Doctoral Dissertation Research: Category Structure of Pitch Accent in the Mid-Western Japanese Dialects

音高是音乐中用来区分音符的资源，也是不同语言声音系统中普遍存在的一个方面。语言使用音高有许多功能，从划分短语和形成问题，如在英语中，到区分单词。本研究调查了六种日语方言的音高重音系统，这些方言在识别和区分单词时严重依赖音高对比。这些方言以特别复杂的音高使用而闻名，称为“志贵”。Shiki被描述为单词开头的局部音高规范，以及贯穿单词的更全面的音高运动。由于目前大多数声音的音位表征依赖于局部表征声音的元素，所以志贵的动态和全局方面对典型的表征方案提出了挑战。这项研究将确定不同方言中志贵对比的性质，这些方言已知表现出不同的志贵清单。日本中部方言的另一个方面是，它们表现出许多不同的志贵和口音系统，形成了一个自然的类型样本，可以区分志贵和口音类型来标记不同的单词。这项研究首先收集了散布在日本中部内陆水道的六个不同品种的不同口音类型的录音。然后将这些录音提交声学分析，并建立音高控制时间过程的动态模型。这些模型将复杂的移动模式简化为一组参数，这些参数可以用来研究它们如何捕捉不同方言中的单词对比。它还将包括一个感知组件，其中这些模型参数用于创建合成语音，以便在稍后的现场工作阶段由这些方言的讲话者进行感知评估。然后，目标是确定说话者在语音产生中控制和在语音感知中注意的基频F0的那些特性。研究结果将告知语言系统如何使用音高和跨语言的音高系统类型学的问题。这项研究也将有助于我们理解说话者如何在不同的口音中保持独特的志记系统，因为在一个相对较小的地理区域内分布着如此多不同而复杂的系统。此外，这项研究有望为音高控制的本质以及如何在音高使用研究中表现出来开辟新的概念。这不仅适用于日语等语言的词汇对比研究，也适用于英语等语言的语调研究。也许最重要的是，这项研究为记录几种濒危的日语方言提供了有价值的目的。例如，位于日本内海的一个小岛上的井岛方言，是唯一已知的保留了与古日语相同类型系统的方言。然而，说真正的井木岛方言的人很少，而且年龄都在70岁以上。这项研究将在这种遗产永远消失之前记录下方言的语言。

Pitch is the resource used to distinguish notes in music, and it is also a ubiquitous aspect of the sound system of different languages. Languages use pitch for numerous functions, from marking off phrases and forming questions, as in English, to differentiating words. This study investigates pitch accent systems in six dialects of Japanese, which are known to rely heavily on pitch contrasts for the identification and differentiation of words. These dialects are notable for a particularly complex use of vocal pitch, called "shiki". Shiki have been described variously as local pitch specifications at the beginning of a word, and as a more global movement of pitch throughout a word. Since most current phonological representations of sounds rely on elements that represent sound locally, the dynamic and global aspects of shiki present something of a challenge to typical representational schemes. This research will determine the nature of shiki contrasts in different dialects that are known to exhibit different inventories of shiki. One other aspect of the dialects of central Japan is that they exhibit many different shiki and accent systems, forming a natural typological sample of the different ways in which shiki and accent types can be differentiated to mark different words. The study begins with the collection of recordings of different shiki-accent types in six different varieties scattered about the inland waterways of central Japan. These recordings are then submitted to acoustic analysis, with the construction of dynamic models of the time-course of pitch control. These models reduce the complex movement patterns to a set of parameters that can be examined for how they capture the word contrasts in the different dialects. It will also include a perceptual component in which these model parameters are used to create synthetic speech for perceptual evaluation by speakers of these dialects in a later field-work stage. The goal, then, is to determine those characteristics of the fundamental frequency F0 that speakers control in speech production and pay attention to in speech perception. The findings will inform the question of how linguistic systems use pitch and the typology of pitch systems cross-linguistically. The study will also contribute to our understanding of how speakers can maintain distinct shiki systems in different accents, since there are so many different and complex systems distributed over a relatively small geographic area. In addition, this research holds promise to open up new conceptualizations of the nature of pitch control, and how it should be represented in studies of pitch usage. This is relevant not only for studies of word contrasts in languages such as Japanese, but also intonational studies of languages such as English. Perhaps most importantly though, the study serves the valuable purpose of documenting several endangered dialects of Japanese. For example, the Ibukijima dialect, localized on a small island in the inland sea of Japan, is the only known dialect that maintains the same type of system as reconstructed for Old Japanese. However, the speakers of an authentic Ibukijima dialect are few in number and over seventy years of age. This study will document the speech of the dialect before this heritage is gone forever.