December 14, 2007

The non-existence of Kilpatrick's Rule

In response to my recent post "Authoritarian rationalism is not conservatism" (12/11/2007), Breffni  O'Rourke has come to the defense of James Kilpatrick. I think Breffni's points are good ones, so I've given the whole note below, along with my response.

But first, a bit of background. Back in July of 2006, James Kilpatrick complained in print about the "horrid" headline "Mass Transit Not An Option for All Drivers", on the grounds that "if mass transit is not an option for 'all' drivers, it cannot be an option for even one driver". He added, "Even a little ambiguity is a dangerous thing. The problem with this Horrid Example is that it creates a nanosecond of uncertainty."

Neal Whitman and I ignored the "nanosecond of uncertainty" business, since a literal application of this idea would put pretty much all of the English language off limits. But we did take literally the assertion that "if mass transit is not an option for 'all' drivers, it cannot be an option for even one driver"; and we took this to be a claim about the interpretation of not and all in phrases of the same general structure as the "horrid" headline. Neal discussed the semantics of the question, in terms of the scope of operators in the corresponding logical formulae; and I added a set of historical citations, showing that Kilpatrick's prescription has been frequently and consistently violated by the likes of William Shakespeare, Herman Melville and T.S. Eliot.

At the time, I got a certain amount of negative email from Kilpatrick partisans; and recently, a related issue came up in Jan Freeman's column in the Boston Globe, which brought another few emails accusing me  of pandering to the vulgar mob, subverting truth by mere empiricism, and so on. So I wrote about the "strange tangle of ideas... according to which an appeal to the historical authority of usage by elite writers is seen as a step in the direction of mob rule", and suggested that Kilpatrick's partisans, though calling themselves conservatives, are actually authoritarian rationalists.

Here's Breffni's note:

First of all, I'm entirely with you on "authoritarian rationalism".

However, I think you were a little unfair to Kilpatrick in the case of the headline "Mass transit not an option for all drivers". I thinnk there really is something going on there. I don't think Kilpatrick explained himself very well (he barely explained himself at all), I don't think he thought the thing through, and certainly he wildly overstated the "horridness" of the headline. Even so, he didn't try to generalise to a "Kilpatrick's Rule": he only complained about *this specific headline*.

So, assuming good faith on his part, and assuming no real grammatical rule is being breached, where does his "nanosecond of uncertainty" come from?

You and Neal Whitman looked at negation and quantification alone, but I think on closer inspection there is a (mild) problem arising from the interaction of scoping ambiguities with a conflicting bracketing preference. In the headline, "not" is part of what I think is apt to be understood as an idiomatic chunk, "[is] not an option" - as in "Failure is not an option", "Surrender is not an option". It wants to be understood and parsed as a unit. On a crude Google search, three out of seven first-page hits for "is an option" (excluding "What is an option?", and also duplicates) are instances of "Failure is an option", which I take to be stunt allusions to "Failure is not an option".

So I think "not an option" is not just the syntactic negation of "is an option". If this is right, then "not an option" in the headline is open to being understood, if only on a fleeting preliminary parse, as occupying a single slot:

"Mass transit is (not an option) for all drivers"

...which conflicts with the intended interpretation:

"Mass transit is not (an option for all drivers)".

This would mean that one *possible* first-reading comparator for the headline "Mass transit not an option for all drivers" is "Mass transit unaffordable for all drivers" - the interpretation Kilpatrick claims to prefer (for polemical purposes, doubtless). The examples you use to counter the putative "Kilpatrick's Law" don't have the same ambiguous bracketing brought about by "not an option". "The solution of the mystery was not known to all" (Trollope) is obviously fine, but the "mass transit" headline has a possible reading that's more akin to "The solution of the mystery was unknown to all" - hence the "nanosecond of uncertainty" that Fitzpatrick felt, and which I think I may also have felt, before spending all this time staring at it. How strongly you feel the conflict will presumably depend on how formulaic "not an option" feels to you, and I imagine that will vary from speaker to speaker.

PS: I tried to use Google to back up my claim about "not an option" being a chunk, but Google declined to endorse me unambiguously:

"is not an option" OR "isn't an option" -"is not an option which" - "is not an option that" -"isn't an option which" -"isn't an option that" -- 264,000

"is an option" -"is an option which" -"is an option that" -"what is an option" -- 205,000

Still, it sounds like a formula to me - a mainly American one.

[Back to me.]

Breffni makes two excellent and related points. The first is that Kilpatrick may not have meant to claim anything about the interpretation of not and all in general. The second is that "not an option" is a common collocation, and this may change the normal scope preferences in examples like the cited headline.

With respect, I disagree with the first point. Although Kilpatrick's ire was aroused by the specific headline in question, he explains the problem in terms that apply to any sentence of similar structure. And it's true that his discussion is brief and allusive, but a similarly brief complaint by John Dryden was part of the process that has led to the slaughter of millions of innocent clause-final prepositions.

As for Breffni's second point, I'm less certain. It's certainly true that "not an option" is a common collocation. For example, from this morning's Google News:

  a/an __ not a/an __ ratio

And obviously a "local" interpretation of such a collocation would be incompatible with wide scope of negation. Thus"Mass transit is impossible for all drivers" can't possibly mean "It's not the case that mass transit is possible for all drivers". So if "not an option" tends to be interpreted as if it were a single word, like impossible, then Kilpatrick would be right about this case.

But there are many common examples of other constructions that point in the opposite direction. For example, contrastive subsitution of option seems to be no problem at all:

Renovation is not an option, it's a necessity.
The County states that recycling is not an option, it's the law.
Firestopping is not an option but a requirement.

Preventive diplomacy is not an option but a necessity

Obviously, forms like "Renovation is impossible, it's a necessity" are completely out of the question here -- but in reading the examples cited above, I can't detect the predicted instant of hesitation, putatively caused by a desire to interpret "not an option" as if it were a single word.

And in phrases of the form "not an option <preposition> all <plural nominal>", the interpretation seems without exception to give wide scope to the negation -- exactly as the surface-structure order predicts:

There is the possibility of camp for school-age children, but it is not an option for all working parents...
But card checking is not an option for all Las Vegas unions.
Out-wintering beef suckler cows on hill grazings is not an option for all farmers but there are situations where it can work well...
It does require some user exertion, meaning the iBOT is not an option for all wheelchair users.
She is generally a proponent of the cageless approach but warns that it is not an option for all dog owners.

Unfortunately this is not an option in all email clients.
Flex time is not an option in all positions ...
This is not an option in all jurisdictions.
This is not an option with all xenobiotics.
Heating substrates to elevated temperatures can help reduce columnar growth, but this is not an option with all polymer substrates.
TIF is not an option on all digital cameras so please check your owner’s manual.

Phrases of this form are reasonably common, and I have yet to find a single one whose interpretation goes in the direction that Kilpatrick claims to prefer.

We'd have to design and implement some reaction-time or eye-gaze experiments to try to determine whether the claimed ambiguity is causing any processing delay. But if the ambiguity were a seriously active one, I'd expect to see a reasonable percentage of examples in which the interpretation goes the other way, as we do for genuine cases of psychologically-valid scope ambiguity.

Posted by Mark Liberman at December 14, 2007 08:04 AM