Semantics and truth relative to a world

My concern in this paper shall be with the idea that truth is in some way relative to a possible world. There is no doubt something right about this sort of slogan, but nonetheless, most of what I have to say about it will be negative. In particular, I shall argue that relativity of truth to a world plays no substantial role in the empirical semantics of natural language as it is standardly done. This is so, I shall argue, even when we focus on truth-conditional semantics carried out in the model-theoretic tradition. Work in this tradition relies on some formal languages whose model theory is typically done using a model-theoretic notion of truth at a world; but I shall argue, this is not an empirically significant aspect of semantic theory. Empirical work is never done in a vacuum. Empirical work in semantics is done not only against a background of the metatheory of formal languages, but also of philosophical ideas about content. Some ways of thinking about content do, I shall grant, show us a notion of truth relative to a world. But, I shall argue, they do not do so in a way that directly impacts empirical work in semantics itself. They offer us important ways of understanding the basic notion of content, but do not indicate any data, or any other sorts of demands on empirical ∗Thanks to Kent Bach, Brit Brogaard, Dave Chalmers, Andy Egan, Cody Gilmore, Carrie Jenkins, Bernhard Nickel, Agust́ın Rayo, Adam Sennet, Paul Teller, and members of the contextualism reading group at the ANU for helpful discussions and comments on earlier drafts of this paper. The final preparation of this paper was done while I was a program visitor at the Philosophy Program in the Research School of Social Sciences of the Australian National University. Thanks to all the people there for gracious hospitality and a wonderful research environment.