Analysing the Influence of first language prosodic features on speech production and perception of Japanese language learners

本研究调查了日语母语者和非日语母语者的语言节奏，使用时长变量根据语言节奏对语言进行分类。我们收集了以汉语、越南语、韩语和法语为母语的人的日语阅读语音数据。我们已经积累了343个包含tokushu-haku的日语单词，它们是双元音、元音、长元音、/CjV/序列和元音。对说话者的日语话语进行时长变量分割和分析：(1)时长比，(2)标准差，(3)变异系数，(4)两两变异指数，(5)标准化语音开始不同步的平均值。对这些持续时间变量的分析表明，单个变量不足以有效区分母语和非母语日语使用者的语音节奏。要区分土著和非土著群体，至少需要两个或三个以上的变量。

This study investigated speech rhythm produced by native and non-native Japanese speakers, using durational variables which are used to categorize languages by their speech rhythm. We collected Japanese read speech data of native speakers of Chinese, Vietnamese, Korean and French. We have accumulated 343 Japanese words containing tokushu-haku, which are geminate consonants, moraic nasals, long vowels, /CjV/ sequences, and devoiced vowels. The speakers' Japanese utterances were segmented and analyzed by durational variables: (1) durational ratios, (2) standard deviation, (3) coefficient of variation, (4) pairwise variability index, and (5) average of normalized voice onset asynchrony.The analyses of these durational variables showed that individual variables were not effective enough to differentiate between native and non-native Japanese speakers' speech rhythm. At least two or often more than three variables were required to discriminate between the native and non-native groups.