Integrating Multimodality in Emotion Moderation: The Effects of Multimodal AI on EFL Speakers’ Boredom, Enjoyment, and Anxiety

ABSTRACT AI‐powered tools are increasingly used in EFL speaking classes to support learners’ emotional regulation, but most studies focus on text‐based AI for content generation or interaction. This overlooks the multimodal nature of learning and teaching, highlighting the potential value of multimodal AI with video and images. Therefore, this study, informed by positive psychology (PP), systematically compared the effects of three interventions: multimodal AI, text‐based AI, and traditional instruction (teacher and peer‐led speaking activities) in terms of foreign language boredom (FLB), foreign language enjoyment (FLE), and foreign language anxiety (FLA). A total of 159 first‐year intermediate EFL university students from three intact classes participated in an eight‐week classroom intervention. The findings revealed that both AI‐based interventions outperformed traditional instruction in fostering more positive emotional experiences. In particular, multimodal AI showed clear advantages in enhancing learners’ enjoyment and reducing boredom, likely due to its integration of visual and auditory stimuli that enriched interaction and engagement. However, this advantage did not extend to reducing FLA, where no significant difference was observed between multimodal and text‐based AI. These results highlight the value of AI multimodality in speaking classes, both for fostering positive emotions and for aligning instruction with the multimodal nature of communication and cognition. The study also shows how technological design shapes affective outcomes, offering practical implications for emotionally responsive AI integration in foreign language (FL) teaching.