Speaker
Description
Speech-recognition technologies are increasingly promoted as tools for supporting spoken language practice in computer-assisted language learning (CALL). However, far less attention has been given to the practical challenges of implementing such tools in authentic educational settings. This study reports on the real-world deployment of a speech-recognition–based application designed to support Japanese medical students’ practice of English medical interview communication.
The application enables learners to role-play scripted doctor–patient dialogues through three stages of production: reading aloud, shadowing, and recall. Learner speech is automatically converted to text and compared with the target script, generating an accuracy score that determines whether the dialogue progresses. The application was released in April 2025 to medical students who consented to participate in the study.
The original research design aimed to compare the effectiveness of app-based practice with traditional study using printed dialogue materials. However, adoption of the application proved extremely limited. Among 258 eligible students, fewer than 20 downloaded the application and only five used it for English practice. Consequently, the planned comparison of learning outcomes was not feasible.
Rather than focusing on learning effects, this presentation examines the structural factors that limited student participation. In particular, voluntary participation requirements and the absence of links between app use and course assessment appear to have significantly reduced engagement. The findings highlight important constraints on the adoption of optional digital learning tools and underscore the need to consider institutional and motivational factors when integrating CALL technologies into formal educational contexts.