Reported in the Sic! (errors) section of Michael Quinion’s World Wide Words #794 this morning:
A health report of 18 July on the BBC site about the risks of not taking physical exercise was spotted by Martin Wynne: “The public needed to be warned about the dangers of inactivity rather than just reminded of the benefits of it.”
Here, it at first appears to refer in inactivity, but a bit of thought will convince you that the writer intended it to refer to activity; but activity is inside the word inactivity, and so would (on many accounts) be unavailable as an antecedent for it. The relevant putative generalization is known as the Anaphoric Island Constraint (AIC): words are “islands” for anaphora; anaphora can’t “reach inside” words. (Brief discussion here; examples of AIC violations here and here.)
But things aren’t that simple.
First observation: someone at the BBC spotted the problem and changed the text:
And they said tackling [the problem] required a new way of thinking, suggesting the public needed to be warned about the dangers of inactivity rather than just reminded of the benefits of being active. (link)
(Now there’s no anaphora, so no possible AIC violation.)
Then some e-mail exchange this morning between Gregory Ward (with other AIC mavens listening in) and me, in which Gregory noted that cases like this were covered in
Ward, Gregory; Richard Sproat; & Gail McKoon. 1991. A pragmatic analysis of so-called anaphoric islands. Lg 67.3.439-74.
where the authors argue that anaphoric islands aren’t a syntactic/semantic phenomenon at all, but a pragmatic one, with the acceptability of particular examples having to do with processing ease considerations linked to context, discourse organization, and the like.
Still, I pointed out, the original example was likely to be mis-read the first time through. Ah, Gregory replied:
once the [in-] prefix is accented (to signal the upcoming contrast), it reads much better.
I agreed, adding:
But that’s unlikely to work in print (though it’s ok in speech). So it’s a good example of a sentence that sounded fine in the writer’s head but won’t work well in print.
Now I was back in familiar territory. From my 2007 posting, “Hearing the sentence in your head”:
From Atul Gawande’s op-ed piece “Let’s Talk About Sex” in the NYT, 5/19/07, p. A25:
Reducing unintended pregnancy is the key — half of pregnancies are unintended, and 4 in 10 of them end in abortion.
The first reading I got was that 4 in 10 pregnancies end in abortion. But whoa, that can’t be right; surely the abortion rate isn’t that high. Gawande must have intended to say that 4 of 10 UNINTENDED pregnancies (2 in 10 of all pregnancies) end in abortion.
I read the sentence in my head with unaccented them, the usual prosody for anaphoric pronouns; with that accenting, the referent of them is pregnancies, pregnancies being the nearest available antecedent NP, an NP moreover in a phrase (of pregnancies) parallel to the phrase (of them) that them is in. Gawande presumably heard it in HIS head with accented them, the accent here signaling that the usual referent-finding procedures don’t apply. (The accented HIS in my last sentence illustrates a different use of accent: to point up a contrast, his head vs. my head.)
Gawande could have made things clear — by putting them in small caps or italics, to indicate accent, or by choosing those or these, which here would convey the introduction of a new discourse referent, one the reader has to identify from the context: unintended pregnancies. Why didn’t he make things clear?
Because he heard the sentence in his head and didn’t realize that he’d have to mark the accent for the reader (or choose a different anaphor).
Here’s an area where a good copyeditor could have saved the day; another pair of eyes in another person’s head would probably have caught the problem with the BBC and the Gawande sentences.