<![CDATA[Bridging the Affective-Linguistic Gap: A Mixed-Methods Exploration of AI-Assisted Speaking Practice and Willingness to Communicate. Objectives: This study examined the effectiveness of AI-assisted learning in enhancing students’ Willingness to Communicate (WTC) and speaking proficiency in an Indonesian higher education context. It also explored students’ perceptions of AI-based tools' support for communicative confidence and readiness during speaking practice. Methods: A convergent mixed-methods design was employed with 60 undergraduate students enrolled in a Business English course. The experimental group received six AI-assisted speaking sessions (100 minutes each) using the SmallTalk2Me platform, while the control group received conventional instruction. WTC and speaking proficiency were measured using parallel pre- and post-tests, and qualitative data were collected through open-ended reflection surveys. Quantitative data were analyzed using descriptive statistics and two-way mixed ANOVA, and qualitative data were analyzed thematically. Findings: Quantitative results showed a significant main effect of time, indicating that students in both groups improved their speaking proficiency and communicative readiness across the semester. However, neither the group effect nor the interaction effect reached significance, suggesting that AI-assisted practice did not produce statistically greater gains than conventional teaching. Qualitative findings, however, revealed perceived affective benefits among AI users, including reduced speaking anxiety, increased confidence, and appreciation of a low-pressure practice environment with instant feedback and opportunities for repetition. Some students also reported challenges, including occasional misrecognition in AI feedback, dependence on stable internet access, and interactions that felt less natural than human communication. Conclusion: The findings indicate that AI-assisted learning did not yield statistically greater gains in speaking proficiency or WTC than conventional instruction. Although qualitative data suggest that some learners perceived changes in their affective experiences during AI-assisted speaking practice, these perceived advantages did not translate into statistically superior outcomes. Overall, the results indicate that the pedagogical impact of AI-assisted learning remains limited within the scope of the present study. Keywords: AI-assisted learning, higher education, language learning, speaking proficiency, willingness to communicate.]]>
Copyrights © 2025
