P. M. S. Hacker on Rudolf Carnap and H. P. Grice

"Carnap, unsurprisingly, was incredulous. His account of analyticity, he remonstrated, was intended as an explication of the philosophical concept of analyticty as applied to ordinary language, which is indeed impredse, since expressions in ordinary language do not have sharply defined meanings. Quine had argued that he did not know whether ‘Everything green is extended’ is analytic or not, and he attributed his uncertainty to the unclarity of the term ‘analytic’ (TDE 32). Carnap objected that the unclarity was due not to the term ‘analytic’, but to the fact that it is unclear in ordinary language whether the term ‘green’ is applicable to a single spatio-temporal point, where a point is construed as lacking extension, since ordinary language does not talk of points thus construed. (Grice and Strawson strengthened Carnap’s point in noting that the same uncertainty attaches to the question of whether it is true that everything green is extended - and Quine could hardly complain that the term ‘true’ is irremediably unclear.) In a constructed language, one lays down meaning postulates in order to ensure clarity. Hence, if it is a meaning postulate that ‘G’ (‘green’) is inapplicable to spatio-temporal points, then ‘(x) (Ux z> ~ Gx)’ is analytic in L, and so is ‘(x) (Gx z> Ex)’ (where ‘U’ signifies ‘unextended’ and ‘E’ ‘extended’). ‘Analytic in L’, Carnap argued, signifies sentences whose truth depends on their meanings alone, and is thus independent of the contingency of facts, or ‘true in virtue of meanings’. Of course, it does not follow that such statements cannot be revoked; the same sentence can be analytic in one system and synthetic in another. An analytic truth is unrevisable only in the sense that it remains analytically true as long as the language rules are not changed. (He might have added that it is anything but obvious what it would be to revise fine’s belief that bachelors are unmarried, without changing the use (or meaning) c of the expression.) The attribution of truth to synthetic sentences may be changed in the light of experience, even though the logical structure of the language does not change. ‘The analytic/synthetic distinction can be drawn always and only with respect to a language system, i.e. a language organised according to explicitly formulated rules, not with respect to a historically given natural language’ (Carnap, ‘Quine on Analytidty’, in Dear Camap, Dear Van, pp. 427-32)."