Thursday, December 10, 2009

Language and Time (and Space)

Sorry about the delay. I've been pretty good about maintaining the every-Monday-and-Wednesday schedule since early October, but what with the end of the semester, grading, packing for an extended trip back to Michigan before the Eastern APA, etc., etc., etc., these things happen. Better late than never, though, so here's a post....

In Quentin Smith's book Language and Time, he argues for a moderate version of the A-Theory. Some background:

The B-Theory of Time holds that "now" is an indexical like "here," serving only to locate the speaker in time. The A-Theory is all about "the reality of temporal becoming," which basically means that there's a fact of the matter about which moment uniquely counts as "the present." The best argument for the B-Theory is that it's hard to square the admittedly compelling, common-sensical intuitions about time encoded in the A-Theory with the empirical deliverances of Einstein's Special Theory of Relativity (STR). The best argument for the A-Theory (beyond whatever immediate intuitive pull it might have) is that ordinary language sentences are tensed--that is to say, they seem to more or less constantly refer to properties of pastness, presentness and futurity--and that if we don't believe in the real objective existence of those properties, we have to either say that all tensed sentences of English and other languages are false (including, for example, the sentences physicists use to express the STR) or we have to come up with a way to paraphrase the sentences (or at least the ones we take to be true) in such a way that they no longer refer to pastness, presentness and futurity.

As it turns out, the project of coming up with such tenseless paraphrases is pretty hard. Smith is an A-Theorist, and he has a lot of fun pointing out the holes in various extant proposals to do this sort of thing. For example, the tenseless date theory says that we can translate "the meeting starts now," tokened at three o'clock as "the start of the meeting (is) at three o'clock." (The parentheses indicate that the (is) is tenseless.) So far, so good. But what, he asks, about "it is now three o'clock"? Apply the tenseless date theory, and we get "it (is) three o'clock at three o'clock." No good. For one thing, "it (is) three o'clock at three o'clock" is a tautology, and "it is now three o'clock" really really shouldn't come out that way. You need to check your cell phone* to see if "it is now three o'clock" is true, but you can be justifiably confident about "it (is) three o'clock at three o'clock" even if you just woke up in a doorless, windowless room with no memory of how you got there and your phone is out of power.

The tenseless token-reflexive theory does better with this example. It says that we should interpret "the meeting starts now" as "the start of the meeting (is) simultaneous with this utterance," which works just as well for sentences like "it is now three o'clock."

Smith, however, thinks that his counter-examples to this are just as good. For example, consider "it is true that it was true that the era devoid of linguistic utterances is present." Isn't this true? I mean, surely, there is language now, and there wasn't in the past. It isn't true now that there isn't any language, but it used to be. Right? issue is that part of what's traditionally at stake between A-Theorists and B-Theorists is that the former take propositions to change truth-values over time (this is the point of developing tense logics) whereas the latter take them to have eternal unchanging truth-values, so Smith is at the very least dancing on the edge of begging the question here.**

Another issue comes to the fore when we start think about what the "it" is that was supposed to be once true and now false. It's certainly not a linguistic utterance like a sentence, since if so it--i.e. a sentence declaring the absence of sentences from the world--would never have been true, since such a sentence could never have been true, for obvious reasons. In fact, it looks like "it" has to be a proposition, and in fact a proposition that existed despite the non-existence of sentences. That is to say, Smith's move here strongly suggests a fairly extreme version of the proposition theory of truth-bearers, whereby we believe not only in the propositions expressed by sentences, but a sort of Platonic realm of un-instantiated propositions as well.

Now, there may be a lot to be said in favor of proposition theory--see the discussion with Emil on this blog a while back--but there's surely a lot to be said against it as well (e.g. if propositions are abstract objects we can't causally interact with, how would we ever find out that they existed?), and if the price of the best philosophical account of how to make sense of the picture of reality suggested by taking our best current science seriously is that we have to abandon belief in propositions, that strikes me as a price worth paying. But someone who starts from a posture more sympathetic to propositions, presentism or both than mine might not be convinced by this. As such, I'll take a stab at pointing out the problem that originally sparked my own slow move away from the A-Theory when I was an MA student at Western Michigan (where I took several classes from Smith, and spent a lot of time thinking about these issues):

On p. 129 of "Language and Time," Smith says that it would be an "interesting task" to provide an account of indexical terms like "I" and "here" in light of his account of "now," but that "this task falls beyond the purview of the present treatise."

Even when I was an A-Theorist***, this passage always struck me as extremely unsatisfying. The more I thought about it, the more dissatisfied I was. Note, for example, that Smith's criticisms of the date theory all go through just as well for what we can call the "place theory" of spatial indexicals. Imagine the following fairly banal conversation:

P1: "I just got here?"
P2: "Where's 'here'?"
P1: "Oh, sorry, 'here' is Miami."

We might be tempted to translate "I just got here," tokened in Miami, to "I just got to Miami." That's pretty clearly not going to work for reasons that exactly parallel the problem with temporal indexicals. "Miami is here" is only true when tokened in Miami but "Miami is in Miami" is true even if tokened in Hong Kong. The best way to go here looks very much like a token-reflexive view where "Miami is here" means "Miami is co-located with the person saying this" or something roughly along those lines. I feel absolutely no inclination to beieve that claims change their truth-value from location to location. Moreover, I feel even less inclination to think that there's some sort of property of "hereness" independently objectively existing in places where no speaker is located. As such, consideration of the sentence "it is true on Mars that 'Mars is here'" gives me no reason whatsoever to believe in propositions that have a special truth-value on Mars that they don't have on Earth, or to give up on the token-reflexive view for the sake of such strange Martian propositions.

Why, exactly, should "nowness" be any different in this respect? If it feels different, is there anything motivating that feeling other than residual folk belief in the claim that there is an ontologically privileged present moment, which is, after all, precisely the bone of contention here? If so, I'm having troubling seeing what.

*Ten years ago, you would have checked whatever those things were that people used to wear on their wrists. I forget.
**I'm putting things in terms of propositions here because that's the traditional way to frame it, not because I necessarily believe in propositions.
***We all get to have a few youthful indiscretions. Hell, just between you and me, there was even a time in my life when I was a libertarian about free will. Don't judge me. I was just a kid, and everyone else was doing it.


punninglinguist said...

The best argument for the A-Theory [...] is that ordinary language sentences are tensed.

I just want to poke my linguist head in and say that if this is the best argument for A-theory, then it's one of those theories that could not have arisen in many other cultures. Many languages (Mandarin, Maya, many others) have no tense system at all. They use aspectual markers, specifying whether the verb refers to a complete/incomplete/stative/repeated/etc action, with past/present/future-ness generally left to be inferred from context.

I also want to ask out of curiosity: if you don't believe in propositions, then to what are truth values assigned?

Take care,

Emil said...

"I also want to ask out of curiosity: if you don't believe in propositions, then to what are truth values assigned?"

Unidentified truth bearers of course!

Joking of course. The above answer is explanatory useless and maybe even false (maybe some theory of truth bearers is true, thus they are identified).

Ben said...


"I also want to ask out of curiosity: if you don't believe in propositions, then to what are truth values assigned?"


punninglinguist said...

I actually like that answer a lot, Ben. But can a question or an exclamation have a truth value? If not, it seems that only sentences that express propositions can have truth values. Where is the line drawn between sentence and proposition?

genna said...

"I also want to ask out of curiosity: if you don't believe in propositions, then to what are truth values assigned?"

The old answer, I believe, is assertions.

Ben said...

Hey Tristan,

Sorry about the delay. Been in transit back to MI and getting set up here for the last couple days. But re: sentences vs. propositions....

As far as I know, no one takes interrogatives or imperatives to be truth-evaluable--there are "imperative logics," but this seems to be logic-by-analogy, and I think the standard story there is that validity is a matter of something like "wish-preservation" rather than truth-preservation--so sentence-based theories of truth bearers are really *declarative*-sentence-based theories of truth-bearers. (E.g. one view is that the bearers of truth are "sentences in contexts" or "precissified (sp?) sentence," and I think that these can be read without any violence to the views of those who advocate these approaches as "declarative sentences in contexts" or "precissified declarative sentences.") So far, so good, since we're making a purely syntactic distinction. The problem at that point is that we don't want to apply truth talk to grammatically well-formed but meaningless sentences, e.g. "Pumpkin pie penetrates Modus Ponens in a purple fashion." Again, as far as I know, no one disagrees with the claim that meaningfulness is a pre-condition for truth-evaluation, so it seems like to really draw out the underlying idea, we should say that the word "sentences" in various sentence theories of truth-bearers should be read "meaningful declarative sentences."

At this point, though, one might worry that I'm cheating. After all, you might not unreasonably ask, isn't the difference between meaningful and meaningless declarative sentences just that the former express propositions and that the latter do not?

Maybe. There are tricky issues here about exactly how to analyze "meaningfulness," but in defense of the distinction, I'd point out three things:

(1) If we can tell a completely empirical psychological story about how certain sentences are such that competent language users have a certain kind of reaction to reading it that seems to track the meaningfulness vs. meaninglessness distinction without cluttering up our ontology with an extra category of objects--propositions--above and beyond sentences, then it seems like from the virtue of simplicity alone, we should do that.

(2) Also, if "propositions" are taken to be abstract object (not located in space or time, causally inert) then it seems to be pretty much impossible for them to be doing any work in that empirical story about how literate people learn things from reading certain sorts of ink scratches. (Granted, there are lots of different theories of propositions, and this criticism wouldn't by any means apply to all of them, at least not in that form, so I'm not inclined to put much weight on it.)

(3) When we look at the views of those who do believe in propositions, we see that while (as far as I know) everyone agrees that only meaningful declarative sentences can express propositions, it's far from the case that everyone accepts that being a meaningful declarative sentence is a sufficient condition for expressing a proposition. A certain variety of truth-value gap theory explicitly denies this (e.g. Kripke at least gestures in this direction in "Outline of a theory of truth"), holding that, e.g. certain paradoxical sentences are meaningful but don't express propositions and as such aren't truth-evaluable. I'm deeply unsympathetic to this move, but it does establish that a skeptic about propositions isn't really begging the question or assuming anything too dubious by separating out meaningfulness from the proposition issue.

(BTW, the point about Mandarin, etc., is an interesting one. I've heard people claim that before, but I've never been confident enough about whether the grammatical sense of "tense" that those languages lacked sufficiently tracked all of the features of languages like English the A-Theorist is talking about to give the B-Theorist much support. It's certainly something I'd like to hear a bit more about.)