Mid-Level Language Processing

9616833这项研究项目将考察人们在阅读、听力和口语过程中做出的快速语法决定。要理解一个句子，读者和听者需要立即对进入的单词如何与其他单词和短语组合做出决定，以便对句子中的“谁在对谁做什么”做出解释。通过这种方式，人们能够以递增的方式保持对每个句子中描述的事件和动作的解释。这项研究将集中于对暂时性句法歧义的处理，这是当理解者必须在一个歧义词或短语的两种可能的语法选择中进行选择时发生的。检查这一快速决策过程有助于揭示语言处理系统是如何组织来咨询语言和非语言信息的。其中一项研究将考察读者和听众如何处理句法模糊的短语。这些实验将检验这样一种假设，即理解者对短语预期解释的初始猜测是通过参考先前的经验来自动确定的，这些经验涉及单个单词在不同句法上下文中出现的频率。实验将集中在这样的假设上，即识别一个单词不仅包括对该单词的含义的快速检索，而且还包括关于该单词在不同类型的短语中出现的频率的概率信息。这项研究将使用各种不同的实验技术，旨在揭示语言加工的时间过程，包括在阅读过程中监控眼睛注视和在听力过程中测量对单词的反应时间。第二个实验将检查说话者是否对他们自己讲话中的句法模糊短语敏感，以及这是否会影响他们造句子的方式。特别是，这些实验将考察当听者错误解释歧义短语的可能性增加时，说话者是否提供有用的短语(韵律)提示。这两条研究路线可能有助于开发通过语言与人类互动的计算系统。此外，进一步了解正常成年人是如何处理语言的，可能会改善对语言障碍患者的治疗和教育。***

9616833 TRUESWELL This research project will examine the rapid grammatical decisions that people make during reading, listening and speaking. The comprehension of a sentence requires readers and listeners to make immediate decisions about how incoming words combine with other words and phrases, in order to develop an interpretation of "who is doing what to whom" in a sentence. In this way, people are able to maintain, in an incremental fashion, an interpretation of the events and actions described in each sentence. This research will focus on the processing of temporary syntactic ambiguities, which occur when a comprehender must choose between two possible grammatical alternatives of an ambiguous word or phrase. Examining this rapid decision process can help uncover how the language processing system is organized to consult linguistic and non-linguistic information. One line of research will examine how readers and listeners process syntactically ambiguous phrases. The experiments will test the hypothesis that a comprehender's initial guess about the intended interpretation of a phrase is determined automatically by consulting prior experiences concerning how often individual words have appeared in different syntactic contexts. The experiments will focus on the hypothesis that recognizing a word includes the rapid retrieval not only of the meaning of the word but also of probabilistic information concerning how often the word has appeared in different types of phrases. This research will use a variety of different experimental techniques which are aimed at uncovering the time-course of language processing, including the monitoring of eye fixations during reading and the measurement of the response time to words during listening. A second line of experiments will examine whether speakers are sensitive to syntactically ambiguous phrases in their own speech, and whether this impacts the way in which they produce sentences. In particular, the experiments will examine whether speakers provide helpful phrasing (prosodic) cues when there is an increased chance that the listener may incorrectly interpret an ambiguous phrase. These two lines of research may help the development of computational systems that interact with humans via language. In addition, furthering the understanding of how normal adults process language may improve the treatment and education of those individuals with language impairments. ***