Unaccusativity and neurocognitive indices of second language acquisition: An ERP study.

摘要

Increased second-language (L2) proficiency is associated with a shift from explicit to implicit processing; however, the neural underpinnings are of this shift are not well understood. Furthermore, it is known that unaccusative verbs cause persistent difficulties in L2 learning. In this study, behavioral and event-related potential (ERP) responses to grammatical and ungrammatical sentences involving unergative and unaccusative verbs, in native and L2 English speakers, were examined in order to investigate whether neural responses changed with increasing proficiency, and to seek evidence regarding Oshita's (1997) Unaccusative Trap Hypothesis. Native English speakers and three groups of Chinese learners of English at different levels of grammatical knowledge (low, mid, high) were shown grammatical sentences containing passive, unergative, or unaccusative verbs, and ungrammatical sentences containing incorrectly passivized unaccusative and unergative verbs. Event-Related Potentials derived from high-density electroencephalogram (EEG) recordings revealed no systematic differences between unaccusatives and unergatives for the native speakers; however, the three learner groups all showed different responses to the verb manipulation. Native speakers revealed a left anterior negativity (LAN) to ungrammatical conditions, followed by a ramplike negativity (the N400/700). However, the three learner groups showed an N180 peak over parietal sensors bilaterally, greatest in amplitude for low-proficiency learners and attenuated for middle and high proficiency learners. Furthermore, a grammaticality effect was seen in the Chinese High and Chinese Low groups, with higher N180 amplitudes associated with grammatical sentences. The N180 is associated with visual working memory, which suggests that non-native speakers recruit non-linguistic mechanisms for L2 processing. These findings indicate that N180 attenuation, together with a shift towards anterior and left-lateralized processing, may provide an index of increased L2 proficiency. In addition, while there was weak behavioral support for the Unaccusative Trap Hypothesis, the ERP results did not reveal a distinction between the processing of unaccusative and unergative sentences.