Electoral overnegation

This post from Amy Hamblin's Community Blog was on the front page of the Obama-Biden web site for a while last night:

We're still awaiting the final results tonight, but one thing is clear — this grassroots movement can never be underestimated. Thank you to everyone who helped us make an astounding 1,053,791 calls today! I know it wasn't easy and many of you kept calling long after you were tired and your voice had grown hoarse, but your calls to get our supporters out the polls helped tip the scales in key battleground states like Pennsylvania and Ohio. [emphasis added]

We've discussed the general phenomenon of overnegation — semantic problems typically arising from the combination of modals, negatives, and scalar predicates —  many times in the past. The particular case of cannot/must not underestimate/overestimate is discussed here, here, and here.

1. Jonathan Mayhew said,

November 5, 2008 @ 12:53 pm

I don't think it's an overnegation. I read it as meaning "we shouldn't underestimate the power of grass roots movements…" In other words, thinking they are less powerful than they are, underestimating them, is a mistake. I see how it could be seen as an overnegation, though, if it is read as "no amount of estimation of such movement is overstated."

[(myl) This is similar to Barbara Wallraff's analysis, discussed here. The trouble with this idea is that it implies that can == should, which isn't true in general, with or without a negative. Modals are protean, but still, "you can't win" is not the same as "you shouldn't win". Or to paraphrase the example that I used in discussing Barbara's suggestion, "I can never jump farther than 15 feet" doesn't mean "I should never jump farther than 15 feet".

It's true that you could fix Ms. Hamblin's phrase by substituting "must" or "should" for "can". But that doesn't imply that in this context, can actually means "must" or "should". ]

2. Chris said,

November 5, 2008 @ 1:21 pm

True, but I disagree with your disagreement with Wallraff, anyway. George Washington wasn't incapable of telling a lie, and it would be absurd for Washington or anyone else to claim that he was so incapable – i.e. to classify the scenario of him telling a lie in the same category as the scenario of him jumping 50 feet. Clearly he meant "I must not tell a lie", whether that alternative meaning is listed in dictionaries or not.

Can == should isn't true in general, but it is true sometimes, including Washington's apocryphal utterance.

You're being excessively rigid, IMO, if you interpret all use of modal "can" as statements of literal physical impossibility, rather than implausibility (The Cubs can't win this year), moral imperative (I cannot tell a lie), illegality (The President can't just break any laws he feels like breaking), and other extended uses.

3. Jonathan Mayhew said,

November 5, 2008 @ 1:33 pm

I concur respectfully with what Chris says. Can covers a wide semantic ground. Even if she should have written should, I don't think she should have written overestimate instead of underestimate. The problem, if there is one, is not with under or over negation but with the choice of a modal that sounds slightly "off." My head hurts!

4. Jean-Sébastien Girard said,

November 5, 2008 @ 1:54 pm

I read it as an understatement of "cannot afford to underestimate", myself.

5. Mark P said,

November 5, 2008 @ 2:14 pm

I have to strain to read the quote as meaning what the author apparently intended instead of the opposite.

I also think the quote attributed to Washington does actually mean that he is incapable of telling a lie, not because it violates the laws of physics but because that is the sort of man (or boy) he is. It certainly doesn't mean that he mustn't tell a lie. That makes it sound more like he is afraid of punishment for lying rather than being willing to accept punishment for tree cutting rather than tell a lie.

6. Chris said,

November 5, 2008 @ 2:31 pm

I forgot to mention before: I think a possible source of the extended usages of can may be speakers drafting it to fill the semantic void left by the decline of ought. (This theory presumes that this usage of can is recent, which I have no positive knowledge on one way or the other.) In a passage written 200 years ago I would not be surprised to see "ought not underestimate" to express the same concept. Why can is used in these senses rather than (or more precisely, in addition to) should or must, I don't know; but the latter aren't subject to the same ambiguity and therefore don't get the same attention.

Also, I would like to point out the phrase "literally can't", used to express actual impossibility (and, in my experience, often occurring after a use of plain "can't"). Surely this is only necessary because plain "can't" can have other meanings besides literal impossibility – you can't deny that.

P.S. The Merriam-Webster Dictionary of English Usage, IIRC sometimes extolled in these virtual pages, comments on "cannot" for "should not, ought not" in their second note on "cannot". They seem to consider it standard English.

[(myl) I don't see anything like this in my copy — do you mean the discussion of "cannot but, cannot help, cannot help but"?]

7. Skullturf Q. Beavispants said,

November 5, 2008 @ 7:29 pm

Of course, I normally understand "You can't win" as "It's impossible for you to win", or, through hyperbole, "It would be very difficult for you to win."

However, I can imagine some contexts where one might say "You can't win" with the meaning of "You shouldn't win." For example, if my friend bets a lot of money on a footrace against someone, and then his opponent gets injured shortly before the race, severely hampering his running ability, it's plausible that I might tell my friend something like "You can't win this race! That would be wrong!" meaning "I don't recommend that you beat this guy!"

8. Chris said,

November 6, 2008 @ 10:28 am

[(myl) I don't see anything like this in my copy — do you mean the discussion of "cannot but, cannot help, cannot help but"?]

No, that's separate; I'm referring to the discussion right after "cannot, can not". I wonder if we are using different editions? I can't find any edition information in my copy but it is copyright 1989 and has the ISBN 0-87779-032-9.

Another example occurred to me last night: Jim Carrey's character in the movie Liar, Liar is cursed with the inability to lie – he inadvertently blurts out the truth. I would say that he literally cannot tell a lie, but I would not interpret Washington as claiming to be under that sort of compulsion. (Obviously Washington hadn't seen the movie, but I think similar concepts of compelled truthfulness date back to mythology.) Washington *chooses* not to lie even when it might benefit him, and it is that choice that establishes him as honest. (Carrey's character is definitely not honest, even though everything he says is true – he tries to lie on many occasions, but fails because of the curse.)

Also, if Washington refuses to lie because of apprehension of punishment, rather than his own conscience or principles, then that isn't very moral, IMO. At this point the discussion leaves the realm of linguistics and enters that of moral philosophy.

I also agree with Tim Silverman's point in the successor thread about goal-directed deontics, which can give rise to non-morally-loaded uses of "must" and permissive "can". I would add that sometimes a presumed unpleasant consequence is appended – "you can't leave the cake in the oven too long or it will burn"; clearly you *could* leave the cake in the oven too long, but this is why you probably don't want to. This may actually be the closest to "cannot be underestimated" – sure, you *could* underestimate it, but then you'd be making a serious mistake, and you wouldn't want to do that, would you?