It's like, so last decade.

Many years ago I wrote a seminar paper that I never published on the uses of non-canonical like. You know, it's like the like that's used by teenage girls. I don't know why I never published it, because what I found in my corpus study was not uninteresting: non-canonical like turned out not to be an unconstrained hedge, as I (and many others) had thought before. No, it was a far cry from ums and uhs and their discourse cousins. What I found in the corpus study of hundreds of thousands of words was that non-canonical like was actually pretty tightly syntactically constrained. There were three contexts where it occurred:

1. Sentence initially: Like, what's going on here?
2. After a verb, especially a copula: He's like, such a poser.
3. After a preposition: Give me a pound of, like, black forest ham.

I found thousands of tokens of non-canonical like and they all fell into one of these three categories, with category (2) accounting for by far the largest number (~75%) of tokens. And after I wrote that paper, I spent the next several months listening for further tokens in the speech around me, and sure enough, I found no counterexamples. Use of non-canonical like indeed seemed to obey the constraints outlined above. Of course, as with most similar items in discourse studies, the conditions set out in the study were necessary but not sufficient to trigger the use of non-canonical like; there were plenty of cases where one of these conditions was satisfied but no like occurred. (Cassie Creswell spent a lot of time talking about how to handle this kind of necessary-but-not-sufficient discourse problem in natural language generation in her dissertation. It's worth looking at if you're interested in this kind of thing, and who isn't?)

Anyway, I found myself thinking about non-canonical like again a few months ago, when I overheard the following conversation about an upcoming sale:

RH: Who told you about it? Do you know when it starts?
KF: My mother's, like, mail guy.

This was clearly a different case than the three I'd found in my initial corpus study. I haven't had time to do a detailed follow-up, but I've been paying more attention since overhearing that snippet, and I've started noticing a proliferation of other apparent counterexamples as well:

1. My mom like doesn't even understand what it's about.
2. She like can't get it through her head that he's never coming back.

So either I missed some cases in what I thought was an exhaustive corpus study in 1997, or the categories of use have expanded since that time. If I have a little extra time, maybe I'll take another look at a corpus study and see about publishing the results if I find anything interesting.

Either way, I'm not sure who else is looking at this right now, but this seems like a topic ripe for (re?)visiting by some interested student of the syntax-pragmatics interface. :)