Measuring context effects in spoken word recognition

测量口语单词识别中的上下文效果

• 批准号: 6835699
• 负责人: JOSE R BENKI
• 金额: $ 7.03万
• 依托单位: UNIVERSITY OF MICHIGAN AT ANN ARBOR
• 依托单位国家: 美国
• 财政年份: 2004
• 资助国家: 美国
• 起止时间: 2004-01-01 至 2006-06-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/6835699
• 关键词: behavioral /social science research tag; clinical research; cues; human subject; model; model design /development; noise; phonology; speech recognition

DESCRIPTION (provided by applicant): The long-term objective of this research is the development of quantitative models of how top-down linguistic knowledge and bottom-up sensory information are combined in spoken word recognition. The research proposed here would provide quantitative measures of the effects of contextual knowledge on the recognition of English monosyllables by modeling whole word recognition rate as a function of phoneme recognition rate. Such models have important applications in helping hearing-impaired individuals understand speech and in the development of machine-based recognition technologies. Preliminary results from a word recognition in noise study quantitatively evaluated the facilitative context effect of scarcity of similar sounding words (i.e., sparse phonetic neighborhoods). These results suggest that neighborhoods defined by word beginnings are more relevant than those defined by word endings. Two experiments are proposed to further these findings and provide the foundation for future investigation of other context effects and longer stimuli. Experiment One compares the contribution to recognition of the beginning and end of words varying in similarity to other words using a variable masking technique. Experiment Two is a complementary study that will test for effects depending on the temporal distribution of phonological information using an auditory prime. In addition to their use in evaluating specific hypotheses based on current theories, the quantitative results from these studies should be useful for evaluating future conceptions of spoken word recognition as well.

描述（由申请人提供）：本研究的长期目标是开发自上而下的语言知识和自下而上的感官信息如何在口语单词识别中结合的定量模型。本文提出的研究将提供定量措施的影响，上下文知识的英语单音节词识别建模作为音素识别率的函数的全词识别率。这些模型在帮助听力受损的人理解语音和开发基于机器的识别技术方面具有重要的应用。噪声中单词识别研究的初步结果定量评估了相似发音单词稀缺的促进性上下文效应（即，稀疏语音邻域）。这些结果表明，由单词开头定义的社区比由单词结尾定义的社区更相关。两个实验提出了进一步的研究结果，并提供了基础，为未来的调查，其他的背景效应和更长的刺激。实验一比较了使用变量掩蔽技术的贡献，以识别的开始和结束的话不同的相似性，以其他单词。实验二是一个补充研究，将测试的效果取决于时间分布的语音信息使用听觉启动。除了用于评估基于当前理论的特定假设外，这些研究的定量结果还可用于评估未来的口语单词识别概念。