Supporting Spoken Interaction in Academic Settings: Students’ Use of Real-Time AI Transcription in EFL Classrooms

Abstract: This study explores EFL students’ use of real-time AI transcription to support spoken interaction in academic classroom settings and examines their perceptions of its role during oral activities. Adopting an exploratory mixed-methods design, the study integrates qualitative data from classroom observations and semi-structured interviews with quantitative data from a self-report questionnaire. The qualitative findings reveal that students employ AI transcription in self-directed and interaction-sensitive ways, such as monitoring speech accuracy, managing communication breakdowns, supporting turn-taking, and reducing anxiety during real-time interaction. These practices indicate that AI transcription functions not merely as a corrective aid but as an interactional resource embedded within ongoing classroom discourse. Quantitative results further show generally positive student perceptions regarding the usefulness of real-time AI transcription for enhancing clarity, confidence, and engagement in spoken interaction. By foregrounding students’ situated practices and experiences, this study contributes to emerging research on AI-mediated spoken communication in EFL contexts. The findings suggest that real-time AI transcription holds pedagogical potential when integrated flexibly and responsively to learners’ communicative needs, offering insights for educators seeking to support spoken interaction through AI-enabled technologies.