Abstract/Summary

An ongoing debate revolves around whether non-native (L2) speakers resolve discontinuous linguistic dependencies, such as subject-verb agreement and reflexive dependencies, similarly to native (L1) speakers. Variable findings across studies have contributed to this ongoing discussion. Among the different factors suggested to explain potential L1/L2 processing differences, is the role that the working memory retrieval mechanism, which subserves dependency processing, may play in L2 processing. While effects associated with this retrieval mechanism, namely similarity-based interference effects, have been extensively investigated in L1 processing, their relevance to L2 processing remains speculative. The question of whether the computation of subject-verb agreement and reflexive dependencies relies on the same memory retrieval mechanism has also been a subject of debate in the L1 processing literature. This thesis aims to examine these two dependencies, subject-verb agreement and reflexive-antecedent dependencies, in parallel experimental settings within the same group of participants at different proficiency levels. This approach offers a valuable means to understand L2 processing patterns across dependencies and brings us closer to reaching a general agreement on L1/L2 processing similarities and differences, while minimizing individual variation. Through two large-scale studies employing offline and online measures, the results generally indicated that L2 speakers’ processing patterns are largely similar to those of L1 speakers. L2 speakers demonstarted sensitivity to grammatical constraints without an increased vulnerability to interference effects compared to L1 speakers. Reduced grammaticality effects, however, were observed specifically for S-V agreement in L2 speakers, while no such attenuation was found for reflexive-antecedent dependencies when compared to L1 speakers. Proficiency analysis revealed that this discrepancy may be driven by the performance of lower proficiency L2 speakers in S-V agreement, contradicting the more target-like processing profile observed in higher proficiency L2 speakers. Taken together, the overall pattern of results suggests that L1/L2 processing follows a similar manner, and any potential quantitative differences can be attributed to individual differences, particularly in L2 proficiency.