How many L’s in “cancelled”?

I went to buy something the other day using a credit card, but I screwed up somehow and the machine ended up cancelling the transaction. It announced this to me in a message that persisted on the screen for an interminable twenty seconds as “The transaction has been canceled.” For those twenty seconds, all I could think about — aside from my lingering fear that perhaps my card had been disabled and now I was never going to be able to get whatever doubtlessly important object I was trying to buy — was that that message just didn’t look right to me.

I’ve always written the past tense of cancel with two L’s. It’s cancelled to me, cancelling as well. Because I’m not as familiar with the canceled spelling, it occasionally triggers a strange “can-sealed” pronunciation in my head. This is presumably because my brain follows one of those standard heuristics of English pronunciation, that a single vowel followed by a single consonant and an e means to make the first vowel long and silence the e. That’s what we have in such words as rile, smote, or gale. And it’s especially prominent to me since it’s in my first name (Gabe).

This pronunciation heuristic is generally followed in tense changes as well; the verb pan becomes panned in its past tense, with two n‘s, to maintain the short a sound. Without the double n, it’d be paned, which I’d pronounce, well, like paned (as in double-paned glass).

And yet I’ve noticed more and more over the years that my countrymen disagree with me. In error messages I see a single l, leaving me even more depressed about the error. The AP Stylebook disagrees with me too. But why? What caused Americans to move away from the general English spelling heuristic?

Ah, much better.

I didn’t know, but if there’s anyone who could shed light on this, it’s Ben Zimmer. He puts it at the foot of Noah Webster, the American Samuel Johnson. Webster compiled the first dictionary of American English, and consciously sought to distance American English from British English, which he saw as corrupted by the aristocracy. Because Webster was codifying American English as a dialect separate from the standards of British English, this gave him the ability to make the changes he saw as appropriate to the American forms.

One of the major changes he wanted made was spelling reform, and so in Webster’s first dictionary (1828, available in searchable form here), we see the beginning of many Anglo-American debates: colour appeared as color, centre was switched to center, and our target cancel was listed with past tense canceled, present progressive canceling, and noun form cancelation.* His idea here was to push for easier or more natural or more accurate (relative to pronunciation) spellings. The u doesn’t get pronounced in colour? Gone. Centre isn’t pronounced cent-ruh? Switch it. Cancelled doesn’t have a double-l sound? Smash ’em together.

Some of Webster’s revisions took over pretty quickly. A quick glance at Google N-grams shows color surging in AmEng in the 1830s, and surpassing colour by 1850. Center took longer, but still surpassed centre by the turn of the century.

But others, like canceled, stayed on the sidelines. Oh, canceled grew in popularity, but it wasn’t until the middle of last century that the two forms evened out, and it wasn’t until the ’80s that canceled finally asserted itself as the more common form.** Personally, I think that sluggishness is because this spelling change doesn’t make as much sense as the others. The second l may be silent, but it tells you not to change the stem vowel’s pronunciation, and thus it has something of a purpose.

What’s interesting about all of this to me is that Webster was primarily a descriptivist, compiling a dictionary wherein he was looking to accurately capture the American form of English. But he prescribed a new spelling for a large set of words, and now his changes, which for years were held in lower esteem, are becoming the thing that prescriptivists demand adherence to.

Unfortunately, in his attempt to simplify matters, Webster introduced new confusion. I don’t see how it’s easier to remember not to put in an extra l when all the similar words double their last letter. And worse, Webster’s changes didn’t fully take. Sure, canceled and canceling are doing fine, but cancelation never caught on. Thus the AP Stylebook (and many other usage guides) have the inflections of cancel as canceled, canceling, cancellation, which is needlessly complicated in my mind.

And so that’s the deal. In American English, single-l canceled is the common form, almost thrice as common as cancelled according to Google N-grams. There will probably be a day where the double-l form will look as old and affected as centre in American English, but that point isn’t here yet. Use whichever form you like more. Me? I like the across the board double-l forms. (Of course I do; I was born just before canceled surpassed cancelled.)

—
*: Webster also included the fun adjectival form cancelated, which I hope to incorporate into my speech in the future.

**: I hope the non-Americans in the audience will forgive my focus on American English. None of these Websterian changes have surpassed the original form in the British English portion of Google N-grams, and I don’t have enough personal experience in non-American Englishes to say anything more than these numbers do.

A debate centering around etymological and logical fallacies

I know it’s become common over these last few posts for me to discuss etymological fallacies, but that’s only because they’re so easy to disprove. They’re like a little vacation for me, a pathetic little vacation I take without moving from in front of my computer.

The current etymologically-motivated complaint I’ve grown tired of is the claim that you can’t say center around. This one’s fun because it involves geometry. (I must confess that I never actually took a geometry class. Instead, I took topology, which is sort of like geometry where the entire world is made of infinitely flexible rubber. This is why I can think of geometry as fun.)

Suppose you have a circle O. The circle gets its name from its center, O, so you might want to say that the circle O is centered at point O. But at the same time, the circle is located all around point O, so I can’t see anything unreasonable in saying that the circle is centered around point O either.

Other people, though, can. And they do. I especially like the exhortation of that last link:

“How can you center around anything? You cannot. You can center on or focus on something, but not around it. Think about it!”

So I started to think about it.  And try as I might, I couldn’t see how you could center on something. Take, for instance, that king of three-dimensional objects, the sphere.  Suppose the sphere X is on point O. Then the sphere X must be above point O, by the definition of on.  To have point O as its center, sphere X must extend equally in all directions from point O.  (More generally, point O must be the average of all points in object X.)  But it can only be the case that X is on O and X has O as its center if X is an impossibly droopy sphere; otherwise the center of X will have to be above O. Even if we move beyond spheres, it’s really only inverted bowl-type mathematical objects that could rest on their own centers. So really, isn’t center on illogical, too? Oughtn’t it to be center at?*

Please tell me you’re thinking that this discussion is really stupid.  It is, isn’t it? After all, if geometric logic really determined the proper choice of preposition in an idiomatic construction, we’d all be saying that this debate centers at a contentious point. And of course, we don’t. The Google n-gram corpus has 4858 examples of “debate centers on” and 1763 of “debate centers around”, but does not have a single attestation of “debate centers at” in a trillion words. Language isn’t geometry, and there is no reason to try to make it so. As the Merriam-Webster Dictionary of English Usage (MWDEU) puts it, “[…] questionable or sound, logic is simply not the point. Center around is a standard idiom […]”

So let’s stop dismissing center around out-of-hand for “logical” reasons and look at it dispassionately. How standard of an idiom is it? Well, it’s a fairly old construction; the OED first attests it in 1868 in Edward Freeman‘s solidly scholarly The History of the Norman Conquest:

(1) “It is around the King..that the main storm of battle is made to centre.”

Google Books has some even older attestations:

(2a) “Clouds of deep crimson centered around him, and one would think, by the glory of his parting, he was loath to deprive the earth of her light […]” [1824]
(2b) “[…] I occasionally acted as chaperon to Miss Jameson, but as my hopes centered more trustfully around one object, my taste for general society diminished […]” [1840]
(2c) “His thoughts returned to Miss Percival; his hopes again centered around her.” [1840]

Is center on any older? Not much. The OED’s first attestation of center on comes from 1789, but this usage is based on the obsolete definition “to converge on”. If we don’t accept that example because of the (subtle) difference in meaning, the next attestation is in 1867. That puts it contemporaneous with the first attestation of center around. Google Books has some older attestations, although they might fit better with the “converge on” meaning:

(3a) “Our hope centered on God in Christ, and our hearts ready to leave the world.” [1775]
(3b) “Had it centered on a monarch, it would have given the means of a vigorous and healthy government; but it never centered on a monarch.” [1834]

MWDEU notes that up through the 19th century, in was the primary idiomatic preposition used with center, alongside a smattering of on, upon, and around. More recent usage has shifted these proportions, with on and around taking precedence in American English and round frequent in British English. (in has really fallen by the wayside.) And between the emergence of center around and grammarians’ first complaints about it in the 1920s, no one seems to have thought it illogical. I guess they just weren’t as good at geometry back then.

Summary: There’s nothing illogical about center around, at least nothing inconsistent with the logic of language. (And center on isn’t a paragon of logic itself.) Regardless of the question of logic, center around has been around for 150 years, and there’s no reason to ditch it now.

—

*: And while we’re at it, why are prescriptivists willing to accept center as a verb in the first place? Don’t they know the verb center comes from the noun center? I thought they hated all the bastard verbs that come from nouns, like access.