The worry is this. If stuff about future contingents is what motivates assessment relativity, then the motivation will apply equally to any symbolic medium with which we convey thoughts about the future. Gestures can be used to convey such thoughts. So can pictures. So they are presumably “assessment relative” too. But rather than going down that path, I’d rather conclude that there is no reason to suppose that there is something language-specific, in the syntax or in the semantics for English (or Swahili or what-have-you) whose job is to allow for assessment sensitivity. Instead, assessment relativity, if it exists, is a pragmatic phenomenon, having to do with our use of symbols in general, not with something as specific as natural language semantics.

One might complain — and Phil Kremer did complain — that just because it shows up elsewhere is not yet a reason to think that there isn’t some doohickey in language that does the job. Witness marking of time, utterer, etc. which can show up in our use of picture and gestures, but is a feature of linguistic expressions too. Fair enough. But there’s an Ockham’s Razor point here: if there is no reason for putting a “context of assessment” slot into natural language expressions, beyond what we convey with our use of language, and if such conveying can be explained by appeal to stuff we need for gestures, pictures and the like, then there’s no need to posit something extra in the linguistic expression. Do not posit semantic rules, or syntactic items, without necessity. Now, we have tense markers; we have agreement markers and pronouns; and so on. These require rules in the semantics of the language. Their manifest existence is the reason for positing doohickeys for time, speaker, etc. But what thing, what item of syntax, in natural language has the job of marking assessment relativity? (Don’t say “Future tense”, because MacFarlane wants assessment relativity to be far more general. And we can get assessment relativity, even with respect to future contingents, from expressions that don’t exhibit future tense.) At a minimum, MacFarlane provides no evidence of items in the language that his “context of assessment” works on.

One might say that we need a doohickey in the syntax or the semantics because assessment relativity has to do with what is asserted, not just with what is implicated. Goes the idea, pragmatics can’t get you asserted content; assessment relativity is about what’s asserted; so we need a “slot” for context of assessment. It’s then an empirical matter where the slot is; but that one is required is beyond question. This won’t work, however, precisely because pragmatics does have a role in determining what is asserted. (Pace my pal Jason, who’s just nuts on this issue.)

In sum, even if assertions are assessment relative, I don’t see why this has any implications for natural language semantics at all. Unlike ‘I’, ‘now’ and the like, assessment relativity has nothing especially to do with natural languages.

http://www.missouri.edu/~kvanvigj/certain_doubts/

The thought is more like this:

Look at the relevant proposition “PA is rectangular” as a relation between the object PA and the concept rectangularity. The question is, what is the concept rectangularity? If you’re looking at Kaplan’s LD system (to take a simple example), rectangularity will be the interpretation of the term “rectangular,” which is a function from world-time pairs to sets of objects (the extension of “rectangular” at that moment).

But perhaps concepts don’t have to be functions from world-time pairs to sets of objects; perhaps the concept itself can be a function from context-world-time triples to sets of objects. Then Jack and Jill will be expressing the same proposition when they say “PA is a rectangle”—it’s the pair of PA and rectangularity—but the truth-value of this proposition can vary with the context in which it’s evaluated—which, remember, need not be the context of Jack’s or Jill’s utterance. This can happen because I’ve built evaluation-dependence into the proposition itself, via the concept of rectangularity.

This may well be simply verbal reshuffling of where the dependence goes, and hence a fancy form of contextualism in disguise. (The way I’ve set it up, even ordinary propositions need to be supplemented by a world and time, and this won’t jibe with many people’s notions of proposition; I certainly don’t want to dictate a use of “proposition” except that I would like to see it used as a verb more.) And I’m pretty sure that this won’t help anyone see how relativism is possible; I still think I’m just saying “Context dependence, but not the kind you’re used to,” which is already in John’s papers. But maybe my stuff about propositions and concepts (I’m not very fond of either term in phi. language) will strike a chord with someone, dunno.

Anyway, I should probably make this my last post on this thread, unless it turns out that I’ve created another misleading impression.

The right description of that view is that the sentence “That is a rectangle” expresses different propositions in different contexts of use, not because any of the lexical items are context-sensitive, but because the structural relation of predication expresses different relations in different contexts. So, you’ve got a different contextual mechanism accounting for the context-sensitivity of the sentence than a contextualist theory of “rectangle”. But it’s contextualism all the same — the same sentence expresses different propositions in different contexts of use, but not for a reason we’re used to seeing.

You want to say that the propositions expressed by the different utterances of “That is a rectangle” are the same, but two people can disagree about the truth of that proposition, and both be right. This is the relativist position, but your explanation hasn’t helped me understand how it is possible. If Jack has one predication relation in mind, and Jill another, then Jack and Jill are not considering the same proposition — they’re interpreting the syntactic predication relation differently. If, on the other hand, they have the same proposition in mind, then they’re interpreting the predication relation the same. I’m at somewhat of a loss here about how your way of going is helping.

It does. Which world does “this” refer to, according to the relativist?

It sounds like the relativist arguing that semantic discourse takes place in the actual world, against which we evaluate the truth values of Anne and Sue’s statements. I am I right to infer that the relativist also posits an Anne-world and a Sue-world, within which which their statements “X is better than Y” could have different truth values than we assign them in the actual world?

Let’s say Anne is a behaviorist and Sue is an objectivist.

Anne says “Abbey Road is better than Master of Puppets.”
Sue says, “No way, man. Master of Puppets is better than Abbey Road.”

Anne thinks that X is a better album than Y iff more humans would take X to a desert island than Y. Sue thinks that X is better than Y iff X the more successful celebration of humans as they can and should be. It sounds like they can’t both be right because it sounds like Sue contradicted Anne. But upon closer examination, it may turn out that the disputants are actually talking past each other.

Jeff—I am thinking of a view on which what is said by the utterance-in-context-of-utterance is (in some sense) that Pennsylvania is a rectangle, and that whether Pennsylvania is a rectangle varies with context of evaluation, because rectangle is a concept whose extension varies with context of evaluation. So it’s context dependence, but not the kind we’re used to. (Here I’m just repeating MacFarlane in a degraded form.)

You’re right that the (4) sentences are weird; I don’t think the relativist can make a case on predicting ordinary-language ability of sentences like that. (Even (4) doesn’t sound natural to me.)The question is going to be, in part, whether there’s some other good reason for adopting that sort of semantics. Relative truth happens to provide a particularly elegant solution to some problems in speech-act theory in Belnap’s branching semantics (that’s the future-contingent stuff), but that merit may not travel far outside Pittsburgh.