The Impact of Segmental Accuracy on Intelligibility

Abstract

Intelligibility over nativeness has been increasingly argued as an appropriate goal for second language (L2) pronunciation teaching (Levis, 2005), yet relatively little is known about the phonological factors that make a nonnative speaker’s speech intelligible. Previous studies on the impact of nonnative English speakers’ segmental (i.e., consonants and vowels) pronunciation accuracy on listeners’ level of actual understanding (i.e., intelligibility) are relatively limited, compared to that of suprasegmental features such as stress and intonation. Therefore, the current study examined the relationship between segmental accuracy (in terms of target segments in minimal pairs) and listener-based intelligibility (i.e., rate of accurate word identification across listeners). Eight native-English listeners were assigned to complete a minimal-pairs forcedchoice task recorded by twenty Korean EFL learners. The results showed substantial and negative correlations between segmental accuracy and listener-based intelligibility. Vowel errors were linked to lower intelligibility overall, but the negative correlation between consonant errors and intelligibility was stronger. Some sounds were not substituted/produced erroneously by the speakers, yet still posed intelligibility problems to native listeners. Descriptive analyses of individual contrasts indicate target phonemes most often misperceived by native listeners. These findings will help make specific pedagogical recommendations for the teaching of English pronunciation.