Metaphor Understanding { a Cognitive Science Perspective

This paper surveys some of the important trends of the research on metaphor comprehension in the cognitive science area. We examine the semantic theories of the metaphor, some pros and (mostly) cons of the very challenged error-recovery model of metaphor comprehension. We present Lakoo's view that metaphors are conceptual rather than linguistic phenomena; we discuss how theories of other tropes are consistent with metaphor theories; and nally we talk about some attempts to develop a computational theory of the metaphor. 1 Motivation A legitimate question the reader of this paper might ask is: why metaphors? Aren't they the exclusive appanage of poets and writers? Is this a study about the psychology of people involved in comprehending literary works? The answer to these both questions is \not necessarily." Indeed, every day language is full of dead or novel metaphors. Metaphors are one of the most valuable sources of new words for a language; and the all English (or other language) speakers make use of metaphors on a regular base. Moreover, metaphors are not only an ubiquitous phenomenon in the natural language, but researchers claim that they actually innuence our way of thinking. There is a famous paper of Reddy Reddy 1979] that shows how our beliefs about language and communication are a consequence of the conduit metaphor which is often used by the English speakers when they talk about language. Lakoo Lakoo 1992, Lakoo et al. 1980] extends the class of \cultural metaphors" and supports this idea with lots of evidence. It is often the case that metaphors help us grasp abstract concepts by expressing them in terms of more concrete notions. A lot of the words denoting abstract nouns were initially metaphors. The main questions we are trying to answer in this paper are: how do people understand metaphors? Is there any special process that makes metaphor understanding diierent than literal word understanding? Is a new meaning created when a metaphor is understood? What are our chances to develop a computational theory of the metaphor and which constraints should it satisfy? What makes a good metaphor good? Equally interesting questions are: why do people use metaphors? What makes them choose a metaphor instead of one or more literal words? Or maybe there is no equivalent literal translation of the metaphor? Although episodically we might touch this second class of questions in our survey, they are not our main focus, partly …