The Impact of AI Chatbot Integration on Vietnamese EFL Learners' Speaking Performance, Fluency, and Confidence

Speaking proficiency remains a significant challenge for Vietnamese EFL learners due to limited practice opportunities, heightened anxiety, and psychological barriers. This study investigated the impact of AI chatbot integration on Vietnamese university students' speaking performance and confidence. Employing a one-group pre-test post-test quasi-experimental design, 38 intermediate-level EFL learners engaged in structured AI chatbot practice three times weekly for eight weeks. Speaking performance was assessed using IELTS-style tests evaluating fluency, grammatical accuracy, pronunciation, and lexical resource, while speaking confidence was measured through an adapted anxiety questionnaire. Results revealed significant improvements in both speaking performance (d = 1.45) and confidence (d = 1.24), with all performance dimensions demonstrating substantial gains. Notably, lower-confidence learners experienced greater confidence increases (M = 9.82 vs. M = 5.67), and performance improvements positively correlated with confidence gains (r = .48). Individual variability analysis identified that while 89.5% showed performance improvement, 10.5% experienced minimal change or decline, attributed to inconsistent practice and technical difficulties. These findings suggest that AI chatbot integration effectively addresses both linguistic and affective dimensions of speaking development in Vietnamese EFL contexts. The results hold important implications for technology-enhanced language pedagogy, particularly for learners experiencing speaking anxiety and requiring additional practice opportunities beyond traditional classroom instruction.