More on but instead

A couple days ago I posted about apparent failures of parallelism in examples like:

Some whales, including the blue whale, don’t have teeth but instead something called baleen.

This is of the form

(1) not S but (instead) NP

and my discussion was built on the assumption that things like (1) — and (2) as well —

(2) not only S but (also) NP

are just examples of ordinary coordination, with the second conjunct “reduced” to a NP.

Although most discussion of coordination assumes that, insofar as possible, all sorts of coordination with and, but, and or are to be related to coordinations of full clauses (with various sorts of “reduction”, that is, suppression of some shared material), this assumption is gratuitous. The alternative is to posit separate constructions in which the conjuncts are stipulated to have different syntactic structures; the semantic relationship between such a construction and a coordination of full clauses is then a matter of the way the semantics of the special construction works.

I was reminded of this alternative analysis when I came across an entirely unremarkable sentence in Scientific American (September 2009, in Davide Castelvecchi’s “Batteries”, p. 73):

It is the electrolyte that makes the battery work, because it allows ions to flow but not electrons, whereas the external circuit allows electrons to flow but not ions.

This has two clauses in it of the form:

(3) S but not NP

the first of which is “it allows ions to flow but not electrons” ‘it allows ions to flow but it doesn’t allow electrons to flow’.

Ah, this is familiar territory (to me, anyway): a type of what’s called “end-attachment coordination” in section 4.5. of CGEL‘s chapter 15, on coordination and supplementation (by Huddleston, Payne, and Peterson). This is the postposing of a coordinate, as in their example

They had found Kim guilty of perjury but not Pat.

(They also give examples with and and or.) Some postposing examples “have the informational status of an afterthought” (p. 1345), but many do not.

They go on to argue (p. 1346f.) that though an analysis that treats such examples as coordinations of clauses (a full clause plus a clause fragment that we might think of as a “reduced” clause) is possible in many cases, there are examples that don’t have a workable analysis in these terms.

Obviously, there are lots of details to work out in getting the semantics right. But if we look at (3) as a special construction involving the coordination of different structures, then the way is open to viewing (1), and (2) as well, in a similar light.

Leave a Reply

%d bloggers like this: