Effects of an automatic speech recognition system with peer feedback on pronunciation instruction for adults

Abstract

This study examined the difference in adults’ pronunciation performance with peer feedback and individual practice when using an automatic speech recognition (ASR) system. The same ASR software was used in both the comparison (n = 31) and the experimental group (n = 33) for 12 weeks. The participants were working adults in Taiwan. During the weekly lesson, the participants dictated a text to the ASR software, Speechnotes, on its website and then practiced misidentified words by themselves (the comparison group) or with feedback from team members (the experimental group). Learners’ pronunciation was measured in reading aloud and spontaneous conversation tasks before and after the treatment. Participants’ satisfaction and perceptions of the ease of use of the software were also assessed at the end of the course. The results revealed a significant difference in the two groups’ pronunciation post-tests, suggesting that peer feedback is more effective than individual practice for the correction of pronunciation. The experimental group was more satisfied than the comparison group with the software. However, no differences in accentedness or perception of ease of use were evident. The findings, ASR software implementation challenges, and implications for adult language learning are discussed.