Examining Morphological Sensitivity and Individual Differences in the Acquisition of Agreement in a Second Language

Abstract

This experiment investigates learner sensitivity to agreement violations involving uninterpretable number and gender features in Spanish. Twenty-two low-proficiency, English-speaking learners and a control group of twelve native Spanish-speakers gave grammaticality judgments after viewing sentences presented rapidly in word-by-word format. The stimuli targeted agreement between nouns in subject positions and adjectives in the predicate. Responses to violations of number agreement, which is instantiated on verbs in English but not adjectives, were compared to violations of gender agreement, which is unique to the L2 Spanish. In line with the Full Transfer/Full Access theory (Schwartz and Sprouse, 1994; 1996), it was hypothesized that learners would demonstrate the same pattern of sensitivity to agreement violations as native speakers, but that they would be less sensitive to gender versus number violations. The question of whether or not individual differences impact grammaticality judgments was also explored, due to the claim of the Fundamental Difference Hypothesis (Bley-Vroman, 1989; 1990) that successful adult L2 acquisition relies on exceptional abilities outside the realm of Universal Grammar (UG). Learners completed the Short Form of the Modern Languages Aptitude Test (Carroll and Sapon, 1959) and the Raven Advanced Progressive Matrices (Raven, 1965) as measures of verbal and nonverbal aptitude, respectively. Results support Full Transfer/Full Access, suggesting that uninterpretable features can be acquired by adult L2 learners, even in the case of gender features that are not present in the L1. Unique L2 features do seem to present a challenge at this stage of acquisition, as the learners performed significantly worse for gender versus number agreement. In line with FTFA, however, performance on gender agreement correlated most strongly with proficiency. Grammaticality judgments were not found to correlate with either verbal or nonverbal aptitude.