I've got friends in low places,
where the whiskey drowns and the beer chases
my blues away.
We start from a structure in which a final constituent is construed with both members of a preceding conjunction, say
Kim selected and Leslie packed the samples.
and then add something else that only makes sense with the second conjunct, e.g.
??Kim selected and Leslie packed the samples up.
It's not clear what the status of these structures is. Ordinary examples like the one that I just constructed seem pretty doubtful to me, but there are plenty of cases like the Garth Brooks lyric that go down fine, at least as long as I don't think about them too closely.
Meanwhile, I keep seeing things that might be called "anti-FLoP coordinations", where rather than the second half of such a conjunction having an extra bit, instead the first half is missing something. Here's one that I remembered to mail myself a link to -- it's from a 2/21/2005 Q&A in the "Online Only" section of the New Yorker, where Michael Spector explains about avian flu:
Second, it can kill and cause severe disease in humans—though, so far, for that to happen a person would have to have been exposed at great length, or have eaten raw, infected poultry. [emphasis added]
The point is that the "full" form of the first conjunct must be something like
...a person would have to have been exposed at great length to infected poultry.
with an extra "to" that's nowhere to be found in the original.
In both conjuncts, an object or indirect object has been placed after another constituent that it might well have preceded, in order to get "infected poultry" into final position to be shared:
...a person would have to have eaten [infected poultry] raw.
...a person would have to have eaten raw [infected poultry].
...a person would have to have been exposed [(to) infected poultry] at great length.
...a person would have to have been exposed at great length [(to) infected poultry].
It's clear from the rest of the transcript that Michael Spector likes this kind of structure, which has traditionally been called "right node raising" to express the sense that a shared final (i.e. "right") constituent has been "raised" so as to be shared by both members of a preceding conjunction:
[A B C] and [D E C] ⇒ [[[A B] and [D E]] C]
There are at least two other examples in the same interview. One was earlier in the same sentence cited earlier, and involves sharing a noun that is the object of a verb in the first conjunct and of a preposition in the second:
Second, it can kill and cause severe disease in humans—though, so far, for that to happen a person would have to have been exposed at great length, or have eaten raw, infected poultry.
In the third case, the shared constituent is a prepositional phrase, connected to noun phrases in both conjuncts:
But by closely monitoring the spread, and by examining the genetic structure of the virus, we can get a sense of how to develop a vaccine and how to make better drugs.
In the first case I cited, the noun phrase "infected poultry" is the object of a preposition in the first conjunct and of a verb in the second one. Michael Spector was apparently so confused by all the shifting around needed to get the constituents in the right order to allow "infected poultry" to be shared, that he didn't notice the little "to" that got lost in the shuffle. In other words, I don't think this is a dialect difference, or an informal construction, or a syntactic change in progress -- it's just a mistake, and I'd bet that Spector would think so too, if it were pointed out to him.
It's surprising that that this got through the New Yorker's editorial process. Although the document is an interview transcript, it's surely been edited at least to the extent of removing filled pauses, false starts and so on, and you'd think that they'd fix this kind of verbal mistake as well.Posted by Mark Liberman at April 11, 2005 09:47 PM