This thesis is an examination of the linguistic approach to modality (also known as 'linguistic conventionalism') - i.e. the view that necessity is to be explained in terms of the linguistic rules that we have adopted. Drawing on an investigation into the history of this approach, I argue against the currently prevalent attitude that it can be dismissed as misguided. The aim, however, is not to argue that the linguistic approach is correct, but, more modestly, to put it back on the table as an interesting and viable research program.

The thesis is divided into three parts. In part A, I articulate a conception of the commitments of the approach based on the ideas that influenced it, how it emerged and developed in the work of the logical positivists, and, in particular, the role it was meant to play in "making a consistent empiricism possible".

Next, in part B, I defend the core ideas of the approach against various objections. Notably, I consider the objection that truth cannot be "created" by convention, the objection that necessities cannot be explained in terms of contingencies, and the objection that determining what the linguistic conventions are, unlike determining what the modal facts are, is a straightforwardly empirical matter.

In part C, finally, I turn to objections which purport to show that there are limits to what can be explained in terms of linguistic convention. Specifically, I consider whether we need to assume a non-conventional distinction between admissible and inadmissible linguistic rules, a non-conventional consequence relation, or a non-conventional starting-point in order to get the linguistic approach off the ground. An overarching question is whether we are forced to take some logic for granted in a way which would undermine the explanatory ambitions of the approach.

I argue that some of the prominent objections rely on misunderstandings, that some can be answered head-on, and that some point to genuine challenges and constraints which put pressure on the linguistic approach, but do not warrant a wholesale rejection of the view. Instead, they point to areas where further work is needed.