Logic & Philosophy of Science Colloquium

Here's an argument that that "I" is a context sensitive expression. Consider simultaneous utterances, u and u', of "I am wearing a hat", one by Stephen and one by Jason. Intuitively, these utterances might disagree in truth-value, contingent upon who is or isn't wearing a hat. It is not unreasonable to conclude that u and u' express distinct propositions, and differ in their truth conditions. Since these semantic differences are not the result of ambiguity (lexical or structural) or vagueness or ellipsis, it is customary to infer that 'I am wearing a hat' is a context sensitive sentence, because it harbors the indexical 'I'. We call this sort of argument a Context Shifting Argument (CSA). All such arguments take this form: we are asked to consider two utterances of a single unambiguous non-vague, non-elliptic sentence. CSA proponents claim that intuitions about these utterances reveal that they: I1: say different things. I2: express different propositions. I3: have different truth conditions. I4: and so, may differ in truth-value. Obviously, these are variations with semantic import; the standard semantic account is to infer that the sentence in question is context sensitive, possibly because it harbors an indexical expression. Philosophers have exploited CSA, e.g., to resolve the standoff between epistemic falliblists and skeptics, to defend a brand of moral relativism, to resolve Sorities and Liar paradoxes, to defend a conservative view about psychological attitude attribution, to explain intuitions about quantifier domain under specification, to provide a workable semantics for attributive adjectives, and so on endlessly. In this paper I will review a number of alleged and powerful cases of CSA. I will then briefly discuss conclusions drawn from such cases. And lastly and most importantly, I will offer a two-part objection of CSA. I first show that were its conclusions true, we should have a range of intuitions that we do not. My diagnosis is that contexts of utterances are systematically under-described in CSA. Only by under-describing various contexts of utterance can an appearance of having singled out semantic context sensitivity even arise. But when fully specified (in the relevant respects) it is easy to see that CSA by itself does not provide semantically significant evidence.