An Empirical Study of the Effect of ASR-Supported English Reading Aloud Practices on Pronunciation Accuracy

Abstract

Automatic Speech Recognition (ASR) technology has excellent potential for improving learners’ English pronunciation. However, the specific effects of ASR on pronunciation improvement are still unclear. Taking learners’ language proficiency into account, the present study aims to investigate the specific effects of ASR-aided reading aloud practices on English pronunciation accuracy. 50 senior high school students from a city in southern China are participants. They are divided into two groups: low- and high-proficient groups. Two tests and specific pronunciation error analysis are used to test students’ pronunciation improvement, while a post-study survey is conducted to explore students’ perception of using ASR. The results indicate that, for low-proficient learners, ASR-supported reading aloud practices help them improve global pronunciation accuracy and correct word error and deletion error; while for high-proficient learners, the training doesn’t improve their global pronunciation but helps them correct deletion error and addition error. Both groups show positive perception of ASR-supported training.