Word meaning and sentence comprehension

词义和句子理解

• 批准号: 155704-2007
• 负责人: McRae, Kenneth
• 金额: $ 1.9万
• 依托单位: University of Western Ontario
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2009
• 资助国家: 加拿大
• 起止时间: 2009-01-01 至 2010-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=422600
• 关键词: Word; meaning; sentence; comprehension

Language is a major part of what makes us human. The broad objective of my research is to understand how people interpret what they hear and read, how this knowledge develops in children, and how it becomes degraded due to brain impairments. My research is roughly divided into two parts. In one part, I and my collaborators study people's memory for the meaning of words that refer to living things such as dog, and nonliving things such as toaster. To do this, we use "neural networks", which are models of mental processing that are written as computer programs. We have built a series of neural networks that produce the meaning of words like toaster given how the word is spelled or pronounced. We study how our models work and then use that knowledge to come up with predictions that we test on human participants. This testing typically involves looking at the time it takes for people to answer questions like, "Does a shirt have a pocket?" or "Does the word dog refer to something you can touch?" We also damage our models by altering the program so that part of the simulated mind no longer works correctly. We then test whether the damaged model mimics the performance of people who have suffered brain damage.

语言是使我们成为人类的主要部分。我的研究的广泛目标是了解人们如何解释他们听到和阅读的内容，这些知识如何在儿童中发展以及如何由于脑部障碍而降级。我的研究大致分为两部分。在一部分中，我和我的合作者正在研究人们的记忆，这些单词的含义是指诸如狗和烤面包机之类的非生命事物。为此，我们使用“神经网络”，它们是作为计算机程序编写的心理处理模型。我们已经建立了一系列神经网络，鉴于单词的拼写或发音方式，它们会产生像烤面包机这样的单词的含义。我们研究模型的工作原理，然后使用该知识来提出对人类参与者测试的预测。该测试通常涉及查看人们回答问题所需的时间，例如“衬衫有口袋吗？”或“狗一词指的是您可以触摸的东西？”我们还通过更改程序来损害模型，以使模拟的一部分不再正常工作。然后，我们测试受损的模型是否模仿遭受脑损伤的人的表现。