词切分的韵律学线索

PROSODIC CUES FOR LEXICAL PARSING

本文是关于汉语普通话词切分的一个实验报告。用无意文重复句和合成句作语音材料,探讨音节时长分布、基频曲线、音节强度以及共振峰等韵律学线索在词切分中的作用。得到的主要结果是,在排除语义和音段层面信息的条件下,听者能够根据句子的韵律学特征进行句子结构的判断和词的切分;在对切分有效的韵律学线索中,最重要的是音节的时长分布模式,其次是基频曲线,音节的强度和共振峰变化对切分没有贡献。

This is an experiment report on lexical parsing of Chinese utterances, Us-ing as stimuli reiterant and synthesized sentences'of six-syllable Chinese shortutterances, the roles in lexical parsing were investigated of such prosodic cuesas distribution patterns of syllable durations, fundamental frequency and inten-sity curves of utterances, and forment transitions of syllables. It was found thatlisteners could identify with high accuracy sentence structures and parse wordsin short utterances without any segment and semantic information, and thatamong the prosodic cues possibly usefull for lexical parsing the most importantwere distribution patterns of syllable durations and fundamental frequencycurves of utterances.