An Online Tool for Practicing Oral Summary: Can an Effective Tool Be Built With ASR?

Orally summarizing academic lectures from notes is a key integrated speaking skill for English for Academic Purposes, yet few studies offer concrete training methods. This study explores the development and effectiveness of an online practice tool using freely available automatic speech recognition (ASR) to support oral summary practice. While ASR is commonly used for pronunciation training, its potential for broader oral skill development remains underexplored. The tool was implemented in a university-wide EAP curriculum in Japan, with data collected from 652 students. Their first, final, and total number of tool usages were analyzed using a Hotelling’s T2 test, which revealed significant improvement in students’ spoken word count, mean length of utterance, and use of textbook target phrases (p < 0.001, ηp2 = 0.384). Higher average scores on these metrics, along with content keyword usage, were associated with stronger end-of-semester oral summary performance. These findings suggest the tool was effective and that its feedback measures were appropriate. However, the study also highlights areas for refinement, including the need to revise keyword lists and improve the scoring model based on user data. The results support the use of ASR-based tools for oral summary training, while emphasizing the importance of data-driven feedback.