ASR-based dictation practice for second language pronunciation improvement

Abstract

In pronunciation learning, there is a need for resources and tools that help students monitor their speech or provide feedback on errors. While researchers have seen ASR-based technologies as potential tools, little attention has been paid to dictation programs, which have been criticized for low levels of recognition, but offer advantages such as accessibility and flexibility. This study examines two groups of learners in a pronunciation workshop: CONV, which had fully face-to-face instruction, and HYBRID, which had half of the instruction face-to-face and half using the computer, practicing production using a dictation program, Windows Speech Recognition. Results show that both groups improved from pre- to post-test and that there were no statistically significant differences between the two groups. Results indicate that dictation programs may be useful as a complement to face-to-face pronunciation teaching, especially if in-class time for pronunciation teaching is limited.