> > Basically, it's this: because rules of grammar
> >(at least to an extent) reflect usage...
>
> To what extent are grammar rules anything *but* descriptive, as opposed to prescriptive?

I'm generally dubious of both sides of arguments about whether some X reflects some Y, or Y reflects X. What I was thinking of when I wrote this was an entirely different issue: does violence in movies and video games reflect violence in life, or does violence in life reflect violence in movies and video games? One outspoken camp of people, most of whom are either morally offended by violence in the media or simply disgusted by it, insists that it causes violence in real life. Another outspoken camp, most of whom are fans of visceral movies and video games, insist that the violence in them is merely a reflection of violence in real life. Both camps reek of personal agenda. Not that that makes them both wrong, but I don't buy either argument. I think it's more realistic to suggest that they each reflect each other. Violence in the media is a reflection of violence in life, but violence in the media nevertheless desensitizes people to real violence, which I believe leads to more real life violence than gamers would like to admit, though less than alarmists like to charge. Of course I'm not implying it's an even 50-50 exchange.

Likewise, I suspect grammar describes more than it prescribes, but isn't it kind of simplistic to say that grammar only serves in one role or the other? If nothing else, the learning of grammar slows change to the way language is used, which is itself a prescription of sorts. Just look at the difference in the way scholars and drop-outs talk for an example. With the usual exceptions on both sides, people who have been taught grammar (whether or not their adherence to the rules after the fact is conscious) speak very differently from those who have not. I know I use language very differently as the result of learning grammatical rules and attempting (consciously or subconsciously) to adhere to them.

But it goes beyond slowing change. Even though I ultimately concede yours and Stephen's argument that the foundation for the split infinitive rule is shaky at best, and that many language scholars refuse to adopt it, who can deny that it hasn't had a prescriptive effect on the way a lot of people talk? Ok, maybe most ignore it, but there are a lot that don't -- more than those who ignore the rule realize, I think. I personally reject it as a rule, and yet the fact that it exists still makes me conscientious about using split infinitives and leads me to concede that, rule or not, split infinitives are awkward and generally best avoided. That's the prescriptive effect a rule of grammar most people don't even accept has had! Consider the larger effect of a rule with a more convincing and accepted basis.

So I think it's a two-way street. More description, but some prescription, too.