THE SEMANTICS-PRAGMATICS DISTINCTION:
WHAT IT IS AND WHY IT MATTERS

Kent Bach, San Francisco State University (IV-B)

The semantics-pragmatics distinction (SPD) has played an important role in philosophy and linguistics since its introduction by Charles Morris in 1938. Philosophers have relied on it to explain certain paradoxes, to defend certain theses that would otherwise be counterintuitive, and to expose dubious ambiguity claims and other confusions of use with meaning. Linguists have treated pragmatics as a bin for disposing of phenomena that would otherwise be the business of semantics (as part of grammar) to explain. Which linguistic phenomena are semantic and which are pragmatic has been a matter of considerable controversy, aggravated by a lack of consensus on what is at issue. These phenomena include indexicality, anaphora, ambiguity and polysemy, vagueness, semantic underdetermination, implicitness, implicature (conventional and conversational), presupposition, non-truth-conditional content, performatives, and illocutionary force. Clarifying the SPD requires explaining the difference between linguistic and extralinguistic information, and untangling various aspects of the multifarious notion of context.

Different formulations of the SPD have relied on oppositions of (1) meaning vs. use (or coded vs. inferred content), (2) truth-conditional vs. non- truth-conditional meaning, and (3) context invariance vs. context sensitivity, but these contrasts, important as they are, either draw the line in the wrong place or leave something out. I propose a formulation whose aim is to take all the relevant phenomena into account and to provide a principled basis for deciding where they belong and for demarcating the division of theoretical labor.

Some would argue that there should be no division of labor, that the SPD embodies a misconception of what is involved in understanding utterances, namely, two distinct stages of processing (corresponding to linguistic and to extralinguistic information), such that the first feeds into the second. A variety of examples, involving such phenomena as semantic incompleteness, conventional implicature, and generalized conversational implicature, have been used to show that pragmatic factors impinge on the semantic. In reply I argue that although these considerations complicate the formulation of the SPD, e.g. by showing that utterances do not always express exactly one proposition, they do not undermine the SPD, for it does not concern the processing of information but merely the information processed.