Effect of consonant manner on L2 speech perception in multispeaker babble noise

摘要

Past research has shown that perceiving speech in non-laboratory environments is challenging for both native and non-native listeners, even if they are able to perform well in laboratory (quiet) environment. The present study reports the impact of manner of articulation on English consonant identification in multispeaker babble noise by Japanese learners of English. Twenty-five American English listeners and 17 Japanese listeners were presented with 23 American English consonants embedded in the context "You are about to hear a_a". Stimuli were presented with and without multipeakerbabble noise. Participants were asked to listen to the stimuli, and to choose the consonant they heard from 23 consonant choices. Three signal-to-noise ratios were adapted for the noisy environments: 10dB, 5dB, and 0dB. Analysis was carried out as Language (Japanese or English), Manner (stops, fricatives, affricates), and Condition (SNR = 10dB, 5dB, and 0dB) as independent variables and Accuracy rates as the dependent variable. Statistical analyses demonstrated significant interactions between Language and Condition, as well as Condition and Manner.