Noncompositional

Multi-level coordination is when you get something like this:

She has been around the world, climbed the tallest mountains, but won’t eat a simple sea cucumber.

Where the first two coordinated verb phrases are inflected to be complements of have (has been, has climbed) while the final one is not (*has won’t eat). For many examples and discussion thereof, you need look no further than here.

Just the other day I happened (by complete accident I assure you) to have my set-top box set to the Disney Channel (by no, or at least little, fault of my own) when a children’s show called Bunnytown started. The first thing I heard was some bunny addressing TV land, saying something like (IIRC):

Are you tired, sick, or have the flu?

You know, this sort of stuff is what the TV ratings system was designed for: TV-7:L-NCC (that’s “7 and over, due to language - non-canonical coordination). Parents need to be able make informed choices about the language they expose their kids to, and as a non-parent, I should think that some kids just aren’t ready for, prepared to deal with, or have the linguistic self-awareness to understand, complex coordinate structures, let alone from cute bunnies.

Geoff Pullum’s newspaper headline noun compound interpretation difficulty post earlier today reminded me of Japanese. Why? Because sometimes the Japanese love really long noun compounds. In fact, it may be that nothing strikes more fear into the heart of a learner of Japanese than seeing those huge Chinese-character compounds. (okay, maybe the million or so pronunciations of a single character ranks up there too). I found that the Ministry of Health, Labor and Welfare is one of the, shall we say, best, users of noun compounds:

非加熱血液凝固因子製剤納入先医療機関名

[pre un-heated blood congealment factor production delivery] medical-treatment organization name

…I think. I’m not sure about the 先 ‘pre’. And that’s not splitting up the N-N compounds like “blood” = “blood liquid” and “congealment” = “congeal lumpify”

I’ve heard some Japanese speakers look at such compounds and say that it’s like it’s Chinese. Which is true to a certain extent: certainly Chinese text only has Chinese characters (maybe some numerals here and there), making it daunting to a reader of Japanese who is used to some kana every now and then to demarcate major phrasal boundaries. But these long compounds are not like normal Chinese texts, because they are just huge compounds, not clauses. The trained (or native) reader of Chinese, I assume, has no trouble sighting the relevant “grammatical” words just as a reader of Japanese reader would. But I will say that, at least for me, reading mostly Japanese over a period of time and then switching to Chinese requires a bit of a transition period before I can get used to seeing nothing but one script.

Wow. Garkov has to be the funniest example I’ve seen in the domain of replacing the content of speech bubbles in comics. And in this case it has the added virtue of being funnier than what used to be there. In almost every case.

The combination of the subject line and first few main text lines of a bit of spam I got this morning was, in fact, slightly coherent.

Subject: Or coherent

Body: increasingly so the longer I m here and the more time I spend at

C A 9N A D/8AN P 0 4H A RM A 5CY

I prefer a colon after coherent and a comma after increasingly so

If you’re a linguist and you haven’t checked out the squibs/essays for the Jorge Hankamer WebFest, I recommend you do so immediately.

One of the squibs is by Language Logger Geoff Pullum, entitled Hankamer Was! It is devoted to exploring the possiblity that verb phrase ellipsis (VPE) in English can be used in cases where what is elided does not have a correspondent in the linguistic context. The original claim, from a paper by Jorge Hankamer and Ivan Sag, is that there is no such possibility. In their made-up context, you have two stage performers, one who appears to be about to chop off his own hand. The other performer cannot say, without any prior linguistic context, *Don’t worry, he never has before, where the meaning is supposed to be “he never has chopped off his own hand” or something like that.

Pullum goes through all the putative examples of verb phrase ellipsis that have been argued to be possible even without a linguistic antecent. He finds only about a dozen, and argues that they are all fixed expressions, and further, involve non-declarative clause types: Don’t!, May I?, and so on. He then issues a challenge: send in your examples of ellipsis without a linguistic antecedent. He’s willing to bet that none will be. (ahem). Let this post be such an entry.

One day last week, while walking home from work with my girlfriend, an interaction like this happened. I’m N, my girlfriend is G.

N and G are walking down the street, and pass a woman walking the other way wearing a half sweater. N points it out.

N: What do you think of those?

G: Those half sweaters? They seem seem silly to me.

N: Yeah. I know someone who has one…

G: I used to want one for a long time, but I didn’t.

At this point I was somewhat in shock that this seems to be a case of VPE with no linguistic antecedent. Unlcear how to get the miraculous thing to happen again, I just said (something like):

N: Uh huh?

G: …

But nothing like the VPE-containing utterance was forthcoming, so like a normal person I reoriented to the content, not the form, of what was being said.

At the outset let me say that I basically agree with Hankamer and Sag (and Pullum) that VPE requires a linguistic antecedent.* There is at least one way to interpret G’s utterance that does not involve VPE: namely that she abandoned the utterance before completing it, and it would have ended with a full verb phrase like I didn’t end up buying one. There was quite a pause between her putative VPE and my “uh huh” — I was waiting to see if anything would happen (”completing” or repairing the sentence, the sky falling, etc.), but nothing did. Note that the abandonment (if that’s what it was) would almost assuredly have been “internally” motivated: G was not eating or drinking anything, or having difficulty breathing, nor did anything absolutely astounding pop into the air in front of us.

I will also note that there is almost a linguistically-provided semantic antecedent in want. That is, wanting something means wanting to have or get something. Some people interpret this as an indication that in the syntactic structure there is an unpronounced HAVE element; others take a more semantically-oriented approach (say, metonymy) or lexical approach (it’s just another sense of want). This is close, but clearly what G didn’t do was buy a sweater, not have one (*I used to want one for a while, but I didn’t (ever) have one).

I unfortunately cannot meet the requirement Pullum set out, namely that two independent observers be there at the time. I can however be rather sure of all the preceding context, at least in that so far as I recall there was no mention of buying anything since we met up at the train station to walk home, and perhaps even that entire day.

In any case, this is my humble submission to the inventory of VPE without a linguistic antecent.

*I “basically” agree because there’s always a bit of me that wonders if the restriction is actually something like “VPE must have an antecedent that is salient in XYZ way” where XYZ always happens when something is introduced by linguistic means, and nearly never by anything else. But I admit I have no way of making this non-circular.