SENTENCE PROCESSING IN SPANISH BILINGUALS: EXPLORING ADJECTIVAL INTERPRETATIONS By CARLIE OVERFELT A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS UNIVERSITY OF FLORIDA 2012
1 2012 Carlie Overfelt
2 To my fam ily, for their love and support
3 ACKNOWLEDGEMENTS I would like to thank my advisors, Dr. Edith Kaan and Dr. Jason Rothman for their endless advice and support throughout this project. I would also like to thank my lab assistant Jose Ceballos for his aid and moral support in carrying out all phases of the study. Without the combination of aid from these three individuals, as well as from family and friends, this project would have been unattainable. You all have my deepest gratitude for all your assistance and patience throughout this process. Thanks g o out to all.
4 TABLE OF CONTENTS page ACKNOWLEDGEMENTS ................................ ................................ ............................... 3 LIST OF TABLES ................................ ................................ ................................ ............ 6 LIST OF FIGURES ................................ ................................ ................................ .......... 7 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 12 2 BACKGROUND ................................ ................................ ................................ ...... 14 Accounts of Second Language Processing ................................ ............................ 14 A Continuation of Rothman et al. (2010) ................................ ................................ 18 3 THE CURRENT STUDY ................................ ................................ ......................... 23 4 METHODOLOGY ................................ ................................ ................................ ... 25 Participants ................................ ................................ ................................ ............. 25 Stimuli/Experimental Design ................................ ................................ ................... 27 Targets ................................ ................................ ................................ ............. 27 M eaning Changers ................................ ................................ ........................... 31 Procedure ................................ ................................ ................................ ............... 33 5 RESULTS ................................ ................................ ................................ ............... 38 Off Line ................................ ................................ ................................ ................... 38 On Line ................................ ................................ ................................ ................... 38 Accuracy ................................ ................................ ................................ ........... 39 Experimentals ................................ ................................ ............................ 39 Meaning changers ................................ ................................ ..................... 40 Reaction Time ................................ ................................ ................................ ......... 43 Questions ................................ ................................ ................................ ......... 43 Experimentals ................................ ................................ ............................ 43 Meaning changers ................................ ................................ ..................... 43 Sentence Level Reaction Times, Word Positions 4 9 ................................ ....... 45 Experimentals ................................ ................................ ............................ 45 Meaning changers ................................ ................................ ..................... 47 Additional investigation ................................ ................................ .............. 47 Only accurate trials ................................ ................................ .................... 56 Top performers ................................ ................................ .......................... 58
5 6 DISCUSSION ................................ ................................ ................................ ......... 59 Confounds ................................ ................................ ................................ .............. 59 Predictions ................................ ................................ ................................ .............. 61 Accuracy ................................ ................................ ................................ ................. 62 Reaction Time ................................ ................................ ................................ ......... 65 Questions ................................ ................................ ................................ ......... 65 Self Paced Reading Word Positions 4 9 ................................ .......................... 66 On line versus off line ................................ ................................ ................ 68 Theoretical implications ................................ ................................ ............. 69 7 CONCLUSIONS ................................ ................................ ................................ ..... 70 APPENDIX A EXPERIMENTAL ADJECTIVES ................................ ................................ ............. 72 Self Paced Reading Items ................................ ................................ ...................... 72 Experimental Adjectives (32) ................................ ................................ ............ 72 Meaning Changing Adjectives (24) ................................ ................................ ... 78 B OFF LINE MATERIALS ................................ ................................ .......................... 82 Language and Education Questionnair e ................................ ................................ 82 Debriefing Questions ................................ ................................ .............................. 87 Vocabulary ................................ ................................ ................................ .............. 88 Semantic Interpretation Task: Meaning Changers ................................ ................. 89 Semantic Interpretation Task: Experimentals ................................ ........................ 91 REFERENCE S ................................ ................................ ................................ .............. 97 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 100
6 LIST OF TABLES Table page 4 1 Participant demographics by group ................................ ................................ .... 27 4 2 Conditions explained by word order in determiner phrase ................................ .. 28 5 1 Off line task scores, by group ................................ ................................ ............. 38 5 2 Accuracy on plausibility questions, by condition, experim entals ......................... 41 5 3 Accuracy on plausibility questions, by group ................................ ...................... 41 5 4 Accuracy on plausibility questions, by condition, meaning changers .................. 41 5 5 Response time to plausibility questions, by condition, experimentals ................. 44 5 6 Response time to plausibility questions, by condition, meaning changers ......... 44 5 7 Mean response time to plausibility questions, by gr oup, meaning changers and experimentals (in ms.) ................................ ................................ ................. 44
7 LIST OF FIGURES Figure page 2 1 Experimental adjective semantics ................................ ................................ ...... 19 2 2 Meaning changing adjective semantics ................................ .............................. 19 2 3 Syntax of adjectival movement ................................ ................................ ........... 21 2 4 Off line tasks from Rothman (2010) ................................ ................................ .... 21 4 1 Full example of an experimental item ................................ ................................ 29 4 2 Semantic interpretation task ................................ ................................ ............... 37 5 1 Accuracy of plausibility questions, comparing meaning changers and experim entals ................................ ................................ ................................ ..... 42 5 2 Average reaction time to end of sentence plausibility question, by condition, experimentals ................................ ................................ ................................ ..... 45 5 3 Average reaction time to end of sentence plausibility question, by condition, meaning changers ................................ ................................ .............................. 45 5 4 Reaction times on word positions 4 9, conditions a and d, experimentals .......... 50 5 5 Reaction times on word positions 4 9, conditions c and b, experimentals .......... 51 5 6 Reaction times on word positions 4 9, conditions F1 and F4, meaning changers ................................ ................................ ................................ ............. 51 5 7 Reaction times on word positions 4 9, conditions F3 and F2, meaning changers ................................ ................................ ................................ ............. 52 5 8 Reaction times for experimental conditions by word position, overall ................. 52 5 9 Reaction times for group 1 by word position, natives, experimentals ................. 53 5 10 Rea ction times for group 2 by word position, heritage, experimentals ................ 53 5 11 Reaction times for group 3 by word position, le arners, experimentals ................ 54 5 12 Reaction times for meaning changer conditions by word position, overall .......... 54 5 13 Reaction times for group 1 by word position, natives, meaning changers .......... 55 5 14 Reaction times for group 2 by word position, heritage, meaning changers ........ 55
8 5 15 Reaction times for group 3 by word position, learners, meaning changers ........ 56 5 16 Reaction times for only accurate trials by word position, experimentals ............. 57 5 17 Reaction times for all trials by word position, experimentals .............................. 58 6 1 Practice question ................................ ................................ ................................ 64
9 LIST OF ABBREVIATIONS AN adjective n oun, pre nominal adjective placement, non canonical word ordering DP/DPs determiner p hrase, the phrase including the determiner, adjective, and noun FA/FAs Full Access account ( s ) gram. g rammatical NA noun a djective, post nominal adjective placement, canonical word ordering NumP number phrase RDA/ RDAs Representational Deficit a ccount ( s ) ungram. ungrammatical UG Universal Grammar WMP word marker phrase
10 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requi rements for the Degree of Master of Arts SENTENCE PROCESSING IN SPANISH BILINGUALS: EXPLORING ADJECTIVAL INTERPRETATIONS By Carlie Overfelt December 2012 Chair: Edith Kaan Major: Linguistics Sentence level processing in non native speakers of Spanish has been a widely studied topic in the field of second language acquisition. This paper attempts to add evidence to a recent area of study that utilizes both off line and on line tasks to compare native speakers and second language learners, in order to build evidence in favor of either accounts of representational deficits, or full access accounts. Participants (n=36) of varying levels of S panish were tested using an adjectival manipulation that has both semantic and syntactic entailments, as first seen in Rothman et al. 2010 1 This previous study found that higher proficiency intermediate level learners of Spanish could attain native like linguistic competence in the feature tested, and this evidence did not support accounts of representational deficits. In the present study, we implemented one of the same off line tasks used previously, and added an on line self paced reading task. This would allow for us to compare both explicit and implicit knowledge of the manipulation in question. We used the same adjectival manipulation, which had very subtle shifts in 1 Rothman, J., Judy, T., Guijarro Fuentes, P., & Pires, A. (2010). ON THE (UN) AMBIGUITY OF ADJECTIVAL MODIFICATION IN SPANISH DETERMINER PHRASES informing debates on the mental representations of L2 syntax. Studies in Second Language Acquisition, 32 (1), 47 77.
11 meaning between pre and post nominal adjective placement, as well as a special se t of adjectives that had very strong changes in meaning when placed either before or after the noun in the determiner phrase. We found that natives and learners showed no statistical differences in either the on line or off line tasks to either the experi mental adjectives or the meaning changing adjectives, though neither group demonstrated any sensitivity to either type of adjective type tested. Instead the participants showed sensitivity only to the conditions containing non canonical word ordering, whe re the adjective appeared first in the DP. We rejected accounts of representational deficits, but only tentatively favored accounts of full access.
12 CHAPTER 1 INTRODUCTION A central debate in the study of second language acquisition is the question of how learners of a second language differ from native speakers of the same language. The way a child learns his first language and the way a teen or adult acquires an additiona l language is not surprisingly very different. In turn, the question is frequently raised as to how these different paths of acquisition affect the way these disparate individuals process the language in question. We know that there are many observable d ifferences in the grammars of L1 and L2 speakers. These have been referred to as optionality, variability, fossilization, among others. Here, we will explore two different accounts that attempted to explain the variances between L1 and L2 grammars: acco unts of representational deficits, and generative accounts such as Full Access. Representational deficit accounts, in general, state that second language grammars are lacking in some sense. The shallow structures hypothesis (Clahsen and Felser, 2006), for example, suggests that L2 grammars are less detailed than those of native speakers. In opposition, other researchers take a more generative approach to second language acquisition, suggesting that both L1 and L2 learners have the same inherent ability to acquire a language, as guided by the principles of Universal Grammar. In particular, we look at accounts of Full Access, which say that learners are able to tap into UG like children learning a second language. The study reported in this thesis is a continuation of Rothman Judy, Gui ja r ro Fuentes, and Pires (2010) that tested native Spanish speakers versus native English speaking learners of Spanish. Using an adjectival placement manipulation, these researchers conducted two off line experiments. These tasks looked at the differences
13 in semantic meaning, which vary between post nominal adjectival placement (canonical) and pre nominal adjective placement, as forced by preceding contexts. Rothman et al. found that advanced learners of Spanish were able to converge on native like processing abilities, and thus concluded that no evidence for accounts of representational deficits was present. In the present study, we intend to use the same adjectival manipulation mentioned above (to be discussed in gr eater detail below), while having both off line and on line tasks for the participants to complete. Should our results confirm those of the previous study, it would lend strong evidence, both off line and in real time, that learners of a second language a re capable of acquiring native like processing abilities, and that their underlying grammar is not deficient or incomplete. The paper will proceed with an overview of relevant research on this topic (Chapter 2), followed by a run through of the methodolo gy and goals of the experiments (Chapter 3 and 4), and will close with an overview of the results, a discussion of the results and final conclusions based on the theoretical framework laid out below (Chapter 5, 6, and 7).
14 CHAPTER 2 BACKGROUND Accounts of Second Language Processing Accounts of representational deficits posit, as the name would suggest, that L2 learners have mental representations that are in some way deficient as compared to native speakers. This assumes that learners of a second lang uage are not capable of reaching a native like ability in their L2. One popular account that falls under this category is the Shallow Structures hypothesis (Clahsen and Felser, 2006). In this model, L2 learners have representations during processing that contain less detail than native speakers. In other words, the processing is less profound. This means that learners rely on more superficial cues to process a phrase, as opposed to deeper syntactic structures. For example, a learner attempting to use the past tense might rely Declarative/Procedural model (2004), which posits a declarative and a procedural memory system, in that L2 learners will process lexical i tems easier than underlying syntactic rules or grammar. The reliance on declarative systems is akin to the reliance sing, coined by Ferreira, Baile y, and Ferraro (2002), is the idea that even native speakers only partially process most sentences, as evidenced by studies on garden path sentences and passives. The difference between the L1 and L2 speakers is that L1 spea kers have the native underlying representations, which L2 learners lack, even though the structure of the processing system is similar.
15 Two other accounts of representational deficits include the Failed Functional Features Hypothesis (FFFH)(Hawkins and Chan, 1997) and the Interpretability Hypothesis (IH)(Tsimpli and Mastropavlou, 2007). The FFFH proposes a declining access to functional categories after a certain critical period. This theory claims that the learner has some access to Universal Grammar but interference from the L1 results in the inability to instantiate features of functional categories that differ from their primary language. These difficulties suggest that the problems are of a syntactic nature. Tsimpli and Roussou, in their 1991 p aper, state that the parameters with the settings as specified by the L1 resist being reset, and that this is due to maturational effects. Thus parameter settings become resistant to change sometime during childhood, or around the time the L1 has been ful ly acquired. Tsimpli and Mastropavlou also argue that differences between the L1 and L2 arise from issues within the narrow syntax of the L2. Their Interpretability Hypothesis, which is seen as an updated version of the FFFH, differs in that it distingui shes interpretable and uninterpretable features. Delving into the breakdown between the Logical Form (LF) and the Phonetic Form (PF), as specified by the Minimalist Program, those features that have PF but no LF become uninterpretable to L2 speakers. Thi s would include those features not instantiated in the L1. These features are also sensitive to a critical period, like the FFFH claims, and constrained by the L1. What these hypotheses and theories have in common, is that they predict a learner will onl y be able to achieve a less profound, surface level understanding of the accessible even partially, this accessibility fades with time, and the learners are left unable to overcome the parametric obstacles set in place by their L1. L1 and L2
16 speakers then are not faced with a different process of learning the new language, but rather lose access to the superset of features available and have to contend with existing knowled ge that competes with the new input. Native like acquisition is ultimately unattainable, according to these accounts of representational deficits. In opposition to accounts of representational deficits, Full Access (FA) models posit complete accessibili ty to UG in adult second language learners. The Full Transfer/Full Access model (Schwartz and Sprouse, 1994, 1996) predicts that L2 speakers have full transfer of their L1 as the initial state of the L2, with exceptions of specific lexical items. Input f rom the L2 forces realizations of parsing failures, and the predicted access to UG then allows the learners to reset the necessary parameters to those of the L2. In this model, UG is not susceptible to maturational effects, and access does not wane. Acce ss to UG also explains why poverty of the stimulus properties, which are properties that L2 speakers acquire without explicit input, can be learned by second language learners. Learners are not said to be completely constrained by the parameters set in th e primary language, and the L1 is usually not affected by the L2. White (1985b, 1986) demonstrated both the transfer and access of FT/FA using the classic example of the null subject parameter with native Spanish (null subject) and French (non null subjec t) speaking learners of English. The studies showed that the English, with the Spanish speakers (whose L1 differed in parameter setting compared to the L2 English) perform ing worse, thus supporting the claims for full transfer and full access.
17 There are several different variants of the FT/FA model. Of interest here is the Full Access Hypothesis, which challenges the idea that full transfer of the L1 is the initial st ate of the L2 (Epstein et al., 1996; Flynn, 1996; Flynn & Martohardjono, 1994). Though this hypothesis rejects L1 transfer, it does not rule out the possibility of the L1 having effects on the second language. White (2003) insists that without L1 transfer learning their first language, though this is not implicitly stated. The role of the L1 in this account is not completely clear. The full access mentioned above st ill applies in this account, meaning that L2 speakers are still UG constrained like L1 speakers. Both the FT/FA and FA, in contrast to accounts of representational deficits, predict the ability of learners to converge on native like abilities in the secon d language. Any notable differences, or optionality, might be attributed to individual variation in speakers or general performance factors as opposed to underlying representational discrepancies. The comparison of the two sets of account is summarized here: Role of L1: Representational D eficit accounts: The L1 is an insurmountable obstacle in learning the L2. Features that are either not instantiated, or are instantiated differently in the L1 will resist resetting in the L2. Full Access accounts: T he L1 is either fully transferred as the initial state of the L2 and is gradually reset with input from the L2 (without affecting the state of the L1, or does not transfer, though interference from the L1 can be overcome. Role of UG: RDAs: No access to UG, or partial access, which has a critical period. FAs: Full access to UG throughout L2 acquisition that does not diminish with maturational effects. Ultimate attainment of L2:
18 RDAs: L2 speakers have incomplete representations and cannot achieve nat ive like ability in the second language. FAs: L2 speakers should be able to reach the same linguistic competence as native speakers. A Continuation of Rothman et al. (2010) The basis of the present study is a continuation of Rothman, Judy, Guijarro Fuentes, and Pires (2010). Rothman et al. investigated the intersection of syntax and semantics in first and second language learners of Spanish by using two off line tasks: a semantic interpretation task and a context based collocation task. 45 English speaking learners of Spanish (21 intermediate, 24 advanced) and 15 native Spanish speakers participated in the experiment. The study used a manipulation of adjectival placement in Spanish, which has subtle meaning changes when the adjective occ urs pre or post nominally. The canonical word order in Spanish is to place the adjective after the noun, unlike English. This post nominal placement tells the reader that the adjective refers to a small subset of the noun it modifies. When the adjective occurs pre nominally, the subset reading is no longer available. Pre nominally, the adjective refers to the noun as a whole, as though it is a general characteristic of the noun. The difference between the two is very fine. Note below (Figure 2 1) : 1a. Post nominal normal Las chicas bonitas The (D, pl, fem) girls (N, pl, fem) pretty (A, pl, fem) 1b. Pre nominal normal Las bonitas chicas The (D, pl, fem) pretty (A, pl, f em) girls (N, pl, fem)
19 Figure 2 1 E xperimental adjective s emantics Many adjectives in Spanish work in this way, but not a ll. Though not used in the Rothman et al. study, an additional group of special adjectives will be use d in the present study. We refer to these as meaning changing adjectives. Whereas the target adjectives have a subtle change in meaning pre and post nominally, the meaning changing adjectives have an entirely diffe rent meaning in the two adjectival placements (thus the name). We have included these in the present study because the semantic change should be much more salient to all levels of proficiency. If we find none of our anticipated results in the experimenta l items, we hope that we will see clear patterns in these meaning changers. Only a small group of adjectives share this meaning changing property. It should also be noted that this type of adjective and its semantic properties is a topic that is explicit ly discussed in second language classrooms, where the other type of adjective is not talked about. Below is an example (Figure 2 2) : 2a. Post nominal meaning changing El amigo viejo The (D, sg, masc) friend (N, sg, masc) old (A, sg, masc) 2b. Pre nominal meaning changing El viejo amigo The (D, sg, masc) old (A, sg, masc) friend (N, sg, masc) Figure 2 2 Meaning changing adjective s emantics We have given much focus to the semantic entail ments of the position of the adjective, and we should also touch on the underlying syntax. Following Bernstein (2001), Rothman proposes the Word Marker Phrase (WMP), which precedes the noun phrase, thus coming in between the determiner and the noun itself The number phrase
20 intervenes between the determiner and noun, where it also precedes the WMP. In romance languages like Spanish, the head noun obligatorily must raise to the head of the WMP, and then to the number phrase in order to check gender and nu mber features. The two word orders emerge from whether the adjective originates as an adjunct to the NP or the NumP. In interpretation 3a (see below), the pre nominal adjective reading will always emerge. In interpretation 3b though, the movement of the noun will cause the noun to move before the adjective, and result in the canonical order. Since English does not have a WMP, there is no obligatory noun raising. Thus, regardless of whether the adjective originates as an adjunct to the NP or the NumP, t he noun will never move ahead of it. An adjective occurring pre nominally has scope over the noun it modifies, thus the interpretation that the adjective is describing the noun as a whole falls out. In the post nominal position, the noun has scope over t he adjective. See example below (Figure 2 3): 3a. valientes incas 3b. incas valientes (generally brave) (as opposed to cowardly)
21 Figure 2 3. Syntax of adjectival m ovement Reprinted by permission from Rothman et. al. 2010. On the (Un) Ambiguity of Adjectival Modification in Spanish Determiner Phrases: informing debates on the mental representations of L2 syntax. (Page 54). Studies in Second Language Acquisition. In the R othman et al. experiment, participants completed two off line tasks, mentioned quickly above: a semantic interpretation task and a context based collocation task. Below are examples of both (Figure 2 4) The semantic interpretation task asks participant s to read a sentence in Spanish, and select the correct English interpretation. The collocation task asks the participants to place the adjective in question (bolded between parenthesis) either before or after the noun according to the context presented i n the surrounding sentence. A. Page 15 (7) B. Page 16 (10) Figure 2 4 Off line t asks from Rothman (2010) A) Semantic Interpretation Task. B) Context Based Collocation Task Reprinted by permission from Rothman et. al. 2010. On the (Un) Ambiguity of Adjectival Modification in Spanish Determiner Phrases: informing debates on the mental representations of L2 syntax. (A: Page 62, Example 7; B: Page 63, Example 10). Studies in Second Language Acquisition. Results from Rothman e t al. indicate that native speakers of Spanish and the advanced learners of Spanish performed equivalently on the tasks. As a whole
22 intermediate speakers did not perform natively, but on further inspection, it was found that if the group of intermediates were divided into two groups based on proficiency, the more advanced intermediate speakers did converge on native like performance, where the low level intermediate speakers did not. Rothman et al. concluded that it was possible for second language learne rs to acquire the syntactic properties of Spanish determiner phrases with the appropriate semantic fall out. This evidence does not support accounts claiming representational deficits in second language learners, since the most advanced participants did s eem to have acquired native like linguistic competence in Spanish. The main departure of the current study from Rothman et al. is the incorporation of on line testing. A strong majority of the existing research focuses on data taken from off line behavi oral experiments. T hose who do compare second language learners to native speakers with both off line and on line tests tend to use ERP techniques (Gabriele, Fiorentino, and Alemn Ba n, in press; Alemn Ban, Fior entino, and Gabriele, 2012), eye tracki ng (Grter, Le w Williams, and Fernald, 2012) or are investigating representational deficits in comparison to computational effects (Hopp, 2009) using different hypotheses, such as the Interface Hypothesis (Sorace and Filiaci, 2005). To our knowledge, lit tle work has been done to compare generative accounts to those or representational deficits in both and on line and off line fashion. This is an emerging field in the study of second language processing, and thus relatively fewer studies have been complet ed at this time. Thus, this is an appropriate area to contribute research to.
23 CHAPTER 3 THE CURRENT STUDY The current study will explore the relationship between second language learners of Spanish and native Spanish speakers in terms of their sentence level processing. We hope to contribute to the current research which attempts to learn if second language l earners can acquire new syntactic structures that will be processed like that of native speakers and have the same underlying configuration. If we find that there are differences between L1 and L2 speakers and/or differences between the outcomes of the on line and off line tasks, our results could fall under one of two possible camps of second language processing theory: representational deficit accounts or full access accounts. As previously stated, this study is meant to be a continuation of Rothman 2010 work. The current study will utilize the same adjectival manipulation mentioned above to compare the processing strategies of native speakers of Spanish and native English speaking learners of Spanish of various, higher levels of proficiency Where this study differs is that it will utilize both off line and on line tasks with a self paced reading task and two off line semantic judgment tasks. If we are able to replicate the results from the previous study, we would expect to find that nati ve speakers perform very well on both the on line and off line tasks, whereas (especially lower proficiency) learners make statistically more errors on one or both of the tasks. The heritage speakers tested here are expected to perform equivalently to the native speakers. We would expect more errors to occur on the self paced reading task than in either of the off line tasks, due to the fact that on line tasks are generally more taxing than off line tasks Our off line tests explicitly ask for a semantic judgment regarding the determiner
24 phrase, whereas the on line task asks a much more implicit question of semantic judgment via the plausibility judgment question referring to the context and target sentences. Errors on the semantic judgment task would be selecting the incorrect interpretation (of the two available), and errors on the self paced reading task would be selecting the wrong plausibility rating on the comprehension question. Additionally, on the self paced reading task, we expect to se e longer readings times for the conditions in which the adjective noun order in the target sentence does not correspond to the meaning supported by the context regardless of the adjectival placement due to the extra processing time taken once the participants real ize the mismatch between the context and the target sentence. In both the on line and off line tasks, we expect to see better scores across the three groups on items to contain the meaning changing adjectives than on the experimental items, keeping in min d that there is a more marked change semantically in those types of adjectives. If the learners do not perform natively on either the on line or off line tasks, we would attribute the problems to representational deficits. However, if the learners perfor m well on the off line or the on line task, we would find evidence in support of full access to UG. Like Rothman et al., we do expect to see results inconsistent with representational deficits. These are the highlights of our hypotheses and expected outc omes. We have laid out all our predictions in the second discussion section in Chapter 6, where we discuss each point.
25 CHAPTER 4 METHODOLOGY Participants This study recruited participants from the University of Florida community. In total, 39 subjects completed the study: 13 native speakers of Spanish, 10 heritage speakers of Spanish, and 13 native English speaking learners of Spanish. Three participants were excluded from analysis because their Spanish background did not conform to the categ ories of native/heritage/learner discussed below and had low proficiency scores with respect to the group they would have been placed into. Therefore the data analysis mentioned below will use 36 participants. These 36 participants were composed of 9 mal es/27 females, 18 25 years of age ( see Table 4 1) and had no reading or learning disabilities. Participants were compensated for their participation either monetarily at $7.50 per hour, or by receiving class credit for research hours. Compared to Rothma n et al., additional measures were used to divide the participants into proficiency levels due to the smaller number of participants. Previously, to qualify as a native speaker, a participant would need to score a 48 or better on the DELE proficiency task Scores on the proficiency task alone made for a very small group of native speakers. Though, it was clear that some speakers were in fact native speakers of Spanish even though they did not score as such on the task, due to the age at which they moved to the US from a Spanish speaking country. Therefore, we also took into consideration the length of time the participants lived in a Spanish speaking country before arriving in the US. The requirements for qualifying as a native
26 speaker of Spanish here i ncluded higher scores on the DELE proficiency task (refer to Table 4 1, ranging 34 49, average 45; excluding one participant 41 49 average 46, which is just shy of reaching the threshold for qualifying as a native speaker) and an extended amount of time li ving in a Spanish speaking country. All native speakers lived outside of the United States until at least age 6 and referred to Spanish as their first and primary language. Inadvertently, this study contained a decent sized group of heritage speakers. T he general definition for heritage speaker used here was a person who grew up speaking Spanish at home, while living in an area where the language outside the home raised in a home where a non English language is spoken, who speaks or merely understand the heritage language, and who is to some degree bilingual in English and s, 2000). Keeping in mind that this study was conducted in the state o f Florida, which has a fairly sizeable Spanish speaking population, it is not surprising that many potential subjects were raised in a heritage language environment. Since these types of Spanish speakers are qualitatively different from either native spea kers or learners of Spanish, they were placed in their own group. We categorized a speaker as a heritage speaker if they grew up in the US, and said that they learned Spanish and English both before age 5. These speakers also noted that, in general, they preferred using Spanish over English at least 25% of the time, and scored decently on the proficiency task (range 35 47, average 42, which qualifies as an advanced speaker).
27 Finally, participants were grouped as learners of Spanish if, of course, they we r e not native Spanish speakers or heritage speakers. This means that they called English their first and primary language, and did not begin learning Spanish until they were exposed to it in school or some other language environment outside the home. Due to low numbers of participants (n=13), we were not able to divide these speakers into various levels of Spanish proficiency. Instead, they were all grouped together. Their proficiency scores ranged from 27 45, with an average of 36, which is intermediat e in level Table 4 1 Participant demographics by g roup Native mean (S t andard Deviation ) Heritage (S t andard Deviation ) Learner (S t andard Deviation ) Preferred use of Spanish (% of time) 44.7 (20.6) 33.5 (10.3) 9.9 (6.4) English onset (age in years) 6.5 (3.6) 2.1 (1.4) 0 (0) Spanish onset (age in years) 0 (0) 0 (0) 10.4 (4.5) DELE proficiency scores (correct answers out of 50) 45 (3.7) 42 (3.5) 36 (6.7) Age (years) 20.8 (1.2) (19 23) 20.3 (1.8) (18 25) 20.4 (1.0) (19 22) Gender (M,F) 4,9 0,10 5,8 Stimuli/Experimental Design Targets The list of adjectives used in the target sentences in the present study was chosen for their frequency and simplicity. We wanted to ensure that all adjectives would be comprehensible to all levels of Spanish speakers, including the lower level Spanish
28 learners. Therefore, several of the target adjectives were cognates in English. The list used is as follows: aburrido apasionado aventurero barato carioso caro dbil delgado deshonesto dif cil divertido elegante enojado estudioso est pido feo fuerte gordo honesto importante influyente inteligente ocupado orgulloso pattico pesado perezoso simpatico talentoso tmido and valiente Each item, meaning each adjective listed above, of the experimental materials cont ained a 4 way set of sentences, or paradigm, which were Latin squared over four separate lists. By 4 way set of sentences, we mean that each item had a grammatical and an ungrammatical (or better stated, a plausible versus an implausible; see discussion b elow in Chapter 6) version of both the pre and post nominal adjectival positioning (refer to Table 4 2) That means that each item (for example delgado ) had the following four target sentences: a) adjective before noun, grammatical; b) adjective before n oun, ungrammatical; c) noun before adjective, grammatical; d) noun before adjective, ungrammatical (see Appendix A for full list of experimental items). Table 4 2 Conditions explained by word order in determiner p hrase Grammaticality Condition 4 5 6 Gram. (plausible/felicitous) a, F1 D Adj N Ungram. (implausible/infelicitous) b, F2 D N Adj Gram. (plausible/felicitous) c, F3 D N Adj Ungram. (implausible/infelicitous) d, F4 D Adj N Every target sentence is preceded by a context sentence. These context sentences forced either the pre or post nominal reading for the target sentence. The context sentence was the same for conditions a and b, as well as c and d, but different between those two pairings. In other words, the same context sentence pre ceded
29 conditions a and b, but a different context sentence preceded conditions c and d. Keep in mind that the only thing that differed in the target sentences was the adjective placement. What made the target sentences ungrammatical, as it was in conditi ons b and d, was that the context sentence forced a pre or post nominal reading, and the target sentence contained the opposite adjective placement. For example, condition b, which has the adjective after the noun in an ungrammatical way, would be ungramm atical because the context had a pre nominal adjective reading, but the target sentence showed a post nominal adjective placement. Below is an a example of one complete set of context and target sentences (Figure 4 1) : Context 1: Todos los estudiantes son muy listos y van mucho a la biblioteca para estudiar y aprender los textos Condition a : Por eso 1 / es 2 / que 3 / los 4 / inteligentes 5 / estudiantes 6 / sacan 7 / buenas 8 / notas 9 / en 10 / los 11 / exmenes. 12 / (all students are intelligent) Condition b : Por eso 1 / es 2 / que 3 / los 4 / estudiantes 5 / inteligentes 6 / sacan 7 / buenas 8 / notas 9 / en 10 / los 11 / exmenes. 12 / ( only the students who are intelligent ) Context 2: Las clases son muy avanzadas, entonces hay pocos estudiantes que entienden el material. Condition c : Por eso 1 / es 2 / que 3 / los 4 / estudiantes 5 / inteligentes 6 / sacan 7 / buenas 8 / notas 9 / en 10 / los 11 / exmenes. 12 / ( only the students who are intelligent ) Condition d : Por eso 1 / es 2 / que 3 / los 4 / inteligentes 5 / estudiantes 6 / sacan 7 / buenas 8 / notas 9 / en 10 / los 11 / exmenes. 12 / (all students are intelligent) Figure 4 1. Full example of an experimental i tem
30 As you can be seen, the same target sentence is used for all four conditions, with the adjective being the only alternating factor. This ensures that the context sentence is the only thing that forces the differences in meaning between conditions a/b and c/d. The targets are purposefully vague enough to follow logically from each of the two context sentences. Since there were four lists into which these items were divided and each item had four conditions mentioned above, it is easy to see that one of each condition of these items went into one of the four lists, with no same condition of any item appearing in the same list. For example each list contained one condition of all the 32 adjectives mentioned above, but none of these conditions were repeated in any list. The 32 adjectives combined with a separate set of nouns to complete a set of 32 determiner phr ases in the target sentences. These DPs, which contained a determiner, noun, and adjective, always occupied the 4 th 5 th and 6 th frame position in the target sentences. Though no adjectives were repeated over the 32 target sentences, a few of the nouns were repeated ( ni os coches estudiantes in order to ensure t hat very strong and appropriate examples of each adjective were present. Much care was given to ensure that these 4 lists were as equal and counterbalanced as possible. The lists were created by counterbalancing the items based on the following criteria : frequency of the adjective; frequency of the noun; length of the adjective; length of the noun; and length of the word immediately following the critical items (either the noun or adjective, depending on whether the adjective was pre or post nominal). Frequency data was taken from the online website Corpus Del
31 Espa ol (www.corpusdelespanol.org). We created 4 groups of the 32 experimental adjectives and nouns (meaning 8 in each group) that were not statistically equivalent on any of the 5 criteria just mentioned. For each group of the 4 groups (since 32 adjective + noun combos breakdown into 8 items in the 4 groups), one of the four conditions mentioned in the previous paragraph (a, b, c, d) went into the 4 lists. This is what was Latin squared over th e 4 lists. In other words, each list contained the 32 experimental adjective + noun combos, which was composed of 8 of each condition (a, b, c, and d), each of which come from a different group of the 4 statistically different groups. We also paired thes e lists in the Latin square design in such a way that the items containing the same noun were in the most opposing conditions as possible. The means that when coche was used twice in a list, it appeared in conditions a and d, or b and c, which had opposin g adjective placement and grammaticality. Each participant saw one of the 4 lists during the self paced reading task. All lists were checked multiple times by native Spanish speakers to ensure correct grammar, spelling, and that the manipulations were a ccurately forcing the correct readings. Meaning C hangers Within the group of fillers, we incorporated a separate group of specific adjectives. As previously mentioned, these are the meaning changing adjectives. Below is a list of the 12 we utilized. alto cierto listo pobre puro raro real seria simple solo unico viejo
32 We selected these 12 meaning changing adjectives, which are tho se that had the strongest difference in meaning pre and post nominally. Each was used twice, with different nouns, making 24 total meaning changing items. We did not feel that there was a large enough set of strong meaning changing adjectives to create a group without repeating any adjectives, so thought it best to double up on these 12 instead. These had the same 4 condition paradigm mentioned above with the target materials, with pre versus post nominal adjective placement and grammatical versus ungram matical. Just like the targets, these meaning changing adjectives were counterbalanced across the 4 lists for length of the noun and adjective within the DP and length of the word immediately following the DP, as well as the frequency of the noun and adje ctive. The groups that went into each of the 4 lists (6 in each list from the 24 total meaning changing adjectives) were found to be statistically unequal in the same aspects of frequency and word length mentioned above with the experimental items. Since we had a repeat of all the meaning changing adjectives, similar to the repeat of a few of the nouns within the target pairings, we made sure that in each of the 4 lists, the repeated adjective, which appeared 2 times in each list, appeared in opposite con ditions (a with d, or b with c). This helped to make these items as different as possible within the lists and to reduce potential effects of repetition on reading times. No nouns were repeated in the meaning changing DPs. Again, the pre nominal conditi ons (both grammatical and ungrammatical) saw the same context sentence, and the post nominal conditions saw the same context sentence, which was different from the context sentence of the pre nominal adjective + noun pairings. Essentially the target pairi ngs and the meaning
33 changing pairings were nearly identical, with the exception that there are fewer meaning changing items in total, and the meaning changing adjectives were repeated fully. So in total, each participant saw 32 target sentences, 24 mean ing changing sentences, and 72 filler sentences for a sum of 128 items. These fillers consisted of 36 plausible sentences, and 36 implausible sentences. By plausible or implausible, we mean logically possible or not, without respect to grammaticality or a ny other aspect. For example, Los das de verano son llenas del sol followed by, Es por eso que estos das son muy fros They contained no pre or post nominal adjectives. We d id not use a 4 way paradigm here. Instead each filler item was solely a context sentence and a target sentence. Procedure This study consists of the following 8 parts, in this order: Informed C onsent Language Q uestionnaire Spanish Proficiency T ask, DEL E Self Paced Reading T ask Debriefing Vocabulary T ask Semantic Judgment T ask Meaning Changers Experimentals
34 Before the experiment began, all participants gave written informed consent, as per University of Florida Institutional Board Protocol. The first task was a variant of the LEAP Q language questionnaire crea ted by Marian, Blumenfield, & Kaushanskaya (2007). The purpose of the questionnaire is to obtain as much information as possible about the linguistic background of each participant. Of particula r interest will be to ascertain: the method with which each participant learned Spanish and English, whether it be through schooling or contact with other native speakers, or any other method; when they began learning each language; the current percentage of usage of each language; when/if they moved to the US from a Spanish speaking country; and general demographic questions such as age, gender, race, etc. Second was a Spanish proficie ncy task. The task used here was the DELE Sp anish proficiency task. This was a 50 question test. The first 30 questions ga ve a sentence with a blank towards the end of the sentence, and 4 possible responses to choose from. The final 20 questions were in the format of a cloze test. The participants read a paragraph, fill ed with 20 blanks, and had 3 options per blank to choose from to complete the sentences. Previously, the level of Spanish was determined as follows: 50 48 corrects answers = native speaker; 47 40 = advanced speaker; 39 30 = intermediate speaker, <29 = be ginner. Question #9 on the cloze test presented a problem. We noticed that the participants consistently chose the incorrect response. After reviewing the test, we decided that we would accept 2 possible responses for that question, meaning anyone that had answered the correct response or the frequently chosen response received credit for that question. As mentioned above, these scores
35 contributed in large part to whether the participant was categorized as native, heritage, or learner, though it was not the only factor. The third, and arguably most demanding task was the self paced reading task, which we have explained above. This task was completed on a computer in a separate room in the lab. Each participant was guided to this room by our native Sp anish speaking assistant, where the participant was seated in front of a computer. The assistant explained the directions to the participant in Spanish, the purpose of which was to put the participants in a Spanish speaking mindset before beginning the t ask. The instructions were also presented on the screen, in Spanish. Once the instructions were explained, the participants went through 8 practice questions while the assistant remained in the room. These practice questions were identical in form to th e fillers, and did not contain any pre or post nominal adjectives. There were 4 plausible and 4 implausible questions. This allowed the participants to ask questions before they continued on to the actual test questions. The experiment itself consisted of 2 blocks of 64 items (context sentence, target sentence, plausibility question), with a short break in between. Since this was a self paced reading task, the context sentence appeared on screen as a whole, while the target sentence appeared word by word as controlled by the participant. The t arget sentence initially appeared as a set of blanks, and with each button click, a new word in the sentence appears while the previous word became a blank line again. After the final word o f the target sentence d isappeared, the participant wa s presented with a plausibility judgment question. They had to determine how plausible the second sentence (target) was with respect to the first sentence (context).
36 The options were: 1 completely plausible; 2 plausible; 4 implausible; 5 completely implausible. We removed the neutral option of 3 to force the participants into making a decision on plausibility in one of the 2 directions. The purpose of these questions was to keep the participants on task, and to test if the y were making correct judgments with respect to grammaticality and adjective placement. Directly after the self paced reading task, the participants completed a debriefing form. This questionnaire allowed the subjects to voice any opinion they had about the task or problems they experienced with the preceding task. This included questions about the font size, length of task, ease of task, etc. Answers to these questions were initially important to resolve any outstanding problems with the task that the researchers may have not noticed or overlooked. Ultimately, the responses were informative, but the task itself did not undergo any changes due to responses. Each participant then completed a brief translation task. This task contained all the adjective s (32) and nouns (29) that were used in the target items of the self paced reading task. All were listed in Spanish, and the participant was asked to provide the English translation. We wished to check that the participants knew the meanings of the targe ts to ensure that any problems that arose in the self paced reading task were not due to lack of knowledge of Spanish. Some of the native Spanish speakers were not able to provide English translations, due to lack of knowledge of the English term, and not a lack of Spanish knowledge. This is something to take into consideration when looking at the scores on this task. The final two tasks were semantic judgment tasks. These tests were based off the Semantic Interpretation task used in Rothman (2010) (See Figure 4 2) Each
37 question showed a sentence, with the DP underlined, and two options for the interpretation of the underlined portion. The structure was the same as in the Rothman study, but new sentences and options for interpretation were created for this study. See below: 1. Mi tio es un pobre hombre Figure 4 2 Semantic interpretation t ask The first of the two semantic judgment tasks tested the meaning changing adjectives and the second tested the target adjectives. In the shorter, meaning changing adjective test with 13 questions, the two options for interpretation were alternated as to wh ether the pre nominal or post nominal interpretation appeared in the first column so as to prevent any correlation between the placement of the adjective in the sentence and the column. For the target adjective test containing 32 questions, the first colu mn consistently had a pre nominal or kind denoting reading, while the second column had a post nominal, set denoting reading. The sentences themselves, in both tasks, had alternating pre and post nominal DPs in no specific order. The sentences in this ta sk were not the same as those used in the self paced reading task. My uncle is unfortunate. The man is not wealthy.
38 CHAPTER 5 RESULTS Off L ine We completed one way ANOVAs to test the means of the three groups in question (natives, heritage, and learners) with respect to their performance on the vocabulary test, the semantic judgment task on meaning changing adjectives, and the semantic judgment task on the experimental adjectives. No significant differences were found among the groups on the three measures: Vocabulary: ( F ( 2, 33) = .474, p = 627); errors in meaning changers semantic task: ( F ( 2, 33) = .833, p = .444); errors in experimentals semantic task: ( F ( 2, 33) = .925, p = .406). Table 5 1. Off line task scores, by g roup Group Vocabulary errors, out of 57 (Standard Deviation ) Meaning Changers errors, out of 13 (Standard Deviation ) Experimentals errors, out of 32 (Standard Deviation ) Natives 2.7 (3.6) 3.0 (0.9) 15.5 (4.1) Heritage 2.0 (1.6) 3.7 (1.4) 15.9 (4.0) Learners 3.4 (3.8) 3.3 (1.4) 12.5 (9.1) On L ine The following statistical analyses, as well as the analyses for the offline data, were completed using version 20 of the SPSS software. In order to be considered statistically significant, we restricted our analyses to only results that were below < .05 We tested both the experimental items and the meaning changers. Our three areas of interest were with the accuracy of the plausibility question, the reaction time of the plausibility question, and the reaction times at word positions 4 9. For all repe ated measure ANOVA tests, we used the univariate analyses.
39 Accuracy Experimentals We completed a repeated measures ANOVA with the within subjects factors of Grammaticality (grammatical a/c or ungrammatical b/d) and Adjective Placement (pre nominally a/d, or post nominally c/b), and a between subjects factor of Group for the experimental conditions to analyze the accuracy of responses to the end of sentence plausibility judgment task. Participants had four options to choose from (1, 2, 4, 5). For gra mmatical target sentences, answers were counted as accurate if they chose either 1 or 2, and inaccurate if they chose either 4 or 5. The opposite was true for ungrammatical target sentences. Mean and standard deviation of the accuracy of the experimental m aterials can be found in Table 5 2 We found a significant effect of Gr ammaticality (F(1, 33) = 878.34 p < .001) and a significant interaction between Grammatica lity and Group (F(2, 33) = 4.53 p = .018). For the effect of Grammaticality, we can see by looking at the means below that the accuracy for the grammatical conditions (a and c) was much higher than those of the ungrammatical conditions (b and d). We then looked at the interaction effect by singling out each group to compare the grammatical versus ungrammatical conditions. All three groups showed significant differences between these conditions: Group 1 natives: (F (1, 12) = 579.62 p < .001); Group 2 heritage: (F (1, 9) = 157.68 p < .001); Group 3, learners: (F (1, 12) = 288.66 p < 001). In all groups, the grammatical conditions were much more accurate compared to the ungrammatical conditions. Table 5 3 below shows percentage accuracy by group, which includes both the experimentals
40 and meaning changers. We did further T tests to l earn if any specific group had a statistically smaller difference between the grammatical and ungrammatical conditions. All effects were insignificant except the differences between the pre nominal conditions (a and d) between groups 1 and 3 (the natives and learners) (t (24) = 3.75 p = .001) Given that every question had two correct possible answers, one being plausible or implausible, and the other being very plausible or very implausible, we conducted a second repeated measure ANOVA to test whether the participants showed any difference in a ccuracy between the extreme responses (1 and 5) compared to the non extreme responses (2 and 4). We were curious to see if the accuracy rates went up at all for the items where the participants responded with a very ( im)plausible answer. The within subj ect factors were plausibility (plausible or implausible) and extremity (1/5 or 2/4), and the between subjects factor used was Group. No significant effects or interactions were found. Meaning changers We completed a repeated measures ANOVA with the wi thin subjects factors of Grammaticality (grammatical F1/F3 or ungrammatical F2/F4) and Adjective Placement (pre/post nominally, F1/F4, F3/F2) and a between subjects factor of Group for the meaning changers. Table 5 4 below shows the breakdown of the accur acy means per condition. This showed a significant effect of G rammaticality (F(1, 33) = 54.50 p < .001), but no other effects or interactions were found. Just as with the experimental items, the grammatical conditions were consistently more accurate tha n their ungrammatical counterparts. No further tests were conducted. To compare the
41 accuracy of the experimental items with the meaning changers, Figure 5 1 is presented below. The previous analyses were collapsed over items, which is commonly referre d to as a by subject (F1) analysis. We also attempted to complete an F2 analysis, which would be by item, collapsing over participants. The F2 analysis was not possible with either the experimentals or meaning changers, since data was not available for e very condition in each group due to lack of having at least one member from each group complete each of the four lists. Table 5 2 Accuracy on plausibility questions, by condition, e xperimentals conditions Mean Standard Deviation a: gram, AN .830 .144 b: ungram, AN .101 .107 c: gram, NA .826 .135 d: ungram, NA .174 .128 Table 5 3 Accuracy on plausibility questions, by g roup Group a b c d F1 F2 F3 F4 1 native .89 .07 .84 .13 .71 .33 .77 .27 2 heritage .81 .11 .85 .15 .62 .33 .60 .42 3 learner .78 .13 .80 .24 .60 .33 .67 .38 Table 5 4 Accuracy on plausibility questions, by condition, meaning c hangers conditions Mean Standard Deviation F1: gram, AN .644 .215 F2: ungram, AN .333 .207 F3: gram, NA .685 .245 F4: ungram, NA .352 .168
42 Figure 5 1 Accuracy of p lausibil ity questions, comparing meaning changers and e xperimentals 0 0.2 0.4 0.6 0.8 1 Experimentals Meaning Changers proportion correct Accuracy of Plausibility Questions a/F1 b/F2 c/F3 d/F4
43 Reaction Time Questions Experimentals Reaction time data for the plausibility judgme nt questions is shown in Figure 5 2 an d Figure 5 5 below. Reaction times were analyzed only for the questions that the participants responded to correctly. This meant that, for certain subjects, no data was available for analysis in certain conditions, particularly in the ungrammatical conditions (b and d). We also ran an Excel macro to remove any data that was more than two standard deviations greater than the overall mean. Above we saw that only approximately 10% of the responses to the ungrammatical questions were accurate. Thus, we must keep in mi nd that the number of analyzable trials is greatly reduced, which in turn reduces the power of the statistics. We conducted a repeated measures ANOVA, using Adjective Placement and Grammaticality as the within subjects factors (two levels in each), and th e between subjects factor of Group. No significant effects or interactions were found. Meaning changers Reaction time data for the plausibility judgment task was also analyzed only for those questions with correct answers in the meaning changers. Again, limited data was available for analysis in certain conditions, such as the ungrammatical conditions (F2 and F4). We conducted a repeated measures ANOVA, using Adjective Placement and Grammaticality as the within subjects factors (two levels in each), and the between subjects factor of Group. Here we found no significant effects, though we found a
44 significant interaction between Adjective Placement and Grammaticality (F(1, 26) = 6.02 p = .021). Further paired t tests showed a significant difference betw een conditions F1 (adjective first, grammatical) and F4 (adjective first, ungrammatical) (t (32) = 2.43 p = .021). By looking at the reaction time means in the Figure 5 3 below, we can see that the F4 condition was much longer than the F1 condition. Th is corresponds to low accuracy in the F4 condition, which is not indicative of a speed accuracy tradeoff. Normally, we might expect that the conditions that have the least accurate results are also the ones with the shortest reaction times, but that is no t what we find here. Just as with the accuracy analyses, we were only able to complete an F1 analysis on subjects, as opposed to an F2 analysis on each item we tested. Many items did not have any data to analyze. Table 5 5 Response time to plausibil ity questions, by condition, e xperimentals conditions Mean (ms) Standard Deviation (ms) a: gram, AN 1667 883 b: ungram, AN 2180 1522 c: gram, NA 1934 1110 d: ungram, NA 2697 2176 Table 5 6 Response time to plausibility questions, by condition, meaning c hangers conditions Mean (ms) Standard Deviation (ms) F1 1942 1440 F2 2318 1427 F3 2297 1695 F4 3135 2769 Table 5 7 Mean response time to plausibility questions, by group, meaning changers and e xperimentals (in ms.) Group a b c d F1 F2 F3 F4 1 1603 1541 1680 3326 1355 2832 2436 3860 2 1616 2586 2032 2710 1825 1858 2042 2949 3 1769 2354 2113 2165 2611 2091 2354 2564
45 Figure 5 2 Average reaction time to end of sentence plausibility question, by condition, e xperimentals Figure 5 3 Average reaction time to end of sentence plausibility q uesti on, by condition, meaning c hangers Sentence Level Reaction Times, Word Positions 4 9 Experimentals 0 1000 2000 3000 d c b a Time (ms) conditions Average Question RT: Experimentals Mean (ms) 0 1000 2000 3000 4000 F4 F3 F2 F1 Time (ms) conditions Average Question RT: Meaning Changers Mean (ms)
46 Word positions 4 9 in the self paced reading sentence were examined for latency of the response for each position. The same cut off procedure (Excel macro) was used for this data as was used previously with the reaction time to the end of sentence questions. Word position #4 was always the determiner, #5 and #6 were either the target noun o r adjective, and positions #7, #8, and #9 were the three words directly following the determiner phrase. Analysis was not conducted beyond word position #9 as we did not expect spillover effects beyond this point. Also, some of our target sentences ended at word position 10, so we also chose word position 9 as our cutoff in order to avoid confounds due to end of sentence wrap up effects. We conducted repeated measures ANOVAs on the conditions that had the same adjective placement (a vs. d, b vs. c), so t hat we would be comparing reading times on the same part of speech. Otherwise, we would essentially be comparing apples and oranges by looking at reading times of adjectives versus nouns, for example when comparing condition a and b, which have alternatin g nouns and adjectives in word positions #5 and #6. For conditions a and d, we conducted a repeated measures ANOVA, using the within subject factors of Grammaticality (two levels) and Word Position (six levels) and the between subject factor of Group. No effect of group was found. We did find a significant effect of W ord Position (F (5, 29) = 60.64 p < .001) and an interaction between Grammaticality and Word Position (F (5, 29) = 7.61 p < .001). Further t tests showed longer response times for the ung rammatical condition (d) at word position #4 ( the determiner, (t (35) = 2.55 p = .015)), and for the grammatical conditions (a) at word positio n #6 (the noun, (t (35) = 2.26 p = .030)) and word position #8 (two places
47 after the determin er phrase ended, (t (35) = 2.66 p = .012)). See Figures 5 4 and 5 8 through 5 11 below. For conditions c and b, we found significant effects of Word Positio n (F(5, 29) = 71.49 p < .001). No other interactions were found. See Figures 5 5 and 5 8 through 5 11. Meani ng changers The same tests were conducted for the meaning changers as for the experimentals. We compared conditions F1 and F4 first (adjective first conditions). Another repeated measures ANOVA was carried out, using the within subject factors of Gramm aticality (2 levels) and Word Position (6 levels), and the between subject factor of Group. No effect of Group was found. We found a significant effect of W ord Position (F (5, 29) = 32.74 p < .001) and a significant interaction between Grammaticality an d Word Position (F (5, 29) = 3.19 p = .016). Further t tests showed longer response times for the ungrammatical condition (F4) at word position #8 (two places after the determine r phrase ended, (t (35) = 2.76 p = .009)), and for the grammatical conditi on (F1) at word positi on #5 (the noun, (t (35) = 2.36 p = .024) and #7 (the word immediately f ollowing the DP, (t (35) = 2.12 p = .041)). See Figures 5 6 and 5 12 through 5 15. Additional investigation Although no significant results were found in the experimentals or meaning changers with respect to the factor Group, we decided to take a look at each of our groups individually for both the experimental items and the meaning changers to see if any patter ns emerged for the sake of future research or if this study were to be repeated
48 (with changes, of course), even though they would not be statistically significant. Below are the graphical representations of the reaction time data, by group, for word posit ions 4 9 in the experimental items (refer to Figures 5 4, 5 5, and 5 8 through 5 11). We can see in the experimentals that the learners and heritage speakers do show differences in conditions a and d, compared to the natives, immediately after viewing the adjective in its non canonical order. Since the adjective occurs in position #5, we see spill over effects at position #6. This effect is delayed in the native speakers it seems, until the 7 th word position. We do see an effect of Adjective Placement i n word #6 for all groups, but as we noted above, we are comparing different word categories, so this difference does not hold much significance. Conditions b and c differ very slightly in any of the groups, at any word position. This is another indicatio n of the lack of perception of grammaticality. Though we do see effects of grammaticality that are significant at word position #7 for the natives, and at word position #6 for the heritage speakers. What is interesting about both of these effects is that the grammatical condition (condition a) of the pre nominal adjective placement has longer reading times than the ungrammatical condition (condition d). We would expect the ungrammatical condition to show increased length of reading time, but that is not what we see. We double and tripled checked the coding of these conditions to be sure this was not a coding error. The only difference in the presentation of these materials is the context sentence, which intends to force a pre or post nominal reading. T his causes no effects in conditions b and c, so it is puzzling as to why we do see effects in a and d. There does not seem to be a logical reason for this effect, and we are open to suggestions that may explain it.
49 Finally we looked at the differences in groups for the meaning changers. Below (in Figures 5 6, 5 7, and 5 12 through 5 15) we see that the peak reaction times occur consistently at word position #7. This effect occurs directly after the DP, similarly to what we saw in the natives for the e xperimental items. We also consistently see that the AN ordered conditions have longer reading times than the NA ordered conditions. The effects of Adjective Placement increase in significance as the proficiency level of the group decreases. We see that there is almost an effect of Adjective Placement in the native speakers (F (1,12) = 4.75 p = .054), there is an effect for the heritage speakers (F (1,9) = 7.23 p = .025), and there is a strong effect fo r the learners (F (1,12) = 38.98 p < .001). This seems to hint at the idea that the more proficient speakers are less affected by the non canonical word ordering than the learners. This agrees, to some extent, with the results we saw with the experimental items. The less proficient speakers, namely th e heritage speakers and the learners, showed a more immediate reaction to the non canonical word ordering, and thus we see longer reading times at word position #6, instead of #7. The natives, though, have much shorter reading times at position #6, likely due to the fact that they are not as affected by the appearance of pre nominal adjectives. There is no effect of grammaticality be tween conditions F1 and F4 in any of the groups at word position #7. So even if the natives may have an easier time processi ng both of the word orders with the meaning changers, no group is sensitive to the manipulations of grammaticality. In word position #8, though, which is two positions after the end of the DP, we see that the native speakers show that the grammatical posi tion, F1, has a shorter reading time than F4, though this is not found to be
50 statistically significant. We were hoping for this effect to be significant across the board, but unfortunately this is the only place we see any sign of it. Between conditions F3 and F2, a repeated measures ANOVA found significant effects of Word Position (F (5, 29) = 25.628, p < .001), but no other interactions. See Figures 5 7 and 5 12 through 5 15. Figure 5 4 Reaction times on word positions 4 9, conditions a and d, e xperimentals 0 100 200 300 400 500 600 4 5 6 7 8 9 time (ms.) word position 4 9 Experimentals: a vs. d, Pre adjectival a d D ADJ N
51 Figure 5 5 Reaction times on word positions 4 9, conditions c and b, e xperimentals Figure 5 6. Reaction times on word positions 4 9, conditions F1 and F4, meaning c hangers 0 100 200 300 400 500 600 4 5 6 7 8 9 time (ms.) word position 4 9 Experimentals: c vs. b, Post adjectival b c D N ADJ 0 100 200 300 400 500 600 4 5 6 7 8 9 time (ms.) word position 4 9 Meaning Changers: F1 vs. F4, Pre adjectival F1 F4 D ADJ N
52 Figure 5 7. Reaction times on word positions 4 9, conditions F3 and F2, meaning c hangers Figure 5 8. Reaction times for experimental conditions by word p osition o verall 0 100 200 300 400 500 600 4 5 6 7 8 9 time (ms.) word position 4 9 Meaning Changers: F3 vs. F2, Post adjectival F2 F3 D N ADJ .00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 4 5 6 7 8 9 time in ms. word position Overall Reaction Times, Experimentals a b c d D ADJ/N N/ADJ
53 Figure 5 9. Reaction times for group 1 by word position, natives, e xperimentals Figure 5 10. Reaction times for group 2 by word p osition, heritage, e xperimentals .00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 4 5 6 7 8 9 time in ms. word position Reaction Times for Natives: Experimentals a b c d D A/N N/A .00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 4 5 6 7 8 9 time in ms. word position Reaction Times for Heritage: Experimentals a b c d D A/N N/A
54 Figure 5 11. Reaction times for group 3 by word position, learners, e xperimentals Figure 5 12. Reaction times for meaning c h anger conditions by word position, o verall .00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 4 5 6 7 8 9 time in ms. word position Reaction Times for Learners: Experimentals a b c d D A/N N/A .00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 4 5 6 7 8 9 time in ms. word position Overall Reaction Times, Meaning Changers a b c d D A/N N/A
55 Figure 5 13 Reaction times for group 1 by word position, natives, meaning c hangers Figure 5 14 Reaction times for group 2 by word position, heritage, meaning c hangers .00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 4 5 6 7 8 9 time in ms. word position Reaction Times for Natives: Meaning Changers F1 F2 F3 F4 D A/N N/A .00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 4 5 6 7 8 9 time in ms. word position Reaction Time for Heritage: Meaning Changers F1 F2 F3 F4 D A/N N/A
56 Figure 5 15 Reaction t imes for group 3 by word position, learners, m eaning c hangers Only accurate trials After we completed the analyses for the word level reaction times, we also decided to restrict the data to only the reactions times for items that contained correct responses on the plausibility questions (See Figure 5 16 below). Perhaps after restricting the data of interest to only the accurate responses, we will see more clear patterns emerge. We collapsed all the data over groups. This was done in place of restricting analyses to those participants who performed well on the more difficult, ungrammatic al conditions, because even the best performers scored below chance in their averages. We found that, even with only the data on the accurate responses, there were only significant effects of Word Position for the data set contai ning a and d (F (5, 24) = 14.17 p < .0 01), c and b (F (5, 15) = 21.33 p < .00 1), F1 and F4 (F (5, 28) = 21.67 p < .001), and F3 v. F2 (F (5, 26) = 16.39 p < .001). No significant interactions .00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 4 5 6 7 8 9 time in ms. word position Reaction Time for Learners: Meaning Changers F1 F2 F3 F4 D A/N N/A
57 Figure 5 17 as well to view differences between only the accurate trials and all trials. Figure 5 16 Reaction times for only accurate trials by word position, e xperimentals 0 100 200 300 400 500 600 700 4 5 6 7 8 9 time in ms. word position All Accurate Trials a b c d D A/N N/A
58 Figure 5 17 Reaction times for all t rials by word position, e xperimentals Top performers Finally, we attempted to isolate some of the top performers on the accuracy questions to further analyze the data and to learn if the better performers presented a different pattern on their reaction times. We created a group of the top 20, and the top 10 performers. Unfortunately, the overall accuracy scores of even the best performers were barely above chance, with an average of .59 overall. We did not feel such low scores warranted further analysis. 0 100 200 300 400 500 600 700 4 5 6 7 8 9 time in ms. word position All Trials a b c d D A/N N/A
59 CHAPTER 6 DISCUSSION Confounds To begin, we wish to acknowledge the confusion that might be caused by the terms grammatical and ungrammatical. Technically, these terms are not accurate, as it is not illegal grammatically to have an adjective occur in either the pre or post nominal posi tion regardless of the preceding context. Better words might be felicitous and infelicitous, or plausible and implausible. We did not change the terminology throughout this paper only because grammatical and ungrammatical are somewhat easier to understan d, and these terms have been used throughout the experiment and analysis, but it should ne noted that they are not entirely accurate terms. With that in mind, it has also been pointed out that the difference between the two forcing contexts, that of a ge neral characteristic of the noun in question and that of a specific group within the larger group of the noun, is not quite so straightforward. It is necessarily the case that the domain of a subset (the post nominal adjectival interpretation) falls withi n that of the superset of the noun in question. Therefore, in the self paced reading portion of the experiment, the ungrammatical NA condition (d) is not exactly ungrammatical or implausible. It is possible for a subset of a noun to fit into a context t hat intends to describe the entire population of the noun. For example, if we use the DP las chicas/bonitas, nominal reading would imply that all girls are pretty, and the post nominal reading would imply that only a certain sm aller group of the girls are pretty. If the context sentence sets a scene saying that all girls are pretty, that being the pre nominal reading, it is still true that a post nominal
60 adjective placement in the target sentence remains within that superset of girls. In other words, if all girls are pretty, then necessarily a small subset of girls, as noted in the post nominal target sentence, is within that overall group of girls, and thus are also consisten t with the pre nominal context. In these conditions, our intention was to very specifically say that a part of the population had the property of the adjective (such as strong or pretty) while the rest of the population did not have said property, but it is logical to assume that some pa rticipants may have thought it plausible (which is what the end of sentence question asked) that this subset fit into the general superset of the noun. This may be a source for the inconclusive nature of the results found in the on line task. Another bia s may come from the practice questions we used. These eight practice questions used very blatant examples of a context that either agreed or disagreed with the moving window target sentence (discussed more below in Accuracy section of discussion). If we were to repeat this study, perhaps a better option would be to instead take the second task in the Rothman et al. study (2010), the context based collocation task, and make it into an on line, computer based task. This would be directly comparable to the previous study, and would possibly avoid many of the confounds we discussed here with the self paced reading task. This study does also acknowledge that the population of participants used here is not ideal. Though extensive methods of recruiting were use d, a large percentage of interested participants were heritage speakers, and with the time given, we were not able to get a large enough group of willing Spanish learners or native speakers. If this study were to be repeated or continued, we would look to increase the number of
61 natives to at least 15 participants and learners to a group of 30+, with a clear division between advanced and intermediate speakers as was seen in Rothman (2010). It is disconcerting that we do not see more clear results with th e meaning changers. They were included in this experiment essentially as a control. We expected to see more robust results, i.e. a more dramatic disparity between grammatical and ungrammatical conditions, with the meaning changers than with the experimen tal items. As we will discuss in further detail below, this is not what we found. The semantic differences between the two adjectival positions in the experimental items is subtle, but the semantic differences between the two adjectival positions with th e meaning changers is not. The fact that even the native speakers were not picking up on this very salient difference points less to faulty design and more to a participant pool that may not be truly native in their linguistic processing. We pointed abov e to a less than ideal population in this study, and this bit of evidence supports that logic. Predictions Accuracy on end of sentence plausibility judgment questions: We expected to find that natives and heritage speakers (groups 1 and 2) would perf orm with greater accuracy than (especially lower level) learners (group 3). We expected to find greater accuracy on meaning changer conditions (F1 F4) than experimental conditions (a d) in all groups. Reaction times: End of sentence plausibility judgm ent questions: We expect ed learners to respond slower to the end of sentence judgment task than natives or heritage speakers. Word Positions 4 9: We expect to see longer reading times in general, across groups, in conditions that are ungrammatical (b and d, F2 and F4) regardless of adjectival placement. These results are expected to be particularly robust in the
62 ungrammatical conditions (F1 and F4). Longer reading times are expected to occur on or soon after, in terms of word position, the ungrammati cal word (either the noun or adjective) is read. On line versus off line tasks: We expect ed natives and heritage speakers to perform well on both the on line and off line tasks. If learners perform equivalently to the natives on either of the tasks, we expect it to be on the off line tasks. On the on line t ask, we did not expect learners to pe rform natively on experimental items, though we did expect them to perform well on the meaning changers. On the off line tasks, we expect ed to find that learners could perform equivalently to natives, as seen in Rothman (2010). Theoretical implications : We expect ed to find evidence in support of Full Access accounts, and against accounts of representational deficits, in line with the findings of Rothman (2010) meaning that the learners would be able to converge on native like linguistic competence on one or both types of tasks. Accuracy There seemed to be a clear pattern in the outcome of the plausibility questions, as far as accuracy is concerned. For both the experimentals and the meaning changers, we found that the participants performed much wo rse on the ungrammatical items (b/d, F2/F4) compared to their grammatical counterparts in both adjectival positions. The participants were not aware of the way in which the context sentences were affecting the interpretation of the target sentences. It would seem that, for the experimental conditions at least, the differences in pre and post nominal contexts were not enough to force the participants into interpreting their relation with the target sentences as implausible. Perhaps overall, the manipulat ion was too subtle, and the
63 participants just assumed that the majority of the experimental sentences were plausible, regardless of the actual grammaticality. Therefore, it would appear that they perform very well on the sentences that were actually plaus ible, and very poorly on the sentences that were not plausible. Bearing in mind the possible confounds mentioned earlier (subset versus superset, inappropriate participant pool, etc.) which li kely have also contributed to ge neral confusion. For the meani ng changers, we can see that the participants showed improvement in the ungrammatical conditions (compared to the experimentals), and at the same time performed worse on the grammatical ones. Both scores came closer to converging on scores that were only at chance, or scores that both neared 50% accuracy. We would expect that with the meaning changers, which have a greater semantic difference between th e two adjectival positions, the participants would score higher/more accurately than with the experiment als, which are subtler in their change in meaning. This showed to be true with the off line tasks, where all groups tested very well. We do see this with the ungrammatical conditions, which were responded to more accurately than the experimentals, though still below chance. It is a bit more puzzling as to why the accuracy on the meaning changers drops with the grammatical conditions (compared to the experimentals), because we might expect accuracy on these conditions to remain high. If the participants were in a sort of auto pilot mode with the experimentals, and just answering plausible to all of them, perhaps when they encountered the meaning changers and were more aware of the change in grammaticality between the pre and post nominal conditions, we wo uld expect scores to
64 drop accordingly on the grammaticals. They no longer assumed that the majority of responses are automatically plausible. were looking for a more dras tic disparity between the context and its target to qualify it as implausible, and therefore they responded overwhelmingly that the pairings were plausible. We could partially attribute this to the subtlety of the manipulation, and partially to the set of practice questions and fillers that were used. In the 8 practice questions, we designed the sentences to be very clear examples of plausible versus implausible relations between the context and target sentence. For example (Figure 6 1) : Context: Los das de verano siempre estn llenos de sol. Summer days are always full of sun. Target: Los das de verano son muy fros. Summer days are very cold. Figure 6 1 Practice q uestion We think it is possible that, with examples like this, the participants may have been expecting great discrepancies, which were not found in our experimental materials. The solution to this problem is a difficult one, since we would need to design practic e sentences that do not directly bring attention to the adjectival manipulation in question, but that do force the participants to pay attention to finer instances of syntactic differences. Since no effects or interactions with Group were found, we can sa y that this generalization extends to all levels of Spanish speakers tested here. Finally, with the accuracy on the plausibility questions, we found that the experimentals had an interaction between Grammaticality and Group. T tests showed that all of the 3 groups in question had very significant differences between the
65 grammatical and ungrammatical conditions. Therefore, our conclusions made here regarding the patterns of accuracy on the plausibility questions extend to not one group in particular, bu t rather all 3 groups. In all, with our expectations in mind, we did not find that the natives or heritage speakers differed from learners, nor did we find higher accuracy rates on the meaning changing items compared to the experimental items. Reaction Time Questions Data on the reaction times to the plausibility questions resulted in largely insignificant findings. In both the experimental items and the meaning changers, no significant effects were found. We did not find any effects or interactions related to Groups either, which would indicate that our natives, heritage speakers, and learners showed no statistical differences. We found a significant interaction between Adjective Placement and Grammaticality in only the meaning changers, which show ed that response times in condition F4 (ungrammatical, NA, average 3135 ms.) were significantly longer than F1 (grammatical, AN, average 1942 ms.). This finding could be due to the fact that this condition had non canonical ordering and was also ungrammat ical. Though, as mentioned in the previous section, it is unclear as to whether the participants picked up on the manipulation of ungrammaticality. We do see, numerically at least, that in both the meaning changers and the experimentals, the reactions ti mes were shorter for the grammatical conditions compared to the ungrammatical conditions. Though, the differences are not statistically significant since we do not see effects of grammaticality. We must also bear in mind again that the
66 number of trials t hat comprised the ungrammatical averages was greatly reduced compared to the grammatical averages, since we removed all trials with incorrect responses. This has the ability to skew our findings, and we do not believe that these results have a large impac t on our conclusions. When looking at the relations between accuracy and response times, we might expect to see if there wa s speed/accuracy trade off, that is, more errors occurring with shorter response times What we find is exactly the opposite. The d ata show that in both the experimentals and the meaning changers, the longest response times occurred in the conditions with the least accuracy ( see Figures 5 1, 5 2, and 5 3). The F4 condition (ungrammatical, NA), for example, had the longest average rea ction time at just over 3 seconds (3135 ms.), and an accuracy percentage of just 35%. This pattern would indicate processing difficulties, which would be in line with what we would hope to see for a condition like F4, which shows an ungrammaticality betwe en the context and the target sentence, as well as an effect of canonicity, since the ungrammatical version of an NA condition is actually pre adjectival, AN. We see this same pattern with condition d (ungrammatical, NA) in the experimentals, showing a co nsistency over all the materials. Checking back in with our expectations, we did not find that learners had slower reaction times to the plausibility questions than either the natives or the heritage speakers. All groups were statistically equal. Self Pac ed Reading Word Positions 4 9 We found significant effects of Word Position for all 4 adjectival based pairings. This may not be surprising, as we would expect to find variance at each word position,
67 considering the differences in word category or word l ength, among other factors. What was more interesting was that we found significant interactions between Grammaticality and Word Position for both of the pairings with the non canonical word ordering (a vs. d, F1 vs. F4), yet not for the pairings with the canonical NA ordering. T tests, though, do not show a consistent pattern as to whether the grammatical or ungrammatical condition was longer, nor do we see a pattern as to which word positions have statistical differences in reaction times between the ex perimentals and meaning changers. The only logical conclusions that we can draw from this outcome is that the participants are not aware of the grammaticality status between the pre nominal adjective conditions. We assume that if they were aware of the d ifferences in grammaticality, one condition would have consistently longer reaction times across the word positions (presumably longer for the ungrammatical conditions). We might also expect to see the same word positions affected in both the experimental s and the meaning changers. Our data shows neither of these anticipated outcomes. Therefore we can only conclude that grammaticality effects are not present. Since we do, however, have consistent interactions with only one of our two options of adjectiv al placement, that being the non sensitivity to the non canonical word ordering, which is present regardless of the grammaticality. This would indicate that grammaticality, or felicitousness, is not being processed accurately, and that the participants are not successfully integrating the context with the target sentence. This reigns true for all groups tested here. Overall, our expectations were not borne out. We had hoped to see longer reading times for ungrammatical conditions, but we instead found longer reading times
68 for non canonical word orders. We also were unable to find more robust findings for meaning changers, as both types of adjectives had similar patterns. On line versus o ff line Off line tests showed that no group performed statistically better than any other group in either of the semantic judgment tasks. We could argue that these findings are in line with the results from Rothman (2010) to a point, since the learners here are able to achieve native like, which we have seen might be questionable). In this study, however, all learners tested, including all intermediate and advanced speakers, performed like natives. Rothman (2010) found a clear divide between high and low level intermediates, and was able to show the point in learning at which learners acquired the syntactic processes underlying the adjectival movement. This was not demonstrable in thi s study (though we also did not have enough intermediate level participants to be able to divide them into high and low level intermediates). As we saw above, effects of Group were not present for the on line task either. Natives and heritage speakers s howed no difference from learners in any area that we tested statistically with regards to the self paced reading task. Our predictions were that learners would converge on native like processing in the off line tasks, but not in the on line tasks. In fa ct, learners scored natively (or equivalent to our group of natives) on both off line and on line testing, but we take this result with a grain of salt knowing that neither group seemed to perceive the adjectival manipulations we attempted to investigate.
69 Theoretical i mplications We found no evidence to support accounts of representational deficits from the results here. Learners did not exhibit statistically unequal results compared to natives or heritage speakers. If we were to accept the representational deficits accounts with the existing results, we would essentially have to conclude that both natives and learners processed these materials in a shallow manner. This undermines hypotheses like Shallow Structures, which state that only the learners have less profound underlying representations as compared to natives. Stating that natives also process these structures in a shallow manner removes the basis from which we are able to declare what is shallow and what is deep processing. Though we believe that conditions are not ide al and that our claims are not irrefutable, we could only conclude from the evidence here that our learners show more evidence in support of Full Access accounts than RDAs. Perhaps our participants did not recognize the semantic differences between the tw o adjectival placements or properly integrate the context, but the fact remains that speakers who learned Spanish as their first language and native English speakers who learned Spanish as a second language both seemed to process the materials equivalently albeit in a faulty manner.
70 CHAPTER 7 CONCLUSIONS The purpose of this study was to investigate the processing abilities of English speaking learners of Spanish compared to native Spanish speakers. Many studies have used off line completion tasks, but the present study sought to also compare results from off line and on line tasks to discern whether learners of a second language have narrow underlying representations that prevent them from native like ultimate attainment, or whether learners are able t o overcome features set in their L1 through continued access to UG, and converge on native linguistic competence. The results here are not conclusive. We found that in the off line tasks, the groups were not statistically different from each other. In c ontrast to the results from Rothman et al., our natives did not perform better statistically than either the heritage speakers or the learners, even though we attempted to replicate the semantic interpretation task. In fact, numerically, the learners scor ed better than either more proficient group. Therefore, we find evidence against accounts of representational deficits. As for the on line tasks, we did not find the effects of grammaticality that we were hoping to find. The accuracy results from the plausibility task showed that all three groups were unsuccessful at differentiating between the grammatical and ungrammatical conditions. We might attribute these errors to biases in the subtleties of the adjectival manipulations, among other confounds in the participant pool and materials mentioned above. We also found almost no significant differences in reaction times to the plausibility questions, regardless of the factor of group. When we look at reaction times for word positions 4 9, the pairings w ith canonical adjective placement showed no differences in grammaticality, and were essentially identical at each word
71 position. The non canonical placement pairings did show effects of grammaticality, though there was not a consistent pattern either in t he word positions affected or in which condition had longer reaction times. This also did not vary significantly by group. We could say that learners in this study were able to achieve native like competence, but this would be misleading to a certain deg ree. None of our groups, including the native speakers, seemed to show sensitivity to the effects of grammaticality in either our experimentals or meaning changers. We can only argue that even if our pool of natives did not perform as well as expected, o ur learners did not fail to reach scores equivalent to those of the natives. We therefore cannot claim that the learners show evidence of having different underlying representations than natives, and we must reject representational deficit accounts. Our results are somewhat different from those found in Rothman (2010), both studies lead to the same conclusion, which is to dismiss RDAs. Before we could fully accept full access accounts, though, we would need evidence that the learners (and the natives, fo r that matter) had fully acquired the syntactic movements underlying the semantic changes in meaning, which we did not find here.
72 APPENDIX A EXPERIMENTAL ADJECTIVES Self Paced Reading Items Experimental Adjectives (32) 1. Aburrida: AN: Todas las pelculas del director duran entre 3 4 horas, y no es posible verlas sin dormirse en la mitad porque nunca tienen un tema interesante. NA: De todas las pelculas del director horrible, la menos interesante dura 3 horas y ensea como la pintura se seca en u na pared. o Es por eso que la aburrida/pelcula del director no gana dinero en el cine. 2. Apasionada: AN: Las mujeres con hijos siempre tienen sentimientos muy fuertes con respecto a sus nios y quieren protegerlos a toda costa. NA: No todas las madres quieren o cuidan a sus hijos, pero las que si, lo hacen con mucha pasin. o Lo obvio es que las apasionadas/madres aman a sus nios. 3. Aventurero: AN: A los nios normalmente les encanta descubrir y probar cosas nuevas sin pensar en los riesgos. NA: De los tres nios Garca, solo a uno le gusta probar y descubrir cosas nuevas, y los otros son muy cautelosos. o Es por eso que los aventureros/nios toman riesgos en su juventud. 4. Barato: AN: El rey colecciona los coches ms caros en el pas. NA: El asi stente del rey tiene algunos coches caros y otros de precios ms bajos, pero solo usa los coches caros para los eventos especiales. o Por lo tanto los baratos/coches no son usados por mucho. 5. Bonito: AN: Las modelos tienen que ser muy bonitas, altas, y delgadas, y por eso reciben proyectos con personas que hacen ropa. NA: Solo las modelos ms bonitas del mundo trabajan con famosas compaas como Gucci o Versace. o Por eso es que los bonitos/modelos sacan las contratos con estos diseadores.
73 6. Cariosa: AN: Una caracterstica importante de ser una monja es tener la compasin hacia otras personas y la voluntad de cuidar y ayudar a los que ms lo necesitan. NA: En comparacin a otras monjas que son menos amables, Madre Teresa era conocida por su amabilidad y generosidad excepcional, y por eso gan el premio Nobel de la paz en 1979. o Es importante saber que las cariosas/monjas se dedican a un fin benfico. 7. Caro: AN: Los diamantes son las piedras ms duras y bonitas del mundo, y por eso, cada diamante cuesta mucho. NA: Los diamantes que se muestran en el Museo Smithsonian, como el Diamante Hope, son los ms espectaculares del mundo y tienen precios igualmente espectaculares, especialmente cuando los comparas a diamantes normales. o Se puede dec ir que los caros/diamantes valen sus precios enormes. 8. Dbil: AN: Todos de los bebs recin nacidos no han desarrollado los msculos necesarios para sostener el peso de su cuerpo y por eso no son muy fuertes. NA: Aunque normalmente los bebs recin nacidos no son fuertes, hay algunos con una enfermedad que causa que sus msculos crezcan mucho ms rpido de lo normal y estos bebs aprenden movimientos mucho antes. o Lo normal es que los dbiles/bebes no se puedan sentar sin ayuda. 9. Delgada: AN: Cada pared en el apartamento es como una hoja de papel, y puedes or las conversaciones de otros pisos. NA: En el complejo de apartamentos, las paredes entre pisos son gruesas, pero las paredes en cada apartamento son como hojas de papel. o Es una lastima qu e las delgadas/paredes no bloqueen el sonido. 10. Deshonesto: AN: Los ladrones nunca dicen la verdad porque ganan dinero y cosas materiales por sus mentiras. NA: Algunos ladrones confiesan la verdad frente a la polica, y otros mentiran hasta la muer te. o Lo obvio es que los deshonestos/ladrones no merecen la confianza.
74 11. Difcil: AN: Es una propiedad inherente de los problemas de clculo que no son fciles y duran mucho tiempo en resolverlos. NA: Ningunos de los problemas do clculo son fciles pero los de clculo avanzado son casi imposibles de resolver. o Queda muy claro que los difciles/problemas toman mucha prctica para entenderlos. 12. Divertido: AN: Todos los videojuegos de Sony ofrecen muchas horas de diversin y siempre se agotan en las tiendas de videojuegos. NA: Algunos de los videojuegos de Sony son populares porque son entretenidos, pero hay otros que son aburridos o que frustran a los jugadores. o Es normal que los divertidos/videojuegos son ms populares en las tiendas. 13. Elegante: AN: Para ser una reina, una mujer tiene que ser refinada, hablar con elocuencia, y siempre comportarse bien. NA: Hay reinas que se comportan mal, y hay otras que son muy refinadas y hablan con elocuencia. o No hay duda que las elegantes/reinas so n buenos ejemplos para las nias. 14. Enojado: AN: Las abejas en general, tienen la reputacin de siempre estar enfadadas y de mal humor, y por eso nadie debe molestarlas. NA: De todos los tipos de abejas, la abeja reina es la ms enfadadiza y llena de hostilidad contra sus enemigos, y no duda en picar. o Teniendo esto en cuenta los enojados/insectos son peligrosos y pueden hacer dao. 15. Estudioso: AN: Todos los profesores pasaron mucho tiempo en las bibliotecas de sus universidades porque les encantaba aprender de su tema preferido. NA: Los mejores profesores de la astrofsica vivieron en las bibliotecas para estudiar, y los que evitaron sus estudios ahora no son profesores. o Queda muy claro que los estudiosos/profesores son importantes en sus universidades.
75 16. Est pido: AN: Todos los criminales hacen tonteras por las cuales los meten en la crcel al final. NA: Hay muchos criminales en el mundo, pero los que estn en nuestras crceles son los que hicieron errores tontos y dejaron pistas de su identidad en el lugar del crimen. o Somos afortunados que los estpidos/criminales no escapan del castigo. 17. Feo: AN: Todos los patitos de la madre pato se ven muy raros y poco atractivos. NA: La madre pato tiene 5 patitos, pero solo uno se ve diferente con pelo irregular y menos atractivo que los dems. o Todos ya saben que los feos/patitos se hacen ms bonitos de adultos 18. Fuerte: AN: Los monos en general, como los orangutanes, son fuertes y muy atlticos. NA: Los orangutanes masculinos son mas fuertes que los femeninos. o Todos saben que los fuertes/orangutanes son un tipo de mono. 19. Gordo: AN: La vida de un cerdo consiste en comer y dormir, y por eso cada da suben de peso ms y ms hasta un peso muy alto. NA: Los granjeros prefieren vender los cerdos ms grandes porque tienen mas carne. o En las granjas los gordos/cerdos ganan mucho dinero para los granjeros. 20. Honesto: AN: Los juicios importantes requieren un juez que sea leal a la ley y al acusado. NA: Hay jueces que tienen motivos fuera del juicio que les motivan a tomar malas decisiones. o Es por eso que los honestos/jueces se encargan de juicios importantes. 21. Importante: AN: En las ciencias y la medicina, todas las reas hacen investigaciones p ara aumentar el nivel del conocimiento y estas investigaciones siempre son muy influyentes y significativas. NA: La investigacin sobre el cncer es mas valiosa que otros tipos de investigacin porque sus xitos salvan y extienden las vidas de los ciudada nos del mundo. o Hoy en da las importantes/investigaciones curan las enfermedades ms serias.
76 22. Inteligente: AN: Todos los estudiantes son muy listos y van mucho a la biblioteca para estudiar y aprender los textos. NA: Las clases son muy avanzadas, en tonces hay pocos estudiantes que entienden el material. o Por eso es que los inteligentes/estudiantes sacan buenas notas en los exmenes. 23. Influyente: AN: Cada decisin que el presidente hace puede cambiar su pas y todas sus decisiones son importan tes para el bienestar del gobierno y los ciudadanos. NA: El presidente toma muchas decisiones, pero las que afectan a todo el pas reciben ms atencin en los medios de comunicacin porque pueden cambiar como vive la poblacin. o Con esto en mente las influy entes/decisiones deben ser discutidas en el gobierno. 24. Ocupado: AN: Las secretarias siempre estn respondiendo a llamadas y haciendo citas, entre otras tareas, y nunca tienen tiempo libre, pero son muy productivas. NA: Hay secretarias que pasan todo el da trabajando, pero otras intentan evitar sus tareas. o Por lo tanto las ocupadas/secretarias son las mejores secretarias. 25. Orgulloso: AN: Todos los padres de bebs recin nacidos siempre quieren hablar de sus nios y mostrar fotos a amigos porqu e estn muy satisfechos con su beb y de si mismos. NA: Para algunos padres, el nacimiento de sus nios es una carga, pero a otros les gusta hablar constantemente de su beb, mostrar fotos, y estn muy contentos. o No es sorprendente que los orgullosos/pad res se hacen una molestia. 26. Pattico: AN: Los equipos de baloncesto en Colorado siempre son horribles y los resultados finales de los juegos son una lstima. NA: Normalmente los equipos de baloncesto en Colorado son horribles, pero los de Denver y Colorado Springs han perdido todos sus juegos y sus admiradores. o Es bastante triste pero los patticos/equipos deben dejar de jugar. 27. Pesado: AN: La compaa de coches llamada Hummer manufactura vehculos grandsimos que pesan mucho y usan mucha ga solina. NA: La compaa Hummer, que normalmente fabrica coches gigantes, ha empezado a hacer coches mas pequeos y ligeros. o Por lo tanto los pesados/coches no son populares ahora.
77 28. Perezoso: AN: En un da normal, los gatos no hacen mucho ms que comer y dormir casi todo del da. NA: A la mayora de los gatos les gusta jugar y correr por la casa, pero hay algunos que solo quieren comer y dormir todo del da. o Es evidente que los perezosos/gatos no son muy activos. 29. Simptico: AN: En la guard era, todos los nios aprenden la importancia de pensar en otros y de compartir con sus amigos. NA: Aunque los instructores en la guardera ensean la buena educacin, solo los nios que son ms agradables practican como compartir. o Por eso a los simpticos /nios les gusta compartir sus juguetes. 30. Talentoso: AN: En sus libros, cada autor sabe como manipular sus palabras de una manera muy potica que entretiene a varios tipos de personas. NA: De todos los autores, hay pocos que tienen la capacidad de escribir libros que duran por los aos. o Por lo tanto los talentosos/autores crean novelas que cautivan a sus lectores. 31. Tmido : AN: Los estudiantes nuevos en la escuela son introvertidos y nunca quieren hacer preguntas a los profesores porque tienen m iedo. NA: De todos de los estudiantes nuevos, solo hay 2 que hablan y hacen preguntas en las clases porque el resto son muy introvertidos. o Con esto en mente los tmidos/estudiantes no entienden bien los temas. 32. Valiente: AN: Los bomberos tienen que poner sus vidas en peligro para salvar las vidas de otros, y muchos les consideran como hroes. NA: Durante un incendio, muchos bomberos se quedaban fuera mientras que algunos entraron a la casa para salvar las vidas de una familia. o Es evidente que los va lientes/bomberos deben ser respetados por su trabajo.
78 Meani n g Changing Adjectives (24) 1, 2. Alto: AN: Juan es un hombre de baja estatura con un trabajo muy importante en el gobierno de Gran Bretaa. NA: Alberto mide 6 pies y 6 pulgadas, entonces le cuesta mucho trabajo entrar por la puerta de su oficina. o Es por eso que el alto/ministro es tan evidente en la oficina. AN: La persona mas importante en la compaa es Jos, porque es el jefe de todos los empleados, pero mide solo 5 pies de estatura. NA: Jos, el empleado con la menos responsabilidad, es la persona ms grande relacionados al techo. o Es por eso que el alto/funcionario se merece todo el respeto. 3, 4. Cierta: AN: L a gramtica de unos artculos tiene que ser revisada antes de publicarlos en el peridico. NA: La mujer tiene problemas con su amigo porque le ha dicho una mentira. o Es por eso que las ciertas/cosas son muy importantes en la vida. AN: Hay tipos de leyes que no tienen sentido y parecen extraas o innecesarias, y hay otras que son importantes para la seguridad de todos los humanos. NA: Aunque que hay leyes en contra de matar o robar, no son necesarias porque son actos inherentemente m alos. o Todos ya saben que las ciertas/leyes no deben ser violadas. 5, 6. Listo: AN: Marco no es muy inteligente pero tiene todos sus materiales preparados para la reunin de maana. NA: El jefe est muy impresionado con la gran inteligencia de Pedro pero no con su falta de preparacin. o Es por eso que los listos/empleados son tan tiles en la oficina. AN: Marco siempre lee el material de la clase el da anterior para estar bien preparado aunque no entiende los temas. NA: Paola lleva muchos aos es tudiando la fsica molecular en la universidad y es la estudiante mas inteligente en el programa aunque nunca hace la tarea para sus clases. o Es por eso que los listos/estudiantes son tan exitosos en la universidad.
79 7, 8. Pobre: AN: La hermana de mi mad re se ha partido el pie al caerse, pero no importa porque tiene mucho dinero para pagar la cuenta del hospital. NA: La mujer lleva das sin poder comprar comida porque no tiene dinero, pero an es feliz. o Por ese motivo la pobre/mujer ha tenido un mal da. AN: El viernes, el carsimo porttil de Pedro se cay en el piso, rompiendo el disco duro, y l perdi todos sus archivos de la tesis de los dos aos anteriores. NA: Cuando Pedro compr su casa nueva, fue despedido de su trabajo mientras que tenia muchas deudas en su tarjeta de crdito. o Despus de todo el pobre/hombre tuvo que empezar de nuevo. 9, 10. Puro: AN: Los vendedores del almacn no solamente venden oro. NA: Los vendedores de oro y plata de alta calidad estn cobrando precios muy altos. o Es por eso que el puro/oro es tan difcil de encontrar. AN: La compaa que vende el agua embotellada no aade otros ingredientes ni qumicos a sus productos. NA: En la compaa que vende el agua embotellado, el agua pasa por muchos filtros y otro s procesos para limpiarlo y purificarlo. o Es por eso que la pura/agua tiene el sabor muy natural. 11, 12. Rara: AN: Cristian y Jos tienen una amistad que no es extraa pero pocas personas han tenido. NA: La mejor amiga de Mariana es una vaca, pero e so ocurre frecuentemente en las granjas. o Por eso pienso que las raras/amistades son muy interesantes en la vida. AN: De vez en cuando, hay una pelcula tan impresionante que se hace un clsico inmediatamente para todos los que la ven y es disfrutada au n mas porque no es abstracta ni extraa. NA: Mi amiga y yo vimos una pelcula muy artstica que era bastante extraa y abstracta, pero tambin bien hecha con temas profundos, como lo normal de este director. o Al fin de ao la rara/pelcula gana el premio Oscar.
80 13, 14. Real: AN: Durante muchos aos, la familia Hernndez vivi en el centro de Madrid limpiando casas y una estudiante quiso hacer un documentario sobre ellos. NA: En la pelcula La Bella Durmiente, la familia de Aurora pas una vida muy triste. o La pelcula muestra una real/familia y todo su pasado. AN: El matrimonio del Prncipe William y Kate Middleton fue una de las bodas mas anticipadas y maravillosas del a o 2011. NA: La pareja Griega va a tener un matrimonio grandsimo con familia de todas partes de los EEUU y Grecia. o Es por eso que el real/matrimonio fue tan costoso y elaborado. 15, 16. Seria: AN: El payaso est sangrando porque se cay de su moto cicleta en camino al trabajo. NA: Durante el funeral de Ernesto, todos los invitados estuvieron callados y respetuosos. o Debemos de manejar la seria/situacin con mucho cuidado y compasin. AN: Julio es muy amable, pero no tiene un sentido de humor b ueno y prefiere trabajar en vez de hacer chistes, y por eso no somos muy amigos. NA: Julio es mi mejor amigo en el mundo porque siempre me ayudara con cualquier problema o situacin mala, y lo hara con una risa y muchos chistes. o Despus de todo mi serio /amigo es una persona muy buena. 17, 18. Simple: AN: La profesora de matemticas intent no enfadarse cuando Andrea solo pregunt una pregunta difcil. NA: Marco tena vergenza de preguntarle varias preguntas fciles a la profesora. o El estudiante pr egunt una simple/pregunta en el primer da de clase. AN: Es muy comn que los estudiantes de Ingles hagan errores con el presente indicativo, porque es el tiempo mas frecuentemente usado. NA: El tiempo mas fcil para aprender en Ingles es el tiempo presente indicativo. o Es por eso que los simples/problemas no les molestan a los estudiantes.
81 19, 20. Sola: AN: Un grupo de amigos fue al museo pero la entrada al museo estaba limitada a una persona por hora. NA: La persona se pasa toda la tarde jugand o sin sus amigos o amigas. o Es por eso que una sola/persona est all esa tarde. AN: El medico, que siempre trabaja por las maanas, normalmente tiene que cuidar a muchos pacientes sin ayuda, y eso enfada a su novia. NA: El medico pas tanto tiempo estudiando y trabajando en el hospital que nunca poda salir o encontrar una esposa, como muchos de sus colegas. o Es por eso que el solo/medico no tiene mucho tiempo libre. 21, 22. nica: AN: Pablo lleg a los Estados Unidos de Brasil hace unos pocos das, entonces tiene un solo amigo y ese amigo es una persona normal. NA: La mujer tiene muchos amigos, pero hay un amigo muy especial que siempre le ayuda cuando tiene problemas. o Es recomendado tener a un nico/amigo para servir de compaa. AN: Au nque el nio tiene una familia grande, no tiene hermanos ni hermanas como el resto de los nios en su barrio. NA: La familia tiene 7 nios, pero solo Carlos sabe como tocar la guitarra y hablar Ingles. o Con esto en mente el nico/nio ha recibido mucha atencin de sus padres. 23, 24. Viejo: AN: Las jvenes Andrea y Laura han sido amigas desde que atendan la escuela primaria. NA: El nuevo amigo de Manuel cumple 75 aos la prxima semana, pero no va a tener una fiesta porque esta enfermo. o Por eso que los viejos/amigos deben ser apreciados y valorados. AN: Los nios Juan y Tomas se conocieron hace muchos aos, y han sido amigos desde siempre. NA: Mis mejores amigos tienen mas de 70 aos, son amigos muy leales y podemos hablar de cualqui er problema y me dan buenos consejos, pero les conoc hace solo un mes. o Es por eso que los viejos/amigos son importante en la vida.
82 APPENDIX B OFF LINE MATERIALS Language and Education Questionnaire Note: Your participation is voluntary; you may decline to answer questions. Age:___ Gender:___ L /R Handed ____ Hispanic: Y/N____ Race:_______ Education 1a. How many years of formal education do you have? _____ Check your highest level of education: [ ] less than High school [ ] Some college [ ] Masters [ ] High school [ ] College [ ] Ph.D./M.D./J.D. [ ] Professional training [ ] Some graduate school [ ] Other:_______ 1b. Have you taken any linguistics classes? If so, how many (roughly)?_____________ 1c. Have you taken any Spanish lang uage classes? If so how many?_______________ a. When did you start taking classes?__________________________________ b. What level are you currently at (ex: Intermediate 1/2)?__________________ 1d. If you did not learn Spanish in a classroom setting how did you learn it? _______________________________________ _______________________________ 1e. What is you major?________________________ Minor?_____________________ Language 2. Have you had, or do you have, any known problems with reading hearing, speech, language or learning?_______________If yes, please explain: ____________________________________________________________________ ____________________________________________________________________ 3. What variety/varieties of English do you spea k or have you learned: [ ] American English, standard [ ] American English dialect/regional variety, namely ____ [ ] Afro American English/Ebonics [ ] British English [ ] British English, dialect/regional variety, namely ____ [ ] Other English variet ies, namely ______ 4. What variety/varieties of Spanish do you speak or have you learned/been exposed to in learning (check all that apply):
83 [ ] Castellano/Spanish from Spain [ ] South American variety, namely ____ [ ] Central American variety, namely _____ [ ] Caribbean variety, namely______ [ ] Other Spanish variety, namely ______ 5. Please list all languages and dialects you know or have been exposed to ( including those you took at school ) in the order of acquisition List y our native language first. If you speak a regional variety or dialect, please list this as well. Language(s), in order of acquisition: Age you first acquired/wer e exposed to the language Number of years you used/have been using the language How you first learned the language (caregivers, school, friends, vacation, Order of current dominance 1= most dominant; 2= next most dominant, etc. 1 2 3 4 5 6 6. In the table below: a. List the percentage of time you are currently and on average exposed to each language/dialect you mention in (4) above. b. When choosing to read a text available in all your languages, in what percentage of cases would you choose to read it in each of your languages? Assume the original was written in a language you cannot read. c. When choosing a language to speak with a person who is equally fluent in all your languages, what percentage of time would you choose to speak each language? Language: a. Percentage current average use b. Preference READING Percentage of time (see instructions) c. Preference SPEAKING Percentag e of time 1 2
84 3 4 5 6 Total: 100% 100% 100% Language(s) spoken Age acquired 1 Mother or other female primary caregiver 2 Father or other male primary caregiver 3 If applicable: Other female caregiver (e.g. stepmother) 4 If applicable: Other male care giver 8. If you were not born in the US, at what age did you move to the States? ___________ In which country or countries did you l ive before you moved to the US? ___________ __ _______________________________________ _______________________________ 9. If you were born in the US, have you ever lived in another country? ______________ In which country or countries have you lived and for how long?___________________________________ ______________________________ For the dialect of Spanish and the dialect of English that you are most proficient in (even if you hardly speak them), please complete the following question s on the attached sheets. Use a separate sheet for each language.
85 Primary English dialect:_____________________ This is my [ ] native; [ ] second; [ ] third; [ ] fourth language List your level of proficiency in the language: 0=N Speaking 0 1 2 3 4 5 6 7 8 9 10 Understanding spoken language 0 1 2 3 4 5 6 7 8 9 10 Reading 0 1 2 3 4 5 6 7 8 9 10 Writing 0 1 2 3 4 5 6 7 8 9 10 List the number of years or months you spent in each language environment years months A country where the language is spoken A family that speaks the language A school/working environment in which the language is spoken Select how much the following factors contributed to you learning the language 0=no contributor 10=most important contributor Interacting with family 0 1 2 3 4 5 6 7 8 9 10 Interacting with friends 0 1 2 3 4 5 6 7 8 9 10 Reading 0 1 2 3 4 5 6 7 8 9 10 Language tapes/self instruction 0 1 2 3 4 5 6 7 8 9 10 Formal classes 0 1 2 3 4 5 6 7 8 9 10 Watching TV/movies 0 1 2 3 4 5 6 7 8 9 10 L istening to radio/music 0 1 2 3 4 5 6 7 8 9 10 Internet 0 1 2 3 4 5 6 7 8 9 10 Other, namely: 0 1 2 3 4 5 6 7 8 9 10 Rate to what extent you are currently exposed to the language in the following contexts 0=never Interacting with family 0 1 2 3 4 5 6 7 8 9 10 Interacting with friends 0 1 2 3 4 5 6 7 8 9 10 Reading 0 1 2 3 4 5 6 7 8 9 10 Language tapes/self instruction 0 1 2 3 4 5 6 7 8 9 10 Formal classes 0 1 2 3 4 5 6 7 8 9 10 Watching TV/movies 0 1 2 3 4 5 6 7 8 9 10 Listening to radio/music 0 1 2 3 4 5 6 7 8 9 10 Internet 0 1 2 3 4 5 6 7 8 9 10 Other, namely: 0 1 2 3 4 5 6 7 8 9 10 In your perception, how much of a foreign accent do you have in the language? (0=no; 5=moderate; 10=pervasive): 0 1 2 3 4 5 6 7 8 9 10 How frequently do others identify you as a non native sp eaker of the language, based on y our accent? (0=never; 5=half the time; 10=always): 0 1 2 3 4 5 6 7 8 9 10 To what extent do you identify yourself with the culture of the people who speak the language? (0=no identification; 10=complete identification): 0 1 2 3 4 5 6 7 8 9 10 Primary Spanish dialect:_____________________
86 This is my [ ] native; [ ] second; [ ] third; [ ] fourth language List your level of proficiency in the language: Speaking 0 1 2 3 4 5 6 7 8 9 10 Understanding spoken language 0 1 2 3 4 5 6 7 8 9 10 Reading 0 1 2 3 4 5 6 7 8 9 10 Writing 0 1 2 3 4 5 6 7 8 9 10 List the number of years or months you spent in each language environment years months A country where the language is spoken A family that speaks the language A school/working environment in which the language is spoken Select how much the following factors contributed to you learning the language 0=no contributor 10=most important contributor Interacting with family 0 1 2 3 4 5 6 7 8 9 10 Interacting with friends 0 1 2 3 4 5 6 7 8 9 10 Reading 0 1 2 3 4 5 6 7 8 9 10 Language tapes/self instruction 0 1 2 3 4 5 6 7 8 9 10 Formal classes 0 1 2 3 4 5 6 7 8 9 10 Watching TV/movies 0 1 2 3 4 5 6 7 8 9 10 Listening to radio/music 0 1 2 3 4 5 6 7 8 9 10 Internet 0 1 2 3 4 5 6 7 8 9 10 Other, namely: 0 1 2 3 4 5 6 7 8 9 10 Rate to what extent you are currently exposed to the language in the following contexts 0=never Interacting with family 0 1 2 3 4 5 6 7 8 9 10 Interacting with friends 0 1 2 3 4 5 6 7 8 9 10 Reading 0 1 2 3 4 5 6 7 8 9 10 Language tapes/self instruction 0 1 2 3 4 5 6 7 8 9 10 Formal classes 0 1 2 3 4 5 6 7 8 9 10 Watching TV/movies 0 1 2 3 4 5 6 7 8 9 10 Listening to radio/music 0 1 2 3 4 5 6 7 8 9 10 Internet 0 1 2 3 4 5 6 7 8 9 10 Other, namely: 0 1 2 3 4 5 6 7 8 9 10 In your perception, how much of a foreign accent do you have in the language? (0=no; 5=moderate; 10=pervasive): 0 1 2 3 4 5 6 7 8 9 10 How frequently do others identify you as a non native speaker of the language, based on your accent? (0=never; 5=half the time; 10=always): 0 1 2 3 4 5 6 7 8 9 10 To what extent do you identify yourself with the culture of the people who speak the lan guage? (0=no identification; 10=complete identification): 0 1 2 3 4 5 6 7 8 9 10
87 Debriefing Questions 1. Was th e font used either too big, or too small? a. What about the style of the letters? 2. Was the self paced reading task too long overall? a. Would an extra break have been helpful? 3. On a scale of 1 7 (1=very easy, 7=very difficult), how hard was the self paced reading task? (the one on the computer) 4. Were you able to keep focus throughout the entire experiment? 5. Did you have any problems with the vocabulary or the wording in either the context sentence or the sentence you clicked through word by word? 6. Did the Spanish used in the self paced reading task seem to favor a specific dialect of Spanish, or did it feel fairly standard to you? 7. Were you able to figure out what the self pa ced reading task was testing? If so, please explain: 8. Did you use any strategies to help you through the self paced reading task? (such as clicking through the words rapidly to answer the question after?) 9. Have you participated in any other linguistic experiments? Or any other experiments? 10. Was there anything else about the experiment thus far that you noticed or would like to comment on?
88 Vocabulary Please provide the English translation of the following words aburrido autor apasionado b be aventurero bombero barato cerdo bonito coche carioso criminal caro decisin dbil diamante delgado estudiante deshonesto equipo difcil gato divertido insecto elegante investigacin enojado juez estudioso ladrn est pido madre feo modelo fuerte monja gordo nio honesto orangutanes importante padres influyente pared inteligente patito ocupado pelcula orgulloso problema pattico profesor pesado reina perezoso secretaria simptico videojuego talentoso tmido valiente
89 Semantic Interpretation Task: Meaning Changers For the following sentences, please indicate the meaning of the underlined portion of the sentence by circling the corresponding box. 1. Mi tio es un pobre hombre 2. La leche pura sabe rico 3. Tomas es mi viejo amigo 4. El examen pregunt los problemas simples 5. El solo abogado trabaja por la noche 6. Maria es una nica nia 7. El hombre es un alto funcionario en la empresa My uncle is unfortunate. Thomas is my friend who is elderly. The exam asked simple questions. The single lawyer works at night. Maria is a unique child. Pure milk tastes good. The man is not wealthy. Just/mere milk tastes good. Maria is an only child. The only lawyer works at night. The exam asked common questions. Thomas is my long time friend. The man is a high ranking official in the business. The man is a ta ll official in the business.
90 8. La princesa vive en un real castillo 9. Mi amiga seria no le gusta salir 10. Vi el mismo actor en la pelicula 11. Las sillas raras estan en la sala 12. La tienda vende las gafas varias 13. Ciertas respuestas son misteriosas The princess lives in a royal castle. The princess lives in a real castle. My true/serious friend. I saw the actor himself in the movie. The strange chairs are in the living room. The store sells different types of glasses. Certain answers are mysterious. My friend who is a serious person. I saw the same actor in the movie. The rare chairs are in the living room. The store sells several glasses. True/Correct answers are mysterious.
91 Semantic Interpretation Task: Experimentals For the following sentences, please indicate the meaning of the underlined portion of the sentence by circling the corresponding box. 1. Los doctores ocupados hacen mas errores. 2. El vino barato tiene un mal sabor. 3. Las elegantes joyas son reservado para la reina. 4. Los incas valientes resistieron a los conquistadores. 5. Los libros aburridos me dan sueo. 6. Las personas simpticas les gusta sonreir. Doctors, who are busy, make errors. Cheap wine tastes bad, not expensive wine. The elegant jewelry is reserved for the queen, not the simple jewelry. Only valiant Incas resisted the conquerors, cowardly Incas did not. Books, which are boring, make me sleepy. Incas, who are known to be brave, resisted the conquerors. The jewelry, which is elegant, is reserved for the queen. The busy doctors, not the ones with free time, make errors. Wine, which is cheap, tastes bad. Boring books, not the exciting books, make me sleepy. People, who are friendly in general, like to smile. Friendly people, not angry people, like to smile.
92 7. Las bonitas chicas son arrogantes. 8. Los contratos importantes necesitan ser firmados. 9. Los orgullosos padres hablan de sus hijos. 10. Los gatos enojados no dudan en pelear. 11. Los nios timidos no dejan el lado de sus madres. 12. Los modelos delgados no comen mucho. Girls, who are pretty, are arrogant. The pretty girls, not the unattractive ones, are arrogant. Contracts, which are important, n eed to be signed. Parents, naturally proud, talk about their children. Cats, who are hesitate to fight Timid children, not brave, side. Children, who leave their Proud parents, as opposed to embarrassed parents, talk about their kids. Important contracts need to be signed, not the insignificant ones. Angry cats, not hesitate to fight. Models, who eat much. Thin models compared to plus size models.
93 13. Los pesados camiones manejan muy despacio. 14. Los ladrones deshonestos le mienten a la policia. 15. Los inteligentes profesores leen mucho 16. Los musicos apasionados escriben sus propias canciones. 17. Los perros perezosos duermen en el sol. 18. Las mujeres debiles no levantan pesas. Semi trucks, which are heavy, drive very slowly. Thieves, who are dishonest, lie to the police. The dishonest thieves, not the honest ones, lie to the police. Heavy semi trucks drive very slowly, compared to the light trucks Professors, that are intelligent, read a lot. Passionate musicians, not indifferent musicians, write their own songs. Lazy dogs, not active dogs, sleep in the sun. Dogs, which are lazy, sleep in the sun. Intelligent professors, as opposed to slow professors, read a lot. Musicians, who are passionate, write their own songs. Weak women, not strong weights. Women, who are physically weak,
94 19. Las gimnastas talentosas practican mucho. 20. Unos estudiosos estudiantes vinieron al congreso 21. Los montaeros aventureros viajan por el mundo. 22. Los gordos cerdos comen todo el dia. 23. Los tigres fuertes dominan los ciervos. 24. Las dificiles matemticas confunden todo el mundo. Talented gymnasts practice a lot, not poor gymnasts. Some studious students came to the conference, not the lazy ones. Gymnasts, who are talented, practice a lot. Some students, who are studious, came to the conference. Mountain climbers, who are adventurous, travel the world. Fat pigs, not skinny pigs, eat all day. Tigers, who are strong animals, overpower deer. The adventurous mountain climbers, not the timid one, travel the world. Pigs, which are fat, eat all day. The strong tigers overpower deer, not the weak tigers. Math, which is difficult, confuses everyone. Difficult mathematics, not the easy kind, confuses everyone.
95 25. Los feos patitos se convier ten en cisnes. 26. Los honestos ciudadanos pagan los impuestos. 27. Los caros diamantes brillan en la luz. 28. Las excusa s pattica s no resuelven los problemas 29. Los pol ticos influyentes se hacen conocidos 30. Para los adultos divertidos la vida es una fiesta The expensive diamonds sparkle in the light, not the cheap ones. Ducklings, which are ugly, become swans. Ugly ducklings become swans, not cute ducklings. The citizens who are honest, not the dishonest ones, pay their taxes. Citizens, who are honest, pay their taxes. Diamonds, which are expensive, sparkle in light. Excuses, which are pathetic, problem. Pathetic excuses problem, unlike good excuses. Politicians, who are influential, become well known. Influential politicians become well known, not the unimportant ones. For adults, who are fun, life is a pa rty. For those adults, who are also fun not boring, life is a party.
96 31. Los e st pido s criminales fueron detenidos 32. Las novias cariosas cuidan a sus novios The criminals, who are stupid, were arrested. The criminals, who are stupid, not smart, were arrested. Girlfriends, who are caring, take care of their boyfriends. The girlfriends who are caring, but not the mean ones, take care of their boyfriends.
97 REFERENCES Alemn Ban, J., Fiorentino, R., & Gabriele, A. (2012). The processing of number and gender agreement in Spanish: An event related potential investigation of the effects of structural distance. Brain Research 1456 49 63. Anderson, B. (2008). Forms of evidence and grammatical development in the acquisition of adjective position in L2 french. Studies in Second Language Acquisition,30 (1), 1 29. doi:10.1017/S0272263108080017 Anderson B ( 2001 ). Adjective position and interpretation in L2 French In J. Camps & C. R. Wiltshire (Eds.), Romance syntax, semantics and L2 acquisition (pp. 27 42 ).Amsterdam : Benjamins. Bernstein J ( 2001 ). The DP hypothesis: Identifying clausal properties in the nominal domain. In M. Baltin & C. Collins (Eds.), The handbook of contemporary syntactic theory (pp. 536 561 ). London : Blackwell Chomsky, N. (1965). Aspects of the theory of syntax Oxford, England: M.I.T. Press. Clahsen, H., & Felser, C. (2006). Continuity and shallow structures in language processing. Applied Psycholinguistics, 27 (1), 107 126. doi:10.1 017/S0142716406060206 Clahsen, H., & Felser, C. (2006). Grammatical processing in language learners. Applied Psycholinguistics, 27 (1), 3 42. doi:10.1017/S0142716406060024 Clahsen, H., & Felser, C. (2006). How native like is non native language processing ? Trends in Cognitive Sciences, 10 (12), 564 570. doi:10.1016/j.tics.2006.10.002 Epstein, S., Flynn, S., & Martohardjono, G. (1996). Second language acquisition: Theoretical and experimental issues in contemporary research. Brain and Behavioral Sciences 19, 677 714. Ferreira, F., Bailey, K. G. D., & Ferraro, V. (2002). Good enough representations in language comprehension. Current Directions in Psychological Science, 11 (1), 11 15. doi:10.1111/1467 8721.00158 Flynn, S. (1996). A parameter setting appro ach to second language acuisition. In W. Ritchie and T. Bhatia (eds.), Handbook of language acquisition (pp. 121 58). San Diego: Academic Press.
98 Flynn, S. and G. Martohardjono. (1994). Mapping from the initial state to the final state: the separation of universal principles and language specific principles. In B. Lust, M. Suer and J. Whitman (eds.), Syntactic theory and first language acquisition: cross linguistic perspectives. Vol. 1: Heads, projections and learnability (pp. 319 35). Hillsdale, NJ: L awrence Erlbaum. Gabriele, A., Fiorentino, R., & Alemn Ban, J. (in press). Examining second language development using event related potentials: a cross sectional study on the processing of gender and number agreement. Grter, T., Lew Williams, C., & Fernald, C. (2012). Grammatical gender in L2: A production or a real time processing problem? Second Language Research, 28(2), 191 215. Hawkins, R. (1997). The partial availability of universal grammar in second language Second Language Research, 13 (3), 187. Hopp, H. (2009). The syntax discourse interface in near native L2 acquisition: Off line and on line performance. Bilingualism Language and Cognition, 12 (4), 463 483. doi:10.1017/S13667289099 90253 Marian, V., Blumenfield, H. K., & Kaushanskaya, M. (2007). The language experience and proficiency questionnaire (LEAP Q): Assessing language profiles in bilinguals and multi linguals. Journal of Speech and Hearing Research, 50 940 967. Rothman, J., Judy, T., Guijarro Fuentes, P., & Pires, A. (2010). ON THE (UN) AMBIGUITY OF ADJECTIVAL MODIFICATION IN SPANISH DETERMINER PHRASES informing debates on the mental representations of L2 syntax. Studies in Second Language Acquisition, 32 (1), 47 77. Schwar tz, B. D. (1998). The second language instinct. Lingua 106, 133 160 Schwartz, B. D., & Sprouse, R. A. (1996). L2 cognitive states and the full Transfer/Full access model. Second Language Research, 12 (1), 40 72. Sorace, A. A. (2011). Pinning down the Linguistic Approaches to Bilingualism, 1 (1), 1 33. Sorace, A., & Filiaci, F. (2006). Anaphora resolution in near native speakers of italian. Second Language Research, 22 (3), 339 368. doi:10.1191/0267658306sr271oa Sprouse, R. A. (2011). The interface hypothesis and full Transfer/Full Access/Full parse: A brief comparison. Linguistic Approaches to Bilingualism, 1 (1), 97 100. doi:10.1075/lab.1.1.16spr
99 Tsimpli, I. M., & Dimitrakopoulou, M. (2007). The interpretability hypothesis: Evidence from wh interrogatives in second language acquisition. Second Language Research 23, 215 242. Tsimpli, I. M., & Roussou, A. (1991). Parameter resetting in L2? UCL Working Papers in Linguistics 3, 149 170. Valds, G. (2000). Spani sh for Native Speakers: AATSP Professional Development Series handbook for teachers K 16 (Vol. 1). New York, NY: Harcourt College Publishers, p. 1. White, L. (2003). Second language acquisition and Universal Grammar. New York: Cambridge University Press. White, L. (1985b). The pro drop parameter in adult second language acquisition. Language Learning 35: 47 62. White, L. (1986). Implications of parametric variation for adult second language Cook (ed.), Experimental approaches to second language acquisition (pp. 55 72). Oxford: Pergamon Press.
100 BIOGRAPHICAL SKETCH Carlie Anne Overfelt was born November 30 th 1985 in Radford, Virginia. The younger of two children, she grew up in Galax, Virginia, graduating with honors from Sc ience and Technology in 2004. She earned her B.A. in Linguistics, with a minor in Spanish fr om the University of Virginia in 2008, and an M.A. in Linguistics from the University of Florida in 2012. A summer abroad during 2007 in Valencia, Spain spurred Carlie to seek work outside the US. Upon graduating from U Va in 2008, she moved to Madrid, S pain to pursue a job in teachin g English as a foreign language. There, she completed a TEFL class in order to earn certification to teach She spent the following year teaching various levels and age groups of learners in Madrid, and traveled throughout t he country. In her free time she also worked with a local international student housing company to translate texts from Spanish to English and continues to work for them sporadically. After returning to the United States in the Fall of 2009, Carlie ap plied and was accepted to the University of Florida, and spent the next year working in Galax, Virginia to prepare for graduate school. In Gainesville, she received the tremendous opportunity to manage the Brain and Language Lab at UF, under the direction of Dr. Edith Kaan. The current paper is the product of work completed in that lab. Following her time in Florida, Carlie plans to look for work in the Northern Virginia/Washington DC area, to be close to family and friends. Her parents, Lloyd and Debb ie, have retired from work in the field of Special Education in Galax, Virginia, and her older brother, Jeremy, is growing in the ranks of college football while currently
101 working as Special Teams Coordinator and strength and conditioning coach for all tea ms at Shepherd University outside DC in Shepherdstown, West Virginia.