Processing Neighbors in Speech Perception and Production

语音感知和生成中的邻居处理

• 批准号: 6905624
• 负责人: MICHAEL S VITEVITCH
• 金额: $ 20.9万
• 依托单位: UNIVERSITY OF KANSAS LAWRENCE
• 依托单位国家: 美国
• 财政年份: 2004
• 资助国家: 美国
• 起止时间: 2004-07-01 至 2009-05-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/6905624
• 关键词: behavior test; behavioral /social science research tag; clinical research; human subject; information retrieval; language; neural information processing; perception; psychoacoustics; semantics; speech recognition

DESCRIPTION: (provided by applicant): The research in this revised proposal builds upon work that was completed under R03 DC 04259 "Phonotactic Information in Speech Production," which examined how the frequency of occurrence of phonological segments in a word (i.e., phonotactic probability) influenced the speed and accuracy of producing spoken words. The long-term goals of the current project are to better understand the processes and representations used to produce and recognize spoken words. Current models of speech production and word recognition propose multiple levels of representation, such as semantic, lexical, and sub-lexical representations. The research in the current proposal will examine how the frequency of occurrence of a stimulus, and the number of representations that resemble that stimulus (i.e., neighbors) influence the speed and accuracy of processing at each level of representation. The experiments in this proposal are unique in that they will examine the influence of frequency of occurrence and of neighboring representations at several levels of representation, including the semantic, the lexical, and the sub-lexical levels. Furthermore, the research in this proposal will examine these influences in both parts of the spoken language system (i.e, during the recognition and production of spoken words). The results of the proposed research will provide insight into the processes and representations used during the recognition and production of spoken words. Investigating the influence of frequency of occurrence and number of neighbors at each level of representation in both the recognition and the production of spoken language will lead to a better understanding of how different outcomes (e.g., competition versus facilitation) may arise given the demands of the different processing systems. Finally, this research will enable us to build better models of the two systems, as well as better describe the interface between the two systems. That is, the work in this proposal is a first step in the needed direction of integrating models of sub-lexical, lexical, and semantic processing and of models of recognition and production.

描述：（由申请人提供）：本修订提案中的研究建立在R03 DC 04259“语音产生中的语音信息”下完成的工作基础上，该工作研究了单词中语音片段出现的频率（即语音发音概率）如何影响产生口语单词的速度和准确性。当前项目的长期目标是更好地理解用于产生和识别口语单词的过程和表示。当前的语音生成和单词识别模型提出了多个层次的表征，如语义、词汇和亚词汇表征。当前提案中的研究将检查刺激出现的频率以及与该刺激相似的表征（即邻居）的数量如何影响表征每一级处理的速度和准确性。本提案中的实验是独一无二的，因为它们将在几个表征层面（包括语义、词汇和亚词汇层面）检查出现频率和相邻表征的影响。此外，本提案中的研究将研究这些影响在口语系统的两个部分（即在口语单词的识别和产生过程中）。拟议的研究结果将提供深入了解在识别和产生口语单词过程中使用的过程和表征。在口语的识别和产生中，调查每一级表征的出现频率和邻居数量的影响，将有助于更好地理解在不同处理系统的要求下，如何产生不同的结果（例如，竞争与促进）。最后，本研究将使我们能够更好地建立两个系统的模型，以及更好地描述两个系统之间的接口。也就是说，本建议的工作是朝着整合亚词汇、词汇和语义处理模型以及识别和产生模型的方向迈出的第一步。