Time Course of Spoken Word Recognition

描述(由申请人提供)：成年人可以识别50,000到100,000个单词，由不同的说话者在不同的声学环境中说出。这种能力是非凡的，因为语音是以一系列持续数百毫秒的瞬时声学事件的形式展开的，没有可靠的词边界线索。尽管口语处理在普通人群中普遍成功，但5%的一年级学生入学时患有某种无法用听力障碍解释的语音障碍。此外，一旦成功地获得了语言系统，它就很容易受到伤害或中风的侮辱(美国有100万成年人患有某种形式的失语症)。口语识别在语言习得和口语理解中起着核心作用，它允许存储与词汇表征相关的丰富的句法、语义和语用知识，并在理解过程中快速获取这些信息。更全面地了解口语单词识别的感知和计算能力，对于促进对正常和异常语言习得和处理的理解是至关重要的。由于语音信号随时间展开，并且单词的声学实现随其局部环境而变化，因此使用嵌入连续语音中的单词在精细的时间颗粒上评估口语单词识别是重要的。该项目已将“视觉世界”视线跟踪确立为检查口语识别的有力工具。我们开发的方法越来越多地被用来解决参与者从婴儿到老年人，以及正常和受损人群的口语处理问题。这项拟议的研究有两个目的。第一个目标是评估一个“数据解释”框架，在这个框架中，连续言语中的词语处理受到基于语境的期望的调节，这(A)影响听者如何解释输入，(B)提供一种快速知觉学习/适应的机制。我们操纵语速和基于话语的信息结构来检验预期如何影响异步线索的实时整合，以及线索权重是如何通过知觉学习来调整的。第二个目标集中于影响视觉世界实验设计、解释和分析的三个新问题：(1)最早的信号驱动的眼动(至少部分地)是由所显示的图片的语音信息介导的，还是主要由口语激活的知觉表征介导的；(2)语音信号中的线索与第一次刺激驱动的凝视之间的最小延迟是多少；(3)凝视是否受状态依赖的影响，如果是，在什么条件下，以及如何在基于事件的统计框架内模拟这些影响。

DESCRIPTION (provided by applicant): Adults can recognize between 50,000 and 100,000 words, spoken by different talkers in varying acoustic environments. This ability is remarkable because speech unfolds as a series of transient acoustic events extending over a few hundred ms, without reliable cues to word boundaries. Despite the ubiquitous success of spoken language processing among the general population, 5 percent of first graders enter school with some type of speech sound disorder that cannot be accounted for by hearing impairment. In addition, once the language system has been successfully acquired, it is susceptible to insult from injury or stroke (accounting for 1 million adults in the U.S. with some form of aphasia). Spoken word recognition plays a central role in language acquisition and spoken language comprehension, allowing for storage of the rich array of syntactic, semantic and pragmatic knowledge that is linked to lexical representations and rapid access to this information during comprehension. A more complete understanding of the perceptual and computational capacities underlying spoken word recognition is essential to advancing understanding of both normal and deviant language acquisition and processing. Because the speech signal unfolds over time and the acoustic realization of a word varies with its local environment, it is important to evaluate spoken word recognition at a fine temporal grain using words embedded in continuous speech. This project has established "visual world" eye tracking as a powerful tool for examining spoken word recognition. The methods that we have developed are increasingly being used to address questions about spoken language processing in participant populations across the lifespan from infants to older adults, and in normal and impaired populations. The proposed research has two aims. The first aim is to evaluate a "data explanation" framework in which processing words in continuous speech is modulated by expectations based on context which (a) affect how listeners interpret the input and (b) provide a mechanism for rapid perceptual learning/adaptation. We manipulate speech rate and discourse-based information structure to examine how expectations affect real-time integration of asynchronous cues and how cue-weights are adjusted through perceptual learning. The second aim focuses on three emerging questions that affect the design, interpretation and analysis of visual world experiments: (1) Are the earliest signal-driven eye movements to pictures (at least partially) mediated by phonological information from displayed pictures or are eye-movements primarily mediated by perceptual representations activated by the spoken word; (2) What is the minimal lag between cues in the speech signal and the first stimulus-driven fixations; and (3) Are fixations affected b state-dependencies, and if so, under what conditions, and how can these effects be modeled within an event-based statistical framework.