Effects of Speech Recognition-based Pronunciation Feedback on Second-Language Pronunciation Ability

Abstract

This study’s goal was to determine whether receiving a particular type of feedback on nativeness of second language accent positively influenced pronunciation over time. Forty-five native speakers of American English of beginning to intermediate Spanish ability were randomly assigned to three groups. The first group was asked to practice Spanish using speech recognition-based software that provided scores of nativeness of pronunciation. The second group practiced with software that was identical but with no feedback indicating pronunciation scores. The third group did not practice with the software. The subjects’ speech was recorded at the beginning of the study and again after three weeks, and scores based on log posterior probabilities were calculated. The speech recognizer outputs these scores, which have been shown to correlate well with human listeners’ non-nativeness judgements. The two practice groups showed a small but statistically significant improvement over the control group. The two practice groups were also compared with each other.