ACOUSTIC AND PERCEPTUAL INVARIANCE IN SPEECH

语音中的声学和感知不变性

• 批准号: 6124944
• 负责人: Sheila E Blumstein
• 金额: $ 17.21万
• 依托单位: BROWN UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 1982
• 资助国家: 美国
• 起止时间: 1982-06-01 至 2003-11-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/6124944
• 关键词: behavioral /social science research tag; clinical research; human subject; language; language development; nonEnglish language; perception; phonology; psychoacoustics; sound frequency; sound perception; speech; speech recognition

The broad goal of this research is to explicate how the properties of sound map on the lexicon. The current proposal focuses on a characterization of the structure of phonetic categories, including both boundary constraints and internal structure, and the mapping of these categories on to lexical form. Boundary constraints of phonetic categories will be explored to determine whether particular acoustic properties called criterial attributes remain stable across a number of sources of variability including the influence of language, the presence or absence of phonemic contrast, the influence of the intrinsic nature of the sound segment (consonant or vowel), the influence of a secondary articulation on the emergence of an acoustic property, and the influence of regressive assimilation and vowel environment. The 'shape' of the internal structure of phonetic categories will be charted by exploring the extent to which languages have 'preferred' or prototypical region for the representation of a phonetic category; the relationship between a criterial attribute and the 'family' of values associated with it to define the best exemplar for that category; and the extent to which to which a range of 'best exemplars' of a criterial attribute may be influenced by speaking rate and the presence of other acoustic parameters. The mapping of phonetic categories on to lexical form will be investigated by assess whether and in which ways the acoustic variations which to and affect phonetic category structure ultimately impact lexical form. Experiments will investigate the role of acoustic variants that do or do not define phonetic categories, the influence of phonetic position and syllable stress, and the influence of prior lexical activation (top-down processing) on the emergence of subphonetic effects on lexical access. Research methods include acoustic analyses of speech and development of quantitative algorithms, computer-editing of natural speech in order to conduct both speech perception and lexical decision experiments. This research should provide a greater understanding of the speech processing system and how it interfaces with higher levels of language processing. Characterizing the speech processing of normal adults serves as the basis for the understanding of pathologies of speech and language, particularly in adults, but also in children.

本研究的主要目的是解释语音的属性是如何映射到词汇上的。目前的建议侧重于语音类别的结构，包括边界限制和内部结构的表征，以及这些类别到词汇形式的映射。语音范畴的边界约束将被探索，以确定特定的声学属性（称为标准属性）是否在许多可变性来源（包括语言的影响、音位对比的存在或不存在、声音片段的内在性质的影响）之间保持稳定（辅音或元音），次级发音对声学特性出现的影响，以及回归同化和元音环境的影响。语音范畴内部结构的“形状”将通过探索语言对语音范畴的表征具有“偏好”或原型区域的程度来绘制;标准属性和与之相关的值“家族”之间的关系，以定义该类别的最佳范例;以及标准属性的“最佳范例”的范围可能受说话速率和其它声学参数的存在影响的程度。语音范畴到词汇形式的映射将通过评估影响语音范畴结构的声学变化是否以及以何种方式最终影响词汇形式来研究。实验将调查的声学变量，做或不定义语音类别的作用，语音位置和音节重音的影响，以及影响先前的词汇激活（自上而下的处理）对词汇访问的亚语音效应的出现。研究方法包括语音声学分析和定量算法的发展，计算机编辑的自然语音，以进行语音感知和词汇决策实验。这项研究应该提供一个更好的理解语音处理系统，以及它如何接口与更高层次的语言处理。表征正常成人的言语处理是理解言语和语言病理的基础，特别是在成人中，但也在儿童中。