LINGUIST List 4.95

Fri 12 Feb 1993

Disc: Subject/Object Asymmetry and Idioms

Editor for this issue: <>

Directory

1. , Subject/Object Asymmetry and Idioms

Dick Hudson asks why the statistical asymmetry between S-V idioms
(not common, apparently) and V-O ones (very common) should, as he
says, "not count" as an argument for some kind of fundamental
asymmetry between subjects and objects. I would say that there
are two points here that need to be considered, which are not
unrelated to each other.

(1) There have been several different theories which have predicted,
falsely as it turns out, that S-V idioms should be not merely infrequent
but impossible, such as the apparently discredited theory that idioms
have to be constituents. And it would certainly seem that Marantz and
Chomsky, at least, formulate their theory of subject-object asymmetry
in the same way.

(2) The very fact that such idioms exist in well-known languages
and yet were missed makes me somewhat hesitant to assume anything
about the possible significance of the rarity of these examples.
Maybe there are a lot more, or maybe there are few in English but
lots in some other languages with otherwise rather similar syntactic
structures. There are lots of apparent anomalies when it comes
to the frequency of different types of idioms. For example, why
should there be (apparently) so many more idioms like 'The jury is out on X',
where the idiom consists of the subject + the verb + a piece but less than
the whole of the VP than there are of the form subject + verb?
Or why should there be so many idioms involving less than the whole
subject phrase (with a nonidiomatic possessor, that is) than there
are involving the whole subject phrase?

In fact, historically, the whole thing is quite weird, because
there are types of idioms which were once claimed to be equally
impossible (like 'The jury is out on X'), but nowadays it seems
that no one is interested in how frequent they are. Actually,
they are (apparently) not all that frequent, but since they are
no longer considered crucial to the issue of subject/object
asymmetry or of the VP, no one seems to care. So, it seems
to me that there is a pattern here:

 First of all, for too long the facts have been used simply
 as a way of arguing for or against some theory which itself
 has little to do with idioms and which does not in general
 help explain anything about them.

 Second, as examples which were once thought impossible are
 uncovered, after a suitable period of trying to minimize
 their significance or to deny their reality, we simply come
 up with a new theory (which no longer predicts them to be
 impossible, but which does not explain why they were sufficiently
 rare so that for a long time everyone agreed they did not
 exist at all, and which, ipso facto, does not explain anything
 about the statistical distribution of different idiom types).

Why should the sole role of idioms in linguistic theory be as
"arguments" or "tests" for theories about constituent structure
and such?