What is putatively inviolable but it got violated anyway?
« previous post | next post »
I was busy throwing out works by Jerry Fodor today (one really has to, every year or two, or one's whole office would eventually become clogged and unusable) when I noticed that the title of his December 2005 Presidential Address to the Eastern Division of the (published in Proceedings and Addresses of The American Philosophical Association 80:2 [November 2006], 11-24) is a violation of the Coordinate Structure Constraint:
and A Posteriori and It Flies South in the Winter?
You might think it would embarrass a famous defender of the idea that we have innate knowledge of universal grammar if he unreflectingly wrote and published a sentence that violated an important constraint of universal grammar. But it won't.
The Coordinate Structure Constraint, as proposed by John Robert Ross in his MIT doctoral dissertation in 1967, says (I paraphrase) that an unfilled grammatical function logically associated with a fronted element (like the interrogative word what in this case) cannot be located inside one of the coordinates in a coordination if the fronted element is outside it. (There is an exception for the case where the fronted element is associated with a grammatical function that is unfilled in ALL of the coordinates, as in What was [[Fermat convinced he had proved __] [and [ Andrew Wiles finally successful in proving __]]]?, but that will not be relevant here.)
The body of the open interrogative that forms Fodor's title is a coordination of clauses, having this bracketing structure:
[ [ x is universally quantified and necessary and a posteriori ]
[ and [ it flies south in the winter ] ] ]
The variable x stands semantically for the thing we seek in answer to the question. Syntactically it is just a gap. In order for the whole title to have the interrogative force intended, the what must have scope over all of the above. That means it is outside the coordination delimited by the outermost brackets. (There is another coordination inside, namely the adjective phrase coordination universally quantified and necessary and a posteriori, but that is irrelevant. The property I am concerned with would still be evident if you replaced that entire phrase by a single adjective such as necessary.) Thus we have this structure for the whole thing:
[ whatx [ [ x is universally quantified and necessary and a posteriori ]
[ and [ it flies south in the winter ] ] ] ]
There is an embarrassment here, for those still capable of that emotion. On the assumption that linguistic nativism is true, neurologically normal speakers and writers shouldn't come out with utterances that violate universal constraints on syntactic structure. Those universal constraints are supposed to be relatively immutable firmware that guides everybody's linguistic acquisition and use. They're not supposed to be just conventions like that you use the smaller fork for the salad. They're supposed to be deep and (if the theory positing them has any real content) in some sense inviolable. Yet Fodor's title violates the Coordinate Structure Constraint just as surely as Ross's original examples, like:
*[ whose taxx [ [ did the nurse polish her trombone ]
[ and [ the plumber compute x ] ] ] ]
Ross's example does sound (deliberately) worse, mainly because he has embellished its wrong-soundingness by building it around a bizarrely non-serious scenario involving a plumber computing an income tax return while a nurse polishes her trombone in the background. (Lack of seriousness was a serious part of the normal methodology of syntacticians back in the sixties. Seriously.) It also differs in having the variable as object in the second coordinate (the compute clause) rather than subject in the first. That might be the key difference.
For I am not saying Fodor has committed a crashing ungrammatical act. It is important that his title sounds perfectly normal. It was accepted at the time without remark by the massed ranks of the American Philosophical Association, including me (I happened to be at the meeting in December 2005).
So what does that mean? It could mean that Ross's statement of the constraint is wrong. Or that there is no such constraint. Or that sometimes we mistakenly process an ill-formed structure without noticing its ill-formedness. I don't know.
All I know is that presenting Jerry Fodor with this clear evidence that he may have inconsistent beliefs (that linguists have vindicated linguistic nativism by identifying numerous constraints like the Coordinate Structure Constraint that correspond to innate restrictions on syntactic form, and that he has uttered a sentence violating one of said alleged innate restrictions) will have no effect on him whatever. He tends to speak of linguistic nativism as "the only game in town" for anyone who wants to understand how human beings could acquire language; and (he will feel) you don't quit the only game in town just because of one puzzling run of the cards. So syntacticians still have some work to do: big fat hairy deal. Linguistic nativism still rules even if linguists haven't figured out how to state the rules by which it rules. That is the sort of thing he will say.
I will just point out one other tiny little thing, which Fodor himself will have missed. (He does not do close reading of texts, even his own. He will not read this either.) When he came back to his title at the end of the lecture he was apparently unable to violate the constraint again. This time he said the same thing in a slightly different and syntactically irreproachable form, referring to "what's universal and necessary and a posteriori and flies south in the winter."
This is different, you see. [I've changed the following, because it was simply mistaken before, as at least two readers spotted. —GKP] The structure involves a binary coordination. The first conjunct itself contains a coordination (is universal and necessary and a posteriori), and the second is another finite verb phrase, flies south in the winter. There is no violation of Ross's constraint, because there is no coordination of clauses in which one contains the gap and the other doesn't. What is associated with the subject function in a single clause that has a coordination of predicates.
Perhaps the constraint guided Fodor the second time around? Like an angel who lets you fall and skin your knee once, but then saves you when you nearly fall a second time? Like I said, I simply don't know how these putative innate constraints are supposed to operate.
Comments are open for those who do.
Bloix said,
December 29, 2008 @ 1:19 pm
Obviously it's incorrect. It's a joke.
John Lawler said,
December 29, 2008 @ 1:30 pm
Presciently, when I came across this post, I had just finished remarking on resumptive pronouns vis-à-vis Ross Constraints in a comment on the previous post, which points, like the sig link above, to a short introduction to Ross Constraints on my website, whose paraphraseology is closer to the original than Geoff's above, though both work fine.
mollymooly said,
December 29, 2008 @ 1:37 pm
What Bloix said. The question already contains a jocular syllepsis; the pleonastic "it" adds some jazz-hands and ba-dum-chhh.
Mary Kuhner said,
December 29, 2008 @ 2:05 pm
Clearly it is possible to produce sentences that violate this constraint, otherwise we would have no examples to use in explaining what the constraint is. So the strong hypothesis "sentences like this can't be produced" was already untenable well before the talk title came about. This leaves us with a weaker hypothesis along the lines of "sentences like this can't be produced by accident." The talk title, as other posters have mentioned, might have been a non-accident. But I don't see how, in the complicated half-conscious ball-juggling of normal speech, we can ever get a clear line between accidents and non-accidents sufficient to test the hypothesis. If I said something like this, how would you be able to determine if I'd said it by mistake, as a quote, as a playful abuse of English, or two or three of the above?
Language Log's analysis of the plagiarism accusations against Kaavya Viswanathan suggests that a person could soak up whole complex sentences and then write them later without (it is argued) being aware that they are quotes. How, even in principle, could we be sure the talk title wasn't such a half-forgotten quote or slant quote? Linguists seem particularly likely to have been contaminated by previous exposure to sentences violating the constraint. (Professional hazard?)
Luis said,
December 29, 2008 @ 2:35 pm
Maybe I am missing something, but doesn't Fodor's title have a parse where the first three conjuncts are subjconjuncts of a larger conjunct that is itself coordinated with the fourth conjunct? Something like this, with __ marking traces:
[[What is [___ A] and [___ B] and [___ C]] and [it D]
where "it" is anaphoric to "what", in the same way as it "which cocktail do you drink constantly because you like its taste?". If this analysis is viable, then there is no CSC violation.
[Luis: The answer to your question has to be no, the analysis is not viable. It makes the title a coordination of an open interrogative ("What Is Universally Quantified and Necessary and A Posteriori?") and a declarative ("It Flies South in the Winter"). But that means that the question mark is in the wrong place! The obvious intent was to make the whole thing semantically a question. So in that case the "what" is extracted from only the first part of it. And your "because" example is not relevant because it's not a coordination: because is not a coordinator, it's a preposition that takes clause complements. —GKP]
Chris said,
December 30, 2008 @ 7:21 am
"On the assumption that linguistic nativism is true, neurologically normal speakers and writers shouldn't come out with utterances that violate universal constraints on syntactic structure. Those universal constraints are supposed to be relatively immutable firmware that guides everybody's linguistic acquisition and use."
This is all kinds of wrong, as Mary Kuhner politely points out. Remember the competence/performance distinction? I, or Jerry Fodor, or anyone else can string words or non-words or growls or whatever together in whatever way we please (though some strings may be trickier to produce and give an interpretation to than others). Even Mary Kuhner's weaker hypothesis that strings such as the one in question can't be produced by accident seems to me to be explicitly contradicted by the linguistics 101 examples of people coming out with ungrammatical strings due to tiredness, drunkenness etc. I'm not taking issue with the point that this weaker hypothesis is near enough untestable, just with the suggestion that either this weaker hypothesis, or the stronger one that Geoffrey attributes to Fodor and other linguistic nativists is any part of standard conceptions of Universal Grammar.
UG (if it exists) constrains grammars absolutely. Grammars constrain usage only partially. The theory of UG has nothing to say about how usage arises out of grammars.
Finally, the Coordinate Structure Constraint is hypothesized to be part of UG. Maybe this hypothesis is false. Maybe the whole idea of UG is a non-starter, but this in no way hangs on the truth or falsity of the Coordinate Structure Constraint, much less on individual violations of it in usage.
Is anything about what I've said unobvious, controversial or wrong?
Oskar said,
December 30, 2008 @ 8:28 am
I hate to be, you know, that guy, but what is Universally Quantified, Necessary, A Posteriori and flying south in the winter? "All migratory birds"?
Mihai Pomarlan said,
December 30, 2008 @ 10:03 am
No, but a colorless green idea. It would, in summer, be sleeping furiously.
Adam Stephanides said,
January 4, 2009 @ 8:32 pm
Fodor's title was clearly inspired, consciously or unconsciously, by Groucho's riddle in Duck Soup: "what is it that has four pair of pants, lives in Philadelphia and it never rains but it pours?"