August 01, 2004

What book did John buy and read the magazine?

I've been unearthing coordinations that I find grammatical, even though they violate some hypothesis about coordination, as in my "Do I misspeak?" posting. The latest find -- (1) "Hyatt Rickeys, which will be demolished and the property turned into a residential development" (which I reported on on ADS-L, 7/28/04) -- is an apparent violation of the Coordinate Structure Constraint, with an extraction only from the first conjunct; (2) "What book did John buy and read the magazine?" is a classic example of this type, and pretty much everyone (me included) judges it to be ungrammatical. But (1) seems fine to me, so it looks like another in a list of types of exceptions to the CSC, for instance things like (3) "What did Harry go to the store and buy?" and (4) "How much can you drink and still stay sober?", which are widely judged to be grammatical (and I agree).

Now, at Neal Whitman's suggestion, I've looked at the "coherence and extraction" section (chapter 6) of Andy Kehler's Coherence, Reference, and the Theory of Grammar (2002), where exceptions to the CSC (also the literature on them) are thoroughly surveyed, then reanalyzed in discourse-structural, rather than purely syntactic, terms -- and there I find Kehler asking us to reassess even the asterisk on examples like (2).

As Kehler puts it (p. 133):

It may... be possible to retain a suitably articulated hypothesis that syntax is autonomous and still account for [data like (2)-(4)], as long as the CSC is neither included within nor is a by-product of the system of grammar rules. Because this view forces us to the conclusion that sentences like [(2)] are perfectly grammatical, however, it brings to light a potentially worrisome situation regarding the manner in which theorists rely on their judgments. Previous researchers have certainly considered sentences like [(2)] to be ungrammatical, and it is at least questionable whether this judgment differs qualitatively from other ungrammaticality judgments upon which researchers commonly construct their syntactic theories. This data should force us to reassess whether the intuitions we have about ungrammaticality really represent syntactic wellformedness, and if they do not, what one might use as a basis for determining what sentences are unacceptable for purely syntactic reasons.

Definitely food for thought, I say.

Some further remarks on (1)... Though this particular sort of example hasn't, so far as I know, been discussed in the literature (even by Kehler), it strikes me as falling pretty easily into Kehler's general "coherence" approach, though the details need to be worked out. The idea would be that there's a relationship of association-in-context between the hotel (Hyatt Rickeys) and the property (the property on which the hotel sits, in fact) -- a relationship much like that between the referent of the dependent noun of a N-N compound and the referent of its head noun (cf. "pumpkin bus", with its wide range of interpretations, or for that matter "hotel property") -- as well as a relationship between the demolishing of the hotel and the transformation of the property into housing, namely that the two are presented as parts of a single unfolding event.

Things are not quite so simple, though. Gapping is crucial to the acceptability of (1). Without Gapping in the second conjunct, the coordination is much less acceptable: (1') "Hyatt Rickeys, which will be demolished and the property will be turned into a residential development". I wouldn't go so far as label (1') as ungrammatical, especially in view of Kehler's cautions above, but it certainly is clunky. In contrast to (1), where Gapping helps things along, compare cases where Gapping helps not at all: (5) "Kim, who ate sushi and Sandy sashimi" is as bad as (5') "Kim, who ate sushi and Sandy ate sashimi". It seems that the special relationship between the referents of the two subject NPs (Hyatt Rickeys and the property, vs. Kim and Sandy) plays an important role here, and in other examples constructed by Clai Rice in a 7/30/04 posting to ADS-L: (6) "the junk cars, which will be destroyed and the tires recycled" (the tires could be the ones belonging to the cars, ones on which they sit in the junkyard, ones that are stored inside the cars, or many other things -- roughly the range of interpretation for "car tires"); (7) "the junk cars, which will be crushed and the birds transferred to the sanctuary" (cf. "car birds"); and (8) "the hotel, which will be demolished and the lake filled in" (cf. "hotel lake").

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at August 1, 2004 03:08 AM