Abstract

Jason Stanley has proposed that we can account for the effects of extralinguistic context on truth-conditional content whilst remaining loyal to a compositional semantics for natural language. This is possible, he argues, because there are covert variables present in the logical forms of certain sentences whose values are fixed relative to contexts, but which do not register in the overt structure of those sentences. In the present article I assess the plausibility of positing such variables in logical form, focusing particularly on the examples Stanley provides in order to corroborate their presence. I argue that these examples are apt for an alternative treatment than that offered by Stanley—one which does not make recourse to covert variables, and is hence more credible. I end by spelling out some consequences of this argument for Stanley’s proposal.