Abstract

There have traditionally been two views as to what makes it the case that a singular term has the prepositional content that it does. According to Descriptivism, the content of a term is descriptive since it is given by a cluster of descriptive properties commonly associated with it. According to Referentialism, the content of a term is singular since it is determined by the object it picks out. It follows that empty terms can have descriptive, but not singular, content. If narrow content is what intrinsic duplicates have in common, then descriptive content is arguably narrow. Singular content, however, is wide since intrinsic duplicates who inhabit different environments express different singular contents by the same terms. On the face of it, the arguments against Descriptivism and Semantic Internalism - the view that content is narrow - seem convincing, but a worry persists, namely how to reconcile Semantic Externalism - the view that content is wide - with the kind of Privileged Access speakers enjoy with respect to the contents of their occurrent attitudes. By thorough examination of those arguments I find space for an intermediate position. What we learn is not that reference cannot go by properties, but rather which properties mediate reference. Kripke's Modal Argument proves that we need rigidified descriptive properties, Putnam's Twin Earth Argument shows that we better include causal properties, and Burge's Arthritis Argument highlights that we frequently invoke properties involving reference to other speakers. What is more, considerations about the behaviour of singular terms in intentional contexts strongly suggest that their propositional contents cannot be exhausted by their referents. By deploying so-called Two-Dimensionalism, as developed by Stalnaker, Kaplan, Evans, Davies, and others, I argue that singular content is knowable only after relevant empirical information about the actual world is in, and so is not subject to Privileged Access. Descriptive content, however, is a priori knowable since it is independent of which possible world is actual. But if that is so, then descriptive content constrained by rigidity, causality and other-dependence must also be a priori. All it takes is knowledge of how to describe various thought experiments. Although the latter kind of content is not object-dependent, as is singular content, it is wide in that it fails to be shared by duplicates who are embedded in distinct physical or social environments. So, we should expect compatibility between Semantic Externalism and Privileged Access only if Rigidified Causal Descriptivism is adopted.