The comprehension and production of non-literal language

该研究项目考察了非字面语言的理解和产出。在这种形式的语言中，表面信息本身与所持有的关于世界的规范知识和/或说话者的意图不一致。尽管这样的比喻普遍存在于自然语言中，但它们对语言理解或产生的计算模型是有问题的，而且往往被证明对大脑损伤的人来说是困难的。此外，对于理解过程中所涉及的过程几乎没有达成共识，更不知道语境因素(语言因素和副语言因素)在生产中用来邀请成功的意图交流。在我们到目前为止的工作中，我们认为，理解非字面语言并不需要专门的语言处理器，一组语言和社会线索在理解中发挥着重要作用，总的来说，约束满足模型似乎具有灵活性，可以解释在档案文献中发现的一些明显相互矛盾的数据。我们打算系统地考察诸如性别、年龄或社会地位等语用因素是否在隐喻、讽刺和谚语的理解和产生中发挥作用，以及这些因素如何与更常见的基于语篇处理的非语用方面(如语义重叠和对即将到来的语篇的明确标记/期望)相互作用。我们将采用一套相当不同的衡量标准：评级、阅读时间、受限的生产任务和ERPs(与特定输入相联系的脑电波)。这些数据与我实验室中已经获得的其他数据将：(1)允许对讽刺加工和产生的替代理论进行独立测试，(2)允许对讽刺的约束满足解释进行直接测试，以及(3)允许基于约束满足原则开发讽刺理解的计算模型。这些数据将对大脑损伤的人产生影响，这些人无法“理解”非字面语言的意义

The research program examines the comprehension and production of non-literal language. In this form of language the surface message alone is inconsistent with canonical knowledge held about the world and/or the intention of the speaker. Although such tropes pervade natural language they are problematic for computational models of language comprehension or production, and often prove difficult for people with brain injuries. Moreover, there is little consensus on the processes involved in comprehension and even less known about the contextual factors (linguistic and paralinguistic) used in production to invite a successful communication of intent. In our work to date we suggest that a specialized linguistic processor is not required to understand non-literal language, that a set of linguistic and social cues play an important role in comprehension and that, in general, constraint satisfaction models appear to have the flexibility to explain the somewhat apparently conflicting data found in the archival literature. We intend to systematically examine whether pragmatic factors such as gender, age, or social status play a role in the comprehension and production of metaphor, sarcasm and proverbial language, and how these factors interact with more commonly studied non-pragmatically based aspects of discourse processing (such as semantic overlap, and explicit markers /expectations of the upcoming text). We will employ a set of quite diverse measures: ratings, reading times, constrained production tasks and ERPs (brainwaves linked to specific input). These data, taken with others already obtained in my laboratory will: (1) permit for independent tests of alternative theories of sarcasm processing and production and, (2) permits a direct test of a constraint satisfaction explanation of sarcasm, and (3) permit for the development of a computational model of sarcasm comprehension, based on constraint satisfaction principles. The data will have implications for people with brain injuries unable to "get" the point of non-literal language