Promoting Oral Communication Among Chinese University EFL Learners Through AI ‐Mediated Interaction: A Mixed‐Methods Investigation

Abstract

Artificial intelligence (AI) is increasingly being integrated into language education, yet evidence regarding its influence on learners' emotional experiences and oral communication remains limited. This mixed‐methods study examined the effects of AI‐mediated speaking practice on university EFL learners' oral anxiety, speaking enjoyment, self‐perceived oral communication competence and willingness to communicate (WTC). A total of 105 intermediate‐level learners participated in an eight‐week quasi‐experimental intervention in which the experimental group engaged in regular AI‐mediated speaking activities, whereas the control group completed conventional self‐directed speaking practice. Quantitative data were analysed using descriptive and structural equation modelling techniques, and qualitative insights were obtained through semi‐structured interviews with 16 participants. Compared with the control group, learners in the AI‐mediated condition reported lower speaking anxiety and higher speaking enjoyment, self‐perceived communicative competence and willingness to communicate. The structural model indicated that the relationships among AI‐mediated practice, emotional experiences, perceived competence and WTC were consistent with the proposed theoretical framework, while interview findings suggested that learners perceived the AI‐supported environment as encouraging repeated practice, greater confidence and communicative engagement. These findings support the potential of conversational AI as a complementary tool for facilitating learners' emotional experiences and perceived communicative readiness in EFL speaking instruction, while highlighting the need for future research using behavioural measures and interaction‐equivalent comparison conditions to further examine these effects.