24 August 2005

sign and trade X for Y

Just a little gapped coordination I noticed today in a headline on TSN.ca:

Senators sign and trade Hossa for Heatley

I know exatcly what it's supposed to mean: the Senators sign Hossa, and then trade him to another team, receiving Heatley in exchange.

I guess I wouldn't have noticed if both verbs had single complements, as in Senators sign and trade Hossa. The actual coordination is a so-called WTF construction, in which the two halves of a coordination are of unlike categories. In this case, one's a single-complement verb and the other is a double-complement verb.

Now, I'm not complaining, but my expectation from the actual coordination is that the complement of the second verb is also understood as the gapped complement of the first verb. This led me to the two following parses:

a. [ [ [sign ___i] and [trade Hossai] ] for Heatley ]

b. [ [sign ___i] and [trade [Hossa for Heatley]i ] ]

In (a), the structure implies that the signing and trading of Hossa were both done for Heatley's benefit (which is not inaccurate). But the placement of for Heatley outside the VP headed by trade suggests that Heatley wanted both actions done, and not necessarily that he was one of the traded parties. (Based on the assumption that if you are "trading X for Y", X and Y are both VP-internal arguments).

In (b), there is this phrase Hossa for Heatley which is understood as the complement of both verbs. But for Heatley seems to be weird as an internal argument for sign.

I can think of four ways of reacting to this:

(1) Dismiss it as ungrammatical or errorful.
(2) Leave it as a product of the special grammar of sports reporting.
(3) Attribute it to headline space restrictions, which preclude Senators sign Hossa and trade him for Heatley.
(4) Adhere to a theory of grammar that allows gapped configurations in which the gapped complement can co-refer with an overt element embedded within the coordinated complement. Thus the following structure:

c. [ [sign ___i] and [trade [Hossai] [for Heatley] ] ]

I'm inclined towards (4). It's not up to me to decide which data is usable and which isn't, which rules out (1). If it were just an aspect of sports reporting, it's still left unexplained. If it's a headline issue, other headlines are clearly longer, as in Champions: Liverpool advances despite loss and Theissen not concerned about Williams in 2006.

(Final possibility: headlines seem to avoid pronouns. In a sample of 44 headlines all clustered together, I found only one with a pronoun: Pistons add Davis to their frontcourt. It's hard to determine how many could have had pronouns but didn't. Looks like the theory implied in (4) needs to let avoid-pronouns-in-headlines trump avoid-wtf-coordinations)

0 Comments:

Post a Comment

<< Home