注意与学习对连续语音神经加工的影响

During speech communication, what the listeners hear is often a continuous sound stream and parsing the continuous sound stream into discrete linguistic units such as words, phrases, and sentences is a critical step in speech comprehension. Recently, it is shown that cortical oscillations on different time scales can entrain to the rhythms of linguistic units of different sizes during speech comprehension. The current project utilizes entrained neural activity to study how attention and language learning modulate neural encoding of hierarchical linguistic structures. Functional magnetic resonance imaging is also employed to elucidate the cortical networks underlying hierarchical speech processing. The first goal of the project is to study how attention modulates speech processing at different levels, e.g., encoding of acoustic features, lexical segmentation, and phrasal structure parsing. Attention will be modulated by presenting distracting sensory stimuli or by engaging the subjects in different tasks that require different levels of processing. The second goal of the project is to study how the brain encodes artificial words that are learned in three steps: statistical learning, explicit learning of the phonological form of words, and explicit learning of the word meaning. Taken together, the project will deep the understanding of how speech is processed in the brain and elucidate how the attention and acquired language knowledge are used to parse speech streams into hierarchical linguistic structures.

日常生活中，语音往往以连续语音流的形式呈现。在语音理解过程中，大脑需要把连续的声音流切割成词汇、短语、句子等离散的语言单元。近期研究发现，在语言理解过程中，大脑皮层中出现与层级语言单元同步的神经振荡，而本项目将利用这种同步神经振荡来研究注意和语言学习如何调节连续语音中的层级结构的神经加工，并通过功能磁共振定位进行层级语言加工的脑网络。在注意调节方面，本项目将研究连续语音加工中听觉编码、分词和语法分析这三个加工阶段如何受到注意调节。实验将利用干扰刺激调节被试的注意焦点并通过不同的任务调节语音加工深度。在语言学习方面，本项目将采用多阶段的学习方式，考察统计学习、语音学习和语义学习如何改变大脑对连续语音中词汇的加工。本项目的研究将增进对连续语音神经加工机制的了解，澄清大脑如何利用注意资源和所习得的语言知识将连续的语音流分割成为离散的语言单元。

日常生活中，语音往往以连续语音流的形式呈现。在语音理解过程中，大脑需要把连续的声音流切割成词汇、短语、句子等离散的语言单元。近期研究发现，在语言理解过程中，大脑皮层中出现与层级语言单元同步的神经振荡，而本项目利用这种同步神经振荡来研究注意和语言学习如何调节连续语音中的层级结构的神经加工。对于语音加工的注意调节，项目一方面发现组字成词的过程很强地受到自上而下的注意调节，在韵律线索缺失的情况下无法自动形成词汇的神经表征；另一方面发现语言理解过程中注意力发生动态变化，而且眼动相关的运动系统参与调节动态注意。对于学习对语音加工的影响，项目发现外显学习比内隐学习可以建立更稳定的词汇表征，而且大脑皮层低频神经振荡可以跟踪利用实验过程中习得的句法规则构建的人工短语。上述项目资助的研究成果发表于Nature Communications, eLife(2篇), Journal of Neuroscience(封面文章)等期刊。