Real-Time Interactive AI and Speaking Anxiety among English Majors in an Exploratory Mixed-Methods Study
This exploratory mixed-methods study examines speaking anxiety and experiences of real-time interactive AI practice among English majors at a college in China. An online survey yielded 98 valid responses from 121 returned questionnaires. Twelve volunteers subsequently participated in a short-term, single-group intervention using a customized Doubao agent for IELTS Speaking Part 3 practice, and four participants completed semi-structured interviews. The survey indicated concern about public speaking and others’ evaluations, alongside frequent use of AI for expression-related assistance. The reported pre–post means declined on five negatively worded anxiety items. Interviewees described accessible practice, reduced perceived evaluation pressure, and a patient interaction pace, while also identifying limitations in feedback and conversational authenticity. These findings suggest potential benefits of AI as a supplementary speaking partner. However, the small convenience sample, lack of a comparison group, short duration, and reliance on self-report limit causal interpretation and generalization. Further work should establish the measurement properties of the adapted questionnaire and examine transfer to human interaction.
Authors
- Meiru Wen (ORCID: https://orcid.org/0009-0005-9099-2393)
- Yanzhen Ma
- Xinyi Wang
Publication Details
- Journal
- Frontiers in Humanities and Social Sciences
- Published
- 2026-11-18
- DOI
- https://doi.org/10.54691/tt2fx597
- Primary Topic
- EFL/ESL Teaching and Learning
- Type
- article
- Field-Weighted Citation Impact
- 0.00