Asian Journal of Education and Social Studies

Taiwanese university EFL students often have limited opportunities to practice English speaking outside the classroom, while large class sizes restrict teachers' ability to provide immediate and individualized oral feedback. As a result, AI-assisted automated speaking assessment has emerged as a potential supplementary tool for enhancing speaking practice and oral proficiency. This study investigated the role of automated speaking assessment in supporting Taiwanese university EFL students’ oral proficiency. Specifically, it aimed to examine whether automated speaking assessment could enhance students’ fluency, pronunciation accuracy, and content organization, and to explore students’ perceptions of automated feedback in large EFL classes. The participants were 46 freshmen enrolled in a two-credit Academic English course at a university in Taiwan. TEEMI, an online speaking practice and assessment platform, was used as a supplementary tool to provide students with repeated speaking practice and immediate automated feedback. The study adopted a case-study design and collected both quantitative and qualitative data, including TEEMI speaking scores and students’ reflective writings. The results showed that students completed 136 speaking records across 11 speaking tasks, suggesting that the platform increased opportunities for oral practice. The proficiency distribution indicated that many students remained at the pre-A2 or A2 level, while some achieved A2+, B1, or B1+ performance. Qualitative findings further revealed that students perceived automated speaking assessment as useful for identifying pronunciation errors, improving fluency awareness, encouraging repeated practice, and reducing speaking anxiety. Among the speaking dimensions, pronunciation awareness appeared to benefit most from automated feedback, followed by fluency and self-monitoring. However, improvement in content organization was less evident, indicating that automated feedback alone was insufficient for developing coherent and well-structured oral responses. The findings suggest that automated speaking assessment can serve as a valuable supplementary tool in university EFL speaking instruction, particularly in large classes where teachers may not be able to provide immediate individualized feedback. Teacher guidance and classroom-based communicative practice remain essential for developing broader oral communicative competence.