Sentence processing in English natives and non-natives: evidence from attachment resolution and agreement processing

How similarly or differently native speakers (L1ers) and non-native speakers (L2ers) resolve attachment ambiguities and compute agreement during online reading has been widely examined to inform the theoretical debates on whether L1 and L2 acquisition and processing are qualitatively different. While L2 attachment resolution has been studied using either globally (e.g., We called the brother of the man who bought himself a book yesterday.) or temporarily ambiguous relative clauses (e.g., We called the brother of the woman who bought himself a book yesterday.), how L2ers resolve attachment ambiguities when reading both types of relative clauses is little known. While previous studies have examined the roles of linguistic information and participant-level individual differences in L1 and L2 attachment resolution separately, whether these factors modulate L1 and L2 processing to a similar extent has not been studied within the same study. Also, while studies that investigated agreement processing in L2ers with an L1 that does not have agreement (e.g., Chinese) have reported contradicting findings, what underlies the existing inconsistencies is unknown. Further, little is known about how number is specified, in terms of on nouns and other constituents such as demonstratives, may facilitate computation of non-local agreement in L1 and L2ers. Thus, this thesis reports three studies that address these issues, using both offline and online tasks. Study 1 showed that despite some quantitative differences, L1 and L2 attachment resolution are guided by the same parsing strategy, indexed by a low attachment preference, and modulated by similar factors, such as individual differences in lexical automaticity. Study 2 and Study 3 further demonstrated that L1 and L2 agreement computation are not fundamentally different and are modulated by additional number marking similarly, even when L2ers’ L1 does not instantiate relevant features. Taken together, the findings from these three studies have lent support to the theories that suggest no qualitative differences between L1 and L2 acquisition and processing