Nonliteral language processing as a constraint satisfaction problem

该研究项目考察了人们是如何理解的，以及他们在何种条件下产生的语言，在这种语言中，一个人所表达的东西与他想表达的东西是不同的。这种语言出现在所谓的非字面语言中，如隐喻(例如……“我的车是个柠檬”)，讽刺（“天哪，你居然是我最好的朋友”），谚语式的演讲（例如“篱笆那边的草更绿”），意图传达（但没有说明）汽车不可靠、朋友不是朋友、嫉妒的例子。理论解释和计算模型都不足以解释我们如何处理这种无处不在的语言，尽管人们认识到，陈述的上下文有助于恢复预期的意思。我们有一种程序化的方法来研究话语语境中非字面语言的处理，它采用了一套融合的方法，这些方法既测量了在各种语境中对这种语言的反应，也测量了我们在阅读过程中对句子的理解。此外，我们将上下文分为与话语本身更相关的上下文（如文本中的语言“暗示”或约束，预先警告一个人即将到来的句子要非字面理解）和与社会文化约束因素更相关的上下文（如当一个句子由女人对男人说，或由牧师对公共汽车司机说时活跃的刻板印象）。我们观察到，社会文化约束在语言加工的早期阶段对理解起着重要作用，这一发现对我们如何加工非字面语言具有重要意义。我们现在通过研究一系列话语和社会文化约束来扩展这些发现，目的是区分语言加工的不同理论解释。除了理论意义外，该研究还具有实际意义，因为非字面语言在脑损伤后尤其受损，遵循一个发展过程，并用于说服性沟通。因此，基础研究不仅在理论上有直接意义，而且在临床和教育上也有直接意义。

The research program examines how people comprehend, and the conditions under which they produce, language in which what one states differs from what one intends to convey. This language occurs with so-called non-literal language such as metaphor (e.g.. "My car is a lemon"), sarcasm ("boy you turned out to be my best friend"), proverbial speech (e.g ""the grass is greener on the other side of the fence"),  instances in which  one is intending to convey (but does not state)  that the car is unreliable, the friend is not a friend and that one is envious. Theoretical explanations and computational models alike are inadequate in their explanations of how we process such pervasive language, although there is a recognition that the context in which the statement is made aids in the recovery of the intended meaning. We have a programmatic approach to studying the processing of nonliteral language in discourse context that employs a converging set of methodologies, both those that measure the reaction to such language in various contexts and those that measure our understanding of the sentence during the act of reading it. Moreover, we separate context into those more tied to the discourse itself (such as the linguistic "hints" or constraints in the text forewarning  a person that an upcoming sentences is to be understood nonliterally) and those more tied to social-cultural constraints factors (such as stereotypes active when a sentence is said by a woman vs a man, or by a priest vs a busdriver). We observe that social-cultural constraints  play a role in comprehension at the earliest moments of processing, a finding with important implications on how we process nonliteral language and are extending these findings now by examining a range of discourse and social-cultural constraints with the aim of discriminating between different theoretical accounts of language processing. In addition to the theoretical import, the research is of practical importance because nonliteral langauge is especially impaired with brain injuries, follows a developmental course and is used in persuasive communication.  Consequently, the basic research should have direct implications not only theoretically but clinically and educationally as well.