Is the notion of analyticity "hopelessly confused"?
The analytic/synthetic distinction has a long and esteemed philosophical history, having its roots, arguably, in Descartes' 'clear and distinct ideas,' then in Hume's distinction between 'relations of ideas' and 'matters of fact,' and explicitly, in Kant and Ayer, where the notion of analyticity as a relation between meanings of words was introduced. It is partly for this reason that Quine's attack on the distinction was so influential and important. I shall argue that, though the distinction is confused, the situation is not hopeless: there does still exist an analytic/synthetic distinction between certain types of proposition, though to hypothesise that this distinction covers all propositions is an error.
The analytic/synthetic distinction is a categorisation of all propositions into two classes, and has been traditionally been thought to be coextensional with the necessary/contingent and the a priori/a posteriori categories. Thus, analytic propositions are traditionally held to be necessary, that is, there is no possible world (or alternatively no possible way in which our world could be) in which they could not be true (or false), and a priori, that is, they can be known independently of experience. Contrariwise, synthetic propositions are traditionally held to be contingent, that is, there are possible worlds (or alternatively there are possible ways our world could be) in which their truth-value differs, and a posteriori, that is, they are known via experience. Though this seemingly biconditional link between analyticity, necessity and a prioricity is traditional, it has been questioned in modern philosophy: Kripke argued that there could be contingent a priori propositions and necessary a posteriori propositions; Kant argued that there are synthetic a posteriori propositions. The apparent coextensionality between analyticity, necessity and a prioricity is perhaps one of the reasons why the analytic/synthetic distinction has been so widely accepted without criticism: it certainly is natural (whether reasonable or not) to suppose coextensionality between them and if this is the case then, prima facie, analyticity seems like a neat explanation of necessarry a priori propositions. However, it would be a digression to discuss whether this link does exist in the way traditionally thought: it will merely suffice to state that, if analyticity is to stand up to scrutiny it must does so on its own, without recourse to necessity or a prioricity, firstly because the link itself is debateable and, more importantly, because it is a separate thesis.