Abstract

A theory of predicate reference is generally expected to provide an account of how a predicate and an object have to be related in order for the former to be applicable to the latter. I argue that such an account cannot be supplied, and put forward an approach to predicate reference that accommodates the consequences of this negative result. ;I focus first on the information-theoretic approach to the task. I argue that this approach would require providing a specification of which properties of each representational event are to be treated as fixing its reference by virtue of their nomological relations to distal properties. I contend that the task of providing this specification faces serious obstacles. ;After considering whether Saul Kripke's sceptical paradox succeeds in establishing that an account of reference is not to be had, I turn to a general difficulty with the very idea that we can identify reference facts. I argue that the way in which the notion of reference is implicated in the task of identifying a realm of facts renders every construal of reference facts fundamentally uninformative. An informative specification of semantic facts would have to be undertaken from a point of view outside language altogether, and the availability of this vantage point is a philosophical illusion. ;This outcome forces us to reconsider the kind of explanation that reference calls for. I put forward an alternative approach, which I call immanent realism. On the immanent realist picture, explaining reference need not involve identifying reference facts. Instead, it proposes to explain the notion by means of an account of how our practice of applying predicates according to ever revisable procedures gives rise to the idea that our predicate applications can be thought of as correct or incorrect--as accurate or inaccurate representations of the objects of predication. The outcome is a version of realism purged of all commitment to a transcendent vantage point from which the connections between the world and our representations of it can be described