## "… not understating the threat"

Preet Bharara, "Asleep at the Laptop", NYT 6/3/2012:

THE alarm bells sound regularly: cybergeddon; the next Pearl Harbor; one of the greatest existential threats facing the United States. With increasing frequency, these are the grave terms officials invoke about the menace of cybercrime — and they’re not understating the threat.

Let's unpack that. "The officials are understating the threat" would mean that on the degree-of-threat scale, the true threat level is higher — more threatening — than the degree of threat cited in the officials' statement(s):

If we negate that assertion — "It's not the case that the true threat level is higher than the degree of threat cited in the officials' statements" = "The true threat level is not higher than the degree of threat cited in the officials' statements" = "The officials are not understating the threat" — then the real threat level must be less than or equal to what the statement(s) say it is:

This is almost certainly the polar inversion of what the author wanted to say, which was that the real threat level is as great or greater than what the officials say — in other words, that "they're not overstating the threat".

As we've often observed, the combination of more than one negative and a scalar predicate, especially when one of the negatives is lexical, is generally too much for our poor monkey brains to handle without diagrammatic assistance.

[Tip of the hat to John Swindle]

## 30 Comments

1. ### Agustin said,

June 5, 2012 @ 1:19 pm

the real threat level must be lower than the statement(s) say it is

Or the real threat level can be equal to what the statement(s) say it is.

Just a bit of grade-school math: A is not less than B, so A could be equal to B, or A could be greater than B.

[(myl) Indeed, as the diagram indicates. But I don't think this is what the author meant to convey.]

2. ### QET said,

June 5, 2012 @ 1:20 pm

And here I thought that figure of speech had an illustrious name: litotes. Also, what if, in not understating, the official is simultaneously not overstating, but accurately stating?

3. ### Stu said,

June 5, 2012 @ 1:24 pm

Almost, but not entirely. If the true threat level is not higher than the stated threat level, that means that the threat level is either lower than the stated threat level (as you noted) OR at the SAME threat level. Read this way ("not understating" = "accurately stating" or "overstating"), Mr. Bharara's statement still ends up working out, though admittedly it does seem to suggest that they're being a little hyperbolic.

That said, I had a heck of a time understanding what the problem was until I realized that he probably meant to say "not overstating" the threat, which works out a little bit more in his point (that the threat is actually higher than OR at the same level as the statements about it).

4. ### Valerie said,

June 5, 2012 @ 1:34 pm

And yet "they're not overstating the threat," meaning they're not exaggerating the threat (and the threat is therefore at least as great as they say it is) seems easier to grasp, doesn't it?

[(myl) Presumably that's what Mr. Bharara meant.]

5. ### mgh said,

June 5, 2012 @ 1:53 pm

google gives many examples of "it is no understatement to say"
I would not have caught this one if I had been reading casually!

6. ### Mr C said,

June 5, 2012 @ 2:05 pm

One slight error in your analysis: "then the real threat level must be lower than the statement(s) say it is" should actually be "then the real threat level must be lower than or equal to the level stated". ~(p=q)

7. ### Mr C said,

June 5, 2012 @ 2:06 pm

Crud, several people beat me to it while I was typing.

8. ### Brian said,

June 5, 2012 @ 2:07 pm

It's impressive: Even though I was reading it on LL, in an article that described it as a misnegation, I still failed to see the problem at first. Although in my case this was at least in part due to lazy reading: my mind automatically substituted "overstating" for "understating", or at least the meaning, and I only realized I had done this when I read Valerie's comment.

9. ### Mr C said,

June 5, 2012 @ 2:07 pm

Argh, also, my mathematical notation came out all stupid looking.

10. ### QET said,

June 5, 2012 @ 2:19 pm

On reconsideration, I now see the writer's error as well. Sigh.

11. ### Dan Hemmens said,

June 5, 2012 @ 3:20 pm

This is an almost meaningless distinction, but I wonder if "understating" here was a substitution for "overstating" or "underestimating".

12. ### Dan Hemmens said,

June 5, 2012 @ 3:21 pm

Wait, I'm talking nonsense, it would still be overestimating, not underestimating. I am an idiot.

I could have sworn there was a valid "under" construction, but there isn't.

13. ### BZ said,

June 5, 2012 @ 3:28 pm

Misunderstating maybe?

14. ### Dan Lufkin said,

June 5, 2012 @ 3:59 pm

Could I see a diagram treatment of "they misunderestimate me"? I think it has to be 3-dimensional.

15. ### Al said,

June 5, 2012 @ 4:43 pm

This monkey brain thinks the writer was going to say "they're not underplaying the threat", but they got a bit confused when moving from less-used "(under|over)play" to much-more-frequently-used "(under|over)state".

16. ### Steven in Spain said,

June 5, 2012 @ 5:08 pm

I think Al's on the right track: the author meant to say "they're not downplaying the threat," i.e. they're giving the threat the proper level of credence it deserves.

[(myl) As far as I can tell, downplay (though a bit vaguer in meaning) has exactly the same logic as understate. The opposite of downplay, more or less, is exaggerate. In this case, "They're not exaggerating the threat" would be congruent with the author's overall message; "They're not downplaying the threat" would flip it in the opposite direction, and is not at all what he meant to convey.]

17. ### Faldone said,

June 5, 2012 @ 5:55 pm

This is a recurring theme here at Language Log Plaza. I wonder, if it's a common usage and even educated LL groupies have trouble seeing that there is a problem without analyzing it with great effort, could it maybe be a problem with the understanding of the grammar of the language rather than a problem with the grammar used by the speaker and the hearer.

18. ### MattF said,

June 5, 2012 @ 7:15 pm

@Faldone

One might even say there's a certain peevishness in the LL coverage of misnegation. On the other hand, one would think that 'not understate' means something different from 'not overstate'. Also, these items always make me reflect on the idea of a 'Miss Negation' beauty contest.

[(myl) One obvious difference between what linguists have to say about misnegation and what peevers have to say about the things that annoy them is that we linguists are interested, not annoyed. There's a well-established and interesting linguistic literature about negation-confusion, focused on trying to figure out the nature and causes of the phenomenon.

There do seem to be some cases where a lexical ambiguity or lexical change is at issue, such as "still unpacked". There are other cases that have become idiomatic, probably through the mechanism of negation by association, like "could care less". There's an argument that some cases involve subtle issues of natural modal logic. And there may be some examples where a residual tendency towards negative concord is at issue.

But in cases like the one discussed in this post, it's hard to avoid the conclusion that multiplex negatio ferblondiat, as Larry Horn put it. And there's a tradition of serious analysis based on this premise, starting with Wason & Reich, "A Verbal Illusion", The Quarterly Journal of Experimental Psychology 31(4):591-97, 1979.]

19. ### EndlessWaves said,

June 5, 2012 @ 7:28 pm

My feeling is that he's not expecting you to take the preceding sentence and a half seriously.

Terms like 'Armageddon' are often used trivially rather than seriously and can frequently be found referring to minor problems. If the officials were to use those terms in a common way they'd be understating the situation – but they're not.

20. ### Viseguy said,

June 5, 2012 @ 7:58 pm

"… these are the grave terms officials invoke about the menace of cybercrime — and they’re not understating the threat."

Was the intended meaning possibly, "… and [in these officials' way of thinking] they're not understating the threat"? It's the "and" that leads me to wonder about this. If the author had truly meant to say "they're overstating the threat", I would have expected the second clause to begin with "but".

21. ### Viseguy said,

June 5, 2012 @ 8:02 pm

Strike my last comment. I misunderestimated my ability to understand what the author likely wanted to say.

22. ### Jerry Friedman said,

June 5, 2012 @ 9:28 pm

It is evidently no litotes to say that misprocessing misnegation is not any less easy than misnegation itself.

23. ### Antariksh Bothale said,

June 5, 2012 @ 9:47 pm

Even I am inclined to think that, in this case, the author meant to convey that the threat has been accurately assessed and stated (as opposed to saying that it has been exaggerated) by the officials and that this won't qualify to be in the list of undeniable mis-negations we otherwise encounter on LL.

24. ### Ellen K. said,

June 5, 2012 @ 10:45 pm

I understood it, when I read it the first time, as the author saying that the threat is at least as great as what these people are saying. And I still think that's what the author meant to convey.

It did not occur to me, upon reading it, that the literal meaning was backwards of what I understood it to be saying. But it didn't take anymore than reading "Let's unpack that." to realize what this issue is.

25. ### Dan Lufkin said,

June 6, 2012 @ 8:28 am

@Jerry F — No, we should not fail to allow any misnegation to go less unchallenged than ever.

26. ### David L said,

June 6, 2012 @ 9:43 am

Even after reading and understanding Mark's explanation, I was perplexed that one part of my brain still wanted to understand "they're not understating the threat" as meaning the same as "they're not exaggerating the threat," even though another part of my brain is well aware that 'understate' and 'exaggerate' mean exactly the opposite. But then I was able to resolve the conundrum. Consider:

If the story had said "they're understating the threat," that would be bad. They are declaring the threat to be less than they know it to be. Lulling us into a false sense of security. Endangering our lives. Not cool! But no, actually, "they're not understating the threat." OK, they're doing their job, they are on the ball. Thank heavens!

On the other hand, if the story had said "they're exaggerating the threat," that would be bad too. They're fear-mongering, trying to scare us into abandoning yet more of our civil liberties. Not cool! But no, actually, "they're not exaggerating the threat." OK, they're doing their job, they are on the ball. Thank heavens!

In short, the two double negative statements have the same positive connotation. "They're not understating the threat" and "they're not exaggerating the threat" mean the same thing iff they are assessing the threat accurately — which is presumably what we want them to do. QED.

(So I agree with what Antariksh Bothale said above).

27. ### erik i said,

June 7, 2012 @ 9:31 am

(In agreement with the above comment): I think that the statement "they're not understaing the threat" is read as "they're not kidding" or "they're not whistling Dixie". The literal meaning of the words has almost no effect on how we understand the phrase, I guess because it's hard to quickly grasp the literal meaning.

[(myl) That's a simplified form of the hypothesis explored and tested by Wason & Reich, "A Verbal Illusion", The Quarterly Journal of Experimental Psychology 31(4):591-97, 1979.]

28. ### Nelida said,

June 7, 2012 @ 6:52 pm

They are not exaggerating the threat – obviously, that is what the author meant to say. Ergo, not "overstating". Simple, my dear Watson.

29. ### Cybersecurity and the threat of concession | The Violence of Nations said,

June 9, 2012 @ 7:48 am

[…] U.S. Attorney Preet Bharara, in the June 3 New York Times. Hat tip to Language Log, which post focused on the grammar of "not understating". I have a more substantive […]

30. ### harrison said,

June 25, 2012 @ 2:16 am

I wonder if the composition of this sentence (or the way we process its intended meaning) has something to do with the common use of litotes in written language. The "not un-", although not etymologically related to constructions like "not unkind," is similar enough that it might fool our poor monkey brains.