Lee-Ann,
I’ve been thinking about this question of “understanding the rules” where grammar is concerned in the context of philosophy of language. Chomsky early took the view that speakers have “tacit knowledge” of the grammar of their languages. According to some, this ran afoul of analytical philosophy’s account of “knowledge” (on the “Standard Analysis”), which is that one can be said to have knowledge if she has a justified, true, belief. I suspect that folks can vary their language use with no conscious knowledge of how they are altering their registers. I’m thinking, for example, of Labov’s research showing how folks in New York were more or less likely to pronounce word-terminal ‘r’s depending on whether the circumstances were more or less formal. These speakers altered their practices, I’m guessing with little or no conscious effort or knowledge. I suspect that writers can vary registers without conscious knowledge, too.

But, it’s not clear to me whether a listener (or reader in written discourse) will know whether the speaker (or writer) knows the rules she is breaking or not. Thus, the ethos question you raised is still more complicated, I think, than you suggested. The writer who wants to ‘break the rules’ in written academic discourse needs to know the rules AND needs to let the readers know she knows them. Williams addresses this in his article by ending it with a challenge to the reader: “find the 100 or so grammar errors I’ve intentionally made here.” There was really no practical way, though, for the reader to know whether a given mistake is one Williams intended. It’s a brilliant move, I think 😉
-Brian