Further to KB's post below, the reason sometimes given for politicians (and others, of course) to turn nuclear to nucular - from Eisenhower to Palin (we think Jimmy Carter had a variant pronunciation, too, though different than "nucular") - is to sound down home, the way that Palin likes to drop her "g"s. It's a keeping-it-real kind of thing.
However, you knew there had to be a field of study about this "issue," and that it would address nucular. William Safire, who was the New York Times' language maven before his retirement, addresses how cognitive linguistics handles "nucular":
The Harvard cognitive scientist Steven Pinker informs me that ''a person mishears a sonorant consonant (like l or r or y) followed by a vowel as a vowel followed by a consonant. This can happen because the two sounds are acoustically similar -- they literally look alike on an oscilloscope. Presumably the respective neural patterns that arise when a person hears them are similar as well.''
Well, okay. There's more, but we think we might stick with the keeping-it-real argument, perhaps mixed with the cowboy way of finding convenient ways to slide easily over syllables - pardner instead of partner, for example - which does align, we think, with Pinker's formulation.
How to fix it?
The way to straighten out your mental dictionary, if you have this ''nukular'' problem, is to train your brain to think of the word not as three syllables but as two words: new and clear.
We like that. New clear - sounds like an advertising campaign waiting to happen. In sum, we're quite all right with nucular - the dictionaries give it play, and it continues to make English a dynamic language, determined by its speakers.
First balloon: "The nucular renaissance promises an emission free, low cost energy source for all." Second balloon: "if he doesn't stop saying nucular, he's gonna get such a pinch." Third balloon: "Ow! Why'd you pinch me!?"