Unpacking EFL students’ engagement with GAI in digital multimodal composing

Abstract

While existing studies underscore the potential of Generative Artificial Intelligence (GAI) to streamline digital multimodal composing (DMC) and elevate product quality, research systematically synthesizing learners’ multi-dimensional engagement with GAI in DMC remains limited. Adopting a qualitative research method, this study investigated eight Chinese EFL undergraduates’ cognitive, behavioral, and emotional engagement with GAI in DMC and the influencing factors. Data sources included semi-structured interviews, written reflections, screenshots, and final DMC products. The findings showed that, behaviorally, students leveraged different GAI tools to accomplish seven major goals, including textual content generation, analytical processing, content reorganization, content refinement, translation, task management, and multimodal content generation. Cognitively, students implemented four strategies to optimize GAI’s unsatisfactory output, namely, manual refining, revising prompts, negotiating with GAI, and leveraging the strengths and offsetting weaknesses of different GAI tools. Students expressed both positive and negative emotions towards GAI integration in DMC due to its affordances and limitations. Both individual and contextual factors influenced students’ engagement, including learner agency, GAI literacy, L2 proficiency, peer recognition, division of labor, and contextual limitations. Based on the findings, we propose a Model of Learner Engagement in GAI-assisted DMC and provide actionable insights for optimizing GAI-integrated DMC pedagogy.