AI-Enabled Instruction for L2 Speaking: Effects on Learners' Oral Proficiency

This study focuses on the application of artificial intelligence (AI) technology in second language (L2) oral instruction, aiming to address the long-standing practical difficulties in L2 speaking pedagogy, including large class sizes which result in limited opportunities for speaking and delayed feedback from instructors, lack of authentic communicative contexts, and high levels of speaking anxiety. This study adopts a quasi-experimental design. Two parallel classes of non-English majors from related disciplines in the first year at the author’s university were selected as research subjects. The experimental class (28 students, 2025 intake, Mathematics and Applied Mathematics) received an AI-driven innovative teaching model, which included pre-class human–computer interactive oral training, in-class in-depth human–computer dialogue and PPT presentations, and after-class human–computer interactive PPT video production. The control class (30 students, 2025 intake, Civil Engineering) followed the conventional oral English teaching approach. The research spanned one semester, and data were collected through final oral examination performance analysis and face-to-face interviews. The results indicate that students in the experimental class significantly outperformed the control class in terms of oral performance with the mean final oral score exceeding that of the control class by 1.47 points (p = 0.002). Meanwhile, AI-assisted oral training also exerted a positive transfer effect on students' comprehensive language proficiency (reading, translation, and writing). This study indicates that AI technology can positively impact L2 oral instruction, suggesting instructors consider revising L2 speaking pedagogy. In addition, the experiment revealed that over-reliance on AI may weaken students' autonomous thinking ability, so targeted suggestions for mitigating such risks are accordingly proposed.