Conversational postulates revisited

CONVERSATIONAL POSTULATES REVISITED J. L. Morgan University ofIllinois Examining Gordon & Lakoff's influential article(1971) onconversational postulates, this paper shows that the term postulate is misleading; that G&L's concept of conversational implicature as a case of entailment is mistaken; and that their view of the interaction of conversational implicature and syntactic rules is based on an incorrect analysis. It is suggested that some of the problems in their paper stem from a pernicious ambiguity of the phrase 'can convey'. In this briefnote I want to examine closely the picture of conversational implicature presented in Gordon & Lakoff's paper 'Conversational postulates' (1971). It is not that textual criticism strikes me as a thing worth doing in itself. Rather, I think that their paper has been influential, and is likely to be an important work in the long run; it is therefore important to point out some areas where their paper is potentially—and no doubt unintentionally—misleading. If my interpretation of G&L's intentions is wrong, then perhaps this paper will serve as a point ofdeparture for a clarification of what they do mean. G&L's paper is inspired in obvious and acknowledged ways by Grace's pioneering work on conversational implicature (1968). They state (p. 63) that their purpose is twofold: first, to outline a way in which conversational principles can begin to be formalized and incorporated into the theory of generative semantics; and, second, to show that there are rules of grammar, rules governing the distribution of morphemes, that depend on such principles. In particular, they examine and attempt to formalize a number of conversational principles that they call 'conversational postulates.' Their intention is that conversational implicature be treated as a case ofentailment; i.e., the conversational implicatures of a sentence are to be derived as entailments of the logical structure of a sentence, together with a representation of context and a set of conversational postulates. The points to which I will address myself are these : the status of G&L's ' conversational postulates' in a theory of pragmatics; the role of entailment and speaker's intentions in their approach; and the interactions of conversational implicature with syntactic rules. 1. First, a minor point: the choice of the term 'conversational postulate' is unfortunate. My copy of Webster's new collegiate dictionary defines 'postulate' as follows: ' 1. A proposition which is taken for granted or put forth as axiomatic; an underlying hypothesis. 2. An essential prerequisite.' Dagobert Runes' Dictionary of philosophy states that a postulate is 'An indemonstrable practical or moral hypothesis , such as the reality of God, freedom, or immortality, belief in which is necessary for the performance of our moral duty.' One can see that the use of the word 'postulate' in the term 'meaning postulate' is in the spirit of definitions like these. A meaning postulate states some semantic property of a word, say, which is basic and arbitrary—which does not follow as a consequence of anything else we 277 278LANGUAGE, VOLUME 53, NUMBER 2 (1977) know about that word, or about the language of which it is a part. Thus a meaningpostulate treatment of the word 'kill' which stated that if it were true that 'X killed Y', then ? died' would also be true, would not be a consequence ofany other fact about 'kill' or about English, but a basic and axiomatic fact about 'kill.' But surely G&L do not intend their 'conversational postulates' in this same strict sense. It is intuitively clear that each of their postulates ought to be not axiomatic, but derivable as a consequence of a general theory of language use like that advanced by Grice; we should consider inadequate any general theory that did not have these 'postulates' as consequences. Indeed, G&L themselves observe (p. 66) that their postulates 1Oa-IOc are not axiomatic, but 'predicted' by their postulate 4L Then it seems that such postulates are merely the result of chains of inference from general conversational principles,1 rather than axioms or postulates in the usual sense—and, as such, are not really an addition to the repertoire of concepts of the theory of conversation, but shorthand for consequences of existing notions. 2. I have a more serious argument with what I take to be the spirit of G&L's approach: namely...