Noncompositional

There is a great literature out there on syntactic blends, most of which I’m frightfully unaware of (but see this LL post by Mark Liberman, the article he cites by Dwight Bolinger (a good and quick ready), and some recent work [like this piece] by Liz Coppock at “that school down the peninsula”). I suppose one of the most infamous of these (or at least the one that has caught some attention recently) is the double-is or is-is construction, as in

The thing is is that this only happens when unplugged.

which (some have said) blends “what (the thing / it) is is that…” and “the thing is that….”

Well, I noticed something searching on the ol’ internet the other day, and I’m not quite sure how to categorize it. It starts with two English idioms: (be) more like it (MLI) and (be) what I’m talking about (WITA). They are similar in meaning: they say that something (an item, an event, a situation) is very close to some ideal, perhaps closer than might have been expected, or closer than previous versions of it (whatever “it” is) had been. Nearly always, they are anchored with respect to “now” (adjusted for viewpoint) and “me.” MLI seems like it might be more flexible with respect to times and points of view than WITA. But back to that later.

Now, blends…well, maybe you can see where this is going. I’ll just give the hits from google:

Read more »

The always-entertaining Geoffrey Pullum has recently remarked on the seeming weirdness of English proper names and definite articles. He notes that though human-referring proper names do not normally appear with determiners, there is a niche of productivity. When the name is modified attributively (like sharp-eyed, attentive-to-language), then a determiner is all required. Except when introducing a character in subject position in which case you can say something like Foolishly brave Qarn Trippian stepped out from the dragon’s cave and triumphantly waved his bloodied hand. (I suspect that some people might also be willing to ignore the formal requirement, as long as you’re introducing a new character, as in and who did I run into but always ridiculously funny Brian Regan. But I could be wrong.)

There are other places in English where the structure determiner-modifier-HumanName can appear, as noted in the LL post. These are cases like the Qarn I knew back in monster-killing camp and the Qarn of Targor, not the Qarn of Pindrell, which are sometimes talked about in terms of conceptual coersion from proper noun to common noun. A more familiar example might be the use of possessives when disambiguating between acquaintances of different people who happen to have the same name (”I saw Jim the other day” — “Oh, the guy in your Onomastics class?” — “No, Julie’s Jim. You know, the one she always has lunch with”)

Read more »

A particular snowclone got my attention a few months ago, so I thought I might as well make public note of it. There are two basic forms that I’ve noticed:

• A play X to B’s Y
• A be X to B’s Y

Examples of the first are, I suspect, rather old, dating back at least as far as to whenever people talked about operating musical instruments or playing dramatic roles “towards” another such performer. The more mundane of these are like this example: a woman plays the bass lines to my guitar riffs… and vice versa. They get more fun when you have metaphoric uses of this same construction. This metaphoric use is also quite old, I think (see examples at the end of the post), but some more modern uses might be called for:

For now it’s winter and the building plays the perfect backdrop to my homage to the tree in this city. All the same, I have the same crippling burden of knowlege and ego to go with it, and scripture constantly plays the hammer to my self-loving nail. So I went home to ask my roommate, he’d know. The poor guy plays the Felix to my Oscar, and OUR show gets weirder each season.

Particularly in the second two examples, what this construction lets you do is give roles to two entities, and then the names of the roles evoke a larger situation where the roles interact in commonly understood ways. Hammer on nail, Felix on Oscar, and so on. (The first example is a more straightforward extension, not relying on any additional cultural frames aside from just “having special roles to be (metaphorically) played out.)

Things get much more interesting when you get rid of play and just allow be. Examples: Read more »

Last night I caught part of CNN’s special 300 Million: Melting Pot or Meltdown, on the booming population of our fair nation. You can look at summary and review of the show, since I can’t seem to find anything about the program on CNN other than this blog entry.

While the content is of course noteworthy, what got my attention was something that the host, Anderson Cooper, said near the beginning of the broadcast. From the transcript:

Right now, we are less than 12 hours away from a milestone with enormous implications. At 7:46 a.m., Eastern Time in the United States, the U.S. population is expected to hit 300 million. 300 million. We have reached this milestone faster than ever before. It took the nation 139 years to get to 100 million mark in 1915. And another 52 years to reach 200 million in 1967. But it will have taken just 39 years to hit the 300 million mark, which means that the country is getting bigger and much bigger much faster.

Note the part that I made bold. We’ve got a comparison being made by the word faster. The item being compared is the (speed) of reaching the milestone of 3300 million people. And the standard of comparison: well, it’s ever before. Now, when you put just before, you can omit the material that might describe what actually happened before. Compare this to Nowadays, threats spread further and faster than ever before. What happened before was “threats spread (far/fast to some degree)”. What’s important in making this sentence work is (i) that we’re talking about generic threats, so there doesn’t necessarily have to be some particular threat that spread both before and now, and (ii), there actually could be such a threat, since threats can spread continuously. A slightly different example would be His apartment is bigger than before. Here there’s not a generic assignment, but a role. That is, there are several incarnations of the role apartment in “his” life (e.g., the one on Broadway, the one on Market, the one back East), and the current incarnation is bigger than the previous role-fillers were. (On another reading, the same physical apartment was made bigger, by annexing neighbouring rooms, for instance).

But to return to Anderson’s sentence, we’ve got a problem. We can’t interpret what we have done before is “reach this milestone,” because (i) there is no generic “this milestone” (since it really means “3300 million,” which is a particular value), (ii) there is no role “this milestone”. (Similarly, you can’t say this apartment is bigger than before to mean that it’s bigger than previous incarnations of the apartment role). Nonetheless, what he said is clear enough, especially if you listen to the next few bits. Basically, we’ve gone from milestone-to-milestone faster than ever before, and this particular time, we got to the 3300-million milestone. And, though this may be obvious, you have to make sure that you are comparing equivalent milestone-distances; you can’t, say, look at how long it took to get from 150 to 250 million, and compare it to the time it took to get from 250 to 300 million.

Now, I’m not sure what principle might tell you that you allowed to say something like that. I mean, you can describe it as a sort of “coercion” from entity to role, induced by the before. It would be sort of (barely) similar to the coercion involved in a sentence like I believed her in five minutes, where the believe is coerced into being an event with an endpoint, rather than a state.

[edit: In an extended brain-o, somehow most quantities were divided by 100 in my head when I wrote this. This has been fixed, with reminders to my brain to be more attentive next time.]

Yesterday on one of my favorite shows, Good Eats, I heard an interesting coordinate structure out of the host’s mouth. The topic of the episode was mayonnaise, and Alton was discussing the properties of emulsions:

How can there be so many opposing forces crammed together, and yet still be stable?

Not to infringe on the territory of those interested in coordination, but notice that the second part, after the and yet is non-finite, so this looks to be a coordination of verb phrases or perhaps TPs, if you believe in sanitary syntactic structure (compare it to in one city there can be a lot of student-oriented stores and still be good places for families, or how can there be unity and still be diversity).¹

However, expletive there cannot be followed by an adjective: *there was still (very) stable. So how is the second phrase licensed? The above sentence does sound a bit weird (weird enough for my pay-attention-to-this-construction alarm to go off). But it gets about ten million times worse if you rearrange the conjuncts: *how can there still be stable even when there are so many opposing forces? A sort of off-the-cuff analysis would be to say that this is really coordination of clauses, and that there is a missing subject in the second “clause”:

…and yet the emulsion still be stable?

But that is pretty darned degraded. I did manage to find cases like

He argues that a person could be morally responsible for a particular action and yet it still be true that that person could not have done otherwise

where the subject, actually an expletive it is realized in the right location (for more parallelism, just imagine that the first conjunct is also a there be construction). But then there is the question of why you have non-finite be in the right conjunct: if they are both top-level (utterance-level) clauses, then (AFAIK) they should both have finite verbs. So you’d have to say that some auxiliary (and I dare you to tell me which one) is also omitted.

My tentative guess is that this is a syntactic blend (call it a performance error if you like, but my feeling is that it is the sort of semi-grammaticalized “error” that should be accounted for even by those who think the competence-performance distinction is worthwhile). Say, a blend of (how can) this emulsion have so many opposing forces crammed together (in it) and still be stable with how can there be so many opposing forces crammed together (in this emulsion). And of course the there be construction and the have construction are very similar semantically, so taking the syntax licensed by one (have) but using the actual form of the other is more likely.

(to clarify: they have missiles pointed at them is paraphrasable as there are missiles pointed at them; the first sort of means “missiles are pointed at them; this is related to them”)

1: Note also the lovely speech-act construction not to VP, but [stuff], which also involves some coordination of unlike types. It’s pretty specialized: it’s really strange to rearrange the conjuncts (seems rather obvious why); also, you would probably have a hard time leaving out the not; but you can use just instead.

Okay, the other construction regarding say. As is often the case with constructions, the phenomenon can go along with a long introduction or an abbreviated one, and since I have a seminar to attend shortly, I’m going with the short one. [interesting that this statement is as true now as it was when I first drafted the post]

The word as can be used to parenthetically introduce a clause that has a proposition-denoting clause missing. Oh, just look at these sentences:

• CxG is lame, as most people told me __
• As (it) __ often happens to me of late, I was about to fall asleep when I remembered something very important that I’d forgotten to take care of.
• Ducks and cows, as you have repeatedly proven __ in your research, have well-defined local dialects.

These have a certain external distribution, the determination of which is left as an exercise to the reader; or read Chris Potts’ papers (in 2002) on the matter, though a counterexample to one of his major claims is already present in this post. Now, what I thought I’d found was a case where using say as the subordinate predicate (i.e., as…say), then you got to put it in places where you couldn’t with other verbs. For instance (from Google searches):

Read more »

The construction that I was thinking of that gives say an otherwise unseen valency is: who says?. A variation is says who? (and possibly, as a response, says me/her/etc!, which also features a fixed, inflected form of the verb).

So the valence alternation that is realized in these constructions is : Speaker/NP/Subj + Content/null/DNI. Here DNI is “definite null instantiation,” aka “null complementation” (with definite interpretation). As far as I can tell, there is no other context where say can take just a subject with a definite interpretation of a null complement. And I also don’t know another verb that acts in the same way that doesn’t already license DNI. So, for instance, who knows? works, but I know! is also normal. *I say is bad, even as a response to “who says that blah blah blah?”

(two exceptions to the above: (so) says you!, which also has some strange agreement marking, and also Well, I say!. One may wish to claim some of these are somehow related)

Don’t get mad, please. =) The other construction (which itself has a special distribution with say) will be put up later.

The recent break in posting will, I’m afraid, likely continue for about a week, as I gear up for a couple of presentations at the ICCG. In the meantime, the assignment is to think about the word say. Outside of quotative inversion, I can think of at least two constructions where say acts exceptionally: one imparts say with an otherwise-unseen valency, and in the other, say imparts upon the construction an external distribution that is otherwise illicit.