AI-Supported University Efl Education for Speaking Confidence and Learner Autonomy

Abstract

This study explores how artificial intelligence (AI)-supported applications contribute to speaking confidence and learner autonomy among Indonesian university EFL students. Employing a qualitative descriptive design, the study involved eight students from the English Education Study Program at Universitas Kristen Indonesia Toraja who had experience using AI-based applications such as ChatGPT and Duolingo for English-speaking practice. Data were collected through semi-structured interviews, observations, field notes, and audio recordings, and were analysed thematically using the interactive framework of data condensation, data display, and conclusion drawing. The findings revealed five interrelated themes: developing speaking confidence through low-pressure practice, receiving immediate corrective feedback, extending speaking practice beyond classroom limitations, developing learner autonomy, and sustaining motivation through interactive application features. Students perceived AI as a flexible and non-judgmental practice environment that enabled repeated rehearsal, pronunciation and grammar correction, self-paced learning, and greater preparedness before speaking with lecturers or peers. The novelty of the study lies in conceptualising AI as a pre-communicative support environment that bridges private rehearsal and authentic human interaction rather than replacing lecturers or classroom communication. However, the findings are limited to participants’ qualitative experiences and do not establish causal improvement in speaking proficiency. The study contributes to AI-supported language education by highlighting the interconnection between psychological readiness, autonomous learning, corrective feedback, and responsible pedagogical integration of AI in university EFL contexts.