Biking home listening to an old Fresh Air podcast from my backlog, I was amused to hear the story of Frank Sinatra giving a grammaticality judgment. Sammy Cahn describes how Sinatra objected to his lyric for the song "The Last Dance."
Here's the relevant bit of the transcript from the Fresh Air site:
CAHN: So when you speak, you would say they're wondering just when we will leave. You wouldn't say, they're wondering just when will we leave. So he said that one, just when we will leave. I said no, it isn't – hold it. I said they're wondering just when will we leave. But till we leave. He said what kind of cockamamie word is…
CAHN: I said no one speaks like that. I said no. I said no one speaks like that, but we aren't speaking, Frank, are we? We're singing, aren't we, Frank? And that's the only time we ever kind of good-naturedly quarreled about a line.
Apparently Cahn shared the judgment but justified the inversion on artistic grounds. Sinatra subsequently did it Cahn's way, not his way.
I also love Cahn's bisyllabic pronunciation of "aren't" here.
Postscript: Of course, embedded subject-aux inversion IS available in many dialects of English, unbeknownst to Sinatra or Cahn, and also to many generative syntacticians until the early 1990s. Rusty Barrett once gave a talk * in which he described trying to report the existence of embedded auxiliary inversion in his hometown's nonstandard dialect in his graduate syntax class — this after the instructor had (proudly) shown that the theory he was teaching entailed that embedded auxiliary inversion was impossible (good for accounting for Sinatra's English, e.g.). The instructor's (jocular) response to Rusty's data point was, "Well, you clearly aren't an English speaker." Possibly intended as a hyperbolic joke, the actual effect of the response was to exclude Rusty and his nonstandard grammar from participation in the development of the theory being constructed in class.
Of course shortly after that, embedded subject auxiliary inversion was discovered by the syntactic literature (in Hiberno English and Belfast English) and shown to have important theoretical implications. It also appears to be on the rise in informal standard American English, judging from my undergraduates' feedback. It COULD have been Rusty who brought those facts to the world's attention. But thanks to a type of cultural insensitivity peculiar to one flavor of linguistic subculture, and, worse, scientific incuriosity, it wasn't.
I've heard similar jokes, and even, to my shame, occasionally made them in my own classes. Strange and wonderful data that your theory can't account for can be uncomfortable to contemplate. All the more reason to treat it with respect, not denial. Correct response? Interest and gratitude.
*Rusty's talk was given at the 2008 Arizona Linguistics and Anthropology Symposium, and was fittingly titled, "Is it any way might you could tell me how come am I not an English speaker? Subject-auxiliary inversion and introspective methodologies in linguistics and anthropology." The handout for the talk linked to above is online, I have discovered, here. He gives an amazing set of data documenting the epistemic content of embedded aux inversion in Ozark English.