Using Acoustic Measurement as an Estimate of Vowel Intelligibility

Intelligibility of second language (L2) speakers is typically measured using dictation, transcription and/or listening comprehension exercises performed by native (L1) speakers (Kang, et al., 2017). However, Koffi’s (2019) Acoustic Masking and Intelligibility (AMI) theory implies that intelligibility of vowels can be measured instrumentally by examining the first formant (F1) as it carries 80% of the acoustic energy found in vowels. The AMI theory also combines Just Noticeable Differences (JND) thresholds and Relative Functional Load (RFL) calculations to gauge severity of masking and intelligibility. To test this theory, I used read speech provided by 32 advanced L1 Arabic speakers of English. Eleven monophthong phonemic vowels of English were extracted from the speech samples and F1 values were manually calculated (n=7,392). Using the AMI approach, the vowel intelligibility is assessed internally and externally. Internal masking analyses focus on whether or not the participants differentiate clearly among the English vowels when they speak. External masking focuses on whether or not the vowels produced by the participants mask the vowels produced by GAE speakers. The comparison of F1 values show that [ɪ], [ʊ], and [æ] being the most problematic vowels for L1 Arabic speakers of English. Additionally, as the RFL of [ɪ] and [æ] are particularly high, it is recommended that the pronunciation of these two vowels be prioritized in instruction. The findings of this study have implications for L2 pronunciation teaching and learning for the large population of L1 Arabic learners of English.