|The purpose of this dissertation is to argue that semantic conventionalism of a, more or less, Dummettian variety is unjustly neglected in contemporary philosophy.
The strategy for arguing this is to make a conjecture about why people ignore it; there seem to be two plausible reasons: 1) there are (what people take to be) obviously preferable candidates on offer; 2) there are (what people take to be) knock-down arguments against semantic conventionalism.
In response to 1), I consider intentionalist Gricean semantics, and argue it is at least no better off than conventionalist theories. Of course, any number of theories could be used oppose semantic conventionalism. But the Gricean theory is seen as particularly strong, and showing that it is no better off makes my case for the viability of semantic conventionalism all the more compelling. For 2) I consider three possible reasons for thinking that conventionalism has been refuted.
Chapter Three concerns the objection that semantic conventionalism depends on the existence of ``luminous'' psychological states, of which there are none (according to Williamson's anti-luminosity argument). I agree with Williamson, and reject luminosity as part of a viable conventionalist theory.
Chapter Four supposes that semantic conventionalist theories depend on the (untenable) analytic/synthetic distinction to avoid collapse into holism. However, I also reject the analytic/synthetic distinction for a more favourable distinction.
In Chapter Five, the objection I consider is that semantic conventionalism involves an epistemically constrained notion of truth and so collapses into incoherence because of the knowability paradox. However, my response to this is that the semantic conventionalist should be happy with such an epistemic account of truth and that it does not lead to the knowability paradox. The paradox can, and is, resolved in this chapter.
So, (1) and (2) are false. The concluding chapter brings together all that we have learned throughout the dissertation about what a defensible version of conventionalism might look like.