BEHAVIORAL AND COMPUTATIONAL STUDIES OF MORPHOLOGY

形态学的行为和计算研究

• 批准号: 6186086
• 负责人: MARK S SEIDENBERG
• 金额: $ 13.29万
• 依托单位: UNIVERSITY OF SOUTHERN CALIFORNIA
• 依托单位国家: 美国
• 财政年份: 1998
• 资助国家: 美国
• 起止时间: 1998-09-10 至 2001-05-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/6186086
• 关键词: behavioral /social science research tag; clinical research; communication behavior; computational neuroscience; computer simulation; human subject; language; memory; neural information processing; phonology; psychological models; reading; semantics; syntax

DESCRIPTION (Adapted from applicant's abstract): This proposal describes research on inflectional and derivational morphology and their role in reading and related tasks. Morphology is the aspect of language concerning the forms of words. The issues to be addressed concern how this information is represented in lexical memory; how it relates to other types of information including orthography, phonology, and semantics; and how it is used in processing. The research utilizes both behavioral studies of skilled language users and computer simulation modeling. The issues are framed in terms of connectionist principles concerning knowledge representation, acquisition, and processing that were developed in previous research on reading simple, monomorphemic words. The proposed research attempts to extend this theoretical framework to the processing of more complex words. The main emphasis of the research is on the quasiregular character of inflectional and derivational morphology: the fact that both systems are productive and can be described as "rule-governed" but also admit many exceptions that deviate from what is predicted from the rules. These phenomena have standardly been treated within dual-route models of lexical access in which there are separate mechanisms for the rule-governed cases and exceptions. Connectionist models provide an alternative approach in which all forms are governed by a single network. The experiments focus on phenomena that differentiate the dual- and single mechanism approaches in both inflectional and derivational domains. The theoretical account is also tested by implementing computational models with the goal of simulating behavioral data in close detail.

描述（改编自申请人的摘要）：该提案描述了 屈折和派生形态学及其作用的研究 阅读和相关任务。 形态学是语言的一个方面 词语的形式。 要解决的问题涉及这些信息如何 表示在词汇记忆中；它与其他类型的关系如何 信息，包括正字法、音系学和语义学；以及它是怎样的 用于加工。 该研究利用了行为研究 熟练的语言用户和计算机模拟建模。 问题是 以有关知识的联结主义原则为框架 先前开发的表示、获取和处理 阅读简单的单词素单词的研究。 拟议的研究 尝试将这一理论框架扩展到更多的处理 复杂的单词。 研究的重点是拟正则 屈折形态和派生形态的特征：事实上，两者 系统是富有成效的，可以被描述为“规则管理”，但也 承认许多偏离规则预测的例外情况。 这些现象已在双路由模型中得到标准处理 词汇访问，其中有单独的机制来管理规则 案例和例外。 联结主义模型提供了另一种方法 其中所有形式都由单一网络管理。 实验重点 关于区分双机制方法和单机制方法的现象 屈折域和派生域。 理论帐户也是 通过实施计算模型进行测试，目的是模拟 详细的行为数据。