Jonathan Owen quotes a passage from The Chicago Manual of Style on the word "nauseous":
The use of nauseous to mean nauseated may be too common to be called error anymore, but strictly speaking it is poor usage.
The first time I came across the idea that "nauseous" means "nauseating" and not "nauseated", I was extremely puzzled, as I was sure that I had never heard "nauseous" used to mean anything but "nauseated." I was born in 1964, so I bet that the "nauseated" meaning had become dominant by the 1970s. (Anyone want to do the corpus work to test that?)
But that's not what interests me here, nor is what interested Owen. As he puts it, "the truly strange assumption is that words have meaning that is somehow independent of their usage." To see this, consider the implications of the sentence if you generalize it:
The use of X to mean Y may be too common to be called error anymore, but strictly speaking it is poor usage.
This is clearly utterly absurd: if this pattern were generally true, then every new meaning Y that develops for any given word X would be "poor usage" even when it had become completely common—and even, perhaps, when an older meaning for X had long since disappeared.
Such an understanding of language completely ignores how language actually works as it develops over time: some old words disappear; some old words develop new meanings; new words are coined; some of those coinages survive; other coinages disappear. And how do we tell what a word means? By looking at how people use the word: when they use X to mean Y, then X means Y, even if it once meant Z. And it is even possible for X to mean Y and Z at the same time.
But not with "nauseous," at least not for me. The people who insist that "nauseous" should only mean "nauseating" totally contradict my linguistic experience—and I doubt that it is just a matter of my idiolect.