Acoustic Feature-based Pronunciation Test Of Non-native Speakers On Chinese Stops

With Chinese culture spreading all over the world, more and more people takeinterest in learning Chinese. Meanwhile, there are some difficulties in learning Chinese.Especially for Chinese pronunciation, the foreign accent of non-native speakers occurs,most due to not mastering the phonological distribution pattern. Much attention is notgiven to pronunciation tests which is part of language testing. During the assessment ofspoken tests, the subjective rating scale method is adopted. The reliability of spokentests is affected greatly because of the individual differences and subjective factors ofraters. However, with acoustic phonetics and experimental phonetics developing, thescientific and convenient analysis tools are provided, which brings into opportunitiesfor pronunciation testing.In order to explore an objective and scientific assessment method of Chinese stops,this thesis paper proposes three research questions.1. How to apply acoustic features of consonants, i.e. VOT, duration of aspiration(AD), and the ratio of duration of aspiration (AD) to duration of vowel (VD) to the testof Chinese stops?2. Can the CAP, the hypothesized three-level rating scale be quantified? Ifpossible, what are the boundaries of each level for scoring Chinese stops?3. What typical errors do non-native speakers of Chinese usually make inpronouncing Chinese stops? Why do non-native speakers of Chinese make such errors?In this thesis, Pratt is adopted to extract the values of acoustic features.Comparative analysis between pronunciation of test participants and standardpronunciation is done. The analysis among the pronunciation correct level, acceptablelevel and problematic level is conducted. The significance of differences is conductedbetween aspiration stops and unaspiration ones by using SPSS.By using Praat, VOT values, AD values and the ratio of duration of VD to ADvalues could be attained. Combined with CAP judging criterion hypothesis of stops,they could be applied to Chinese stops testing. However, CAP, the hypothesizedthree-level rating scale couldn’t be quantified in this study. Still some progresses aremade. As for aspiration duration, the value ranges of three different levels of stop/b/,which are correct level, acceptable level, and problematic level, could be distinguishedfrom each other. As for the ratio of aspiration duration to vowel duration, the valueranges of three different levels of stop/b/and/k/don’t overlap and could be distinguished. Besides, the typical pronunciation mistakes made by non-nativespeakers are mispronouncing unaspiration stops as aspiration stops, shorter duration ofaspiration, shorter voice onset time, smaller differences between aspiration stops andunaspiration stops, and poor phonological distribution pattern. Sometimes, the durationof initial consonant of a Chinese syllable gets too long or the duration of final soundgets too long so that the balance is broken.