Page 133 - Encyclopedia of Philosophy of Language

ceeding in making conformity fully rational. The rea- son for failure in this respect is that common knowledge of conditional preferences (that is, pref- erence for conformity given that most others conform) does not yield a sufficient reason for believing that others will in fact conform, and hence not a sufficient reason for conformity on one's own part.
3. LewisonLanguageConventions
On Lewis's conception, a possible language is an abstract entity, roughly an infinite set of sentences together with meanings and grammatical moods (this is more precisely specified in terms of functions, utter- ance occasions, and possible-worlds semantics). The notion of a convention enters the picture when it comes to explaining what makes a possible language the actual language of a population. The idea is that a possible language L is the actual language of a popu- lation P in case there is a convention, among members of P, of truthfulness in L. In conforming to such a convention, a member of P assertively utters a sen- tence of L only if that sentence is true (with respect to the utterance occasion), given the interpretation of that sentence as a sentence of L.
In a later paper, (Lewis 1975), with a slightly differ- ent definition of the concept of a convention, the con- vention of truthfulness in L is replaced by a convention of truthfulness and trust in L. In either case it is intended that Grice's analysis of com- munication intentions shall result as a special case of conformity to a convention.
To this proposal it is, inter alia, objected that either it presupposes that members of P can already think, independently of their capacity to speak L, or else the general convention reduces to a convention of truthfulness and trust simpliciter, which, if a con- vention at all, belongs to the moral rather than to the linguistic order.
4. Is Language Conventional?
Lewis has claimed that the view that there are con- ventions of language is a platitude, something only a philosopher would dream of denying. It has, indeed, been denied by philosophers, but the more interesting kind of dissent has the form of denying that any fun- damental properties of languages, or linguistic prac- tices, can be acceptably explained by appeal to conventions, or rules.
4.1 Quine on Conventions and SemanticRules
As early as the mid-1930s, W. V. O. Quine attacked the view of the logical positivists that logical truth is conventional. The idea was that a logically true sentence, or rather a logically valid sentence schema, is true (valid) either directly in virtue of a convention,
which holds of axioms, or indirectly, in virtue of consequences of conventions, which holds of theorems. The conventions in question govern the use of, and thus determine the meaning of, logical expressions such as and and if-then, or their symbolic counterparts. The first part of Quine's main objection (Quine 1976) was this: in order to arrive at the logical validity of the theorems, from the statements of the conventions, one must make inferences, and in those inferences one must already make use of logical properties of expressions occurring in those state- ments,particularlyif-then. Sincelogicalpropertiesof expressions are determined by conventions it seems that a further convention is required, and so on. To this it may be replied that the problem arises only under the assumption that a convention must be for- mulated in advance of being adhered to, and that this assumption is false. Quine agrees that this reply is reasonable, but the second part of his objection is that appeal to unstated conventions runs the risk of reducing the notion of a convention to an empty label. This argument has been very influential. However, it has been objected against Quine that conventions governing the use of logical expressions should not have the form of if-then sentences but, for example, of deduction-rule schemata of other kinds. This reply is to some extent effective against the first part of Quine's argument, but not at all against the second.
The empty-label theme was further pursued. Quine (1980) has inveighed against Carnap and others that notions such as synonymy, analytic truth, and necess- ary truth cannot be explained by appeal to semantic rules, simply for the reason that the notion of a sem- antic rule is as much in need of explanation as the other notions. Without such an explanation no more is known than that a semantic rule is something stated on a page under the heading 'semantic rules'.
It might be said that what characterizes definitions, logical and mathematical truths, and other statements thought of as expressions of rules, or of consequences of rules, is that they are immune to revision; no obser- vation, come what may, can render them false. Quine has stressed, however, that when the need to revise a theory arises because of new observations, then any part of that theory can be dropped, including state- ments that were once adopted as definitions; no state- ment is ultimately immune to revision. Moreover, there is no difference in principle between, say, a revision of logic and a revision of quantum theory, no acceptable reason for saying that the one is a change of rules and the other a change of theory.
Quine has expressed appreciation of Lewis's analy- sis of the notion of a convention (in the foreword to Lewis 1969), but the problems of synonymy and related notions are not eliminated, since such notions are already made use of in Lewis's semantics of poss- ible languages.
Convention
Ill