November 15, 2004

Seems like, go, all

Email arrived this morning from Maryellen MacDonald, responding to last week's post on seems like.

I was thinking that your novel seems like examples must be related to quotative like as I was reading your post (though I didn't know that this is what they were called until you told me). Then I got thinking about the other quotative tems go and all, as in:

And then I go, "Get out!"
And he was all, "Whoa!"

My brief thinking about this led me to believe that be+all and be+like are pretty interchangeable as quotatives, but not in the "seems" constructions. "All" admits a very narrow usage with seems (as in He seems all anxious and stuff.), basically meaning "completely," but it's never had any of the usages that you discuss for "seems like". Thus the new usages for "seems like" that you note seem to stem from the combination of a) the previous "seems like" usages and b) the quotative usage. The quotative usage by itself isn't enough, otherwise you'd see things like: He seems all I really have a case.
The closest I found (though not looking exhaustively) was this theory of mind piece, though not quite the right usage:

one minute he seems all I like you and the next I'm just out of the picture.

I realize you were suggesting that the quotative usage of "like" was allowing its extension out of preexisting "seems like" constructions. I'm like all agreeing and stuff, I guess.

Part of what's going on here, I think, is that there are two possible sources for "like" in such cases. We noted this ambiguity in connection with God's late 2003 revelation to Pat Robertson about the outcome of the recent election.

One kind of like is used to introduce clauses or noun phrases, roughly in the way that words like after do:

He ran like the devil was chasing him.
He ran like a deer.

while the other kind of like is a particle that can be inserted almost anywhere in any phrase, without modifying the syntactic relationships of the words around it:

"... her and her, like, five buddies did, like, paint their hair a really fake-looking, like, purple color."

[Example from Muffy Siegel's paper "Like: The Discourse Particle and Semantics" (J. of Semantics 19(1), Feb. 2002)].

I suspect that a careful study of the phenomena (which I have certainly not made) would reveal that these alternative construals are playing a role in the on-going changes in various uses of like, including quotative "be like". As Maryellen suggests, it's also likely that the same ambiguity is playing a role in changes in some related constructions in which like does not always appear. If you analyze "he seems like S" as a case of like-the-particle, then "he seems S" ought to be an alternative. Since this is a perfectly sensible English syntactic frame (e.g. "he said S"), why not go with it? And "he seems S" would logically be an alternative form of "he seems that S". As I explained in last Thursday's post, I haven't gone down this road myself, but I can see how someone might.

One of the interesting questions here is how these alternatives work psychologically. For someone (like me) who uses both kinds of like, does the planning and execution of a given example necessarily fall into one category or the other, or am I sometimes (or always) in a mixed state, kind of like electrons in the Copenhagen interpretation of quantum mechanics?

[Update: Maryellen responded:

The dominant view of language production is that lexical items can be in your "mixed" state, that is with several alternatives partially activated and competing for selection in the utterance, but possible syntactic structures are never in a state of partial activation--speakers are thought to develop only one as a function of the lexical items that have become activated (among other things). Thus on that view, the answer to your question depends on whether you're referring to like the word or the various syntactic structures it might appear in.

I don't happen to accept this syntax/lexical dichotomy and think that there's good evidence for partial activation of both alternative structures and words during production, but since you asked a psycholinguistic question, I wanted to be sure to include the mainstream view.

I agree with her -- and it seems to me that the "mainstream" view is not entirely coherent. In nearly all cases, different lexical items carry with them different syntactic as well as semantic structures. So how can alternative lexical options be simultaneously activated without alternative structural options also being in play? You could invent a model in which collateral inhibition operates strongly among structures but not among words, but this would imply that structures are never intrinsically lexicalized. Which seems wrong.

(Apologies to general readers for the the allusive "inside baseball" musings here, and to psycholinguists for unlicensed and ill-informed encroachment on their territory... As always, the Language Log Marketing Department stands ready to refund your subscription fees, cheerfully and in full.)]


Posted by Mark Liberman at November 15, 2004 08:27 AM