Language Log

That'll teach me ...

... to try to post brief, quick things to Language Log.  Either I have to do it up brown or I have to pepper the posting with qualifications, provisos, and warnings.  So I seriously messed up on the teach you to posting, in at least two ways, and now I have a pile of e-mail messages about the details.

The two things: (1) by not contextualizing and elaborating on my observation that I had noticed the undernegation in


I gave the impression that I was claiming the idiom was new to me and that it was in fact recent; (2) by giving the shorthand gloss 'teach X not to VP', I gave the impression that I thought this was a full analysis of how the idiom is used.

I got lots of e-mail messages (astonished or corrective) on each of these points, and on a couple of others.  Here's the history ...


Item (1).  Linguists are forever "noticing" familiar things -- that is, noticing that some familiar phenomenon has an interesting twist to it.  What are the regularities connecting the form and meaning of these expressions?  How might they have come about?  And so on.  That's what happened to me with


Yeah, sure, I was familiar with this way of talking, and had been for a long time.  But I was moved to reflect on the fact that such expressions, which lack any negative elements, are nevertheless understood as having negative import: they are indeed about teaching and learning, but the lesson to be learned is NOT to VP.  (Yes, there's more.  I'm saying only that this is PART of the story, not that it's the whole story.)

So we have an idiom (which I'll call Teach To), in which the meaning of the whole cannot be read off from the meanings of the parts using regular compositional principles of meaning.  And Teach To is undernegated: no negative in the syntax, negative import in the semantics/pragmatics.  (Undernegation comes up every so often here on Language Log, since it was first mentioned under that name by Ben Zimmer in December 2005; search on "undernegation" for the full set of postings.)

In my first version of the posting, I said that we hadn't mentioned Teach To before, but that was wrong.  Somehow my search on the site failed to pick up the 2004 Chris Potts piece in which Teach To got a brief mention, but Mark Liberman quickly set me straight, and I revised my posting.

Still, I failed to include the boilerplate:


And I failed to state EXPLICITLY that the usage was familiar to me and had (as far as I knew) been around for some considerable time.  Very bad step.  (Students of implicature might want to look at what went wrong here.)  So I got tons of e-mail, some from old friends, expressing disappointed bewilderment about how dense I was about English usage.

My mail did include some cites:

NOAD2 has much the same entries as the Oxford dictionaries Cauchi cites, but with (of course) a different illustration for the "informal" Teach To: I'll teach you to mess with young girls!

In fact, it turns out that the OED has the relevant sense, not marked as colloquial or informal:

Moral: always check out the dictionaries.  (It's not in AHD4, by the way.)

Item 2.  Another pile of e-mail came in about how the point of Teach To us to encourage the addressee to AVOID VPing by alluding to the unfortunate consequences of VPing.  That's certainly what Teach To conveys.  When I say


I convey that I should avoid VPing (posting brief, quick things to Language Log) because doing so leads -- in fact, has led -- to unfortunate consequences (in this case, being widely misunderstood).

When someone says


they (probably) convey that the addressee should avoid VPing (talking back to them) because doing so will lead to unfortunate circumstances (there is a threat of forthcoming punishment for the recent action of VPing).

This is still far from the full story, but it's better than my offhand gloss.

For more turns and twists, see the Linguist List discussion from 4 October through 3 November 1993, initiated by a query from Laurie Bauer: postings 4.795. 4.822, 4.844, 4.859. 4.873, 4.874, 4.884, 4.898, 4.909.  (Thanks to David Denison for pointing me to these exchanges.)  Like most mailing list discussions, this one wanders a good bit, but many good points are made.

Item 3 (not previously announced): odds and ends.  Here in passing are a few topics from my mail.

3.1.  Learn.  Lots of readers pointed out that they allow, or prefer, learn instead of teach in the Teach To idiom (as in the Wind in the Willows quote above).  This is just what you'd expect in varieties that have learn 'teach'.  But there may be people who don't generally use learn 'teach' but nevertheless have learn as part of a package with undernegation, in a kind of importation of elements from vernacular speech.

3.2.  Frequency.  Many readers wrote to tell me that Teach To is enormously frequent, much more frequent than any explicitly negative counterpart (note Andrew McGuinness's comment above).  I'm sympathetic.  But: I no longer trust ANYONE's off-hand estimates (even my own) of the frequencies of variants; your/my impressions are so biased by expectations, beliefs, selective attention, etc. that they can't be trusted.  Ya gotta do the numbers.

That is, I suspect my correspondents are right, but that has to be demonstrated.

Also, in this case, it's hard to know how to factor in the enormous number of occurrences of will teach X to VP that are entirely literal (an offer: "I'll teach you to factor polynomials").

3.3.  Sarcasm.  Some readers wrote to say that Teach To is sarcastic: it has a characteristic intonation and a characteristic import, it is claimed.

Well, as with could care less, Teach To CAN have such an intonation and/or such an import, but it very often doesn't have either.  That is, the idiom isn't intrinsically sarcastic.  Like all sorts of other expressions, it can have sarcasm attached to it.

(David Eddyshaw attempted to connect the absence of negation to sarcastic intonation, in much the way that Steve Pinker proposed for could care less.  The idea is that the intonation marks the speaker's intent, so an explicit negative isn't necessary.  I refer you to the Language Log discussions of could care less for a critical treatment of this idea; an archive of postings is here.)

3.4.  Will.  In the Linguist List discussion, John Lawler noted that the will is part of the idiom.  As with the could of could care less or the would of would rather, a modal is part of the idiom.

3.5.  Negative polarity.  In the same Linguist List discussion, John Lawler pointed out that though Teach To is implicitly negative, it doesn't license negative polarity items (like anyone and anywhere):


Contrast the implicitly negative deny:


So there are several different flavors of implicit negation.

Final note: this has been a quick pass through some of the complexities of Teach To -- more adequate than my unfortunately brief original posting, but still far from the whole story.  Still, I think it's useful to say some things on the topic, without trying to say everything that could be said.

Posted by Arnold Zwicky at January 6, 2008 03:21 PM