This should have a title, I guess
D. Dowty ("Toward a Semantic Analysis of Verb Aspect and The English "Imperfective" Progressive"):
As it has been argued … that the futurate progressive of (26) ["John is leaving town tomorrow"] semantically involves some notion of planning, it might seem that the event of leaving described in (26) may, after all, be 'in progress' in this loose sense. Though this line of thinking may have merit, to pursue it would quickly lead us into the very difficult but fascinating questions of how humans conceive of events as grouped together into causally and temporally related 'meta-events' involving intentions as well as actions, and I doubt that such investigations would lead us to productive results in model-theoretic semantics anytime soon (p 67)
One might be tempted to say on reading this "that's why he's a linguist, not a philosopher", but even granting that one might think that a linguist's interest in language should have something to do with the way humans communicate, and that his or her ambition shouldn't be to advance the state of model theory but to pursue routes, even routes that hold little immediate promise for model theory, that will lead to enlightenment about that communication. But if you like model-theoretic semantics then that's that, I suppose. (Though the last paper I read about this, Landman's "The Progressive", had the amusing feature that the analysis, rich in lambdas and backwards Es and whatnot, in the end turned on what it is reasonable to believe might happen.) More worrying to me is the actual proposal he gives:
[PROG φ] is true at I and w iff there is an interval I' such that I ⊂ I' and there is a world w' for which φ is true at I' and w', and w is exactly like w'sic, I assume this means "including"] I.
Of everything you might worry about here, I want to ask this: what sort of a thing is φ? It is both something that can be operated on to yield a progressive sentence such as "John is crossing the street", and also something that is truth-apt. The perhaps most natural candidate, "John crosses the street", doesn't work, because even though it can be true or false it can't be given the right sort of reading. Another candidate that will flip truth values in the way we want could be "John crossed the street" (going this route would incline you, perhaps, to the way of thinking about events and processes R. Stout gives in Things That Happen Because They Should[1]), but then, since Dowty for independent reasons wants the interval I to be the smallest "over which the appropriate change-of-state takes place", John will only be crossing the street at the moment at which it becomes true that he has crossed it (nor will the so-called imperfective paradox hold). It is presumably (I can't recall the precise reasoning) for considerations along these lines that Galton has his progressive-creating operator operate not on propositions but on what he calls event radicals—here it would be John-CROSS-the-street—which are not truth-apt. Or perhaps φ isn't a non-progressive proposition at all, but rather a closed-progressive proposition, that is, one in which the truth of the present licenses the inferences to the truth of the future perfect. (This seems generally to be given as the licensed inference from the simple past/past progressive to the perfect, but I don't think it makes much difference.[2]) I am pretty certain that's not what Dowty is thinking of, but it might work. It doesn't help very much with understanding the progressive, certainly. It would actually be somewhat similar to Galton's official account, and I think it has a weak point in that "smallest interval" thing: without the smallest interval all sorts of crazy propositions are true (for instance, my house is burning down), but its inclusion creates a problem involving identifying what the interval over which the change of state taking place is, or would be like, and when it begins—which you might think just reduces to knowing when the analysandum is true. (An example: it is not really clear at what point a house burning down starts burning down. I mean at what point the house is burning down. The drapes being on fire doesn't get it.) I suspect that we will end up thrown back onto the problem of how humans group events (or, since I don't mind countenancing them, processes) together, even without intentions.
[1] You would think a book with this title would be similar to the one Carnap wrote about what he learned from Col. Parker, but no.
[2] Actually there is this difference: From "he is pushing a cart" you can infer both "he will have pushed a cart"[3] and "he has pushed a cart", but even if you could infer "he will have drawn a circle" from "he is drawing a circle", you couldn't infer "he has drawn a circle". So the past-tense case is simpler.
[3] On the reading of this where all it means is that it will in the future be the case that he has pushed a cart, and not (what is really the more natural reading) that there will have been a cart-pushing by him between now and some point in the future, which possibly started before now but will, if so, extend past now.
Comments
on 2009-03-09 1:26:24.0, Daniel Lindquist commented:
"From "he is pushing a cart" you can infer both "he will have pushed a cart""
Maybe I'm missing something from the context for this footnote, but isn't the invalidity of this sort of inference one of Thompson's main points? You can be crossing the street without it ever needing to become the case that you have crossed the street, for you might get hit by a bus partway through. This won't mean that it's false that you were crossing the street: that was just what you were doing, when you were interrupted. That's the important difference between the imperfective and the perfective. The imperfective is compatible with never finishing the job.
Or maybe I'm just reading "infer" too strongly. It follows logically that he will have been pushing a cart; it follows from an additional premise that he will finish the job that he will have pushed a cart. This additional premise it the sort of thing one would have a defeasible warrant to, I think. Or at least it's the sort of thing one often has good reason to think true, when one thinks that the related progressive imperfective judgement holds true.
Also, I'm not sure why the drapes being on fire can't show that the house is burning down. Maybe the fire spreads from them to the rest of the place. After all, I can be playing a poker game before I've looked at any cards or bet anything, and while I'm pouring myself a drink. I don't need to have made any progress for the progressive to be true -- its truth lies in it just being that I am moving towards making progress, ceteris paribus, not that I've already done stuff (for it was already true as soon as the doing of stuff began, and the doing of stuff would not have begun were it not true).
But then, it's not clear to me what the most natural thing to say about the progressive is, after several hours of Ford's seminar on the topic. I really cannot remember how much of what Ford's said is in "Naive Action Theory" and how much is extra-"Life and Action"-al.
and, further, on 2009-03-09 9:44:17.0, ben wolfson commented:
Thompson's point applies to "he is crossing the street", because you don't need to have reached the end of the street to be crossing it. But if you're pushing the cart, then you have pushed the cart—there's no endpoint you have to have reached for the perfect formulation to become true. (Unless you think that you can, really, be pushing the cart, without yet having touched the cart—and I'm very sympathetic to that claim, but it's one to which a lot of reasonable resistance can be offered, mostly in the form of the observation: but look, you aren't doing anything yet! But if you do accept that, then it's true, that inference fails also.)
The footnote regarding the reading of "he will have pushed the cart" is meant to deny even the need for the extra premise—all it means (and admittedly this is an unnatural reading, but if you're a linguist interested in tense logic or whatever, it's probably the reading you're using) is that in the future it will be true that he pushed the cart. And that's true even if it all it means is that in the future it will still be true that he pushed the cart—I mean true of the same cart-pushing. We don't need the success condition since if he is pushing the cart now he's already succeeded in doing some cart-pushing, modulo the possible twist above.
I'm not sure why the drapes being on fire can't show that the house is burning down. Maybe the fire spreads from them to the rest of the place.
Sure, that could happen. I myself wouldn't say, from the drapes being on fire, that the house is burning down (maybe if the drapes were soaked in kerosene), because that seems, to me, to be too localized still. (And in this case it's not as if the house is planning to burn down and taking this first step to its aim.) If the drapes don't work for you, the immediate fire-involving precursor might.
I'm not sure how you want to cash out the distinction between "making progress" and "moving towards making progress". That sounds like making progress. This: "the doing of stuff would not have begun were it not true" is also a claim I want to make, but it's also one to which you can expect the insistence that before it was true—before the doing of the stuff began—all that was true was that you intended to do the stuff, not that you were doing whatever doing the stuff amounted to.
and, further, on 2009-03-23 20:01:43.0, Daniel Lindquist commented:
oh hey, I forgot about this comment thread. I wish I got e-mailed when new comments hit.
"But if you're pushing the cart, then you have pushed the cart—there's no endpoint you have to have reached for the perfect formulation to become true. (Unless you think that you can, really, be pushing the cart, without yet having touched the cart—and I'm very sympathetic to that claim, but it's one to which a lot of reasonable resistance can be offered, mostly in the form of the observation: but look, you aren't doing anything yet! But if you do accept that, then it's true, that inference fails also.)"
In Ford's action seminar, he was pretty adamant that you can be opening the window without having touched the window (and so without the window having been opened). This was supposed to be brought out most clearly by the poker example. The "you're not doing anything yet" objection seems to rest on an equivocation: I can be opening the window while I am not yet opening the window (because I am currently taking my sweater off, say).
Actions like cart-pushing are weird, since there's no set endpoint to them. Someone can have pushed the cart even if he never finished pushing the cart. This is different from crossing the street, then -- one hasn't crossed until one is on the other side, and then one is no longer crossing. One's finished phi-ing IFF it is true that one has phi-ed.
"(And in this case it's not as if the house is planning to burn down and taking this first step to its aim.)"
I don't see why we need to attribute planning to the house to attribute "agency" here. For one thing, the house isn't burning anything, the fire is. The house is the patient. To say that the house is burning (when so far it's just the drapes -- NB that "burning" here is intransitive --) is just to think that the drapes' burning is the first step on the way to the house burning -- that this would be the way things would go if they continued as they are currently going. They are the means by which the house is being burned down, though the agent here is not acting as an exercise of practical reasoning. (This seems to account for your distinguishing the drapes case from the kerosene-soaked drapes case. Only in the latter does it seem like this would be a means to the end of the house burning down, i.e. it would be an answer to the question "how did the house burn down?".)
"I'm not sure how you want to cash out the distinction between "making progress" and "moving towards making progress". That sounds like making progress."
I can be moving towards making progress at Phi-ing without having made any progress on Phi-ing, since I can not be doing anything that serves as a means to the end of Phi-ing. I can be playing poker without having played any poker.
Thinking on these examples some more, I suspect that there is not a uniform treatment of all this possible. Sometimes the truth of the progressive Phi-ing implies the truth of the perfective Phi-ed, and sometimes not, it seems. Nor can I think of any clear way to distinguish the cases.
man it has been two weeks since you responded and you got back in like eight hours, I am bad at this internet thing
and, further, on 2009-03-23 20:30:10.0, ben commented:
Actually, a lot of the above is what I think.
However, I will nevertheless have a longer comment later (but not two weeks later!).
and, further, on 2009-04-06 18:02:05.0, ben wolfson commented:
I'm so close to having made myself a liar I don't think I can stand to let the opportunity pass by.
and, further, on 2009-04-13 17:27:50.0, ben wolfson commented:
I can however report something that I didn't notice before, namely, that Dowty is really wedded to model theory:
"It has been suggested to me by David Lewis that perhaps [PROG $\phi$] should be defined as true in case $\phi$ will be true in that possible world similar to the actual one in which the natural course of events' obtains. This may indeed be correct, but I presently see no way of making natural course of events' precise in model-theoretic terms."
So why bother thinking about it further, right?