Processing of tense morphology and filler-gap dependencies by Chinese second language speakers of English

Dong, Zhiyin
Journal Title
Journal ISSN
Volume Title
University of Delaware
There is an ongoing debate in the field of Second Language Acquisition concerning whether a fundamental difference exists between the native language (L1) and adult second language (L2) online processing of syntax and morpho-syntax. The Shallow Structure Hypothesis (SSH) (Clahsen and Felser, 2006a, b) states that L2 online parsing is qualitatively different in that L2 speakers build a "shallower" syntactic representation with less hierarchy and fewer details than L1 speakers, and therefore cannot use purely structural principles effectively. Instead, L2 speakers rely mainly on semantic and lexical information to process sentences. This thesis evaluates such claims by examining how L2 speakers process complex filler-gap (FG) dependencies and tense morphology in real time. Specifically, it tests whether L2 speakers (a) posit abstract syntactic traces in filler-gap (relative clause) constructions, and (b) use morphological decomposition rules when processing tense morphology. The method of Event Related Potentials (ERP) is used because its indexes reveal the nature of the underlying processing mechanism (i.e., syntactic vs. semantic). In addition, the roles of L2 proficiency, L1 interference, and working memory capacity are examined. The results indicate that the learners are able to produce native-like brain responses for tense morphology processing. However, their ERP indicators differ drastically from the native controls in resolving filler-gap dependencies in relative clauses, a construction with highly abstract elements. As the SSH predicts, the L2 speakers resort to semantic and lexical information without building syntactic traces. Furthermore, this parsing pattern is not affected by proficiency and working memory differences. While the empirical evidence is largely in line with the SSH, a different theoretical conclusion than that of the SSH was reached. Assuming the view that the parser and the grammar are in one system, it is concluded that L2 shallow parsing is limited to constructions with highly abstract elements with no overt reflex and persists only when this mechanism does not interfere with successful meaning computation.