Training segmental productions to improve second language intelligibility

Abstract

Training procedures that purport to improve the intelligibility of the speech of second language learners are often not validated by research. This study examines the effectiveness of English segmental intelligibility training for native Mandarin Chinese-speaking learners. Ten Mandarin Chinese speakers trained on 16 American English phonemic contrasts. Half of the participants trained on 16 vowel errors common in Mandarin-accented English and the other half on 16 common consonant errors. Each participant received 15 h of computer-based, minimal word-pairs training. Native American English listeners judged the intelligibility of pre- and posttraining recordings of isolated words read by the trainees in a two-alternative, forced-choice identification task. Listeners also judged the intelligibility of sentences produced by trainees and recorded pre- and posttraining. Results showed that training produced significant improvement in the intelligibility of isolated words included in the training and also in phonologically similar words not included in training. Vowel and consonant training participants made similar word-level improvement. The effect of word-level training on improvement in sentence intelligibility, while small, was also significant when results from participants in both vowel and consonant training groups were considered together. These results encourage continued exploration of structured computer-assisted pronunciation training to improve second language intelligibility.