?

Log in

No account? Create an account
Trevor Stone's Journal
Those who can, do. The rest hyperlink.
Syntax Bush 
23rd-Feb-2009 06:36 pm
I *kiss* linguists
Is there a sensible syntax tree for A rose is a rose is a rose? The parse that makes most sense to me is
{a rose is [a rose} is (a rose])
which is not a tree. But the parse
((a rose) is ((a rose) is (a rose)))
is NP is VP at the top level, and that's not generally grammatical (e.g., "Obama is the president lives in the White House").
Comments 
24th-Feb-2009 01:55 am (UTC)
Looking for consistency in English is like looking for diamonds in tap water.
24th-Feb-2009 02:11 am (UTC)
Poetry and syntactic sensibility do not good bedfellows make.
24th-Feb-2009 07:08 am (UTC)
I suppose that brings up the much larger question: How important is syntax, really? If the poetry is perfectly clear but the syntax model gets all muddled, what shall we do on Saturnalia next?
24th-Feb-2009 08:30 am (UTC)
Very not syntax important is. To some extent, at least. The language processing centers of our brains are able to do all kinds of gymnastics around what should logically be insurmountable obstacles -- when someone is talking to you on the phone and they cut out every other word, you often -- figure -- what -- wanted -- say. Poetry is very similar, there are still some basic syntactic rules that are being followed in the "Rose is a rose" quote, it's not actually that far off of a natural, grammatical sentence.

The question, retargeted to what I think is the more intriguing angle, would be: How does semantic content seem to transcend so much of our syntactic structure? And the answer to that is, we really don't know yet. There are some very intriguing theories on it, but it's one of those things that is not really currently testable, except by coming up with examples that seem to fit the various hypotheses.

One of my favorite examples is the following, completely grammatical, sentence:

"Well, more people have been to Russia than I have."

At first glance, this seems to make perfect sense to most people. But when asked to explain what it means, they look at it again and have that "...wait, but..." moment with it. This has even been referenced among linguists as proof of an autonomous syntax, that operates outside the bounds of the parsing of meaning, and vice versa. I don't know if I can go quite that far, but I think it speaks volumes for the fact that sentences can be grammatical gibberish, just as easily as they can be ungrammatical but crystal clear. (The more subtle example from one of Chomsky's earliest linguistic texts is the now-iconic "Colorless green ideas sleep furiously," but this one demonstrates a different aspect of the separation of grammar and semantics -- the former example truly has no discernable meaning that isn't conjecture, whereas this sentence does contain a kernel of meaning, albeit coated in some poetic license and self-contradiction.)

Also, syntax is a very fluid element cross-linguistically. There are many languages where word order bears little or no governance, and all meaning is conveyed through inflectional affixes or other means of marking case onto individual words. English is one of the stricter word-order syntax structures, but even we have some fluidity available to us while remaining "grammatical," and meaning can persist even beyond those restrictions.

And now my final thought on the whole thing, you say "the poetry is perfectly clear," but is it? That single line has been interpreted and analyzed in many different ways, and I believe part of that is that the semantics are given a freer rein when the syntax is disjointed and pulled away from the expected constructions, as poetry often does. I believe that one of the most important aspects of poetry is precisely that -- it allows for language to be put through contortions and acrobatics that allow new potential semantic interpretations to surface because while it cannot rely on common ground and prior construction recognition in the way that conversational grammatical speech does, it is not restricted by them either.

Edited at 2009-02-24 08:31 am (UTC)
24th-Feb-2009 09:59 am (UTC)
Hmm. A somewhat naive question: is the Russia sentence grammatical? How would one diagram it?
24th-Feb-2009 09:41 pm (UTC)
That's not a naive question at all.

The problem with this sentence is that it's clearly not... SOMETHING... because it's gibberish. And yet most people are happy to accept it as not only grammatical but meaningful, until they're asked to really point out what the sentence means. But unlike other sentences that have similar effects (such as 'garden path sentences' like "the horse ran past the barn fell," or "throw me out the window my sweater," or other similar language illusions), this sentence will continue to appear grammatical even after you have seen through the 'illusion' of it and recognize it as nonsense.

So... I don't remember off the top of my head how much ling background you have, but I seem to remember you having at least a little bit. I will use a little ling jargon here and there, but feel free to ask for clarification if I assume too much familiarity on your part. I'm going to work in the X-bar syntax structure, as I feel it's more descriptive in this case. To simply tree-diagram this sentence doesn't, I think, really address the awesomeness of it.

The sentence is an example of what linguists have begun lately to term 'Escher sentences' (a reference to the plausible-yet-impossible Escher staircases and other depth illusion illustrations). This is because you can diagram them to some extent, but at some point the diagramming curls back in on itself and creates the optical illusion of grammaticality but leaving no room for semantic mapping that makes any sense.

So, in a way, the sentence isn't "grammatical," but only insofar as no meaning can be mapped to it.

Take for example the following sentence: "More Americans have been to Hawaii than Britons have." That makes perfect sense, and is grammatical.

NP[More Americans] VP[[have been to Hawaii] PP[than Britons have]]

(I have omitted some other minor diagramming such as the VP's initial prepositional phrase structure "to Hawaii", and focused on the main issue at hand, for ease of discussion.)

Alternately, here is another perfectly grammatical sentence: "They have been been to more countries than I have."

NP[They] VP[[have been been to more countries] PP[than I have]]

In the first sentence, there has been an ellipsis from the [Britons]-headed VP, as the uninflected 'have' at the end indicates. The ellided subordinate VP would be PP[than I VP[have been to Hawaii]].

(The sentence would in fact be more prescriptively grammatical if the entire VP-clause were ellided: "They have been to more countries than I.")

Try to replace this ellision to our Escher sentence, however, and:

"More people have been to Russia than I have [been to Russia]."

This obeys all of the grammatical rules of the Hawaii example sentence, and yet the collision of using a singular quantity in this type of comparative structure creates that Escher staircase effect. But the (greatly simplified) X-bar syntactic diagram of the sentence would be:

NP[More people] VP[[have been to Russia] PP[than I have]]

This is a perfectly grammatical construction, as shown by the other two sentences, and yet it is actually a Frankenstein's monster of the two, creating the illusion of perfectly acceptable construction but obliterating all possible meaning. Weird, huh?

Oh and PS -- I've added you to my friends list, and look forward to more LJ (and maybe even real-world) hangouts.
25th-Feb-2009 07:45 am (UTC)
Wow. It's weird that the rules of grammar don't actually throw up any red flags when you pull that Frankenstein technique. But now that I think about it, I suppose language is a sufficiently complex formal system for Godel's Incompleteness Theorem to apply, so it would make sense after all that you can construct nonsensical sentences without the rules stopping you.

Where do you hang out IRL these days, anyway? I've been sort of hermiting.
25th-Feb-2009 07:48 am (UTC)
I know, it's one of my favorite things about linguistics -- these bizarre little hiccups that we discover, that show all kinds of unpredictable responses. Brains are so cool.

re: hermiting, I hear you. I have been too, to some extent. I hang my hat in SJ, and have been mucking about all over. It would be great to see you sometime.
24th-Feb-2009 10:13 pm (UTC)
you say "the poetry is perfectly clear," but is it?

Stein's original utterance is just as clear as "Rose is a rose; a rose is a rose" which fits a syntax tree.

The thrust of my idea was "How necessary is syntax to communicative understanding?" What if we think of syntax/grammar as one tool of many for use in communication? "When things get complex, apply grammar to keep everything clear." In that view, syntax parsing as a step in a pipeline is the wrong approach. Instead, you'd try to parse if simpler semantic extraction techniques didn't work or you'd run a parse simultaneous to other angles (neural net style).
This page was loaded May 21st 2018, 1:08 pm GMT.