IMITATION--A TOOL FOR STUDYING SPEECH PERCEPTION

我们研究的总体目的是利用成年人模仿他们所听到的言语的表现出的倾向作为一种工具来探索言语感知的本质。在最初的研究中，我们探讨了言语感知对象(即声学或发音的)的性质，这是言语研究领域中争论很多的一个话题。作为基础研究的一个重要课题，这个问题本身就很有意义。它对应用也有影响，例如，对于成功的语音识别设备的设计。我们打算证实并利用这样一种趋势，即人类似乎不得不模仿他人的语言。我们还建议探索当模仿发生时被模仿的是什么。如果我们最初的发现表明知觉对象是声学的或听觉的，那么我们就会问，模仿是表面声学形式还是更抽象的规范化形式。如果最初的发现表明发音的感知对象，那么我们就会问表面的发音或语言手势是否被模仿。我们还建议探索影响模仿便利性和保真度的各种因素(特别是语言学习的种类)。我们项目的一个重要研究工具是电磁发音测量仪(Emma)，它将允许我们跟踪受试者的手势，并将它们与他们所跟踪的语音的手势联系起来。然而，其他涉及声学测量、感知判断或两者兼而有之的技术也将被用来评估模仿的保真度。我们希望我们的研究对关于感知对象的辩论做出有益的贡献，从而促进基础研究和应用研究的进步。

The overall aim of our research is to use the demonstrated tendency for adults to imitate the speech they hear as a tool to explore the nature of speech perception. In initial studies, we explore the nature of objects of speech perception (i.e., acoustic or articulatory), a topic that is much debated in the field of speech research. The question is of interest in its own right as an important topic for basic research. It also has implications for application, for example, for the design of successful speech recognition devices. We intend to confirm and then exploit the tendency that humans appear to have to imitate the speech of others. We also propose to explore what is imitated when imitations occur. If our initial findings suggest that perceptual objects are acoustic or auditory, then we will ask whether imitations are of surface acoustic forms or of more abstract normalized forms. If initial findings suggest articulatory perceptual objects, then we will ask whether surface articulations or linguistic gestures are imitated. We also propose to explore the kinds of factors (especially varieties of language learning) that influence both imitation facility and fidelity. An important research tool for our project is the electromagnetic articulometer (EMMA) that will allow us to track the gestures of our subjects and relate them to those of the speech that they shadow. However, other techniques involving acoustic measurement, perceptual judgments or both will also be used to assess imitative fidelity. We expect our research to contribute usefully to the debate about perceptual objects, and, therefore to advances in both basic and applied research.