From gaze-based evidence of strategic processing in multimodal L2 speaking task: Towards mixed-method approach

Abstract

Despite the growing attention to the use of multimodal prompts in speaking assessments, evidence on the effect of input modes on real-time strategic behavior beyond fixation counts remains underexplored. This eye-tracking study investigated gaze behaviors of 43 adult speakers as they engaged with text, image, and video prompts during the preparation and response phases, focusing on gaze patterns across modes and phases, individual differences, and the correspondence between gaze-based and self-reported strategies. Ten gaze indicators across 10 areas of interest were analyzed using repeated-measures, mixed ANOVAs, principal component analysis, and correlation analysis. Results showed mode- and phase-specific features. Exploratory analysis also indicated individual differences (gender and proficiency) in gaze behavior across modes. Gaze-based strategic components suggest mode dependence of strategic gaze allocation and no significant associations with self-reported strategy use, after controlling for gender, proficiency, and task performance. This finding might suggest a complementary view on the construct of strategic competence. Implications for the cognitive design and validation of a multimodal speaking assessment for L2 learners were discussed.