The distinction between literal meaning, the domain of semantics, and inferred or implicated meaning, the domain of pragmatics, is examined. Grice's (1975) theory of implicated or conveyed meaning as part of an overall account of cooperative communicative behavior is evaluated. A range of data is presented which challenges Grice's theory. Discourse conventions and context determination are highlighted in an analysis of pragmatic inference. The data suggest that pragmatic inferencing plays a much greater role in natural language comprehension than is generally realized. Once the domain of pragmatic theory is extended, Grice's principles of conversation are inadequate to account for comprehension. Other approaches to the problems of inferred meaning are examined. It is concluded that selected aspects of these alternatives must be combined in order to arrive at a satisfactory account of pragmatics. (RW)