Auxiliary Reduction: it all depends on the context

[Arnold M. Zwicky]

On Sep 19, 2004, at 1:53 PM, Dennis R. Preston wrote, following up on
my report of "And you're?", contrary to the usual constraints on
Auxiliary Reduction:

>> While I suspect that arnold is right to seek for phrase structure
>> generalizations and good behavior of certain related morphophonemic
>> processes in most cases, I wonder if the search for a integrated
>> general account of such discourse ellipsis facts will be fruitful.
>> Seems to me that, given an appropriate discourse context, amost any
>> element might appear.
>
> Teacher: What are these things?
>
> Kid: A ball, a doggie, a..,a...
>
> Teacher: A...
>
> But I would not go on to seek integration of that performance with
> almost anything else I know about articles (except to predict what
> sort of item the teacher expected to find in the post-a slot), and
> bare articles are surely as rare as contractions in phrase final
> position.

perhaps you'll think this is over-subtle of me, but i think there are
two rather different things going on here.  what dInIs cites is a
simple attempt by someone to prod their interlocutor to produce the
next word or words.  i think what was happening in the exchange from
CSI was a conventionalized type of "quiz" or "test" question (such
questions are used in classroom situations, but not only there), in
which the prod is for a *constituent*, not merely some word or sequence
of words.

there are two types.  one has a (prosodically marked) wh word in the
position of the questioned constituent, as in the wonderful Monty
Python quiz show exchange:  "Now Karl [Marx], number two. The struggle
of class against class is a *what* struggle?"   Karl Marx: "A political
struggle."  or, in an introduction context: "I'm N N.  And you're
*who*?"  or: "So you went down into the basement *why*?"

the other type is elliptical, with the questioned constituent omitted,
as in our "And you're?"  or: "So you went down into the basement
because?"  i believe that the ellipsis must be sentence-final; the wh
word can occur in any position open to (in situ) wh words in the
language.

so wh quiz questions are not available for constituents that have no
corresponding wh word.  particles for instance.  "You came
*what*/*how*?" can't be used as a quiz question asking about which
particle is appropriate in combination with the verb "came" (came out,
came over, came in, came up,...); that's because there's no wh word for
questioning particles.  [note: there are also reclamatory questions,
which have somewhat different syntax and pragmatics.]

i believe that elliptical quiz questions are subject to the same
constraint; "You came?" is just as bad a quiz question as "You came
*what*/*how*?"

in any case, i'm claiming that quiz questions have a syntax and
pragmatics of their own, and are not mere prods for additional words.
the two types of quiz questions are, in some sense, constructions of
english.  so the fact that some of them don't obey the ordinary
constraints on AuxRed really does make a puzzle.

well, having thought about this a bit, i do have an answer to the
puzzle, though it might not satisfy everybody.  the answer requires
making a distinction between (at least) two types of regularities in
the pairing of syntactic form with semantics/pragmatics.  call the two
collections of regularities Module 1 (M1) and Module 2 (M2).  [in
public presentations for many years now, i've been calling M1 the
Grammar and M2 the User's Manual, following chris culy's usage in his
1996 Language Variation and Change article on null objects in english
recipes (which began as a 1987 qualifying paper at stanford).  but i
won't insist on those labels here, because they confuse, mislead, or
distress many people.]

the crucial thing is that expressions licensed by regularities in M2
can violate constraints applicable in M1: transitive verbs that require
direct objects in M1 (like "make") can lack them according to M2
(recipe: "Make into small balls"); finite clauses that require subjects
in M1 can lack them according to M2 (english is not a pro-drop
language, but subjects can be omitted in certain kinds of informal
writing: "Saw Kim last night."); and so on.

in my very large file of Rejected Abstracts, there's one on the
conditions on "gonna", which are (in M1) more restrictive than those on
"goin(g) to".  in particular, "gonna" is subject to an Explicit Head
Condition, requiring an explicit head verb that's a form of "be":
with my thesis goin(g) to be finished soon,...
*with my thesis gonna be finished soon,...
nevertheless, in informal-speech truncated sentences of several types
(licensed in M2), "gonna" can occur without a form of "be":
Anyone gonna eat that shark steak?
Gonna take a sentimen'tal journey.

the big generalization is: stipulations in M2 override those in M1.
and that applies to elliptical quiz questions with respect to the
constraints on AuxRed.

arnold, waxing heavily theoretical