Although generative artificial intelligence (AI) tools show significant promise for foreign language learning, there is still a pressing need for empirical research on their implementation and effectiveness. In this quasi-experimental mixed-methods study, we examined the impact of speaking practice with avatar-assisted artificial intelligence (AI) chatbots, specifically the Call Annie application, on the English-speaking skills of pre-service English language teachers and their perceptions of AI chatbots as speaking partners. A total of thirty-five first-year ELT students in Türkiye participated in the study, with nineteen in a teacher-led control group and sixteen in an AI-based experimental group practicing with Call Annie for eight weeks. Data consisted of pre- and post-test speaking tests, the Foreign Language Speaking Anxiety (FLSA) questionnaire, and scales measuring willingness to communicate (WTC) and self-perceived communication competence (SPCC). Quantitative data were analyzed using group comparisons and repeated-measures ANOVA, while qualitative data were analyzed via inductive thematic analysis. Results indicated similar improvements in speaking skills and decreases in FLSA across groups, suggesting that these gains may be associated not only with the type of intervention but also with the repeated speaking practice involved in the study. While the control group showed slightly higher WTC and SPCC, the differences were minimal. Qualitative findings revealed that participants preferred the teacher as a speaking partner and attributed their speaking anxiety primarily to themselves. Overall, the findings suggest that avatar-assisted AI chatbots can serve as viable alternatives for speaking practice in resource-limited educational contexts, although they do not appear to offer a clear advantage over teacher-led interaction.
Aydın et al. (Sat,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: