January 02, 2004

The theology of phonology

In a previous post, I quoted a note from Robert Beard in which he came out four-square as a language moralist, and identified what is "proper" and "right" in pronunciation with what is "consistent".

Now, I'm in favor of language standards. Many of my colleagues consider me dangerously right-wing on this question. However, I think it's unwise to use ethical metaphors to justify arbitrary cultural norms. If a man should wear a necktie in court, it's not because there is something intrinsically immoral about an open collar.

In contrast, Prof. Beard argues eloquently that the standards he is defending are not arbitrary social conventions, but rather consequences of basic linguistic principles. In particular, he suggests that "proper" pronunciation is not a matter of how well-spoken people talk, but rather a question of what is "consistent", by which he means something like "characterized by regularity in the relations among the forms, sounds, and meanings of words". Alas, if morality requires consistency in this sense, then we are all deep-dyed linguistic sinners, every one of us.

Here's what he wrote:

We do not consider language a democratic process here at yourDictionary. So, even if the majority of US citizens pronounce "nucleus" [nyu-klee-us] and "nuclear" [nyu-ku-lar], it doesn't make it phonologically right, which we take to mean simply "consistent." Generally, we simply point out the inconsistency and tell our visitors they may be consistent or talk like the folks around them, whichever pleases them.

One of our most popular projects on our website is out "100 Most Often Mispronounced Words" which include both "nuclear" and "jewelry." It is popular because the educated people who visit our site are convinced that there are proper and improper ways to pronounce words and they, by and large, prefer the former.

There are plausible arguments for enforcing consistency in syntax, though it can be tricky to decide what the principles should be. In semantics, the truth should certainly be something we can calculate without taking a poll. But in morphophonemics -- the relationship between the form and sound of words -- the idea that standards are determined by fundamental laws is a surprising one. To see why, let's take a simple example from standard English.

The plural of loaf is loaves, the plural of thief is thieves. However, the plural of oaf is not oaves, and the plural of chief is not chieves.

Quite a few words ending in /f/ work like loaf, voicing the final /f/ in the plural: calf, dwarf, half, hoof, knife, leaf, life, loaf, scarf, self, sheaf, shelf, thief, wolf.

A somewhat larger number of words work like oaf, letting the final /f/ stand unchanged in the plural: belief, chief, clef, cliff, coif, cuff, gaff, goof, handkerchief, kerf, midriff, muff, oaf, pontiff, proof, puff, reef, relief, riff, ruff, sheriff, skiff, sniff, snuff, standoff, stiff, tariff, tiff, whiff. As far as I know, all words where final /f/ is spelled /ph/ or /gh/ also fail to voice the final consonant of the stem in the plural: epitaph, glyph, graph, morph, nymph, seraph, sylph, triumph, etc.; and cough, laugh, rough, tough, trough.

Some words are variable: in my speech, hoof, roof, beef, turf, and wharf sometimes pluralize thiefishly and sometimes chiefishly. In a few cases, it depends on what you mean. If a staff is a stick, its plural is staves, but if a staff is a set of employees, its plural is staffs.

Any fair-minded observer will agree, I think, that we have here an inconsistent relationship between word structure and word pronunciation. But is this a moral problem? Should you call for Sancho Panza, mount Rocinante and ride off to restore consistency to the plural of English nouns ending in /f/?

Well, I don't see any volunteers, at yourDictionary or elsewhere. The most obvious reason is that this sort of partial inconsistency -- what Mark Seidenberg calls quasi-regularity -- is ubiquitous in English and in every other language. Enforcing regularity in morphophonemics is like trying to clean sand off the beach:

The Walrus and the Carpenter
Were walking close at hand;
They wept like anything to see
Such quantities of sand:
"If this were only cleared away,"
They said, "it would be grand!"

"If seven maids with seven mops
Swept it for half a year.
Do you suppose," the Walrus said,
"That they could get it clear?"
"I doubt it," said the Carpenter,
And shed a bitter tear.

There is an interesting and important controversy among psycholinguists about where this quasi-regularity comes from and what it means. James McClelland, Mark Seidenberg and others think that quasi-regularity arises because the (partial) regularities are emergent properties of connectionist networks; Steven Pinker, Michael Ullman and others think that quasi-regularity arises because there are two distinct and competing brain mechanisms whose functions overlap, one a (temporal/parietal-lobe) semantic memory system for looking things up, and the other a (frontal-lobe and basal ganglion) procedural memory system for figuring things out.

In both theories, human speech is east of morphophonemic Eden. There are forces leading to regularization and forces leading to exceptionality. If you think that consistency is next to godliness, both theories -- like the facts of language -- force you to confront phonological original sin. And with respect to the morality of inconsistent pronunciation, let him who is without sin cast the first stone.

Posted by Mark Liberman at January 2, 2004 08:56 AM