According to the research (and I trust that the speaker did a good job looking at all the research), it does no good (and may do harm) for an instructor to mark or correct grammar in student papers. The research looked at lots of different strategies: code and look it up, check and have the student correct it, and on and on. And none of it, NONE OF IT, actually worked to get students to write better or more grammatically. None of it.I found this research conclusion depressing and, with all due respect to Bardiac and the researchers, counterintuitive. How do they "get it" over time if you can't tell them about errors? Asking them to clarify is great, and I totally agree with that part--but how can they ever get better if they don't know something is wrong?
As I said in the comments over there, this may be true for some or even the majority of students, but there's a motivated minority (or maybe majority) that wants to know and wants to get better. They're the ones who say "but how can I correct this?" or "how can I change it so it sounds better?" or even "thanks--no one has ever told me that that was a problem before." I've heard all these from students, and I'll bet you have, too. Heck, I've said them all when I was a student, and I'll bet you have, too.
I read a book about medical school by Perri Klass one time years ago, and she talked about how medical students learn to do some simple procedures, like stitching up a minor wound. The phrase she used was that the students were supposed to "watch one, do one, teach one"--the idea being that once they'd done the first two parts, they could teach it to someone and learn by doing the teaching. They didn't get a chance to say "nothing works" or "I can't get this" or "why do we have to learn this anyway?" or "you're ruining my creative spark with all your nattering on about complete sentences"; they just had to do it. Of course, they were a self-selected, motivated group--they were in medical school, after all--but still, couldn't something like this be tried?
And one piece of anecdata out of several I could mention: last year, I had a student who wrote convoluted sentences to say obvious things. We spent a lot of time together going over ze's drafts before ze turned them in, discussing individual sentences, and so on. Those meetings are like interviews: you're interviewing students about what they are saying, and the words that come out of their mouths are often an awesomely clear version of what's not written on the paper. So we work with that, and then we look at another draft, and so on. And the writing gets better. That's the part that everyone, even the researcher, agrees works, it seems: one-on-one time with students talking about their writing. Ze also improved in terms of the comments given in our class's draft workshops--the "teach one" part of the equation.
But if I hadn't said anything about the student's sentences, how would ze have known? And if ze had not been persistent and motivated, would hir writing have changed? I'd say persistent, motivated student + discussions about drafts + indicating that there are fixable problems on the draft = the possibility of writing success.