Language Log

Sprinkled under the radar

Back in March, Joel Wallenberg e-mailed me a stunning antedating of the GoToGo construction -- as in "She's going to San Francisco and talk on firewalls" 'She's going to go to San Francisco and talk on firewalls', with the go of prospective be going to and motional go (in the GoAndVP construction) telescoped into a single word going -- from the 1920s back to 1864.  Chris Waigl then supplied examples from early in the 20th century.  So it looks like GoToGo has been around for quite some time, but without attracting attention or eliciting comment, until Charles Hockett noted "the recent colloquial pattern I'm going home and eat" in his 1958 textbook (p. 428).  It lasted a century or so completely under the radar, and (so far as I can tell) got only this one blip until close to the end of the 20th century, when David Denison (and later Laura Staum and I) began studying it.  Nobody even complains about it.  How could the construction maintain itself over such a long period of time without being noticed?


Let's start with the Wallenberg e-mail of 3/16/06, in which he quoted from an August 7, 1864 column Mark Twain wrote for the newspaper the San Francisco Morning Call:


And then the Waigl examples, all from quoted (but fictional) conversation:


Then there is one in Sherwood Anderson's Winesburg, Ohio (1919) and one in F. Scott Fitzgerald's The Great Gatsby (1925) and quite an assortment (collected by Denison) in movie dialogue.  If you look hard enough, you can find attestations all through the 20th century.  And if you ask people for judgments (in a careful way, as Staum did) you'll find a significant minority, maybe as many as 20%, who have little or no trouble with GoToGo sentences.  (I am in this minority, and was stunned to discover, five years ago, that most other people judge GoToGo sentences to be flatly unacceptable.  Most, but not all, of my colleagues in linguistics at Stanford.  My own daughter, even!  Oh, sharper than a serpent's tooth!)

So you can get unacceptability judgments when you explicitly ask people to judge sentences.  And every so often a linguist, or someone else unusually attentive to the details of language, notices that there's something remarkable about GoToGo sentences: they coordinate a finite VP with a non-finite (base-form) VP.  Otherwise, GoToGo sentences escape notice.  Why?

First, GoToGo sentences are rare, for two reasons.  One reason is that most speakers don't produce them at all.  The other reason is that, even if you're a GoToGo speaker, there just aren't that many occasions when you want to put together all the parts of a GoToGo clause: a clause making a future assertion with be going to (rather than will), about motion, with the specific motion verb go and with an expressed goal for that motion, and with the motion seen as the first part of a two-part event, the second part of which is also explicitly expressed (by a coordinated VP).  You could go for weeks or months without wanting to express such a thought in such a way.

Now, suppose you're NOT a GoToGo speaker.  Every so often -- maybe every few months, or even more rarely -- an example will come up.  The intentions of the speaker or writer will be clear, and the sentence will be quite close to what you might have produced yourself (with be going to or will plus motional go in the GoAndVP construction).  You probably will unconsciously take the example to be a minor inadvertent error and silently "fix" it in processing, the way people do with most speech errors.  So you won't notice anything.  (There's an alternative response, which I'll take up in a moment.)

From the other side, GoToGo speakers won't notice that you don't use the construction, because they understand the alternatives you do use, and because people don't notice small gaps -- or, sometimes, large gaps -- in other people's productions.  (If you're a need/want+Ved speaker -- "The garden needs watered" -- you can go for decades without realizing that lots of other people don't use this construction, ever.)

The result is that the construction can stay under almost everybody's radar, for any amount of time.

But how is it maintained from generation to generation?  Maintenance depends on the occasional person's having a response other than tacit correction to an occurrence of GoToGo.  It is, of course, possible for people -- young children, in particular -- to take this occurrence as evidence that the language has a construction (with non-parallel coordination) that they hadn't come across before; this way, the stock of GoToGo speakers can be constantly replenished, though it will stay small.

(It's also possible for the construction to be created afresh, by some kind of analogy, as Hockett suggested, or by telescoping, as I suggested.  If the 40-year gap between 1864 and 1904 isn't eventually filled by the kind of attestations we see from 1904 on, then we'll have to assume at least two independent innovations.  But others could have occurred.  There's really no way to tell.)

Since occurrences of the construction are so rare, we can expect that its spread to new speakers will be close to random, with few if any associations with social groups or contexts; normally, there just won't be enough examples for a learner to posit any sort of pattern in who uses the construction and on what occasions.  The construction will be (very lightly) sprinkled across the social landscape.  As far as I know, this is the case. 

In general, very low-frequency constructions can be expected to show social sprinkling.  It's not inevitable -- there could be a kind of in-group fashion for a particular construction -- but it's very likely.

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at May 3, 2006 06:32 PM