The Impact of AI‐Assisted L2 Learning on Learners’ Emotions: A Meta‐Analysis

ABSTRACT The scope of research on AI‐assisted second/foreign language (L2/FL) learning has expanded beyond learning outcomes to focus on learners’ emotional experiences. Despite their proliferation, existing empirical findings are inconsistent, and few studies have systematically synthesized the impacts of AI‐assisted L2 learning on learners’ emotional experiences. This meta‐analysis scrutinized 31 empirical studies with 50 effect sizes to examine the impact of AI‐assisted L2 learning on learners’ emotions. Results from a three‐level model indicated that AI‐assisted L2 learning significantly increased positive emotions ( g = 0.489, p = 0.004) and reduced negative emotions ( g = −0.428, p = 0.048), with effect sizes of approximately medium magnitude. In terms of specific emotion constructs, AI‐assisted L2 learning was found to significantly reduce boredom ( g = −0.760, p = 0.042), whereas effects on enjoyment and anxiety were not significant. Furthermore, moderator analyses identified several significant factors influencing positive emotions, including participant source, AI function, interactivity modality, feedback type, targeted L2 skills, and intervention duration, while no significant moderators were detected for negative emotions. Participants from language institutes, as well as interventions employing adaptive learning systems, assessment and monitoring systems, text‐based interactivity, hybrid feedback, writing and comprehensive tasks, and interventions lasting longer than eight weeks, demonstrated greater effectiveness. This study advances the understanding of the emotional dimension of AI‐assisted L2 learning and underscores the importance of strategically leveraging AI tools in educational practice to enhance learners’ emotional experiences and engagement.