L1 and L2 Processing of Filler-Gap Dependencies in Mandarin Chinese Speakers: Evidence from Self-Paced Reading Times and Event-Related Potentials

Abstract

Language processing is a complex task requiring our brain to rapidly integrate various linguistic information from strings of unfolding words and quickly build comprehension. While previous studies have shown that native speakers (L1) employ both incremental and predictive processing in real time (e.g., Clifton & Frazier, 1989; Altmann & Kamide, 1999), debates persist regarding how second language speakers (L2) process sentences in real time and whether their processing mechanisms align with or differ from those of native speakers. In the field of L2 processing, some scholars believe that L1 and L2 processing share qualitatively similar mechanisms, and differences in L2 processing profiles arise from individual factors or language features (e.g., Dallas et al., 2023; Sabourin et al., 2014; Sabourin & Stowe, 2008; Berghoff, 2022; Grüter & Rohde, 2021). Others argue for qualitative differences between L1 and L2 processing (e.g., Clahsen & Felser, 2006; Pakulak & Neville, 2011), and these differences are especially apparent when the sentence structure includes filler-gap dependencies, a non-local syntactic structure requiring parsers to establish a connection between a displaced filler and its corresponding syntactic gap over distance (e.g., Marinis et al., 2005; Dong et al., 2023). The primary goal of this dissertation thus is to examine how L1 and L2 speakers process filler-gap dependencies in real time and whether they show qualitatively similar or different processing profiles and underlying neural mechanisms in response to both syntactic and semantic cues.

Three experiments were conducted in the dissertation. Experiment 1 used a self-paced reading paradigm to investigate how L1 English speakers and L2 speakers with L1 Mandarin Chinese process filler-gap dependencies with varying dependency lengths (long vs. short) and filler plausibility (plausible vs. implausible). The results showed that both L1 and L2 groups, regardless of their L2 proficiency and years of immersion, exhibited increased reading times at the verb, the position that indicates a potential syntactic gap, yet their sensitivity to filler-verb mismatches was observed only in the second post-verbal position. These findings indicated active and sequential processing of syntactic and semantic information. Then, using the same paradigm and within-subject factor manipulations, Experiment 2 investigated whether native Mandarin Chinese speakers process Chinese topic structures similarly to how English speakers process filler-gap dependencies. The results revealed that native Mandarin Chinese speakers, like English speakers, actively searched for gaps, evidenced by elevated reading times at the first verb they encountered and sensitivity to semantic mismatches shortly thereafter. Previous research has shown that active gap-filling strategy can be interpreted as a response to the characteristics of wh-dependencies. (Wagers & Phillips, 2009). Thus, the findings lend support to the existence of movement-derived filler-gap dependencies in Mandarin Chinese, despite its wh-in-situ nature. Lastly, Experiment 3 used event-related potentials (ERPs) to examine the brain responses of L1 and L2 speakers with the manipulation of filler plausibility and gap licensor types, which were either a transitive verb or a preposition. The results showed no overall group difference. However, while L2 speakers with high proficiency showed native-like but attenuated and delayed predictive processing, L2 speakers with low proficiency showed no evidence of prediction and experienced syntactic reanalysis.

Overall, this dissertation revealed qualitatively similar processing mechanisms between L1 and L2 speakers, with L2 differences being explained by individual factors. Moreover, the findings lent support to a movement-derived account of Chinese topic structures, despite Chinese being a wh-in-situ language.