Proceedings of the Annual Meeting of the Cognitive Science Society

This study examines how L1-Mandarin-Chinese L2-Korean learners comprehend constructions expressing transitivity within a good-enough processing framework. Using picture selection with webcam eye-tracking, we compare learners with native speakers in two studies. Study 1 crosses voice (active vs. passive), word order (canonical vs. scrambled), and verb position (verb-final vs. verb-initial). Learners are accurate for canonical actives, but perform below chance on passives and show reduced target looks in critical regions, consistent with a persistent agent-first bias and limited use of case-marking and passive morphology. Study 2 obscures case markers to isolate verbal morphology; learners favour agent-first interpretations and show only weak sensitivity to passive morphology. Across two studies, proficiency predicts more target-like performance, especially in non-canonical configurations. These findings highlight the role of cue alignment/(re)weighting in L2-acquisition pathways and a proficiency-mediated shift towards more algorithmic parsing, suggesting the dynamic and adaptive nature of L2 good-enough processing.