Abstract

Bernard Williams's essay ‘Wittgenstein and Idealism’ argues that that the conventionality of language entails the dependence of the truth of sentences and ultimately of corresponding states of affairs as truth‐makers on the existence of thinking subjects. Peter Winch and Colin Lyas try to avoid William's paradox by distinguishing between the existence conditions of a sentence and its assertion. The Winch‐Lyas solution is criticized and a stronger Winch‐Lays resistant version of Williams's paradox is proposed. A more satisfactory countercriticism is given, involving an ineliminable modal fallacy occurring in the paradox inference, that arises because of the argument's invalid combination of categorical and counterfactual assumptions and conclusions.