Meaning” in Analysis, 10.5. The point of Goodman’s article (for our present purposes) is that it applies strict speaking quite strictly, yet within the framework of the traditional ‘realist’ assumptions about sameness, with the result that its thesis constitutes a reductio ad absurdum of the traditional demand for a fixed criterion of sameness (in this case sameness of meaning); that is to say, for a criterion which is quite independent of purpose and context. (This is one reason why I mention the article: its conclusions indirectly confirm mine. But I should acknow­ ledge that Goodman-or so I believe-deliberately constructed his thesis as a reductio ad absurdum, not however of the demand for a fixed criterion of sameness but of attempts to subject ordinary (unformalized) languages to logical analysis.)

The salient features of the article can be outlined as follows: Its object is to find a fixed criterion by which to judge whether “ two names or predicates in an ordinary language have the same meaning” . However, under the influence of the prestige of Strict Speaking, Goodman adopts the unstated assumption that a criterion of sameness (of meaning or of anything) is unworthy of its name unless it never says things are the same when they are ‘strictly’ (i.e. in some context) different, al­ though it may say things are different when in some context they are the same. Given this lop-sided assumption, given also Goodman’s determination to adopt a fixed criterion of some sort, he is eventually forced by his own properly strict speaking to the inevitable conclusion. Since any two things are different in some context, he finds that the only way he can ensure a criterion which never says things are the same when they are strictly different is to adopt one which invariably says they are different. Accordingly, he does adopt such a criterion. His


argument finally concludes that the only criterion which fulfils the postulated conditions is one which apparently proves that no two names or predicates in an unformalized language ever have the same meaning. Rudner then adds an extra twist by showing that, by the same criterion, even two word-tokens of the same word-type (such as ‘rose’ and ‘rose’) are different in meaning. (Note that one of two assumptions must be abandoned: (1) the assumption that we must have a fixed criterion (i.e. the Universal Context assumption); (2) the assumption that some words in an unformalized language sometimes have the same meaning. Goodman abandons the second assumption, together with the useful notion ‘sameness of meaning* as distinguished from ‘identity of meaning by definition’. In this book I abandon the first assumption, together-I admit-with a certain feeling of security. But this feeling is after all illusory.)

For our present purposes the most interesting feature of Goodman’s argument is that our acceptance of the criterion he finally adopts de­ pends upon prior acceptance of two arbitrary logical decisions. They are decisions which I think we would not accept (as I hope to show in a moment) so to speak in vacuo> but which we are quite likely to accept when the way for them has been carefully prepared by Goodman-when in fact he has altered the mental environment so as to make what would otherwise be obviously inadmissible seem admissible after all.