Practicing with AI Conversational Agents and the Reduction of English-Speaking Anxiety: Evidence from First-Year Undergraduates at a Bangladeshi Public University

Abstract

Reducing speaking anxiety is essential for developing second-language speaking skills. This paper presents an evidence-based study on the incorporation of AI conversational agents to alleviate English-speaking anxiety among first-year undergraduate EFL learners at a semi-urban public university outside the capital city of Bangladesh. The study examines students' and teachers' perspectives and experiences regarding the use of AI tools for developing English-speaking skills and investigates how these tools can facilitate confidence-building in speaking English. A total of 151 first-year undergraduate students from different disciplines of a public university and five ELT professionals from public and private universities participated in this mixed-methods convergent design study. Data were collected through questionnaires, pre- and post-intervention speaking performance recordings from a group of ten purposively selected participants, and semi-structured teacher interviews. The data were analyzed using both quantitative and qualitative methods. The findings suggest that a careful and systematic approach to incorporating AI conversational agents into English language courses may support learners' English-speaking development. In particular, AI conversational agents may help alleviate English-speaking anxiety, boost learners' confidence, and improve their speaking fluency.