The influence of signal-to-noise ratio and listener’s native language on vowel intelligibility.
Tessa Bent, Diane Kewley-Port · The Journal of the Acoustical Society of America · 2008
In a previous study, intelligibility rankings of American English (AE) talkers were different for native and non-native listeners [Bent, et al., J. Acoust. Soc. Am. 122, 3016 (2007)]. To equate overall performance, the native and non-native listeners were tested at two different signal-to-noise ratios (SNRs). Therefore, the differences in intelligibility rankings between the two listener groups may have been affected by the listener’s native language and differences in SNR. The current study explored this issue by testing across-talker differences in AE vowel intelligibility for AE listeners and Korean listeners learning AE as a second language. Listeners heard recordings of ten AE vowels in /bVd/ context produced by ten talkers under three SNRs: −8, −5, and −3 dB. The words were presented for identification in a 10-alternative forced choice task. Results showed that when AE and Korean listeners’ were tested under the same SNR, the across-talker intelligibility scores were highly correlated for the two listener groups. However, Korean listeners were less accurate at vowel identification compared to the AE listeners. Furthermore, Korean listeners’ intelligibility rankings of the talkers were more variable across SNRs whereas the AE listeners were more consistent. (Work supported by NIH DC02229 and T32-DC00012).