SGER: MEG Studies of Speech Perception

这项研究使用脑磁图成像技术来研究大脑是如何表现和使用语言类别的。最终，本研究将探索从语音到语义的各个层次的语言分析。目前的项目从语音和语音分类开始——语音的独特特征——以及它们是如何在听觉皮层的语音感知中实现的。这项初步研究的主要目的是比较大脑对声音的反应，而不是言语的语言特性。例如，（英语中的）浊音停顿和不浊音停顿在声学上是通过发声起始时间（VOT）来区分的，VOT是一个连续的物理维度。然而，对语音的感知并不是沿着VOT维度连续的。在这个维度上，声音是被绝对地听到的，它们要么属于发声的范畴，要么属于无声的范畴。一个类别内相对较大的VOT差异难以察觉，而跨越类别边界的相对较小的差异则容易察觉。很明显，听觉系统和大脑并没有将VOT的物理连续维度线性地映射出来，也没有将音高和响度映射成对数线性。从动物实验的结果中提出了分类知觉可能的神经生理基础。这个项目询问人类的反应在多大程度上类似于动物模型，以及除了哺乳动物听觉系统常见的身体反应之外，是什么构成了语言类别感知。

MARANTZ This study employs MEG imaging techniques to investigate how the brain represents and uses linguistic categories. Ultimately this research will explore all levels of linguistic analysis from the phonetic to the semantic. The current project starts with phonetic and phonological categories - distinctive features of speech sounds - and how they are realized during speech perception at auditory cortex. The primary goal of this preliminary research is to compare the brain's reaction to the acoustic as opposed to the linguistic properties of speech. For example, voiced and voiceless stops (in English) are distinguished acoustically by voicing onset time (VOT), a continuous physical dimension. Perception of speech sounds, however, isn't continuous along the VOT dimension. Along this dimension, sounds are heard categorically, as belonging to either the voiced or voiceless categories. Relatively large differences in VOT within a category are difficult to perceive while relatively small differences that cross the boundary between categories are easy to perceive. It is clear that the auditory system and brain do not map the physically continuous dimension of VOT linearly, or log-linearly as pitch and loudness are mapped. Possible neurophysiological bases of categorical perception have been suggested from the results of experiments with animals. This project asks to what extent the human response resembles the animal models and what constitutes linguistic category perception over and above the physical response common to the mammalian auditory system.