Making words disappear or appear: A neurocognitive and behavioral investigation of effects of speech rate on spoken word recognition

理解人类如何理解语音是一个悬而未决的具有挑战性的问题，部分原因是不同的说话人、方言和语速等因素在语音信号中引入了大量的时间和频谱变异性。本研究的重点是时间语境对语段、音节和单词知觉的影响。研究的结果可以提供对涉及语速中断的疾病(例如构音障碍、口吃、帕金森病和失语症)的治疗的洞察，提供改进语音技术应用的方法(例如，增强的自动语音识别、更自然的发音的计算机生成的语音)，并导致与涉及理解口语的大脑机制相关的新发现。研究人员还将让学生参与这项研究，其中包括那些在该项目上合作的本科生。调查人员将测试对语音感知中时间现象的不同解释。他们提出了在词汇和语音层面控制现象的两种相互作用的认知机制，每种机制都由不同的神经计时机制驱动。假设词频效应主要来源于自上而下、言语特有的时间预期，而音速效应则源于自下而上、非言语特有的瞬间节奏预期。这一假说将通过心理声学研究、脑活动的非侵入性测量和理论建模进行评估，以确定语音时间特性的神经表征所揭示的加工特征。

Understanding how humans comprehend speech is an unsolved and challenging problem, in part because factors such as different speakers, dialects, and speaking rates introduce a great deal of temporal and spectral variability into the speech signal. The focus of this research is on the influence of temporal context on perception of segments, syllables, and words. Results of the research may offer insights into treatment of disorders that involve disruption of speech rate (e.g., dysarthria, stuttering, Parkinson's disease, and aphasia), inform approaches to improve speech technology applications (e.g., enhanced automatic speech recognition, more natural sounding computer-generated speech), and lead to new discoveries related to brain mechanisms involved in understanding spoken language. The investigators will also involve students in the research, including those from a primarily undergraduate institution collaborating on the project.The investigators will test different accounts of temporal phenomena in the perception of speech. They propose two interacting cognitive mechanisms controlling phenomena at lexical and phonetic levels, each driven by a different neural timing mechanism. The hypothesis is that effects of lexical rate primarily stem from top-down, speech-specific temporal expectancies, while phonetic rate effects originate in bottom-up, transient rhythmic expectations that are not specific to speech. This hypothesis will be assessed using psychoacoustic studies, non-invasive measures of brain activity, and theoretical modeling in order to identify the processing characteristics revealed by neural representations of temporal properties of speech.