| Since the seminal work of Alan Juffs, second language (L2) sentence processing has become a central research topic in the field of L2 acquisition. Beyond a general hunger for new data offered by new experimental techniques and designs, intellectual concerns with system-wide understanding of first-language (L1)-L2 differences extended to the processing system, have brought about this development in the field. Thus, one body of research characterizes L2 sentence processing in a general modular processing system, with L1-L2 differences arising from processing speed, affected by differences in grammar, prosody, and lexis, with fixed limited resources. Others, in contrast, claim that while L1 sentence processing heavily relies on the computation of a detailed syntactic analysis of the input in real time, L2 sentence processing relies instead on thematic, lexical, and contextual information, as learners generally fail to compute syntactic representations that are as detailed as native speakers'. The present study examines the degree to which L2 sentence processing -- like L1 sentence processing--includes syntactic computations in real time. It focuses on the processing of (apparent) long-distance anaphoric dependencies with reflexive pronouns like himself versus pronouns him by Chinese learners of English with intermediate proficiency. Anaphora resolution of this type in native English requires a series of local syntactic computations, providing a particular useful test of the nature of L2 sentence processing. This is because it appears that processing of such dependencies in native grammar complies with certain syntactic constraints, namely Principle A and Principle B of the Binding Theory. This study consists of three reading time experiments using Linger (Doug Rohde, MIT). If syntactic computations mediate the resolution of anaphors and pronouns in L2 sentence processing, reading times (RTs) should reveal specific contours, with specific local binding theoretic computations. These computations should occur independently of learner's knowledge of context, as this has been found in English native speakers. RT data collected from native speakers and intermediate Chinese-speaking learners of English reveals crucial effects, pointing to a structurally driven processing mechanism. |
