Arnold Zwicky's Blog

On the 14th, Ron Butters posted to ADS-L with a complaint about his AOL spell- and grammar-checker. He typed

I will have to convince the writer to give us a quick revision of her article.

and the grammar-checker spit (or perhaps spat) back that this use of convince had an “inappropriate preposition”, and suggested persuade instead.

Butters said he vaguely remembered

that some old-time prescriptivists condemn the use of “convince” as a verb meaning “persuade,” but this seems bizarrely old-fashioned–and “preposition” has nothing to do with it.

The proscription against convince with an infinitival VP complement is in fact a “zombie rule” (like the ones discussed here), a proscription that has died in practice but continues to lumber about in odd corners of usage advice.

First, “preposition”. Well, yes, we were all puzzled by this. The program could have said that the verb was inappropriate, Or, as Larry Horn noted on ADS-L, it could have said that the subcategorization frame — that is, the construction — was incorrect for this verb (not that grammar-checkers know from subcategorization frames, but you can see the point). In fact, the problem, if there were one, would be with the combination of verb choice and construction choice, neither of these being primary.

So where did “preposition” come from? Almost surely from the infinitive marker to. In this use, it’s certainly not a preposition — though hundreds of years ago it was, and because of this history many dictionaries list it in an entry for the preposition to.

In any case, the issue here has to do specifically with convince (and persuade) with an infinitival complement, conveying impulsion towards action. There’s no issue with things like

I convinced them that the earth is flat.
I convinced them of the truth of the phlogiston theory.

(or with the variants of these with persuade). The question is about things like

I convinced them to leave.
They were convinced to leave.

[It can be hard to talk about these matters. The manuals are inclined to assume the rightness of the variant they prescribe, so that they’ll talk about “using convince instead of persuade“, or, more blatantly, “using convince where persuade would be correct”. If you talk about things this way, it’s going to be hard to have a rational discussion with you about the choice between convince and persuade. Your mind is already made up.]

MWDEU has an excellent detailed article on convince and persuade. There’s a tradition going back to the 19th century for distinguishing the two verbs by meaning, with convince referring to mental acceptance and persuade to mental acceptance leading to action. Indeed, there were complaints about persuade being used in situations where no component of resultant action is involved (as in persuading someone that the earth is flat, as opposed to persuading someone that they should go to graduate school); but writers in general seem not to have heeded this advice, then or now.

However, until about 50 years ago, there were no reverse complaints — because, apparently, convince was not used (in print) with an infinitival VP complement until the 1950s! (No doubt it had been used this way in speech for some time before then.) And very quickly critics began complaining about the construction. But it spread, and by 1989 MWDEU was labeling it “a fully established usage”. And over the years the American Heritage Dictionary usage panels moved from being in favor of preserving the proscription against it to a majority accepting the usage; these panels tend to be conservative in their judgments, so that their votes are likely to reflect (explicit) usage opinions of a generation or more before, rather than the current practice of good writers.

Here’s the path of development:

1. There is an innovation — in this case, the extension of convince to the infinitival-complement construction, by analogy to persuade, a verb that it already shared both syntax (use with that and of complements) and semantics with. (You might ask why the innovation didn’t happen earlier — the basis for the analogy had been around for some time — but such questions are, for the most part, unanswerable.)

2. The innovation spreads, eventually to “better writers”. There’s then a competition between the older variant and the innovation, which can be resolved in a number of different ways: in particular, the variants can co-exist (while being differentiated subtly in their semantics, differentiated stylistically, or differentiated by being preferred by different groups of speakers) or one can spread at the expense of the other. Convince and persuade with infinitival complements seem to be in a stable equilibrium at the moment; below I will suggest that they are, or at least can be, subtly different semantically.

3. A usageist backlash develops. Usage critics are generally antagonistic towards innovations (on the grounds that they introduce unnecessary variants, eliminate useful distinctions, introduce ambiguities, whatever) and are likely to pounce on them.

4. The innovation continues to spread nonetheless. In the case at hand, MWDEU concludes its discussion of innovative convince by saying, “in another generation perhaps no one will care”. The proscription against it has become a zombie rule.

Similar paths have been taken by the mass determiner a lot (“I have a lot of difficulty with this idea”) vs. much (“I have much difficulty with this idea”) — Language Log discussion here and here — and the sentence-adverbial (speaker-oriented) hopefully — some Language Log discussion here — and many other innovations. (A lot is a 19th-century thing, hopefully, like convince to, a 20th-century thing.)

Now for a few words about the semantics of convince to vs. persuade to. I’ve heard two suggestions about a subtle difference between them; these could both be true for some people, or could be true for different people. The first is that we have another case of Y = X + something (Language Log discussion here): convince to is persuade to plus a suggestion of something like ‘against resistance’. The second is that persuade to might have a stronger component of linguistic effort than convince to — more involvement of organized argument. Such subtle differences would show up as a slight preference for one or the other of the variants in certain circumstances; for instance, both would predict a slight preference for persuade to over convince to with reflexive objects, as in

I convinced/persuaded myself to take the job.

But such subtle differences in (non-truth-functional) import are notoriously hard to detect; reflecting explicitly on them is likely to produce artefacts in judgments.

This entry was posted on January 16, 2009 at 8:13 pm and is filed under Usage. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.