Individual Differences and the Role of the L1 in L2 Processing: An ERP Investigation

Our study uses event-related potentials (ERP) to investigate two factors in the second language (L2) processing of agreement: the role of number and gender features in the native language (L1), and the impact of individual differences between learners. Previous ERP studies investigating L2 agreement vary as to whether they find native-like processing (e.g., Rossi et al., 2006; Tokowicz and MacWhinney, 2005; Ojima et al., 2005; Sabourin, 2003). Additionally, recent studies have found variability between groups of learners with regard to development of native-like responses to agreement violations (McLaughlin et al., 2010; Tanner et al., 2009). We further examine these issues by testing both learner aptitude and sensitivity to number and gender agreement violations in L2 Spanish by native speakers of English. An analysis of sensitivity to violations in three types of agreement allow the comparison of responses to (a) number features on verbs (el barcosg sg/ *flotanpl), which are similar in the L1 English, (b) number features on adjectives (la islasg sg /*rocosaspl), a category where agreement is not instantiated in English, and (c) gender agreement on adjectives (la islafem fem/*rocosomasc), which involves a feature that is unique to the L2. Theories of L2 acquisition differ with regard to claims about the potential for native-like acquisition, a debate that centers on distinct claims regarding the role of a proposed universal set of features and parameters for language, or Universal Grammar (UG) (Chomsky, 1965; 1980; 1981). The Interpretability Hypothesis (Tsimpli and Mastropavlou, 2007; Tsimpli and Dimitrakopoulou, 2007) claims that adult learners have access to UG only through their L1 and are unable to incorporate features of the L2 that are not instantiated in the L1 if those features are uninterpretable . For example, the gender and number features of adjectives in Spanish are argued to be uninterpretable (e.g., Carstens, 2000) and thus should be acquirable by L2 learners only if those features are also present in the L1. For English-speaking learners of Spanish, number agreement should be acquirable to native-like levels, but not gender agreement. The Full Transfer/Full Access approach (Schwartz and Sprouse, 1994, 1996), on the other hand, claims that L1 features and parameter settings are influential in the early stages of L2 acquisition, but that L1 properties can be abandoned in favor of other UGconstrained settings as needed in order to accommodate L2 input. Number and gender features in L2 Spanish should both be acquirable by English-speaking learners, but at lower levels of proficiency, native-like processing of gender agreement may not be well-developed. The extent to which learners exhibit native-like acquisition of agreement features has largely been investigated using behavioral measures (White et al., 2004; McCarthy, 2008; Franceschina, 2002; Montrul et al., 2008), but the availability of brain imaging technology has recently made possible the brain-level comparison of agreement processing in learners and native speakers of a language. Particularly promising is the use of EEG technology to measure event-related potentials (ERPs). ERPs are a measurement of electrical activity in the brain time-locked to a specific stimulus. When averaged