April 12, 2005

Astounding Coordinations (continued)

Eric Bakovic, spurred by Neal Whitman and Mark Liberman, continues our conversation on Astounding Coordinations, with examples that seem to involve coordination of unlikes and/or remainders in coordination that are understood in different ways with different conjuncts.  I've been collecting various sorts of failure-of-parallelism examples for a while now; here are some further contributions to the conversation -- a couple of relatively routine examples, plus some apparently involving coordinations of a clause having a subject gap with a clause having an object gap.

1.  Ordinary failures of parallelism.  First, from MacNeil & Cran, Do You Speak American? (2004), p. 61:  "Kirk Arnott, assistant managing editor [of the Columbus Dispatch], is the language cop or watchdog of the Dispatch.  He believes in informal and conversational language, and that his paper should be as conversational as possible, to be accessible and clear to readers."

On the surface, this is:

... believes [ [in NP] and [that S] ]

with a PP and a complement clause treated as parallel.  Somewhat more ingeniously, you could claim that the structure is really:

... believes in [ NP and [that S] ]

(with automatic suppression of the "in" of "in that S" when the remainder is distributed over the conjuncts), though that still has an NP and a complement clause treated as parallel.  And there's the question of whether the two conjuncts are interpreted with the same verb believe.

Next, there's a Remington shaver commercial, heard on television 12/21/04:  "... designed for closeness, comfort, and to clean itself automatically".

There are two issues here.  One is what James Cochrane, A Little Book of Bad English, calls "not enough ands", as in his example (pp. 91-2): "The scandal was headline news, seriously damaging the credibility of the president, the Republican Party, and giving a considerable boost to the lagging Democrats."  (Cochrane, p. 93, maintains that "errors of this kind have begun to crop up regularly only in the last twenty or thirty years", a claim of recency that I doubt, though I'm not yet in a position to dispute it with data.)  In the Remington commercial, the and problem, if it is one, can be easily fixed: "... designed for closeness and comfort, and to clean itself automatically".

Then this example is like the Dispatch one.  On the surface, it's:

... designed [ [for NP] and [to VP] ]

with a PP and an infinitival VP treated as parallel.  Or, if the structure is

... designed for [ NP and [to VP] ]

(with automatic suppression of the "for" in "for to VP" when the remainder is distributed over the conjuncts), we have a NP and an infinitival VP treated as parallel.  In any case, I have no trouble interpreting the two conjuncts with the same verb design, but others might find even this problematic.

As a side note, I should point out that instances of "faulty parallelism" involving either... or are very common indeed, to the point where it seems absurd to treat all of them as ungrammatical.  Among the examples supplied by Merriam-Webster's Dictionary of English Usage (p. 434) is this fine specimen from Darwin's Origin of Species: "... the stripes are either plainer or appear more commonly in the young".  Here either is located "too low"; for perfect parallelism we'd need: "... the stripes either are plainer or appear more commonly in the young".

It turns out that there's a pretty considerable literature on the placement of either, which I was made aware of last Friday by Philip Hofmeister's public presentation of a Stanford qualifying paper.

2.  On to subject and object gaps.  Yesterday, Bruno Estigarribia wrote Ivan Sag and me with an example from the New York Times, April 8, 2005, "Maybe Less Use of the Prescription Pen" by Anahad O'Connor, Business section:

"When you compound these drugs, that means the heart won't see it, and the stomach won't see it," she said. "So for people who I'm not going to give a cox-2 and also have a history of ulcers, the way around it is to take the anti-inflammatory and make it into a cream."

Estigarribia commented: "It took me near 3 seconds to parse the sentence (or maybe more, I wasn't timing it, I sure was confused). What is going on here?"

I replied that coordination of a clause with an object gap ("I'm not going to give a cox-2") and a clause with a subject gap ("also have a history of ulcers") is usually judged ungrammatical, though there's some question about what condition bars it.  And I provided two further examples that were discussed on the newsgroup sci.lang back in November 2004:

(1) ... the "Control Panel" (which you presumably have to know is there and how to get to)...

(2) [...] New Mexico, which the president leads but was still uncalled as of noon Wednesday...

(In example (2), you have to accept that the writer intended "the president leads New Mexico" to mean something like 'the president leads in New Mexico'.)

Ivan added that, if he remembered right, Gerald Gazdar's 1981 paper ("Unbounded dependencies and coordinate structure", Linguistic Inquiry 12.155-84) treated these as ungrammatical, though that conclusion was challenged in a paper presented at a summer LSA meeting (maybe College Park in 1982).  I'm working on tracking that paper down.

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at April 12, 2005 02:01 PM