Non-literalness and non-bona-fîde in language: An approach to formal and computational treatments of humor

The paper is devoted to the study of humor as an important pragmatic phenomenon bearing on cognition, and, more specifically, as a cooperative mode of non-bona-fide communication. Several computational models of humor are presented in increasing order of complexity and shown to reveal important cognitive structures in jokes. On the basis of these limited implementations, the concept of a full-fledged computational model for the understanding and generation of humor is introduced and discussed in various aspects. The model draws upon the authors ' General Theory of Verbal Humor, with its six knowledge resources informing a joke, and on SMEARR, a sophisticated semantic-network-based computational lexical environment. The relevance of the approach to the interpretation, generation, and cognitive structure of humor is discussed in the broader context of the nature of the cooperative non-bona-fide modes of communication.