Using multiple measures to document change in English vowels produced by Japanese, Korean, and Spanish speakers: the case for goodness and intelligibility.

Purpose: This study examined the effectiveness of using goodness ratings and intelligibility scores to document changes in vowel production following pronunciation training. The relationship between listener perceptions of goodness and intelligibility was also examined.

Method: Fifteen English language learner speakers (5 Japanese, 5 Korean, and 5 Spanish) participated in 16 sessions of vowel-focused pronunciation training. Pre- and posttraining judgments of 10 English vowels in /hVt/ context were conducted by 25 monolingual English speakers who served as listeners. Listeners judged vowel intelligibility using a 10-alternative forced-choice task and rated goodness using a 5-point Likert scale.

Results: Goodness ratings and intelligibility scores captured improvement in the accuracy of several vowels following training. However, some vowels that received better mean intelligibility scores received poorer mean goodness ratings following training. The relationship between goodness ratings and intelligibility scores revealed that vowels such as /æ/ and /ʌ/ were more dependent on goodness for intelligibility than vowels such as /i/ and /e/, which were highly intelligible even when they received poor goodness ratings.

Conclusion: English vowels differ with respect to the importance of goodness for accurate identification by listeners. As such, clinicians should examine both goodness and intelligibility when measuring change following pronunciation training.