Creation of a Multi-Modal Eye-Tracking Dataset for L2 Idiomatic Processing
Eye-Tracking; Idiomatic Expressions; Second Language (L2); Psycholinguistics; Cognitive Effort.
This work investigates the cognitive processing of idiomatic expressions by non-native (L2) English speakers, analyzing how different proficiency levels influence the reading of non-literal language. The research utilizes eye-tracking to monitor university students, stratified from beginner (A1) to proficient (C2) levels, regarding stimuli that contrast figurative and literal usages. The study employs fixation and regression metrics to measure cognitive effort and processing difficulty in real-time. Analyses indicate an inverse correlation between proficiency and the number of regressions: the C2 group demonstrates stability and efficiency, suggesting direct access to meaning, while intermediate levels (B1/B2) exhibit heterogeneous patterns and high rates of re-analysis, evidencing the effort regarding the non-compositional nature of idioms. The work consolidates an annotated dataset that connects physiological metrics to psycholinguistic theories of second language acquisition.