Saturday, May 05, 2007

Clitics and affixes

This afternoon, I caught the last few minutes of CBC's And Sometimes Y, hosted by Russell Smith's. (These half-hour programs on English tend to be much better than Smith's occasional language pieces in the Globe & Mail.) In the bit I caught, author Michael Winter was justifying his abandonment of quotation marks and most apostrophes in his novel (novels?). Smith asked him about his practice of writing don't as dont. He said something to the effect that he just didn't see the point in constantly reminding readers that two words had been sandwiched together.

Indeed, the apostrophe does commonly occur in situations that can be rewritten as two words. But mostly it's there to stand in place of missing letters. It turns out that Winter actually knows this. On his blog, he writes,
"I just decided one day, what's up with this symbol that tells the reader of a contraction. That two words have been sandwiched and a letter or two left out. Why do I have to remind the reader of this grammatical omission."
Still, Winter's characterisation isn't entirely accurate. The fact that something can be decomposed into two words is merely coincidental to the apostrophe. Consider workin', ma'am, and gov't. In most of the seemingly combined words, the apostrophe doesn't even come at the boundary of the two words. It does in can|'t, but not in words like is|n't or does|n't.

I didn't write this to nitpick Winter's minor misstatement (he's got some good points and makes them in an engaging way; you can listen here from about the 20-minute mark.) No, the interesting part of this is that the n't ending on auxiliary verbs isn't really a contraction at all. Rather, it's an inflection.

In contractions, we have constituents, like the 're in we're, that are called clitics. It's often hard to distinguish between clitics and bound morphemes such as -ing, or -tion, but there are a number of good reason (discussed in Pullum & Zwicky, 1983) to argue that the n't ending has become an inflectional affix. I'll look at a few here.
  1. Clitics are promiscuous, attaching to a wide range of words, but affixes are picky and arbitrary.
  2. Phonological and semantic changes in the root are more typical of affixes than of clitics.
First, here are some clitics: 's = is, 're = are, 've = have. These can attach not just to nouns, but to prepositions, verbs, adjectives, even conjunctions:
  • The horse you rode in on's leaving.
  • something specific happening which relates to the press release but's not part of it
Affixes, in contrast, only go with certain classes of words: plural -s only with nouns, -ing only with verbs, and n't only with auxiliary verbs. It won't attach to other verbs (e.g., *maken't). In fact, it won't even attach to all auxiliaries. You find amn't only in certain dialects and *ben't just doesn't work at all.

Second, when the clitics attach, you rarely have to change the pronunciation of your root. For example, you don't get something like I have, changing to *ee've or any other such thing. With affixes, however, you have oddities like won't, don't and ain't.

Perhaps the most relevant point for ESL teachers, is that you can't always just switch a not to an n't. In sentences like the following, the auxiliary simply isn't directly followed by the not.
  • Can you not see what's before your eyes?
  • *Can youn't see what's before your eyes?
  • Can't you see what's before your eyes?
And sometimes even when there is a not right after a verb that can be an auxiliary, the n't chooses to go with the actual auxiliary. (fixed Aug 17, 2008)
  • Could they have not done something else?
  • *Could they haven't done something else?
  • Couldn't they have done something else?
There's more in the paper, but that's enough for now.

No comments: