<![CDATA[Pronunciation remains one of the most challenging skills for EFL students in Indonesia, particularly at the elementary school level where teachers often lack English expertise and class time is limited. This study examined the effectiveness of the ELSA Speak application as an AI-powered pronunciation learning tool in addressing these challenges. Using a true experimental design with pre-test and post-test measures, the study was conducted at UPT SD Negeri 79 Gresik involving nine sixth-grade students randomly assigned to experimental (n=5) and control (n=4) groups. The experimental group received three treatment sessions using ELSA Speak, while the control group followed conventional instruction. Results revealed a substantial and statistically significant improvement in pronunciation scores in the experimental group compared to the control group (p=0.016), with the experimental group demonstrating a dramatic gain from pre- to post-test (p0.001). These findings indicate that ELSA Speak's features — particularly its accurate pronunciation modeling and real-time automated feedback — effectively compensate for the constraints of conventional EFL instruction in elementary settings. This study contributes novel evidence for the application of mobile-assisted language learning at the elementary school level in Indonesia, a context previously underexplored in the literature. Practically, integrating ELSA Speak into classroom instruction is recommended as a feasible and impactful strategy to enhance pronunciation learning, provided that device access and internet connectivity are adequately supported.]]>
