Ambiguity is an inherent characteristic of natural language use. Many words, for example, have multiple meanings, e.g., the word port can mean a wine, a harbor, the left side of a ship, etc. Sentences can also have multiple interpretations, e.g., who is doing the visiting in the sentence visiting enemies can be dangerous? Despite such ambiguities in natural language, people are able to understand one another reasonably well. How people decide among alternative interpretations has been studied extensively and is now fairly well understood. However, there is an additional type of ambiguity that remains unexplored and is not well understood: the ambiguity between literal and nonliteral interpretations of utterances. How do people choose among the several possible interpretations of any utterance? Traditional theories of nonliteral language use do not go beyond the notion that people will search for a nonliteral interpretation whenever a literal interpretation does not make sense in context. How this search is done remains a mystery. We plan to study how people arrive at nonliteral interpretations in various contexts. We will focus on three important types of nonliteral language: metaphor, irony and idioms. The central question is, how do people combine the literal meanings of utterances with the conversational context so that a speaker's intended meaning is understood? The first step will be to determine the communicative functions of these types of nonliteral language use. With respect to metaphors, for example, why do speakers choose to use the metaphor form, such as my job is a jail instead of the presumably simpler literal form, my job is like a jail? The uses of metaphor, irony and idioms will be assessed by determining what people understand and remember when various forms of communication are used. Information transmission will be assessed, as well as connotations and perceptions of speakers' intentions and attitudes. The second step will be to identify the contextual cues that people use to understand utterances as literal or nonliteral. We have already shown that people will arrive at nonliteral interpretations even when literal interpretations do make sense, suggesting that the standard literal-first theory is incomplete, at best. The overall goal of the research program is to develop a more adequate description of how people understand nonliteral language. This is a necessary component of the theory of language use and comprehension. Such a theory, in turn, is necessary for developing more effective educational methods and materials, and for developing workable machine language recognition and comprehension devices.
