The impact of ChatGPT-assisted instruction on speaking performance, motivation, and self-regulated learning: a quasi-experimental study of intermediate EFL learners in China

Purpose: This study examined the effects of ChatGPT-assisted speaking instruction, compared with conventional communicative instruction, on the speaking performance, motivation, and self-regulated learning (SRL) of intermediate Chinese EFL learners. Design/methodology/approach: A quasi-experimental pretest-posttest control-group design was conducted with 143 intermediate EFL learners across four private language institutes in Chongqing, China. Over a 12-week period, the experimental group engaged in 10 ChatGPT-assisted speaking sessions featuring interactive practice and automated formative feedback, whereas the control group received conventional communicative speaking instruction. Data were collected using validated speaking assessment rubrics, an EFL motivation questionnaire, and a self-regulated learning scale. One-way analyses of covariance (ANCOVA) were performed to assess post-intervention differences while controlling for baseline pretest scores. Findings: After controlling for pretest performance, the ChatGPT-assisted group demonstrated statistically significant improvements over the control group in speaking performance (F (1,140) = 32.06, p < .01) and self-regulated learning strategies (F(1,140) = 505.32, p < .01) However, the between-group difference in motivation did not reach statistical significance (F(1,140) = 3.29, p = .072, ηp2 = .02), though adjusted descriptive means favored the intervention cohort. Originality/value: While generative AI tools have attracted considerable attention in language education, empirical research evaluating their concurrent impact on linguistic output and psychological/metacognitive attributes remains limited. This study contributes empirical evidence regarding how interactive generative AI can foster autonomous, regulated learning behaviors alongside oral proficiency development, providing practical and pedagogical insights for curriculum designers and EFL educators integrating AI into communicative language classrooms.