N recent weeks, throughout the public discussion of Iraq's weapons capabilities, Americans have heard the word "nuclear" pronounced "NOO-kyuh-luhr" by the president, many members of Congress, a general or two and even a few news anchors.

Of the many language controversies that arouse passions, no other ó not "hopefully," not the split infinitive, not "most unique" ó seems to bother people as much as this. Even though this pronunciation is now included as a variant in all major American dictionaries, a usage panel convened for the "Harper Dictionary of Contemporary Usage" rejected it by a factor of 99 to 1. Steve Kleinedler, the pronunciation editor of the "American Heritage Dictionary," said complaints about this variant are the most frequent comment he gets. Merriam-Webster editors have written a special form letter to respond to those who write in to criticize the inclusion of this pronunciation.

Yet the use of "NOO-kyuh-luhr" is not uncommon, even among prominent and educated people, including four of the nation's last 10 presidents: Dwight D. Eisenhower, who was widely admonished for it; Gerald R. Ford; Jimmy Carter, who did graduate work in nuclear physics, and also used "NOO-kee-yer"; and now George W. Bush.

There are two main explanations for the altered pronunciation. The process of swapping two sounds is common. An old example is the word "third," which used to be "thrid" (like "three"). The pronunciation as "third," with the "r" and the vowel transposed, didn't become standard until the 16th century. A modern example is the word "comfortable"; its usual pronunciation, "CUMF-ter-buhl," doesn't seem to bother anyone, though the "r" and "t" are transposed.

Linguists call this process metathesis, and through it even highly educated people can unknowingly create pronunciations like "INT-ruh-gul" for "integral" or "ir-REV-uh-lent" for "irrelevant." However, metathesis ó in the case of "nuclear," the transposition of the "l" with its following vowel ó still doesn't answer the "why?"

The real reason is the powerful force of analogy. Words ending in a two-syllable "klee-er" are extremely rare; besides nuclear, the only marginally common such word is cochlear. But a "kyuh-luhr" pronunciation is found in many common words: spectacular, particular, muscular, circular, molecular and others.

Analogy is responsible for a number of linguistic forms, from spelling ("miniscule" for "minuscule," by analogy with "mini-") to pronunciation ("et cetera" becomes the metathetic "EK-set-er-uh," for the many common words with "ex" overwhelm the one beginning with "ets") to grammar (the plural of book should be beech, not books ó just as goose becomes geese ó but for the influence of the mass of words forming their plural with "s").

SO why is "NOO-kyuh-luhr" so particularly irritating? It's hardly a new pronunciation ó the "Oxford English Dictionary's" first example is from 1943, or around the time that nuclear was becoming a familiar word to nonscientists. A driving reason is that it's so common among people who hold prominent positions in society. Most people don't get upset by pronunciations like "PO-lice" or such words as "ain't" as long as these common utterances occur only in dialectal use. But when nonstandard forms like these get used in mainstream contexts, they get noticed. And while folksiness can be acceptable in nontechnical contexts, mispronouncing a technical term is rarely considered O.K. It is viewed as a sign of ignorance, not quaintness.

But in the end, language changes because of the people who use it, not by appealing to any external factor, like spelling or history. People criticizing "NOO-kyuh-luhr" should also condemn "CUMF-ter-bull," yet they don't. Merriam-Webster's form letter about "nuclear" spends only two sentences discussing the word itself: most of the letter is an explanation of why spelling is not a valid basis for determining pronunciation. It offers the example of the words electric, electricity and electrical, in which the one letter "c" represents three different sounds.

In other words, "NOO-kyuh-luhr" is a lost cause, and no amount of pleading on the grounds of technical context or historical novelty can expunge its use. But as Enid Pearsons, the former Random House pronunciation editor, once said, "If we're so upset about `NOO-kyuh-luhr' proliferation, we can only hope that technology will come up with a new source of energy that we can all pronounce."