A few days ago I wrote to a small group of coordination-and-negation enthusiasts with the following but instead example (from my own hand, in e-mail):
I got constant piles of outraged mail about [my] not having a clickable address, but instead one that had to be typed in by hand.
(that is, ‘… but instead having one that had to be typed in by hand’). I was sure that I had written similar things on other occasions, and in fact I saw no problem with it, despite its being formally non-parallel (the second conjunct is missing a verb, though the first conjunct has one).
My correspondents didn’t recall having seen discussions of such not …. but instead examples, but they turn out to be extraordinarily common (and unremarkable). Despite being formally non-parallel, they are entirely grammatical. Some examples might be a bit awkward, but that seems to be a matter of processing difficulties rather than grammaticality.
And in fact these examples are closely related to another case that has gotten a lot of attention in the usage literature: not only … but also.
Background: there are not … but instead examples that are entirely parallel. Some have the same verb in the two parts:
It is a very interesting fact that turtles don’t have teeth, but instead have horny ridges that are serrated and sharp on their upper and lower jaws. (link)
… taxing benefits might be “a good way to ensure that people don’t have these big Cadillac plans but instead have more sensible plans.” (link)
Some have different verbs:
Earthworms don’t have lungs, but instead breathe through their skin. (link)
In a lawsuit filed the same month, Plotkin contends that LaBan didn’t eat a strip steak, but instead had a “steak sandwich without bread.” (link)
However, non-parallel examples can be googled up by the millions (in raw hits). A very small sampling:
Some whales, including the blue whale, don’t have teeth but instead something called baleen. (link)
Strangely, roughly five percent of these insomnia sufferers actually sleep much better than they realize. Why? They don’t have insomnia, but instead a condition known as Sleep State Misperception, or Pseudoinsomnia. (link)
I didn’t see this perhaps apocryphal fish, but instead something just as remarkable – pristine habitat. (link)
We didn’t see many people begging for money, but instead many entrepreneurs selling carved wooden hippos or copper bracelets they had made. (link)
The fish were carefully kept in plastic bags with oxygen supplied. We didn’t plan to eat them, but instead sell them alive. [i.e., 'instead planned to sell them alive'] (link)
The last of these examples strikes me as a bit awkward, but still grammatical. Most of the examples I googled up would not, I think, attract anyone’s notice.
MWDEU makes a similar point about “failure of parallelism” in not only … but (also) examples (“They not only tell lies but bad lies”, from Jowett’s translation of The Dialogues of Plato). Complaints about non-parallelism here go back to the 19th century, but despite this, the “the nonparallel construction is common enough to pass almost unnoticed” and “it creates no confusion or misunderstanding”. MWDEU concludes,
It is clear that constructions which are not precisely parallel are as much a part of standard English as those which are precisely parallel.
MWDEU isn’t claiming that all non-parallel constructions are standard English, only that some are. (Though in fact you could press the case that most non-parallelisms are not offenses against grammar, but present problems of other sorts. The topic is very complex, in part because the usage literature is remarkably confused about what constitutes faulty parallelism. For more discussion, see my long Language Log posting from 2005 on faulty parallelism.)
People sometimes explain to me that formally non-parallel examples like the ones treated above might pass notice, but are still “technically” errors in grammar. When I ask them for the authority that pronounces them errors, they sometimes cite Strunk & White (1959), where we find the following rule (italics and bold face as in the original):
15. Express co-ordinate ideas in similar form.
That’s not a bad piece of practical advice for writers (parallelism helps readers to process written material, though it’s far from the only relevant factor in easing processing) — but what S&W mean by “similar form” is not at all clear, so it’s tough advice for writers to follow. In any case, how did this rule come to be seen as a rule of grammar?
But so it has, which is probably what leads people to think of non-parallelism as “technically” a mistake in grammar. Or maybe they believe that the Language Lords have ruled on the matter, and that’s the end of it. The Language Lords have a lot to answer for.