Third colloquium Semantics and Philosophy in Europe SPE3 BOOK OF ABSTRACTS

The existence of context-sensitive general terms raises some issues that have only recently begunto be addressed. One of these is the nature of context-sensitivity itself: is it a purely (“narrow”)psychological phenomenon, internal to the speaker? Is it in part social? Or is it perhapsecological, in the sense of pertaining to the relationship between a community of speakers andtheir physical environment? This paper defends the view that context-sensitivity is an ecologicalphenomenon, a view that I call “Ecology”. I argue that certain predicates in natural language arecontext-sensitive (or, in some cases, are not context-sensitive) because of facts about the natureof our physical environment, such as the structure of space-time. For example, physicists tell usthat simultaneity is relative to a frame of reference. It is because of this fact that ‘simultaneous’and related words (‘before’, ‘after’, ‘in five minutes’, etc.) are frame of reference-sensitive. Butthe structure of space-time is a metaphysically contingent matter: it would have been possible, inthe broadest sense of “possible”, for the world to be Euclidean, and if the world had beenEuclidean, ‘simultaneous’ and related words would not have been context-sensitive even thoughwe, the users of these words, would have been intrinsically (more or less) exactly as we actuallyare. Not all examples are based on the relativity of simultaneity, however. If one is sufficientlyimaginative, it is not difficult to conjure up examples of pairs of worlds w and w’ inhabited,respectively, by communities c and c’ which are intrinsically indiscernible and speak a languagevery much like English, such that a particular word is context-sensitive as used by c but its twinas used by c’ is not, owing to differences between the laws of w and w’. I will illustrate how suchexamples could be constructed using location-sensitive words for times: ‘summer’, which ishemisphere-sensitive, and ‘noon’, ‘3 p.m.’ and the like, which are time-zone sensitive. If Ecology is true, as I maintain, some interesting consequences follow. One obviousconsequence concerns the utility of various tests for context-sensitivity that have been proposedin the literature: if Ecology is true, then an expression may fail every one of the proposed testsand yet be context-sensitive. Another consequence is a puzzle for those who maintain that allcontext-sensitivity must be represented in logical form: since Ecology says, in effect, thatcontext-sensitivity is not in the head, it follows that either it is not the case that all context-sensitivity is represented in logical form, or logical form itself is not in the head. I regard this asan addition to the (already quite weighty) reasons to be externalist about logical form (cf. Ludlow“Externalism, Logical Form and Linguistic Intuitions”), but I will leave the defense of LF-externalism for another paper; the puzzle alone is interesting enough. I anticipate two sorts of objections. There are those (like Á. Pinillos) who opt for a relativisticsemantics for ‘simultaneous’ and the like, and there will be those who simply deny what I take tobe the data. I argue that the relativist solution is ad hoc: neither of the usual motivations forrelativism – disagreement data and the “operator argument” – are present in the cases I discuss.While speakers who are unaware of the physical facts will think they are disagreeing when theyreport simultaneity facts from different frames of reference, what matters is whether we, who areaware of the physical facts, think they are disagreeing, and at least I have no inclination toattribute disagreement to such speakers (and this judgment seems widely shared). As for the“operator argument”, syntactic considerations tell against the view that ‘in my frame ofreference’ and the like are sentential operators. The data-denier is more difficult to deal with. She will insist that (e.g.) in both the Euclidean andSR worlds, ‘simultaneous’ and the like are context-sensitive in the same way (have the sameKaplanian character), but that in the Euclidean world all speakers are in the same frame ofreference, so contexts within the Euclidean world never differ with respect to the coordinate towhich ‘simultaneous’ (etc.) are sensitive. I have two replies to this. First, it is not clear how toimplement this strategy with the other examples that I claim support Ecology (consider thehemisphere-sensitivity of ‘summer’). The second worry is more serious: since the data-deniermaintains that the Kaplanian characters of expressions supervene on certain intrinsic (“narrowpsychological”) properties of their users, she must countenance a lot more hidden context-sensitivity than the advocate of Ecology does. For it is a consequence of the data-denier’ssupervenience thesis that if it is possible for a community of speakers to be intrinsically just likeus and use a certain expression E with a Kaplanian character that is sensitive to a contextualparameter P, then E, as we actually use it, has a P-sensitive character. The existence of remotepossibilities threatens to render our expressions sensitive to features of context we would neverhave imagined (and perhaps even cannot imagine) them to be sensitive to. Insofar as the data-denier is motivated by a desire to minimize non-obvious context-sensitivity, she should switchallegiances.