PERCEPTUAL, LINGUISTIC & COMPUTATIONAL BASES OF DYSLEXIA

该提案描述了基于发展性阅读障碍的研究， 利用对正常和受损读者的行为研究， 建模，以确定导致文字处理缺陷的因素。 有相当大的不确定性有关的基础诵读困难; 现有的证据暗示了几个因素， 视觉感知和听觉感知障碍。该提案 介绍了行为研究，将检查组成部分的文字阅读 技能的目标是确定是否不同类型的 损伤会导致单词水平处理的缺陷， 常见于诵读困难症单词识别是一项复杂的技能绘画 几种类型的知识，过程和认知能力。 未能获得与年龄相适应的阅读技能可能源于 这些方面的任何缺陷。提供的理论框架 连接主义模型被用作生成预测的基础 关于这些不同来源的阅读障碍的损害。这一框架 已经产生了文字处理的明确计算模型。它还 提供了一种强有力的方法来检查受损行为的基础： 可以通过在工作环境中引入各种类型的异常来模拟 正常性能的模型。这使我们能够确定不同的来源 的阅读障碍，并产生预测， 实验研究中使用的阅读和相关任务。四- 将对3-4年级的阅读障碍者进行为期一年的纵向研究， 进行了大量的测量，以评估正字法， 语音，视觉和听觉感知和其他组成部分， 阅读。年龄、字数匹配的正常阅读者对照组 识别技能也将被跟踪4年。数据使用 为了生成指示每个个体的缺陷区域的对象简档， 以及他们行为特征的稳定性和预测有效性 加班费将予以评估。检查关键组件的其他实验 更深入的技能将在休赛期进行， 纵向研究这些实验将集中在（a）假定的缺陷 在与诵读困难相关的言语感知中，以及（B）单词- 水平变量（如频率、拼写声音规律性和 阅读障碍者与正常者词和非词命名比较 读者模拟的目的是了解 词汇系统的损伤导致了受损类型 在实验研究中观察到的行为。行为数据将是 与联结主义模型有关，这些模型模拟了正常的 非典型发展。模型在不同词项上的错误模式 类型将被比较的错误和反应时间的阅读障碍 儿童对这些词的类型作为一个测试的有效性的理论 模型

This proposal describes research on the bases of developmental dyslexia, using behavioral studies of normal and impaired readers and computational modeling to identify factors that give rise to word processing deficits. There is considerable uncertainty regarding the bases of dyslexia; existing evidence implicates several factors including phonological, visual-perceptual, and auditory perceptual impairments. The proposal describes behavioral studies that will examine components of word reading skill with the goal of identifying determining whether different types of impairments contribute to the deficits in word level processing that are commonly observed in dyslexia. Word recognition is a complex skill drawing on several types of knowledge, processes, and cognitive capacities. Failures to acquire age-appropriate reading skills could derive from impairments in any of these areas. The theoretical framework provided by connectionist models is used as a basis for generating predictions concerning these different sources of dyslexic impairment. This framework has yielded explicit computational models of word processing. It also provides a powerful way to examine the bases of impaired behavior: they can be simulated by introducing various types of anomalies in a working model of normal performance. This permits us to identify different sources of reading impairment and generate predictions concerning their effects on the reading and related tasks used in the experimental studies. A four- year longitudinal study of dyslexic readers in grades 3-4 will be conducted using a large battery of measures to assess orthographic, phonological, visual- and auditory-perceptual and other components of reading. Comparison groups ~f normal readers matched on age and word recognition skill will also be followed for 4 years. The data will be used to generate subject profiles indicating deficit areas for each individual, and the stability and predictive validity of their behavioral profiles overtime will be assessed. Additional experiments examining key component skills in more depth will be conducted during off-months for the longitudinal study. These experiments will focus on (a) putative deficits in speech perception associated with dyslexia and (b) the effects of word- level variables (such as frcquency, spelling-sound regularity and imageability) on word and non-word naming among dyslexic and normal readers. The goal of the simulations is understand how specific impairments in the lexical system give rise to the typed of impaired behavior observed in experimental studies. The behavioral data will be related to connectionist models that simulate detailed aspects of normal and atypical development. The models' patterns of errors on different word types will be compared to the errors and reaction times of dyslexic children on these word types as a test of the validity of the theoretical model.