LINGUIST List 5.184

Sat 19 Feb 1994

Disc: *These men and women

Editor for this issue: <>

Directory

1. Leo Obrst, Re: *These men and women
2. , These man and woman

Concerning J.B. Johannessen's posting and the subsequent discussion
(5.115, 5.123, 5.136) on examples such as the following, I would like
to comment briefly.

 1. a. There was/*were a man and two wo men in the room.
 b. There were/*was two women and a man in the room.
and
 2. a. A man and two women were/*was in the room.
 b. Two women and a man were/*was in the room.

In my dissertation (U. Texas-Austin, 1993, "Coordination and Concord
in Generalized Categorial Grammar"), I look at some of these kinds of
examples, and advocate a theory to explain them in terms of pronominal
incorporation within a gen. categorial framework. In this view, verbs
in languages with subj/verb agreement have a null or
partially-specified incorporated pronominal in one argument slot
(possibly the remnant of topicalization processes like
left-dislocation ala Givon (1975) and others). The verb, which is itself
a function, thus contains a function-valued argument. Another way of
looking at this is that s/v agreement is the result of function
composition along with a semantic/pragmatic resolution of features,
the latter process computing a subsumption point, beneath which lies
possibly a range of interpretations (of the discourse referents).
Pleonastic/expletive pronominals occurring in existential or
presentative constructions like (1), having null or minimal featural
content, compose with the verb prior to the verb's combining with the
(postverbal) NP, which has non-null featural content. Variation in the
agreement acceptable to speakers in (1a), for example, could be due to
the order of combination of the postverbal conjuncts with the verb
(and resolution of the featural content), as represented in the
following (abstracting away from the kinds of rules involved here):

3. a. there was a man and two women
 pro verb np conj np
 ---------- -------
 x y
 ----------
 z
 ---------------
 s

 b. there were a man and two women
 pro verb np conj np
 ---------- ---------------
 x y
 ---------------
 z
 ---------------
 s

In addition, perhaps some of the variation in the inverted forms are due
to there being at least two differently typed conjunctions: the ordinary
general conjunction and an NP-only comitative-like form (the latter
perhaps accounting for the "farthest-conjunct" strategy in some
languages and constructions, as reported by Corbett (1991), et al.)

Any "explanation", I think, also has to address questions of topic/focus,
desubjectivization, etc.

Leo Obrst

Perhaps I should rephrase my point to make it clearer: the very
fact that in order to account for *these man and woman, we need
to look at the features of the conjuncts is inconsistent with
the spirit of the constituent-structure view of syntax, although
admittedly the correct languages can apparently be generated.
The whole point of constituent structure is that you put pieces
larger than a morpheme or a word and smaller than the whole
sentences together and then assemble those into yet bigger pieces
and so on. In other words, once you put 'man and woman' together
you have that as a piece and should thenceforward operate with
that piece without looking back. Of course, Bloomfield could
have been wrong to adopt the constituent structure model if
it turns out that languages allow themselves to look back. What
I referred to as the dependency model is simply a more general
model in which you CAN look back. Of course, you can "cheat"
by encoding the internal structure of constituents into features
which you attach to these constituents, but that is precisely
what the constituent structure idea was supposed not to do.