Key points are not available for this paper at this time.
Effective pronunciation instruction is essential for improving EFL learners’ communicative competence and overall learning satisfaction. However, traditional classroom practices often lack consistent, individualized feedback, limiting learners’ progress—especially in developing accurate pronunciation. While AI-powered tools have shown promise in informal L2 learning, few empirical studies have examined their integration into structured EFL pronunciation instruction—particularly classroom-based applications of intelligent CALL (ICALL) tools such as ChatGPT, a generative language model, and SpeechAce, an automatic speech recognition (ASR)-based feedback system. This study addresses this gap by investigating the effects of integrating these tools into classroom instruction to improve EFL learners’ pronunciation accuracy—specifically, segmental features (phoneme accuracy) and suprasegmental features (word stress patterns)—and L2 motivation, including Current L2 Self, Ideal L2 Self, and Intended Learning Effort. A sequential mixed-methods design was employed with 71 undergraduate EFL learners in Saudi Arabia. Quantitative results showed significant improvements in segmental pronunciation (phoneme accuracy at both word and sentence levels), with limited gains in suprasegmental stress patterns. Motivation increased significantly in Current L2 Self and Learning Effort; however, no significant within-group change was found in Ideal L2 Self, likely due to the experimental group’s slightly higher baseline score. Qualitative findings supported these trends, with students reporting positive perceptions of AI tools, citing real-time feedback, personalized practice, and increased engagement. This study contributes empirical evidence to the growing field of AI-assisted pronunciation instruction and supports the pedagogical viability of integrating tool-enhanced feedback into structured EFL contexts.
Abdelhalim et al. (Mon,) studied this question.