1 SYNTACTIC DEVELOPMENT OF THE SPANISH SUBJUNCTIVE IN SECOND LANGUAGE ACQUISITION: COMPLEMEN T SELECTION IN NOMINAL CLAUSES By LAURIE A. MASSERY A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009
2 2009 Laurie A. Massery
3 To Richard, Mom, Dad, Lisa, Michelle, Michael, and Sonya
4 ACKNOWLEDGMENTS I thank Dr. Gillian Lord Dr. Theresa Antes, Dr. Joaquim Camps, and Dr. Eric Potsdam for their invaluable knowledge and expertise in the fields of Second Language Acquisition and Syntax. Their continued guidance and input th roughout this venture has made this project possible. I also thank Claudio Fuentes A.B.D. fo r his assistance in the statistical analysis and model building in this project. His knowledge and proficiency in statistics has allowed us to make important conclusions in our field of study.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES ...........................................................................................................................7LIST OF FIGURES .........................................................................................................................9LIST OF TERMS ...........................................................................................................................11ABSTRACT ...................................................................................................................... .............13 CHAP TER 1 THE SPANISH SUBJUNCTIVE ...........................................................................................151.1 Introduction ...................................................................................................................151.2 Modal and Syntactic Explanati ons of the Spanish Subjunctive .................................... 151.2.1 Evolution of Deontic Modality .........................................................................171.2.2 Ambiguity in Epistemic Modality ..................................................................... 191.2.3 Deontic and Epistemic Moda lity in Nominal Clauses ...................................... 211.3 English Irrealis ..............................................................................................................231.4 Generative Grammar ..................................................................................................... 261.4.1. Principles and Parameters ................................................................................. 271.4.2 Syntactic Structures ...........................................................................................311.4.3 Functional Categories........................................................................................341.4.4 L1 Transfer ........................................................................................................461.5 Summary of Goals ........................................................................................................ 511.6 Conclusion ............................................................................................................... .....562 ACQUISITION OF THE SUBJUNTIVE .............................................................................. 582.1 Introduction ...................................................................................................................582.2 L1 and L2 Acquisition Patterns.....................................................................................592.3 L2 Acquisition of the Spanish Subjunctive ...................................................................622.4 Production Studies ....................................................................................................... .632.5 Input Studies ............................................................................................................ .....682.6 Conclusion ............................................................................................................... .....773 METHODOLOGY ................................................................................................................. 793.1 Introduction ...................................................................................................................793.2 Participants ....................................................................................................................793.3 Instruments ....................................................................................................................803.3.1 Magnitude Estimation (ME) ............................................................................. 803.3.2 ME Data Collection ..........................................................................................82
6 3.3.3 L2 Mood and Conjugation Task (L2MAC) ...................................................... 86 3.3.4. Other Information.............................................................................................. 873.4 Scoring Procedures ....................................................................................................... 883.5 Conclusion ............................................................................................................... .....894 RESULTS AND DISCUSSION ............................................................................................. 914.1 Introduction ...................................................................................................................914.2 Magnitude Estimation Task .......................................................................................... 914.3 L2MAC .........................................................................................................................964.3.1 Results for Epistemic and Deontic Modality in Nominal Clauses .................. 1014.3.2 Modal Analysis of ES .....................................................................................1044.3.3 Epistemic Context Analysis ............................................................................ 1064.4 Decline in Accuracy at the Intermediate Level ........................................................... 1084.5 Research Question 1 ....................................................................................................1114.6 Research Question 2 ....................................................................................................1144.7 Summary of Findings ..................................................................................................1165 CONCLUSION .................................................................................................................... .1185.1 Introduction .................................................................................................................1185.2 IL Representations ......................................................................................................1185.3 Contributions and Implications ................................................................................... 1245.4 Limitations of Study .................................................................................................... 1275.5 Pedagogical Implications ............................................................................................ 1315.6 Closing .................................................................................................................. ......135 APPENDIX A ME (MAGNITUDE ESTIMATION TASK) ........................................................................ 138B L2MAC (L2 MOOD AND CONJUGATION TASK) ......................................................... 140C SI (SYNTACTIC IDENTIFICATION TASK) ....................................................................141D EXPLANATION OF SI ....................................................................................................... 143E BACKGROUND FORM AND QUESTIONNAIRE ........................................................... 145LIST OF REFERENCES .............................................................................................................146BIOGRAPHICAL SKETCH .......................................................................................................155
7 LIST OF TABLES Table page 1-1 Idiosyncratic behavior of epistemic verbs in Spanish subjunctive modality ..................... 202-1 L1/L2 parallel subjunctive acquisition patterns .................................................................603-1 Syntactic environments used in ME .................................................................................. 864-1 Results for ME Task ...................................................................................................... ....924-2 Means and standard deviations for ME and corresponding p values ................................. 934-3 Compared Syntactic Environmen ts: p values adjusted by Tukey ...................................... 944-4 ME Environment ranking .................................................................................................. 954-5 Raw Scores for L2MAC ....................................................................................................974-6 Sample scores for L2MAC (mood and conjugation abilities) ...........................................974-7 Summary of tests for significant effects of the variables included in the model for L2MAC ......................................................................................................................... .....994-8 Least Squares Means for effect course for L2MAC ........................................................ 1004-9 LS Means of the scores for L2MACcontrolling by course .............................................. 1014-10 ANOVA table for the model consideri ng course, ability, and environment as variables ...........................................................................................................................1024-11 P values for differences in m odal environment adjusting by Tukey ............................... 1024-12 Multiple Comparisons: Acquisition of DS, EI, ES modalities for L2MAC .................... 1024-13 P values for difference between course s for L2MAC controlling by DS environment ... 1034-14 LS Means for the interactions of course and modal environment ................................... 1034-15 P values for difference between cour ses for L2MAC controlling by ES and EI environments .................................................................................................................. ..1054-16 LS Means for the interac tions of course and modal environment (ES Scores only) .......1064-17 LS Means for the interactions of course and modal environment ................................... 1074-18 Possible subordinate structures in Spanish ...................................................................... 112
8 4-19 Possible subordinate structures in English ....................................................................... 1135-3 Sample presentation of English irre alis and Spanish subjunctive structures ...................135
9 LIST OF FIGURES Figure page 1-1 A syntactic representation of I want him to buy milk. ........................................................ 321-2 A syntactic representation of quiero que pro compre leche. ............................................321-3 A syntactic representation of IL s yntax during L2 subjunctive acquisition: IP complement quiero t venir .............................................................................................361-4 A syntactic representation of IL syntax during subjunctive acquisition (* Quiero t vienes .) ............................................................................................................................ ...371-5 Unspecified FPs .......................................................................................................... .......381-6 A syntactic representation of IL syntax during subjunctive acquisition (* Quiero t vienes. ) ............................................................................................................................. ..451-7 A syntactic representation of Spanish subjunctive syntax ( Quiero que bailes. ) ................ 521-8 A syntactic representation of L1 English irrealis syntax ( I want you to dance ). ...............521-9 A syntactic representation of IL s yntax during L2 subjunctive acquisition (* Espero t venir *I hope you to come) .........................................................................................531-10 A syntactic representation of IL s yntax during L2 subjunctive acquisition (* Espero t venir *I hope you to come) .........................................................................................541-11 A syntactic representation of IL s yntax during L2 subjunctive acquisition (* Espero t vienes I hope you come) .............................................................................................. 553-1 Sample reference line demonstration ................................................................................. 834-1 L2MAC raw scores (mood + conjugation) out of 30 ......................................................... 994-3 Line graph of DS, ES, and EI mood scores .....................................................................1094-4 Example of error production and co rrection in developing IL systems ........................... 1104-5 Complexity of Spanish Modal System ............................................................................ 1165-1 A syntactic representation of IL s yntax during L2 subjunctive acquisition: IP complement (* quiero t venir ) ........................................................................................ 1205-2 A syntactic representation of IL s yntax during L2 subjunctive acquisition (* Espero t venir ) ............................................................................................................................ 120
10 5-3 A syntactic representation of IL syntax during subjunctive acquisition (* Espero tu vienes. ) ............................................................................................................................. 1225-4 A representation of IL synt ax during subjunctive acquisition (* espero t venir ) ........ 123
11 LIST OF TERMS 1. COMPLEMENTIZER: que that 2. CP: Complementizer Phrase (i.e., que comas that you/he/she eat) 3. DEONTIC MODALITY: utterances that connote influen ce, desire, obligation, indirect commands (i.e., desear to desire, obligar to obligate, querer to want, recomendar to recommend) 4. EPISTEMIC MODALITY: utterances that connote varied degree s of truth value a nd/or levels of reality (i.e, creer to believe, pensar to think, dudar to doubt, imaginarse to imagine, etc.) 5. FAILED FUNCTIONAL FEATURES HYPOTHESIS (FFH): learners are unable to acquire functional features in L2 6. FUNCTIONAL CATEGORIES (FC): categories such as DP, IP, and CP which are responsible for the morphological feature ch ecking of lexical items (V, N, Adj, Adv, P) to ensure accurate structure building in a given language 7. INTERLANGUAGE (IL): learners devel oping language system 8. IP: INFINITIVAL PHRASE (i.e., comer to eat) 9. IRREALIS MODALITY: difference between factual and non-factual information 10. L2MAC (L2 mood and conjugation abilities task): ta sk used to collect modal identification and conjugation data from learners to m easure learners ability in these areas 11. LEXICON: a list of words in a given language and their idiosyncratic linguistic properties 12. LINEAR MIXED MODEL: used to capture possible correlation in the responses of each participant 13. LS MEANS: a term used in statistics that refers to the statistical procedure necessary for computing the mean of the av erage in an unbalanced group 14. MAGNITUDE ESTIMATION (ME): a procedure originally used in psychophysics to measure perception using visual stimuli 15. MARKEDNESS THEORY: a feature that separa tes less complex and more complex linguistic features in languages (i.e., subjunctive = complex, indicative/infinitive = less complex) 16. MARKEDNESS DIFFERENTIAL HYPOTHESIS: unmarked features in L1 are more likely to transfer to L2 (i.e., indicativ e and infinitival verb forms)
12 17. MINIMALIST PROGRAM: the framework that comes after Principles and Parameters that adds distinctions between lexical and functional category development and places greater emphasis on the acquisition of feature speci fication as part of lexical knowledge 18. MODALITY: acutely describes the intent of the speaker and her level of commitment to what is being uttered 19. MOOD: grammatical category that requires mor phological and syntactic agreement (i.e., CP + indicative/subjunctive morphology) that correspond to the intent of the speaker 20. MISSING SURFACE INFLECTION HYPOTHESIS: learners have access to functional categories in second language acquisition even though such categories are not always present in L2 production 21. PRINCIPLES AND PARAMETERS: the internally focused linguisti c framework that that revised specifications of what constitutes innate capaci ty to include more abstract notions of general principles and constr aints common to human language as part of a Universal Grammar 22. S1: subject of principle clause 23. S2: subject of subordinate clause 24. SUBJUNCTIVE: a grammatical category that must be represented morphologically and syntactically indicating desire, influence, obligation, possibilit y, and low levels of certainty 25. UNIVERSAL GRAMMAR: innate linguistic knowledge that is suggested to consist of principles and parameters common to all human languages 26. V1: matrix verb found in principle clau se of a complex s yntactic structure (S1, V1 + S2, V2) 27. V2: subordinate verb found in the second clause of complex syntactic structure (S1, V1 + S2, V2) 28. X-BAR THEORY: shows hierarchical structures th at all human languages adhere to.
13 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SYNTACTIC DEVELOPMENT OF THE SPANISH SUBJUNCTIVE IN SECOND LANGUAGE ACQUISITION: COMPLEMEN T SELECTION IN NOMINAL CLAUSES By Laurie A. Massery August 2009 Chair: Gillian Lord Major: Romance Languages If you ask students taking Spanish as a fore ign language which grammar point they find most challenging, many of them will probably sa y the subjunctive. This may be because the more common use of the subjunctive in English is captured by non-subjunctive structures like infinitival or null CP [+indicative] structures. From a generative perspective, the present study investigates second language (L2) syntactic and semantic development of the Spanish subjunctive at the University leve l. The results of the study reveal that learners at all levels of instruction continued to transfer first language (L1) syntactic ru les of English irrealis to L2 structures during acquisition. For this study, a total of 160 English-speak ing learners studying Spanish as a second language were recruited. Magnitude Estimation (M E) was used to gage learners syntactic intuitions about the Spanish subjunc tive by requiring them to analyze a series of structures in L2. We found that in their judgment of the sentences, learners transferred the [+ optional] overt CP rule in English to Spanish. In addition to thes e syntactic analyses, we used a guided production task which measured learners semantic performan ce as well. The data showed that learners were less successful at identifying epistemic modality (+/reality, existence, etc.) in nominal clauses,
14 as sentences using deontic modality (+/obligation, desire, influence, etc.) received substantially higher scores. The data were also analyzed in terms of learnability and second language grammatical capabilities as well as in terms of implications for the L2 classroom. The pedagogical implications of this study include the need to capitalize on learne rs ability to access functional categories in nominal clauses and make the [-optiona l] overt CP rule in Spanish more salient to learners.
15 CHAPTER 1 THE SPANISH SUBJUNCTIVE 1.1 Introduction The Spanish subjunctive is one of the m ost, if not the most, difficult grammar points for English speaking learners of Spanish as a second language to conque r (Collentine 1997, 1998; DiFranco 2005; Farely 2004; Gumestad 2006; Jelsinki 1977; L ubbers-Quesada 1996, 1998; Terrell & Hooper 1974; and Terrell, Ba ycroft and Perrone 1987). This is due most likely to the fact that: (1) the Spanish subjunctive require s understanding the concept of mood; (2) the concept of mood is not clear to learners in their first langua ge because authentic subjunctive structures are evasive in English and are not freq uently used in colloquial speech as in Spanish; and finally (3) most English speakers captu re subjunctive meani ng through non-subjunctive sentence structures that include indicative morphology in compleme nt clauses (CPs), infinitival clauses, and gerunds. This project addresses thes e issues and applies them to the acquisition of the Spanish subjunctive illustrating the complicated process that L2 learners of Spanish must face when learning this difficult grammar point. This chapter presents various theoretical definitions of, and explanations for, the Sp anish subjunctive including relevant terminology pertaining to modal dichotomies such as epistemic versus deontic. It also offers a detailed discussion about how the syntactic structure of the Spanish subjunc tive works in nominal clauses and finally, compares the Spanish subjunctive to equivalent irrealis forms found in English. 1.2 Modal and Syntactic Explanations of the Spanish Subjunctive In this section, we discuss theoretical term inology and prio r research conducted on the Spanish subjunctive in order to provide the reader with the information needed to better comprehend both the purpose and findings of this study. To begin, we show that many researchers have taken on the monolithic topic of the Spanish subjunctive and that the majority
16 of these researchers have referred to the s ubjunctive as mood (Bla ke 1983, 1985; Collentine 1995, 2003; Givn 1994; Lubbers-Quesa da 1996, 1998; Faingold 1998, 2000; inter alia). Understanding the term mood is fundamental for comprehensive subjunctive analysis. It is defined in this study as a grammatical category th at is represented mor phosyntactically and used to indicate desire, influence, obligation, possibility, probability and certainty. Within this grammatical category of mood, however, exists a semantic subcategory that more accurately identifies the illocutionary point, or rather the attitude of the speaker, known as modality. Modality acutely describes the in tent of the speaker and her level of commitment to what is being uttered. Perhaps this is why many resear chers have chosen to study the subjunctive (Borganovo and Prevost 2003; Givn 1994; Lazano 1995; and Prez-Leroux 1998) by evaluating it in terms of opposition between epistemic (i.e., degree of truth, level of reality) and deontic (i.e., influence, desire indirect imperative, etc.) modalities. The terms epistemic and deontic modality create an umbrella under whic h all other dichotomies may be categorized. For example, the distinction between assertion vs non-assertion (Terrell and Hooper 1974; Goldin 1974; Majias-Vicandi 1994) dubita tive vs. optative/jussive (Foster 1982; Struderus 1995), and experience vs. non-experience all fall under episte mic modality (i.e., possibility, probability). Other connotations such as desire, influence, a nd obligation pertain to the category of deontic modality. These numerous and varied dichotomous re lationships show that (1) the subjunctive is part of a contrastive system that cannot exis t independently of the indicative mood; and (2) epistemic and deontic modalities are innate, modal properties that belong to specific lexemes. In the following paragraphs, we present a detailed analysis of the indicative/subjunctive modal contrast. We begin by stating that the indicative is used primaril y in affirmations to indicate high-certainty truth value or transmit informati on while the subjunctive is used to connote
17 influence, desire, and possibilit y. The indicative/subjunctive cont rast is necessarily broken down further into separate subcategories known as deontic and epistemic modalities under which, as mentioned previously, all other dichot omous relationships are categorized. 1.2.1 Evolution of Deontic Modality The following paragraphs look firstly at how the indicative mood, in its m ost basic syntactic and pragmatic form, can evolve into complex sentence structur es of deontic modality. We begin with simple indicative clauses in sent ences (1) and (2) and fini sh the analysis with compound structures using subjunctive verb conjugati ons in subordinate clauses in examples (5) and (6). The following sentences are examples of how the indicative mood is used in simple syntactic structures to transmit information (i.e., devoid of subjectivity) 1. Maya habla con su profesor. Maya speaks-3p-sing. with her professor. Maya speaks with her professor. 2. Maya vive en Espaa. Maya vive-3p-sing.in Spain. Maya lives in Spain. Examples (1) and (2) are strict ly factual, information-driven sentences. They are merely transmitting information from one interlocutor to another. In sentences (3) and (4) below, however, the reader is exposed to more pragmati cally complex structures that connote desire. These structures can be analy zed as belonging to the gramma tical, morphosyntactic category [+ indicative] and the semantic subcategory [+ deontic]. Sentences (3) and (4) do not require subjunctive morphology because there is only on e subject found in each sentence. The lexemes quiere (wants 3S) and prefiere (prefers 3S) do, however, possess deontic modality. 3. Maya quiere hablar con su profesor. Maya wants-3p-sing.to-speak with her professor. Maya wants to speak to her professor.
18 4. Maya prefiere vivir en Espaa. Maya prefers-3p-sing. to-live in Spain. Maya prefers to live in Spain. In opposition to examples (1) and (2), examples (3 ) and (4) show that the interlocutors are no longer simply engaged in an informative conve rsation but are actually connoting the subjects desire by using verbs such as quiere (want) and prefiere ( prefer ). In examples (5) and (6) below, the matrix verb of the main clause sele cts a subordinate, complementizer phrase (CP) as the direct object rather than an infinitival phras e (IP) as seen in structures (3) and (4). In the following examples, we see that each sentence contains a more complex syntactic structure containing two subjects: (S1) and (S2). Each of these examples also possesses the grammatical mood [+ subjunctive] and semantic modality [+ deontic]. 5. Maya quiere que pro hable con su profesor. Maya want-3p-sing. [that she speak-3 p-sing.-subj. with her professor.] Maya wants me to speak with her professor. 6. Maya prefiere que su familia viva en Espaa. Maya prefers-3p-sing. [that her fam ily live-3p-sing.-subj. in Spain.] Maya prefers that her family live in Spain. In these examples, the complexity of (5) and (6) is not only reflected in the pragmatic and syntactic construction of the sentence, but in th e morphology of the subordinate verb as well. In sentence (5) the verb hable (I speak), where hable cannot refer to S1, is conjugated in the subjunctive to reflect the influe ntial relationship of S1 over S 2. In sentence (6), the verb viva (to live) is conjugated in the subjunc tive to reflect the influence of S1 over S2. In both (5) and (6), the use of the subjunctive in the subordinate clause is a reflection of S1 influence. Therefore, in order to briefly summarize the examples analyzed in this section and the nature of deontic modality, we suggest that deontic modality, as manifested with in the subjunctive mood, follows three essential rules: (1) there are always tw o subjects which belong to a complex sentence
19 structure separated by the complementizer que (that); (2) S1 belongs to the matrix clause while S2 belongs to the subordinate clause; and (3) the matrix verb (V1) appear s in the indicative while the subordinate verb (V2) appears in the subjunct ive. In this next section we discuss epistemic modality. Our goal in the following section is to demonstrate why epistemic modality, as part of the dichotomous semantic system, is more complex than deontic modality as a result of its reliance on interlocutor inte rpretation and expression. 1.2.2 Ambiguity in Epistemic Modality Epistem ic modality is difficult to analyze because verbs used in these contexts are subjected to pragmatic and morphos yntactic changes that reflect vary ing levels of truth value as perceived (and manipulated) by the speaker. In other words, when truth value is high, the speaker uses the indicative mood in the subordinate clause to illustrate that she believes the information to be true; if the speaker employs the subjunctive in the second clause, sh e is suggesting that she is uncertain or doubtful about the information being transmitted. To complicate matters even further, lexical selection also becomes a factor when verbs like dudar (to doubt) and negar (to negate) are introduced alongside verbs such as creer (to believe) and pensar (to think). Once the negative lexical item no is included for example, epistemic modality can become seemingly unstructured. Therefore, verbs like dudar select the subjunctive in the subordinate clause when used without no. When such verbs are used with no, though, the indicative is required. The verbs creer and pensar work in the opposite way: when used without the negative lexeme no, they select the indicative in the subordinate clause, and when they appear after no, the subordinate clause requires the subjunctive. To better illustra te these nuances, we provide the following table 1-1. As seen in these examples (7a 10b) th ere are minor, yet important, morphosyntactic changes that correspond to the lexi cal selection of the principle ve rb. In (7a) for example, the
20 verb duda (he/she doubts) requires a CP complement whose verb must be conjugated in the subjunctive. Examples (10a) and (10b) work exactly the opposite of the verb dudar. The subjunctive is used when the negative lexical item no is inserted into the syntax before the verb. In (10a), the verb pensar (to think) selects a CP clause in which the verb appears in the indicative habla (he/she speaks). Example (10b), howev er, shows that the implementation of no requires the verb of the subordinate clause to appear in subjunctive hable (he/she speak). Table 1-1. Idiosyncratic behavior of epis temic verbs in Spanish subjunctive modality Use of indicative Use of subjunctive (7a) Juan no duda que Martn habla ingls. John does not doubt (is positive) that Martin Speaks English. (7b) Juan duda que Martn hable ingls. John doubts that Martin Speaks English. (8a) Juan no niega que Juan habla ingls. John does not negate that (agrees that) John speaks English. (8b) Juan niega que Martn hable ingls. John negates (does not agree) that Martin speaks English. (9a) Juan cree que Martn habla ingls. John believes that Martin speaks English. (9b) Juan no cree que Martn hable ingls. John does not believe that Martin speaks English. (10a) Juan piensa que Martn habla ingls. John thinks that Martin speaks English. (10b) Juan no piensa que Martin hable ingls. John does not think that Martin speaks English. Under the umbrella of epistemic modality, exam ples (7a) (10b) can be split further into two distinct subcategories: a ssertable and non-assertable info rmation. According to Lunn (1992) potentially assertable informa tion must have two qualities: in the speakers opinion, it must be both reliable as to truth value and informative as to new value. Information that is believed to be flawed in truth value or flawed in new value is unlikely to be asserted (4 30). In addition, if the interlocutors have no prior knowledge of the in formation being transmitted, they have had no time to form an opinion. Consequently, what is be ing transmitted is strictly information (i.e., use of indicative). Once the interloc utors have had the informati on for any amount of time though,
21 subjective opinions may form allowing the subjunc tive to be used as a rhetorical mechanism to discuss information rather than simply state it in an objective and informative manner. After having analyzed all of the examples in section 1.2, several broa d generalizations can be made about the Spanish language: (1) mood is a grammatical category with morphosyntactic representation under which the semantic subcate gories deontic and epistemic are accessed; (2) deontic and epistemic modalities are able to acco mmodate all other dichotomous analyses within the indicative/subjunctive, grammatical mood distinction; (3) le xical properties of V1 determine complement selection; (4) nominal clauses of low epistemic and deontic modality always select a CP complement with a V2 subjunctive; (5) subo rdinate clauses of deontic modality in Spanish are not affected by the inse rtion (or lack thereof) of the negative lexical item no because the nature of deontic modality always semantically, morphologically, pragmatically, and syntactically requires subjunctive. With the above generalizations that we have made regarding the Spanish subjunctive grammatical mood system and its modal subcategories, we are better equipped to apply these observations to the sy ntactic environment of nominal clauses. 1.2.3 Deontic and Epistemic Modality in Nominal Clauses It is nearly impossible to discuss the Spanish subjunctive without incorporating som e discussion of the syntactic structures that are used with it. Much of the research devoted to the syntactic structure of the Spanish subjunctive is organized in terms of clausal structures, including complement (also known as nominal), adverbial, and rela tive (Collentine 1995, 1998, 2003; Lubbers-Quesada 1998) clauses. This st udy focuses on the complexity of nominal subjunctive structures, given their ability to accommodate both deontic and epistemic modalities in subordinate clauses. The structure of the Spanish subjunctive in nominal clauses involves a complex syntactic structure that must include both a principle and subordinate clause The subordinate clause (also
22 known as a complement phrase or CP) acts as the dire ct object of the matrix, or principle, verb as demonstrated in the following example: 11. Elena quiere que Sofa estudie lingstica. Elena want-3p-sing. [that Sophia study-3p-sing.-subj linguistics] Elena wants Sophia to study Linguistics. Sentence (11) clearly shows th at the subordinate clause of the example in Spanish, beginning with the complementizer que (that), occupies the space of the direct object of the main verb. While the joining of two clauses may not seem very complex, the process is syntactically more intricate than it appears. On ce the learner has structur ed the sentences, she must now pay attention to the mo rphosyntactic requirements of the subjunctive. In other words, the morphology of the conjugated verbs found in the principle and subordinate clauses must reflect the desires or actions of the parties involved. For example, the verb querer (to want) in the main clause must appear in the indicative quiere reflecting that Elena wants something (V1 = fact/indicative). In the s ubordinate clause, the verb estudiar (to study) must be conjugated in the subjunctive estudie reflecting the influential and hypothe tical nature of the complement phrase (V2 = hypothetical/subjunctiv e). Other examples of deontic modality found in nominal clauses emerge when the matrix clause uses verbs such as sugerir (to suggest), recomendar (to recommend), aconsejar (to advise), desear (to desire), rogar (to beg), and requerer (to require), for example. Complications gene rally surface, however, when the learner is confronted with epistemic verbs such as pensar (to think), creer (to believe), dudar (to doubt), negar (to negate), suponer (to suppose), or ver (to see). Even more confusing are verbs like suponer that are allowed to employ either mood in the subordinate clause. Take the following sentences for example:
23 12. Alejandro supone que Marta toma clases en la Universidad. Alejandro supposes-3p-sing. [that Mart ha takes-3p-sing.-subj .classes at the university.] Alexander supposes that Martha takes classes at the university. In sentence (12), one can see that according to the speaker, Alexander highly suspects that Martha is taking classes at the university. However, the speaker may construct the same sentence using the subjunctive in the s ubordinate clause as demonstrated by the following example. 13. Alejandro supone que Marta to me clases en la Universidad. Alejandro supposes-3p-sing. [that Mart ha takes-3p-sing.-subj .classes at the university.] Alexander supposes that Marta takes classes in the university. In this case, the subjunctive is used to imply that it is very unlikely that Martha is taking classes at the university at this time (i.e., low truth value, low epistemic levels). For learners of Spanish as a foreign language, distinguishing be tween epistemic and deontic modality within the grammatical category of the subjunctive mood is ch allenging. Not only is this true because of the complexity of the structure itself, but because mood distinction is not regularly discussed in English. In order to better understand the depth of this problem, it is necessary to analyze the Spanish subjunctive alongside the English subjunctive and other ir realis structures. In doing so, we need to pay particular a ttention to the more common English equivalents to the Spanish subjunctive including infinitival structures and in dicative morphology in CP clauses. In the next section, we discuss the term irrealis in greater detail and demonstrate its importance in analyzing English examples in this study. We also provide examples of equivalent English and Spanish structures that better illustrate the modal-sy ntactic differences that exist between these two languages. 1.3 English Irrealis One of the most common problem s for English-speaking learners of Spanish is understanding what its equivalent structures are in English. This occurs because in English the
24 notion of irrealis modality is captured in a variety of different structures. In this project, we define irrealis as the difference between factua l and non-factual information. This definition is therefore much less restrictiv e than the grammatical term mood, as used in Spanish. Syntactically, the term irrealis is more flexible than the term mood because it can accommodate infinitival, gerund, and subordinate clause analyses. Semantically, the term irrealis is able to account for both epistemic and de ontic speaker attitudes. The first example provided in this section is sentence (14) below. As mentioned previously, even though the auth entic subjunctive mood in English is mostly obsolete, a few structures do remain. In fact, an authentic English subjunctive structure follows the same syntactic pattern as Spanish, though sentences with this structure tend to sound overly formal in Modern English. Take the following example. 14. It is necessary that John arrive tomorrow. While the sentence in (14) is grammatically correct, it sounds artificial in casual speech. The more frequent ways of transmitting the same message in colloquial English are given in (15): 15. (a) I need John to arrive tomorrow (b) John needs to arrive tomorrow. In both (15a) and (15b), the notions of S1 in fluence and desire are implied and represent more common uses of English irrealis at pr esent. Unlike Modern English, Old and Middle English used the subjunctive with much gr eater frequency. Over time, however, these subjunctive structures have been replaced with ot her forms such as infinitives and indicatives, for example. This is not true for Spanish. In Spanis h, the subjunctive is used regularly in colloquial speech so that sentences such as Quiero que vengas a mi casa (I want you to come to my house) and No creo que sea as (I dont believe that it is so ) are quite common in everyday
25 conversation. Consequently, L2 learners of Span ish, more familiar with English irrealis syntax, often try to either reduce complex sentences in Spanish to fit simpler English templates or use potentially null CPs with the indicative mood; a topi c which is discussed in greater detail in latter sections of this chapter. The following examples of deontic modality fu rther illustrate the complexities of the English irrealis system. The verb prefer in sentence (17) for exampl e, exemplifies the authentic subjunctive in terms of syntactic environment. The verbs hope and want in sentences (16) and (18) however, use non-subjunctive morphology, which as we stated earlier, falls under the notion of irrealis modalit y. The English verb hope like its Spanish equivalent espera selects a CP complement. In English, however, the subordinate verb emerges in the indicative; unlike its Spanish counterpart. 16. (a) I hope that he is going to the party. (b) I hope that he goes to the party. (c) *I hope that he go to the party. The verb prefer in English also selects a CP complement but always requires the use of the subjunctive as shown in example (17). 17. (a) They prefer that he go to work. (b)* They prefer that he goes to work. The verb want in English selects an IP complement. The use of any finite verb in this context would be incorrect. 18. (a) I want him to go to the store. (b) I want that he go/goes to the store. The above analysis illustrates the syntactic co mplexity of irrealis modality in English and the intricate differences that exist between th e structures. It also shows the optionality of that in English shedding light on the lack of shared sy stematic behaviors between many of these verbs
26 making it difficult to observe clear patterns. Based on the information presented here, it is fair to suggest that ambiguous differences that exist between the Spanish subjunctive mood and the English irrealis system are a combination of morphosyntactic and semantic features. Therefore, we provide a more in depth analysis of these distinctions to more clearly illustrate the idiosyncratic syntactic and semantic behavior s of these two languages. Moreover, for this analysis, it is paramount that the reader unders tand how such behaviors work as part of an intricate modal system which is, above all, driven by speaker attitude. As a result, in order to analyze these behaviors, we adopt the ideas of Generative Gra mmar (GG). GG is applicable to all languages allowing linguists not only to make generalizations about properties shared by all human languages, but also to observe their di stinct and individual patterns. The following section provides information regarding GG and more specifically, Principles and Parameters framework (P&P), which taken together, provide theoretical explanations of the semantic and syntactic differences that exist between the E nglish irrealis system and Spanish subjunctive mood. 1.4 Generative Grammar In the present study, we use a generative approa ch originally proposed by Chom sky in the late 1950s. Using a generative approach to gr ammar analysis, we attempt to identify a finite number of explicit rules that wi ll help us to predict and explai n interlanguage grammars during L2 subjunctive development in Spanish. A core prin ciple of Generative Gra mmar (GG) is that all sentences have structure. The Structure De pendency Principle (Cook 1988, p.6) indicates that operations on sentences such as movement requi re a knowledge of the st ructural relationships of the words rather than their linear sequence. Syntactic structures will be important in what follows because, by proposing interlanguage (IL) st ructures, we can gain insight into learners errors and provide theoretical expl anations as to why such errors occur. In this project, concepts
27 of the GG approach are used to help us explai n data collected by second language learners of Spanish. In the following section, we develop con cepts with the Principles and Parameters (P&P) framework of GG that will be important for the remainder of the work. 1.4.1. Principles and Parameters Structu ral similarities and differences acro ss languages are accounted for by a set of principles and parameters. Princi ples are invariant conditions that hold for all languages. Even though every language obeys these principles, all languages are not the same because there are parameters within the system and different lang uages can set these parameters differently. In other words, parameters are the language-specific options that re sult in cross-linguistic variation. One such parameter is the Head Parameter which determines the position of endocentric phrasal heads with respect to their complements. Langua ges can set this parameter differently. English places heads on the left as in drive a car where drive the head of the VP, appears on the left. Like English, Spanish is also a head-initial la nguage. Languages like Japane se are head-final or right headed and the head follows its complement. English and Spanish differ in another way, how ever. In Spanish, verb phrases such as conduce un coche (drive a car) do not require an overt subject and can function as a complete declarative sentence on thei r own. In English, the VP drive a car must have an overt subject as in I drive a car or Jon drives a car. The difference is a result of the Pro-Drop or Null Subject Parameter which allows the subject of a sentence to be a null pronoun, called pro. Despite this difference in allowing or not allowing a null pronominal subject, both languages adhere to the Extended Pr ojection Principle (EPP) which st ates that clauses must have subjects. In Spanish, pro can serve this function. Take the sentence Come pizza (She eats pizza) for example. This sentence is grammatical in Spanish. However, the subject is not overt and must therefore be pro in order for the EPP to be satisfied: pro come pizza In English, in contrast, the
28 Pro-Drop Parameter is set negatively and pro cannot be used to satisfy the EPP. Thus Eats pizza is ungrammatical; an overt subject, is required. The EPP is an extension of the Projection Principle, which says that lexical characteristics of individual words must be uniformly satisfie d throughout the syntax. In the English sentence She rides bicycles for example, the verb rides is transitive in that it must be followed by a direct object noun phrase (NP). The le xical entry for the verb rides in this case would be rides [ __NP] demonstrating that the lexical propert ies of this particular verb synt actically require that it have a direct object NP followi ng it. In the sentence Marcos quiere que Noelia vaya a la fiesta (Mark wants Noel to go to the party) for example, the grammar requires that quiere (he/she wants) select a complementizer phrase (CP) as the direct object. Within the subordinate clause, the verb must occur in the subjunctive as seen in vaya (go). In modern American English, one translation of this sentence is Mark wants No el to go to the party in which the verb wants selects an infinitive. The traditional analysis with GG is that this infinitive is an inflectional phrase (IP) rather than a CP as in Spanish. Both want and quiere are verbs of deontic modality as discussed previously. The difference is that distinct clausal complement type s are chosen as the direct object of the matrix verb in each language. In other words, the syntactic behavior of quiere and want shows that each lexeme occurs in a distinct syntactic environment. With this example, we have illustrated the critical role of the lexicon in syntax. In the following paragr aphs, we suggest that the underlying difficulty in L2 acquisition stems from a combination of semantic and unsystematic syntactic differences that exist between English and Spanis h. In order to formalize these differences, we use subcategorization frames in lexical entrie s. Before proceeding, however, we discuss the contents of a lexical entry.
29 According to Cowper (1992), a lexical entry contains four components: (1) phonological information (speaker must know how to pronounce the word in question); (2) morphological information (speaker must know the correct form of the verb); (3) semantic information (speaker needs to know what the word means); and finall y (4) syntactic information (speaker needs to know how to construct the predicate). In complia nce with these 4 essential steps, we use the following subcategorization frames to show how English irrealis and Spanish subjunctive structures compare and contra st both semantically and synt actically. The Spanish verbs esperar, querer, and prefer are compared to thei r English counterparts hope want and prefer We chose to analyze these verbs because they are common in the English irrealis and Spanish subjunctive mood systems. These verbs also provide a variety of structural differences that represent an assortment of constructions in these two languages. In syntax, there is no direct way to show which mood is requ ired in a subordinate clause using subcategorization frames because subcateg orization is local. As a result, mood selection must be demonstrated through a series of steps that are local1. Take the following example of the verb querer to want in Spanish: querer, V, [ __CP [sjv]] that, C[sjv], [ __IP [sjv]] I[sjv], [ __VP [sjv]] Note: sjv = subjunctive This series of lexical entries encode that querer want selects a subjunctive complementizer phrase (CP). The head of this CP C, in turn selects a subjunctive inflectional phrase (IP). The head of this IP, I in turn, selects a VP whose head is inflected with subjunctive morphology as seen in the above sub categorization frame for the verb querer to want. In 1 Information regarding complex subcategorization frames in the subjunctive were provided by Dr. Eric Potsdam through personal communication.
30 Spanish, all three verbs, querer, esperar and preferir possess the same subcategrization frames in subjunctive contexts. Unlike Spanish, however, the subcategorizati on frames of the equivalent English verbs (i.e., want hope and prefer ) are all different from each other and therefore behave idiosyncratically. We return to the verb want for example. want V, [IP [to] The verb want in English selects and IP (inflectional phrase) with the head to as a complement rather than a CP (complement phrase) followed by the subjunctive in the subordinate clause. In the sentence I want Jon to go to the store Jon, the subject of the subor dinate clause, is assigned accusative case by the main verb want rather than nominative Case by I. The verb hope in English, however, follows essentially the same structure as it would in Spanish. The difference being that hope, unlike esperar, requires the embedded verb to appear in the indicative rather than the subjunctive. hope V, [ __CP [ind]] that, C[ind], [ __IP [ind]] I[ind], [ __VP [ind]] Note: ind = indicative Out of the three English irrealis verbs identified in this section, prefer is the only matrix verb (V1) that selects a CP complement and requires that the s ubordinate verb (V2) appear in the subjunctive mood, as in I prefer that you be/are on time Prefer is one of the few verbs in English that uses the exact same structure as its Spanish equivalent. prefer V, [ __CP [sjv]] that, C[sjv], [ __IP [sjv]] I[sjv], [ __VP [sjv] Note: sjv = subjunctive
31 In the following section, we discuss phrase struct ure. In particular we discuss X-Bar Theory and how it is useful in helping us to create possible structural re presentations of IL development in Spanish subjunctive acquisition. By using X-Ba r theory, we are able to see the syntactic difference between L1 and L2 in structural representations and provide the reader with plausible representations of IL. 1.4.2 Syntactic Structures The following diagram s apply the above information to provide the reader with accurate L1 and L2 syntactic structures. In addition, the subcat egorization frames illustrate some intricate behaviors that have been discussed throughout this chapter which we apply here to corresponding diagrams in order to better illustrate these differen ces. The first figure, figure 1-1, shows the syntactic representation of the English irrealis sentence I want him to buy milk In English, V1 want selects an IP complement in which to is generated in the head of IP and the DP him is generated in the Spec IP pos ition of the subordinate clause. Him is assigned accusative case from the head of VP. The Span ish equivalent of this sentence ( I want him to buy milk ) however, does not follow the same pattern as in English. The sub category frame for the English verb want is as follows: want V, [ ___IP [to]]. The Spanish example, Quiero que pro compre leche, requires an additional functio nal category. The subordinate CP is required in Spanish for the overt realization of que. Quiero que l compre leche. quiero V, [ __CP [sjv]] que C[s]v], [ __IP [sjv]] [[s]v], [ __VP [sjv]].
32 Figure 1-1. A syntactic representation of I want him to buy milk. Figure 1-2. A syntactic representation of quiero que pro compre leche.
33 The most important difference between these tw o structures is the use of the functional category CP which has more recently become a fo cal point in L2 lexical development. As a result of this shift, it is suggested that learners must c ontrol the idiosyncratic behaviors of individual lexemes before L2 parametric values can be set and full acquisition can occur in a given language. For example, Herschensohn (2000) uses the analysis orig inally introduced in Speas (1990, p. 9) which shows the importance of lexical acquisition in L2. According to Speas, a learner needs to acquire the phonological, sema ntic, syntactic, and morp hological information of a lexeme in order to reach native-like prof iciency in the target la nguage. Although Speas uses the word eat in English, we apply his analysis to the verb espero (I hope) in Spanish. Therefore, in order for full acquisition of a given lexeme to take place, learners would need to acquire the following properties: (1 ) the phonological representation of espero is /espro/; (2) the meaning of espero is I wish; (3) syntactically this le xeme is categorized as [+V N], and finally, (4) morphologically espero refers to the first person singular (1PS) of the verb esperar (to wish). Notice that in our syntactic analysis of espero we show that this verb can select a CP as its complement. Such a selection is necessary in Spanish subjunctive contexts. In latter sections of this chapter, we discuss how th e incomplete acquisition of lexemic properties contributes to mismatched features in L2 and shed light on the importance of functional categories in L2 development. The acquisition of functional ca tegories (FCs) in language deve lopment is relevant to the present study as we will be looki ng at different theories regardi ng learners access to functional categories during L2 acquisition. In the following sections we discuss several possibilities regarding such access. Among thes e possibilities we include the following: (1) learners do not have access to FCs as proposed by the Failed Fu nctional Features Hypothesis (FFFH) presented
34 in Hawkins and Chan (1997); (2) learners gra dually develop IP and CP complements in L2 (Vainikka and Young-Scholten 1998) ; and finally (3) learners have access to IP and CP complements even though such features are not always overtly present in learner production. This is referred to as the Missing Surface In flection Hypothesis or MSIH (Prvost and White, 2000). We will argue in favor of the latter hypoth esis as our study supports the existence of FCs in L2. 1.4.3 Functional Categories Generative gramm ar makes a distinction between lexical cate gories and functional categories (FCs). To clarify the distinction betw een functional categories and lexical categories, we refer to Radford (1997) who claims that nouns verbs, adjectives, adverbs, and prepositions typically have descriptive content and so are contentives (98). In opposition to th ese categories, however, exist functors (i.e., functional categories) which Radfor d defines as words that have an essentially grammatical functi on (38). In other words, functors are words that exist within grammar but have no relation to th e outside world. Included in this list are determiners (ex: DP the), complementizers (ex: CP that ), infinitival particles (ex: IP to ), and auxiliaries (ex: do) (p.38). Learners access to such categories during SLA has been a topic of current research, according to Haznedar (2003), and are therefore discu ssed in greater detail in the present chapter. In the following paragraphs, we look at differe nt theories that have emerged over the years regarding FCs and their transferab ility in SLA. The first of thes e theories is called the Failed Functional Features Hypothesis (FFFH) which is defined by White (2003) as the claim that adult L2 learners are unable to acquire features of functional categories [Det, Infl, Comp] which differ from those realized in L1 (276). As a re sult of these lacking feat ures, some research has suggested that learners look for other ways to acquire the target language. The Failed Functional Features Hypothesis (FFFH), discussed in Hawkin s and Chan (1997), concluded that L2 learners
35 do not have access to functional f eatures. This conclusion is base d on their study that looked at the acquisition of CPs in L2 English of L1 Cantonese speakers with knowledge of standard written Chinese. The participants of this study were separated into thr ee specific groups that consisted of an elementary, intermediate, and an advanced group of learners. The majority of these learners had approximately six or more years of exposure to English, most of which took place in the classroom. Their principle assump tion was that L2 acquisition of L1 feature specifications would only be possible if such features were present (either the same or similar) in both languages. This being their main assumption, one of Hawkins and Chans main goals was to test whether or not the Chinese-speaking par ticipants had the ability to acquire overt morphophonological properties of complementizers in L2 English. They chose a grammaticality judgment test to collect the data as their interests did not include, as they refer to it, absolute performance (203). The items used in this gr ammatical judgment task were specifically designed to test the acquisition of CP morphol ogy in L2. The results indicated that beginning learners were much less accurate in their interp retation of CP use in L2 English (56%). The intermediate group showed 67% accuracy while the advanced group showed 79% accuracy. Hawkins and Chan concluded that as the Chines e speakers proficiency increased, so did their ability to acquire CP morphology in complement position. With this said, however, results from the items that tested learners unconscious know ledge of subjacency violations (i.e., limitations on movement) in English, showed a severe decrease from the elementary level to the advanced level. This led Hawkins and Chan to propose that even as proficiency in learners L2 continued to progress and become increasingly more English-like, their IL systems did not mirror those of L1 English grammars. Although their results may seem rather discouraging in this respect, Hawkins and Chan suggest that learners find other ways to achieve proficienc y in L2. In order to
36 compensate for FC deficiency in L2, Hawkins and Chan propose that learners do the following in order to advance in the target language: t hey [learners] will map morphophonological forms from the L2 on to L1 feature specifications. The performance of L2 learners with such grammars would display properties of a familiar kind: L1 s yntax with L2 lexical items (216). In other words, researchers should expect to see Englishlike structures with Spanish vocabulary as seen in figure 1-3. Therefore, based on the above anal ysis, we suggest that the following figure (1-3) provides the reader with a plausi ble representation of what lear ners IL might look like during L2 subjunctive acquisition. Figure 1-3. A syntactic representa tion of IL syntax during L2 subjunctive acquisition: IP complement quiero t venir In figure 1-3, we see that CP is missing from the structure. In theory, this missing CP does not affect the main clause. The problem exists in the subordinate clause because the verb
37 incorrectly selects for an infinitival IP rather th an the correct finite CP. The question mark in I represents the fact that we do not know exactly what is in I, a notion we return to in latter sections of this chapter. In figure 1-4 we use indicative morphology instead of the infinitive verb form. Figure 1-4 is still incorrect, however, beca use the functional category CP is missing. Both diagrams 1-3 and 1-4 represent common errors th at L2 learners of Spanish make during the acquisition process. The errors make sense if we accept the idea that learners do not have access to CPs in L2. The unavailability of CPs in L2 is highly unlikely, though, considering that CPs are available in L1 English. Figure 1-4. A syntactic representation of IL syntax dur ing subjunctive acquisition (* Quiero t vienes .) Structures 1-3 and 1-4 may be expected representations of IL during subjunctive acquisition if the reader accepts FFFH, for exam ple. Such representations are not the only options, however, as Vainikka and Young-Sc holten (1996) present a different view.
38 Vainikka and Young-Scholten ( 1996) propose that functional projections are gradually acquired, which they explain in several ways. First, Vainikka and YoungScholtens suggest that functional categories emerge as a result of a VP-b ased grammar in which L2 learners transfer lexical VPs from L1 to L2. In turn, these VPs serve as the foundation for learners initial L2 grammars. Second, Vainikka and Y oung-Scholten present the possibili ty of a gradual process in which learners build their grammars from th e bottom up. Included in this gradual process, however, is the appearance of unspecified Functiona l Phrases (FPs). FPs are thought to emerge at the IP level after the VP has been transferred fr om L1. The unspecified FP allows the verb to move even though phi features (i.e ., person, number, tense) do not match. From this perspective, we can explain why mismatched subject-verb agreement often occurs in L2 production, for example. To better illustrate this idea, we refer to Whites (2003) structural interpretation of this process. Figure 1-5. Unspecified FPs As discussed previously, and more clearly demo nstrated in figure 1-5, FPs emerge at the IP level. Problems occur when the unspecified FP f eatures do not match the properties of a given lexeme (see figure 1-3). According to Vainikka and Young-Scholten, however, eventually these FP projections evolve into IPs, ultimately giving way to CP comple ments. As a result, researchers should expect that CPs will be the last of the functi onal categories to be acquired.
39 Vainikka and Young-Scholten supp ort their theory with eviden ce from a longitudinal study that analyzed naturalistic data from adult L2 learners of German. The participants were L1 speakers of Korean, Turkish, Italian, and Spanish. The researchers used a wide variety of tasks to elicit oral production data; all of which were ta pe-recorded and analyze d. By the end of the study, over several hundred samples were collected from each speaker. The results of the data show that all learners had transferred L1 headedness properties from L1 to L2. For example, for Korean and Turkish speakers, L2 acquisition was faster because both lan guages, like German, possess head-final VPs. In other words, Korean and Turkish participants began acquiring L2 by transferring L1 VP properties to L2 with the understanding that German, like their native Korean and Turkish languages, was also a head fina l VP language. Unlike the Korean and Turkish participants, however, the Italian and Spanish par ticipants had more difficulty transferring such properties. Italian and Spanish l earners were required to switch their headedness parameter from head initial to head final. With this said, though, Vainikka and Young-Scholten claim that in early stages of L2 acquisition, the Turkish and Ko rean participants showed few signs of verb raising which they operationalized as a verb preceding VP mate rial (18). For the Spanish and Italian participants, verb ra ising depended on the position of negation phrases and temporal adverbs (See also Pollock 1989). Vainikka a nd Young-Scholten reported that the Turkish and Korean participants showed evidence of verb raising only 14% of the tim e while the Spanish and Italian speakers showed no indication of verb move ment at all which, as stated previously, is linked to access to FCs. For example, Bongiovanni the Italian participan t recruited for this study, produced the following sentence in his L2 German (18): 19. Nein en matina nix essen. No in morning [Ital.] not at INF I dont eat in the morning. (Nein, morgens esse ich nicht(s))
40 In this example, we see that the VP appear s after the negation phras e which indicates no sign of verb raising. Therefore, due to the overall lack of movement among participants of this study, Vainikka and Young-Scholten suggested that L2 learners did not have access to functional categories at the time of the study. In additi on, Vainikka and Young-Scholten claim that their data showed no signs of auxili aries, modals, agreement patterns, complementizers or WHmovement; all of which one would expect to notice if FCs were available during initial stages of L2. Epstein, Flynn, and Martohardjono (1996) also show support for the gradual development of functional projections disc ussed in Vainikka and YoungScholten. In this study, over 50 participants were recruited for a pr oject that required them to complete a series of imitation tasks in L2 which would measure learners use of IP and CP complements in their production. The participants included over 30 chil dren and approximately 18 adults. Overall, learners ability to imitate sentences with IP projections was approxi mately 20% better than their ability to imitate sentences with CP projections. This greater accur acy in IP production can be taken to mean that learners had not completely acqui red CPs at the time of the study. Thus far we have presented the reader with several theoretical options in relation to the role of FCs in L2. Among these options we have included FFFH (Hawkins and Chan 1997) and VP-based L1 transfer (Vainikka and Young-Sc holten 1996). In the following paragraphs, we present The Missing Surface Inflection Hypothesis (M ISH) which is supported by the data of the present study. MSIH (Prvost and White 2000) proposes that learners have knowledge of functional categories even though their production does not al ways reflect such competence. In addition, learners are thought to have full access to FCs ev en during the initial stag es of acquisition, unlike
41 the theories presented above. Prvost and Wh ite (2000, p.113) suggest that learners have problems with the realization of surface morphol ogy such that they resort to non-finite forms (see also Haznedar and Schwartz, 1997). Th is implies that even though accurate surface inflection is not always visible in L2 production, researchers should not take this to mean that learners lack syntactic competence and access to functional categories. In their study, Prvost and White conducted a longitudinal case study of four adults who were learning second languages. Of this group, two of the participan ts were learning French as a second language while the other two were learning German. After conducting a series of oral interviews with each participant during a one to two year period, Prvost and White found that even the more advanced learners of the group had the tendency to use non-finite forms during their oral interviews. In other words, contrary to what one might expect, learners use of non-finite verb forms was not limited to early stages of acquisiti on. In addition, they show that non-finite verb forms were raised in some cases. Such phenom ena was discovered when learners produced sentences in which a verb appeared before a negation phrase (as discussed in Vainikka and Scholten 1996). The V Neg order was considered a raising environment while the Neg V order was considered a true infinitive environment. Prvost and White show that there was systematic V Neg order in their study indicati ng early availability of FCs. Th is order is illustrated in the following example (117). 20. ich studiere nicht I study-1S not Prvost and White also found that learners use of non-finite verb forms demonstrated variability meaning that in some cases, an infi nitive verb would appear before the Neg phrase showing that learners were using infinitival forms in finite contexts. We will make reference to
42 these findings in latter chapters as our study su pports the conclusions reported in Prevost and White (2000). Herschensohn (2001) supports the findings of Prvost and White (2000). In this study, two American high school students served as partic ipants for the project. The data collection procedure consisted of a series of oral inte rviews over a six month period. The results of Herschensohn mirror those of Prvost and White (2000). The results show that even though both participants had progressed in their overall mo rphological ability through out the duration of the study, they both continued to use non-finite verb fo rms in finite contexts, even at more advanced stages of L2 development. Herschensohn concluded that the use of infinitival forms in SLA is rather accidental, incorrect infl ection that is a result of inte rlanguage processing difficulties (292). In addition, she suggests that infinitives in L2 are different from infinitival verb forms in L1 because, as shown previously in Prvost a nd White (2000), L2 learners use finite and nonfinite forms interchangeably in some cases. Haznedar (2003), like Herschensohn (2001) and Prvost and White (2000), advises against linking inaccurate surface morphology to the lack of functional features in L2. Hazneder's suggestion is based on her study of two particip ants. Haznedars study is similar to the two previous studies mentioned above in that it too was a longitudinal study that included a series of tape recorded interviews. Her approach was sligh tly different from the others, however, in that she focused on Case-markings to gage the morphol ogical and syntactic accuracy in L2. The first study was conducted on Erdem, a child with L1 Turkish who was learning English as a second language at the time of the study. According to Haznedar, Erdems ability to accurately assign nominative Case to subjects in his speech producti on indicated that IPs must have been available for feature checking. In addition, sh e adds that Erdems use of CPs was also fairly productive.
43 In her second study, Haznedar collected oral production data from an American student named John who was learning Turkish as a second language at the time of the study. The data was gathered over a period of 5 months in whic h each session lasted anywhere from 60 to 90 minutes. Like Erdem, John was successful in assigning nominative Case to subjects, again showing support for the availability of IP in L2. John was not as successful with other types of Case, however. The error rate for the Case marki ngs of accusative, dative, locative, and ablative Case were especially high while genitive case was considerably lower. The results of Haznedars studies show evidence agains t Vainikka and Young-Scholtens VP-based proposal that IL grammars do not have access to FCs during initial stages of L2 acquisition. They also present ev idence against Hawkins and Chan (1997) who suggest that FCs do not exist in L2. Instead, Hazed ar shows that the L1 Turkish speaker (Erdem) had very few Case errors meaning that feature checking in IP must have been available at the time of the study. More specifically, she suggests that if IPs had not been availa ble to the learner, he would have committed more nominative Case errors in his production. The data revealed the opposite, however, showing that this par ticipant (Erdem) had almost no errors of this type. Although in Johns production, accusative, da tive, locative, and ablative Case assignment was highly inaccurate, nominative Case assignment was highly accurate. Such findings suggest that IP must have been available at the time of the study, as we saw with Erdem. Otherwise, John would not have used nominative case successfully to the extent that he did. In other words, there was enough evidence in these two studies to support the existence of FC s in L2. Based on her studies, Haznedar concluded that the l ack of functional elements should not be taken as evidence for the absence of the associated functional categories ( 148). Such findings are consistent with those of
44 Prvost and White (2000), and Herschensohn (20 01), and add additional support for the present research. The next article we discuss in this secti on comes from Sorace (2000). She claims that the MSIH hypothesis predicts the following in second language acquisition: (1) finite forms are truly finite and thus occur only in finite cont exts whereas (2) non-finite forms are sometimes genuinely non-finite and sometimes used as a defa ult substitute for finite forms (99). Sorace offers a very clear picture of the MSIH perspective and the goals of this theory. In addition she claims that: learners knowledge of the syntactic conse quences of finiteness seems to be in place despite the frequent surfacing of non-finite form s in finite contexts. This indicates that a clear distinction should be made between knowledge of abstract features, which is unimpaired, and knowledge of surface morphology, which can remain problematic in L2 development (99). In other words, as suggested previously, the theo ry claims that L2 learners have access to functional categories, like CPs for example, even though such access is often not visible in L2 production. To better illustrate the concepts presented in Prvost and White (2000), Herschensohn (2001), and Haznedar (2003), we pr esent the reader with figure 1-6. In this diagram, we insert a null CP complement to repr esent IL structures. Figur e 1-6 is the predicted IL representation for the sentence s found in figures 1-3 and 1-4. We present figure 1-6 as an expected repres entation of IL during s ubjunctive acquisition based on the studies mentioned here. In the pr esent study, we agree with Prvost and White (2000), Herschensohn (2001) and Haznedar (200 3) that the lack of overt morphological production does not indicate a lack of functional categories in L 2. Such a syntactic deficiency does not make sense considering that learners ha ve access to CPs (and FCs in general) in their L1. The studies in this section have shed light on the relationship between syntax and morphology (or lack thereof) and provided a variet y of ways in which FCs are acquired in L2.
45 Figure 1-6. A syntactic representation of IL syntax dur ing subjunctive acquisition (* Quiero t vienes. ) We argue in favor of MSIH as the findings of the present study support this hypothesis and will be discussed in greater detail in chapter 4. Thus far, our analysis has in cluded a discussion of mood, wh ich has presented the reader with some of the syntactic and modal issues that L2 learners of Spanish encounter when acquiring the subjunctive. We have also argued in favor of FC st udies such as those of Prvost and White (2000), Sorace (2000), Herschensohn (2001), and Haznedar (2003), that provide information about learners access to FCs in L2. It is at this point in our analysis that we raise the following question: if learners have access to FCs in L2, as we have argued here, and surface morphology does not always represent such access, th en is it possible that L1 transfer plays a role in learners L2 subjunctive development? The studies discussed in this section have already
46 identified problems in L2 including overt mo rphological production and accuracy of complex syntax. With this said, however, we have not addressed the deeper issue as to why such phenomena occur which is why we propose L1 tran sfer. MSIH provides support for the idea that learners have access to structures already acquired in their L1. Therefore, if L1 transfer is indeed at play in L2, we may be able to explain IL grammars and pr ovide support for our structural representations. In order to do so though, we need further evidence that L1 transfer is a viable option for our analysis. Based on this necessity, we discuss L1 transfer in greater detail and look for further evidence of L1 syntactic transfer in previous L2 studies in the following section. 1.4.4 L1 Transfer Ellis (1994) m entions two type s of transfer in language acquisition. The first type is referred to as borrowing transfer (where the L2 influences L1) while the second is referred to as substratum transfer (see p. 310). Borrowing tr ansfer occurs when L2 influences L1 and substratum transfer occurs when L1 influences L2. For the purposes of our study, we focus strictly on substratum transfer in which a learne rs L1 affects L2 development. Before presenting these studies, however, in the paragraphs below, we discuss some of the characteristics of L1 transfer based on Zobl (1980, p.472). Such characteri stics have helped us identify the existence of L1 transfer in the current study and also in previous research. Based on previous studies, Zobl (1980) compile s a series of characteristics and error types that can be used to distinguish L1 transfer from other developmental issues. Among these he includes the following factors: (1) er rors reflect learners use of L1 as a crutch at low level of L2 proficiency; (2) errors reflect use of L1 as a source of hypotheses about L2; (3) errors reflect inability to separate the two languages; (4) errors result from L1 habits; and (5) errors represent an interlingual genera lization. We use these characte ristics as indications of L1
47 transfer to analyze the studies me ntioned here and again in latter chapters to discuss the results of our study. White (1992) conducted a study that looked at L2 parameter sett ing of native speakers of French with L2 English. White explains that in French, finite verbs raise past the negation phrase (NegP) which she illustrates us ing examples such as the one seen in the following example. 21. Jean naime pas Marie. Jean no like-3p-sing.neg marie John doesnt like Marie. Verb raising in French can also be seen in sentences that include adverb phrases (AdvP) in which the adverb is also post-verbal. If we look at the sentence found in example (23), Jean regarde souvent la television (John often watches television ) we notice that the adverb souvent (often) appears after the verb regarde (watches). 22. Jean regarde souvent la television. Jean watches-3p-sing. often television. John often watches television. In English, however, adverbs a nd negation phrases do not appear after thematic verbs as shown in the translations of the French examples in examples (22) and (23). A native speaker of English would not say *John likes not Marie or John watches frequently television, for example. White suggests that if learners raised verbs in L2 English over NegP or AdvP, then this would serve as evidence of possible L1 transfer in L2 acquisition. In order to carry out her study, she recruited a total of 97 participants which were separated into two distinct groups. The larger of the tw o groups received instruction regarding question formation while the other learned about adve rb placement in English. White included both a preference task, used to measure learners comp etence, and an oral production task, used to measure learners syntactic ability in L2. Both tasks included negation phrases (NegP) and
48 adverb phrases (AdvP). It was s uggested that if learners raised thematic verbs over NegP or AdvP in L2, learners were transferring L1 parametr ic values of French to L2 English. The data revealed that the learners were successfully able to formulate questions in L2 English meaning that they knew English verbs di d not raise. This means that learners were aware that long movement from the nucleus of VP to the nucleus of CP was prohibited in English. With this said, however, learners did accept optional short movement of the verb past the adverb in L2. Short movement such as this shows evid ence of L1 transfer of French to English considering that verbs in English do not move. White (1985) looks at the appl ication of the Pro-Drop Parameter of L1 Spanish speakers learning English as a second language. Accordi ng to White, three syntactic properties are associated with this parameter: (1) subject-verb inversion in declarative sentence structures as seen in Canta Paco (Paco sings); (2) that-trace effects as in Quin dijo que lleg tarde ? (Who said that he arrived late); and (3) missing subjects and subject pronouns as seen in Conduce bien (he/she drives well). He r study looked at how frequently nativ e speakers of Spanish and French learning English as a second language identified such properties as acceptable or unacceptable in L2. In her study, the Spanish group was considered the experimental group while the French group was added as the control group. The French group was used as the control group because French, unlike Spanish, does not allow for missing subjects and is therefore not a Pro-Drop language. From this perspective, French and English share more properties than English and Spanish. Consequently, the results of the Spanis h group were compared to those of the French control group. In doing so, White wanted to observe whether or not the application of L1 Spanish parametric values in L2 ultimately caused transfer errors.
49 In order to collect data for her study, White used a grammaticality judgment test which was distributed to a total of sevent y three adult learners of ESL. Fifty four of the participants belonged to the native Spanish speaking group while the remaining nineteen made up the French control group. Both groups were broken down further in to five distinct pr oficiency levels. Level 1 was considered a low proficiency group (i.e., beginners) and Level 5 was referred to as the high proficiency group (i.e., advanced learners). The results of he r study show that some of the properties of the Pro-Drop Parameter transferred fr om L1 to L2 while others did not appear to carry over. The results between the Spanish and French group for the missing subject property of the Pro-Drop Parameter was significant. The data show that native speakers of Spanish learning English at the beginning levels accepted sentences with missing subjects more frequently than those at more advanced proficiency levels. Th e French group, on the other hand, even at the beginning levels, rarely accepted sentences with missing subjects in English as grammatical. Based on these findings, White ul timately concluded that rese tting L1 parameters in L2 contribute to transfer errors. Eubank (1994) conducted an empirical study whic h suggests that L1 lexical and functional projections transfer to L2. It is proposed in hi s analysis that headedness properties associated with such functional projections also transfer from one language to another while strength and agreement features do not. In other words, not all properties are transferable from L1 to L2. In order to make this claim, Eubank bases his di scussion on data compiled from previous studies including Pollock (1989), Belleti (1990), and Ch omsky (1991) among others. In his analysis, he looks at the L2 acquisition of English by L1 Fren ch learners. We have already mentioned that English and French both require overt subjects. French and English are different, however, in terms of feature strength of V. As in Spanish, verbs raise in French as a result of strong features.
50 Using Pollocks Split Infl hypothesis, Eubank analy zes verb raising in French. He suggests that French verbs undergo long movement meaning that finite verbs originate in V (under VP), move to T (under TP), and ultimately land in Agr (unde r AgrP). In French, nonfinite verbs may also undergo short movement from V (under VP) to T (under TP). They do not, however, move to Agr (under AgrP). In his analysis, Eubank suggests, that L1 French learners of L2 English have interlanguage systems that possess neither the str ong agreement features of their L1 nor the weak agreement features of their L2 at beginning st ages of acquisition. This underspecification of features may explain why learners apply an optional verb raising rule in L2 creating ungrammatical structures in English. Such struct ures include learners pr eference for non-finite verb forms in declarative sentences, as predic ted by Eubank, which he later supports with data collected in Gerbault (1978). The final study we mention in this section, Collentine (1995), deals exclusively with the Spanish subjunctive and the complex syntactic stru ctures that accompany this mood. We review this study only briefly here as it is discussed in greater detail in the fo llowing chapter. This study is useful in showing how L1 affects both m odal and syntactic development in L2 Spanish. Collentine (1995) looks at intermediate-level l earners mood selection in nominal clauses. The results of his study show that part icipants relied on L1 syntax in L2 production tasks. This was based on observations that even though learners attempted to form complex structures using the Spanish subjunctive in their L2, they actually only produced a seri es of simple sentences that were mostly attached by commas. Morphological errors were also abundant showing that learners had the tendency to use indicative morphology rather than subjunctive morphology. Collentines study, like the previous studies ment ioned by White and Eubank, show that learners have the tendency to transfer L1 properties over to L2 fitting Ellis explanation of L1 transfer.
51 1.5 Summary of Goals One of the goals of this study is to provide the reader plausible IL representations that more clea rly illustrate the developmental issues L2 learners face when acquiring the subjunctive. On the following pages, we provide the reader with a total of five diagrams. The first two diagrams represent common Englis h irrealis and Spanish subjunc tive structures used by native speakers of each language. The latter structures, however, serve as representations of IL systems during subjunctive acquisition. The syntactic e rrors shown in these diagrams include V1 selection of an IP complement with V2 [-finite], V1 selection of an IP complement with V2 [+ indicative], and finally, V1 select ion of a null CP complement with V2 [+ indicative]. Logically, other syntactic representations can be fathomed. For example, V1 selection of null CP complements with V2 [+ subjunctive] or V1 selection of IP complements with V2 [+ subjunctive] are also possible re presentations. The diagrams we have chosen to analyze here, however, are more relevant to the current study. On the following pages, we compare the latter three diagrams to the first two in the series so that the reader can identify the errors and processes that we will be referring to in the following chapters. We begin this discussion by analyzing figure 1-7. In figure 1-7 we look at the correctly structured Spanish example of Quiero que pro bailes (I want you to dance). In this structure, we observe that the ma in clause consists of one subj ect and one verb (S1 + V1). V1 ( quiero ) appears with indicat ive morphology and selects a CP complement. In the subordinate clause we find S2 ( pro) and V2 (bailes) in which the subordinate verb appears in the subjunctive mood. Both the subject of the principle and subordi nate clauses receive nominative Case in this deontic, nominal structure.
52 Figure 1-7. A syntactic re presentation of Spanis h subjunctive syntax ( Quiero que bailes. ) In figure 1-8, the English equivalent to Quiero que bailes (I want you to dance), we see a different structure. In the English version V1 selects an IP complement rather than a CP complement. Figure 1-8. A syntactic repr esentation of L1 Eng lish irrealis syntax ( I want you to dance ).
53 Now that we have presented a few examples of grammatical structures in L1 and L2, we turn to plausible representations of IL. Based on the previous studi es mentioned in this chapter, we suggest that the following IL representati ons are highly accurate Recall that we support MSIH and L1 transfer. Based on these hypotheses, we discuss IL structures that include both IP and CP complement structures. We later argue in favor of structur es that include CP complements which we attempt to support in chap ters 4 and 5 with the results of our study. In figure 1-9, for example, we notice that V1 espero I hope selects an IP complement under which V2 appears in the infinitive ( venir ). We need to ask ourselves whether or not learners actually do have access to FCs as discus sed in previous studies (see section 1.4.3). The reader should notice that we have placed a questi on mark in the head of IP in the subordinate clause of figure 1-9. The question mark represents the fact that we do not know exactly what is in I. In other words, we dont know if learners perceive the verb as [+ finite] or [-finite]. Figure 1-9. A syntactic represen tation of IL syntax during L2 subjunctive acquisition (* Espero t venir *I hope you to come)
54 Figure 1-10 provides the reader w ith nearly the same structure as 1-9 but with one very important difference. In figure 110, we consider an alternative representation that includes a null CP. In this representation, we leave the verb venir (to come) in the infinitive based on the analysis given in Prvost and White (2000). Recall that Prvost and Wh ite suggest that the absence of surface inflection (i.e., infinitival verb forms) does not mean that learners are actually using the non-finite form in non-finite contexts. Rather, non-finite verb forms are being used as default forms exhibiting properties of finite verb s (109). This means that in some instances, learners are using infinitive stru ctures as they would finite verbs. The point here, however, is not that this structure will result in an ungrammatical surface construc tion. Instead, the purpose is to consider the possibility that learners use infinitival verb forms in finite contexts. Figure 1-10. A syntactic representation of IL syntax during L2 subjunctive acquisition (* Espero t venir *I hope you to come)
55 Figure 1-11 is another example of errors that we suggest occur in learners IL systems. The reader should notice, however, that this struct ure is formed exactly lik e an English irrealis structure that would be used with the verb hope for example. As discussed previously, the verb hope selects a CP complement in which the overt appearance of that is optional as seen in I hope (that) he comes to the party. We also notice that in this struct ure, the indicative mood is used in the subordinate clause. In addition, this struct ure shows that the lear ner is applying the [+ optional] overt CP rule in English to the ta rget language. Regardless of production errors, though, we strongly suggest that lear ners have access to FCs in L2, even if they are not overtly available in performance tasks. Figure 1-11. A syntactic representation of IL syntax during L2 subjunctive acquisition (* Espero t vienes I hope you come)
56 The figures shown in this section compare accu rate syntactic structures in Spanish and English to representations of IL during L2 subjunctive acquisition. We also present these figures as a way to follow L2 processes and show areas of potential difficulty in L2 development. For example, it has been argued in previous studies that surface inflection is an indication that functional categories are available in interlanguage systems and th at learners have unconscious knowledge of functional categories (Prvost and White 2000, p.103). In addition, Prvost and White suggest that even though adult second lang uage learners have problems including overt FCs in surface structures, this does not mean th at their IL systems lack such categories. The diagrams we have provided in this chapter ta ke such possibilities in to consideration as well as provide representations of other error types found in L2. 1.6 Conclusion After having identified the possible sem antic and structural difficulties between the Spanish subjunctive and its English irrealis equivalents, the reader was presented with a detailed analysis of Spanish and English modal systems. Our investigation is useful to researchers because it applies theoretical info rmation to the acquisition of co mplex syntax (i.e., S1, V1 + S2, V2) using intuition data in second language acquisition. It also clarifies difficult terminology and combines the grammatical notion of mood with the semantic notion of modality simultaneously to systematize and organize structural contrasts that exist between English and Spanish. Furthermore, it explores the structures us ed by English speakers acquiring the Spanish subjunctive and follows learners evolving IL systems using both theoretical explanations and concrete examples to explain specific phenomena in L2 subjunctive development. Moreover, L2 Spanish instructors will benefit fr om this research and from the examples presented in this study because many of them do not know which Englis h structures are comparable to the Spanish subjunctive. As a result, instructors often misinform their students by telling them that the
57 subjunctive simply doesnt exist in English, wh ich is untrue, as demonstrated repeatedly throughout the literature. In addition, this chapter has provided an overview of previous explanations of the Spanish subjunctive and ha s identified three fundame ntal concepts that dominate this literature: mood (i ndicative vs. subjunctive), modality (deontic vs. epistemic), and syntax (IP vs. CP). The following chapters address the difficulty of L2 Spanish subjunctive acquisition. Chapter 2 discusses previous work conducted on L2 acquisition of subjunctive structures. Chapter 3 describes the methodology and data colle ction procedures followed for this study. The findings and results are discusse d in Chapter 4. Chapter 5 briefl y summarizes our findings and presents implications for the classroom, as we ll as suggestions for future research. We remind the reader that we have chosen a generative approach for this study. Such an approach allows us to analyze human language as a general system with unique, language-specific options that explain why not all languages share the same surface structure. In addition, P&P framework along with X-Bar Theory has provided us with a solid foundation from which to analyze our findings. By using X-Bar Theory, we were ab le to consistently propose plausible IL representations allowing us to make assump tions about learners syntactic intuitions and therefore identify problems learne rs face during L2 acquisition.
58 CHAPTER 2 ACQUISITION OF THE SUBJUNTIVE 2.1 Introduction As discussed in Chapter 1, the Spanish subjunctive is am ong the most challenging grammatical concepts of Spanish grammar for L1 speakers of English to acquire (Collentine 1997, 1998; DiFranco 2005; Farely 2004; Gume stad 2006; Jelsinki 1977; Lubbers-Quesada 1996, 1998; Terrel & Hooper 1974; Terrell, Baycroft a nd Perrone 1987). As a way to facilitate the learning process and reduce the amount of information a learner is exposed to at a given time, textbooks often explain the Spanish subjunctive th rough syntactic analyses that are divided into three distinct clausal types: nominal, adverbial, or adjectival structures. Some studies also conduct investigations using clausa l distinctions (Blake 1985) while others incorporate specific Second Language Acquisition (SLA) theories or pedagogical techniques, ranging from Generative Grammar (Cook 1988; Pollock 1989; Wh ite 2003) to Processing Instruction (Farley 2001, 2004; Collentine 1998). The goals of this chapter are as follows: (1) to show similarities in first and second language modal and syntactic development; (2) to discuss relevant studies that identify problems facing second language learne rs of Spanish when acquiring epistemic and deontic modalities within the subjunctive mood; a nd (3) to briefly review the role of functional categories in L2. Certainly one might ask why we have chosen to mention L1 studies of the Spanish subjunctive in our analysis. To answer th is question, we suggest that L1 studies show similar developmental patterns to th ose of L2 studies. With this said, however, our goal is not to prove that L2 learners follow the same patterns as L1 learners. Rather, we highlight studies that emphasize subjunctive complexity drawing parallel s between L1 and L2 de velopmental patterns.
59 2.2 L1 and L2 Acquisition Patterns There are a num ber of reasons why learners make errors in L2. Floyd (1990) and Faingold (1998, 2000) for example, address subjunctive phe nomena that occurs mainly in Mexico and Spanish-speaking areas of the United States. Ba sed on previous work, the purpose of Faingold (1998) was to analyze the development of synt actic processes in rela tion to mood in first language acquisition, language learni ng, history of language, and language change in progress. In order to carry out his study, Faingold compiled a multitude of data retrieved from morphological, phonological, psycholinguistic, sociolinguistic, and historical studie s from his own work. He also included research provided by authors such as Blake (1983), Goldin (1974), Pishwa (1989, 1991), and Wode (1983); inter alia Using the data taken from th ese projects, Faingold analyzed the modal and syntactic development among Span ish speakers in the United States, Mexico, Argentina, and Spain. The compiled data show th at Mexican children ofte n used the indicative instead of the subjunctive in oral speech; an error often committed by L2 learners of Spanish as discussed in chapter 1. Faingold (1998) is supported by Lazano (1995) who suggests that it is not until late childhood that a Spanish speaki ng child utilizes the subjunc tive of doubt and not until adolescence, that the difference between a dogma tic belief and a skeptical doubt are sorted out (98). In this study, Lazano shows that both L1 and L2 learners of Spanish either left out or substituted subjunctive structures with indicative ones. Lazano was also able to identify several important language patterns shared by both L1 an d L2 learners of Spanish, further demonstrating that accurate subjunctive use requires time to develop in both L1 and L2 language. These patterns are briefly summarized and more clearly illustrated in Table 2-1.
60 Table 2-1. L1/L2 parallel subj unctive acquisition patterns L1 subjunctive acquisition patterns L2 subjunctive acquisition patterns Native Spanish-speaking children acquire the indicative mood before the subjunctive mood Adult L2 learners of Spanish acquire the indicative mood before the subjunctive mood Native Spanish-speaking children have the tendency to substitute the subjunctive with the indicative Subjunctive neutralization: Adult L2 learners of Spanish have the tendency to substitu te the subjunctive with the indicative Subjunctive neutralization: Native Spanish-speaking children use the indicative or the infinitive with greater frequency Adult L2 learners of Spanish use the indicative (including future indicative) and infinitive with greater frequency Table 2-1, based on the data and results documented in Faingold (1998), provides the reader with a brief illustration and summary of shared L1 and L2 subjunctive acquisition patterns suggesting that both L1 and L2 l earners follow similar developmen tal patterns. According to the studies we have discussed here, L1 and L2 learners show an overall greater use of infinitives and indicative verbs. Blake (1983) analyzed L1 acquisi tion of the Spanish subjunctive in all of the clausal environments such as deontic-compl ement, adverbial, relative, and epistemiccomplement structures. He found that the epis temic modality in complement clauses of doubt, attitude, and assertion tended to be the most probl ematic for L1 learners. According to Blake, epistemic modality in relation to verbs of doubt a nd attitude do not completely develop until later in the acquisition process. He goes on to suggest that indirect commands (deontic-complement), adverbial, and adjectival clauses appeared to be less difficult. Conse quently, the only clausal environments in which participants demonstrat ed adult-like comprehension and used by the age 10 were adverbial and deontic-complement clauses. Collentine (2003) for example, also suggests that Spanish speakers do not master mood sel ection until adolescence, when sociolinguistic factors pressure them to conform to linguistic norms (75). Collentines analysis was based on a
61 study conducted by Gili & Gaya (1972) that anal yzed first language acquisition of the Spanish subjunctive by preschool childr en from Mexico and Puerto Ri co. The results of this study showed that children confronted the subjunctive by employing seman tic strategies that basically limited their options to deontic modality (as in querer que, want that) and adverbial phrases that made reference to the future (as in cuando hable when I speak). Taken together, these L1 studies collectively sustain the idea that first la nguage acquisition of the Spanish subjunctive is difficult even for native Spanish speakers. Such claims are based on findings that show Spanishspeaking children continue to overuse and substitute the subjunctive with infinitival or indicative morphology until latter stages of acquisition. To briefly summarize, this section has id entified several important problems that L1 learners of Spanish face during subjunctive acquis ition resulting in the further identification of crucial issues that impede L1 (and some areas of L2) subjunctive acquisition. Such issues include the following: (1) substitution of subjunctive mood with indicative mood or infinitive verb forms; (2) inability to master epistemic modality in early development; and (3) limited syntax. Moreover, as mentioned previously in this chapte r, native speakers of Spanish generally acquire epistemic modality during the latter stages of la nguage development meaning that the indicative and infinitive forms are obviously acquired be fore subjunctive mood. Native English speaking learners of Spanish as a second language, lik e native Spanish speakers, also acquire the subjunctive later in the developmental process. Even though the subjunctive is taught early on in elementary Spanish, usually within the first or second semester, L2 learners have difficulty acquiring the Spanish subjunctive even at the more advanced levels of instruction. However, after learning how difficult it is for L1 learne rs of Spanish to acquire subjunctive mood, this challenge should come as no surprise to the reader. The studies discussed in the following
62 sections address some of the theoretical and instru ctional issues that L2 learners are confronted with when acquiring the Spanish subjunctive. Our study provides evidence for L1 transfer which is discussed in greater detail in chapter 4. In the next sec tion, we focus exclusively on L2 subjunctive acquisition studies. 2.3 L2 Acquisition of the Spanish Subjunctive The problem that secon d language learners f ace when acquiring the Spanish subjunctive is two-fold: the first problem being syntactic (i.e., complex clauses) and the second being semantic (i.e., deontic vs. epistemic). In order to expl ain such phenomena, we refer to concepts of Generative Grammar and various subtheories based on this a pproach. We have found this approach to be very useful in inves tigating L2 subjunctive development. The Role of Failed Functional Features Hypothesis (FFFH) in SLA: In this section we briefly return to, and summarize se veral studies already discussed in chapter 1 in order to make them more relevant to the pr esent study. In doing so, we demo nstrate how functional categories such as IP and CP are partially responsible for parametric differences that exist between English irrealis and Spanish subjunctive syntax. In chapter 1, we intr oduced several hypotheses that address L2 acquisition of FCs. In addition, we pr esented the reader with studies that had been conducted on this issue in L2 acquisiti on. Vainikka and Young-Scholten (1994, 1996) for example conducted a study using production data which illustrated learne rs inability to use functional categories in oral speech. Recall that this inability was interpreted as an indication that L2 learners were unable to access functors in their IL. As a resu lt, they suggest that in L2 acquisition, only lexical words tr ansferred from L1 while function words did not. They also claimed that during the first stag es of L2 acquisition, learners IL included non-specific FPs (i.e.,
63 functional projections) which later evolved into more specific AgrPs1 (micro-projections within IP) and CP projections (i.e., FP IP CP). In agreement with the studies mentione d in chapter 1, Radford (1990, 1995) made comparisons between L1 and L2 acquisition of FC s. Radford claimed that in his study of first language acquisition, children appear ed to lack both IP and CP functional categories. He also suggested that the acquisition of L1 functiona l categories occurred in the following order: VP IP [finite verb] CP[+ finite verb]. In other words, L1 learners acquire the VP [+ lexical] first and later IP and CP projections become av ailable. VPs in L2, however, select unspecified functional projections (FP) which later assume other forms such as IP and CP projections with increased exposure to the target language. If Radf ords theory is accurate, then L2 learners have an additional step in their development. In other words, rather than VP selecting an IP or CP directly, learners pass through a de velopmental stage in which unsp ecified FPs appear before IP and CP complements. As stated in chapter 1, we do not agree with idea that FCs ar e unavailable in L2. Rather, we support the findings of Prvost and White (2000), Herschensohn (2001), and Haznedar (2003) who suggest that L2 produc tion does not reflect lack of syntactic competence and FC accessibility. Instead, production data is often in complete meaning that surface morphology is not always available. In keeping with this idea, the following studies exem plify the limitations of production data which raise concerns about premature L2 production. 2.4 Production Studies Authors such as Terrell, Baycroft and Perrone (1987), Collentine (1995) and Stokes, Krashen, and Kartchner (1998) have conducted research using me thodologies that include oral, 1 Pollock (1989) separates IP into two microprojections that include Agr P and T (i.e., agreement features and tense features.
64 written, and input-based tasks. Each of these studie s reported having conc erns about requiring students to produce subjunctive structures premat urely due to underdeveloped syntax in L2. In addition, two of these studies show evidence of L1 syntactic transfer which, as mentioned previously, may contribute to the sy ntactic simplicity in L2 production. Terrell, Baycroft, and Perrone (1987) suggest that the subjunctive should be introduced later in L2 development of Spanish. In their study, they show that learners require time to learn the subjunctive and use it in or al speech. They claim that le arners do not learn how to use complex clauses until the latter part of a begi nning or intermediate course. In this study, the researchers investigate learners interlanguage during first year Spanish in order to analyze learner strategies for using the subjunctive orally. They also look for evidence of comprehension, or lack there of, during conversat ion. In order to ca rry out this study, the researchers recruited students from the University of California, all of whom had studied Spanish for at least two years. The class that the participants were en rolled in normally included an explicit presentation of the grammar point being covered. The presen tation was usually followed by practice exercises that included both oral and/or written activities whic h were later applied in conversation. Prior to conducting their interviews, the researchers scored exams that had been administered to the students on the subjunctive. Ov erall, the test scores were very promising and the students were satisfied with their grades and with what they had learned in the class. According to Terrell et al., the av erage score of the class on this particular exam was 23 out of 25. Although learners were confiden t in their scores and knowledge of the Spanish subjunctive, the researchers wanted to observe subjunctive accuracy in spontan eous conversation. In order to do this, the researchers required participants to an swer a question (given in English) in Spanish. Out of twenty possible questions designed to e licit the Spanish subjuncti ve, the learners were
65 asked to randomly select one. Several of the questions used in the study are as follows: (1) What sort of person would you like to marry?; (2) What are your plans for the future?. According to our interpretation of this study, a correct answer to the first question would require a learner to use the present subjunctive in a subordinate relative clause as in Busco una persona que sea simptica (I am looking for someone nice) or Quiero casarme con alguien que tenga buen trabajo (I want to marry someone with a good job ), for example. In order to answer the second question, a grammatically correct answer c ould have included the present subjunctive in a subordinate adverbial phrase as in Cuando me gradue, quiero ser enfermera (When I graduate, I want to be a nurse). The resear chers collected data from a to tal of six hours of taped oral interviews. The subjunctive exampl es produced by the learners were scored as either correct, incorrect, and required but not used (23). The total num ber of items analyzed by the researchers was eighty one. Out of these eighty on e items, only a total of te n sentences, or twelve percent, showed correct use of the Spanish subjun ctive. Rather than producing sentences like the ones shown above, the learners c onstructed a series of simple sentences not connected by overt CPs (see page 24): 23. *Mis padres quieren mi venga aqu. My parents want-3p-plural [ me come-3p-sing.-subj here.] My parents want me to come here. 24. Qu quieres mi diga ? What do you want-2p-sing [ me to-say?] What do you want me to say? According to the researchers, only 10% percen t (8 out of 81) of the sentences analyzed were comprehensible to native sp eakers of Spanish. These results contrast dramatically from the high test scores discussed previously on which learners achieved an average of 92% accuracy. The researchers also claimed that out of the eigh ty one items analyzed for the study, fifty two of
66 them were sentences that required, but were not used with, the Spanish subjunctive. The majority of these sentences (35/52) were constructed as noun clauses by l earners. Twenty seven of these thirty five items were semantically categori zed as volition, which in our analysis, fall under deontic modality. The remaining eight sentences were semantically cate gorized as doubt, which we refer to as epistemic. There were only a total of thirteen adverbial cl auses and four relative clauses produced by learners. This shows the productivity of nominal clauses and deontic modality in L2 production. Based on the results of their study, Terre ll et al. concluded that after one year of Spanish at the unive rsity level, learners had not ac quired the rules of the Spanish subjunctive sufficiently to be able to use it correctly in oral speech. The results of Terrell et al. are similar to those of Collentin e (1995), discussed below, in th at both studies conclude that learners are unable to use the su bjunctive in oral speech. In add ition, both studies show evidence of L1 syntactic rules being applied in L2. For example, both studies show evidence that the [+ optional] overt CP in English is applie d in L2 production da ta (see Terrell et al p.22). Collentine (1995) conducted a study that looked at the development of complex syntax and mood selection at the intermediate level Spanis h. In order to carry out this study, Collentine chose epistemic subjunctive development in noun clau ses as the focal point of his research. He used two distinct types of oral production tasks to determine how L2 learners at the intermediate level used the subjunctive in oral speech. The fi rst task was referred to as the conversational task which included a 10 minute discussion be tween the researcher and each individual participant (n = 40). The second task, called the controlled oral producti on task, presented the participants (n = 38) with a set of drawings that included distinct contexts used to elicit the subjunctive mood in NP clauses. The results of th e first task showed that the majority of the 804 clauses produced by the learners were simple sentence structures such as Juan come (John eats)
67 or Yo trabajo (I work) employing the indicative (517/804 = 64%). With this said, however, Collentine stated that these sentences were, neve rtheless, frequently juxt aposed in a paratactic fashion, giving the effect of complex utterances as seen in the following examples documented in his study (128): Juan quiereyo voy (*John wantsI go) instead of Si Juan quiere, yo voy tambin (I John wants, I will also go) (127). Anot her example of such an occurrence included *no pienso es justo (*I dont think is fair) instead of no pienso que sea justo (I dont think its fair) in which the complementizer que (that) was missing and two indicative verbs appeared side by side. As for the remaining 36% of the participants (287/804), their sentence structure production was labeled biclausal (i .e., coordinate structures, NPs, if clauses) with a greater use of nouns than verbs as in Carlos come una ensalada, Yo estudio en la biblioteca (Charles eats a salad, I study in the library) (128) The use of such biclausal stru ctures gave the illusion that learners at the intermediate level were able to create complex sentence structures when in reality, they were actually adjoining simple phrases. The second task, which focused on both syntactic and modal development, was constructed to specifically elicit NP structures of both epistemic and deontic modality. The data revealed that most of the partic ipants actually used NP sentence structures (64%). Unfortunately, however, one third (36%) of these NPs were simplified constructions missing the complementizer que subordinate clauses, etc. According to Collentine, the simplified structures were used by a number of participants to av oid having to use subjunc tive forms; a common occurrence among L2 learners of Spanish. The data also demonstrated that most of the learners syntactic constructions involved parataxis (i.e., simple sentence structures appearing side-byside) as seen in the following example (129): Carla dice: ven aqui (Carla says: come here). The overall simplification of complex syntax and high noun to verb ratio were discouraging for
68 Collentine. Ultimately, he conclude d that the participants of this study had not yet reached a fully developed syntactic stage. The re sults also showed that overall the participants were more comfortable using the indicative mood in oral sp eech and preferred to use less advanced syntax such as coordinate or single clause structures during conversation. More relevant to the present study though, is Collentines observation that many of the learners seemingly relied on principles of English syntax to complete the controlled speech task. This suggests that learners either used simpler IP complements or applied the L1 rule of [+ optional] overt CPs instead of the required [optional] overt CP rule in Spanish. Therefor e, this study not only rais es questions about how learners acquire the subjunctive, but also how L1 interferes with L2 acquisition of the subjunctive. Collentine (1995) is especially releva nt to our research because our project builds upon his observations that learners fre quently rely on L1 irrealis stru ctures in L2 to fit the more frequently used L1 templates. In the presen t study, such templates include subordinate IP complements with nonfinite verbs and CP compleme nts with V2 finite [+ indicative] verbs. Moreover, because we have access to large amo unts of cross-sectional data, we are better equipped than previous studies to monitor both modal a nd syntactic evolution as it occurs across levels. In the following section, we look at res earch that focuses on learners understanding of the Spanish subjunctive rather than their production of this m ood. Stokes, Krashen, and Kartchner (1998), Salaberry (1996), and Leow, Egi, Nuevo, & Tsai (2003) all chose to look at the effects of input on L2 subjunctive development. 2.5 Input Studies The results revealed in S tokes et al. (1998) sugge sted that free readi ng had a positive effect on L2 subjunctive competence (i.e., knowledge of language, not performance) over time. This study included a total of 59 students of Spanis h, all of whom were studying Spanish at the University level. In order to carry out their study, the re searchers requir ed the participants to take
69 a test that would measure their competence in the Spanish subjunctive. The test consisted of audio recorded sentence completion which asked participants to co mplete nineteen phrases such as Siempre estudio cuando Los profesores generalmente dan examenes para los estudiantes Espero conocoer a algn profesor que .. In addition, learners filled out questionnaires asking them to include informati on regarding their overall exposure to Spanish prior to the study, amount of formal study in high school, how much time the learner had spent abroad in a Spanish speaking country, and whether or not they were awar e that the subjunctive was the focus of the test. The data were analyzed using parametric statistics which showed that there was no correlation between th e amount of formal instruction and study of the subjunctive. Therefore, the results indicate that formal instru ction did not affect learners performance on the subjunctive test. A multiple regres sion test showed, however, that free reading in Spanish was a significant factor in determini ng a learners level of competence in relation to the Spanish subjunctive. In other words, reading impacted learners subjunctive acquisition while formal instruction did not. The researchers also claimed that raw scores indicated that it would take approximately one thousand hours of reading to raise learners scores on the test. Even though this sounds like a large number, there is reason to support the idea that reading improves subjunctive acquisition. In additi on, even though the re searchers clearly imp lied that learners would need to be motivated enough to engage in sufficient free reading in order to be successful, their results did show that high levels of input as a result of reading, had a positive impact on learners subjunctive competence. Leow, Egi, Nuevo, and Tsai (2003) and Salabe rry (1996) each took the notion of reading to a more advanced level through either textually-enhanced examples or simplified input. Leow et al. tested the effect of te xtual enhancement by using data e licitation, otherwise known as think
70 aloud protocols, to test the e ffects of textual enhancement in second language acquisition. In order to carry out this study, a to tal of 188 learners were recrui ted. The target linguistic forms analyzed were the present perfect and presen t subjunctive in Spanish, neither of which the participants had been formally exposed to prio r to the study. By the end of the study, a total of 116 participants were eliminated from the study based on poor attendance, inability to produce usable material for data analysis, not following directions, etc. The remaining 72 participants were divided into two groups: a textually enhanced group (i.e., experimental group) and and unenhanced group (i.e. control group). The material s used for eliciting the target structures included a text which learners were asked to read aloud and an assessm ent task. The assessment task was later used to measure learners intake of the target structures. The reading text included ten present perfect items and ten present subjun ctive items. The texts used for each of the experimental groups included augmented target item s that would increase learners awareness of the structure. By textually e nhancing subjunctive examples th rough underlining, bolding, italics, etc., Leow et al. predicted th at the learners attention woul d be drawn to the subjunctive structures in question, thus maki ng them more salient. Using a para metric t-test, the researchers discovered that there was no significant difference in the amount of noticing reported in either group. As a result, the researcher s concluded that the textually enhanced structures did not significantly increase intake or aff ect learners overall performance. Salaberry (1996), on the other hand, researched the pedagogical value of simplified input by controlling the quantity and th e level of input the learners received. He did this by manipulating texts so that the m odified version of the original, more complicated texts contained both simpler lexemes and reduced syntactic struct ures. For example, the vocabulary used in the simplified passages was heavily based on Eng lish cognates and employed fewer idiomatic
71 expressions. The syntactic structures that Salabe rry included in these texts avoided using the ProDrop Parameter. This means that every sentence included an overt subject. Further simplification included the repetition of subj unctive lexical cues such as deseo (I desire) thr oughout the text he used in the study. In manipulating and simplif ying these texts, the goal of the study was to improve learners comprehension of the Spanish subjunctive and increase their L2 grammar. The instruments used to collect data for this study included a pre and post test, each of which contained a total of twenty-eight sentences (14 subjunctive items and 14 distracter items). All sentences were exactly the same on both tests and presented to the learners using a multiple choice format. The instruments differed, however, in the order in whic h the sentences were presented. All items on both tests were completely randomized. In order to carry out this study, Salaberry divided a group of 85 college student s into 3 separate groups: simplified input group (which received modified, simpler texts), th e non-simplified input group (which received unmodified texts containing more difficult vocabulary and syntax), and the contro l group (which received texts devoid of subjunc tive examples). Using an ANOV A, Salaberry found that there was no significant difference between the pretest and posttest. What he did find, however, was that intermediate level learners performed better than beginning level learners. In addition, he discovered that overall, both e xperimental groups scored highe r on the posttest than on the pretest. Considering that the data showed no between-group significance, Salaberry was unable to demonstrate that input reduc tion and textual simplification ha d any impact on the learners subjunctive acquisition. In fact, the control groups scores were comparable to both of the experimental groups indicating th at there was essentially no acqui sition value in presenting the learners with simplified input. Taken together, these studies suggest that free, unstructured
72 reading, textually enhanced input, and simplified input are unlikel y to facilitate or enhance subjunctive acquisition for the aver age L2 learner in Spanish. Input Processing : The types of instruments presented in the previous section have not been the only instruction-based instruments used in subjunctive research. Some of the research conducted on the Spanish subjunctive has investigat ed the effects of Proce ssing Instruction in L2 subjunctive acquisition (Collentine 2002; Carroll 2004; Gass 1997; and VanPatten 1996, 2004, among others). Processing Instruction (PI) is defi ned in Collentine (1998) as an input-oriented approach to grammar instruction promoting the intake of grammatical properties (576). PI focuses on comprehension skills rather than pro duction skills and requires learners to follow rigid instructions that assist them in making form-meaning connections. According to Collentine (1998) Processing Instruction enco urages learners to notice a gr ammatical structures formal and semantic properties (580). The principle difference between more traditional grammar instruction and PI is that PI is an input-based method as opposed to an output-based method. In other words, learners are require d to focus on sentence interpreta tion rather than morphosyntactic production. In order to achieve this PI requires that learners be presented with the following components: (1) explanation of the relationship between a given form and the meaning it can convey; (2) information about processing strate gies, showing learners how natural processing strategies may not work to their benefit; and (3) structured input activities in which learners are given the opportunity to process form in the in put in a controlled environment more conducive to making form-meaning connections (see VanPatte n 1996, p. 60, 67-68). Structured input, when used correctly within the PI framework, is benefici al for learners because intake is thought to be extracted from the input. Some researchers who have included Processing Instruction in their research have obtained successful results from their data analyses (Lee 1987, Farley 2001, and
73 Farley 2004), implying that learners competence was positively impacted by PIs structured activities. Collentine (1998) looked at the benefits of Processing Instruction (PI) in Spanish subjunctive development. To carry out his study, Collentine recruited a to tal of fifty-four L1 English speakers studying Spanish at the university level. All participants were enrolled in second-semester Spanish courses at the time of the study. The fifty four learners were randomly assigned to three distinct groups including a control group (n = 18), a PI group (n = 18), and an output-based instruction group (n = 18), which we will refer to here as OBI. The treatment for the PI and OBI groups was different only in the way it was processed. Pack ets including a review of subjunctive forms, an overview and comparis on of subjunctive and indi cative forms in relative clauses, and communicative and ro te practice exercises were dist ributed to the OBI group. These tasks included fill in the blank activities, senten ce completion tasks, student/student conversation exercises and teacher/student conversation exer cises. The PI group received a review of subjunctive forms and information regarding relative clauses in both declarative and interrogative sentences, explanati ons regarding indicative and subj unctive differences in relative clauses, and practice activities rele vant to the above factors. The PI treatment packet was created to raise awareness of the Span ish subjunctive by incorporating more explicit information and processing strategies. With this said, however, both groups were exposed to approximately the same vocabulary items, number of nouns and verbs. The control group was the only group not exposed to the Spanish subjunctive at the time of the study. Collentine also included a pretest and posttest to measure the success of the PI and OBI treatments. Each pre and posttest consisted of an interpretation and production task made up of twenty items. A distracter exercise, which included a series of open-ended que stions, was distributed to the learners between the two tasks
74 in order to reduce the chance of priming effects. On the inte rpretation task, one point was awarded for each correct answer and zero points were awarded for an incorrect answer. On the production task, one point was awarded for the co rrect mood and zero points for incorrect mood. In order to analyze the data, a two-way analys is of covariance (ANCOVA) was used by the researcher in which the pretreatment subjunctive test means and pretest means were considered covariates in the posttest means. The ANCOVA analysis revealed that there was a main effect for group while the post hoc Scheffe showed that results for both groups were nearly the same. Collentine suggested that both the PI and OBI groups performe d better on the posttests. In addition, considering that the rese archer was unable to show th at the PI group performed better than the OBI group, he concluded that the PI gr oup did not benefit any more than the OBI group. Instead, he found that both treatment types increased learners ability to perform and interpret structures in the Spanish subjunctive. Farley (2001) on the other hand, presents diffe rent findings to those of Collentine (1998). He looked at how meaning-based output instru ction (MOI) and Proce ssing Instruction (PI) affected learners ability to interpret and produce the Spanish subjunctive of doubt. In order to carry out his study, Farley recruited a total of 29 L1 English sp eakers at the university level enrolled in a fourth-semester Spanish course at the time of the study. Farley claims that by using fourth semester students, learners would know more vocabulary and ther efore learners would have less difficulty performing in Spanish. Prior to the study, the learners had not been formally exposed to the Spanish subjunctive during their ti me at the university. In order to ensure that participants ability level was homogenous, Farley eliminated participants who received sixty percent or more on the pretest. A score reaching sixty percent or more on the pretest that he administered before beginning the study indicate d that learners knowledge of the subjunctive
75 was too advanced for the study. The remaining part icipants were randomly assigned to either the MOI or PI treatment groups. The MOI group included a total of twelve part icipants while the PI group was assigned a total of seventeen. The materi als used for the treatm ent included a total of two instructional packets. The packet created for the PI group incorporated eight structured input activities accompanied by an information sheet th at contained explicit information regarding the Spanish subjunctive. Such information included stru ctural content of the subjunctive, location in the sentence, semantic environment, and learning strategies. Learners we re also warned that subjunctive verb conjugations are redundant meaning that they reflect the doubt or uncertainty already expressed in the principle clause. The MOI group, like the PI group, received th e same packet and explanations. The difference between these two groups, however, wa s found in the execution of the task. The MOI group, for example, was required to produce subjunctive present tense structures in complex clauses while the PI group was required only to make choices about the sentences in front of them. According to Farley, both groups were expo sed to formatting that would increase saliency of the subjunctive forms. The treatment period lasted a total of two days. Two posttests were completed by the participants. The first was distributed the day after the treatment was completed while the second was administered on e month later. The pre and posttests included a series of interpretation and production tasks. The data showed that both groups scores increased from the pretest to the first posttest on the in terpretation exercise. On the second posttest, the MOI scores were considerably lower than those presented by PI. Both groups showed improvement from the pretest to the first postte st on production exercises as well. Using a twoway ANOVA with repeated measures, PI and MOI were considered the between-subjects factor while the three tests (i.e., pretes t, posttest 1, and posttest 2) were labeled as the within subjects
76 factor. The ANOVA analysis revealed that time had a significant effect on learners performance meaning that both groups benefitted from increas ed exposure to the Spanish subjunctive. With this said, however, Farley states that ther e was a significant intera ction between time and instruction type on the interpretation task whic h showed that PI perf ormed better than MOI overall. A two-way ANOVA was also used for the posttest analyses which showed that both treatments improved learners ability to produce the subjunctive in subo rdinate clauses. There was no significant effect for type of instruction on the production exercise. Although the results of these stud ies differ, they all indicate that subjunctive acquisition appears to rely on, at least to some extent, the ability to in terpret mood. Moreover, Terrell, Baycroft, and Perrone (1987) indi cate that rote exercises do not demonstrate whether or not learners actually understand the irrealis connotation of subjunctive phrases nor do they facilitate subjunctive use in oral speech. These studies, along with Collentine (199 5), also show that learners tend to produce nomina l clauses considerably more than relative, adverbial, or impersonal clauses. Terrell, Baycroft, and Perrone make this point clear in their study as does Collentine (1995). Massery (2008c) provides addi tional support for the findings for Terrell, Baycroft, and Perrone (1987) and Collentine (1995). For this study, a total of 155 oral and written examples were analyzed using an ANOVA an d post-hoc Tukey test which ultimately revealed that learners were capable of successfully producing (on thei r own) the Spanish s ubjunctive in deonticnominal clauses. The data were collected from 11 learners who were randomly chosen from a large data base provided by Dr. Jo aquim Camps from the University of Florida. The results of this study showed that in oral production, learners produced the subjunc tive in nominal clauses
77 115 times, with 61% accuracy. Learners produced adverbial clauses 18 times, with 9% accuracy and impersonal clauses 14 times, but with only 4% accuracy. Taken together, these studies influenced our decision to focus on nominal clauses of deontic modality in our task de sign. Moreover, as a result of th e limited production data found in many of these studies, we chose to rely on gramma ticality judgment tasks, rather than production tasks, to avoid such restrictions. We address these issues, among ot hers, in chapter 3. 2.6 Conclusion Based on this review of Spanish subjunctive acquisition, and of the process of acquiring com plex syntax and modality, we suggest that acquisition of the Spanish subjunctive is comprised of complex syntax (i.e., S1, V1 + S2, V2) and morphology. Therefore, it is no mystery as to why L2 learners of Spanish are faced with a difficult task when acquiring the subjunctive. In order to remedy such difficulties, we suggest the that L2 learners of Spanish need exposure to the following components: (1) Eng lish irrealis structures ; (2) familiarity of parametric distinctions that exist between L1 a nd L2, and how such differences result in unequal surface structures; (3) comprehension of epistemi c and deontic modality; and (4) identification of functional categories. In addition, we have c ontinuously drawn parallels between input and output based studies, which despit e their varied test designs, present si milar results. Taken together, the studies mentioned in this chap ter have allowed us to make the following generalizations: (1) learners pr oduce nominal clauses with greater frequency than other clauses in oral speech; (2) deontic modality is acqui red before epistemic modality; (3) L2 production data is limited and does not reveal learners competence; (4) the role of input in L2 subjunctive development is inconclusive. Although these studies have provided us with ample information regarding the Spanish subjunctive and L2 acquisition, ther e are still some important aspects are missing from this area
78 of research. For example, none of these studies test a wide variety of syntactic environments. Collentine (1995) shows us that learners produc e combined simple sentences using commas. Terrell, Baycroft, and Perrone (1987 ) indicate that learners are ab le to produce nominal clauses of deontic modality. Together, th ese studies show us that production data is limited in L2. With such limited production, it is difficult to make sound assumptions about what IL may actually look like during subjunctive acqui sition. Another problem include s the limited amount of class levels studied at one time. Studies have the tende ncy to focus on either beginning or intermediate learners exclusively which also limits our access to IL development. We have taken these issues into account in order to desi gn a study that would include the following components: (1) a variety of instruction levels (i.e ., crossectional test de sign); (2) competence tasks to test a wide variety of syntactic environments; and (3) nominal clauses as they are appear to be the most prolific clausal structures in previous L2 studies Such factors will allow us to test a variety of different classes and levels that could give us a better idea of L2 subjunctive evolution in average university classrooms. In turn, our design is used to identify acquisi tion patterns by creating plausible IL representations of L2 subjunctive development.
79 CHAPTER 3 METHODOLOGY 3.1 Introduction The studies reviewed in the prev ious chapter ha ve all made contributions to the field of L2 acquisition of the Spanish subjunctive and have provided researchers with many explanations regarding interlanguage development However, none of this work has looked at the L2 acquisition of the Spanish subjunctive cross-sec tionally, nor has it speci fically addressed L2 acquisition of the Spanish subjunctive in term s of IP and CP complement selection in comparison to L1 irrealis structures. As a resu lt, the following research questions form the foundation of the present research: How does the subordinate marker que (that) impact learners judgment of grammaticality in complex sentence structures in L2? With what accuracy do L2 learners at different levels of instruction of Spanish correctly identify mood and conjugate verbs using corresponding morphology in nominal clauses? 3.2 Participants One hundred and sixty North Am erican native sp eakers of English at the University of Florida were recruited for this st udy. All participants were students enrolled in vary ing levels of Spanish as a foreign language at the time of the study. The classes se lected for this crosssectional study were (1) SPN (Beginning Span ish II); (2) SPN 2201 (Intermediate Spanish 2); (3) SPN 3300 (Advanced Grammar and Compos ition); (4a) SPN 4420 (Advanced Composition and Syntax); and (4b) SPN 4780 (Spanish P honetics). These sections were chosen as representative of 1st, 2nd, 3rd, a nd 4th year levels of instructi on. Two classes at each level were selected for the study based on scheduling, availa bility, and willingness of the instructors to permit research in their classes. A total of eight classrooms were visited over a two week period.
80 At the start of the study, all of the participants had already been introduced to the Spanish subjunctive one or more times depending on the le vel. The instructors who allowed their classes to be used for data collection were informed exactly what class day and which period(s) the researcher would administer the activities. All pa rticipants were between the ages of 18 and 23 and all were native speake rs of English studying Spanish as a second language at the time of the study. Gender was not controlled for in this study because we chose to analyze the data from a generative perspective. As a result we did not look at specific so ciolinguistic variables such as gender, age, race, etc. 3.3 Instruments In order to answer the research questions presented in the previous section, the data for this study were collected through two instrum ents: an interpretation task and a production activity1. In the following sections, we discuss each of th e three data collection instruments in detail, including information about how each task was introduced to the participants of the study. In the latter portions of this chapter, we discuss the st atistical procedures used to analyze and interpret the data. The ME task is discussed first because it was distributed to learners on the first day of data collection. L2MAC is discussed second, not only because it was distributed on the second day, but also because of its overall complexity. 3.3.1 Magnitude Estimation (ME) The ME task was distributed on the f irst da y of the study. ME is a technique which was originally used in psychophysic s to estimate factors such as brightness or loudness and compare them with a physical stimulus. More recently, ho wever, this method has been used in linguistics 1 We included a third task entitled Syntactic Identification (SI). Unfortunately, due to po ssible priming effects, the results were skewed and therefore invalid. A full explanation of SI task design, scoring procedure, and corresponding RQ can be found in Appendices C and D.
81 and adapted to accommodate acceptability judgmen t tasks based on the notion that grammatical judgments can be obtained when learners are exposed to linguistic stimuli. In linguistics, ME can be described as a modernized way to monitor subt le linguistic intuitions and to numerically rank items as X is greater than Y. In other words, ME is a procedure whereby participants are asked to rank a stimulus by stating how much better or worse the stimulus is from the previous one (Mackey and Gass 2005, p. 359). It has received a good deal of attention in SLA since the mid 1990s, including studies conducted by Zhang (1995), Bard, Robertson, and Sorace (1996), Southwood and Flege (1997), Gass, Mackey, Alvare z-Torres, Fernndez-Garcia (1999), Keller and Asudeh (2001), and Featherson (2005a, 2005b). ME allows the researcher to address L2 data with greater accuracy than the more commonly used interval scales. Analyzing the grammaticality of Spanish subjunctive structures using confining, traditional scal es is difficult because the choi ce of mood is controlled and manipulated by the speaker meaning that it requi res a more flexible and gradient numerical range. Interval scales, for example, require some type of numerical ranking among various items, which is advantageous because they allow res earchers to measure linguistic data quantitatively. However, these scales implement somewhat rigid and artificial ranking systems that are controlled by the researcher rather than the pa rticipant. The rigidness of the more traditional interval scale requires participants to confine th eir intuitions to potentially restrictive scales. Such limitations stifle the participants ability to interpret linguistic data on a more intuitive level, ultimately putting in jeopardy the goal of the acceptability judgment tasks themselves. ME, on the other hand, combines the be nefits of the more traditional interval scale (ranking and mathematical quantification) with linguistic intuition data. By allowing participants to select numerical values that represent the stimuli as they perceive them, the researcher gives each
82 participant the freedom to interpret the data on a more personal level and thereby provide a more accurate analysis of their judgment. This is ad vantageous for researchers because grammatical intuitions are gradient meaning that they function on an arbitrar y slope that only the individual participant can create and fully control. From this perspective, it would be unnatural to confine such intuitions to a traditional seven or ten po int scale controlled by someone other than the participants themselves. To our knowledge, there ar e currently no studies th at use ME in relation to L2 acquisition of the Spanish subjunctive. Ther efore, with this study, we expand the uses of ME in SLA to explore L2 subjunctive intuitions. In the paragraphs that follow, we briefly explain how ME works and how it was presented to the participants prior to the distribution of the actual tests used for data collection. 3.3.2 ME Data Collection The first day of the study was dedicated strict ly to the ME acceptability judgment task. Before the task was administer ed, a brief presentation about how ME was originally used in psychophysics to measure perception using visual stimuli was introduced to the participants using distinct line leng ths. Following a study conducted by Keller and Asudeh (2001), the participants were asked to assign a number to the first line (also referred to as the reference line) that appeared on the board. After choosing this number, the learners were asked to judge the other two lines in relation to the reference line. The values assigned to the lines were entirely participant driven. In other words, the partic ipant could have assigned 300, 30, or 3 to the reference line if that is how their intuition guided them. Howe ver, the remaining lines should have been judged in proportion to the reference line. For example, using figure 3-1 as the visual stimulus, if the learner assigned 300 to the re ference line, the second line should have been assigned 150 and the final line 75. Once the lear ners had chosen numbers to represent their
83 perceptions of each line, we provided them with our analysis (see Figure 3-1), using the number 30, to clarify the process. ___________________ (Line 1: reference line = 30) __________ (Line 2 = 15) _____ (Line 3 = 7.5) Figure 3-1. Sample reference line demonstration By using this example, participants saw how this method was used in psychophysics which allowed us to more effectively explain how these same principles could be applied to ME tasks in linguistic studies. Before begi nning the test, the learners were asked to complete a practice test, allowing them to apply the principles of ME they had been exposed to. The practice test included a total of 10 sentences in English that varied in gramma ticality. All of the sentences provided in the practice exam were presented to the learners in E nglish as a way to avoid altering their intuitions of Spanish in any way prior to the distribution of the actual test used for data collection. We also wanted to be sure that they understood ME. The ME task itself consisted of 60 examples in Spanish plus the reference sentence El chico come hamburguesas en el parque (The boy eats hamburgers in the park), for a to tal of 61 sentences. We chose to use the phrase El chico come hamburguesas en el parque because we did not want to run the risk of altering learners intuition about the subjun ctive. We were afraid learners would discover that the task was based on the subjunctive and therefore chose a sentence that was less likely to reveal its purpose. As we are the first to analyze L2 subjunctive intuition da ta using ME, to our knowledge, we had no direct refere nces from which to draw upon in terms of what the reference sentence should consist of in Spanish. With this said, however, we do provide examples of other SLA studies that have designed ME tasks. Featherson (2005), for example, told learners that the first sentence in a series of items was considered the reference line. All other items wo uld be rated in proportio n to the score assigned
84 to the reference line. He used the sentence Jackie sent immediately the le tter to her cousin as the reference line in the practice exam that he administered before the actual task which was given in German. This sentence is not a native-like English sample as shown by the incorrect placement of the adverb. The second item presented to the learners Jackie sent the letter to her cousin immediately is completely grammatical. Th e third example, however, Jackie sent to her cousin immediately the letter is the l east grammatical of the three sentences. Southwood and Flege (1999), like Featherson (2005) used ME to judge second language acquisition data, only this time, is was used to judge the perception of fo reign accents. They, like Featherson, also used a modulus that could be considered mi ddle of the range (338). This middle of the range modulus, as they refer to it, consisted of an English sentence read by an Italian speaker whose accent was ne ither native-like nor incomprehe nsible in L2. As shown here, Featherson and Southwood and Flege chose referen ce examples that were neither completely correct nor incorrect. In our study, the ME task requi red participants to assign a num ber to the reference sentence which appeared first in a series of sentences. The reference sentence used, El chico come hamburguesas en el parque (the boy eats hamburgers in the park), did not include the grammatical structures that were being test ed. It did, however, provi de learners with a grammatically correct structure in Spanish. This sentence was also chosen as the reference sentence because it was easy for participants at all levels to understa nd on both a syntactic and semantic level. There were a total of six distracter sentences that used stru ctures unrelated to the subjunctive interspersed randomly throughout the items. The remaining 54 examples were chosen because of their syntactic structure and are referred to as environments 1, 2, 3, 4, 5, and 6 in this study. These environments were ordered randomly so the students would remain unaware
85 of the various structures being examined. The si x syntactic environments used are presented in table 3-1. Note that only several of these stru ctures yield accurately structured subordinate clauses in Spanish, although the others yield phrases common in lear ner interlanguage. For example, environment 1 (+ que + subjunctive) is a grammatical subordinate structure in Spanish as seen in espero que vaya a la fiesta (I hope he/she goes to the party). Environment 3 (+ que + indicative), however, renders a co rrectly structured subordinate clause in English, and in Spanish, when epistemic verbs posses the following semantic properties: [+ existence] and [+ truth] Take the following example: 25. Pienso que va a la fiesta. I think-1p-sing.that he/she 3p sing. go to the party. I think (that) he goes to the party. From this example, we can see that in both la nguages, CP structures are used with V2 [+ finite] verb forms. In English, however, the overt CP is optional and V2 appears in the indicative mood. Unlike English, Spanish always requires V2 subjunctive morphology in deontic-nominal environments that include S1 and S2. The existen ce of a [+ optional] overt CP in learners L1 raises questions as to how learners analyze L2 syntax and in turn, how the subordinate marker que affects their sense of grammaticality in L 2. For example, are learners reducing complex clauses in L2 to [que ] IP structures or is the [+ optional] overt CP rule being transferred from L1 to L2? We address these questions among others in chapter 4. Before we can proceed further, however, we need to complete the present an alysis regarding the syntactic environments analyzed in our study. Environments 1-3 includ ed the lexical item que (that) while environments 4-6 did not. It was necessary to expose the learners to structures that included a nd excluded this lexical cue in order to measure the impact it had on the pa rticipants perception of grammaticality, thereby
86 addressing research question 2: H ow does the subordinate marker que (that) impact learners judgment of grammaticality of comple x sentence structures in L2? Table 3-1. Syntactic environments used in ME Environment Formula Example I want John to speak. 1 + que + subjunctive Quiero que hable Juan. 2 + que + infinitive *Quiero que hablar Juan 3 + que +indicative *Quiero que habla Juan. 4 que + subjunctive *Quiero hable Juan. 5 que + infinitive *Quiero hablar Juan. 6 que + indicative *Quiero habla Juan. This question was based on the hypot hesis that learners would fi nd structures that used que (as in environments 1-3) less grammatical than th ose that did not use que (as in environments 46) because the latter environments reflect the more commonly used English irrealis constructions. 3.3.3 L2 Mood and Conjugation Task (L2MAC) We presented L2MAC (see appendix 1) on the second day of the study. L 2MAC was designed to investigate the first research quest ion: With what accuracy do L2 learners at different levels of instructi on correctly identify mood and conjugate verbs using corresponding morphology in nominal clauses? In an effort to answer this question, this task required the participants to indicate the mood of each item by writing the lett er I for indicative or S for subjunctive, and to conjugate the verbs found in th e subordinate clauses according to their modal interpretation of the sentence. This exercise consisted of 15 sentences that were constructed according to the following three semantic environments: (1) epistemic subjunctive phrases (ES); (2) epistemic indicative phrases (EI); and (3 ) deontic subjunctive phrases (DS). The dual component test design required learners not only to conjugate the verb in the subordinate clause, but also to indicate how they identified the mood of each sentence. This design allowed for the
87 separate analysis of intended mood as well as morphological accuracy of their conjugations. L2MAC was constructed in this way to avoid relying strictly on performance data (i.e., production) because knowledge of the participants modal competence is also important to this investigation. Therefore, we can see which mood the participants intended to use even if their choices were not reflected mor phologically. For example, if the participant was exposed to the sentence Mabel sugiere que Alejandra ________ (romper) su blackberry (Mable suggests that Alexandra break her blackberry) and she chose S as the mood and *rompe (he/she breaks) in the indicative as opposed to rompa in the subjunctive, we know th at she was able to accurately identify mood in L2 even though the incorrect verb conjugation did not reflect her modal competence. Learners ability to identify mood is important in th is study, especially considering that learners production is often li mited as discussed in chapter 2. 3.3.4. Other Information We chose to administer the ME task first in order to avoi d priming effects. If we had distributed L2MAC prior to the ME task, it was likely that learners, especially at the more advanced levels, would have realized the pur pose of the study. Such recognition would have made learners aware of the subjunctive focu s of ME thus altering their intuitions. Upon completion of these tasks, part icipants filled out a language background form which asked them to indicate whether or not they had chosen Spanis h as a major or minor, to list previous courses taken in Spanish, and finally, to specify how much exposure they had to the Spanish subjunctive prior to the study. Because the language backgro und form required learners to include specific information about their previous experience with the Spanish subjunctive, it was administered after L2MAC in order to avoid pot entially influencing the participants performance on the tasks. On the second and final day of th e study, we also revealed to the learners the nature, purpose, and goals of the project using a PowerPoint presentation. In a ddition, we also took the time to
88 answer any questions participants had about our study. Overall, learners asked very few questions pertaining to the lingu istic aspect of our study and instead, took the opportunity to express their personal struggles with the Spanish subjunctive. 3.4 Scoring Procedures In this sectio n, we show how each variable was labeled and categorized, what types of tests were run, which statistical models were followed, and finally, how these components were used together to analyze and inte rpret the data of this study. Each task followed a distinct set of scoring procedures. For the ME task, each participant chose numbers that reflected their individual reac tion to the stimuli presented to them during the task. In this activity, as expl ained previously, the participan ts were allowed to judge the grammaticality of the sentences they read by assigning each sentence a score based on a scale that they themselves had created. In order to an alyze this test statisti cally, however, the numbers had to be made comparable to each other be fore using an ANOVA. This was done by scaling and centering the data by using Excel. In order to center the data, we took the average score of each individual test first and then s ubtracted the average of these numbers. The difference was divided by the standard deviation of the set of observations. As a result, all of the observations were centered at zero causing the results to be s ubject to homogenous variability. The homogenous variability is what made the numbers analogous and therefore allowed us to more accurately observe the participants behavior on the test. After the original scores had gone through the pretreatment process, the results were analyzed using a linear-mixed model which was used to capture a possible correlation in the responses of each participant. The name linear-mixed comes from the fact that some of the effects analyzed in a given study may be fixed while others may be random. In this particular project, the linear-mix ed model considered the test to be the fixed factor while the participants were analyzed as the random factor. The tests and participants were
89 labeled this way because participants exposure to the Spanish subjunctive prior to this study varied while the tests remained unaltered throughout the duration of the study. In L2MAC, there were a total of 15 sentences. Each of these senten ces contained a modal identification component which asked participants to indicate the mood of the sentence and another which asked them to conjugate the ve rb of the subordinate clause. One point was awarded for accurate mood choice and another fo r accurate verb conjugation. Each participant could obtain up to two points per sentence maki ng the maximum possible score a total of 30 points (15 sentences x 2 points per sentence). Agreement mo rphology, however, was not taken into consideration. For example, if the target form of the verb comer (to eat) was to be conjugated as coma (1st, 3rd person singular, subjunctive) in the subordinate clause and the participants answer was comas (2nd person singular subjunctive) 1 point was awarded for accurate subjunctive verb conjugation. The statistic al test used to analyze this data was a Two Way ANOVA which considered mood + conjugati on ability as factor 1 and course level as factor 2 using the following formula: prop = course + mood + course*mood (interaction) + student (random) + error. After running the Tw o Way ANOVA, a post hoc Tukey-Kramer test was conducted to observe the cross-sectiona l development between factor 1 (mood and conjugation ability) and factor 2 (c ourse level). In addition to ma king separate conclusions about factors 1 and 2, we were able to see whether or not modal development and production skills evolved together or separately, a nd whether or not learners were better at interpreting mood than conjugating verbs at any time duri ng interlanguage development. 3.5 Conclusion This chapter presented th e reader with in formation regarding instrument design and distribution, showing that each instrument wa s tailored specifically to address the IL development of the Spanish subjunctive. ME wa s designed to investigate learners syntactic
90 competence in L2 subjunctive. L2MAC was de signed to examine participants ability to accurately choose mood and conjugate verbs in nominal clauses. Having explained how the data was collected, scored, and statisti cally quantified, we turn now to the results and interpretations of the data. In chapter 4, we discuss the result s of each task in detail while revisiting and answering the two research questions presented at the beginning of this chapter. Finally, in chapter 5, we discuss the overall success of the project, goals and objectives for future research, and the pedagogical implications of this study.
91 CHAPTER 4 RESULTS AND DISCUSSION 4.1 Introduction W ith the previous chapters discussion of how the activities were us ed for data collection, their distribution, and scoring procedures in mind, th is chapter discusses and interprets the results of each individual task. The tasks are discusse d in the following order: Magnitude Estimation (ME) and L2 Mood and Conjugation Abilities (L2MAC ). ME is discussed first providing readers with information regarding learners syntactic in tuitions in L2. L2MAC, the more traditional of the two tasks, helps to explain modal and morphosyntactic struggles that learners face during L2 acquisition of the Spanish subjunctive. L2MAC allo ws us to make reference to specific modal and conjugation errors in the data. Such examples help clarify more abstr act concepts presented in the ME analysis. The following sections presen t the results along with th e statistical analysis of the various factors included in each activity. Ch apter 5 places the results of this study within the broader empirical and theore tical context in order to draw conclusions about our findings. 4.2 Magnitude Estimation Task ME allows us to analyze learners subjunctive intuitions in a way that, to our knowledge, has not been tested prio r to this study. By using ME, we were able to show that the functional category que was either missing or null in learners st ructural development. Furthermore, this task design required learners to rely on their grammatical intuitions and shed light on learners competence rather than their performance abilitie s; a feature that sets ME apart from more traditional task designs in L2 subjunctive ac quisition. In this section, we show how the participants rated the distinct s yntactic environments tested in ME and illustrate the impact that the functional category que (that) had on learners grammatical intuitions.
92 Recall that ME allows each participant to a ssign scores according to their own arbitrary scale (see chapter 3). Therefore, in order to ma ke the scores comparable, we standardized the data for each participant. The reader is also reminded that eight grammatical sentences (4, 20, 25, 38, 41, 46, 51, and 61) were eliminated prior to the st atistical analysis of th is task as our focus was only on participants reaction to ungrammati cal structures. By using only ungrammatical structures, we were able to test a much wider variety of syntactic environments in the task, including those used in English. We used a Two Way ANOVA model in which we included the participants as a random effect. This means that we included the participan ts in the model as an effect and tried to capture a possible relations hip between the responses of each learner. The results show that course was not a significant fa ctor by itself; however, it should be kept in the model because its interaction with environment is significant. The current statistical model used for this analysis does not allow us to see exactly how course a ffects the results. We dont exactly know what would happen if we took it out of the model. However, as shown below, course does have an effect on the interaction with synt actic environment and le vel of instruction. Table 4-1. Results for ME Task Factor F P-value Course 0.21 0.9307 Environment (syntactic environments 1-6) 159.73 <.0001* Environment*Course 3.19 <.0001* In chapter 3, it was hypothesized that pa rticipants would ra nk environments 1 (+ que + subjunctive), 2 (+ que + infinitive), and 3 (+ que + indicative) lower than environments 4 (que + subjunctive), 5 (que + infinitive), and 6 (que + indicative). It was al so suggested that this ranking order would demonstrate the tendency that learners have to reduce complex syntax (i.e., S1, V1 + S2, V2) to IP structures or [+ op tional] overt CP complements with indicative morphology. The perceived correctness of such e nvironments would show a preference for
93 English-like syntax over complex L2 structur es. Our intuitions were partially correct. Environments 5 (que + infinitive) and 6 (que + indicative), which are the most English-like structures, received the highest scores from learners. Environments 1 (+ que + subjunctive), 2 ( + que + infinitive), and 3 (+ que + indicative), however, were ra nked between environments 5/6 and 4. These environments (1, 2, and 3) also received approxima tely the same score showing no significant difference between them. Environment 4 (que + subjunctive) was ranked the lowest of all six environments and therefore consider ed to be the least grammatical structure by learners. Table 4-2 shows that there are stat istical differences between each of the six environments. In other words, each environment is significant in the model. These differences, however, exist as a result of each environments in teraction with course. Course is not shown in this table, however, because c ourse by itself provides only large p values and is therefore not considered significant in the model. Table 4-2. Means and standard deviations for ME and corresponding p values Effect Environment Estimate Standard error DF t value p value Environment 1/+ que + subjunctive -0.19 0.03 780 -6.02 <.0001* Environment 2/+ que + infinitive -0.13 0.03 780 -4.18 <.0001* Environment 3/+ que + indicative -0.07 0.03 780 -2.09 0.0365* Environment 4/que + subjunctive -0.4 0.03 780 -13.51 <.0001* Environment 5/que + infinitive 0.52 0.03 780 16.75 <.0001* Environment 6/que + indicative 0.53 0.03 780 16.96 <.0001* Table 4-3 on the following page summarizes th e results of ME using Tukeys post hoc test to adjust the p values. In order to understand th e results provided in table 4-3 the reader must know how to read it. The syntactic environments in column A indi cate that there is a difference between any two environments that appear in this column. For example, in row 3, we notice that there is an interaction between environments 1 and 4 as demonstrated by the low p value (p = <.0001). The estimated difference provided in colu mn B is positive indicating that environment 1
94 (+ que + subjunctive) is the environment with th e higher value. In or der to clarify this information, each environment that assumes the highe r score appears with an asteric (i.e., in row 3, environment 1 received the highe r score). In table 4-3 we noti ce that there are no statistical differences between syntactic environments 1 (+ que + subjunctive), 2 (+ que + infinitive), and 3 (+ que +indicative), or between environments 5 (que + infinitive) and 6 (que + indicative). Environment 4, however, was statistically different from the other 5 environments and therefore appears separately in table 4-4. Table 4-3. Compared Syntactic Envir onments: p values adjusted by Tukey A B C Compared syntactic environments Estimated difference p value 1. 1 (+ que +subjunctive) 2* (+ que +infinitive) -0.06 0.7831 2. 1 (+ que +subjunctive) 3*(+ que +indicative) -0.12 0.0620 3. 1* (+ que +subjunctive) 4 (que +subjunctive) 0.23 <.0001 4. 1 (+ que +subjunctive) 5*(que +infinitive) -0.07 <.0001 5. 1 (+ que +subjunctive) 6* (que +indicative) -0.07 <.0001 6. 2 (+ que +infinitive) 3*(+ que +indicative) -0.07 0.6804 7. 2 (+ que +infinitive) 4 (-que +subjunctive) 0.29 <.0001 8. 2 (+ que +infinitive) 5*(que +infinitive) -0.66 <.0001 9. 2 (+ que +infinitive) 6*(que +indicative) -0.67 <.0001 10. 3 (+ que +indicative) 4 (que +subjunctive) 0.36 <.0001 11. 3(+ que +indicative) 5*(que +infinitive) -0.59 <.0001 12. 3 (+ que +indicative) 6*(que +indicative) -0.60 <.0001 13. 4 (que +subjunctive) 5*(que +infinitive) -0.95 <.0001 14. 4 (que +subjunctive) 6*(que +indicative) -0.95 <.0001 15. 5 (que +infinitive) 6*(que +indicative) -0.01 1.0000 In order to more concisely organize the information shown in table 4-3 we provide the reader with a summary of the ME rankings in ta ble 4-4. The ranking of the six environments, as shown in table 4-4, allows us to make the fo llowing observations: (1 ) structures containing infinitive or indicative verbs in the subordinate clause were scored the highest by pa rticipants if the complementizer que was not included. We remind the reader that environments 5 (que + infinitive) and 6 (que +indicative) are the most English-like structures illustra ting a preference for syntax that resembles learners L1; (2) structures containing the complementizer que received
95 the second highest ranking in which the subordi nate V2 did not alter learners sense of grammaticality; and (3) structures containing a subj unctive verb in the subordinate clause that did not include the complementizer que were ranked as the least grammatical structures among learners. Table 4-4. ME Environment ranking Environment Rating Environment 5 (que + infinitive) Environment 6 (que + indicative) Highest ranked structures (i .e., perceived as the most grammatically acceptable s yntactic environments by participants) Environment 1 ( + que + subjunctive) Environment 2 (+ que + infinitive) Environment 3 (+ que + indicative) Middle ranked structures (i.e ., perceived as moderately grammatical environments by participants) Environment 4 ( que + subjunctive) Lowest ra nked structures (i.e., perceived as the least grammatically acceptable environments by participants) Based on these observations, we propose that learners identified subjunctive morphology in subordinate clauses as requiring the complementizer que in complex syntactic structures (i.e., S1, V1 + S2, V2). We suggest that this indicates that learners had access to functional categories in L2. In other words, if learners did not have access to FCs, they would have marked environment 4 (que + subjunctive) as grammatical along with environments 5 (que +infinitive) and 6 (que + indicative). Even in English a structure resembling environment 4 (que + subjunctive) would require at least a null CP. Take the sentence I prefer (that) you go tomorrow for example. Even if the speaker chooses not to use the overt CP, the structure must include it. Therefore, we have ev idence of L1 transfer. In chapter 5, we return to these results and apply them to X-Bar theory. After having analyzed and disc ussed the results for the ME ta sk, we are ready to discuss L2MAC. With L2MAC, we gain a clearer view of learners semantic abilities and their
96 perceptions of deontic and epistemic modalities. In the following section, we present the reader with the results of L2MAC and revisit the research questions that motivated this study. In addition, we answer each question by applying the results and findings in our discussion. 4.3 L2MAC L2MAC wa s the more traditional task presented to the participants of this study. In this sense, it more closely resembles the design formats of data collection instruments used in previous L2 subjunctive acquis ition studies. L2MAC also mirrors the mostly mechanical ways learners tend to practice the subjunctive in fore ign language textbooks. The design of this task is practical in that it requires learners to id entify mood and conjugate verbs simultaneously. Therefore, the main purpose of this task was to test modal and morphological skills in L2. While it is important to continue to test and research aspects of L2 Spanish subjunctive acquisition in this way, it is not necessarily a new technique in the field of SLA. With this said, however, L2MAC is designed to observe how well learners were able to both choose the mood of a given sentence and complete guided production task s that focused on V2 morphology. Having given the learners the oppor tunity to identify mood and conjugate verbs in the same exercise, we were able to measure L2 subjunctive competence cros sectionally along with an y potential interaction between course level (i.e., variable 1) and ability level (i.e., variable 2). L2MAC required participants to indicat e the target mood of each item (I for indicative or S for subjunctive) and then to conjugate the s ubordinate verb according to the chosen mood. A total of two points could potentially be scored for each item. One point was awarded for accurate verb conjugation and another point was awarde d for accurate choice of mood. When looking at the verb conjugation and mood choice, it was observe d that the participants rarely lost points on one aspect without losing points on the other. This was interpreted to mean that modal and conjugation abilities could not be teased apart. In other words, participants who accurately
97 indicated the target mood of a given item were also able to equally conjugate the verb accordingly. Conversely, participants who were unable to accurately choose the target mood could not accurately conjugate th e verb. Most mood and conjugation scores were similar for each participant. Table 4-5 provides th e reader with the raw scores fo r L2MAC. Here we see that all levels scored nearly the same on mood and c onjugation. The conjugation scores for Beginning Spanish II (1131) are slightly lower than the mood scores. If we refer to the literature provided in chapters 1 and 2, however, such morphological inaccu racy may exist as a result of mismatched features in IL. These notions will be discussed in greater detail in chapter 5 and applied using XBar theory. Table 4-5. Raw Scores for L2MAC Level Total mood Total conjug Total 1131 (Beginning Spanish II) 8.95 8.48 17.43 2201 (Advanced-Intermediate Spanish) 7.77 7.77 15.55 3300 (Advanced Grammar and Composition) 9.20 8.85 18.05 4420 (Advanced Composition and Syntax) 9.68 9.84 19.53 4780 (Spanish Phonetics) 8.92 8.83 17.75 Table 4-6 provides the reader with the results for five of the participants mood and conjugation scores for L2MAC. We chose to show scores from one participant at each level as representations of common behavior among the learners. We see from this table that learners received identical or similar scores on mood and conjugation which, in only one case, varied more than two points. Table 4-6. Sample scores for L2MAC (mood and conjugation abilities) Participant Total mood Total conjugation Total P1131A 10 9 19 P2201B 7 7 14 P3300C 8 7 15 P4420D 10 10 20 P4780E 10 11 21
98 As mentioned previously, ta ble 4-5 shows the raw scores for each individual group on L2MAC. In the last column, we see that Advanced Composition and Syntax (4420) received the highest score of all the classes. Advanced Gr ammar and Composition (3300) received the second highest score. We also observe that Spanish Phonetics (4780) received lower scores than the other two advanced levels. We remind the re ader that Advanced Grammar and Composition (3300) and Advanced Composition and Syntax (4420) are language classes while Spanish Phonetics (4780) is a content class meaning that grammar is not the focal point of the course. As a result, the learners enrolle d in the language courses were exposed to Spanish grammar examples more regularly than th e learners in the Spanish Phonetic s course. Table 4-5 also shows that Beginning Spanish II (1131) received higher scores than Advanced Intermediate Spanish (2201). At the time of the study, learners in Be ginning Spanish II (1131) ha d just begun learning the Spanish subjunctive. The Advanced Interm ediate course (2201), however, was working on other grammar topics that did not include the s ubjunctive. In addition, we suggest that learners at the intermediate level may have scored lower on this task as a result of IL reconstruction; a concept we will discuss in latte r sections of this chapter. In figure 4-1, we apply the raw scores of L2 MAC to a bar graph in order to more vividly illustrate the similar (and in most cases identic al) mood and conjugation scor es for all levels of instruction. This figure clearly shows that e ach courses mood and conjugation scores were identical or nearly identical and co uld therefore not be teased apart. Having been presented with the raw scores for L2MAC, the reader is equipped to follow the statistical analysis of this task which we begin by presenting the reader with table 4-7. In table 4-7, it is observed that cour se is the only significant variab le in the model as shown by the
99 low p value (p = 0.0007). The large p values for abili ty and course*ability i ndicate that they are not significant variables in the model. Figure 4-1. L2MAC raw scores (m ood + conjugation) out of 30 To further analyze the results of L2MA C, a Two Way ANOVA was performed. A Two Way ANOVA was chosen because ther e were two distinct variables: course (variable 1) + ability (mood + conjugation skills/variable 2). Table 4-7. Summary of tests for significant effect s of the variables included in the model for L2MAC Source F Value P value Course 4.96 0.0007* Ability (L2 mood + conjugation scores) 0.29 0.5929 Course*ability 0.23 0.9224 Table 4-8 provides the results of this test. We considered a model that would include group effect (i.e., levels of instruction), ability eff ect (i.e., L2 mood + L2 conjugation skills), and interaction. LS Means1 was used to estimate the mean effect of the cells becau se the number of participants in each group was different. After adjusting multiple comparisons through Tukey post hoc tests, we found significant differences between Advanced Intermediate Spanish (2201) 1 LS Means is a term used in statistics that refers to the statistical procedure necessary for computing the mean of the average in an unbalanced group.
100 and Advanced Composition and Syntax (4420), a nd between Advanced Intermediate Spanish (2201) and Advanced Grammar and Composition (3300). Table 4-8. Least Squares Means fo r effect course for L2MAC Courses (1131) Beginning Spanish II (2201) Advanced Intermediate Spanish (3300) Advanced Grammar and Composition (4420) Advanced Composition and Syntax (4780) Spanish Phonetics (1131) Beginning Spanish II 0.2633 0.5954 0.0574 0.9920 (2201) Advanced Intermediate Spanish 0.2633 0.0111* 0.0004* 0.2020 (3300) Advanced Grammar and Composition 0.5954 *0.0111 0.5715 0.9300 (4420) Advanced Composition and Syntax 0.0574 *0.0004 0.5715 0.2489 (4780) Spanish Phonetics 0.9920 0.2020 0.9300 0.2489 Table 4-8 does not, however, allow us to s ee how well the participants in these groups performed on L2MAC. Therefore, we added tabl e 4-9 which provides the actual scores for each course level. In table 4-9, we observe that Advanced Intermediate Spanish (2201) received the lowest score (8.03) while Advanced Composition a nd Syntax (4420) received the highest (9.76). Though it is not surprising that the Advanced Composition and Syntax cl ass (4420) received the highest score, it is puzzling that Advanced In termediate Spanish (2201) received the lowest score. We continue this discussion further in latter sections of this chapter. In table 4-9, we also notice that Spanish Phonetics (4780) received lower scores than Advanced Grammar and Composition (3300) and Advanced Composition a nd Syntax (4420). The differences between these groups are not significant, however, and will therefore not be addressed further.
101 Table 4-9. LS Means of the scores for L2MACcontrolling by course Course Score LS mean Beginning Spanish II (1131) 17.44 (/30) Advanced Intermediate Spanish (2201) 16.06 (/30) Advanced Grammar and Composition (3300) 19.34 (/30) Advanced Composition and Syntax (4420) 19.52 (/30) Spanish Phonetics (4780) 17.76 (/30) Recall that participants could receive up to a total of thirty points on L2MAC. L2MAC was specifically designed as a dua l component task that allowed us to analyze the interaction between mood and conjugation skills (i.e., ability = L2 mood skill s + L2 conjugations skills) and course level. With L2MAC, we also wanted to look specifically at how learners performed and perceived particular modal environments (i.e., DS, ES, EI). As menti oned previously, earlier studies show that deontic modality in nominal cl auses is the most prolific syntactic-semantic environment among L2 learners. In our study, we wa nted to see if learners competence revealed the same tendencies. In order to do this, we separated the deontic subjunctive (DS) from the epistemic subjunctive (ES) and epistemi c indicative (EI) modal environments. 4.3.1 Results for Epistemic and Deonti c Modality in Nominal Clauses In this section of the L2MAC analysis, we focus on modality (i.e., acquisition of epistemic and deontic modality). Table 4-10 provides the ANOVA for the effects of the model and shows the sum of squares, providing information re garding the interaction between the different variables. From this table, we see that c ourse is significant (p = <.0003) and that modal environment (DS, ES, EI) is significant (p = < .0001). The interaction (i.e., combined effects) between course and environment is also signific ant (p = <.0001) suggesting that course level may have affected learners ability to choose th e correct mood (DS, ES, EI). Ability (L2 mood skills + L2 conjugation skills) is however, not sign ificant in this model even through interactions (i.e., combined effects) with environment, course, and course*environment.
102 Table 4-10. ANOVA table for the model considering course, ability, and environment as variables Source of Variance DF Sum of Squares Mean Square F Value P values Course 4 33.51 8.38 5.33 0.0003* Ability 1 0.96 0.96 0.61 0.4344 Environment (DS, ES, EI) 2 557.55 278.77 177.35 <.0001* Ability*Environment 2 4.93 2.46 1.57 0.2090 Course*Ability 4 2.13 0.53 0.34 0.8513 Course*Environment 8 90.01 11.26 7.16 <.0001* Course*Ability*Environment 8 3.41 0.43 0.27 0.9753 Note: ability = L2 mood skills + L2 conjugations skills Table 4-11 shows that there are statistical differences between all modal environments (DS, ES, EI) meaning that every environment is significant in this model. What this table does not allow us to observe, though, is which modality received the highest score and which received the lowest. Table 4-11. P values for differences in modal environment adjusting by Tukey Semantic Environment 1 (DS) 2 (EI) 3 (ES) DS <.0001* <.0001* EI <.0001* 0.02* ES <.0001* 0.02* Therefore, in order to retrie ve this information we look at table 4-12. Table 4-12 shows that learners performed more accurately on DS environments following results from Collentine (1995) and Terrell, Baycroft, and Perrone (1987 ). ES, on the other hand, received the lowest score. Recall that there were a total of 15 sentences in L2MAC. These 15 sentences were divided into the three modal subcategories (DS, ES, EI) making each individual environment proportional to 5. Table 4-12. Multiple Comparisons: Acquisiti on of DS, EI, ES modalities for L2MAC Semantic environment Score (LS means) DS (deontic subjunctive) 4.06 (/5) EI (epistemic indicative) 2.51 (/5) ES (epistemic subjunctive) 2.23 (/5)
103 We already know from the high score (4.06) pr esented in table 4-9 that DS modality was the most accurate semantic environment among learners. This is why we have decided to analyze DS by itself. In addition, we also notice that there is not a large numerical distinction between ES and EI environments which makes the epistemic analysis more intricate. Therefore, the epistemic discussion is broken down further in to two se parate analyses: modal analysis of ES and epistemic context analysis, which ar e discussed after our DS analysis. Table 4-13. P values for difference between cour ses for L2MAC controlling by DS environment Course Beginning Spanish II (1131) DS Adv. Intermed. Spanish (2201) DS Adv. Grammar and Comp (3300) DS Adv. Grammar and Syntax (4420) DS Spanish Phonetics (4780) DS Beginning Spanish II (1131) DS 1.0000 0.1223 0.0069* 0.0697 Advanced Intermediate Spanish (2201) DS 1.0000 0.1103 0.0063* 0.0597 Advanced Grammar and Composition (3300) DS 0.1223 0.1103 0.9784 1.0000 Advanced Grammar and Syntax (4420) DS 0.0069* 0.0063* 0.9784 1.0000 Spanish Phonetics (4780) DS 0.0697 0.0597 1.0000 1.0000 Table 4-13 looks exclusively at DS. In this ta ble, we notice that th ere are two significant interactions here. The first of these differen ces is shown between Beginning Spanish II (1131) and Advanced Composition and Syntax (4420) while the second occurs between Advanced Intermediate Spanish (2201) and Advanced Composition and Syntax (4420). Table 4-13, however, does not tell us how each course performe d. In order to find this information, we need to look at table 4-14. Table 4-14. LS Means for the interactions of course and modal environment Course Environment Score/LS mean Beginning Spanish II (1131) DS 3.63 (/5) Advanced Intermediate (2201) DS 3.54 (/5) Advanced Grammar and Composition (3300) DS 4.21 (/5) Advanced Composition and Syntax (4420) DS 4.58 (/5) Spanish Phonetics (4780) DS 4.35 (/5)
104 In table 4-14, we see that Beginning Span ish II (1131) received a score of 3.63 while Advanced Intermediate Spanish (2201) receiv ed a score of 3.54. Advanced Composition and Syntax (4420), however, received a score of 4.58, which is the highest score of all the levels. These results clearly show that Advanced Comp osition and Syntax (4420) performed the best out of all three courses mentioned in the DS analysis. Moreover, we observe a slight decrease in accuracy at the intermediate level. Although there is no significant difference between Beginning Spanish II (1131) and Advanced Intermediate Spanis h (2201), these slightly lower scores at the intermediate level are worth addr essing, which we do in section 4.4. First, however, we continue with our modal analysis, this time focusing exclusively on epistemic modality. For organizational purposes, we divide our ep istemic analysis further into two distinct analyses: modal and epistemic context analysis. The modal analysis focuses strictly on the acquisition of one specific modality crossectionally and observes patterns within a specific semantic sub-category. In the epistemic context analysis, however, we focus on each individual level of instruction and the interaction between ES and EI within a particular context. 4.3.2 Modal Analysis of ES We begin this discussion with the m odal anal ysis of ES. Table 4-15 shows that there is a statistical difference between Beginning Spanis h II (1131) ES and Advanced Intermediate Spanish (2201) ES. There is also a statistical difference between Beginning Spanish II (1131) ES and Spanish Phonetics (4780) ES. Table 4-16 shows that Beginning Spanish II (1131) ES received higher scores (2.77) than Advanced Intermediate Spanish 2201 (1.63). Beginning Spanish II (1131) also scored higher ( 2.77) than Spanish Phonetics 4780 (1.96) on ES environments.
105 Table 4-15. P values for difference between c ourses for L2MAC controlling by ES and EI environments Course Level BEG SPN II (1131) EI BEG SPN II (1131) ES ADV INT SPN (2201) EI ADV INT SPN (2201) ES ADV GRM COMP (3300) EI ADV GRM COMP (3300) ES ADV COMP (SYN) 4420 EI ADV COMP SYN (4420) ES SPN PHON (4780) EI SPN PHON (4780) ES BEG SPN II (1131) EI 0.5570 0.9945 0.0550 0.9999 0.9996 0.4925 1.0000 0.9995 0.9469 BEG SPN II (1131) ES 0.5570 0.9999 <.0001* 0.1044 0.9947 1.0000 0.7390 0.9998 0.0243* ADV INT SPN (2201) EI 0.9945 0.9999 0.0018* 0.7693 1.0000 0.9956 0.9942 1.0000 0.3388 ADV INT SPN (2201) ES 0.0550 <.0001* 0.0018* 0.4307 0.0024* <.0001* 0.4743 0.0098* 0.9899 ADV GRM COMP (3300) EI 0.9999 0.1044 0.7693 0.4307 0.8861 0.1294 1.0000 0.9176 0.9999 ADV GRM COMP (3300) ES 0.9996 0.9947 1.0000 0.0024 0.8861 0.9579 0.9993 1.0000 0.4592 ADV COMP SYN (4420) ES 1.0000 0.7390 0.9942 0.4743 1.0000 0.9993 0.6003 0.9989 0.9987 SPN PHON (4780) EI 0.9995 0.9998 1.0000 0.0198 0.9176 1.0000 0.9929 0.9989 0.5449 SPN PHON (4780) ES 0.9469 0.0243* 0.3388 0.9899 0.9999 0.4592 0.0323* 0.9987 0.5449 *p = <0.05 This information reveals a decrease in pe rformance in Spanish Phonetics (4780) on ES environments. Table 4-15 also shows that ther e are significant differe nces between Advanced Intermediate Spanish (2201) ES, Advanced Grammar and Composition (3300) ES, and Spanish Phonetics (4780) ES. If we turn to table 4-16, we see that the scores for these courses are not
106 what one might expect, just as in the analysis of Beginning Spanish II (1131) ES. Recall that Beginning Spanish II (1131) ES received a total of 2.77, while Advanced Grammar and Composition (3300) ES received a total of 2.53. As shown previously, Spanish Phonetics (4780) scored only 1.96. These results indicate an overall steady decline in accuracy among the groups in ES environments. The pattern thus far has indicated a steady decrease in ES accuracy crossectionally rather than an increase. With this said, however, the scores received by Advanced Grammar and Composition 3300 (2.53) show a slight increase from Advanced Intermediate Spanish 2201 (1.63) though the sc ores drop again at Advanced Grammar and Syntax 4420 (2.26) and again in Spanish Phonetics 4780 (1.96). Overall, there does not appear to be a clear pattern for epistemic environments. Table 4-16. LS Means for the inte ractions of course and modal environment (ES Scores only) Course Environment Score/LSMEAN Beginning Spanish II (1131) ES 2.77 (/5) Advanced Intermediate (2201) ES 1.63 (/5) Advanced Grammar and Composition (3300) ES 2.53 (/5) Advanced Composition and Syntax (4420) ES 2.26 (/5) Spanish Phonetics (4780) ES 1.96 (/5) 4.3.3 Epistemic Context Analysis We now begin the second half of the analysis, which as stated previousl y, is referred to as the epistem ic context analysis. In table 4-15 we s ee that there is a statistical difference between Advanced Intermediate Spanish (2201) EI and A dvanced Intermediate Spanish (2201) ES (p = 0.0018). Advanced Intermediate Spanish (2201) is the only course that shows a statistical difference between the ES and EI modal enviro nments. In table 4-17 we see that Advanced Grammar and Composition (3300) EI received a score of 2.16 while ES received a score of 2.53. In this table we also observe that Advanced Composition and Synt ax (4420) EI received a score of 2.92 while ES received as score of 2.26. In Spanish Phonetics (4780) EI, like Advanced
107 Intermediate Spanish (2201) EI and Advan ced Composition and Synt ax EI (4420), received higher scores than their ES counterparts. Beginning Spanish II (1131) on the other hand, performed in the exact opposite wa y. In fact, Beginning Spanish II (1131) EI scores (2.33) were lower than their ES scores (2.77), as seen in Advanced Grammar and Composition (3300). Therefore, we see that for Advanced Interm ediate Spanish (2201), Advanced Grammar and Syntax (4420), and Spanish Phonetics (4780), th e EI contexts were sc ored higher than ES environments. The results for Beginning Sp anish II (1131) and Advanced Grammar and Composition (3300), however, are the exact opposite showing that ES scores were slightly higher than their EI counterparts. In terms of ES and EI developm ent, there does not seem to be any clear developmental pattern for either submodality. The only thing we can claim with any certainty is that, overall, EI scores we re slightly higher than ES scores. Figure 4-3 combines and applies the information found in tables 4-14 and 4-17 to a line chart which illustrates the overall decline in moda l accuracy. Earlier in our discussion, however, we noticed a general decline in accuracy at the intermediate level on L2MAC as well. Recall that the L2 conjugation analysis showed that lear ners in Advanced Inte rmediate Spanish (2201) presented the lowest scores of all five groups. Therefore, in the following section, we address this decline at the intermediate level brie fly before continuing with our analysis. Table 4-17. LS Means for the interactions of course and modal environment Course EnvironmentScore/LS mean Beginning Spanish II (1131) EI 2.33 (/5) ES 2.77 (/5) Advanced Intermediate (2201) EI 2.60 (/5) ES 1.63 (/5) Advanced Grammar and Composition (3300) EI 2.16 (/5) ES 2.53 (/5) Advanced Composition and S yntax (4420) EI 2.92 (/5) ES 2.26 (/5) Spanish Phonetics (4780) EI 2.56 (/5) ES 1.96 (/5)
108 4.4 Decline in Accuracy at the Intermediate Level In order to explain this d eclin e in modal and grammatical accuracy, we refer to the Ushaped language diagram (Kellerman 1985a, 1986). U-shaped development can be explained as L2 learners brief decline in accuracy as a resu lt of increasing IL sophistication and complexity (Lightbown 1985). In other words, at the time of our study, it is posit ed that the participants at the advanced-intermediate level were in the proc ess of restructuring their IL systems when L2MAC was administered. According to Lightbown (1985) an increase in error rate in one area may reflect an increase in complexity or accuracy in anothe r, followed by an overgeneralization of a newly acquired structure, or simply by a sort of overloa d of complexity which forces a restructuring, or at least a simplification, in anot her part of the system (177). In the current study for example, Beginning Spanish II (1131) scored higher than Advanced Intermediate Spanish (2201) on L2MAC on two of three modal interpretation ta sks (DS and ES) as indicated by the higher scores. It was the case in this particular study, however, that l earners in Beginning Spanish II (1131) had just been introduced to the present Spanish subjunc tive shortly before the study was administered. Therefore, alt hough this drop at the Advanced Intermediate level (2201) is alarming at first, we can suggest the following: (1) learners in Begi nning Spanish II (1131) performed better on L2MAC because of recent expos ure to the present Spanish subjunctive; and (2) the lower scores support U-Shaped development. If we apply the results retrieved from these two groups to Lightbowns claim above, they do a ppear to fit the U-shaped pattern. The idea presented in Lightbown and Spada (2006), is base d on the Kellerman U (Kellerman 1986) which suggests that as learners knowledge of L2 increases, they pass through a period of decreased accuracy.
109 Figure 4-3. Line graph of DS, ES, and EI mood scores To better illustrate this idea, we use the example espero que [ pro] bailes (I hope you dance) and apply the U-shaped diagram proposed by Kellerman. Moreover, in order to make this example more relevant to th e current study, we apply the course levels most affected by this brief decline: Beginning Spanish II (1131), Advanced Intermediate Spanish (2201), and Advanced Grammar and Composition (3300). The Kellerman model illustrates how learners scores drop temporarily and rise after increased ex posure to L2. The correct form of the sentence espero que [ pro] bailes is shown in the upper left hand corner of the U-shaped diagram in figure 4-4. This example represents the successful output learners produce when they are first introduced to the deontic subjunctive in nominal clauses, for example. At the bottom of the diagram, however, as in L1, we see the incorrect form (i.e., V2 = indicative) showing a decline in learner accuracy. According the figure shown below, after this brief decline in accuracy, learners are able to use the subjunctive correctly once they have incorporated the st ructure in to their IL systems.
110 Figure 4-4. Example of error production and correction in devel oping IL systems The sentence espero que [ pro] bailas (I hope you dance) in point (2) for example, is a possible representation of learners tendency to overgeneralize rules or verb forms in L2 during the learning process. Point (1) for example, in th e upper left hand corner of the U, represents learners accurate use of the subjunctive when th ey are first introduced to it. The idea is that learners have the ability to use the subjunctive mood correctly when they are first exposed to this grammar point. As the IL system progresses, ho wever, errors occur as a result of increased sophistication as shown in point (2 ). Point (3), located at the top right hand side of the diagram, represents learners ab ility to semantically and morphos yntactically use th e subjunctive mood correctly once they have moved thro ugh the acquisition process successfully. Although L2MAC is the more traditional of the two tests designed for the current research, we were able to te st learners mood and conjugation abilities, which, as stated previously, are combined into one variable referred to here as ability. We were also able to show that learners received nearly identical sc ores on modal interpretation as they did on verb
111 conjugation. We confirmed that DS modality in nominal clauses consistently received higher scores than EI and ES semantic environments. In addition, we addressed the overall decline in accuracy at the intermediate level and suggested that learners grammatical accuracy increased again in latter stages of ac quisition showing support for Kell erman (1986) and Lightbown and Spada 2006). With these ideas in mi nd, and after having presented th e reader with the statistical analysis of our data, we revisit the research questions that motiv ated this study. Our goal here is to answer each question by applying the results and data presented in the previous section. First, we address RQ1 which we answer using our findings from ME. Second, we answer RQ2 referring to our fi ndings from L2MAC. 4.5 Research Question 1 The first research question asked how does the subordinate m arker que impact learners judgment of grammaticality of co mplex sentence structures (S1, V1 + S2, V2) and was tested using ME. Results of this task suggest that the complementizer que had an effect on the participants syntactic interpretation of the st ructures. We found that the appearance of que in syntactic environments 1 (+ que + subjunctive), 2 (+ que + infinitive), and 3 (+ que + indicative) received the second highest ac ceptance ranking. In addition, we discovered that there was no significant difference between these environments meaning that any complex structure that included the complementizer que received approximately the same score. With this said, however, the results also showed that environment 4 (que + subjunctive) was sc ored as the least acceptable environment revealing that learners were able to recogni ze that a subjunctive verb in the subordinate clause should be preceded by the complementizer que in nominal clauses. The most interesting conclu sions that are drawn from th e results of ME, however, lie within the ranking of environments 5 (que + infinitive) and 6 (que + indicative). Recall that environments 5 (que + infinitive) and 6 (que + indicative) were ranke d as the highest (i.e.,
112 most grammatical) of the six environments. If we consider the possibility that L1 syntactic transfer is at play, as we discu ssed in chapter 1, we can suggest th at the participants of this study ranked environments 5 (que + infinitive) and 6 (que + indicative) as the most grammatical because they are the structures most commonly used in L1 irrealis. In order to verify this claim, we analyze English syntax in greater detail in the following paragraphs. We show that our analysis fits the criteria of L1 transfer discussed in Zobl (1980) (see section 1.4.4). In order to identify which of these six syntactic environments yields to grammatical utterances, we need to analy ze them according to language. In the following discussion, we analyze these subordinate clauses as if they were preceded by a c onjugated verb in the indicative mood. By discussing each environment as part of a whole sentence, the structure becomes clearer. In table 4-18 for example, we see that only environments 1 (+ que + subjunctive) and 3 (+ que + indicative) are accurate structures in Spanish. Why then, were these structures not ranked higher by learners? The firs t answer to this question comes from the results of ME which, as stated previously, reveals that learners found English-like structures (i.e., environments 5 and 6) to be more grammatical than the other syntac tic environments. With this said, however, which of the ME syntactic environments analyzed here are actually acceptable in English? Table 4-18. Possible subordinat e structures in Spanish Environment Structure Possible subordinate environment Example 1 + que + subjunctive Yes Quiero que hable Juan. 2 + que + infinitive No *Quiero que hablar Juan 3 + que +indicative Yes Creo que habla Juan ingls. 4 que + subjunctive No *Quiero hable Juan. 5 que + infinitive No *Quiere t hablar con Juan. 6 que + indicative No *Quiero habla Juan.
113 To answer this question, we provide the reader with table 4-19 which offers examples of grammatical subordinate structures in English. Fo r this analysis, we are also considering overt CPs that precede indicative verbs. Table 4-19. Possible subordinat e structures in English Environment Structure Possible subordinate environment Example 1 + that + subjunctive Yes (unco mmon) I prefer that he go now. 2 + that + infinitive No *I prefer that he to go now. 3 + that +indicative Yes I believ e that he speaks English. 4 that + subjunctive Yes I prefer he go now. 5 that + infinitive Yes I want him to go now. 6 that + indicative Yes I believe he speaks English. In table 4-19, we notice that there are more e nvironments that represent accurate syntax in English than in Spanish. In English only environm ent 2 (+ that + infinitive) is unacceptable when preceded by an overt CP, for example. And although authentic subjunctive constructions, like that of environment 1 (+ that + subjunctive), so und artificial in English, they do exist, and are therefore grammatical. Environments 5 (that + infinitive) and 6 (that + indicative) are the most commonly used structures in English irrea lis. This perhaps explains why learners rated these environments as the most acceptable in L2. Furthermore, by looking at table 4-19, we notice that English permits a wider variety of synt actic structures, two of which include overt CP complements. Therefore, if L1 English speakers learning Spanish have access to CPs in L1, shouldnt they have access to CPs in their L2? Several studies have argued that functional categories, such as CPs for example, are accessible during L2 acquisition from the very beginning. Epstein, Flynn, and Martohardjono (1996) for example, suggest that based on the re sults of their studies on child second language acquisition of English and adult second language acquisition of Tu rkish, the lack of functional
114 elements should not be taken as evidence for the absence of the associated functional categories (148). Prevost and White (2000) agree with Epstein et al. as they state that variability in adult L2 performance does not reflect a de eper lack of functional categorie s or features associated with tense and agreement. Rather, L2 learners have difficulties with the overt realization of morphology (104). In other words, and more specific to our study, is the suggestion that even though learners did not rate structures with ove rt CPs as more grammatical than structures without overt CPs, this does not mean that learne rs lack access to CP f unctional categories in L2. This provides further evidence that the problem is not that learners lack access to CPs in L2. Rather, learners are unaware of the [+ optional] overt CP rule in English. As a result, learners transfer this [+ optional] overt CP rule from English to Spanish. In our study, we conclude that L1 transfer of the [+ optional] overt CP rule to L2 caused learners to accept ungrammatical syntactic environments as grammatical. By comparing tables 4-18 and table 4-19, we better understand w hy learners ranked environments 1-6 as they did on the ME tas k. For example, not only are environments 5 (que + infinitive) and 6 (que + indicative) grammatical structur es in English, they are the most commonly used. Consequently, lear ners ranked environments 5 (que + infinitive) and 6 (que + indicative) as the most acceptable syntactic environments in L2. Environment 4 (que + subjunctive), on the other hand, was ranked the l east acceptable, indicating knowledge of FCs, while environments 1 (+ que + subjunctive), 2 (+ que + infinitive), and 3 (+ que + indicative) were ranked between the 5/6 and 4. 4.6 Research Question 2 The second research question asked with what a ccuracy do L2 learners at different levels of Spanish correctly identif y m ood and conjugate verbs us ing corresponding morphology in nominal clauses? In relation to RQ2, the results of L2MAC were interpreted to mean that mood
115 and conjugation skills could not be teased apart. The results also showed that the more advanced levels performed better overall than the beginni ng and intermediate levels on L2MAC. Such performance demonstrates that modal and synt actic abilities did increase with time and/or continued exposure to Spanish. We also discussed the fact that learners rece ived higher scores on DS modality and that there were no clear acqui sition patterns for EI and ES modalities. The difficulties learners have with epistemic modality ma y be attributed to its layered and gradient (a term used by Givn 1994) properties. As a result, though we cannot support this claim with our data, we suggest that EI and ES require greater time to develop. Figure 4-5 better illustrates epistemic complexity. Figure 4-5 provides the reader with a summary of the Spanish modal system. This diagram more clearly illustrate s the complexity of both grammatical mood and semantic modality in Spanish. Notice here that epistemic modality is more complex than deontic modality. Epistemic modality is split in to two di stinct branches; EI (epistemic indicative) and ES (epistemic subjunctive). Fo r EI modal environments (i.e., creer to believe), learners must identify the following lexemic properties of V1: [+ reality], [+ existence], [+ belief], and [subjunctive]. In other words, in order to identify a context as EI, learners need to know that EI lexemes refer to something factual, or perceived as factual, by interlocutors. ES environments, on the other hand, work in the opposite way. In ES contexts, learners find lexemes that possess the following features: [reality], [existen ce], [belief], and [+ subjunctive] and therefore require knowledge of non-existence and low truth value. In the next section, we summarize our findings before continuing to chapter 5 where we apply them to X-Bar Theo ry and revisit the IL structures presented in chapter 1.
116 Spanish Modal System Figure 4-5. Complexity of Spanish Modal System 4.7 Summary of Findings In this section, we answered both of the research questions that m o tivated this study. In answering these research questions, we found that the results of the pres ent study are compatible with the tenets and predictions of the following theories: L1 transfer and Missing Surface Inflection Hypothesis (MSIH). L1 transfer helped to explain how L1 irrealis morphosyntaxic properties are transferred from L1 to L2 during acquisition. MSIH was used to suggest that even though learners considered syntactic environments 5 (que + infinitive) and 6 (que + indicative) as the most grammatical structures on ME, this does not reflect learners inability to access functional categories in their IL. Therefor e, to suggest that functional categories are missing in L2 would not be an accurate conclusion. Recall that table 4-19 shows that there are structures in English that require (null) CPs m eaning that we cannot claim learners lack access to functional categories in L2 if th ey already have access to FCs in L1. What does seem to be happening, however, is that learne rs do not make the connection between required overt CPs in
117 Spanish and optional overt CPs in En glish. In other words, learners appear to transfer the L1 [+ optional] overt CP rule of English over to Spanish. In chapter 5, we revisit the syntactic struct ures previously introduced in chapter 1 and apply them to our results and fi ndings. We also discuss the cont ributions our study has made to the field of SLA, suggest directions for futu re research and pedagogical enhancement, and conclude with a brief summary of our goals and general findings of the present study.
118 CHAPTER 5 CONCLUSION 5.1 Introduction The purpose of this research was to investig ate the acquisition of the Spanish subjunctive in nom inal clauses by native speakers of English. In the process of doing so, we have looked at and reviewed a variety of structural issues relevant to L2 s ubjunctive acquisition including the lexical selection of IP and CP complements in nominal clauses. We have also addressed the morphosyntactic and semantic developmental diffi culties facing L2 learners using a generative approach to grammar analysis. In addition, we used Chomskys X-Bar Theory and the Principles and Parameters framework to help us explai n the L2 acquisition pr ocess of the Spanish subjunctive. In chapter 1, we proposed a series of plausible IL structures thought to represent learners developing systems in L2. Some structur es used IP complements while others used CP complements. In this chapter, we return to th ese diagrams and argue that our data support IL structures that include CP complements. We show that learners have access to FCs in L2 because they have such access in L1. Furthermore, we show that learners are transferring the L1 [+ optional] overt CP rule to L2 which explains w hy they use not only the indicative and infinitive morphology, but also why [ que ] structures were rated as high ly acceptable. In the following section, we briefly return to the most relevant syntactic structures pres ented in chapter 1 and review them in relation to our findings. In the latter sections of this chapter we review the contents of the present project, discuss peda gogical implications, and make suggestions for future research. 5.2 IL Representations Recall that in chapter 1, we pr esented a series of syntactic diagram s which we introduced as expected IL representations. Among these re presentations, a few of the structures were
119 comprised of subordinate clauses with the followi ng properties: (1) null CP + V2 infinitive; (2) null CP + V2 indicative; and (3) IP + V2 infin itive; (4) IP + V2 indicative. We used these diagrams to show that learners were either using null CPs in subordinate clauses or reducing complex syntax in L2 to fit commonly used IP templates in English. We argue here, however, that the findings of our study suppor t the existence of CPs in IL. Th is means that learners are not reducing complex sentence structures to fit reduced IP templates. In stead, we argue that learners are actually including null CP co mplements in their structures. Collentine (1995) discovered that what appear ed to be complex sentence production in L2 at the intermediate level was actua lly a series of simple sentences structures. In our study, unlike some of the previous studies mentioned in chapte r 2, we have data that supports learners access to complex sentence structures. In other wo rds, even though production data often shows no signs of such access, as shown in Terrell, Ba ycroft, and Perrone (1987) and Collentine (1995), our data shows that learners are including null CP s in their structures. On the following pages, we revisit IL representations that were introdu ced in chapter 1 and discuss their plausibility based on the results discussed in chapter 4. We revi ew IL structures that include both IP and CP complements arguing in favor of the latter. Figures 5-1 and 5-2 are based on ME syntactic environment 6 (que + infinitive). Recall that learners rated this environment as one of the most acceptable structures in the series of syntactic environments. Figures 5-1 and 5-2 pres ent the possibility of V1 selection of an IP complement. If we assume that figures 5-1 and 5-2 ar e accurate IL representations, we need to show that CPs are inaccessible in L2. This woul d have been a legitimate assumption if learners had considered environment 4 (que + subjunctive) as equally as acceptable as environments 5 (que + infinitive) and 6 (que + indicative) in ME.
120 Figure 5-1. A syntactic representa tion of IL syntax during L2 subjunctive acquisition: IP complement (* quiero t venir ) Figure 5-2. A syntactic represen tation of IL syntax during L2 subjunctive acquisition (* Espero t venir )
121 This was not the case, however, as discussed in chapter 4. In fact, learners considered environment 4 (que + subjunctive) as the least grammati cal structure of the six syntactic environments presented to them. The low ranking of environment 4 (que + subjunctive) indicates knowledge of CP in L 2. Based on this analysis, we c onclude that figures 5-1 and 5-2 cannot be accurate representations of IL. Cons idering figures 5-1 and 5-2 as expected IL representations would be highly in accurate because such representations conflict with the results of our study. Figures 5-3 and 5-4, on the other hand, are cons idered highly expected IL representations. These figures more accurately reflect the results of our data as they include the possibility of CP complements. For example, in figure 5-3, we sugge st that learners are us ing a null CP with V2 [ind] morphology. This structure mirrors common i rrealis structures in English as seen in I hope (that) you come where the overt CP is optional. Even though learners often resort to using indicative morphology instead of subjunctive morphology in equivalent L2 structures, at least now we have data that provides support for CP acce ss. In addition to showing that learners have access to complex clauses in L2, we suggest that they are transferring the L1 [+ optional] overt CP rule to L2. In other words, if learners are accessing CP structures in IL, it must be that CP complements are not represented overtly. With such morphological inaccuracies in L2, we can also ask questions about how learners are us ing infinitive and indicative verbs in L2. For example, if we conclude that learners have access to CP complements, should we assume that learners use finite [+indicative] and infin itive verbs interchangeably? Take the English equivalent to espero t bailas (I hope (that) you dance), fo r example. While in Spanish the accurate construction includes [optional] overt CP + V2 [sjv], the English version requires
122 [+ optional] overt CP + V2 [i nd]. Conversely, a sentence such as I want you to go in English requires an IP + V2 [-finite] complement structure while the Spanish equivalent of this sentence follows the same pattern as the previous Spanish example (i.e., [-optional] overt CP + V2 [sjv]). Considering that both environments 5 and 6 were rated as the most highly acceptable syntactic structures in the series, this may be an accurate assumption. The only difference between figures 5-3 and 5-4 is the use of th e indicative verb vienes in figure 5-3 and the use of the infinitive verb venir in figure 5-4. We consider the possibility that learners sometimes use infinitives as they would indicatives. If our analysis of CPs is accurate, then this analysis makes sense. In other words, even if the learner is not conjugating the verb accurately, this does not mean that the learner is actually using this verb as an authentic infinitive. Figure 5-3. A syntactic representation of IL syntax dur ing subjunctive acquisition (* Espero tu vienes. )
123 Figure 5-4. A representation of IL sy ntax during subjunc tive acquisition (* espero t venir ) We are able to support the id ea that figures 5-3 and 5-4 are possible IL representations because we are basing them on intuition data retr ieved from ME. ME has provided us with solid evidence that learners have access to FCs in L2. Th erefore, if learners have access to FCs in L2, this means that they have access to the comple x syntactic architecture necessary to construct subjunctive sentences. So why th en, do learners continue to le ave out overt CPs in oral and written production as shown in previous studies (see Terrell, Baycroft, and Perrone 1987 and Collentine 1995)? And why do learne rs continue to use inaccura te morphology? Once again, the data from ME has supported the id ea that learners are transferring L1 rules to L2. This suggests that learners use infinitive and indicative verb s interchangeably in finite contexts leaving mismatched features in I. To provide another example, we refer to 5-3 *Espero t vienes.
124 Notice that this sentence, though it is incorrect in Spanish, follows the exact same pattern as its English equivalent I hope (that) you come. This provides us with evidence that the following process is taking place: Application of L1 rule [+ optional] overt CP L2 [-optional] overt CP rule = *error 26. *Espero t come. I hope-1p-sing. [you-2p-sing.-ind. come.] I hope you come. Our ability in this study to present expected IL repres entations that include CP complements and provide support for such structures is one of the strongest contributions that our study has made to the field of SLA. By usi ng ME, we have gained access into learners IL systems revealing access to complex syntactic stru ctures in L2. In the following section, we consider other ways this study ha s contributed to the field of SL A. In doing so, we discuss how we may improve this study, or si milar studies, in the future. 5.3 Contributions and Implications The current project has shown that Magnitude Es timation is able to tap into learners intuitions revealing learners pref erence for English-like syntactic structures in L2. ME has also indicated that such structures in clude null CPs. Previous studies, such as Terrell, Baycroft, and Perrone (1987) and Collentine (1995), used prod uction data to draw conclusions about IL and were therefore unable to separate performance ab ilities from learners actual knowledge of the language. These studies did not show evidence of learners access to complex syntax in L2. Performance data can be problematic as discus sed in Hawkins and Chan (1997) and therefore may not present an accurate picture of L2 ac quisition. The idea that th ere is a distinction, however, between competence and performance requi res further explanation. In order to expand on this point, we turn to Radford (1997). Radfor d defines performance as a term which denotes observed language behaviors, e .g. the kind of things people actually say when they speak a
125 language, and what meanings they assign to sentences produced by themselves or other people. Performance can be impaired by factors such as tiredness, drunkenness, etc (268). Competence, on the other hand, refers to a pers ons innate knowledge of langua ge. So why do we feel that a test that would analyze learners competence in stead of their performance was the better choice for this study? In the present project, we agree with Radford (1997) and his definition of performance. Like Radford, we suggest that perf ormance data does not always reflect learners actual knowledge of the language. For example, in the production studies mentioned throughout this project, not one of them would be able to make the clai m that learners intuitively find structures with V2 subjunctive morphology in a [ que ] syntactic environment ungrammatical. This is a very important discovery because it provides additional support for learners access to complex syntax. We can also make this claim with greater confidence because we were able to test a wide variety of syntactic environments. In L2 production data, it is common to observe missing CPs and subjunctive morphology, as shown in Collentine (1995), for example. What is not common, however, is evidence indicating that learners access null CPs in L2. By using ME, we were able to test a variety of very speci fic syntactic environments which clearly showed learners preference for L1 structures. In production data, as Radford points out, we all make occasional slips of the tongue, or occasionally misinterpret what someone else says to us. However, this doesnt mean that we dont know our native language, or d ont have competence (i.e., fluenc y) in it (2). Therefore, by studying learners competence throu gh intuition data, we are not relying on learners limited ability to produce subjunctive structures in L2, as was the case in Terrell, Baycroft and Perrone (1987), and Collentine (1995). For this particular study, we suggest that competence data was less likely to be affected by fati gue, or other issues that would be more prevalent in average L2
126 classrooms, and therefore problematic. Furthermore, the intuition-based data retrieved by ME in the present research has made it possible to attemp t to construct plausible representations of IL systems during subjunctive acquisition. We would not have been ab le to make these assumptions had we relied on production data. Had we used production tasks exclusively, we would have been limited to, and dependent upon, only what learners provided us with. As mentioned previously, both oral and written production at the undergraduate level is usually very limited and does not allow testing of specific syntactic environments. With this said, however, L2 production data is not without merit. We return to this notion in la tter sections of this chapter. With such limited production, it would have b een difficult for us to make conclusions about learners intuiti ons. For example, Collentine (1995) gi ves examples of the production data that he gathered for his L2 subjunctive acquisi tion study (for a full di scussion see chapter 2). Recall that Collentine concluded th at learners at the intermedia te level were unable to produce complex subjunctive structures in L2. He found that participants often used parataxis instead of complex syntax as shown in the foll owing examples (Col lentine 1995, p. 128). 27. *No pienso es justo. No I think 1p-sing. is3p fair. I dont think its fair. 28. John quiere yo voy. (a) John wants 3p-sing. I-1p go. (b) John wants me to go. 29. Carlos come una ensalada, yo estudio en la biblioteca. (a) Carlos eats 3p-sing. a salad, I1p study in the library. (b) Carlos eats a salad, I eat in the library. In examples 20-22 we observe samples of what learners are capable of producing on their own. However, what we cannot observe is whether or not learners are transferring L1 rules to L2. In order to be able to make this claim, we need evidence that learners intuitively select
127 English-like structures from a seri es of plausible L1 and L2 co mplement structures. This is exactly what we were able to achieve in th e present study. We suggest that by accessing what complement structures learners perceive as acceptable and unacceptable in their L2, we can access the root of the production errors. This way, we can ask why learners make such errors as opposed analyzing what errors they make (i.e., competence rather than performance). In our study, we found that learners found English-like s yntactic structures more acceptable in L2. Contrary to ME, L2MAC required learners to conjugate verbs according the modal environment of each item. L2MAC was a very controll ed task meaning that we were able to test three distinct modal environments (DS, ES, EI) w ithin nominal clauses. It would have been very difficult to elicit such specific modal contexts fr om participants in a full production task. In addition, deontic and epistemic subjunctive modalities have not previ ously been discussed to this extent in L2 Spanish studies to our knowledge. We have taken this study one step further than many others by splitting epistemic modality into EI and ES subcategories. For example, we provide evidence that even at adva nced levels of L2 subjunctive ac quisition learners continue to struggle with the complexity and gradient prope rties of epistemic modality. Although this study was successful in helping us to paint a clearer pi cture of learners IL sy stems and reveal related syntactic-modal issues, there were limitations that need to be addressed. Therefore, in the following section, we identify the limitations of our research a nd offer suggestions for future studies. 5.4 Limitations of Study In this section, the lim itations of the presen t research are discusse d along with suggestions for future research. Among these suggestions are the inclusion of native Spanish speakers (NSS) as a control group, superior non-native speak ers of Spanish (SNNS) as an additional experimental group, the incorporation of producti on tasks designed to observe verb raising (or
128 lack thereof) in L2, and finally, equal distribution of both matrix (V1) and subordinate verbs (V2) in ME tasks. Native speakers of Spanish (NSS) and superi or non-native speakers (SNNS) of Spanish should have been included in this study. The NSS group should have been added as a control group while SNNS should have been included as a superior L2 group. By including a NSS group as the control group, it would have been ea sier to make conclusi ons about L2 acquisition of the Spanish subjunctive. By comparing the results of the NSS group to the SNNS group, we would have been better equipped to compare the accuracy of L1 and L2 subjunctive intuitions. For example, what if L1 and L2 Spanish speaker s presented opposite resu lts? What if the NSS group found environments 1 (+ que + subjunctive) and 3 (+ que +indicative) to be the most grammatical as opposed to 5 (que + infinitive) and 6 (que + indicative)? Such results would further confirm the validity of the present study and support our findings. Conversely, what if the SNNS group presented results simi lar to those of the NSS group? In this case, we could have made the claim that learners no longer rely on L1 at superior levels. On the other hand, if the SNNS group continued to rank environments 5 ( que + infinitive) and 6 (que + indicative) as the most grammatical structures, we could have made the claim th at, even at highly advanced stages of acquisition, learners continue to rely on L1. The inclusion of the NSS and SNNS groups are not the only factors th at we would include in a replication or similar study. We would al so incorporate additional oral and written production tasks to more effectively illustra te the difference between competence and performance abilities in L2. As explained in prev ious chapters, we chose to use ME so that we could test a wide variety of sp ecific linguistic environments th at learners would not produce on their own. With this said, however, more L2 pe rformance data would have helped us to make
129 clearer assumptions about learne rs syntactic producti on abilities. Furthe rmore, White (1992) suggests that the combination of the oral product ion and preference tasks, like those used in her study, offered reasonable insight into learners IL systems (285). Therefore, a combination of both competence and production data may have incr eased the validity of th e overall test design. Production tasks have also been useful in pr evious studies conducted exclusively on the Spanish subjunctive as shown in Terrell, Baycro ft, and Perrone (1987) and Collentine (1995), for example. Recall that by using two different forms of production tasks, Terrell et al discovered that even though learners could successfully co mplete subjunctive sent ences in rote grammar exercises, they could not use the subjunctive in oral speech. By using oral production tasks, Collentine found that learners at the intermediate level we re combining simple sentence structures that originally app eared to be complex. He was able to make the claim that even though learners at the Advanced-I ntermediate level appeared to be producing complex sentences, a close analysis revealed that learners performance was missi ng necessary syntactic components such as the overt CP, for example. In our study, a production task would have allowed us to see whether or not learners actually produced Eng lish-like sentences in oral speech or in a free writing task. It would have been in teresting to see if learners used English-like stru ctures on such tasks and whether or not their output directly mirrored their competence. Moreover, production data might also have provided further insigh t into learners verb raising strategies and perceptions of finite an d non-finite V2 forms. It is suggested that future studies control for verbs in bot h the principle and subordinate clauses. Although we do not have data to support this notion, there is the possibility that learners grammatical judgment was affected by different le xemes. As discussed in chapter 1, each lexeme encodes distinct semantic and syntactic proper ties and may therefore re quire equal distribution
130 on ME tasks. Some verbs, for example, have ir regular conjugations in the indicative and/or present subjunctive (i.e., ir ) which may alter learners perception of grammaticality (see Collentine 1997). Most importantly, however, as di scussed in chapter 1, is the idea that each lexeme possesses distinct linguistic properties and therefore behaves idiosyncratically (see Herchenson 2000 and Speas 1990). In the present study, the distribution of verbs in ME is organized according to environment. The verbs ar e categorized according to the specifications of the environment in which they appeared. Take the verb encontrar (to find) in environment 2 (+ que + infinitive), for example. Generally speaking, this verb is considered a stem-changing verb because the mid-back vowel o, found in the second syllable, changes from a vowel to a diphthong in the present subjunctive (i.e., ella encuentre (she finds). 1. Syntactic environment 1 (+ que + subjunctive): (2x) estudie (he/she studies), coma (he/she eats), (1) compre (he/she buys), (1) viva (he/she lives), tenga (he/she has) 2. Syntactic environment 2 (+ que + infinitive): seguir (to follow), cantar (to sing), hablar (to speak), encontrar (to find), vivir (to live), buscar (to look for) 3. Syntactic environment 3 (+ que + indicative): se casan (they get married), comprende (he/she understands), existen (he/she exists), se mudan (they move), llega (he/she arrives), sigue (he/she continues) 4. Syntactic environment 4 (que + subjunctive): lean (they read), sea (it, he, she is), suba (he/he rises), tenga (he/she has), ayude (he/she helps), se preparan (they prepare themselves), adelgace (he/she gets thinner/loses weight), fume (he/she smokes) 5. Syntactic environment 5 (que + infinitive): buscar (to look for), encontrar (to find), sobrevivir (to survive), pagar (to pay), tener (to have), hablar (to speak), ser (to be), estar (to be), conocer (to know) 6. Syntactic environment 6 (que + indicative): fuma (he/she smokes), tiene (he/she has), es (he/she is), compra (he/she buys), toma (he/she takes), vuelve (he/she returns), tomamos (we take), tiene (he/she has) In this analysis, however, we categorized the infinitive form of the verb encontrar (to find) as an ar verb because in the infinitive form, st em changes are not required as the second syllable is not accentuated. The verb tener (to have) in environment 4 (que + subjunctive),
131 though, was categorized as an irregular verb b ecause the present subj unctive form does not follow the morphological patterning of a regular er in the subjunctive. If we look at the use of verbs in environmen ts 1-6, we notice that the distribution is unequal. For example, in syntactic envi ronment 1, we indicate that the verb estudie was used twice by placing a (2x) to the left of the verb. Whether or not the innate lexical properties of these subordinate verbs affect ed the participants grammati cal judgment of the syntactic environments is unknown and therefore it is not possible to make any co nfirmations regarding this issue. With this said, however, we remind the reader that syntactic environments 5 (que + infinitive) and 6 (que + indicative) were ranked as more gr ammatical than the others. This is worth mentioning because environment 5 (que + infinitive) did posse ss the largest overall quantity of regular (i.e., as defined in this study) ar,er, andir verbs. A total of nine verbs pertaining to this category were us ed in environment 5. Environment 6 (que + indicative), however, possessed only 5 regular verbs which wa s equal to the amount of regular verbs found in environment 1 (+ que + subjunctive). Despite the unequal distribution of V2 verbs, we actively avoided the use of uncommon and irregular verbs as much as possible in order to make the task accessible for participants at all levels of instruction. Now that we have identified the limitations of our study, we turn to the pedagogi cal implications. As shown in chapter 4, and discussed in the present chapter, we have evidence that learners are appl ying the L1 [+ optional] overt CP rule in L2. This means that we need to capitalize on such information and use it in L2 classrooms. 5.5 Pedagogical Implications Spanish subjunctive instruction has undoubtedly changed over the last 30 years. Exam ples of such pedagogical evolution can be seen in the studies mentioned here. Blake (1985), for example, draws upon previous L1 and L2 subjunctiv e acquisition research that includes Bergen
132 (1978), Blake (1983), Gili & Gaya (1961), among others, to o ffer suggestions for teaching subjunctive in the L2 classroom. In doing so, Blake states that on the basis of both ease of learning and frequency counts from adult speec h, we would argue that the classroom teacher should concentrate on commands [deontic nominal] and adverbial expressions (e.g., para que, a que, cuando, hasta que, aunque, despus que, antes que) in the beginning levels (172). According to Blake, presenting the subjunctive in groups such as these situates them in similar semantic categories that can be described as yet-to-be-realized-event s. Collentine (1997) emphasizes the need to raise the communicative value of the subjunctive by creating techniques that would help learners make connections betw een the formal and semantic properties of the subjunctive. More recently, however, based on previous input studies, Ma ssery (2008a) focuses on the importance of incorporating interpretation tasks into classroom and homework activities and proposes that modal identification abilities emerge prior to morphosyntactic production. Such an approach would alter the focus of subjunctive instruction from production to modal identification. This type of instruction would involve exercises that require students to read texts, check boxes, circle answers, underline concepts, etc; based on a related text or listening passage. These are just a few of the many studies that have contributed to the pedagogical enhancement of L2 subjunctive. Taken together, th ey provide instructors w ith a variety of ways to compare and contrast English and Spanish st ructures through readi ng, contrastive analysis, and PowerPoint presentations. Even with all of this information, however, something is still missing in the L2 classroom as the subjunctive remains to be somewhat of linguistic mystery. Based on the current study, we encourage teachers to present nominal clauses of deontic modality prior to other modal-syntactic environmen ts, especially at lower proficiency levels. If instructors present the subjunctive to learners for the first time using DS nominal contexts
133 ( Quiero que comas mejor I want you to eat better) for example, it is suggested that learners will be exposed to complex syntactic structur es without having to concentrate on difficult semantic nuances like those found in ES ( No creo que conozcas a Juan, I dont think you know John) and EI ( Creo que conoces a Juan I think you know John) environments. This suggestion is supported by the resu lts of L2MAC that showed learners scored higher on DS contexts than ES contexts. In addition to our results recall that previous research such as Terrell, Baycroft, and Perrone (1987) and Collentine (1995) have shown that nominal clauses are the most prolific in L2 production. This suggests that learners ar e better able to use nominal structures in L2. Even with sugge stions such as these, however, the subjunctive st ill remains one of the most difficult and complex components of Spanish grammar for L2 learners of Spanish.As as result, in order to facilitat e subjunctive acquisition in the language classroom, one must not assume that technological bells and whistles wi ll work their magic overnight, nor should one assume that dry, rote, production tasks are suffici ent either. Instead, perh aps it is a combination of all of the above. In every cla ssroom it is ultimately the decision of the instructor as to how she will conduct the class and what activities she will use with her students. The use of contrastive analysis in English a nd Spanish when discussing the subjunctive is also thought to be a useful pedagogical strategy. In the present study, ME shows that learners find English-like structures more grammatical than complex, L2 structures. Therefore, instructors should capitalize on su ch information and use examples in English when presenting Spanish subjunctive structures. In other words, if an instructor were to ask learners what the Spanish equivalent is to the sentence I want you to go to the store (i.e., IP complement), based on our findings, they might say *[pro] quiero tu ir a la tienda instead of [ pro] quiero que [ pro] vayas a la tienda If the instructor can use such errors to point out the sy ntactic difficulties
134 learners have as a result their L1 intuitions, subjunctive structures might become more salient to learners. By presenting learners with irrealis structures and au thentic subjunctive examples in English, it is further suggeste d that learners might better unde rstand how the subjunctive works in both languages (Feeny 1978; Kilr oe 1988). In addition, learners should be made aware that English uses a variety of struct ures to capture irrealis meaning while Spanish almost always follows the same pattern. This is something that l earners had not fully incorporated into their IL systems at the time of the study. If learners continue to a pply the [+ optional] overt CP rule in Spanish, a deductive approach might help learners to visualize the structural differences between L1 and L2. It appears that learners access CP structures in IL without mee ting the overt requirements of Spanish. In other words, L2 learners may be acce ssing CP, as they would in English, but they are not phonologically and orthographically producing que as a result of L1 optionality. A deductive approach in this case would allo w learners to see that their mist akes are essentially minor and in need of some tweaking. We sugge st providing learners with ex amples of the three English irrealis verbs that we discussed in chapter 1 ( hope prefer and want ) and their corresponding syntactic structures. These verbs could be presented alongside their Spanish equivalent structures. A side-by-side comparison of the English verbs hope prefer and want and the Spanish verbs esperar preferir and querer, would more clearly illustrate their intricate lexemicmorphosyntactic differences and idiosyncratic behaviors. The following table provides an example of how we might present such in formation in an L2 Spanish classroom.
135 Table 5-3. Sample presentation of English ir realis and Spanish subjunctive structures English irrealis [+ optional] overt CP rule Spanish subjunctive [optional] overt CP rule I hope you dance. Espero que bailes. I hope that you dance. Espero que bailes. She prefers you dance. Prefiere que bailes. She prefers that you dance. Prefiere que bailes. They want you to dance. Quieren que bailes. We recommend emphazing the [+ optional] overt CP rule in English by bolding that 5.6 Closing As m entioned previously, there are several important contributi ons this project has made to the field of second language acquisiti on. ME is a tool that allows the participan ts to take control of their reactions to linguistic stimuli unlike the more traditio nal grammaticality judgment tests for example. Bard, Robertson, and Sorace (1996) support the strength of ME by saying that its [ME] validity comes from intra-subject consistency, which is easily achieved with instructions that encourage subj ects to make full use of the numerical scale in expressing their impressions (63). Therefore, by introducing ME to the grammaticality judgment testing of L2 subjunctive syntax, we ha ve increased the level of sophistic ation in L2 subjunctive testing. Prior to this study, to our knowledge, no re search had specifically looked at the development of L2 subjunctive syntax from this perspective. ME gave learners the freedom to choose numerical representations that more accura tely represented their individual perception of grammaticality. We found that a wide range of numbers was used by learners showing that ME did not confine participants judgmen ts to rigid scales as seen in previous studies. The variety of numbers used in ME to represent learners in tuitions showed how different each participants perception of grammaticality actually was. For example, participant #P1A chose 100 as the highest possible score while #PB2 selected 60 to reflect her perception of the most acceptable structure. Participant #PC3 chose 20 while par ticipant #P4D chose 50 as the highest possible
136 score. In addition, in chapter 4, we discussed how difficult it is to elicit the Spanish subjunctive in oral and written performance tasks. Such difficulties make it challenging to test specific syntactic environments or isolate particular stru ctures in L2. Often times as shown in Collentine (1995), learners end up using paratactic or bi -clausal structures jo ined by commas. Such structures give only the illusion of complex syntax in L2 production. We averted such problems by relying on tasks designed to test learner competence instead of production. In the present study we have shown that even at advanced levels of instruction, learners had not accepted L2 syntactic structures in Sp anish as grammatical. Our findings illustrate learners tendency to rely on L1 syntax, even at advanced levels of instruction, providing evidence of L1 transfer in L2 subjunctive development. With this said, however, we have also provided strong evidence that lear ners have access to CP compleme nts in L2 meaning that IL is capable of generating complex syntax. Taken together, we are able to suggest that the [+ optional] overt CP rule in L1 is transfe rred to L2 during subjunctive development. The approach adopted here has shed light on is sues that learners encounter when presented with the Spanish subjunctive. We have learned th at even at advanced levels of instruction, learners found Englishlike null CP structures to be more acceptable than complex L2 overt CP structures. In addition, we pr ovide solid evidence that functional categories, and most importantly null CPs, were accessible to learners at the time of the st udy. We base this on the low ranking of environment 4 (que + subjunctive). If learners we re unable to access FCs in L2, they would have ranked environment 4 (que + subjunctive) as hig hly acceptable along with environments 5 (que + infinitive) and 6 (que + indicative) in ME. This is also an indication that learners have difficulty recognizing que as [optional] in L2 syntax.
137 Many of the studies that we have discussed in this project ha ve looked at the production, or lack thereof, of functional categories in L2. Our study has taken the opposite approach. The principle benefit of our approach was that we were able to suggest that learners intuitions led them to rate L1 subjunctive syntax as the most acceptable structures in L2. We can suggest with confidence that L2 learners of Spanish have difficulty acquiring the Spanish subjunctive for both syntactic and semantic reasons. Th e learners syntactic problem is rooted in the transference of the [+ optional] overt CP rule in English to L2 Spanish while learners semantic complications are rooted in epistemic modality. It is indisputable that L2 acqui sition of the Spanish subjunctive is difficult and requires time to develop. With this said, however, the problems causing such difficulties are actually quite simple to identify. This study has effectively demonstrated that the problems for L2 learners reside essentially in the L1 and L2 s yntactic differences and the modal intricacies of epistemic and deontic modality.
138 APPENDIX A ME (MAGNITUDE ESTIMATION TASK) (DAY #1) Activity 1: Choose the number that you think best reflects the grammatical accuracy of the reference sentence. Then judge all the other sente nces against this sentence. For example, if you consider the reference sentence to be perfect a nd you give it a 30, the other sentences below with either be as good as just as the reference sentenc e, or they will be considered less than 1. El chico come hamburguesas en el parque. ______ 2. Maribel proclama que su matrim onio seguir bien. ______ 3. Pablo presume que Juana estudie ms. ______ 4. Juan Pablo recomienda aparcar detrs de su apartamento. ______ 5. Ellos no anuncian que se casan maana. ______ 6. Ella no asume que Natalia cantar bien. ______ 7. Claudia no demuestra que comprende el tema. ______ 8. Isaac supone que ella coma sus verduras en la cena. ______ 9. Elena requiere que yo hablar con mi amiga. ______ 10. Pedro intuye Eduardo buscar algo mejor. ______ 11. Manuel piensa que Antonio estudie ms ac. ______ 12. Leonor no intuye que existen problemas entre ellos. ______ 13. Marcos propone que yo compre los boletos hoy. ______ 14. Julin sugiere que ella encontrar mejor trabajo. ______ 15. Los profesores suplican los estudiantes lean ms en casa. ______ 16. Los estudios permiten que se encontrar buen trabajo. ______ 17. No admiten que Ral viva en Nueva York todava. ______ 18. Benjamn no piensa Luis ir a sobrevivir all. ______ 19. Jos no demuestra sea capaz de terminar el proyecto. ______ 20. La polica ruega a los ciudadanos que cie rren las puertas por la noche. ______ 21. Los marineros necesitan el gobierno les suba el sueldo. ______ 22. Jaime nos asegura no fuma marihuana. ______ 23. Luis no asume que siempre tenga r azn como Roberto. ______ 24. La familia desea que los vecinos se mudan de casa. ______ 25. No me imagino que esa informacin sea la verdad. ______ 26. Lorenza suplica su hermana pagar la cuenta. ______ 27. Julio niega tenga deudas. ______ 28. La maestra no piensa la administracin ayude suficiente. ______
139 29. Paco supone el presidente tiene razn. ______ 30. El profesor quiere sus alumnos se preparen mejor. ______ 31. El seor Hernndez exige los empleados trabajan de noche. ______ 32. David me asegura que tener suficiente dinero. ______ 33. David no niega Donato hablar bien espaol. ______ 34. Mara recomienda que Benjamn lle ga por la tarde. ______ 35. El gato bebe leche cocina en la. ______ 36. Pelayo cree Ins es buena amiga. ______ 37. Santiago confirma que sus notas ser buenas. ______ 38. Martn declara que Ivonne se enamora de l. ______ 39. Ana Mara percibe Lucas adelgace rpidamente. ______ 40. Isaac no anuncia sus padres estar divorciados. ______ 41. Iaki considera que la situacin est difcil. ______ 42. A Noelia le gusta que lleguen temprano. ______ 43. La madre desea que su hija sigue con sus estudios. ______ 44. No sospecha que Antonio vivir en Boston. ______ 45. Isabela sugiere su hermana compra la casa de a lado. ______ 46. La seora Smith admite que Juan habla por telfono en la oficina. ______ 47. No creo Carmen toma la clase en serio. ______ 48. Carmen sabe que los hermanos de Pablo llegarn hoy. ______ 49. Ella no piensa Paco vuelve por la tarde. ______ 50. Dudo que ngela buscar a su novi o all. ______ 51. Ricardo prefiere que Ana cocine por la noche. ______ 52. Lucas no demuestra tener talento. ______ 53. Ftima entiende Enrique tenga mucho trabajo. ______ 54. Renaldo dice que azul color favorito es. ______ 55. Lorenzo sospecha Sebastin fume tabaco con sus amigos. ______ 56. Maribel ha hecho su tarea para hoy. ______ 57. Tatiana prefiere tomamos leche con el desayuno. ______ 58. Antonio sostiene Juan Pablo conocer a Alejandro. ______ 59. Julin no expresa su esposa tiene inte rs en salir maana. ______ 60. Francisco cereales come en la cocina. ______ 61. Ellos esperan cenar con sus tos.
140 APPENDIX B L2MAC (L2 MOOD AND CONJUGATION T ASK) (DAY #2) Activity 2: On the line to the left of each sentenc e, indicate whether the mood of the verb should be in indicative (I) or subjunctive (S). Once you have chosen the mood of the sentence, choose the appropriate form of the verb. 1. ______Luis no sospecha que Mara (estar) _________________ prepara ndo una sorpresa. 2. ______Ricardo exige que Laura (vender)_________________ su coche. 3. ______La profesora no cree que Rodrigo (mirar) _________________ las noticias. 4. ______ Elena quiere que Penlope (cocinar) _________________ p escado con limn. 5. ______El jefe no duda que sus empleados (trabajar) _________________ siempre. 6. ______ Los hijos no admiten que sus padres (discutir) _________________ mucho en casa. 7. ______Carlos no intuye que Marco (tomar) _________________ mucho alcohol. 8. ______ Pablo presume que Julin (correr) _________________ todos los das. 9. ______Mabel sugiere que Alejandra (romper) _________________su blackberry 10. ______ Supongo que Roberto (escribir) _________________ sus respuestas. 11. ______ Enrique imagina que su novia (comer) _________________ bien en Espaa. 12. ______Jorge desea que su hija (a prender) _____________ ____ alemn. 13. ______No anuncio que mi esposo (vivir) _________________ en otro estado. 14. ______ Juan prefiere que Marta (asumi r) _________________ sus responsabilidades. 15. ______Ronaldo demuestra que (recibir) _________________ bien a los invitados.
141 APPENDIX C SI (SYNTACTIC IDENTIFICATION TASK) (DAY #2) Activity 3: Choose the sentence in Spanish that provides the most grammatical equivalent to each sentence that appears in English. Please read all of the choices carefully before making your decision. Make sure to only choose one option. 1. Maria wants me to go to the store. _____ Mara quiere que yo voy a la tienda maana. _____Mara quiere que yo ir a la tienda maana. _____Mara quiere yo vaya a la tienda maana. _____Mara quiere que yo vaya a la tienda maana. _____Mara quiere yo ir a la tienda maana. _____Mara quiere yo voy a la tienda maana. 2. Lorena does not believe that Mi chael spends a lot of money. _____Lorena no cree que Miguel gaste mucho dinero. _____Lorena no cree que Miguel gasta mucho dinero. _____Lorena no cree Miguel gasta mucho dinero. _____Lorena no cree Miguel gastar mucho dinero. _____Lorena no cree Miguel gaste mucho dinero. _____Lorena no cree que Miguel gastar mucho dinero. 3. John requires Martha to work every weekend. _____Juan exige que Marta trabaje cada fin de semana. _____Juan exige que Marta trabajar cada fin de semana. _____Juan exige que Marta trabaja cada fin de semana. _____Juan exige Marta trabajar cada fin de semana. _____Juan exige Marta trabaja cada fin de semana. _____Juan exige Marta trabaje cada fin de semana. 4. Paul does not demonstrate that his knowledge is extensive. _____Pablo no demuestra su conocimiento es extenso. _____Pablo no demuestra que su conocimiento es extenso. _____Pablo no demuestra que su conocimiento ser extenso. _____Pablo no demuestra su conocimiento sea extenso. _____Pablo no demuestra que su conocimiento sea extenso. _____Pablo no demuestra su conocimiento ser extenso. 5. I hope that Jacob studies for the test. _____Espero que Jacobo estudiar para el examen. _____Espero Jacobo estudie para el examen. _____Espero Jacobo estu diar para el examen. _____Espero que Jacobo estudia para el examen. _____Espero Jacobo estudia para el examen. _____Espero que Jacobo estudie para el examen. 6 Francis does not admit that Manuel is at his house. _____Francisco no admite que Manuel est en su casa. _____Francisco no admite Manuel est en su casa. _____Francisco no admite que Manuel estar en su casa. _____Francisco no admite Manuel est en su casa. _____Francisco no admite Manuel estar en su casa. _____Francisco no admite que Manuel est en su casa.
142 7 She suggests that David attend the meeting. _____Ella sugiere que David asistir la reunin. _____Ella sugiere David asistir la reunin. _____Ella sugiere que David asiste la reunin. _____Ella sugiere que David asista la reunin. _____Ella sugiere David asiste la reunin. _____Ella sugiere David asista la reunin. 8. Elaine doubts that Beln lives with her parents. _____Elena duda Beln vivir con sus padres. _____Elena duda que Beln viva con sus padres. _____Elena duda Beln viva con sus padres. _____Elena duda que Beln vive con sus padres. _____Elena duda Beln vive con sus padres. _____Elena duda que Beln vivir con sus padres. 9. The general commands his soldiers to attack the enemy. _____El general manda sus soldados ataquen el enemigo. _____El general manda que sus soldados ataquen el enemigo. _____El general manda que sus soldados atacan el enemigo. _____El general manda sus soldados atacan el enemigo. _____El general manda que sus soldados atacar el enemigo. _____El general manda sus soldados atacar el enemigo. 10 Matt does not think that his boss is a good person. _____Mateo no piensa que su jefe es buena persona. _____Mateo no piensa sea buena persona. _____Mateo no piensa es buena persona. _____Mateo no piensa que su jefe sea buena persona. _____Mateo no piensa ser buena persona. _____Mateo no piensa que su jefe ser buena persona. 11. Martha demands that her son look for work _____Marta demanda que su hijo busca trabajo. _____Marta demanda que su hijo busque trabajo. _____Marta demanda que su hijo buscar trabajo. _____Marta demanda su hijo busca trabajo. _____Marta demanda su hijo busque trabajo. _____Marta demanda su hijo buscar trabajo. 12. Mabel does not suspect that her husband works for the CIA. _____Mabel no sospecha que su esposo trabaje para la CIA. _____Mabel no sospecha que su esposo trabaja para la CIA. _____Mabel no sospecha que su es poso trabajar para la CIA. _____Mabel no sospecha su esposo trabajar para la CIA. _____Mabel no sospecha su esposo trabaje para la CIA. _____Mabel no sospecha su esposo trabaja para la CIA.
143 APPENDIX D EXPLANATION OF SI SI was inclu ded because we wanted to comp are English irrealis and Spanish subjunctive structures simultaneously in this study. For this reason, SI was developed as a syntactic identification task designed to test learners ability to equate English irrea lis structures to their Spanish counterparts in the subjunctive. This te st was created in orde r to answer research question 3: How accurately do L2 learners at di fferent levels of instru ction identify Spanish subjunctive structures with their equivalent forms in Modern Englis h? In this task, the learners were presented with a total of 12 English irrealis structures, each followed by a set of six Spanish subjunctive sentences. The participants were asked to choose which Spanish sentence most accurately and correctly conveye d the meaning of the English sentence. Every item in both Spanish and English belonged to a specific semantic category (i .e., ES = epistemic subjunctive, EI = epistemic indicative, and DS = deontic su bjunctive) which formed part of a specific morphosyntactic environment. The reader will notice that these are the same syntactic environments presented to the learners in ME. Item # 1 for example, presents the participants with the following sentence in English Martha demands that her son look for work followed by six possible Spanish interpretati ons as seen in Figure 1. Maria wants me to go to the store. _____Mara quiere que yo vo y a la tienda maana. _____Mara quiere que yo ir a la tienda maana. _____Mara quiere yo vaya a la tienda maana. _____Mara quiere que yo vaya a la tienda maana. _____Mara quiere yo ir a la tienda maana. _____Mara quiere yo voy a la tienda maana. The goal of this task was to see whether or not the participants could accurately identify the Spanish equivalents of the English irr ealis structures presented to them.
144 SI only required the participants to choos e one option. They received one point for a correct answer (i.e., environment 1) and zero point s for an incorrect choice (environments 2-6). In the SI task, the researcher wanted to observe learners ability to interpret English and Spanish syntax. In order to analyze the da ta of all five course levels, an ANOVA was used to see whether or not any differences existed between the levels. Th e data analysis for this test also took into consideration the syntactic envir onments that were presented to the learners. Unfortunately, the results of this task were unreliable due to the or der in which the tasks were administered. It is suggested that L2MAC may have prepared particip ants for this task. In other words, potential priming effects may have caused learners to an ticipate subjunctive examples on the SI task resulting in unreliable data.
145 APPENDIX E BACKGROUND FORM AND QUESTIONNAIRE (DAY #2) YOU MUST RET URN THIS FORM TO LAURIE MASSERY I. SPANISH MAJOR OR MINOR ___Major ____ Minor ____Double ____Other II. BACKGROUND What, if any Spanish classes have you taken at the university level? COURSE TITLE COURSE NUMBER III. EXPOSURE TO THE SPANISH PRESENT SUBJUNCTIVE Please explain briefly the amount of exposure and/or formal inst ruction you have had regarding the Spanish present subjunctive
146 LIST OF REFERENCES Bergen, J. (1978). One Rule for the S panish subjunctive. Hispania, 61 (2), 218-234. Belletti, (1990). Generalized Verb Movement: Aspects of Verb Syntax Torrino, Italy: Rosenberg & Sellier Press. Berne, J. (1990). A comparison of teaching for proficiency with the natural approach: Procedure Design and Approach Source: Hispania, 73 (4), 1147-1153. Bard, E., Robertson, D., and Sorace, A. (1996). Magnitude Estimation of Linguistic Acceptability. Language, 72 (1), 32-68. Blake, R. (1983). Mood selection am ong Spanish-speaking children, 4-12. Bilingual Review, 10 (1), 21-32. Blake, R. (1985). From research to the classroom: Notes on the subjunctive. Hispania, 68 (1), 166-173. Borgonovo, C. & Prvost, P. (2003). Knowledge of polarity subjunctive in L2 Spanish. In B. Beachley, A. Brown, & F. Conlin (Eds.), BUCLD 2: Proceedings of the 27th Annual Boston University Conference on Language Development (pp. 150-161). Somerville, MA: Cascadilla Press. Carroll, S. (2004). Acquisition by Processi ng Theory: A theory of everything? Bilingualism: Language & Cognition, 7 (1), 23-25. Chomsky (1957). Syntactic Structures The Hague: Mouten. Chomsky, N. (1970). Remarks on nominalization. In R. Jacobs & P. Rosenbaum (Eds.), Readings in English Transformational Grammar (pp. 184-221). Waltham, MA: Ginn. Chomsky, N. (1991). Some notes on the economy of derivations and representations. In R. Freidin (Ed.), Principles and Parameters in Comparative Grammar. (pp. 417 454).Cambridge, MA: MIT. Press, Collentine, J. (1995a). The Development of Subjunctive and Complex-Syntactic Abilities Among FL Spanish Learners. Hispania, 78 (1), 122-135. Colletine, J. (1995b). The Development of Complex Syntax and Mood-Selection Abilities by Intermediate-Level Learners of Spanish. Hispania, 78, 123-135. Collentine, J. (1997). The Effects of Irregula r Stems on the Detection of Verbs in the Subjunctive. SAL, 1, 3-23.
147 Collentine, J. (1998a). Processing Instruction and the Subjunctive. Hispania, 81, 576587. Collentine, J. (1998b). Cognitive principles and CALL grammar instruction: A mindcentered input approach. CALICO Journal, 15 (1-3), 1-18. Collentine, J. (2002). On the Acquisition of the subjunctive and authentic processing instruction: A response to Farley. Hispania, 85 (4), 879-888. Collentine, J., Collentine, K., Clark, V. & Fruginal, E. (2002). Subjunctive Instruction Enhanced with Linguistic Instruction. In J. Lee, K. Geeslin, & J.C. Clements (Eds.), Proceedings of the 4th Hispanic Linguistics Symposium (pp. 32-45). Sommerville, MA: Cascadilla Press. Collentine, J. (2003). The development of subjunctive and complex-syntactic abilities Among FL Spanish lear ners. In B. Lafford & R. Salaberry (Eds.), Studies in Spanish Second Language Acquisition: The State of Science (pp. 74-97 ). Washington D.C.: Georgetown University Press. Collentine, J. (2004). Learning context and it s effects on second language acquisition. Studies in Second Language Acquisition, 26 (2), 153-171. Cook, V. (1988). Chomskys Universal Gra mmar: An Introduction Oxford: Blackwell. Cowper, E. (1992). A concise Introduction to Syntactic Theory Chicago: University Press of Chicago. Di Franco., C. (2005). Enseanza de la gramtica y atencin a la forma. Federacin Internacional de Asociaciones de Prof esores de Espaol (F IAPE). Congreso internacional: El espaol, lengua del futuro 1-15. Toledo, Espaa. Eckmann, F. (1977). Markedness and th e contrastive analysis hypothesis. Language Learning, 315-330. Ellis. R. (1994). The Study of Second Language Acquisition Oxford: Oxford University Press. Eubank, L. (1994). On the transfer of pa rametric values in L2 development. Language Acquisition 3 (3) 183-208. Epstein, S. D., Flynn, S. & Martohardjo, G. (1996). Second language acquisition: Theoretical and experimental i ssues in contemporary research. Behavioral and Brain Sciences, 19 (4), 677-758.
148 Faingold, E. (1998). The acquisition, leaning, change, and history of Spanish mood in Mexico and the US (with refere nce to South America and Spain). Pepiere Zur Linguistick, 58, 137-156. Faingold, E. (2000). A systemic analys is of the Spanish subjunctive. Papiere zur Linguistik, 62-63 (1-2), 89-104. Farley, A. (2001). Authentic processing in struction and the Sp anish subjunctive. Hispania, 84 (2), 89-99. Farley, A. (2004). The relative effects of processing instruction and meaning-based output instruction. In B. VanPatten (Ed.), Processing Instruction: Theory, Research, and Commentary (pp. 143-168). Mahwah, NJ: Erlbaum Associates. Featherson, S. (2005a). That-trace in German. Lingua, 115 1277-1302. Featherson, S. (2005b). Universals and grammatical ity: Wh-constraints in German and English. Linguistics, 43 (4), 667-699. Feeny, T. (1978). Using classroom objects in teaching the subjunctive with impersonal expressions. Hispania, 6 (2), 328-329. Finer, D. (1991). Binding Parameters in Second Language Acquisition. In L. Eubank (Ed.), Point Counterpoint: Universal Grammar in Second Language Acquisition (pp. 351-374). Amsterdam: Johns Benjamins. Flege, J. (1997). Effects of experience on non-native speakers pr oduction and perception of English vowels. Journal of Phonetics, 25 437470. Floyd, M. (1990). Sentence Complexity a nd Clause Subordination in Children's Spanish. Hispania, 73 (2), 488-497. Foster, D. (1982). Internal contradi ctions of a Spanish subjunctive. International Review of Applied Linguistic s in Language Teaching, 20 (2), 131-137. Gass, S. (1997). Input, Interaction and the second language learner. Mahwah, NJ: Erlbaum. Gass, S., Mackey, A., Alvarez-Torres, M., & Fern ndez-Garca, M. (1999). The effects of task Repetition on linguistic output. Language Learning 49 (4), 549-581. Gerbault, J. (1978). The Acquisition of English by a Five Year Old French Speaker, Masters Thesis, University of Ca lifornia, Los Angeles. Gili Gaya, S. (1961). Ortografa y prctica. Barcelona: Publicaciones y Ediciones Spes. Gili Gaya, S. (1972). Estudios de len guaje infantil. Barcelona: Bibliograf.
149 Givn, T. (1994). Irrealis and the Subjunctive. Studies in Language, 18 (2), 265-337. Goldin, M. (1974). A Psychological Perspective of the Spanish Subjunctive. Hispania, 57 (2), 295-301. Gudmestad, A. (2006). L2 Variation and the Spanish Subjunctive: Linguistic features predicting mood selection. In C. Klee & T. Face (Eds.), Selected Proceedings of the 7th Conference on the Acquisition of Spanish and Portuguese as First and Second Languages (pp. 170-184). Somerville, MA: Cascadilla Proceedings Project. Haznadar, B. (2003). Missing Surface Inflect ion in adult and child L2 acquisition. In J. Liceras et al (Eds). Proceedings of the 6th Generateive Approaches to Second Language Acquisition Conference (pp. 140-149). Somerville, MA: Cascadilla Proceedings Project. Haznadar, B. & Schwartz, B. (1997). Are there optional infinitives in child L2 acquisition? In E. Hughes & A. Greenhill (Eds). Proceedings of the 21st Annual Boston University Conference on Language Development (pp. 257-268). Somerville MA: Cascadilla Press. Hawkins, R. & Chan, C. (1997). The partial av ailability of Universal Grammar in second language acquisition: The failed functional features hypothesis. Second Language Research, 13 (3), 187. Herschensohn, J. (2000). The Second Time Around: Minimalism and L2 Acquisition Philadelphia: John Benjamins. Herschensohn, J. (2001). Missing inflection in second language French: Accidental infinitives and other verbal deficits. Second Language Research, 17 (3), 273-305. Hilles, S. (1991). Access to Universal Gr ammar in Second Language Acquisition. In L. Eubank (Ed.), Point Counterpoint: Universal Grammar in the Second Language (pp. 305-338). Amsterdam: Johns Benjamins. Hornstein, N., Nunes, J. & Grohmann, K. (2005). Understanding Minimalism Cambridge: Cambridge University Press. Jelsinki, J. (1977). A new look at teaching the Spanish subjunctive. Hispania, 60 (2), 320-326. Keller, F. & Asudeh, A. (2001). Constraints on linguistic coreferenc e: Structural vs. pragmatic factors. In J. Moore & K. Stenning (Eds.), Proceedings of the 23rd Annual Conference of the Cognitive Science Society (pp. 483-488). Mahwah, NJ: Erlbaum.
150 Kellerman, E. (1979). The problem with difficulty. Interlanguage Studies Bulletin Kellerman, E. (1985). If at first you do succeed. In S. Gass & C. Madden (Eds.), Input in Second Language Acquisition (pp. 345-353). Rowly, MA: Newbury House. Kellerman, E. (1986). An eye for an eye: Cr osslinguistic constrai nts on the development of the L2 lexicon. In E. Kellerman & S. Smith (Eds.), Crosslinguistic Influence in Second Language Acquisition (pp. 35-48). New York: Pergamon. Kilroe, P. (1988). Making grammar explicit in the classroom: An illustration using the Spanish subjunctive. Texas Papers in Language Education, 1 (2), 90-100. Krashen, S. (2003). Explorations in Language Acquisition and Use New Hampshire: Heinemann. Lazano, A. (1995). Cognitive development, deontic and epistemic subjunctives. Hispanic Linguistics, 6/7 94-115. Lee, J. (1987). Comprehending the Spanish Subjunctive: An Information Processing Perspective. Modern Language Journal, 71 (1), 50-57. Lee, J. & VanPatten, B. (2003). Making Communicative Language Teaching Happen. New York: McGraw Hill. Leow, R. (1999). The role of attention in second/foreign la nguage classroom research: Methodological issues. In J. Gutirr ez-Rexach & F. Ma rtnez-Gil (Eds). Advances in Hispanic Linguistics (pp. 60-71). Somerville, MA: Cascadilla Press. Leow, R., Egi, T., Nuevo, A. & Tsai, Y. ( 2003). The roles of textual enhancement and type of linguistic item in adult L2 learners comprehension and intake. Applied Language Learning, 13, 93-101. Lightbown, P. (1985). Great Expectations: Se cond-language acquisition research and classroom teaching. Applied Linguistics, 6 173-189. Lightbown, P. & Spada, N. (2006). How Languages are Learned Oxford: Oxford University Press. Lubbers Quesada, M. (1996). Interlanguage variability of the Spanish subjunctive. Estudios de Lingustica Aplicada, 14 (23-24), 358-369. Lubbers Quesada, M. (1998). L2 acquisiti on of the Spanish subjunctive mood and prototype schema development. SAL, 2 (1), 1-23.
151 Lunn, P. (1992). The evaluative function of the Spanish subjunctive. In J. Bybee and S. Fleischman (Eds.), Modality and grammar in discourse : Typological studies in language (pp. 429-449). Amsterdam: John Benjamins. Mackey, A. and Gass, S. (2005). Second Language Research: Methodology and Design Mahwah, NJ: Erlbaum. Martohardjono, G. & Gair, J. (1993). Apparent UG inaccessibility in second language acquisition: Misapplied principles or principl ed misapplications? In F. Eckman (Ed). Confluence:Llinguistics, L2 acquisition and speech pathology (pp. 79-103). Amsterdam: John Benjamins. Massery, L. (2008a). Were th e subjunctive that easy. The Journal of Communication and Education Language Magazine, 7 (6), 18-19. Massery, L. (2008b). PowerPoint technol ogy in the second language classroom. The International Journal of Tec hnology, Knowledge, and Society, 4 (3), 139-143. Massery, L. (2008c). The Spanish subjunctive: Order, Acquisition, and Instruction. Paper presented at Hispanic Linguistic Symposium, (HLS), Quebec. Mejias-Vikindi E. (1994). Assertion and Speakers Intent ion: A pragmatically based account of mood in Spanish. Hispania 77 (4), 892-902. OGrady, W. (1991). Language acquisition and the pro-drop phenomenon: A Response to Hilles. In L. Eubank (Ed.), Point Counterpoint: Universal Grammar in the Second Language (pp. Amsterdam: John Benjamins. Perez-Leroux, A. (1998). The Acquisition of mood selection in Spanish relative clauses. Journal of Child Language, 25 (3), 585-604. Pishwa, H. (1989). Erwerb der deutschen kongruenzregel. Unpublished doctoral dissertation. TuBerlin. Pishwa, H. (1991). Variability in the acquisition of the German agreement rule in syntactic contexts. Papiere zur linguistic, (44/45), 29-47. Pollock, J. (1989). Verb movement, Universa l Grammar, and the structure of IP. Linguistic Inquiry, 20 365-424. Prevost, P. & White, L. (2000a). Accounti ng for morphological variation in second language acquisition: Trun cation or missing inflection? In M. Friedemann & L. Rizzi (Eds.), The acquisition of syntax (pp. 202-235). London: Longman. Prevost, P. & White, L. (2000b). Missing surf ace inflection or impa irment in second language? Evidence from tense and agreement. Second Language Research, 16 (2), 103133.
152 Radford, A. (1990). Syntactic theory and the acqu isition of English syntax Oxford: Blackwell. Radford, A. (1995). Children, archit ects or brickies? In D. MacL aughlin & S. McEwen (Eds.), Proceedings of the 19th Annual Boston University Conference on Language Development (pp.1-19). Sommerville, MA:Cascadilla Press. Radford, A. (1996). Towards a structure-building model of acquisition. In H. Clahsen (Ed.), Generative perspectives on language acquisition (pp. 43-90). Amsterdam: Johns Benjamins. Radford, A. (1997). Syntax: A Minimalist Introduction Cambridge: Cambridge University Press. Rutherford, W. (1982). Language typology and language transfer. In S. Gass & L. Selinker (Eds.), Language Transfer in Language Learning. Rowley, MA: Newbury House. Salaberry, R. (1996). The pedagogical value of simplified written input in L2 acquisition. Cornell Working Papers in Linguistics, 14 125-150. Salaberry, R. (1997). The role of input and output practice in second language acquisition. Canadian Modern Language Review, 53 (2), 422-451. Saville-Troike, M. (2007). Introducing Second Language Acquisition. Cambridge: Cambridge University Press. Schane, S. (1994). Illocutionary verbs, subj ect responsibility, and presupposition: The indicative versus the subjunctive in Span ish. In P. Hashemipour, R. Maldonado, & M. Van Naerssen (Eds.), Studies in Language learning and Spanish Linguistics in Honor of Tracy D. Terrell ( pp. 361-374). New York: McGraw-Hill. Sorace, A. (2000). Syntactic opti onality in non-native grammars. Second Language Research, 16 293-102. Southwood, M., & Flege, J. (1999). Scaling foreign accent: Direct magnitude estimation versus interval scaling. Clinical Linguistics and Phonetics, 13 (5), 335-349. Speas, M. (1990). Phrase Structure in Natural Language Boston: Kluwer Academic Publishers. Stokes, J. (1988). Some factors in the acquisi tion of the present subjunctive in Spanish. Hispania, 71 (3), 705-710. Stokes, J. & Krashen, S. (1990). Some Fact ors in the Acquisition of the present subjunctive in Spanish: A re-analysis Hispania, 73 (3), 805-806.
153 Stokes, J., Krashen, S., & Kartchner, J (1998) Factors in the acquisition of the present subjunctive in Spanish: Th e role of reading and study. ITL Review of Applied Linguistics, 121-122. Studerus, L.(1995). Some unresolved issues in Spanish mood use. Hispania, 78 (1), 94104. Terrell, T. & Hooper, J. (1974). A semantica lly based analysis of mood in Spanish. Hispania, 57 484-494. Terrell, T., Baycroft, B. & Perrone, C. (1987). The subjunctive in Spanish interlanguage: Accuracy and co mprehensibility. In B. VanPatten, T. Dvorak, & J. Lee (Eds.), Foreign Language Learning: A Research Perspective (pp. 19-31). NewYork: Newbury. Terrell, T. (1991). The Role of Grammar In struction in a Communicative Approach. The Modern Language Journal, 75 (1), 52-63. Tsimpli, I. & Roussou, A. (1991). Parameter-resetting in L2? UCLA Working Papers in Linguistics, 3 149-169. Vainikka, A. & Young-Scholten, M. (1994). Dir ect access to X-Bar Theory: Evidence from Turkish adults learning German. In T. Hoekstra & B. Schwartz (Eds.), Language acquisition studies in generative grammar (pp. 265-316). Amsterdam: Johns Benjamins. Vainikka, A. & Young-Scholten, M. (1996a). Gradual development of L2 phrase structure. Second Language Research, 12 7-39. Vainikka, A. & Young-Scholten, M. (1996b). The early stages in adult L2 syntax: Additional evidence from Romance speakers. Second Language Research, 12 140-76. Vainikka, A. & Young-Scholten, M. (1998). Mor phosyntactic triggers in adult SLA. In M. Beck (Ed.), Morphology and its interface in L2 knowledge (pp. 89-113). Amsterdam: John Benjamins. VanPatten, B. (1986). Second Language Acquis ition Research and the Learning/Teaching of Spanish: Some Research Findings and Implications. Hispania 69 (1): 202-216. VanPatten, B. (1993). Grammar teaching for the acquisition-rich classroom. Foreign Language Annals, 26, 435-450. VanPatten, B. (1996). Input Processing and Grammar Instruction in Second Language Acquisition Norwood, New Jersey: Ablex Publishing.
154 VanPatten, B. (2004). Processing Instruction Theo ry, Research, and Commentary. Mahwah, NJ: Erlbaum. White, L. (1985). The pro-drop paramete r in adult second language learning. Language Learning 35: 47-62. White, L. (1990) The verb-movement parameter in second language acquisition. Language acquisition, 1 (4), 337-360. White, L. (1991). Adverb placement in sec ond language acquisition: Some effects of positive and negative evidence in the classroom. Second Language Research, 7, 133-161. White, L. (1992). Long and short verb m ovement in second language acquisition. Canadian Journal of Linguistics, 37 273-286. White, L. (2003). Second language acquisition and universal grammar. Cambridge: Cambridge University Press. Wode, H. (1983). Papers on Language Acquisition, Language Learning, and Language Teaching. Heidelberg: J. Groos Verlag. Zhang, S. (1995). Semantic differentiation in the acquisition of English as a second language. Language Learning, 45 (2), 225-249. Zobl, H. (1980a). The formal and developmental selectivity of L1 infl uence on L2 acquisition. Language Learning 30 (1), 43-57. Zobl, H. (1980b) Developmental and transfer errors: Their common bases and (possibly) Differential effects on subsequent learning. TESOL Quarterly 14 (4), 469-479. Zobl, H. and Liceras, J. (1994). Functi onal categories and acquisition orders. Language Learning, 4, 159-180.
155 BIOGRAPHICAL SKETCH Laurie Massery received her Ma ster of Arts (M.A.) degree in Hispanic Linguistics and Second Language Acquisition from the University of Massachusetts in 2004. She received her Ph.D. from the University of Florida in 2009, al so in Hispanic linguistics and second language acquisition. Within the field of second language acquisition, her specialty and primary focus is the syntactic development in L2.