Mr. Verb isn't working much these days, but a bunch of the Madison crew -- including several members of Team Verb -- are up here in beautiful Banff at the Germanic Linguistics Annual Conference, meeting with the Forum for the Study of Germanic Languages (our British sister organization) and SHEL (Studies in the History of the English Language).
I just heard a talk about usage by Don Chapman called "Lost battles and the wrong end of the canon: Attrition among usage prescriptions". He started by noting that linguists normally simply dismiss the usage canon as crazy, 'screwball', 'unprincipled', and consisting of 'undemonstrated' stuff. He figures, though, that this is a subject for linguistic study. So, he then asks this: How much does the canon change over time? Interesting question.
Half of the prescriptions, he finds, are 'one-offs' -- found once and not again. I'm more surprised to see that huge numbers of old prescriptions have changes in the last decade or so: Chapman finds that the canon is very volatile, with less than a quarter of those proposed in manuals being maintained over time. In fact, two of the biggest manuals (Garner and Peters) show lots of change. (The ppt went by too fast to get the numbers down systematically.)
That is, while prescriptivists often see themselves as fighting a 'lost battle', the target is constantly changing. And the prescribers seem to assume a static canon and we linguists too ignore change in the prescriptions. Chapman argues that prescribers are starting to justify what they include. He's actually optimistic, concluding that this could lead to a better fit between usage and advice.