101,12 %); and LO

(81, 10%). Furthermore, accuracy data consisted in 223 feminine tokens and 248 masculine tokens.

4.1 RT Analysis

4.1.1 Results: Monolingual, D+ and D- Bilinguals

Mean picture naming times indicated that monolinguals were on average faster than the two bilingual groups with both high and low frequency nouns. Between the bilingual group, the more Spanish-dominant bilingual group (D+ bilinguals) was on average faster than the less

Spanish-dominant bilingual group (D- bilinguals) who showed, on average, the slowest naming times. Also, numerically, the greater difference in naming times between high and low frequency nouns was shown by the less Spanish-dominant group. In terms of the morphology effect, we can observe that monolinguals were numerically faster naming transparent nouns with both high- and low- frequency nouns. However, for the bilingual groups, they seemed to name transparent nouns slightly slower than opaque nouns with high frequency nouns. With low frequency nouns, we can observe that both bilingual groups were numerically faster with transparent nouns. This

difference was more visible for the D- group. The naming times difference between low frequency opaque and transparent nouns was more visible with the less Spanish-dominant group.

Below, Table and Figure 4-1 show the mean picture naming times and standard error (SE) for

Experiment 1 RTs analysis.

Table 4-1. Experiment 1 mean RTs in ms. and SE split by frequency and morphology Group Freq. Morph. Means SE Mono H O 867.09 11.0 Mono H T 828.65 11.7 Mono L O 1201.77 21.6 Mono L T 1152.92 17.7

D+ H O 946.46 11.6 D+ H T 956.46 17.7 D+ L O 1663.92 20.1 D+ L T 1644.15 19.3

D- H O 1047.38 22.6 D- H T 1069.03 17.1 D- L O 1933.71 49.7 D- L T 1844.79 41.4

ms.

Figure 4-1. Experiment 1 mean RTs and SE split by frequency and morphology

To assess the statistical significance of the results, we conducted a repeated-measures

ANOVA using R (R core team, 2015, version 3.1.3). We had a 2 x 2 x 3 factorial design with the within-subjects factors of Frequency (High versus Low) and Morphology (Opaque versus

Transparent) and the between-subjects factor of Group (Monolinguals, Bilinguals D+, Bilinguals

D-). The results of this model are presented in Table 4-2.

Table 4-2. Experiment 1 three-groups 2 x 2 x 3 repeated-measures ANOVA output Effect DFn DFd F p ges Group 2 70 89.645 <0.001*** 0.568 Frequency 1 70 532.791 <0.001*** 0.722 Morphology 1 70 6.27 <0.001*** 0.006 group:Frequency 2 70 42.457 <0.001*** 0.292 group:Morphology 2 70 0.791 0.457 0.001 Freq:Morph 1 70 1.372 0.245 0.001 group:Freq:Morph 2 70 1.186 0.311 0.002

As the output shows, we find a main effect for Group (F(2,70) = 89.645, p = 0.001),

Frequency (F(1,70) = 532.791, p = 0.001), and Morphology (F(1,70) = 6.27, p = 0.001. There is also an interaction by Group x Frequency (F(2,70) = 42.457, p = 0.001). Because of this interaction, we will further explore the groups separately in the following sections, as well as conduct a bilingual comparison.

To summarize, the Group factor was significant and looking at Table and Figure 4-1, numerically the direction of the effect indicates that monolinguals were faster than both bilingual groups and within the bilingual group the more Spanish dominant group named faster than the less dominant Spanish group. Frequency was also significant indicating that lower frequency nouns were named slower than higher frequency nouns; however, the magnitude of this effect differs depending on the group as indicated by the significant interaction between Frequency and

Group. Finally, Morphology was also significant, indicating that transparent nouns were overall named faster than opaque nouns. We will explore further this interactions with the analysis of the subgroups.

4.1.2 Results: Monolingual Group

Looking to the individual performance of the monolingual group we observe that mean picture naming times showed that on average low frequency nouns were named slower than high frequency nouns for the monolingual population. Numerically, the monolingual group was faster with transparent nouns than opaque nouns with both high and low frequency nouns. We observed that the magnitude difference between opaque and transparent nouns was larger with low frequency than high frequency nouns.

To assess the statistical significance of the results, we conducted a repeated-measures

ANOVA using R (R core team, 2015, version 3.1.3). We had a 2 x 2 factorial design with the

within-subjects factors of Frequency (High versus Low) and Morphology (Opaque versus

Transparent). The results of this model are presented in Table 4-3.

Table 4-3. Experiment 1 monolinguals 2 x 2 repeated-measures ANOVA output Effect DFn DFd F p ges Frequency 1 32 62.519 <0.001*** 0.457 Morphology 1 32 5.741 <0.001*** 0.011 Freq.:Morph. 1 32 0.028 0.867 0.00005

As the output shows, we find a main effect for Frequency (F(1,32) = 62.519, p = 0.001) indicating that monolingual speakers were faster with high frequency nouns. Morphology

(F(1,32) = 5.741, p = 0.001) was also significant corroborating an effect of transparent morphology for monolinguals, that is, transparent nouns were read faster than opaque nouns.

There was no significant interaction between Frequency and Morphology.

4.1.3 Results: D+ and D- Bilinguals

Mean RTs showed that overall the more dominant Spanish bilinguals were, on average, faster than the less dominant Spanish group with both high- and low- frequency trials.

Numerically, mean picture naming times revealed a frequency effect. The magnitude difference naming high frequency and low frequency nouns was larger for D- bilinguals. The magnitude difference was greater for the D- group. In terms of morphology, high frequency opaque nouns were named slower than high frequency transparent by both groups. However, with low frequency nouns, we observe the reverse pattern, low frequency transparent were named slower than low frequency opaque nouns. Table 4-1 and Figure 4-1 above illustrates these individual numerical patterns.

To assess the statistical significance of the results, we conducted a repeated-measures

ANOVA using R (R core team, 2015, version 3.1.3). We had a 2 x 2 x 2 factorial design with the

within-subjects factors of Frequency (High versus Low) and Morphology (Opaque versus

Transparent) and the between-subjects factor of Group (Bilinguals D+, Bilinguals D-). The results of this model are presented in Table 4-4.

Table 4-4. Experiment 1 bilingual group comparison 2 x 2 x 2 repeated-measures ANOVA output Effect DFn DFd F P ges

Group 1 38 19.987 <0.001*** 0.233

Frequency 1 38 690.656 <0.001*** 0.827

Morphology 1 38 1.570 0.217 0.002 group:Frequency 1 38 6.882 <0.001*** 0.045 group:Morphology 1 38 0.835 0.366 0.001

Freq.:Morph. 1 38 2.684 0.109 0.005

Group:Freq.:Morph. 1 38 0.670 0.417 0.001

As the output shows and looking at the direction of the results in Table and Figure 4-1,

Group was significant (F(1,38) = 19.987, p = 0.001) indicating that D+ bilinguals were faster than D- bilinguals. Also, Frequency was significant (F(1,38) = 690.636, p = 0.001) indicating a frequency effect in all participants. Also, the interaction between Group and Frequency (F(1,38)

= 6.882, p = 0.001) was significant indicating that frequency effects were greater in the less

Spanish dominant group. We now present the analysis of each bilingual group separately to explore further this interaction.

Table 4-5. Experiment 1 D+ bilingual group 2 x 2 repeated-measures ANOVA output Effect DFn DFd F p ges Frequency 1 22 790.753 <0.001*** 0.897

Morphology 1 22 0.153 0. 699 0.0009

Freq.:Morph. 1 22 0.619 0.439 0.004

Table 4-6. Experiment 1 D- bilingual group 2 x 2 repeated-measures ANOVA output Effect DFn DFd F p ges Frequency 1 16 203.376 <0.001*** 0.775

Morphology 1 16 1.825 0.195 0.005

Freq.:Morph. 1 16 2.263 0.151 0.008

As the output shows, for both groups there are only main effects of Frequency.

To summarize, D+ bilinguals have a greater magnitude effect of Frequency than the D- group as shown by the group means in Table 4-1. Both bilingual groups showed Frequency effects where both groups overall named higher frequency noun trials faster than low frequency noun trials.

4.2 Accuracy Analysis

As reported above, accuracy analyses were conducted on the proportion of correct responses out of correct responses and determiner noun mismatches, that is, when participants name the noun with the incorrect determiner. A response was accepted as correct if participants could name the picture that appeared on the screen with the correct article el or la. Responses with un or una were also accepted. Accurate responses that were outside the RT frame (3000 ms.) or were not recorded by the program were also coded for the accuracy analysis. Correct responses after self-correction were accepted for the accuracy analysis. Answers that were not clear due to lack of audibility or unclear pronunciation were discarded. The independent

variables investigated were Morphology and Frequency effect (within-subject variable) and three different linguistic groups (between-subject variable): Monolingual speakers, more dominant

Spanish-English bilinguals (D+ bilinguals) and less dominant Spanish-English bilinguals (D- bilinguals). Below, we present the results of the accuracy analysis.

4.2.1 Results: Monolinguals and D+ and D- Bilingual Group

Mean accuracy rates reveal that monolingual scored at ceiling. Among the bilingual group, D- bilinguals were, on average, less accurate than D+ bilinguals. The D- group showed the largest frequency effects. Mean accuracy for noun morphology revealed that on average bilingual participants were more accurate with transparent nouns than opaque nouns. The larger magnitude difference between opaque and transparent nouns were seen with low frequency nouns.

Figure 4-2. Experiment 1 mean accuracy and SE split by frequency and morphology

100

To assess the statistical significance of the results, we conducted a repeated-measures

ANOVA using R (R core team, 2015, version 3.1.3). We had a 2 x 2 x 3 factorial design with the within-subjects factors of Frequency (High versus Low) and Morphology (Opaque versus

Transparent) and the between-subjects factor of Group (Monolinguals, D+ bilinguals, D- bilinguals). The results of this model are presented in Table 4-7.

Table 4-7. Experiment 1 three-groups 2 x 2 x 3 repeated-measures ANOVA output Effect DFn DFd F p ges group 2 74 84.946 <0.001*** 0.434 Frequency 1 74 43.229 <0.001*** 0.149 Morphology 1 74 15.393 <0.001*** 0.043 group:Frequency 2 74 29.507 <0.001*** 0.194 group:Morphology 2 74 9.550 <0.001*** 0.053 Frequency:Morphology 1 74 5.750 <0.001*** 0.010 group:Freq:Morph. 2 74 2.564 0.083 0.009

As the output shows and based on Figure 4-2, accuracy patterns all factors and interactions were significant except the interaction of Group x Frequency x Morphology. We have a main effect for Group (F(2,74) = 84.946, p = 0.001). Frequency (F(1,74) = 43.229, p =

0.001) was also significant and the interaction between Group and Frequency (F(2,74) p =

29.507). The interaction between Morphology x Frequency was also significant (F(1,74) =

5.750, p = 0.001). Given these interactions, we conducted separate analyses per group as well as a bilingual comparison analysis.

4.2.2 Results: Monolingual Group

As we can see in Figure 4-2 monolinguals scored at ceiling, which resulted in null results for our ANOVA model.

101

Table 4-8. Experiment 1 monolinguals 2 x 2 repeated-measures ANOVA output Effect DFn DFd F p ges

Frequency 1 32 NaNNaN 0

Morphology 1 32 NaNNaN 0

Freq.:Morph. 1 32 NaNNaN 0

4.2.3 Results: D+ and D- bilingual group

Mean accuracy showed that within the bilingual group D+ bilinguals were overall more accurate than D- bilinguals. Observing the mean accuracy for both bilingual participants, it seems that D- bilinguals showed a greater frequency effect. Mean accuracy for the Morphology effect showed that both groups were more accurate with transparent nouns.

To assess the statistical significance of the results, we conducted a repeated-measures

ANOVA using R (R core team, 2015, version 3.1.3). We had a 2 x 2 x 2 factorial design with the within-subjects factors of Frequency (High versus Low) and Morphology (Opaque versus

Transparent) and the between-subjects factor of Group (D+ bilinguals, D- bilinguals). The results of this model are presented in Table 4-9.

Table 4-9. Experiment 1 bilingual group comparison 2 x 2 x 2 repeated-measures ANOVA output Effect DFn DFd F P ges Group 1 42 47.995 <0.001*** 0.276 Frequency 1 42 42.937 <0.001*** 0.235 Morphology 1 42 15.289 <0.001*** 0.074 group:Frequency 1 42 15.092 <0.001*** 0.097 group:Morphology 1 42 4.288 <0.001*** 0.022 Freq:Morph 1 42 5.712 <0.001*** 0.018 group:Freq.:Morph. 1 42 0.463 0.499 0.001

102

As the output shows and based on the numerical patterns observed in Figure 4-2, we found that the group factor was significant (F(1,42) = 47.995, p = 0.001) indicating that D+ bilinguals were more accurate than the D- bilinguals. Also, we found Frequency main effect as well as an interaction by Group x Frequency (F(1,42) = 15.289, p = 0.001). Also, Morphology effect (F(1,42)=15.289, p = 0.001) was significant as well as the Frequency x Morphology interaction (F(1,42) = 5.712, p = 0.001) indicating a different morphology effect based on the frequency of the noun. We will further subdivide the two bilingual groups to further explore these interactions.

Table 4-10. Experiment 1 D+ bilingual group 2 x 2 repeated-measures ANOVA output Effect DFn DFd F p ges

Frequency 1 23 14.209 <0.001*** 0.151

Morphology 1 23 4.997 0.035* 0.055

Freq:Morph. 1 23 4.219 0.051 0.028

Table 4-11. Experiment 1 D- bilingual group 2 x 2 repeated-measures ANOVA output Effect DFn DFd F p ges Frequency 1 19 29.655 <0.001*** 0.322 Morphology 1 19 10.381 <0.001*** 0.102 Freq.:Morph. 1 19 2.596 0.123 0.018

Both analyses result in a main effect for Frequency for D+ bilinguals (F(1,23) = 14.204, p = 0.001) and D- bilinguals (F(1,19)=29.665, p = 0.001) as well as a main effect of Morphology for D+ bilinguals (F(1,23) = 4.997, p = 0.001) and D- (F(1,19) = 10.381, p = 0.001).

4.3 Summary of Results Experiment 1

For the RTs analysis of Experiment 1 we can observe that in the three-group analysis the

Group factor was significant. In the comparative analysis of both bilingual groups the Group

103

factor was also significant indicating an advantage in naming times for the more Spanish dominant group. We found frequency effects that were particularly larger for both bilingual groups. We could observe the largest frequency effects in the D- bilingual group. With regard to the effect of morphology we only found a main effect of morphology in the monolingual group in the RTs analysis. No significant interaction between morphology and frequency was found or any effect of morphology in either of the bilingual groups.

For the accuracy analysis, monolinguals scored at ceiling. Among the bilinguals, D- bilinguals were less accurate than D+ bilinguals, who had the lowest accuracy rates. Both bilingual groups show larger frequency effects, particularly the less dominant bilingual group showed the largest frequency effects. Also, we found a main effect of morphology in both groups indicating that both participants were more accurate with transparent nouns.

104

CHAPTER 5 EXPERIMENT 2 RESULTS: PICTURE DESCRIPTION

This experiment was a picture description task. Participants were presented with a picture stimulus on the screen and were instructed to use borroso/borrosa (“blurry-ADJ.MASC.SG. /blurry-

ADJ.FEM.SG.”) or claro/clara (“clear- ADJ.MASC.SG. /clear- ADJ.FEM.SG”) to describe what appeared on the screen only using the adjective. As in experiment 1, RTs were recorded from the onset of the picture. After experiment 2, a vocabulary task was carried out to assure that participants were agreeing with the correct referent. Only responses with the target referent were included in both analysis.

In experiment 2, RTs analyses were also conducted on correct responses. A response was accepted as correct if participants could describe the picture that appeared on the screen using claro/a (clear-ADJ.SING.MASC/FEM). Correct responses after self-correction were not accepted.

Answers which were not clear due to lack of audibility or unclear pronunciation were discarded.

Accuracy analyses were conducted on responses in which participants produced only claro/a (clear-ADJ.SING.MASC/FEM). Accurate responses that were outside the RTs frame (3000 ms.) or were not recorded by the program were also coded for the accuracy analysis. Responses that were not clear due to lack of audibility or unclear pronunciation were discarded.

The independent variables investigated in both analyses were Morphology and Frequency

(within-subject variables) and Group (between-subject variable): monolingual speakers, more dominant Spanish-English bilinguals (D+ bilinguals) and less dominant Spanish-English bilinguals (D- bilinguals). We carried out repeated measures ANOVAs to obtain the statistical significance of the data. Below, section 5.2 outlines the results for the RT analysis and section

5.3 outlines the results for the accuracy analysis.

105

Monolingual data

From a total of 1495 possible trials, accuracy data consisted in 1289 data points

(adjective matches and mismatches with the correct referent). Inaccurate trials consisted in 17 data points (1 masculine, 16 feminine; 1 HO, 4 LT and 11 LO tokens). Skipped trials or non- identifiable responses consisted in 206 data points (13%) from the total possible trials. From a total of 1495 RTs data consisted in 1237 data points. Data points outside the 3000 ms. window, non-recorded responses, skipped trials, inaccurate or non-identifiable responses constituted 258 data points (17%). Target responses were unequally distributed over the 4 conditions in experiment 2: High frequency transparent nouns: HT (328, 26%); High Frequency Opaque

Noun: HO (327, 26%); Low frequency transparent nouns: LT (316, 24%); and Low frequency opaque nouns: LO (301, 23%). Also, accurate responses consisted in 647 feminine tokens 625 masculine tokens.

Bilingual data

For D+ bilinguals, from a total of 1054 data points, accuracy data for this group consisted of 895 data points (adjective matches and mismatches with the correct referent). Inaccurate data consisted in 104 data points (85 feminine, 19 masculine; 38 LO and 17 LT tokens). Skipped trials or non-identifiable responses constituted 159 data points (15%). From a total of 1056 data points, RTs data consisted in 755 data points. Data points outside the 3000 ms. window, non- recorded responses, skipped trials, inaccurate or non-identifiable responses constituted 301 data points (28%). Target responses were unequally distributed over the 4 conditions for the D+ bilingual group: HT (239, 31%); HO (212, 27%); LT (161, 20%); and LO (178, 22%). Also, accurate responses consisted in 379 feminine tokens and 411 masculine tokens.

106

For D- bilinguals, from a total of 880 data points, accuracy data for this group consisted of 639 data points (adjective matches and mismatches with the correct referent). Inaccurate responses constituted 169 data points (141 feminine, 29 masculine; 8 HT, HO 34, 55 LT, 73 LO tokens) Skipped trials or non-identifiable responses constituted 241 data points (28%). From a total of 880 data points, RTs data consisted in 416 data points. Data points outside the 3000 ms. window, non-recorded responses, skipped trials, inaccurate or non-identifiable responses constituted 435 data points (49%). Target responses were unequally distributed over the 4 conditions for the D- bilingual group: HT (177, 38%); HO (146, 32%); LT (63, 13%); and LO

(83, 17%). Also, accurate responses consisted in 202 feminine tokens and 266 masculine tokens.

5.1 RT Analysis

5.1.1 Results: Monolingual, D+ and D- Bilingual Group

Overall mean accuracy showed that high frequency conditions were responded to more accurately than low frequency conditions. Both bilingual groups had greater difference in accuracy between high frequency and low frequency nouns than the monolingual group and, among the bilingual groups, the D- bilingual group showed the greatest difference. Numerically, among the bilingual groups, the slowest naming times were seen by D- bilinguals who had the slowest naming times naming adjectives on low frequency nouns trials. The greater magnitude difference between high frequency and low frequency noun trials was produced by both bilingual groups. Looking at the average times for Morphology and Frequency together, we can observe that all groups named more slowly adjectives low and high frequency opaque noun trials than adjectives on low and high frequency transparent trials. Among all groups, less dominant bilinguals produced the slowest naming times with LO nouns. The greater magnitude difference between opaque and transparent noun trials was observed with monolingual speakers with low frequency nouns.

107

Table 5-1. Experiment 2 mean picture naming times and SE split by frequency and morphology Group Freq. Morph. Means SE Mono H O 1084.27 9.48 Mono H T 1021.5 9.77 Mono L O 1410.09 19.7 Mono L T 1252.45 18.4

D+ H O 1033.34 10.3 D+ H T 1022.71 9.38 D+ L O 1759.06 13.8 D+ L T 1757.93 16.4

D- H O 1214.52 15.9 D- H T 1165.33 12.6 D- L O 2137.3 39.7 D- L T 2103.49 39.6

ms.

Figure 5-1. Experiment 2 mean RTs and SE split by frequency and morphology

To assess the statistical significance of the results, we conducted a repeated-measures

ANOVA using R (R core team, 2015, version 3.1.3). We had a 2 x 2 x 3 factorial design with the within-subjects factors of Frequency (High versus Low) and Morphology (Opaque versus

108

Transparent) and the between-subjects factor of Group (Monolinguals, D+ bilinguals, D- bilinguals). The results of this model are presented in Table 5-2.

Table 5-2. Experiment 2 three-groups 2 x 2 x 3 repeated-measures ANOVA output Effect DFn DFd F p ges group 2 73 43.071 <0.001 *** 0.327 Frequency 1 73 545.040 <0.001 *** 0.668 Morphology 1 73 32.000 <0.001 *** 0.063 group:Frequency 2 73 57.908 <0.001*** 0.299 group:Morphology 2 73 8.909 <0.001*** 0.036 Freq.:Morph. 1 73 5.906 <0.001*** 0.012 group:Freq.:Morph. 2 73 2.610 0.08 0.011

As the output shows and based on the numerical patterns of Table and Figure 5-1 the

Group factor was significant (F(2,73) = 43.071, p = 0.001). Frequency was also significant and also the interaction between Group and Frequency (F(2,73) = 57.908, p = 0.001) indicating that the impact of Frequency was different between the groups. Morphology was also significant

(F(1,73) = 32, p = 0.001) and also the interaction between Frequency and Morphology (F(1,3) =

5.906, p = 0.001) indicating a different effect of Morphology depending on the noun frequency.

Furthermore, the significant interaction between Group and Morphology (F(2,73) = 8.909, p =

0.001) also indicates a difference in Morphology effect for each group. Because of these results, we will now conduct separate ANOVA models for each subgroup as well as one for the bilingual group together.

5.1.2 Monolingual Group

To briefly summarize again the numerical patterns, looking at the average times, mean picture naming times showed that, on average, adjectives on low frequency noun trials were named slower than adjectives on high frequency noun trials. Looking at the average times of

109

morphology and frequency together, as in experiment 1, high frequency and low frequency opaque nouns were named slower than their transparent counterpart. The greater magnitude difference between opaque and transparent noun trials was seen with low frequency nouns.

To assess the statistical significance of the results, we conducted a repeated-measures

ANOVA using R (R core team, 2015, version 3.1.3). We had a 2 x 2 factorial design with the within-subjects factors of Frequency (High versus Low) and Morphology (Opaque versus

Transparent). The results of this model are presented in Table 5-3.

Table 5-3. Experiment 2 monolinguals 2 x 2 repeated-measures ANOVA output Effect DFn DFd F p ges Frequency 1 33 195.238 <0.001*** 0.532 Morphology 1 33 33.331 <0.001*** 0.135 Freq:Morph. 1 33 10.418 <0.001*** 0.039

As the output shows and based on the numerical patterns, Frequency was significant

(F(1,33) = 195.238, p = 0.001), Morphology (F(1,33) = 33.33, p = 0.001) as well as the interaction of Morphology and Frequency (F(1,33) = 10.418, p = 0.001) were significant indicating that the effect of morphology was different depending on the lexical frequency of the noun. Because this interaction we run a pairwise tests in order to test against all pairs.

Table 5-4. Experiment 2 monolinguals pairwise test Estimate Std. Error t value Pr(>|t|)

l.o - h.o == 0 354.73 34.34 10.330 <0.001 ***

h.t - h.o == 0 -53.86 34.34 -1.569 0.4

l.t - h.o == 0 187.24 34.34 5.453 <0.001 ***

110

Table 5-4. Con’ Experiment 2 monolinguals pairwise test Estimate Std. Error t value Pr(>|t|)

h.t - l.o == 0 -408.59 34.34 -11.899 <0.001 ***

l.t - l.o == 0 -167.49 34.34 -4.878 <0.001 ***

l.t - h.t == 0 241.10 34.34 7.021 <0.001 ***

As the output shows, we find that there is a significant difference for all pairs except for the contrast between transparent and opaque high frequency items. The significant interaction between low frequency opaque and transparent nouns indicates that monolingual participants were faster with low frequency transparent nouns revealing a positive effect of regular morphology with low frequency nouns. We did not find a significant effect of morphology for the high frequency nouns.

5.1.3 Results: D+ and D- Bilingual Group

Numerical patterns showed that, on average, D+ bilinguals were faster than D- bilinguals naming adjectives on high and low frequency noun trials. The greater magnitude difference between high and low frequency nouns was seen by D- bilinguals. For the Frequency and

Morphology effect, adjectives on high and low frequency transparent noun trials were named faster than adjectives on high and low frequency opaque noun trials by both groups. For both groups, the magnitude difference was not very different indicating that frequency does not appear to differentially affect the Morphology effect.

To assess the statistical significance of the results, we conducted a repeated-measures

ANOVA using R (R core team, 2015, version 3.1.3). We had a 2 x 2 x 2 factorial design with the within-subjects factors of Frequency (High versus Low) and Morphology (Opaque versus

111

Transparent) and the between-subjects factor of Group (D+ bilinguals, D- bilinguals). The results of this model are presented in Table 5-5.

Table 5-5. Experiment 2 bilingual group comparison 2 x 2 x 2 repeated measures ANOVA output Effect DFn DFd F p ges Group 1 40 13.800 <0.001*** 0.116 Frequency 1 40 375.352 <0.001*** 0.731 Morphology 1 40 11.335 <0.001*** 0.042 group:Freq. 1 40 9.485 <0.001*** 0.064 group:Morph. 1 40 12.383 <0.001*** 0.046 Freq:Morph. 1 40 1.420 0.24 0.005 group:Freq.:Morph. 1 40 3.291 0.077 0.013

In this model and based on the direction of the results from Table and Figure 5-1 we find that there is a main effect for Group (F(1,40) = 13.800, p = 0.001) indicating an advantage in naming times for the more Spanish dominant group. Frequency (F(1,40) = 375.352, p = 0.001) was also significant along with Group x Frequency interaction (F(1,40) = 9.485, p = 0.001) indicating that the more Spanish dominant group was faster than the less Spanish dominant group and that the effect of frequency was different per group. Morphology was also significant along with Group x Morphology interaction (F(1,40) = 12.383, p = 0.001) indicating a different morphology effect for each group. As a result, we will conduct separate ANOVAs for each group and follow-up with any interactions.

Table 5-6. Experiment 2 D+ bilingual group 2 x 2 repeated-measures ANOVA output Effect DFn DFd F p ges

Frequency 1 23 385.459 <0.001*** 0.825

Morphology 1 23 0.1784 0. 676 0.0009

Frequency:Morph. 1 23 0.256 0.617 0.001

112

As we can observe, Frequency (F(1,23)=385.459, p =0.001) was significant, however,

Morphology was not significant for the more Spanish-dominant group in addition to the frequency effect.

Table 5-7. Experiment 2 D- bilingual group 2 x 2 repeated-measures ANOVA output Effect DFn DFd F p ges Frequency 1 17 125.670 <0.001*** 0.684

Morphology 1 17 12.379 <0.001*** 0.108

Freq.:Morph. 1 17 2.413 0.138 0.025

For the D- group, there is a main effect for Frequency and a main effect for Morphology.

Based on the numerical patterns, the morphology effect indicates a positive effect of morphological regularity for the less Spanish dominant group.

5.2 Accuracy Analysis

To briefly summarize, a response was accepted as correct if participants could describe the picture that appeared on the screen using only the adjective claro/a (clear-ADJ.SING.MASC/FEM) and in the vocabulary test they indicated the target referent with the correct article. Accurate responses that were outside the RT frame (3000 ms) or were not recorded by the program were also coded for the accuracy analysis. Correct responses after self-correction were accepted for the accuracy analysis. Answers that were not clear due to lack of audibility or unclear pronunciation were discarded. The dependent variable for the analyses was proportion of correct responses. Below, the accuracy results are provided.

5.2.1 Results: Monolingual, D+ and D- Group

Overall mean accuracy showed that both bilingual groups had a greater magnitude difference between high frequency and low frequency nouns than the monolingual group and,

113

among the bilingual groups, the D- bilingual group showed the greatest magnitude difference.

Numerically, the less Spanish dominant group showed the lowest accuracy rates, particularly, with low frequency nouns. Looking at Morphology and Frequency together all groups seemed to be more accurate with high frequency transparent noun trials. Monolingual speakers were more accurate with high and low frequency transparent noun trials. However, both bilingual groups were more accurate with high frequency transparent nouns and less accurate with low frequency transparent nouns.

Figure 5-2. Experiment 2 mean accuracy and SE split by frequency and morphology

To assess the statistical significance of the results, we conducted a repeated-measures

ANOVA using R (R core team, 2015, version 3.1.3). We had a 2 x 2 x 3 factorial design with the within-subjects factors of Frequency (High versus Low) and Morphology (Opaque versus

Transparent) and the between-subjects factor of Group (Monolinguals, D+ bilinguals, D- bilinguals). The results of this model are presented in Table 5-8.

114

Table 5-8. Experiment 2 three-groups 2 x 2 x 3 repeated measures ANOVA output Effect DFn DFd F p ges Group 2 75 129.414 <0.001*** 0.506 Frequency 1 75 130.006 <0.001*** 0.359 Morphology 1 75 6.059 <0.001*** 0.012 group:Freq. 2 75 36.609 <0.001*** 0.240 group:Morph. 2 75 0.150 0.860 0.0006 Freq:Morph. 1 75 14.813 <0.001*** 0.042 group:Freq.:Morph. 2 75 6.852 <0.001*** 0.039

As the output shows and based on the direction of the results from Table and Figure 5-2, we have a main effect of Group (F(2,75) = 129.414, p = 0.001) and Frequency (F (1,75)

=130.006, p = 0.001) as well as the interaction between Group and Frequency (F(2,75)= 36.609, p = 0.001) indicates a differential effect of Frequency depending on the group. Morphology

(F(1,75) = 6.059,p = 0.001) was significant as well as the interaction between Frequency and

Morphology (F(1,75) = 14.813, p = 0.001) indicating a differential effect of Morphology based on the lexical frequency of the noun referent. Finally, the 3-way interaction between frequency, morphology, and group was significant (F(2,75) = 6.85, p = 0.001).We conducted separate analyses per subgroup as well as a bilingual comparative analysis to further explore these interactions.

5.2.2 Results: Monolingual Group

Briefly summarizing the numerical patterns observed in Figure 5-2, mean accuracy for noun frequency showed a tendency to be more accurate with adjectives on high frequency than low frequency noun trials. On average, monolinguals seemed to be more accurate with transparent nouns. We can observe a greater magnitude difference between transparent and opaque noun trials with low frequency nouns.

115

To assess the statistical significance of the results, we conducted a repeated-measures

ANOVA using R (R core team, 2015, version 3.1.3). We had a 2 x 2 factorial design with the within-subjects factors of Frequency (High versus Low) and Morphology (Opaque versus

Transparent). The results of this model are presented in Table 5-9.

Table 5-9. Experiment 2 monolinguals 2x 2 repeated measures ANOVA output Effect DFn DFd F p ges Frequency 1 33 4.754 0.036* 0.038 Morphology 1 33 3.144 0.085 0.018 Freq:Morph 1 33 0.057 0.811 0.0001

As the output shows and based on the direction of the numerical patterns, Frequency (F

(1,33) = 4.753, p = <0.001) was significant.

5.2.3 Results. D+ and D- bilingual group

Mean accuracy showed that numerically the more Spanish-dominant bilinguals were more accurate than the less Spanish-dominant bilinguals, particularly with low frequency nouns.

The difference between high and low frequency noun trials is larger for the less Spanish- dominant group. With adjectives on low frequency noun trials, both bilingual groups seemed to more accurate with low frequency opaque noun trials. With high frequency noun trial, both bilingual groups showed the opposite trend, they were more accurate with high frequency transparent noun trials.

To assess the statistical significance of the results, we conducted a repeated-measures

ANOVA using R (R core team, 2015, version 3.1.3). We had a 2 x 2 x 2 factorial design with the within-subjects factors of Frequency (High versus Low) and Morphology (Opaque versus

Transparent) and the between-subjects factor of Group (D+ bilinguals and D- bilinguals). The results of this model are presented in Table 5-10.

116

Table 5-10. Experiment 2 bilingual group comparison 2 x 2 x 2 repeated measures ANOVA output Effect DFn DFd F p ges group 1 42 137.860 <0.001*** 0.413 Frequency 1 42 158.436 <0.001*** 0.574 Morphology 1 42 2.930 0.094 0.008 group:Frequency 1 42 16.269 <0.001*** 0.121 group:Morph. 1 42 0.094 0.760 0.0002 Frequency:Morph. 1 42 19.447 <0.001*** 0.124 group:Freq.:Morph. 1 42 0.522 0.473 0.003

As the output shows, we find a main effect of Group (F(1,42) = 137.860, p = 0.001) indicating that the more Spanish dominant bilingual were less accurate than the less Spanish dominant bilingual group. Frequency (F(1,42)=158.436,p = 0.001) was also significant as well as the interaction between Group x Frequency (F(1,42) = 16.269, p = 0.001) indicating a different effect of frequency depending on the group. The interaction between Frequency and Morphology was also significant (F(1,42) = 19.447, p = 0.001).We will further subdivide the two bilingual groups to explore these interactions.

Table 5-11. Experiment 2 D+ bilingual group 2 x 2 repeated-measures ANOVA output

Effect DFn DFd F p ges Frequency 1 23 37.296 <0.001*** 0.359 Morphology 1 23 1.027 0.321 0.004 Freq:Morph. 1 23 6.348 0.019* 0.078

Table 5-12. Experiment 2 D- bilingual group 2 x 2 repeated-measures ANOVA output Effect DFn DFd F p ges Frequency 1 19 162.964 <0.001*** 0.764 Morphology 1 19 2.133 0.160 0.015 Freq:Morph. 1 19 16.443 <0.001*** 0.207

117

Both analyses result in a main effect for Frequency as well as an interaction by

Frequency x Morphology. These interactions will now be tested with pairwise comparisons for each subgroup.

Table 5-13. Experiment 2 pairwise comparison for D+ bilinguals Estimate Std. Error t value Pr(>|t|) l.o h.o == 0 -0.12879 0.041 3.097 0.0137 * h.t h.o == 0 0.10227 0.041 2.459 0.0731 . l.t h.o == 0 -0.19167 0.041 4.608 <0.001 *** h.t l.o == 0 0.23106 0.041 5.556 <0.001 *** l.t l.o == 0 -0.06288 0.041 1.512 0.4347 l.t h.t == 0 -0.29394 0.041 7.067 <0.001 ***

We have several significant contrasts. Importantly, there is a marginal effect of HO and

HT indicating a differential effect of morphology, particularly, D+ bilinguals were more accurate with high frequency transparent nouns revealing a positive effect of transparent morphology. We did not find any significant effect of morphology in low frequency nouns.

Table 5-14. Experiment 2 pairwise comparison for D- bilinguals Estimate Std. Error t value Pr(>|t|) l.o h.o == 0 -0.289 0.036 7.954 < 0.001 *** h.t h.o == 0 0.142 0.036 3.927 0.001 ** l.t h.o == 0 -0.375 0.036 10.330 < 0.001 *** h.t l.o == 0 0.431 0.036 11.880 < 0.001 *** l.t l.o == 0 -0.086 0.036 2.376 0.090 l.t h.t == 0 -0.518 0.036 14.256 < 0.001 ***

118

Almost all contrasts are significant, with the contrast between opaque and transparent low frequency noun trials being marginal (more accurate on low frequency opaque noun trials).

Similar to D+ bilinguals, this group was more accurate with high frequency items than low frequency items. Similar to D+ bilinguals, this group was more accurate with high frequency transparent nouns. For low frequency items, although a marginal effect, this group indicates a trend to be less accurate with transparent nouns.

5.3 Summary of Results Experiment 2

For the reaction times analysis of Experiment 2 we found similar patterns as experiment

1, within the bilingual group more Spanish dominant bilinguals showed faster naming times than the less Spanish dominant bilinguals. We observe larger Frequency effects, but this effect was more salient in both bilingual groups, even larger frequency effects for the D- bilingual group.

Unlike in Experiment 1, we found a Frequency and Morphology interaction effects in the monolingual group. Monolingual results revealed that they were faster with low frequency transparent noun trials. We found a main effect of morphology in the less Spanish dominant group but no morphology effect in the more Spanish dominant group. In the accuracy results of experiment 2 we also found that monolingual speakers were more accurate than both bilingual groups. D+ bilinguals were more accurate than the less Spanish dominant group. . Frequency effects were larger in the bilingual groups than in the monolingual group, particularly the less

Spanish dominant bilingual group showed the largest difference between low and high frequency. No effect of Morphology for the monolingual group probably due to ceiling effects.

We found for the D+ bilingual group a frequency and morphology interaction effect

(approaching significance). Numerically more accurate responses were given with adjectives on high frequency transparent compared to opaque noun trials; no effect of morphology was seen for the low frequency noun trials. The D- bilinguals showed a significant frequency and

119

morphology interaction effect revealing that they were more accurate with transparent than opaque nouns especially in the high frequency conditions.

120

CHAPTER 6 DISCUSSION AND CONCLUSION

The present dissertation examined whether noun morphology and lexical frequency have a role in the lexical retrieval of gender features in oral production. Particularly, these variables were manipulated in our study to provide further evidence on the different outcomes of adult bilingualism, studying two different HS populations that differ in their amount of exposure to

Spanish while growing up. It is worth noting that both bilingual populations were born in the

USA and both groups self-reported to be dominant in English. Nevertheless, they belonged to different HS generations, and aggregated by group, they also reported to differ in their exposure to Spanish on a daily basis.

We administered two oral production tasks to a group of monolingual speakers of

Spanish from Santiago de los Caballeros (Dominican Republic) and two groups of HSs of

Spanish from Florida (United States) where RTs were measured along with accuracy. The stimuli, particularly a set of color pictures in the oral production task (both in the picture naming task and the picture description task) consisted of high and low frequency nouns with opaque or transparent morphology.

Section 6.1 provides a brief summary of the results and the interpretation of the findings.

In Section 6.2 we provide the details of other results found. Next, Section 6.3 presents a conclusion of the findings and the last three sections present the implications, contributions of these findings and future directions.

6.1 Results, Hypothesis and Discussion

First, I provide a table with the main theoretical models, their predictions and whether we found evidence that supports these hypotheses in the present investigation. Second, the research

121

questions are revisited, and further details are provided of the supporting and contrasting results of the present results and the tested predictions.

Table 6-1. Summary of theories, predictions and findings Theories Predictions Mono Biling. Reliable Cue Main effect of morphology (no frequency effect) Hypothesis  

Lexico-Syntactic Main effect of frequency (no morphology effect) Access   Hypothesis Main effects of morphology and frequency or frequency and morphology interaction (facilitatory Two Routes effect of transparent nouns) Hypothesis  

Interaction between frequency and morphology; more accurate in HO than HT, more accurate with Dual Mechanism LT than LO. Model Faster naming times with HO than HT and faster

naming times with LT than LO  

Bilinguals slower than monolinguals. Larger Frequency-Lag frequency effects in the bilingual group, even Hypothesis larger for the less Spanish dominant group  

Due to the main effects of morphology and frequency in the Experiment 1 RTs analysis and the interaction between frequency and morphology in the Experiment 2 RTs analysis showing in both experiment a facilitatory effect of transparency we find evidence for the Two-

Route hypothesis (Gollan & Frost, 2001) in the monolingual group. Interestingly, we observed the same pattern of results in the bilingual group accuracy analysis, but not in the RTs analysis.

We also find evidence for the Frequency-Lag Hypothesis (Gollan, Slattery, Goldenberg, Van

Assche, Duyck, Rayner, 2011; Emmory, Petrich and Gollan, 2012) in agreement processing in

122

both of the bilingual populations studied. In contrast, we find no evidence for either the Reliable

Cue or the Lexico-Syntactic Access Hypotheses, and the interaction did not match the predictions of the Dual Mechanism Model (i.e., as applied to regular/irregular morphology). We now turn to addressing each research question in greater detail.

RQ (1): What is the role of noun morphology in gender agreement processing?

We predicted different outcomes based on different hypotheses about the impact of morphological cues in gender access. According to the Reliable cue hypothesis (see Section

2.6.1; Bates et al., 1996; Taft and Meunier, 1998) morphological information is the most reliable cue to access gender. Assuming this hypothesis, participants were expected to retrieve gender faster with transparent nouns than with opaque nouns, regardless of the frequency of the noun, showing that gender is more easily retrieved with the help of morphological cues. Therefore, we expected a main effect of morphology. However, our data did not provide such results.

The second possible outcome is that gender could be retrieved without the help of noun morphological endings via an abstract lexico-syntactic route, which we called the lexico- syntactic view. In this case, following previous studies that found no effect of morphological cues in nouns (see Section 2.6.1; Badecker et al, 1995; Miozzo and Caramazza, 1997; Vigliocco et al., 1997), we expected that gender would be retrieved faster in high frequency words regardless of a noun’s morphological ending, that is, a main effect of frequency. However, results in the present investigation did not find evidence of this prediction.

Another outcome was predicted by the Two-Route hypothesis (Gollan & Frost, 2001) and according to this view we should either observe both a main effect of frequency and morphology or an interaction between morphology and frequency, possibly resulting in a reduced frequency effect in transparent nouns. We found main effects for both frequency and morphology for

123

monolinguals in the Experiment 1 RTs analysis, indicating faster naming times for transparent nouns and for high frequency nouns. In Experiment 2, we found a morphology by frequency interaction for monolinguals in the RTs analysis and further analyses indicated that they were faster with adjectives in low frequency transparent noun trials than with low frequency opaque trials indicating a facilitatory effect of noun transparency. For bilinguals, we observed a similar pattern to the one in the monolingual RTs analysis. Both groups had a main effect of morphology in the Experiment 1 accuracy analysis, being more accurate with transparent nouns. In

Experiment 2, we observed an interaction with frequency and morphology and further analyses indicated that both bilingual groups were more accurate with high frequency transparent nouns

(D+ bilingual reaching significance). Based on these results, we can clearly state that participants make use of both the lexico-syntactic and also morphophonological information in line with studies that found evidence of these two routes to access gender (Cafarra et al., 2004) and corroborating the two-route hypothesis proposed by Gollan and Frost (2001).

RQ (2) Are the morphology effects similar to other dual-route models?

We suggested an alternative view of the role of noun morphology based on the studies carried out with regular and irregular forms and the observation that these forms are processed by different processing mechanisms (Pinker 1990, 1999). According to this view, irregular forms, in our study opaque nouns, will be stored as whole forms in memory, while for regular nouns, in our study transparent nouns, rule application will take place. Therefore, the predictions of this view are that with high frequency nouns, irregular or opaque nouns will demonstrate an advantage as seen in faster naming times or higher accuracy rates since they are not computed but quickly retrieved from memory. On the other hand, low frequency transparent nouns should show an advantage since morphological regularity should result in faster application than

124

retrieving low frequency opaque nouns from the lexicon. However, in line with previous studies testing this hypothesis on gender morphology in Portuguese (Resence and Mota, 2017), we did not find evidence of a frequency and morphology interaction in Experiment 1 RTs analysis. We found a morphology by frequency interaction in Experiment 2 indicating, in line with the hypothesis, that participants were faster with low frequency transparent nouns than low frequency opaque nouns. However, we did not see effects of morphology with the high frequency trials. We also did not find any evidence of similarity to other dual mechanisms to process opaque and transparent nouns in the bilingual speakers’ accuracy results.

RQ (3): Do different linguistic populations differ in the use of noun morphology when accessing gender?

Our predictions were based on previously reviewed studies (see Section 2.6.1) that investigated monolinguals and the effect of noun morphology. For the HSs groups, based on previous off-line research that found morphology effects in accuracy, we hypothesized that HSs would reveal morphology effects (Alarcón, 2011; Montrul, 2013; Montrul, 2008) in both accuracy and RTs analysis.

For the bilingual groups, we found that morphology effects were absent in experiment 1 in the RTs analysis. For the RTs analysis in experiment 2, interestingly, we found a main effect of morphology in the less dominant group, but not in the more dominant group. For accuracy, in experiment 1, we found a main effect of morphology indicating that both HS groups were more accurate with transparent nouns. In Experiment 2, morphology x frequency was significant indicating that the effect of transparency was greater in high frequency noun trials. Interestingly, the less dominant HS group was more accurate with low frequency opaque nouns than low

125

frequency transparent nouns, in contrast with the predictions of any of the hypotheses we presented.

We can deduce from our findings that morphological cues are utilized by monolingual speakers and HSs in a different way. We did not find any effect of morphology in any of the bilingual groups in Experiment 1 RTs analysis and in Experiment 2 (RTs analysis) we only found effects in the D- bilingual group. However, we observed the facilitatory effect of noun morphology in the accuracy analysis of both experiments for the bilingual speakers (as a main effect in Experiment 1 and as an interaction in Experiment 2). We cannot conclude that lack of results in the RTs analysis represent a bilingual deficit since we observed morphological effects in accuracy across both experiments, although the exact pattern of effect was different.

With regard to the results found in our HSs groups, we can state that, in line with previous results that found transparency effects in bilingual speakers (Alarcón, 2011; Montrul,

2013; Montrul, 2008), we found that our bilingual groups also utilized morphophonological information in the processing of gender agreement as seen in the accuracy results. Furthermore, we observe a difference between the bilingual and monolingual group as seen by the lack of morphology effects by the bilingual groups in the RT data. However, there is no framework or hypothesis that accounts for these processing differences between monolingual and bilingual speakers with noun morphology. Many researchers take accuracy to be reflective of representation while RTs of processing. Therefore, based on this assumption we can state a difference in processing between HSs and monolingual speakers an assumption that has already been considered by the proponents of other frameworks that compare HSs and monolinguals speakers, such as the Frequency-Lag Hypothesis (Gollan, Slattery, Goldenberg, Van Assche,

Duyck, Rayner, 2011; Emmory, Petrich and Gollan, 2012) that accounts for the different

126

underlying processes of HSs with noun lexical frequency. However, to explain the differences in gender morphological processing between these two group or even within the HSs group, it is necessary to design an experiment exclusively targeting gender morphology so as to obtain RT data (See Section 6.4).

RQ (4): Does noun frequency impact the retrieval of gender features across different linguistic populations?

Lexical frequency was manipulated primarily to test whether morphological cues are used more prevalently than the frequency information stored in the lexicon when retrieving gender. However, due to the limited research on how lexical frequency affects the retrieval of gender, the present investigation also investigated the role of lexical frequency in monolingual and heritage speakers when accessing grammatical gender.

The effect of noun frequency in gender access with monolingual speakers of Spanish has been tested only by Navarrete, Basagni, Alario & Costa (2007) in a gender decision task and picture description task (similar to the ones used here) using a sequence of determiner + copula

+ adjective (new/old). Their results corroborated frequency effects when accessing gender (See section 2.8.2. for a review of the study). Therefore, based on these results, our research question was whether noun frequency plays a role in accessing gender in determiner-noun and noun- adjective agreement. We hypothesized that the frequency effect will be replicated and extended from the Navarrete et al. (2007) findings to new agreement contexts. We hypothesized overall faster response times for higher frequency than lower frequency noun trials, regardless of assignment or agreement operation.

Findings in the present study confirm an effect of lexical frequency in both Experiments

1 and 2. The effect of frequency was overall present in the three group RTs analysis of

127

Experiment 1. Although, numerically, RTs in Experiment 2 were larger than in Experiment 1 we also observed the same frequency effects in the results of Experiment 2. Monolingual speakers of

Spanish showed faster naming times with high frequency nouns similar to the results found in

Navarrete et al. (2007). Monolingual speakers scored at ceiling or near ceiling in both experiments, so this explains the lack of frequency effects in accuracy.

The overall results provide more evidence on effects of lexical frequency in gender access. These results also expand Navarrete’s et al. (2007) results to two different agreement operations, determiner-noun agreement and noun-adjective agreement.

The impact of frequency in different linguistic populations, even in bilinguals with different language dominance, has recently been the focus of many studies and hypotheses.

According to the Frequency-Lag hypothesis and its antecedent, the Weaker Links hypothesis

(Gollan & Silverberg, 2001; Gollan et al., 2002; Gollan et al., 2005; Gollan, Montoya, Cera, &

Sandoval, 2008; Sandoval, Gollan, Ferreira, & Salmon, 2010), noun frequency effects show a greater magnitude difference in bilinguals than monolinguals. In turn, less dominant bilinguals show the highest magnitude frequency effects between high frequency and low frequency nouns.

Additionally, monolinguals have the highest activation with high frequency words. Based on this hypothesis, we wanted to investigate whether noun frequency plays a role in the retrieval of gender features in bilingual speakers with different linguistic dominance.

We hypothesized that, similar to previous research with monolingual speakers

(Navarrete, Basagni, Alario & Costa, 2007) and previous research on lexical access with monolinguals and bilinguals (Gollan & Silverberg, 2001; Gollan et al., 2002; Gollan et al., 2005;

Gollan, Montoya, Cera, & Sandoval, 2008; Sandoval, Gollan, Ferreira, & Salmon, 2010), we will find a frequency effect in the retrieval of gender features. Monolinguals will show the fastest

128

naming times with high frequency and low frequency nouns and the smallest magnitude difference between high frequency and low frequency nouns. Among the bilingual groups, the bilingual group with highest dominance will show the fastest naming times with high and low frequency nouns and consequently the smaller difference between high and low frequency nouns. In other words, the Spanish-dominant bilingual group will show greater activation of HF words than the less Spanish-dominant bilingual group, also leading to faster agreement responses than the less dominant group. The less Spanish-dominant bilingual group will show the least activation with HF nouns, thus the slowest response times. This impact will be more visible with

LF words where the group will show the smallest activation levels showing slower response times than the other groups.

As hypothesized, we found in Experiments 1 and 2 that both bilingual groups were slower than monolingual speakers, especially the D- bilingual group. These results confirm the so-called bilingual lexical frequency “disadvantage” in lexical access. Furthermore, the interaction between both of the bilingual groups and noun frequency demonstrated that both bilinguals showed larger frequency effects, particularly the less dominant bilingual group, as predicted. Also, both bilingual groups seemed to be particularly less accurate than monolingual speakers with low frequency nouns. The accuracy analysis also showed that the less Spanish- dominant group was also less accurate than the more Spanish dominant bilingual group.

In line with our predictions and according to the Frequency-Lag Hypothesis (Gollan,

Slattery and Rayner, 2012; Emmory, Petrich and Gollan, 2012), we found greater frequency effects in both bilingual groups than the monolingual group. In turn, we observed the greatest

Frequency effects with the less Spanish-dominant group. There are no previous studies that provides evidence of the linguistic outcomes of two HSs both USA-born and dominant in

129

English that differ in the amount of exposure of Spanish and/or frequency, and when comparing both bilingual groups’ results, we could clearly observe the difference in lexical access in the two agreement operations investigated between these two bilingual groups.

6.2 Other Results

6.2.1 Gender assignment and gender agreement

Research, particularly in second language acquisition, has provided inconclusive evidence of the differences between determiner-noun and noun-adjective agreement. On the one hand, definite articles appear consistently prenominally with their corresponding noun. On the other hand, the number of adjectives that can go with a noun are very numerous and less obligatory. Therefore, it is more likely that determiner-noun constructions are stored in memory since determiners are more salient than adjectives that also occur post-nominally. Because of prior inconclusive findings and the different structural and linguistic characteristics of both agreement operations we aimed to provide some evidence of the difference of these two agreement processes.

The results for the monolingual speakers provide some different results possibly evidencing the different status of these operations. Numerically, average naming times were larger for Experiment 2 than Experiment 1. Statistically, we also found differences between the two experiments. For the monolingual group in Experiment 1, we found a main effect of morphology and, in Experiment 2, with noun-adjective agreement, we found an interaction between frequency and morphology not found in experiment 1. For the bilingual group accuracy results, we found the same patterns of results found in both experiment RTs analysis.

We additionally observed that the number of mismatches in Experiment 2 was larger than in Experiment 1 (see section 6.2.2), also in line with offline studies that found more errors in noun-adjective than determiner-noun agreement (Martinez Gibson, 2011).

130

We cannot draw any specific conclusions on the differences of these two operations.

These results might be evidence of different mechanisms for these two agreement processes.

However, it might be simply derived from the syntactic differences of these two operations, a prenominal and a postnominal operation, the fact that between the two operations, the noun adjective agreement modifies its word-form according to the gender information of the referent or just the frequency in which these two operations appear in oral production. Further research needs to be done comparing pre-nominal and post-nominal adjectival agreement to provide further evidence.

6.2.2 Masculine default in HSs of Spanish

Previous studies, predominantly in the field of second language acquisition, that investigated grammatical gender in Spanish found an overuse of masculine in determiners and modifiers naming this phenomenon as the “masculine default” (Studies with HSs: Montrul et al.

2013; Studies with L2ers: Fernández-García 1999; White et al. 2004; Shlig 2003; Hawkins 1998;

Bruhn de Garavito & White 2000). In the present study, participants also overuse the masculine as the default option in the determiner (Experiment 1) and the adjective (Experiment 2). In

Experiment 1 the more Spanish dominant bilingual group produced 38 mismatches and 32 were a feminine noun with a masculine determiner, el-MASC serpiente-FEM (the snake) corroborating the overuse of masculine. The less dominant bilingual group produced double the mismatches of the

D+ group; specifically, out of 98 mismatches, 87 were masculine determiners with a feminine noun. In Experiment 2 participants produced more agreement mismatches than in Experiment 1.

Since in Experiment 2 only the adjective was produced, without the referent noun, participants, particularly, heritage speakers, might have ventured to guess more often the answer than in

Experiment 1. In experiment 2, D+ bilinguals produced 104 mismatches and out of these mismatches 85 tokens were the result of the use of masculine default, e.g. (nube-FEM) claro-MASC

131

“clear cloud”. The D- bilinguals produced 169 agreement mismatches and out of these mismatches 141 were the result of the use of masculine default. It is worth noting that in both experiment most of the mismatches were produced with low frequency nouns.

As we can see, bilingual participants overproduced the use of masculine with determiner and to a greater extent with modifiers.

6.3 Conclusion

In the present dissertation monolinguals and Spanish-English bilinguals that differ in

Spanish dominance named pictures with high- and low-frequency nouns and with different morphological endings (opaque and transparent) for the purpose of determining how these two variables (frequency and morphology) affect gender access across different linguistic populations. Based on different accounts on the role of word form information in gender access we predicted the following different outcomes: (1) a morphology over frequency effect (Reliable cue hypothesis), (2) a frequency over morphology effect (Lexico syntactic access view) and (3) a third outcome with a main effect of morphology and frequency or a possible interaction between frequency and morphology with a facilitatory effect of transparency (two routes hypothesis).

Alternatively, we also suggested a fourth outcome (4) based on a framework that poses a different mechanism to process opaque and transparent nouns (Dual Mechanism Model). Results in Experiment 1fit into the Two-Route Hypothesis predictions (Gollan and Frost (2001) in both of our agreement operations and for both monolingual and bilingual speakers. Also, in line with other research that found transparency effects in HSs (Montrul, 2008, 2013; Alarcón, 2011;

Foote, 2015) we found the same results of word regularities in both bilingual groups accuracy analysis.

One of the questions that arises from the different results in RTs and accuracy analysis and Experiment 1 and 2 between monolingual and HSs of Spanish is why gender information

132

and agreement becomes non-monolingual like in HSs. One of the possible explanations is that the input that they receive from their parents might have attrited and lost the canonical gender features of Spanish and therefore we observed the difference in the number of agreement mismatches between bilingual and monolingual speakers. This is more likely for 3rd generation

Spanish speakers in the present investigation, i.e., those whose parents were born in the USA.

However, this might also be true for 2nd generation HSs whose parents were born in a Spanish speaking country and then immigrated to the US. In some cases, if the environment does not support the use of Spanish, speakers might shift to the dominant societal language and some of the features of the first language might be subject to change. However, we can also argue that these HSs might have learned and acquired the canonical features of gender in Spanish, however, like other bilingual linguistic phenomena, such as the loss of subjunctive mood in HSs of

Spanish (Lipski,1993; Silva-Corvalán, 1994), we might be in the presence of the loss or simplification of grammatical gender in Spanish. By comparing the two HS groups in our experiment we were able to see a possible generational change/s of a particular linguistic phenomenon. In our results, we also observed that the use of masculine default was more prominent in the less Spanish dominant group providing evidence of a simplification process in which the default gender is used. Although the morphological data did not provide processing evidence of a difference between both bilingual groups, the frequency results provide solid patterns and results of a possible generational shift. It is worth noting that not all D- bilingual were 3rd generation but most of this group consisted in 3rd generation HSs. Therefore, we cannot state with certainty that this is generational phenomenon, but we can clearly confirm that this is due to difference between these two groups in frequency of use and linguistic dominance in

Spanish.

133

For the frequency effect, our results are compatible with the Frequency-Lag Hypothesis

(Gollan, Slattery and Rayner, 2012; Emmory, Petrich and Gollan, 2012). Our results revealed that bilingual performance differs from the monolingual norm but also from bilingual to bilingual providing evidence on the variability of linguistic skills within bilingual speakers. In both experiments, we found frequency effects and this frequency effect was larger in the less

Spanish dominant group. We can conclude that language dominance due to limited exposure to

Spanish has a differential has an impact in the production of agreement in Spanish.

As mentioned above, we also encountered differences between the two agreement operations in both bilingual and monolingual speakers. The main differences were a main effect of morphology in Experiment 1 and a morphology and frequency interaction in Experiment 2.

Numerically, all participants showed slower naming times in Experiment 2 than in Experiment 1.

Furthermore, all groups have more agreement mismatches with noun-adjective agreement than with determiner-noun agreement. Based on these results we can observe that for all participants noun-adjective agreement was cognitively costlier as seen in slower naming times and a larger number of mismatches. However, as mentioned, we cannot determine, without further research, what is the source of these differences encountered.

6.4 Implications and Future Research

The present investigation reveals processing differences between the two HSs groups we investigated. Both bilingual groups have similar backgrounds, being born in the United States and self-reporting to be dominant in English. Moreover, they also scored similarly in the objective measures, such as the Spanish proficiency test (DELE) and the English proficiency test

(MELICET). Nevertheless, our RTs and accuracy results reveal differences between these two groups. The implications of these results are important for future research with HSs since in most research, even some reported in the present investigation, HSs are grouped together based on

134

self-reported dominance in their heritage language or using proficiency tests that might not provide very reliable results on their Spanish dominance since most heritage speakers do not have much formal instruction in their HL while growing up. Therefore, future research should take into account these underlying differences in an apparently similar HS group. Furthermore, these results are also relevant for educational purposes. Heritage speakers tend to be placed in the

HL classroom based on their results on a grammar test. These tests give advantage to those heritage speakers who had more formal education in Spanish, but they do not provide an accurate representation of the fluency or dominance of the bilingual speaker. As a result, educators combine bilingual students who can write and read well in Spanish with other students who are more orally fluent in Spanish. Nevertheless, our results demonstrate that despite similar scores from the Spanish proficiency test, our results reported a difference between these groups.

Apart from the differences observed between both bilingual groups we also observed a lack of morphology effect for the bilingual group in the RTs analysis of Experiment 1 and

Experiment 2 (only for the D+ bilingual group). As we mentioned, we can state that this is not due to a representational deficit of the gender morphological endings since we found morphology effect in the accuracy analysis in both experiments. However, it is necessary to carry out further investigation to obtain RT data of a morphology effect in bilinguals and find out a possible difference in processing between monolingual and bilingual speakers. As we observed in our results, HSs missed a great amount of experimental trials with low frequency nouns, as seen by the skipped and mismatched trials. Therefore, in order to obtain morphology results it will be necessary to design an experimental stimulus excluding low frequency nouns.

Furthermore, below I consider other experimental limitations that might have confounded the results and should be taken into account for future research.

135

Future research should also consider some methodological issues that were present in the present investigation. For instance, the subjectivity of individual lexical frequency and the fact that it can vary from speaker to speaker even in monolingual contexts. Furthermore, when a bilingual has limited input in the HL their ability to speak fluently may vary from one semantic domain to another, the frequency of use of a word (i.e., its lexical frequency) can be drastically different from the monolingual “norm”. Nevertheless, researchers rely upon monolingual corpora to measure lexical frequency. For this reason, it is necessary to mention that data loss could have been diminished in the present investigation with a more reliable tool to measure bilingual language frequency. Future research could consider this issue and measure bilingual frequency carrying out a norming study where these lexical items are presented, and bilingual speakers’

RTs are measured to obtain a correct frequency effect according to their own experience.

Another issue to consider, as reported in section 6.2.2., is the use of masculine default by bilingual speaker since gender may be processed differently for masculine and feminine items.

Therefore, any future experiments should additionally add grammatical gender (masculine, feminine) as an additional variable. Similarly, the results may have changed had we calculated the subjective gender assignment of the experimental nouns by the bilingual participants.

Lastly, results have provided more evidence of the differences on the agreement operations tested in the present investigation. Hence, future research with prenominal and postnominal adjectival agreement will help to elucidate whether the processing differences rely on the structural position of the syntactic elements we investigated (prenominal and postnominal) or the difference in frequency of occurrence between these two agreement operations.

6.5 Contributions of Findings

The findings of this study are valuable in that they provide information that contributes to our understanding of the cognitive processes in HS adult bilingualism. Particularly bilingual

136

speakers of Spanish that, despite having acquired Spanish from childhood and in a naturalistic way, purportedly differ from the monolingual norm in their linguistic performance. More importantly, since our results provide evidence of the differences in processing within these two bilingual populations, it is of main relevance of testing linguistic dominance in an apparent homogenous bilingual group for future research. The present dissertation provides solid results of the linguistic outcomes of two groups of US-born bilinguals. Despite both bilinguals self- reporting to be dominant in English and reporting greater use of English on a daily basis, they differed in their performance in both experiments. Despite these apparent group similarities in profile, their linguistic dominance affected their performance in different ways. It is worth noting that, as Grosjean (2007) highlights, the number of linguistic domains might have decreased in these bilingual speakers due to language shift from Spanish to the main societal language.

Therefore, the reduced number of linguistic domains resulted in less linguistic exposure, less input and a lesser fluency in Spanish. As seen in this study, less exposure to the HL implied less linguistic dominance in the heritage language and greater frequency effects due to a weaker connection with lexical items. This represents a contribution to the field of HL bilingualism since it describes the underlying processes in oral production within a bilingual group that differed in terms of language exposure and language dominance. It also contributes to research on HS bilingualism from the perspective of linguistic variation among HSs and the relationship between differences in access to input and language shift, language change and language loss from generation to generation. In the context of bilingualism in the United States, especially in

Florida, one of the states with the largest Hispanic communities in the USA, these findings are of main importance as they underscore the need for language policies that will foster HL maintenance.

137

APPENDIX A MONOLINGUAL PARTICIPANTS

Place of residence Gender Age Education M1 Dominican Republic F 22 Bachelor, in progress M2 Dominican Republic F 19 Bachelor, in progress M3 Dominican Republic F 18 Bachelor, in progress M4 Dominican Republic F 18 Bachelor, in progress M5 Dominican Republic F 20 Bachelor, in progress M6 Dominican Republic M 18 Bachelor, in progress M7 Dominican Republic F 22 Bachelor, in progress M8 Dominican Republic F 22 Bachelor, in progress M9 Dominican Republic F 22 Bachelor, in progress M10 Dominican Republic M 27 Master, in progress, M11 Dominican Republic M 24 Bachelor, in progress M12 Dominican Republic F 19 Bachelor, in progress M13 Dominican Republic F 24 Bachelor, in progress M14 Dominican Republic F 19 Bachelor, in progress M15 Dominican Republic F 25 Bachelor, in progress M16 Dominican Republic F 19 Bachelor, in progress M17 Dominican Republic F 26 Bachelor, in progress M18 Dominican Republic F 24 Bachelor, in progress M19 Dominican Republic M 18 Bachelor, in progress M20 Dominican Republic F 20 Bachelor, in progress M21 Dominican Republic F 20 Bachelor, in progress M22 Dominican Republic M 24 Bachelor, in progress M23 Dominican Republic F 19 Bachelor, in progress M24 Dominican Republic M 19 Bachelor, in progress M25 Dominican Republic F 19 Bachelor, in progress M26 Dominican Republic F 22 Bachelor, in progress M27 Dominican Republic F 21 Bachelor, in progress M28 Dominican Republic F 18 Bachelor, in progress M29 Dominican Republic M 22 Bachelor, in progress M30 Dominican Republic F 18 Bachelor, in progress M31 Dominican Republic M 22 Bachelor, in progress M32 Dominican Republic M 19 Bachelor, in progress M33 Dominican Republic M 19 Bachelor, in progress M34 Dominican Republic F 20 Bachelor, in progress Total 34 24F/10M 20.8 33 Bachelor/1MA

138

APPENDIX B BILINGUAL PARTICIPANTS

Place of residence Gen. Gender Age Education B35 Hollywood, FL 2nd M 19 Bachelor, in progress B36 Miami, FL 2nd F 20 Bachelor, in progress B37 Ocala, FL 3rd F 20 Bachelor, in progress B38 Fort Lauderdale 2nd M 21 Bachelor, in progress B39 Weston, FL 2nd F 20 Bachelor, in progress B40 Miami, FL 2nd M 20 Bachelor, in progress B41 ft. Lauderdale 2nd F 19 Bachelor, in progress B42 Miami, FL 2nd M 21 Bachelor, in progress B43 Lakeland, FL 2nd M 20 Bachelor, in progress B44 Miami, FL 2nd F 21 Bachelor, in progress B45 Queens, NY 2nd M 19 Bachelor, in progress B46 Palm beach, FL 2nd M 20 Bachelor, in progress B47 NY city & Weston, FL 2nd F 20 Bachelor, in progress B48 Miami, FL 2nd F 20 Bachelor, in progress B49 Talahasse, FL 3rd M 21 Bachelor, in progress B50 Rochester, NY/Orlando 2nd F 19 Bachelor, in progress B51 Orlando, FL 2nd M 20 Bachelor, in progress B52 Miami, FL 2nd F 19 Bachelor, in progress B53 NY city 3rd F 21 Bachelor, in progress B54 Hialeah, FL 2nd F 20 Bachelor, in progress B55 St cloud, FL 2nd M 20 Bachelor, in progress B56 Miami, FL 3rd F 20 Bachelor, in progress B57 Clearwater, FL 3rd F 20 Bachelor, in progress B58 Lake city, FL 3rd M 19 Bachelor, in progress B59 Pembroke pines, FL 3rd F 19 Bachelor, in progress B60 Miami, FL 2nd F 20 Bachelor, in progress B61 Bronx, NY 2nd F 19 Bachelor, in progress B62 Orlando, FL 2nd M 20 Bachelor, in progress B63 Lake city, FL 3rd F 18 Bachelor, in progress B64 Tampa, FL 3rd M 19 Bachelor, in progress B65 Miami, FL 2nd F 20 Bachelor, in progress B66 Jupiter, FL 2nd F 21 Bachelor, in progress B67 Cape Canaveral, FL 3rd F 19 Bachelor, in progress B68 Fort Myers, FL 2nd F 20 Bachelor, in progress B69 Crystal clear, FL 3rd F 20 Bachelor, in progress B70 Tallahassee, FL 2nd M 20 Bachelor, in progress B71 Orlando, FL 2nd F 21 Bachelor, in progress B72 Hialeah, FL 2nd F 20 Bachelor, in progress B73 Weston, FL 2nd M 19 Bachelor, in progress B74 Orlando, FL 3rd F 21 Bachelor, in progress B75 Miami, FL 2nd F 18 Bachelor, in progress B76 Fort Lauderdale, FL 2nd M 19 Bachelor, in progress B77 Tampa, FL 2st M 21 Bachelor, in progress B78 Jacksonville, FL 3rd F 20 Bachelor, in progress Total 27, 2nd 27F/17M 19.84 44Bachelor (in progr.) 44 17, 3rd

139

APPENDIX C EXPERIMENTAL SEQUENCE SAMPLE

Two sequences introducing the experimental stimuli used in Experiment 2. In this experiment, participants were instructed to use claro/a (clear-masculine or feminine according to the stimulus presented), first sequence, or borrosa (blurry-masculine or feminine according to the stimulus presented), second sequence, to describe what appeared on the screen.

140

APPENDIX D CONSENT FORM IN SPANISH

Informe de Consentimiento

Procesamiento de características léxico-sintácticas de género en nativos y hablantes de herencia de español

Propósito de la investigación: El propósito del estudio es examinar los patrones de habla de los hablantes nativos de español y los hablantes de herencia del español. Qué se le pedirá al participante: Se le pedirá que rellene un cuestionario de competencia (DELE) en español, un cuestionario sobre su/s lenguas, un test de vocabulario y tres tareas lingüísticas que consistirán en nombrar unas imágenes o describir el color u otras cualidades del objeto que aparece en la pantalla (viejo-nuevo). Tiempo necesario: 1 hora y media Riesgos y beneficios: No hay compensación para su participación en el estudio. No hay riesgos asociados con este proyecto de investigación. Puede que se beneficie al participar en este estudio por tener la oportunidad de contribuir a un mayor entendimiento del lenguaje. Aprenderá más sobre su lenguaje y la manera en que lo usa. Su participación en esta investigación ayudará a proveer una comprensión nueva de los patrones del lenguaje bilingüe. Confidencialidad: Su identidad se mantendrá confidencial en la mediada prevista por la ley. A su información se le asignará un número de código. No se usará su nombre en ningún informe. Participación voluntaria: Su participación en este estudio es voluntaria. No hay una penalización por no participar. Derecho de retirarse del estudio: Usted tiene el derecho de retirarse del estudio en cualquier momento sin consecuencia alguna. Con quién ponerse en contacto si tiene preguntas sobre el estudio: Dámaris Mayans, Graduate Student, Department of Spanish and Portuguese Studies, 170 Dauer Hall, P.O. Box 117405, Gainesville, FL 32611-7405, tlf: 352-519-2184. Dr. Ana de Prada Pérez, Associate Professor, Department of Spanish and Portuguese Studies, P.O. Box 11545, Gainesville, FL 32611-7405, tlf: (352) 294-7451. Con quién ponerse en contacto sobre sus derechos como participante en el estudio: IRB02 Office, Box 112250, Universidad de Florida, Gainesville, FL 32611-2250; tlf: 392-0433. Acuerdo: He leído el procedimiento descrito arriba. Acepto de forma voluntaria participar en la investigación lingüística. He recibido una copia de esta descripción.

Participante:

Nombre y firma: ______Fecha: ______

Investigador Principal:

Nombre y firma:______Fecha: ______

141

APPENDIX E MONOLINGUAL SPEAKERS’ LBQ

Formulario de historia lingüística

Instrucciones: Complete el formulario en español.

I. Historia personal

Nombre

Fecha de Nacimiento

Sexo (elija uno) Hombre Mujer

Lugar de nacimiento (Ciudad, País)

Lugar de residencia actual (Ciudad, Estado)

Lugar de residencia de los últimos 10 años (Ciudad, Estado). Sí ha vivido en más de un sitio, indíquelo.

Lugar de nacimiento del padre (Ciudad, País)

Lugar de nacimiento de la madre (Ciudad, País)

Lugar de nacimiento del abuelo (Ciudad, País)

Lugar de nacimiento de la abuela (Ciudad, País)

Si ha vivido en otros lugares además de su país de origen, escriba una lista de los lugares

Lugar de residencia Tiempo de residencia

II. Historia de cómo adquirió su/s lengua/s.

Español Otra ¿Cuál? ______¿Cuál es su primera legua(s)? (Sí considera que ambas, marque las dos)

¿Cuál es su lengua dominante?

¿Qué lenguas se hablaban en tu casa cuando eras pequeño (0 a los 8 años)? ¿En qué lenguas se daban las clases en PRIMARIA?

142

… y en la SECUNDARIA?

… y en la UNIVERSIDAD?

III. Uso de lenguas.

Use la siguiente escala para contestar a las siguientes preguntas sobre el uso de su/s lengua/s.

1. Todos los días 2. Algunos días 3. Pocas veces a la semana 4. Una vez a la semana 5. Una o dos veces al mes 6. Una o dos veces al año 7. Casi nunca 8. Nunca

¿En español? Otra ¿cuál? ______

Con qué frecuencia escribe Con qué frecuencia habla Con que frecuencia escucha música Con qué frecuencia lee periódicos, revistas o libros Con qué frecuencia ve programas de televisión y películas

IV. Habilidad en su/s lenguas.

Usando una escala del 1 al 7, donde 1 es habilidad mínima, 4 moderada y 7 como la de un nativo/a, evalúe su habilidad en su/s lenguas en las siguientes categorías:

En español Otra ¿cuál? ______

Su habilidad para ….HABLAR

…ENTENDER

… ESCRIBIR

… LEER

… PRONUNCIAR

… GRAMÁTICA

… HABILIDAD TOTAL

143

APPENDIX F BILINGUAL SPEAKERS’ LBQ

Formulario de historia lingüística LOS INVESTIGADORES COMPLETARÁN ESTA PARTE Número de participante: Instrucciones: Complete el formulario en español. Generación: Primera / Segunda/ Tercera

III. Historia personal

Nombre

Fecha de Nacimiento

Sexo (elija uno) Hombre Mujer

Lugar de nacimiento (Ciudad, País)

Lugar de residencia actual (Ciudad, Estado)

Lugar de residencia de los últimos 10 años (Ciudad, Estado). Sí ha vivido en más de un sitio, indíquelo.

Lugar de nacimiento del padre (Ciudad, País)

Lugar de nacimiento de la madre (Ciudad, País)

Lugar de nacimiento del abuelo (Ciudad, País)

Lugar de nacimiento de la abuela (Ciudad, País)

Si ha nacido fuera de EEUU, ¿a qué edad llegó a los?

Si ha vivido en otros lugares además de los Estados Unidos, escriba una lista de los lugares

Lugar de residencia Tiempo de residencia

IV. Historia de cómo adquirió sus lenguas.

Español Inglés

¿Cuál es su primera legua(s)? (Sí considera que ambas, marque las dos)

¿Cuál es su lengua dominante?

¿Qué lenguas se hablaban en tu casa cuando eras pequeño ( 0 a los 8 años)?

144

¿En qué lenguas se daban las clases en PRIMARIA?

… y en la SECUNDARIA?

… y en la UNIVERSIDAD?

V. Uso de lenguas.

A. Use la siguiente escala para contestar a las siguientes preguntas sobre el uso de las lenguas.

1. Todos los días 2. Algunos días 3. Pocas veces a la semana 4. Una vez a la semana 5. Una o dos veces al mes 6. Una o dos veces al año 7. Casi nunca 8. Nunca

¿En español? ¿En inglés? Con qué frecuencia escribe Con qué frecuencia habla Con que frecuencia escucha música Con qué frecuencia lee periódicos, revistas o libros Con qué frecuencia ve programas de televisión y películas

¿Con qué frecuencia visita zonas en las que sólo se habla español?

B. Use la siguiente escala para contestar a las siguientes preguntas sobre el uso de las lenguas.

1. Siempre inglés 2. Más inglés que español 3. La misma cantidad 4. Más español que inglés 5. Siempre español

Número

En casa ¿qué lengua se usa más?

¿En qué lengua habla más con sus amigos más cercanos?

¿En qué lengua habla más con sus compañeros de clase?

¿En qué lengua habla más con sus compañeros de trabajo?

145

C. Usando la escala anterior del 1 al 5, responda las preguntas sobre con quién usa sus lenguas ¿Quién? Relación y edad Lengua en la que le habla a esta Lengua en la que esa persona le persona habla Ejemplo: madre, 50 años 5 5

VI. Habilidad en sus lenguas.

Usando una escala del 1 al 7, donde 1 es habilidad mínima, 4 moderada y 7 como la de un nativo, evalúe su habilidad en ambas lenguas en las siguientes categorías:

En español En inglés Su habilidad para HABLAR …ENTENDER … ESCRIBIR … LEER … PRONUNCIAR … GRAMÁTICA … HABILIDAD TOTAL VII. Bilingüísmo

Normalmente, ¿mezclas español e inglés (conocido como Spanglish)? Si_____ No_____ Si tu respuesta anterior fue Sí, responda a las siguientes preguntas usando la siguiente escala.

1. Siempre uso Spanglish 2. Aveces uso Spanglish 3. Nunca uso Spanglish

Número ¿Habla Spanglish con sus padres? ¿Habla Spanglish con su(s) hermano/a(s)? ¿Habla Spanglish con sus abuelos? ¿Habla Spanglish con sus amigos? ¿Habla Spanglish en su trabajo? ¿Habla Spanglish en sus clases (universidad, college, etc.)?

146

APPENDIX G MELICET (Adapted)

Participant: SECTION 1: Grammar

INSTRUCTIONS: Choose the word or phrase that best completes the conversation.

1. “What time will we arrive in San Francisco?” “I’m not sure, because I don’t know ______from here.”

choose one

2. “Did George enter the photography contest?” “No, but if he had, I think he ______.”

choose one

3. “What’s the matter?” “I feel ______out.”

choose one

4. “May I bring you a cup of tea?” “I prefer coffee ______tea.”

choose one

5. “Have you ever gone to Tahiti?” “No, but I have ______for a long time.”

choose one

6. “Will you come to my party on Saturday?” “______I’d like to, I can’t.”

choose one

7. “Don’t forget to pay the rent tomorrow!” “Please remind ______in the morning.”

choose one

8. “Susan plays the piano very well.”

147

“______that, she’s an excellent singer.”

choose one

“The ______over there.”

choose one

10. “Mark isn’t very smart, is he?” “Actually, he’s smarter than he ______to be.”

choose one

11. “What do you think of American football?” “I think it’s ______sport.”

choose one

12. “What shall we do about this problem?” “John suggests ______a meeting.”

choose one

13. “Where did you get those curtains?” “My wife made them ______an old tablecloth.”

choose one

14. “Do you like sugar in your coffee?” “Yes, ______better.”

choose one

15. “Why did John refuse to pay for his dinner?” “Because ______two hours by the time he was served.”

choose one

16. “When is the meeting going to begin?” “______Fred comes, we can get started.”

choose one

17. “Does John have a lot of accidents at work?” “Yes. He isn’t ______he should be.” choose one

148

18. “Did David enter the writing contest?” “Yes, he thinks he has ______.”

choose one

19. “Does Barbara have a difficult job?” “Yes. She is responsible ______many important decisions.”

choose one

20. “You gave me the wrong amount of money.” “How ______? I gave you what you asked for.”

choose one

21. “Will Bill’s report be ready by Friday?” “No, I don’t think he ______it by then.”

choose one

22. “When will this paint be dry?” “Not long. This is very ______paint.”

choose one

23. “Does Sue like circuses?” “Yes, the clowns always make ______.”

choose one

24. “Did you do well on the history test?” “No. I studied all night ______failed.”

choose one

25. “How do those shoes fit?” “My feet are too big ______them.”

choose one

26. “Do Mary’s children help with the housework?” “Yes, if she asks ______.”

choose one

149

27. “Where’s the box I asked for?” “Over there, ______on the table.”

choose one

28. “Let’s plan a picnic for Saturday.” “______it rains?”

choose one

29. “Is Lynn going to buy a new suit?” “Yes, she’s looking for a suit like ______.”

choose one

30. “That movie isn’t very good.” “Just wait. The best part ______.”

choose one

SECTION 2: Cloze

INSTRUCTIONS: Read the passage, then select the word which best fills the blank in both grammar and meaning.

Color is such a constant part of our environment that we tend to ignore its messages. Many people with perfect vision suffer 31 a sort of cultural color blindness. But 32 unnoticed color influences feelings as well. 33 of experiments with both infants and 34 indicate that blue light tends to 35 activity and produce a state of restfulness. 36 more tense a person is, the 37 blue will act as a tranquilizer. Red, 38 the contrary, excites the nervous system, 39 that if this page were printed 40 red paper, electrodes attached to your skin 41 show a definite increase in muscle 42 , restlessness, and eye movements compared with 43 reactions to the white page. Studies 44 found that patients in hospital rooms 45 red or other bright colors require 46 attention from nurses than patients in 47 painted in more subdued colors. Furthermore, 48 has been found that school children 49 more alert and learn faster in 50 painted rooms. However, this is unfortunately accompanied by an increase in restlesness and noisiness.

[Adapted from MELICET test]

150

APPENDIX H DELE (Adapted) Participante: SECCION 1: Texto Incompleto

INSTRUCCIONES: Complete el siguiente texto eligiendo para cada uno de los huecos una de las tres opciones que se le ofrecen.

NIÑOS SALUDABLES

Los padres siempre se están preguntando cómo conseguir que sus hijos sean unos niños talentosos y sanos y las soluciones pueden estar más cerca de lo que creemos. Ni tónicos, ni vitaminas, ni cursos de lectura veloz pueden conseguir tantos resultados en los niños 1 la práctica constante de hábitos saludables, Un sueño reparador, una alimentación sabia, 2 a una actividad física constante y el control del estrés son claves a la hora de potenciar habilidades naturales de los más pequeños.

3 contrario de lo que se creía, el sueño está lejos de ser una fase de hibernación mental. 4 que se descansa es la musculatura, pero en el cerebro se inician procesos fisiológicos fundamentales 5 el adecuado funcionamiento del niño, indispensables en la prevención de 6 enfermedad. El sueño es como el supermercado de noche, al momento del 7 no se apagan las luces, 8 que se encienden muchas más para limpiar las instalaciones y reponer los productos.

No solo 9 vital para el niño dormir las horas recomendadas, también que lo 10 a la hora del crepúsculo, pues en ese momento se 11 la disminución gradual de su actividad y la cantidad de estímulos que acuden a su cerebro desciende.

En la comida están los nutrientes básicos, 12 cumplen importantes funciones estructurales. 13 nacimiento en adelante, el niño obtendrá de ahí la materia prima para formar su cerebro y organismo. Si se 14 un niño talentoso, lo primeo es aplicar en 15 mismo las normas de alimentación saludable.

Las frutas, por ejemplo, deben consumirse más 16 tres veces al día, no hay que permitir que el yogur, otro gran alimento, les 17 protagonismo en la dieta de los chicos.

A pesar de los conocimientos, padres con las mejores intenciones se han topado con la barreara del gusto. Pero la preferencia por la comida sana también se puede educar, acostumbrándolos desde pequeños y explicando el 18 siempre.

El cuerpo humano está diseñado para moverse. Pero, en la actualidad, el sedentarismo ha limitado el crecimiento intelectual y emocional. Para evitarlo es crucial que los niños 19 una actividad

151

física constante, en forma sistemática. Lo preferible es la práctica de un deporte, por ejemplo, el tenis de mesa, que le 20 mucho al niño en términos de coordinación y estrategia. [Adaptado de El Mercurio, Chile] [Adapted from DELE test] SECCION 2: Vocabulario

INSTRUCCIONES: elige el significado de la palabra en negrita.

21. Tengo la impresión de que los libros que yo tenía de pequeña están dispersos por la casa de mis padres.

elige uno

22. Estábamos en plena reunión y, de buenas a primeras, la directora empezó con el tema de la subida de impuestos.

elige uno

23. Es un club muy exclusivo. Tiene una contraseña para poder entrar en determinados días.

elige uno

24. Esa decisión es inapelable; ahora que, si tú quieres, puedes hablar con Juan a ver qué te dice.

elige uno

25. Es necesario restituir el honor de esa persona porque, si no, no querrá asistir a una reunión con todos los demás representantes.

elige uno

26. Llegamos al aeropuerto a las tres y a duras penas cogimos el avión, no sin antes hablar por teléfono con una de nuestras familias.

elige uno

27. Decidieron tener una conversación previa a la firma del tratado para limar asperezas.

elige uno

28. En medio de los exámenes el hijo de Marta tuvo un bajonazo: por eso sigue preparándose para ellos.

elige uno

29. La situación familiar hizo que mi abuelo tomara cartas en el asunto en aquella época.

152

elige uno

30. Con ese aspecto de pasmado, es el mejor escritor de su generación.

elige uno SECCION 3: Gramática INSTRUCCIONES: Elige la opción correcta para cada una de las siguientes oraciones.

31. En la compañía se está decidiendo estos días si ______nuevos horarios para los trabajadores.

elige uno

32. María no era de la opinión de que ______todos a casa de Juan, pero al final fuimos.

elige uno

33. En las vacaciones en Brasil gasté mucho dinero, más ______pensaba: es que era todo tan bonito...

elige uno

34. A Luisa le dio ______decir que tenía sueño y se fue a casa.

elige uno

35. No tenemos ______idea de qué habrá podido pasar en la última jornada de Bolsa porque hemos estado de vacaciones.

elige uno

36. Yo creo que a Carlos no le gustó nada que ______en su casa sin avisar.

elige uno

37. No estoy dispuesta a irme sin que ______la verdad.

elige uno

38. No sé si a Clara ______han devuelto ya las maletas que perdió en el aeropuerto.

elige uno

39. ¿Dónde han estado los chicos toda la tarde, que no los he visto? - No sé, ______porque mañana tienen un examen importante.

153

elige uno

40. ¿Vas a asistir a la inauguración de la nueva sede?- Si tengo tiempo, ______hoy.

elige uno

41. Yo ______tú, hablaría con ella, es lo mejor para aclarar la situación.

elige uno 42. Ella le dijo que, si de verdad la ______, se lo demostrara.

elige uno

43. El hecho ______lo eliminaron de la lista de candidatos todavía no está claro.

elige uno

44. ______que se traslade a vivir a esta casa estará encantado con el paisaje alrededor.

elige uno

45. Nadie conseguirá aprobar ese examen ______se prepare a conciencia: es muy duro.

elige uno

46. ______salir de casa, se dio cuenta de que había dejado las llaves dentro.

elige uno

47. ______haber sabido que ibais a venir, habríamos preparado más comida.

elige uno

48. Había mucha gente que quería acudir al estreno de la película, ______decidiéramos ir otro día a verla.

elige uno

49. Cuando llegamos a la oficina ______15 personas esperando para hablar con nosotros.

elige uno

50. Nuestros hijos ya son mayores. ______arreglan muy bien en casa solos. elige uno

154

APPENDIX I LIST 1 AND 2 FILLERS WITH LEXICAL FREQUENCY

LIST 1 LIST 1 LIST 2 LIST 2 FILLERS FREQ. FILLERS FREQ. FILLERS FREQ. FILLERS FREQ. Lágrima 0.05 Campana 11.859 Lámpara 0.13 Tomate 6.96293 Melón 1.63432 Araña 12.125 Dedal 0.33141 Volante 7.66475 Estuche 1.77728 Bebé 13.243 Pañal 0.40614 Tambor 7.85320 Mantel 1.86826 Camisa 13.724 Enchufe 0.46138 Tanque 8.08389 Chef 1.88450 Dragón 14.793 Biberón 0.477625 Ratón 8.52902 Saxofón 2.44011 Naranja 14.959 Calcetín 0.526363 Rana 8.59400 Exclamación 2.82351 Botella 15.618 Jirafa 0.659578 Sillón 8.61675 Hélice 2.979 Fiebre 17.05 Tetera 0.711564 Martillo 8.99690 Mochila 3.04770 Estatua 17.353 Panal 0.922759 Oliva 9.254 Farol 3.24265 Nido 18.253 Jarrón 0.952002 Molino 9.610994 Cuchara 3.72352 Cañón 18.260 Hamaca 1.146951 Robot 9.692222 Menú 3.85024 Aceite 25.427 Sandía 1.172944 Piscina 10.38754 Bigote 3.9249 Bomba 29.463 Cebra 1.276917 Violín 11.54099 Maquillaje 4.01920 Medalla 32.085 Tractor 1.319156 Burro 11.72619 Maleta 4.042 Pared 33.235 Guisante 1.3419 Avestruz 11.75219 Langosta 4.16541 Cueva 37.284 Cereza 1.410132 Volcán 11.83991 Jamón 4.21415 Espada 39.308 Tacón 1.432876 Beso 12.54823 Delfín 4.63654 Montaña 42.151 Taza 1.432876 Ola 14.56595 Ballena 4.85423 Café 43.759 Tapón 1.462119 Vaca 14.62444 Eclipse 4.86398 Pintura 43.827 Escorpión 1.598 Navidad 16.49595 Cebolla 5.30911 Pecho 45.406 Broche 1.621327 Vela 19.79 Collar 5.35785 Avión 48.100 Jarabe 1.624576 Oveja 20.92454 Abeja 5.53980 Bosque 50.186 Peine 1.692809 Oreja 21.01227 Diente 5.65352 Jardín 53.074 Lechuga 1.696058 Anillo 23.69282 Uva 6.18 Ángel 59.284 Sartén 1.696058 Explosión 24.48562 Tejado 6.23837 Madera 60.460 Yate 1.894256 Tumba 24.79753 Aguja 6.33909 Carretera 65.528 Girasol 2.043717 Silla 24.80403 Hacha 6.55029 Cadena 68.664 Foca 2.105451 Tigre 26.4741 Caracol 6.85571 Hospital 68.781 Jarra 2.137942 Rama 28.10517 Limón 6.96738 Perro 69.970 Ardilla 2.371881 Gato 38.5577 Cohete 7.54128 Estación 86.888 Talón 2.423868 Catedral 39.53894 Diamante 7.57702 Suerte 90.589 Piña 2.488851 Círculo 40.903 Pirámide 7.70049 Vino 91.899 Canoa 2.716 Ventana 41.56641 Flauta 7.77522 Radio 104.08 Palmera 3.050954 Barco 42.73936 Manzana 8.12613 Copa 117.71 Toalla 3.057453 Cárcel 45.60186 Falda 8.33732 Televisión 143.18 Tiburón 3.408361 Clave 65.22 Mosca 8.48678 Oro 150.32 Paraguas 3.749522 Cumbre 70.14921

155

CON’ LIST 1 AND 2 FILLERS WITH LEXICAL FREQUENCY LIST 1 LIST 1 LIST 2 LIST 2 FILLERS FREQ. FILLERS FREQ. FILLERS FREQ. FILLERS FREQ. Balcón 9.19185 Paz 215.82 Graduación 4.630043 Piedra 79.51002 Mansión 10.1536 Iglesia 216.32 Olla 4.773005 Parque 88.31522 Cinturón 10.4850 Mar 258.56 Tortuga 5.634031 Clase 128.34 Saco 10.9301 Rey 328.45 Cubo 5.744502 Calle 145.9519 Cisne 11.1153 Río 358.93 Trompeta 5.78 Dinero 173.2903 Pantalón 11.4435 Mundo 680.96 Helado 5.87446 Ciudad 657.5538 Hormiga 11.8106 Cuna 10.644 Pulmón 6.01743 Sobre 2125.134

156

APPENDIX J VOCABULARY TEST EXPERIMENT 2/STIMULI 1

157

APPENDIX K VOCABULARY TEST EXPERIMENT 2/STIMULI 2

158

LIST OF REFERENCES

Abney, S. P. (1987). The English noun phrase in its sentential aspect (Doctoral dissertation, Massachusetts Institute of Technology). Alarcón, I. (2009). The processing of gender agreement in L1 and L2 Spanish: Evidence from reaction time data. Hispania, 814-828. Alarcón, I. V. (2011). Spanish gender agreement under complete and incomplete acquisition: Early and late bilinguals' linguistic behavior within the noun phrase. Bilingualism: Language and Cognition, 14(3), 332-350. Alarcos, E. (1999). Gramática de la lengua española. Madrid: Espasa Calpe. Alario, F. X., and Caramazza, A. (2002). The production of determiners: Evidence from French. Cognition, 82(3), 179-223. Antón‐Méndez, I., Nicol, J. L., & Garrett, M. F. (2002). The relation between gender and number agreement processing. Syntax, 5(1), 1-25. Audring, J. (2008). Gender assignment and gender agreement: Evidence from pronominal gender languages. Morphology, 18(2), 93-116. Badecker, W., Miozzo, M., & Zanuttini, R. (1995). The two-stage model of lexical retrieval: Evidence from a case of anomia with selective preservation of grammatical gender. Cognition, 57(2), 193-216. Bates, E., Devescovi, A., Hernandez, A., Pizzamiglio, L. (1996). Gender priming in Italian. Perception & Psychophysics, 58(7), 992–1004. 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. Biran, M., & Friedmann, N. (2012). The representation of lexical-syntactic information: Evidence from syntactic and lexical retrieval impairments in aphasia. Cortex, 48(9), 1103-1127. Birdsong, D., & Flege, J. E. (2001, November). Regular-irregular dissociations in L2 acquisition of English morphology. In Proceedings of the 25th annual Boston University conference on language development (pp. 123-132). Boston, MA: Cascadilla Press. Bock, K., & Cutting, J. C. (1992). Regulating mental energy: Performance units in language production. Journal of memory and language, 31(1), 99. Bock, K., & Eberhard, K. M. (1993). Meaning, sound and syntax in English number agreement. Language and Cognitive Processes, 8(1), 57-99. Bock, K., & Levelt, W. J. M. (1994). Language production: Grammatical encoding. In M.A. Gernsbacher (Ed.), Handbook of Psycholinguistics, 945-984. London: Academic Press. Bock, K., & Miller, C. A. (1991). Broken agreement. Cognitive psychology, 23(1), 45-93.

159

Butterworth, B. (1989). Lexical access in speech production. In W. Marslen-Wilson (Ed.), Lexical representation and process (pp. 108-135). Cambridge, MA, US: The MIT Press. Bybee, J. L. (1985). Morphology: A study of the relation between meaning and form (Vol. 9). John Benjamins Publishing. Bybee, J. (1995). Regular morphology and the lexicon. Language and cognitive processes, 10(5), 425-455. Caffarra, S., Janssen, N., & Barber, H. A. (2014). Two sides of gender: ERP evidence for the presence of two routes during gender agreement processing. Neuropsychologia, 63, 124- 134. Caramazza, A. (1997). How many levels of processing are there in lexical access? Cognitive neuropsychology, 14(1), 177-208. Caramazza, A., & Miozzo, M. (1997). The relation between syntactic and phonological knowledge in lexical access: evidence from the tip-of-the tongue phenomenon. Cognition, 64(3), 309-343. Caramazza, A., Miozzo, M., Costa, A., Schiller, N. O., & Alario, F. X. (2001). A cross-linguistic investigation of determiner production. Language, brain, and cognitive development: Essays in honor of Jacques Mehler, 209-226. Carroll, S. (1989). Language acquisition studies and a feasible theory of grammar. Canadian Journal of Linguistics/Revue canadienne de linguistique, 34(4), 399-418. Chomsky, N. (1993). A minimalist program for linguistic theory. The view from Building 20: Essays in linguistics in honor of Sylvain Bromberger, ed. by Kenneth Hale and Samuel Jay Keyser, 1-52. Chomsky, N. (1995). Categories and transformations. The minimalist program, 219, 394. Clahsen, H. (1998). Linguistic perspectives on specific language impairment. Department of Language and Linguistics, University of Essex. Corrêa, L. (2001) Uma hipótese para a identificação do gênero gramatical com par-ticular referência para o português. Letras de Hoje 36 (3), 289-295. Costa, A., Alario, F. X., & Sebastián-Gallés, N. (2007). Cross-linguistic research on language production. Oxford handbook of psycholinguistics, 531-546. Costa, A., Caramazza, A., and Sebastian-Galles, N. (2000). The cognate facilitation effect: implications for models of lexical access. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26(5), 1283. Costa, A., Kovacic, D., Franck, J., and Caramazza, A. (2003). On the autonomy of the grammatical gender systems of the two languages of a bilingual. Bilingualism: Language and Cognition, 6(3), 181-200.

160

Costa, A., Miozzo, M., & Caramazza, A. (1999). Lexical selection in bilinguals: Do words in the bilingual's two lexicons compete for selection? Journal of Memory and language, 41(3), 365-397. Cubelli, R., & Perizzi, S. (1996). Selective loss of grammatical gender of nouns in aphasia. Paper presented at the Istituto di Psicologia del CNR, Roma, December 10. Cubelli, R., Lotto, L., Paolieri, D., Girelli, M., & Job, R. (2005). Grammatical gender is selected in bare noun production: Evidence from the picture–word interference paradigm. Journal of Memory and Language, 53(1), 42-59. Cuza, A., & Pérez-Tattam, R. (2016). Grammatical gender selection and phrasal word order in child heritage Spanish: A feature re-assembly approach. Bilingualism: Language and Cognition, 19(1), 50-68. De Garavito, J. B., & White, L. (2002). The second language acquisition of Spanish DPs: The status of grammatical features. In The acquisition of Spanish morphosyntax (pp. 153- 178). Springer, Dordrecht. Dell, G. S. (1986). A spreading-activation theory of retrieval in sentence production. Psychological review, 93(3), 283. Dell, G. S. (1988). The retrieval of phonological forms in production: Tests of predictions from a connectionist model. Journal of memory and language, 27(2), 124-142. Dell, G. S., Burger, L. K., & Svec, W. R. (1997). Language production and serial order: A functional analysis and a model. Psychological review, 104(1), 123. Dewaele, J. M., & Véronique, D. (2001). Gender assignment and gender agreement in advanced French interlanguage: A cross-sectional study. Bilingualism: Language and Cognition, 4(3), 275-297. Dotan, D., & Friedmann, N. (2007). From seven dwarfs to four wolves: Differences in the processing of number words and other words. Language and Brain, 6, 3-17. Dotan, D., & Friedmann, N. (2010). Words and numbers in the phonological output buffer. Procedia-Social and Behavioral Sciences, 6, 82-83. Duchon, A., Perea, M., Sebastián-Gallés, N., Martí, A., Carreiras, M. (2013) EsPal: One-stop Shopping for Spanish Word Properties. Behavior Research Methods, 45: 1246-1258. Eberhard, K. M. (1997). The marked effect of number on subject–verb agreement. Journal of Memory and language, 36(2), 147-164. Eberhard, K. M., Cutting, J. C., & Bock, K. (2005). Making syntax of sense: number agreement in sentence production. Psychological Review, 112(3), 531. Emmorey, K., Petrich, J. A., & Gollan, T. H. (2012). Bilingual processing of ASL–English code- blends: The consequences of accessing two lexical representations simultaneously. Journal of memory and language, 67(1), 199-210.

161

Emmorey, K., Petrich, J. A., & Gollan, T. H. (2012). Bilingual processing of ASL–English code- blends: The consequences of accessing two lexical representations simultaneously. Journal of memory and language, 67(1), 199-210. Federmeier, K. D., Wlotko, E. W., & Meyer, A. M. (2008). What's ‘Right’ in Language Comprehension: Event‐Related Potentials Reveal Right Hemisphere Language Capabilities. Language and linguistics compass, 2(1), 1-17. Fernández-García, M. (1999). Patterns of gender agreement in the speech of second language learners. In Advances in Hispanic linguistics: Papers from the 2nd Hispanic linguistics symposium, 3-15. Somerville, MA: Cascadilla Press. Ferreira, F., & Swets, B. (2002). How incremental is language production? Evidence from the production of utterances requiring the computation of arithmetic sums. Journal of Memory and Language, 46(1), 57-84. Field, F. Bilingualism in the USA: The case of the Chicano-Latino community. Vol. 44. John Benjamins Publishing, 2011. Finneman, M. D. (1992). Learning agreement in the noun phrase: The strategies of three first- year Spanish students. International Review of Applied Linguistics in Language Teaching, 30, 121–137. Foote, R. (2014). Age of acquisition and sensitivity to gender in Spanish word recognition. Language Acquisition, 21(4), 365-385. Foote, R. (2015). The production of gender agreement in native and L2 Spanish: The role of morphophonological form. Second Language Research, 31(3), 343-373. Gahl, S., & Yu, A. C. (2006). Introduction to the special issue on exemplar-based models in linguistics. The linguistic review, 23(3), 213-216. Garret, M. F. (1980). Levels of processing in sentence production. Speech and talk, 177-220. Garrett, M. F. (1975). The analysis of sentence production. The psychology of learning and motivation, 9, 133-177. Garrett, M. F. (1976). Syntactic processes in sentence production. New approaches to language mechanisms, 30, 231-256. Gennari, S. P., & MacDonald, M. C. (2009). Linking production and comprehension processes: The case of relative clauses. Cognition, 111(1), 1-23. Gollan, T. H., & Frost, R. (2001). Two routes to grammatical gender: Evidence from Hebrew. Journal of Psycholinguistic Research, 30(6), 627-651. Gollan, T. H., Montoya, R. I., Cera, C., & Sandoval, T. C. (2008). More use almost always means a smaller frequency effect: Aging, bilingualism, and the weaker links hypothesis. Journal of memory and language, 58(3), 787-814.

162

Gollan, T. H., Montoya, R. I., Fennema-Notestine, C., & Morris, S. K. (2005). Bilingualism affects picture naming but not picture classification. Memory & cognition, 33(7), 1220- 1234. Gollan, T. H., Slattery, T. J., Goldenberg, D., Van Assche, E., Duyck, W., & Rayner, K. (2011). Frequency drives lexical access in reading but not in speaking: The frequency-lag hypothesis. Journal of Experimental Psychology: General, 140(2), 186. Gonon, M. H., Bruckert, R., & Michel, F. (1989). Lexicalization in an anomic patient. Neuropsychologia, 27(4), 391-407. Gollan, T. H., Weissberger, G. H., Runnqvist, E., Montoya, R. I., & Cera, C. M. (2012). Self- ratings of spoken language dominance: A Multilingual Naming Test (MINT) and preliminary norms for young and aging Spanish–English bilinguals. Bilingualism: Language and Cognition, 15(3), 594-615. Green, J. N. (1988). Spanish. In M. Harris, & N. Vincent (Eds.), The Romance languages, 79- 130. New York: Oxford University Press. Grosjean, F. (1997). The bilingual individual. Interpreting, 2(1), 163-187. Grosjean, F., & Li, P. (2012). The psycholinguistics of bilingualism. John Wiley & Sons. Harris, J. W. (1991). The exponence of gender in Spanish. Linguistic Inquiry, 22(1), 27-62. Hawkins, R., & Chan, Y-C. (1997). The partial availability of Universal Grammar in second language acquisition: the ’failed features’ hypothesis. Second Language Research, 13, 187-226. Hellige, J. B. (1990). Hemispheric asymmetry. Annual review of psychology, 41(1), 55-80. Hellige, J. B., & Sergent, J. (1986). Role of task factors in visual field asymmetries. Brain and cognition, 5(2), 200-222. Indefrey, P., & Levelt, W. J. (2004). The spatial and temporal signatures of word production components. Cognition, 92(1-2), 101-144. Jackendoff, R. (2002). English particle constructions, the lexicon, and the autonomy of syntax. Verb-particle explorations, 67-94. Jescheniak, J. D., Schriefers, H., & Hantsch, A. (2003). Utterance format effects phonological priming in the picture-word task: Implications for models of phonological encoding in speech production. Journal of Experimental Psychology: Human Perception and Performance, 29(2), 441. Jescheniak, J. D., Schriefers, H., Garrett, M. F., & Friederici, A. D. (2002). Exploring the activation of semantic and phonological codes during speech planning with event-related brain potentials. Journal of Cognitive Neuroscience, 14(6), 951-964.

163

Kemmer, L., Coulson, S., & Kutas, M. (2014). Grammatical number agreement processing using visual half-field paradigm: an event-related brain potential study. International Journal of Psychophysiology, 91, 88–103. Kempen, G., & Hoenkamp, E. (1987). An incremental procedural grammar for sentence formulation. Cognitive science, 11(2), 201-258. Kempen, G., & Huijbers, P. (1983). The lexicalization process in sentence production and naming: Indirect election of words. Cognition, 14(2), 185-209. Koenig, O., Wetzel, C., & Caramazza, A. (1992). Evidence for different types of lexical representations in the cerebral hemispheres. Cognitive Neuropsychology, 9(1), 33-45. Kohn, S. E., & Melvold, J. (2000). Effects of morphological complexity on phonological output deficits in fluent and nonfluent aphasia. Brain and Language, 73(3), 323-346. Kulke, F., & Blanken, G. (2001). Phonological and syntactic influences on semantic misnamings in aphasia. Aphasiology, 15(1), 3-15. La Heij, W., Mak, P., Sander, J., & Willeboordse, E. (1998). The gender-congruency effect in picture-word tasks. Psychological Research, 61(3), 209-219. Levelt, W. J. (1989). Self-monitoring and self-repair. Speaking: From intention to articulation, 12, 458-499. Levelt, W. J. (1992). Accessing words in speech production: Stages, processes and representations. Cognition, 42(1-3), 1-22. Levelt, W. J. M. (1989). Working models of perception; five general issues. In B. A. S. Elsendorn & H. Bouma (Eds.), Working models of perception, 489-503. London: Academic Press. Levelt, W. J. M., Praamstra, P., Meyer, A. S., Helenius, P., & Salmelin, R. (1998). An MEG study of picture naming. Journal of Cognitive Neuroscience, 10, 553-567. Levelt, W. J., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical access in speech production. Behavioral and brain sciences, 22(1), 1-38. Lipski, J. (1993). Creoloid phenomena in the Spanish of transitional bilinguals. Spanish in the United States: linguistic contact and diversity, 155-182. Lleó, C. (1998). Proto-articles in the acquisition of Spanish: Interface between phonology and morphology. Modelle der Flexion, 18, 175-195. Lloret, M. R., & Viaplana, J. (1998). Variació morfofonològica. Variants morfològiques. Caplletra. Revista Internacional de Filologia, 25, 43-62. López Ornat, S. (1997). What lies in between a pre-grammatical and a grammatical representation: evidence on nominal and verbal form-function mappings in Spanish from 1; 7 to 2; 1. Contemporary perspectives on the acquisition of Spanish, 1, 3-20.

164

MacWhinney, B. (2001). The competition model: The input, the context, and the brain. Cognition and second language instruction, 69-90. Martínez-Gibson, E. A. (2011). A comparative study on gender agreement errors in the spoken Spanish of heritage speakers and second language learners. Porta Linguarum: revista internacional de didáctica de las lenguas extranjeras, (15), 177-193. Meyer, A. S. (1996). Lexical access in phrase and sentence production: Results from picture– word interference experiments. Journal of memory and Language, 35(4), 477-496. Meyer, A. S., & Belke, E. (2007). Word form retrieval in language production. In The Oxford handbook of psycholinguistics. Miceli, G., & Caramazza, A. (1988). Dissociation of inflectional and derivational morphology. Brain and language, 35(1), 24-65. Montrul, S. A. (2004). The acquisition of Spanish: Morphosyntactic development in monolingual and bilingual L1 acquisition and adult L2 acquisition (Vol. 37). John Benjamins Publishing. Montrul, S. A. (2008). Incomplete acquisition in bilingualism: Re-examining the age factor (Vol. 39). John Benjamins Publishing. Montrul, S. A. (2012). Is the heritage language like a second language? Eurosla Yearbook, 12(1), 1-29. Montrul, S., & Potowski, K. (2007). Command of gender agreement in school-age Spanish- English bilingual children. International Journal of Bilingualism, 11(3), 301-328. Montrul, S., Davidson, J., De La Fuente, I., & Foote, R. (2014). Early language experience facilitates the processing of gender agreement in Spanish heritage speakers. Bilingualism: Language and Cognition, 17(1), 118-138. Montrul, S., De La Fuente, I., Davidson, J., & Foote, R. (2013). The role of experience in the acquisition and production of diminutives and gender in Spanish: Evidence from L2 learners and heritage speakers. Second Language Research, 29(1), 87-118. Montrul, S., Foote, R., & Perpiñán, S. (2008). Gender agreement in adult second language learners and Spanish heritage speakers: The effects of age and context of acquisition. Language Learning, 58(3), 503-553. Montrul, S., Foote, R., & Perpiñán, S. (2008). Gender agreement in adult second language learners and Spanish heritage speakers: The effects of age and context of acquisition. Language Learning, 58(3), 503-553. Navarrete, E., Basagni, B., Alario, F. X., & Costa, A. (2006). Does word frequency affect lexical selection in speech production? Quarterly Journal of Experimental Psychology, 59(10), 1681-1690.

165

O'Rourke, P. L., & Van Petten, C. (2011). Morphological agreement at a distance: Dissociation between early and late components of the event-related brain potential. Brain research, 1392, 62-79. Pascual Cabo, D. (2013). Agreement reflexes of emerging optionality in heritage speaker Spanish. [electronic resource]. [Gainesville, Fla.]: University of Florida, 2013. Patterson, K. and Shewell, C.(1987) Speak and spell: dissociations and word-class effects. In M. Coltheart, G. Santori and R. Job, The cognitive neuropsychology of language, 273-294, London:Lawrence Erlband. Pinker, S. (1991). Rules of language. Science, 253(5019), 530-535. Pinker, S. (1999). How the mind works. Annals of the New York Academy of Sciences, 882(1), 119-127. Polinsky, M. (2006). Incomplete acquisition: American Russian. Journal of Slavic linguistics, 191- 262. Polinsky, M. (2011). Reanalysis in adult heritage language: New evidence in support of attrition. Studies in second language acquisition, 33(2), 305-328. Prévost, P., & White, L. (2000). Missing surface inflection or impairment in second language acquisition? Evidence from tense and agreement. Second language research, 16(2), 103- 133. Ralph, L., Sage, K., Green, H., Berthier, M. L., Maritnez Cuitin, M., Torralva, T., and Patterson, K. (2011). El-La: the impact of degraded semantic representations on knowledge of grammatical gender in semantic dementia. Acta Neuropsychologica, 9(2), 115-1132. Resende, N. C. A., & Mota, M. B. (2017). A behavioral study to investigate the processing routes of grammatical gender in Brazilian Portuguese. Revista de Estudos da Linguagem, 25(3), 1367-1395. Roelofs, A. (1992). A spreading-activation theory of lemma retrieval in speaking. Cognition, 42 (1-3), 107-142. Rosinski, R. R., Golinkoff, R. M., and Kukish, K. S. (1975). Automatic semantic processing in a picture-word interference task. Child Development, 247-253. Runnqvist, E., Gollan, T. H., Costa, A., & Ferreira, V. S. (2013). A disadvantage in bilingual sentence production modulated by syntactic frequency and similarity across languages. Cognition, 129(2), 256-263. Sagarra, N., & Herschensohn, J. (2010). The role of proficiency and working memory in gender and number agreement processing in L1 and L2 Spanish. Lingua, 120(8), 2022-2039. Schiller, N. O., and Caramazza, A. (2003). Grammatical feature selection in noun phrase production: Evidence from German and Dutch. Journal of memory and language, 48(1), 169-194.

166

Schmid, M. S., Köpke, B., Keijzer, M., & Weilemar, L. (Eds.). (2004). First language attrition: Interdisciplinary perspectives on methodological issues (Vol. 28). John Benjamins Publishing. Schmitt, B. M., Münte, T. F., & Kutas, M. (2000). Electrophysiological estimates of the time course of semantic and phonological encoding during implicit picture naming. Psychophysiology, 37, 473 –484. Schneider, W., Eschman, A., & Zuccolotto, A. (2002) E-Prime User’s Guide. Pittsburgh: Psychology Software Tools Inc. Schriefers, H., and Teruel, E. (2000). Grammatical gender in noun phrase production: The gender interference effect in German. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26(6), 1368. Schwartz, B. D., & Sprouse, R. A. (1997). Transfer: A tradition in transition. Presentation at the American Association of Applied Linguistics. Semenza, C., Mondini, S., & Cappelletti, M. (1995). A specific difficulty for the syntactic properties of non-countable nouns. Brain and Language, 51(1), 9-11. Silva-Corvalán, C. (1994). Language Contact and Change: Spanish in Los Angeles. Oxford University Press, 198 Madison Avenue, New York, NY 10016-4314. Smith, M., & Wheeldon, L. (1999). High level processing scope in spoken sentence production. Cognition, 73(3), 205-246. Starreveld, P., & La Heij, W. (2004). Phonological facilitation of grammatical gender retrieval. Language and Cognitive Processes, 19(6), 677-711. Steele, S. (1978). Word order variation: A typological study. Universals of human language, Vol 4, Syntax, ed. by Joseph Greenberg, 585-623. Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of experimental psychology, 18(6), 643. Taft, M., & Meunier, F. (1998). Lexical representation of gender: A quasiregular domain. Journal of Psycholinguistic research, 27(1), 23-45. Teschner, R. V., & Russell, W. M. (1984). The gender patterns of Spanish nouns: An inverse dictionary-based analysis. Hispanic Linguistics, 1(1), 115-132. Ullman, M. T. (2001). The neural basis of lexicon and grammar in first and second language: The declarative/procedural model. Bilingualism: Language and cognition, 4(2), 105-122. Urrutia, M., Domínguez, A., & Álvarez, C. J. (2009). Gender activation in transparent and opaque words. Psicothema, 21(1). Valdés, G. (2005). Bilingualism, heritage language learners, and SLA research: Opportunities lost or seized?. The Modern Language Journal, 89(3), 410-426.

167

Valdés, G. (2014). Heritage language students: Profiles and possibilities. In Handbook of heritage, community, and native American languages in the United States, 41-49. Routledge. Valenzuela, E., Faure, A., Ramírez-Trujillo, A. P., Barski, E., Pangtay, Y., & Diez, A. (2012). Gender and heritage Spanish bilingual grammars: A study of code-mixed determiner phrases and copula constructions. Hispania, 95(3), 481-494. Van Berkum, J. J. (1996). The psycholinguistics of grammatical gender: Studies in language comprehension and production. Nijmegen: Max Planck Instituut voor Psycholinguïstiek. Van Berkum, J. J., Brown, C. M., Zwitserlood, P., Kooijman, V., & Hagoort, P. (2005). Anticipating upcoming words in discourse: evidence from ERPs and reading times. Journal of Experimental Psychology: Learning, Memory, and Cognition, 31(3), 443. Van Turennout, M., Hagoort, P., & Brown, C. M. (1998). Brain activity during speaking: From syntax to phonology in 40 milliseconds. Science, 280(5363), 572-574. Vigliocco, G., & Franck, J. (1999). When sex and syntax go hand in hand: Gender agreement in language production. Journal of Memory and Language, 40(4), 455-478. Vigliocco, G., Antonini, T., & Garrett, M. F. (1997). Grammatical gender is on the tip of Italian tongues. Psychological science, 8(4), 314-317. White, L., Valenzuela, E., Kozlowska–Macgregor, M., & Leung, Y. K. I. (2004). Gender and number agreement in nonnative Spanish. Applied psycholinguistics, 25(1), 105-133. Yip, V., & Matthews, S. (2007). The bilingual child. Ernst Klett Sprachen. Zaidel, E. (1983). A response to Gazzaniga: Language in the right hemisphere, convergent perspectives. American Psychologist, 38(5), 542-546.

168

BIOGRAPHICAL SKETCH

A native of Alicante, Valencian Community (Spain), Dámaris Mayans Ramón obtained her Bachelor Degree in English Philology at the University of Alicante (Spain). In 2011, she obtained a Master Degree in Spanish and a Certificate in Translation Studies at the University of

Illinois, Urbana-Champaign (Illinois). Subsequently, she pursued her Ph.D. in Romance

Languages at the University of Florida specializing in Hispanic Linguistics and obtained her doctoral degree in 2018. Her primary research interests are psycholinguistic aspects of bilingual and multilingual acquisition with a focus on adult heritage speakers´ morphosyntactic processing. Dámaris currently holds a position of Assistant Professor at Colby College.

169