May 31, 2005

How soon before we see the complexities?

reducedq My posting on soon before missed at least one important complexity, which correspondents have now pointed out: how soon before in the examples I gave has how soon modifying before, but there are plenty of elliptical questions in which how soon does not: How soon before we have to leave? 'How soon will it be before we have to leave?'  These elliptical questions, which I believe are unproblematically acceptable, change the Google statistics somewhat, but without obscuring the main point I wanted to make with them.

More important, they provide a possible source favoring how soon before (with how soon modifying before) even for speakers who reject soon before otherwise.

In addition, one correspondent has suggested looking at future-oriented sentences like How soon before midnight will they meet? -- my earlier examples, like How soon before midnight did they meet?, were all in the past tense -- to see if their "basic query" (e.g., 'How soon will they meet?') improves their acceptability.  Whether or not this idea pans out, it is true that the Google examples of both types are heavily future-oriented.

I begin with e-mail from Chris Maloof, who pointed out the many elliptical questions among the how soon before cites that a Google web search provides.  (Marilyn Martin also offered an elliptical question example.)  These are of the form how soon  +  before-clause, and they lack both a subject and a verb.  On the other hand, the examples with how soon modifying before are just ordinary interrogatives, with fronted how soon before X (where X is a clause, as in (2b,c) below, or a NP object, as in (2a,d)), followed by a clause (in inverted or uninverted order, depending on whether the whole thing is in a main or subordinate clause: (2a-c) vs. (2d) below).  Some examples from Google:

Elliptical questions:

(1a)  "How soon before I can ski?" Ankle injuries are common...

(1b)  How soon before every state has conflicting laws on the subject? The states can't currently agree upon ages at this time...

(1c)  And how soon before we will see weirder instruments like Futures being traded on virtual currencies?

(1d)  If I order now, how soon before I get it?

Ordinary interrogatives:

(2a)  How soon before a grant deadline should I submit a protocol?

(2b)  How soon before I travel can I apply for my WHM visa?

(2c)  How soon before the quarter begins may a student be placed in homestay?

(2d)  ... they will help you determine what book to write, how quickly to write it, and how soon before publication you need to start your marketing efforts.

The elliptical questions should be generally acceptable, since they don't have soon (with its usual component of afterness) in combination with before.  So far as I know, this is the case, but it needs examination.  (At this point, I'm hoping to encourage someone else to take up soon before as a project.  My plate is pretty full.)

It turns out to be no easy task to estimate the relative frequencies of the two types; the Google cites are full of repetitions and near-repetitions.  (Many of the ordinary interrogatives are, like (2a-c) above, from faq's, which tend to have similar form.)  My first impression -- again, this should be investigated further -- is that the two types are roughly even, which means that the number of relevant how soon before hits should be cut roughly in half, and the relevant after/before ratio roughly doubled.  Even with this adjustment, the frequency of how soon before is still hugely less than the frequency of soon before without modification by how.  There's still something to be explained.

But the elliptical questions might not just be confounding data; they might have something to say to us.  They provide a pool of acceptable clauses beginning with how soon before and might therefore boost the acceptability of ordinary interrogatives of this form, even for people who don't otherwise accept soon before.  Something to consider.

Finally, Marilyn Martin has suggested looking at future-oriented sentences like How soon before midnight will they meet? to see if their "basic query" (e.g., 'How soon will they meet?') improves their acceptability.  This is, in effect, a suggestion that the future-oriented examples might be treated as amalgams of a how soon question (How soon will they meet?) with a neutral duration question (How long before midnight will they meet?).  I'm dubious about this suggestion, because the past-tense examples could be given a similar analysis (How soon before midnight did they meet? = How soon did they meet? + How long before midnight did they meet?), so I would predict no difference in acceptability between past and future examples.  Something else for someone to look at.

Still, the ordinary interrogatives from Google are, heavily future-oriented; the examples in (2) are all in the present tense, understood with a future orientation relative to the temporal reference point. (The elliptical questions are all future-oriented.)   Of course, the future orientation pretty much comes along with the genre of most of the Google examples, so it remains to be seen whether there is any actual association between how soon before interrogatives and future orientation.

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at May 31, 2005 01:52 PM