Language-and talker-dependent variation in global features of native and non-native speech

We motivate and present a corpus of scripted and spontaneous speech in both the native and the non-native language of talkers from various language backgrounds. Using corpus recordings from 11 native English and 11 late Mandarin-English bilinguals we compared speech timing across native English, native Mandarin, and Mandarin-accented English. Findings showed similarities across native Mandarin and native English in speaking rate and in reduction of the number of acoustic relative to orthographic syllables. The two languages differed in silence-to-speech ratio and in the number of words between pauses, possibly reflecting phrase-level structural differences between English and Mandarin. Non-native English had a significantly slower speaking rate and lower rate of syllable reduction than both native English and native Mandarin. But, non-native English was similar to native English in terms of silence-to-speech ratio and was similar to native Mandarin in terms of words per pause. Finally, some talker-specificity in terms of (non)optimal speech timing appeared to transfer from native to non-native speech within the Mandarin-English bilinguals. These findings provide an empirical base for testing how language-dependent, structural features combine with general features of non-native speech production and with talker-dependent features in determining foreign-language speech production.