语音感知的心理基础：上下文基频信息对声调感知的影响

项目名称： 语音感知的心理基础：上下文基频信息对声调感知的影响

项目编号： No.11474300

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 数理科学和化学

项目作者： 彭刚

作者单位： 中国科学院深圳先进技术研究院

项目金额： 70万元

中文摘要： 语音现象千变万化，然而听者却能够非常顺利地获取说话人想要传达的语言信息。这里面包含了一个叫说话人归一化的过程。这个过程是如何实现的呢，即听者如何将具有纷繁变化的声学信号映射到特定的语音范畴？这是揭示言语感知的一个基本问题。一般听觉机制理论认为语音感知与非语音感知都是基于同样的听觉机理，只不过是语音信号进一步和语义信息进行了关联。根据此理论，只要上下文语音语境与非语音语境共享了某个特征（比如基频），则该特征对靶刺激（比如说靶音节的声调）的加工具有同样的影响。但已有文献中的结果具有较大的争议性。鉴于声调具有易操作性的特点，本项目将以声调的说话人归一化现象为研究对象，利用感知实验，通过操弄语境类型、记忆以及注意力等变量进一步系统地考察声调说话人归一化的机制，并据此来探讨语音感知的机制。该项目的研究成果将为揭示人类言语听觉感知机制提供重要信息，也将加深我们对一般知觉加工的理解。

中文关键词： 语言声学；听觉感知；心理声学；声调感知；上下文信息

英文摘要： Despite the vast variation in speech production, listeners can recognize the intended words without much difficulty. This process is known as 'talker normalization'. How is the talker normalization implemented? In other words, how does the mapping between varying speech signal and linguistic categories take place? This is a fundamental question in speech perception. The general auditory mechanism claims that the processing of both speech and nonspeech sounds shares the general auditory (or perceptual) mechanism. The only difference is that speech signal further connects to semantics. According to the above theory, the same feature (e.g., fundamental frequency) no matter what the carriers are contributes equally to the perception of the target stimulus (e.g., the tone of the target syllable). However, the relevant results in the literature are highly controversial. Since tone, whose acoustic correlate is the fundamental frequency, can be easily manipulated, this project focuses on tone normalization. A series of perception experiments with manipulation of context type, memory and attention will be conducted to systematically examine the psychological bases of context-dependent talker normalization in lexical tone perception. Findings of this project will shed light on the mechanisms of speech perception, leading to a richer understanding of perception in general.

英文关键词： Acoustic phonetics;Auditory perception;Psychoacoustics;Tone perception;Contextual information