Abstract

1. For simplicity, let the domain of our first-level quantifiers, ‘∀ x’ and so on, be words, and in particular just those words which are adjectives. And let the adjective ‘heterological’ be abbreviated just to As is well known, one cannot legitimately stipulate that Why not? Well, the obvious answer is that if is supposed to be an adjective, then this alleged stipulation would imply the contradiction But contradictions cannot be true, and it is no use stipulating that they shall be. Do we need to say anything more? Russell would apparently wish to add this: the stipulation is illegitimate because it attempts to define the word by a quantification over all words, and the word is itself taken to be one of the words quantified over. That is how this supposed definition would infringe his vicious-circle principle . 1 It is of course true that if our quantification is taken to range only over words of a certain kind, say first-order adjectives, and if the word is taken to be a word of a different kind, say a second-order adjective, then the contradiction is avoided. For the supposed stipulation then has no implications for whether is or is not true of any second-order words. Russell imagines the stipulation to be strengthened by adding that is to be true only of first-order words, and hence that it is not true of itself. But no contradiction now results from this. However, one must ask: did we need this explanation in terms of the VCP? It may seem helpful, because there is an initial temptation to say that the original stipulation could not be illegitimate. This is because it is thought of as introducing a new word, and explaining what it means, and one is apt to think that we …