June 21, 2011
Posted by Jay Livingston

There it was again, this time in a report from the Senate Foreign Relations Committee quoted today without linguistic comment by David Brooks:
The unintended consequences of pumping large amounts of money into a war zone cannot be underestimated. [emphasis added]
Usually I know what the writer means, but this time I had to go back and reread Brooks’s column up to that sentence. Yes, the committee meant the opposite of what it was literally saying. Literally, the sentence means that even if you think that there are zero unintended consequences, that’s still not an underestimate. No matter how low the estimate, you still cannot underestimate.

When I first mentioned this logical error (here), I hadn’t realized how prevalent it was. Today, Googling “cannot be underestimated” returns 2,070,000 pages, and I would guess that at least 2 million of them are like the one Brooks quotes.

A couple of those pages, though, are from Language Loggers, who have two explanations.* One is that the speaker means to say “should not be underestimated” rather than “cannot be underestimated.” Nice try, but it seems unlikely that people are thinking “should not” but saying “cannot.” Is there any other linguistic context in which people frequently confuse those two phrases? Offhand, I cannot think of any (or do I mean, I should not think of any?).

More likely is our inability to grasp multiple negatives. Putting things in the positive makes them much easier to understand. A simple single negative is also clear. But, beyond that, as I remember from algebra and multiplying polynomials, keeping track of those minus signs gets tricky. Sentences, especially in a government report, should not be written so that nobody can fail to misunderstand them.

That still leaves unanswered the question as to why we so often prefer to phrase things in the negative. As I said in that earlier post, I especially try to avoid negative phrasing when I’m writing True-False and multiple-choice test items. It’s unfair to students to require them to answer False when that means negating a negative

Brooks’s column, as you might guess, was about the contrary effects of all the money the US is now spending in Afghanistan. Did Brooks also, in the Bush-Cheney years, write about the unintended consequences of the billions spent in Iraq? The number of his columns on that topic cannot be underestimated.

* Ben Zimmer summarized these earlier this year in his New York Times Magazine language column (here).


PCM said...

I could care less.

Jay Livingston said...

Peter, I’m not going to look it up, but I’m pretty sure the Language Log has weighed in on this one too. Their non-prescriptivist view is, in my oversimplified version, that if everybody knows what it means, then that’s what it means, regardless of the literal logic. So when I hear “I could care less,” I would never presume to correct the speaker. I just silently feel a tad superior.

Anonymous said...

I think I'm in favor of the first interpretation (can/should).

This tends to be substituted when people are trying to add emphasis to prescriptive statements. For instance: 'You CAN'T name your son Alice! He'll be teased!'.

I think it's a pretty short step from that to to 'you can't underestimate how much trouble that name will cause him in school', and then it ends up in more formal places

Jay Livingston said...

Anon, As I said in the post, what makes me skeptical about this explanation is that I can’t think of any other context where people say “cannot” when they mean “should not.” But I can think of other contexts where the multiple negatives get things confused. In the earlier post I mentioned those news reports of Supreme Court decisions (“The Court failed to overturn a lower court decision reversing the inability of states to disallow . . .”) I can never figure out who won.