One of the essays in Consider the Lobster addresses exactly this sort of question. It's a review-essay on A Dictionary of Modern American Usage by Brian Garner and you can find it here. (Don't forget to read the footnotes -- they're the best part). "Tense Present: Democracy, English and the Wars Over Usage" is pretty entertaining reading for people like us, because it makes fun of both sides in our discipline's debate over the place of grammar in the literature classroom, that is, of both Prescriptivists who believe that grammar rules are transcendently and universally correct, and those Descriptivists who believe that the rules for making meaning with words are essentially innate, and that everything else is governed by social context. Wallace tries to carve out a position somewhere between the two, in the process providing an introduction to terms and issues that is comprehensible to the proverbial educated lay reader, (and probably anathema to linguists). I recommend it to you, not because I agree with everything it says, but because it's funny, and because it provides a common vocabulary for us to discuss marking grammar errors in upper level course assignments.
Now I know that Arnie is a long time-Prescriptivist, especially on the subject of commas, and probably Kim Blank too, while I am (perhaps) a moderate Descriptivist, and have definitely found myself delivering the same speech to students that Wallace describes in the latter part of the article. I know that in the past some of our first-year instructors have flunked papers for one comma splice, or one misplaced modifier, even though our colleagues in the sciences regularly published peer-reviewed papers with these ubiquitous errors completely unremarked. What's the value of correcting a student's split infinitive, when professional writers commit them all the time and still make money? Are Susan Doyle's copy-editing students taught to correct all the grammar errors in their subject-text, or merely to correct to the appropriate usage level for a particular publication? And is there a universal standard of "correct" written English, or merely, as Wallace says, a provisional one based on audience?