Abstract
In research on second language (L2) processing, the processing of reference has been highlighted as a domain of particular difficulty, but the source of the difficulty is not well understood. The present study examines whether differences in the pronominal systems of the first language (L1) and L2 impact processing. We take a novel approach, testing a group of intermediate-advanced L2 learners in both their L1 (Mandarin Chinese) and L2 (English), allowing us to directly examine whether L2 learners show similar or different patterns when processing the L1 and L2. We also test a group of L1 English speakers. The study focused on two topicality-related factors, subjecthood and pronominalization, that have been shown to increase the prominence of an entity in the discourse, making it more likely that an entity in subject position (subjecthood) or an entity that has been referred to with a pronoun (pronominalization) will be considered as an antecedent for a subsequent pronoun. We developed a picture verification task with visual-world eye-tracking in both English and Chinese. This task provides a measure of both pronoun interpretation and online processing. Results showed subtle differences in how subjecthood and pronominalization are weighted in English and Chinese as L1s: pronominalization played a stronger role in L1 Chinese than in L1 English both in the interpretation measure and in the eye-movement data. The results for the L2 English learners showed an interesting pattern in which their results were more similar to the L1 English results on the measure of pronoun interpretation, but were more similar to the L1 Chinese results in the eye-movement data. These results show successful use of discourse cues in L2 pronoun interpretation but differences between L1 and L2 speakers during processing. It is proposed that decreased sensitivity to morphosyntactic information that is not present in the L1 (case on pronouns) leads to differences in L2 referential processing, in line with proposals that L2 learners face challenges with integrating different kinds of linguistic information online, particularly morphosyntactic information.