|Abstract:||The present study aims to investigate the nature of processing mechanisms in adult first-language (L1) and second-language (L2) sentence processing, specifically, whether adult L2 learners are capable of developing native-like sentence processing strategies. By testing a head-final language (Korean) with relatively free word order, this study also examines whether and how scrambling (switching word order between subject and object of a transitive verb) affects L1 as well as L2 sentence processing (Miyamoto & Takahashi, 2002; Yamashita, 1997). The main purpose of the present study is to examine how L1 and L2 speakers integrate linguistic input and build their sentential representations in online processing of canonical and scrambled sentences.
Previous research on L2 sentence processing disagrees as to whether and how L2 processing differs from L1 processing (e.g., Clahsen & Felser, 2006; Hopp 2006, 2010; McDonald, 2006; Sorace & Filiaci, 2006). Clahsen and Felser (2006) proposed the Shallow Structure Hypothesis (SSH), which predicts that L2 learners will rely on non-syntactic cues (e.g., plausibility or heuristic word order) instead of morphosyntactic cues (e.g., case-marking information) in sentence processing, and build less detailed syntactic representations. However, McDonald (2006) and Hopp (2006, 2010) attributed L1-L2 differences to processing difficulties rather than learners’ incomplete grammatical representations (the processing capacity approach, PCA). The PCA predicts that learners with advanced (or near-native) proficiency and/or high working memory spans could perform more like native speakers. Sorace and Filiaci (2006) proposed the Interface Hypothesis (IH) to account for the non-target-like response patterns found by near-native speakers, which predicts even near-native speakers will fail to perform like native speakers if integration of syntactic knowledge with other cognitive domains (e.g., pragmatics) is required.
The present study tests the three approaches by investigating whether L2 learners of Korean use morphosyntactic (case-marking), heuristic word order, and/or non-syntactic (plausibility or information structure) information when processing scrambled sentences (i.e., the filler-gap dependencies), as compared to native Korean speakers. This study also considers whether such factors as proficiency and working memory capacities play a role in attaining native-like processing. Three experiments are implemented in the present study; the main task is an online self-paced reading task for all three, which measures participants’ online comprehension of sentences. Additionally, an agent identification task tests L2 learners’ offline knowledge of word order and case markers in Korean, and an offline acceptability task measures participants’ knowledge of information structures in Korean. The results of the present study confirmed what the processing capacity approach (PCA) predicts, L1 and L2 differences can be attributed to processing difficulties rather than learners’ incomplete grammatical representations. Thus, the findings of the study suggest that L2 processing is not qualitatively different from L1 processing.