How multimodal GenAI enhances EFL learners’ language achievement: a parallel mediation analysis on the roles of psychological capital and emotion regulation

The incorporation of multimodal Generative AI (GenAI) into teaching is transforming how students engage with language beyond conventional text-based instruction. Although previous studies have examined the instructional benefits of AI-based language learning, less attention has been paid to the emotional and psychological mechanisms through which multimodal GenAI affects EFL learners’ language achievement. To address this gap, and drawing on Self-Determination Theory (SDT), this study examines how multimodal GenAI improves EFL students’ language achievement through the parallel mediating effects of emotion regulation (ER) and Psychological Capital (PsyCap). Using a quasi-experimental design, 230 Chinese EFL students were assigned to either a GenAI-integrated experimental group using image-, text-, and audio-supported activities or a text-based control group. Over eight weeks, both groups completed curriculum-aligned language learning activities. Language achievement was measured using the Preliminary English Test (PET) as both a pretest and a posttest. The parallel mediation model showed that ER and PsyCap jointly mediated the effect of multimodal GenAI on EFL learners’ achievement. ER reflected students’ regulation of immediate emotional responses, while PsyCap captured enduring psychological resources that supported achievement. This study explains the dual emotional-psychological mechanisms through which multimodal GenAI supports EFL learning outcomes. These results offer theoretical and practical implications for designing cognitively supportive and psychologically empowering GenAI-based EFL instruction.