Language Log

In or under

It's a working assumption of linguists that when there are alternative expressions, the choice between them is neither completely free nor completely determined.  Extreme prescriptivists would like to maximize determination, and to make the basis for determination explicit: in any context, only one alternative should be acceptable; there should be a good reason for this choice; and it should be possible to articulate that reason.  Pick either "in the circumstances" or "under the circumstances", and be ready to justify your choice as the right one.

Linguists, studying language in actual use, point out that there is an enormous amount of variation in these choices (variation between speakers, and within a given speaker, variation in different linguistic and non-linguistic contexts); that the choices are mostly matters of (rather subtle) preferences rather than crisp decisions; that these preferences interact with one another in complex ways; and that hardly any of these preferences are easily accessible via conscious reflection.  You can ask people whether they would choose "in" or "under", and why, and with the options presented side by side this way, they're likely to express a preference and to produce some rationale for it, but there's absolutely no reason to think that their accounts are reliable.  You have to look at what they actually do. (Judgments based on side-by-side comparisons are tricky and context-dependent even in situations where you might think they'd be straightforward, like the Pepsi Challenge discussed by Malcolm Gladwell in Blink, pp. 158-9.)

In my recent posting on "in the circumstances" vs. "under the circumstances", I started to move beyond side-by-side comparisons and into some (admittedly crude) statistics on actual use, uncovering a modest preference, in a Google web search, for "under" as opposed to "in", contrary to the advice of some prescriptivists.  Now I've gone a bit further, and it looks like there's some interesting texture to the variation in this preposition choice.

A brief recap: the raw Google webhit figures were:

Now, I'm quite sure that there are between-speaker differences in the choice of "in" vs. "under" with "circumstances", but corpus searches like this aren't going to find them.  And for individual speakers, I suspect that there is variation according to modality, style, and a number of other "external" factors, but again searches like this won't turn them up.  What we can look at are factors that have to do with linguistic context, and here we strike paydirt very quickly.

First, MWDEU claims that when "circumstances" means 'financial situation', "in" is the preposition of choice, and "under" is rare.  Googling supports this claim:


This overwhelming preference for "in" extends to "circumstances" in the sense of 'personal situation' in general.  With some possessive determiners:


(There might be something in the difference between the pronouns, but I'm not exploring that today.)

On the other hand, the modest preference for "under" extends to PPs with the demonstrative "these" (my thanks -- I guess -- to Elizabeth Zwicky for suggesting that I look at determiners other than "the"):


and then balloons for interrogative/relative "which" and interrogative "what":


Now a surprise.  For demonstrative "those", "in" is preferred:


As it turns out, the figures for "in" here are somewhat inflated by occurrences of "those circumstances" with a following relative clause in "where" or "in which", a context in which -- another surprise -- "in" is almost categorically preferred to "under":


Removing these occurrences from the overall "those" count leaves:


There's still a fairly sizable preference for "in" with "those", in contrast to "these". 

(There are only small numbers of other relative clause types with "those circumstances", for example:


I wouldn't try to draw any conclusions from these small differences.)

When we turn to quantity determiners, there seems to be a general preference for "in" over "under":


which becomes very strong for "many":


and almost categorical for "a few":


but is utterly reversed, in favor of "under" (almost categorically), for "no":


In summary: the Google data suggest that "under" is preferred to "in"


but that "in" is preferred to "under"


This just scratches the surface of the phenomenon, but it's enough to indicate that several effects are probably going on.  As usual, the facts of usage are complex, subtle, sometimes surprising, and not easy to derive from first principles.

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at October 28, 2005 05:33 PM