Rick Detorie's One Big Happy for 10/27/2008:
We find it amusing when an apparently logical generalization about word formation goes badly wrong, as Joe's idiosyncratic inference does in this strip.
A couple of days ago, Eugene Volokh discussed a much more widespread case of failed linguistic analogy: "Hisself, My Son, and a Thought About Prescriptivism":
My five-and-a-half-year-old used ["hisself"] a few days ago, and I gently corrected him. We say "himself," I said, not "hisself." I'm a descriptivist when it comes to determining what is "correct"; but I want my child to learn not just any correct way of speaking, but the way that is going to best help him get ahead in life, which sometimes mean the mode of speaking that is most satisfying to self-described "purists." [...]
But of course I also wanted my boy to get a sense of the patterns in the language, so I pointed out the analogies — "herself," "themselves," "myself." Wait a minute! It's "myself," not "meself," and "ourselves," not "usselves"; the first-person reflexive uses the possessive ("my" and "our") followed by "self" or "selves." But the others use the objective ("him," "her," "them") and not the possessive.
And what is it that tells us that "myself" and "himself" are right, while "meself" and "hisself" are wrong? Not any supposed inner logic of the language, it seems to me, but simply usage: "Myself" and "himself" are standard among educated English speakers, at least outside narrow regional dialects, and "meself" and "hisself" are not. What is right to say in English is what educated English speakers say.
This is a cogent explanation of a point that you'd think would be clear to any rational adult who has thought about the analysis of language. But as Prof. Volokh observes, the argument still has to be made:
[W]hen I hear prescriptivists argue using what I think of as "logical prescriptivism" — this spelling or usage is right and that is wrong because of some inner logic of the word, or because of an analogy to other words — I remember examples like this.
(If you're skeptical that any well-informed people really do believe in "logical prescriptivism", see "The theology of phonology", 1/2/2004.)
In the case of word formation and its connections to sound and sense, logic is an especially unreliable guide. And of course the English writing system is a complex pattern of overlapping historical layers with sporadic intrusions of reform, for which the appropriate mode of analysis is more geological than logical.
But as we move to larger units — phrases, sentences, rhetorical or narrative structures — the combinatoric explosion of possibilities makes it less and less likely that any particular pattern can have idiosyncratic properties. Logic tells us that some sort of regular and productive system must mostly take over.
And since different languages may differ syntactically while supporting the same patterns of logical reasoning, this grammatical system can't simply be the logic that we use in reasoning about it. (Though there's an old and seductive idea that all linguistic diversity might somehow be just the logical consequence of lexical differences.)
Linguists have been analyzing and trying to model these grammatical systems for about 2,500 years now, and the best models are pretty good. But it remains true, as Horace told us, that even the best grammatical models must defer to usus / quem penes arbitrium est et ius et norma loquendi ("custom, in whose power is the decision and right and standard of language").
This is a sadly undemocratic fact, since it means that no reliable short-cut, in the form of a short list of simple rules, is available to those trying to master a standard language. There's no real alternative to doing a lot of reading and listening.
[Of course, some people are just confused, even about phrase-sized patterns; and in those cases, we happily (and correctly) prescribe a dose of logic. ]