INTONATION AND SENTENCE PROCESSING

描述（研究者摘要）：自然语言呈现出广泛的 任何语言理解系统都必须解决的一系列理解问题 对峙 成功地理解语言需要一个数字， 不同类型的语言和非语言信息 处理和解释。 当前的心理语言学模型几乎 完全依赖于单一的语言信息来源， 句法或语义。 这项拟议的研究将审查两个额外的 可能被利用的信息来源：韵律 口语的特点和统计数据， 单词在一个句子中同时出现。 本研究的第一个目的是 理论：韵律和节奏影响的实证研究 人类句子理解的统计变量将提供 对语言理解模型的限制。 第二个目标是 实用：对于有听觉缺陷或语言障碍的人， 计划研究的结果将提供关键信息， 补救训练方法的设计，并改善语音到文字 翻译系统和文本到语音生成系统。

DESCRIPTION (Investigator's Abstract): Natural language presents a wide range of comprehension problems that any language understanding system must confront. The successful comprehension of language requires that a number of different types of linguistic and non-linguistic information be processed and interpreted. Current psycholinguistic models focus almost exclusively upon a single source of linguistic information, either syntactic or semantic. The proposed research will examine two additional sources of information that might potentially be exploited: the prosodic features of spoken language and the statistical regularities with which words co-occur in a single sentence. The first aim of the research is theoretical: empirical investigations of the effects of prosodic and statistical variables on human sentence comprehension will provide constraints on models of language understanding. The second aim is practical: for people with auditory deficits or speech problems, the results of the planned research will provide information critical both to the design of remedial training methods, and to improved speech-to-text translation systems and text-to-speech production systems.