?

Log in

No account? Create an account
 
 
09 Maj 2004 @ 12:25
A nearly Escherian Bushism  

"Exact Change," Libby Stephens's usually obnoxious column in the Saturday Star, points out that George Bush's non-apology to the Arab world contains the following Bushism, which is remarkably similar to the sort of "Escher sentence" that has recently been receiving attention from some of the linguists at Language Log (viz., Geoff Pullum, David Beaver, and Mark Liberman), from Kai von Fintel, from Fernando Pereira, and from entangledbank (here):

The American people are just as appalled at what they have seen on TV as Iraqi citizens have. The Iraqi citizens must understand that.

"If any Iraqis actually do understand that," Stephens writes, "we wish they'd translate for us." I think that's a bit unfair; for once, it's really pretty obvious what Bush intended to say. Surely what he had in mind was:

The American people are just as appalled at what they have seen on TV as Iraqi citizens are.

He just made the elementary processing error of repeating the more recent auxiliary verb have instead of the copula are from the matrix clause:

The American people are just as appalled at what they have seen on TV as Iraqi citizens have.

Now, the canonical Escher sentences are things like More people have written about this than I have and More people have been to Russia than I have. These are a bit more complicated, and the real mystery is not how they arise (which they don't necessarily do very often), but rather why they seem perfectly sensible at first glance. Part of it seems to be that they very nearly mean two different perfectly reasonable things—that first one, for instance, very nearly means "More people than I have written about this," and also very nearly means "Other people have written more about this than I have" at the same time. And, as the Language Loggers have already pointed out, no local red flags go up as the sentence sneaks into your brain syllable by syllable or saccade by saccade: in order to notice that it's incoherent, you have to analyze the whole thing.

I don't have a theory of Escher sentences here, but I am reminded of a couple of things:

  1. I remember reading in Carson Schütze's The Empirical Base of Linguistics (which I have not to hand, or I'd look this up and get it right) about a study in which subjects were asked to evaluate the sentence I never heard a green horse smoke a dozen oranges. Many at first said it was ungrammatical, but changed their minds when it was pointed out to them that it expressed a true proposition. This is interesting first because it shows how hard it is for people to tease apart grammaticality and truth in evaluating sentences, and secondly because it shows how the inclusion of a few unexpected combinations of words can trick people into thinking that a sentence is ungrammatical even when it is both grammatical and true. (There may also be somehing in there about the difficulty of detecting negation; note that it's the never that makes the sentence true, at least for most of us.) Escher sentences seem to do the same thing in reverse: they look so normal and so plausible that we don't notice, at first, that they're really uninterpretable. (This still doesn't explain why they look so much more normal and plausible than other ungrammatical sentences, though.)

  2. This one goes back to the notion that an Escher sentence very nearly means two different things at once. From Lewis Carroll's preface to The Hunting of the Snark:

    Humpty-Dumpty's theory, of two meanings packed into one word like a portmanteau, seems to me the right explanation for all.

    For instance, take the two words "fuming" and "furious." Make up your mind that you will say both words, but leave it unsettled which you will say first. Now open your mouth and speak. If your thoughts incline ever so little towards " fuming," you will say "fuming-furious;" if they turn, by even a hair's breadth, towards "furious," you will say "furious-fuming;" but if you have the rarest of gifts, a perfectly balanced mind, you will say "frumious."

    Supposing that, when Pistol uttered the well-known words

    "Under which king, Bezonian? Speak or die!"

    Justice Shallow had felt certain that it was either William or Richard, but had not been able to settle which, so that he could not possibly say either name before the other, can it be doubted that, rather than die, he would have gasped out "Rilchiam!"

    Somehow I can't help feeling that Escher sentences are to syntax and semantics what frumious Rilchiams are to phonotactics and the lexicon. That's not an explanation, either, of course, but it's an excuse to quote from The Hunting of the Snark, which is worth something in itself.

Escher in Lego
by A. Lipson & D. Shiu
(via Geek Notes)


"Belvedere"


"Ascending
and descending"


"Relativity"


"Waterfall"

 
 
 
Tishiewahooweena on 9. Maj, 2004 22:23 (UTC)
In the field of linguistics, is it ever said that linguistics is related to psychology? Whenever you make these entries, I think, "Maybe I missed my calling. I don't know what half of these words mean, but the parts I do understand are fascinating!" Then, I realize that you are talking about how we think.

Really, though. Fascinating!
Q. Pheevrq_pheevr on 10. Maj, 2004 10:14 (UTC)

Yes, indeed. The question of why these Escher sentences seem fine at first glance is really a psycholinguistic question, and psycholinguists investigate questions of this sort using much the same experimental methodology as in other areas of psychology.

The kind of theoretical linguistics I do focuses more on describing and explaining the systems of language itself, in a more abstract and absolute sense. The kinds of questions we ask go something like: "What makes a string of sounds qualify as a grammatical sentence of language L?" In asking this sort of question, we tend to abstract away from the fact that some sentences are easier than others for speakers to evaluate. But there is always the understanding that what we are talking about is part of the larger question of how the human mind works, and that the kinds of answers we give to these questions must ultimately also be compatible with the results of the psycholinguists.

Consider for example an ambiguous sentence like:

We sliced the pizza with the pepperoni.

The theoretical linguist would note that the grammar of English assigns two different structures to this sequence of words—one in which the prepositional phrase with the pepperoni modifies the noun pizza, and one in which it modifies the verb sliced (a nonsensical reading, unless you have very sharp pepperoni), and would potentially be interested in working out exactly what those structures were and how they end up meaning what they do. The psycholinguist might investigate which one of these structures someone hearing the sentence would find more salient, and why, or how much longer it takes to process an ambiguous sentence than an unambiguous one.

Tishiewahooweena on 10. Maj, 2004 17:13 (UTC)
That is so cool!

FTR, the most salient reading for me was the sharp pepperoni version.

Maybe when I feel like taking a night class, I'll see if any of the linguistics classes would be open to me.
Q. Pheevrq_pheevr on 10. Maj, 2004 17:45 (UTC)
FTR, the most salient reading for me was the sharp pepperoni version.

It seems to work that way for most people, which is interesting, since that's the reading that makes less sense pragmatically. This kind of ambiguity is also a big deal in computational linguistics: if you're designing a program that processes natural language, it needs to be able to pick the appropriate structure (or, better yet, to recognize the ambiguity and figure out which is the right structure). Depending on what the goal of the project is, you might want the program to pick the sensible reading, or you might want it to act like a human being.

Tishiewahooweena on 10. Maj, 2004 21:23 (UTC)
Wow, that seems incredibly complex, because frequently, we humans can't quite figure out what another human might find sensible.

The pizza sentence is probably one of the easy ones, as it seems to be a matter of common phrases, groupings of words, or [substitute correct linguistic term here] and the expectations they create. Afterall, who says, "pizza with pepperoni"? It's a pepperoni pizza, fer cryinoutloud!
-entangledbank on 11. Maj, 2004 01:29 (UTC)
So you refine the experiment: We sliced the pizza with extra anchovies. But then for me the bad reading is sort of ruled out by the same anomaly as in: We sliced the pizza with an extra knife. It's hard to develop truly ambiguous constructions, i.e. ones where both readings work perfectly naturally with a single change of word. We sliced the pizza with too many anchovies/knives. Nope, still not right for knives.
Q. Pheevrq_pheevr on 11. Maj, 2004 09:28 (UTC)

Well, how about this:

We sliced the pizza with green peppers.

Green peppers, unlike pepperoni, anchovies, or even mushrooms, don't seem to prefer to be prenominal (?the green pepper pizza).

Another example from the literature (I believe I heard it from Graeme Hirst):

The landlord painted the walls with cracks.
(Anonym) on 11. Maj, 2004 08:50 (UTC)
Anna Phor
"We sliced the pizza with the pepperoni."

I actually get a phonological difference that resolves the ambiguity in this. When the pepperoni is the instrument I can hear the clause boundary -- I'm not sure that I can adequately describe all that's going on there, but "with" receives more stress and there may be some heavier aspiration on "pizza" when it belongs to the heavier noun phrase.

Oddly, I can't hear a difference in my favourite stock ambiguous sentence, which is "Mary saw John on Mauna Loa with a telescope".

The Escher sentences are interesting -- but I wonder if they are really "Escher"? Escher drawings work by being coherent at the local level but impossible when viewed as a whole; and they have multiple interpretations which pop out depending upon which part of the picture that you are looking at.

A sentence like: "More people have written about this than I have" has the first Escher quality; local but not global coherence. But I don't get multiple interpretations ... the only thing I can get this to mean is "[Other] People have written more about this than I have written about this". Also I've just noticed that it has a collective interpretation; it's the totality of writings by other people which exceeds my writing, but it's not necessarily true that any individual has written more about it than I have.

I'm not so worried about how the sentence gets understood -- people assume it's understandable and go for the closest match. (Although I think that's another commonality with Escher pictures. People assume that they are coherent, and I'll bet that there are sometimes people who don't realize that they aren't until it's pointed out to them.) I think the interesting question is what's going on when you produce one of these.
Q. Pheevrq_pheevr on 11. Maj, 2004 09:37 (UTC)
Re: Anna Phor
A sentence like: "More people have written about this than I have" has the first Escher quality; local but not global coherence. But I don't get multiple interpretations ... the only thing I can get this to mean is "[Other] People have written more about this than I have written about this". Also I've just noticed that it has a collective interpretation; it's the totality of writings by other people which exceeds my writing, but it's not necessarily true that any individual has written more about it than I have.

Hm.... I wonder if More people have written about this than I have could arise from a valiant attempt to depassivize something like More has been written about this than I have written about this?

(Anonym) on 11. Maj, 2004 11:32 (UTC)
Re: Anna Phor
I think that's on the right track; although I'm not sure that folk try to depassivize on the fly like that.

The passive does however allow a collective reading whereas I can't find a way to favour a collective reading in the active. "People have written more about this ..." gives me the interpretation "there are several people, each of whom has written more than I have".

I'm wondering if its not some way of trying to shoehorn "people" into feeling like a mass noun? People is an irregular plural, so maybe [more + plural without s] is a template for a collective/mass reading; and then when the speaker gets to the place that "more" should properly be in the sentence, he or she has already used it and so it doesn't get slotted in?

I've no idea what to do with "More people have been to Russia than I". ("Russia has been gone to more often than I have gone to Russia"? Eek.)

I can't even parse the Russia example; but when I first saw "more ...written .." it took me a few minutes to work out what was wrong with it.