Existing research has widely examined the effects of AI-assisted speaking training on language performance and learning outcomes; however, the experiential mechanisms through which AI-assisted speaking training affordances (AISTA) shape learners' communication-oriented behaviors remain underexplored. In particular, limited empirical evidence explains how such affordances are translated into learners’ willingness to communicate (WTC) via flow experience (FE). Drawing on ecological affordance theory and flow theory, the present study conceptualizes AISTA as a core construct and proposes a moderated mediation model in which AISTA influences WTC through FE, with learner agency (LA) capturing individual differences in AI-supported speaking contexts.
Survey data were collected from 247 Chinese EFL university students following a four-week AI-assisted speaking practice and analyzed using structural equation modeling (SEM) with latent main effects and interaction terms. Bias-corrected bootstrapping with 5000 resamples was employed to test mediation and moderation effects. The results showed that AISTA was positively associated with FE, and FE was positively associated with WTC. Flow experience emerged as a key mediator linking AISTA to WTC. Moreover, learner agency significantly moderated the experiential pathway, yielding a pattern of conditional partial mediation: the indirect effect of AISTA on WTC via FE was significant for learners with moderate to high levels of agency but not for those with low agency.
By elucidating the affordance–experience–behavior mechanism underlying AI-assisted speaking training, this study extends ecological affordance theory and flow theory to AI-supported second language learning contexts and highlights learner agency as a critical condition for translating technological affordances into communicative motivation.
扫码关注我们
求助内容:
应助结果提醒方式:
