Tracing Longitudinal Changes in Learners' Experience and Performance During Microsoft Copilot Voice‐Supported English Speaking Practice

Abstract

This article reports a one‐group time‐series study of Microsoft Copilot Voice (MCV) as a structural scaffold in a mandatory university English course in China. Sixty‐eight environmental engineering undergraduates participated in an MCV‐supported speaking sequence embedded in regular instruction. Oral performance and six dimensions of second language learning experience (L2LE)—positive emotions, negative emotions, engagement, relationships, meaning, and accomplishment—were measured at three time points: pre‐task, main task, and post‐task. Results from latent growth curve modeling showed significant positive slopes for oral performance and the positive L2LE dimensions, together with a significant negative slope for negative emotions. While these findings highlight the potential of MCV to create a dynamic, supportive environment that enhances learners' affective experiences and oral proficiency in L2 learning, they are best interpreted as developmental patterns during an MCV‐supported instructional sequence rather than evidence of MCV's independent causal effect, considering the single‐group design of this study. Pedagogical directions are offered for using generative AI voice tools as rehearsal spaces while preserving teacher guidance, peer interaction, and critical reflection.