Grosslanguage validation of computational models of eye-movement control of reading with extensions to oral reading and reading-span tasks

阅读眼动控制计算模型的粗语言验证，并扩展到口语阅读和阅读跨度任务

• 批准号: 21950621
• 负责人: Professor Reinhold Kliegl, Ph.D.
• 金额: --
• 依托单位: Lehrstuhl Trainings- und Bewegungswissenschaft
• 依托单位国家: 德国
• 项目类别: Research Grants
• 财政年份: 2006
• 资助国家: 德国
• 起止时间: 2005-12-31 至 2009-12-31
• 项目状态: 已结题
• 来源: https://gepris.dfg.de/gepris/projekt/21950621?language=en
• 关键词: Grosslanguage; validation; computational; models; eye

This project is part of an ECRP proposal implementing common and local research goals. The common research goal is to establish a crosslanguage corpus of eye movements and fixations using a common protocol consisting of easy reading material of about 1000 words from a minimum of 50 readers with norms available for word frequency and word predictability from sentence context. The collaborative project realizes such a data base for the English, French, and German language. These data will be used for the crosslanguage evaluation of three fully implemented computational models of eye-movement control during reading (E-Z Reader, Split- Fovea-LATER, SWIFT). In a second phase these data will allow for systematic model evaluations. The local research goal is to collect eye movements for this data base under complex reading conditions (i.e., during oral reading and during reading-span tasks) to increase the percentage of fixations in second-pass reading (i.e., fixations after regressions to previously fixated parts of sentences). Aside from yielding badly needed information about eye-movement statistics in such complex reading tasks, these data will allow us to extend the SWIFT model to account for effects related to the eye-voice span during oral reading and for effects related to working-memory load.

该项目是实施共同和地方研究目标的ECRP提案的一部分。共同的研究目标是使用一个共同的协议建立一个关于眼球运动和注视的跨语言语料库，该语料库包括来自至少50名读者的约1000个单词的易读材料，并提供词频和句子上下文中的单词可预测性的标准。该协作项目实现了英语、法语和德语的这样一个数据库。这些数据将用于对阅读过程中眼动控制的三个完全实现的计算模型(E-Z Reader，Split-Fovea-Late，SWIFT)进行跨语言评估。在第二阶段，这些数据将使系统的模型评估成为可能。当地的研究目标是收集这个数据库在复杂阅读条件下的眼动(即，在口语阅读和阅读跨度任务期间)，以增加第二遍阅读中的凝视百分比(即对句子先前凝视部分的回归凝视)。除了在如此复杂的阅读任务中产生急需的眼动统计信息外，这些数据还将使我们能够扩展SWIFT模型，以解释与口头阅读过程中眼睛-声音跨度相关的影响，以及与工作记忆负荷相关的影响。