|290-3||Context Sensitivity in Semantics||MacFarlane||W 2-4||234 Moses|
Words like ‘I’, ‘here’, ‘now’, ‘this’, and ‘yesterday’ are generally recognized to be context-sensitive in a way that ‘cow’, ‘table’, and ‘fifteen’ are not. Their contribution to what is expressed by sentences containing them depends systematically on features of the context in which they are used. In the wake of David Kaplan’s seminal work on the logic and semantics of indexicals in the 1970s, it has become popular to represent this context sensitivity formally by relativizing sentence truth to a “context of use.”
Kaplan’s work was focused fairly narrowly on standard indexicals and demonstratives. Recently, however, philosophers and semanticists have been busily extending the bounds of semantic context sensitivity to other kinds of expressions, including gradable adjectives like ‘tall’ and ‘flat’, counterfactual and indicative conditionals, epistemic modals like ‘might’ and ‘possibly’, propositional attitude verbs, and terms of epistemic assessment like ‘know’ and ‘justify’. Accompanying the first-order discussion of these expressions (which is often motivated by philosophical as well as semantic concerns) has been considerable methodological discussion about just where to draw the line between semantic and pragmatic sources of context sensitivity. In addition to various moderate positions that draw the line in different places, two extreme positions have been defended: conservatives have argued that semantic context sensitivity is limited to the canonical indexicals, while radicals have argued that it infects all language to such an extent that formal semantics is impossible. In the first part of the seminar, we will try to sort out what is at stake in these debates.
In the second part of the seminar, I will argue for a generalization of Kaplan’s framework, in which truth is relativized not just to a context of use but also to what I call a “context of assessment.” I will argue that this generalization is needed in order to make good semantic and philosophical sense of epistemic modals, terms of epistemic assessment (like ‘know’), predicates of personal taste (like ‘fun’), tense (in an indeterministic framework), indicative conditionals, and possibly other bits of language. These expressions, I will argue, are context-sensitive, but not in the familiar way. Instead of being “use-sensitive,” they are “assessment-sensitive.” In addition to working out a formal framework for the description of assessment sensitivity, we will grapple head-on with the philosophical difficulties raised by assessment sensitivity and the kind of “relative truth” it requires.