Neural mechanisms for phonological alternation with high and low productivity - a case study on tone sandhi

在当今世界上使用的大约7000种语言中，保守估计大约有一半是声调的，因为音高的使用本身可能会改变一个单词的意思。这份名单包括在美国经济和国家安全中发挥关键作用的语言，如汉语和越南语，以及许多美洲原住民语言，如纳瓦霍语和切诺基语。更好地了解这些语言的语言特性是提高与这些语言使用者的交流效率和为以英语为母语的人开发更好的这些语言学习材料的有意义的第一步。这个项目调查了两种汉语--普通话和台语--的声调语言模式，使用最先进的神经语言学技术来揭示有声调的单词是如何在说话者的大脑中呈现的。它贡献了方法论创新，将允许通过神经测量来跟踪音调产生过程的不同阶段。此外，这个项目涉及一家美国机构与香港和台湾两个著名的神经语言学实验室之间的国际合作。建立的联系将为美国和国际学生提供教育机会。这个项目的重点是声调交替模式，即一个声调在不同的语境中采取不同的语音形式，也被称为“变调”。行为研究表明，变调模式以不同的方式延伸到新词中，即根据模式的语音性质具有不同的产出力。这表明，不同的机制可能参与了这类语音变化的加工。本研究使用事件相关电位(ERPs)来研究普通话和台语中一些变调模式在语音产生过程中的编码时程。假设非产出性变调模式主要由在词汇提取的较早时间窗(大约在产生开始后150-225ms)内参与神经活动的词汇机制来辅助，而产生性变调模式主要由在较晚的语音编码时间窗(400-600ms)内参与神经活动的计算机制来辅助。我们还假设，词汇机制的神经活动对词汇频率很敏感，而计算机制的神经活动对词汇频率不敏感。这个项目将是第一个研究与产出性和非产出性语音变化相关的假定计算和词汇机制背后的神经活动。对语音交替的神经编码的深入理解将为语音产生的理论模型和语音能力的形式模型提供信息。这个项目的发现还可以与词法分解的问题相联系，如规律性和语义透明度，以实现对语言生产力的神经基础的更全面的理解。该奖项反映了NSF的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

Of the 7,000 or so languages spoken in the world today, a conservative estimate of around half of them are tonal, in that the use of pitch itself may change the meaning of a word. This list includes languages that play key roles in the economy and national security of the US such as Chinese and Vietnamese as well as many Native American languages such as Navajo and Cherokee. A better understanding of the linguistic properties of these languages is a meaningful first step in increasing the communication efficiency with speakers of these languages and the development of better learning materials for these languages for native speakers of English. This project investigates the linguistic patterns of tone in two Chinese languages--Mandarin and Taiwanese--using state-of-the-art neurolinguistic techniques to shed light on how words with tones are represented in the speakers' brains. It contributes methodological innovations that will allow the different stages of the production processing of tone to be tracked by neurological measures. Moreover, this project involves international collaboration between a US institution and two prominent neurolinguistic laboratories in Hong Kong and Taiwan. The established connections will provide educational opportunities for students both in the US and internationally. This project focuses on tonal alternation patterns whereby a tone takes different phonological forms in different contexts, also known as "tone sandhi." Behavioral studies have shown that tone sandhi patterns extend to novel words in different ways, i.e., have different productivity, depending on the phonological nature of the pattern. This suggests that different mechanisms may be involved in the processing of this type of phonological alternation. This project uses event-related potentials (ERPs) to examine the time-course of the encoding of a number of tone sandhi patterns in Mandarin and Taiwanese in speech production. It is hypothesized that unproductive sandhi patterns are primarily subserved by a lexical mechanism that engages neural activities in an earlier time window of lexical retrieval (approximately 150-225 ms after the initiation of production), while productive sandhi patterns are primarily subserved by a computation mechanism that engages neural activities in a later time-window of phonological encoding (400-600 ms). It is also hypothesized that the neural activities of the lexical mechanism are sensitive to lexical frequency, while those of the computation mechanism are less so. This project will be the first to investigate the neural activities underlying the putative computation vs. lexical mechanisms associated with productive and unproductive phonological alternations. The increased understanding of the neural encoding of phonological alternation will inform both the theoretical models of speech production and formal models of phonological competence. Findings of this project can also be linked to issues of morphological decompositionality such as regularity and semantic transparency to achieve a fuller understanding of the neural bases of linguistic productivity.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.