«Learning a Gendered Language: L2 Acquisition and Relative Processing Costs of Spanish Grammatical Gender and Number Alexander M. Hirsch Yale ...»
Running head: LEARNING A GENDERED LANGUAGE Hirsch 1
Learning a Gendered Language:
L2 Acquisition and Relative Processing Costs of Spanish Grammatical Gender and Number
Alexander M. Hirsch
Submitted in partial fulfillment for the degree of Bachelor of Arts in Linguistics
LEARNING A GENDERED LANGUAGE Hirsch 2
One of the central debates in the field of language acquisition concerns the possibility of nativelike language acquisition begun after the end of the so-called “critical period.” Various hypotheses have been developed to explain the drop-off of naturalistic language acquisition skills after this period, as well as the specific deficiencies shown by many, but not all, late second language (L2) learners. To be able to say something useful about such a broad subject, this study’s focus is narrowed to acquisition of certain grammatical features absent from the subjects’ native language but present in the L2. Using a self-paced reading task, this study measures the ability of English-L1 late Spanish learners to process Spanish gender and number mismatches across determiner-noun pairs. This study was carried out in order to determine these speakers’ ability to acquire linguistic elements not present in their native language, to provide insight into the architecture underlying processing of these two similar but distinct elements, and to determine how transfer affects this ability.
LEARNING A GENDERED LANGUAGE Hirsch 3 Acknowledgements I would like to thank my advisor, Professor Maria Piñango, for her insightful comments and strong encouragement. I would also like to thank my parents, Gary and Adrienne, for their never-ending support and equally never-ending patience.
LEARNING A GENDERED LANGUAGE Hirsch 4 Learning a Gendered Language L2 Acquisition and Relative Processing Costs of Spanish Grammatical Gender and Number The below research seeks to answer the question of whether second language learners can acquire aspects of language not present in their native language to the same extent as native speakers. Specifically, this study examines whether native English speakers can attain the ability to process gender mismatches, not present in English, to the same extent as natives, and how this processing ability compares to their ability to process number mismatches, which are present to a certain extent in English. I will begin this paper by providing a brief introduction to the field of child language acquisition in order to provide context for a discussion of second language acquisition research, much of which is based primarily on study of child language acquisition. From here I will discuss past theories of second language acquisition (SLA): I will give a brief picture of what I see as the landmark hypotheses of the field, in order to provide some insight into the history of SLA investigation, as well as grounding in which to root a discussion of the field’s current state.
I will continue by discussing some current, competing hypotheses on the nature and possible extent of L2 acquisition and their relative merits. This section will serve to narrow the overbroad field of SLA to the quadrants to which the current research belongs. I will then move on to a discussion of the previous research on gender and number acquisition, especially in Spanish, in order to provide a background upon which to place the present study. Finally, I will present my experiment and hypothesis, where they stand with reference to the present state of SLA inquiry, and why the current experiment is important
1. Early Exploration into SLA 1A. The Critical Period Hypothesis The existence of a “critical period” of language learning has been the subject of an ongoing debate since it was proposed in a landmark paper by Penfield & Roberts (1959). The so-called “Critical Period Hypothesis” (CPH) states that after a certain discrete period, usually stated as lasting from birth until puberty or later adolescence, it is not possible for a child without a first language (L1) to learn to speak any language with the degree of fluency and command exhibited by a native speaker.
Though not specifically intended by the original proposers of the hypothesis, the CPH has been extended to second language learning, with similar implications. That is, adherents to the CPH with regard to second language (L2) acquisition contend that late L2 learners (i.e.
second language learning is begun after the Critical Period) can never reach the fluency of a native speaker, or will never be able to totally eliminate their native accents.
Experimentally, late L2 learners have been shown to perform worse than natives or early L2 learners on a variety of linguistic tasks. For example, in a study by Johnson & Newport (1989), subjects performed more poorly on a grammaticality judgment task if they began learning their L2 (in this case, English) after the age of 8 than younger children, who performed at ceiling levels. Furthermore, speakers who began learning an L2 when they were 12-14 years of age performed worse than younger children, and roughly on par with older speakers. Interestingly, heritage speakers who learn a language at a young age but lose the language after infancy are generally more capable than L2 learners (Alarcon 2011). These results suggest that this critical period ends definitively between the ages of 12 and 14 after the requisite language acquisition skills have begun to taper off beginning
language becomes much more difficult to acquire. However, these results make no predictions as to the possibility (versus likelihood) of end-state or nativelike L2 acquisition when begun after the critical period.
1A. Beyond the critical period: Failed Functional Features. Many refinements to the CPH, as well as many alternatives, have been proposed in the last 60-plus years. With respect to L2 acquisition, the majority of subsequent hypotheses push in one of two directions; they either refute the claim that a discrete Critical Period blocking late language development exists, or they affirm Penfield & Roberts’s hypothesis and explain the precise effects of a Critical Period on the language faculty that would lead to such a loss of learning potential. 1 The Failed Functional Features Hypothesis (FFFH), proposed by Hawkins & Chan (1997), refines the CPH with respect to its position, or lack thereof, on L2 acquisition. It posits that L1 acquisition entails the fixing of certain “functional features,” which underlie various aspects of the L1. From this hypothesis, two predictions follow about the behavior of L2 learners. First, beginning late L2 learners are expected to directly overlay L2 morphology and lexicon onto L1 syntax. From this starting point, L2 learners will progressively refine the interlanguage syntax (the underdeveloped L2 before L2 completion) until it approximates the L2 syntax as closely as possible. However, Hawkins and Chan propose that certain functional features from the L1 will not be revised to fit the L2; they will exist instead as failed functional features that deviate from those of native speakers.
As an early hypothesis of the age cutoff phenomenon, the CPH does not give a detailed explanation of what
Though attractive in certain respects, the FFFH has not held up under empiricallymotivated scrutiny. Numerous examples have surfaced over the intervening years of late L2 learners whose syntax is essentially nativelike and does not contain flaws along the lines of those predicted by the FFFH. For example, White and Genesee (1996) find no difference in performance between late near-native and native speakers on tasks designed to test access to Universal Grammar, which is thought to be subject to Critical Period effects, and White, et al. (2004) finds that native English Spanish-L2 learners can acquire Spanish grammatical gender to the same extent as French natives, findings that directly contradict the FFFH. Luckily, other hypotheses explaining Critical Period-like effects have emerged subsequent to the FFFH.
1B. Transfer and the FTFA Hypothesis. The Full Transfer Full Access (FTFA) Hypothesis, proposed by Schwartz & Sprouse (1996), is one such hypothesis, both drawing on the CPH and FFFH and differing from these predecessors in important ways. The authors argue that adults acquiring a second language proceed with acquisition beginning with the principles gained from acquiring their first language, like the FFFH. Furthermore, all the principles from the L1 are transferred to the initial state of the L2. Indeed, central to the FTFA argument is the proposal that the end state of L1 acquisition and the beginning state of L2 acquisition are identical within a given learner. L2 development therefore must be a process of refining or changing the initial grammar imported from the L1, so that the resulting interlanguage grammar reflects some mixture of the end state of the L1 and the end state of the L2. FTFA posits, then, that learners of an L2 do not begin anew with a blank slate but rather carry any grammatical principles from the L1 into their initial conception
decipher and produce the L2. Because the L2 is acquired by modifying the transferred grammar of the L1 until it becomes the final grammar of the L2, the L2 learner is never without a grammar but rather possesses a different (but not degenerate) grammar from the target syntax. As a result, errors in morphological inflection, for example, are ascribed not to deficits in the syntax but rather to difficulty in the realization of that syntax so that it conforms to the target construction. Herschensohn, Stevenson & Waltmunson (2005) provide evidence for the FTFA in this respect, finding that L2 learners’ syntax develops more rapidly and accurately than their morphology.
In contrast with the CPH, the FTFA Hypothesis does not argue for a decrease in
language-learning ability over time, and in some senses argues against such degeneration:
if an adult begins to learn a second language with many similarities to their first language—a Spanish speaker learning Portuguese or an English speaker learning Dutch, for example—FTFA predicts that transfer from the first language should trivialize the amount of information needed to achieve fluency in the second language, as there will be substantial overlap between the L1 and the L2. Furthermore, learning novel structures will not be any more difficult during L2 than L1 acquisition, since these structures will simply be added atop previous ones. White, et al. (2004), support this account, finding that native English speakers learning Spanish are able to acquire gender to the same extent as native Spanish speakers. However, flaws in the study’s methodology, such as inadequate categorization of subjects with respect to proficiency and lack of online testing, limit the applicability of the its conclusions.
1C. How much is transferred? The extent of transfer between a first and second
suggests that correspondence between the L1 and L2 is the most important factor in L2 acquisition, implying that transfer is highly important to L2 learning (Bialystok 1997).
However, previous hypotheses, such as Vainikka & Young-Scholten (1994)’s Minimal Trees hypothesis and Eubank (1993/1994)’s Weak Transfer Hypothesis do not agree with Schwartz & Sprouse’s idea of full transfer: Vainikka & Young-Scholten suggest that only lexical categories and the linear orientation of these categories transfers from the L1 grammar, while Eubank proposes that functional categories (and their linear orientations) transfer as well. Importantly, neither Vainikka & Young-Scholten nor Eubank allow that the strength of inflection associated with functional categories transfers between L1 and L2, and it is here that these authors differ from Schwartz & Sprouse. This is to say that no matter the specific version of the weak transfer hypothesis in question, proponents of its various varieties unite in their view that the bare syntactic structure—e.g. functional projections and categories—is left intact in the grammatical transfer from the L1 to the nascent L2, but morphologically-driven features like agreement (gender, number, etc.) do not transfer to the initial L2 grammar.
1D. Mechanisms underlying weak transfer effects. These weak transfer hypotheses necessarily reject a unified morphosyntactic conception of language learning, since they posit that only a certain parts of language—the syntax—may transfer, while other parts are left behind. At the same time, this kind of hypothesis implies that a specific dissociation must exist between syntax and morphology (at least as they are defined within such a framework; White, et al. 2004). Perhaps what is transferred are the structural aspects of language—linguistic components that are primarily organizational and
which has been described as a system of signs and associated meanings (e.g. de Saussure 1916).
Rule-based components, on the other hand, would be left behind. According to Anderson (1992) and others, morphology is one example of a “rule-based component”: it is a system of rules governing related words, not a system of roots and concatenated morphemes as some have characterized it. Anderson’s system places morphology firmly in the rule-based category, diverging from previous theories that described it in a structural light. Perhaps a more obvious member of this class is phonology, a system of rules governing the sounds of related words.
At least intuitively, this kind of categorical division makes sense. Structure building or association, which combine existing parts to form a greater whole, should involve a fundamentally different process from rule application, which applies a rule or function to an input and transforms it into a new but related output. In terms of cognitive organization, there seems to be a general division in the brain between rule-based and structure-based aspects of language.