Intonational Correlates of Discourse Structures

IRI-9443461格罗兹，芭芭拉哈佛大学，235,052-24个月。(与交叉学科活动联合资助--总奖金240,752美元)语调与语篇结构的相关性这是2013财年设立的一个标准奖的第二部分资助，作为一项周转拨款，用于开展对格罗兹-西德纳结构语篇分析理论的实证验证，以及语调指标与全球和局部语篇结构的相关性的研究。研究对象是一个来自教学和叙事语体的多说话人语料库(以确保结果和实验的共性)；语料库被主观地标注结构特征，并使用信号分析方法独立和客观地标注语调特征，目的是建立这两种标记之间的关联。积极的、公认的相关性将允许使用那些指示算法中的语调变化的语篇特征，从而在质量更高的普通合成语音中产生相应的语调变化。这个项目的结果应该建立语篇结构和语篇的语音决定的符号特征之间的关系。

IRI-9443461 Grosz, Barbara Harvard University $235,052 - 24 mos. (Jointly funded with Cross Disciplinary Activities - Total Award $240,752) Intonational Correlates of Discourse Structures This is the second part funding of a standard award initiated in FY93 and processed as a swing grant to conduct research on the empirical verification of the Grosz-Sidner theory of structural discourse analysis, and on the correlation of intonational indicators with both global and local structures of discourse. The subject of study is a multiple speaker corpus (to assure generality of results and experiments that can be replicated) of discourse from instructional and narrative genres; the corpus is annotated subjectively for structural characteristics and, independently and objectively, for intonational features using signal analysis methods, with the purpose of establishing the correlation between the two taggings. A positive, well-established, correlation would allow the use of those discourse features indicating intonational variations in algorithms that produce corresponding intonational variations in synthesized speech of quality higher ordinary. The results of this project should establish the relationship between discourse structures and the phonetically determined signal features of discourse.