AI-assisted speaking instruction and English as a foreign language learners’ oral accuracy and fluency: a quasi-experimental study

Abstract

This study investigated the effects of artificial intelligence (AI)-assisted speaking instruction using a generative AI (GenAI) chatbot on Chinese vocational college English as a foreign language (EFL) learners’ oral accuracy and fluency. A quasi-experimental pretest-posttest control group design was adopted with 80 students. The experimental group engaged in AI-mediated speaking activities, while the control group received conventional instruction. Oral performance was assessed using analytic rubrics adapted from international English language testing system (IELTS) criteria. Results showed significant improvement in oral fluency for the experimental group, while gains in accuracy were not statistically significant. These findings suggest that GenAI chatbots provide interaction-rich environments that enhance fluency development but require complementary form-focused support to improve accuracy. Implications are discussed for integrating AI tools into vocational EFL speaking instruction.