One of my biggest pet peeves in scholarly prose is the habit of advancing an argument by citation of the dictionary definition of one or more of your key terms.
Is this really all that analytically valid, or just rhetorically decorative, maybe even a little fetishistic? Maybe even a little insecure? It's a non-structural buttress in almost all cases, present merely for the appearance of greater support.
Could it be a hold-over from the older, more philological orientation of literary studies? I'm inclined to think not because it is used so rarely in that mode. One of the few adequate uses of this rhetorical device I've seen recently came in Jenny Davidson's Breeding, where she actually traces the word through a succession of dictionaries. Diachronic uses (reminding the reader of an archaic definition would be another good use) are entirely reasonable, but the synchronic is just word-dressing.
I think that the O.E.D. (it's almost always the O.E.D. that is used) is quite simply a sort of name-check authoritative reference that gets dropped in rather like a flavoring particle in German, identical in function to the frequent footnoted references you see to big-named figures that take the form "Foucault makes a similar point in regards to the panopticon…" These aren't so much ways of building up the argument as showing that your argument is neighborly with someone important. This is talismanic, not analytical.
Because really, how often is the precise articulation of the denotative essence of a word a revelation? Aren't we as readers usually capable of evaluating whether a scholar is using her terms in a manner consonant with standard definitions? If there is a specific aspect of the word's definition which needs highlighting, can't the scholar simply define her own terms, and then we as readers can figure out if she's using it legitimately? If the term is itself so vexed that an O.E.D. citation is "required" to pin it down in univocal terms, maybe the scholar should actually talk through that complexity rather than pre-empting it by citation.
Actually, I think it's just a case of scholars being too tentative, feeling like they can't begin working with specific terms unless they're drawing them from somewhere else—they need someone else to say all the words they want to work with before they can begin working with them. You come up with a set of terms that you want to play with throughout your article or your monograph, and you need some way to introduce them, and you feel awkward just saying "Here's what I'm going to do with 'empire'" and letting the reader decide if you're making sense.
Sorry, I was just flipping through an introduction to a book that I'm really eager to read, and the author feels it necessary to use the O.E.D. to define two of the words in the title in order to draw out an opposition or a paradox that could simply have been stated flatly. As I said, it's a pet peeve—I don't disagree with the definitions, or the paradox the author foregrounds between them; I just don't find it necessary to wield the O.E.D. in such a way, and a little insulting to both of our intellects.