Effects and contributing factors of an automatic speech recognition dialogue platform on international students’ Chinese speaking ability

This sequential explanatory mixed-methods study examined whether an artificial intelligence (AI)-driven automatic speech recognition dialogue platform improves international students’ Chinese speaking ability and how affective and technological factors influence it. One hundred and thirteen intermediate learners completed pretest and posttest surrounding 9 role-play sessions; scores rose significantly. Multiple regression showed that willingness to communicate (WTC) promoted gains whereas speaking anxiety impeded them, together explaining 58% of posttest variance; motivation, self-esteem, perceived usefulness and perceived ease of use were non-significant. Thematic interviews with 12 volunteers corroborated the quantitative pattern: authentic, low-stakes AI dialogues heightened speaking ability, but drill repetition and scant social presence dampened sustained motivation. These findings underscore the primacy of affective readiness in technology-assisted language learning, extend speech-recognition evidence to Chinese contexts, and offer practical guidance: learners should leverage AI self-assessment to rehearse confidently; teachers should nurture WTC and mitigate anxiety; administrators can deploy AI tools to ease instructor shortages while maintaining human interaction; and developers should focus on affective factors with richer social features to sustain engagement.