repost of 9 Oct 96 message from K.P.Mohanan

[Mary Dalrymple]

Apparently a lot of people on the list did not receive a copy of
K. P. Mohanan's recent message about COMP, so I thought it would be
worthwhile to repost the message to make sure everyone gets it.  I'm
sorry if this message duplicates what you have already received.

 - Mary


Subject: COMP and OBJ
To: lfg at lists.stanford.edu
Date: Wed, 9 Oct 1996 08:19:41 +0800 (SST)
From: "Mohanan K. P." <ellkpmoh at leonis.nus.sg>

	When we read Chris Pinon's message, we found ourselves
agreeing with him. But when we read Joan's message, we found
ourselves agreeing with her as well, though Joan disagrees with
Chris P.  We would like to make an attempt to explain what looks
like a paradox, by separating the different issues that appear to
underlie the ChrisP-Joan debate.

Let us begin with the following question:

QUESTION 1: Do syntactic frameworks need to represent grammatical
functions like subject and object in terms of entities not needed
independently for the representation of categories like NP and VP?
(`categories' = classes of words and phrases).

	The answer given by the Aspects framework and 1981 GB
framework is `No'. These notions, according to Aspects, can be `read
off' from the independently required representions of the categories
of S, NP, VP, and V. In contrast, the answer given by LFG and current
GB is `Yes'. LFG represents them as SUBJ and OBJ, while (current) GB
represents them as `spec of agr-s' and `spec of agr-o', where `spec'
and `agr-o' do not refer to classes of words or phrases. We take it,
therefore, that LFG and current GB have no disagreements on question
1. (Note: We are not saying that the LFG and GB ways of expressing the
notions of subject and object are notational variants: we are not
ruling out the possibility that there are some empirical consequences
to this difference.)

QUESTION 2: Are grammatical functions stipulated in the lexical
subcategorisation of veqrbs?

	Early LFG stipulated grammatical functions in the lexical
entries of verbs. Once LMT evolved as part of LFG, however,
grammatical functions are no longer stipulated in LFG, but derived
from argument structure through linking rules. This means that what is
stipulated are the lexical semantics of verbs and argument structure,
but not grammatical functions. (This result is important for the
points we wish to make.) Current LFG and current GB have no
disagreements on this score.

QUESTION 3: Granted that (a) we need to distinguish between
grammatical functions and grammatical categories, and (b) we do not
stipulate grammatical functions in the subcategorisation of verbs, is
it necessary to distinguish between OBJ and COMP?

	Question 3 was the one that we (=M, A, T) have been trying
to address. We first interpreted Chris P.'s message as addressing
question 3, but Joan's response suggested to us that he was
probably referring to question 1.

	Our own answer to question 3 is `No'. The primary argument
(Argument 1) that we advanced in our first message was this. If we
agree that (a) grammatical functions are not stipulated but are
inferred/calculated from argument structure, and (b) LMT is good
theory for calculating grammatical functions from argument structure,
then the distinction between object and complement is illegitimate,
because the feature system in LMT ([r], [o]) does not provide for this
distinction. None of the supporters of COMP have refuted this
argument. The seconary argument (Argument 2) that we advanced was that
if all COMP's are S's and all OBJ's are NP's, then the distinction
betwen OBJ and COMP is redundant: whatever we need to say about COMP
and OBJ can be stated as S-OBJ and NP-OBJ.

	Needless to say, one can maintain the distinction between OBJ
and COMP by complicating current LMT. As we understand it, this was
the move that Chris Culy was in favor of. His point was that there
exists a distinction between the clausal second arguments of `find'
and `discover', and that this distinction is best captured as COMP
vs. OBJ. He is therefore prepared to complicate LMT to derive the
necessary distinction between OBJ and COMP. Thus, he abandons the idea
that OBJ in English cannot be S's, and hence abandons the basis of our
Argument 2. He accepts the proposition that current LMT does not
permit the distinction between OBJ and COMP, and is willing to revise
LMT in such a way that it does.

	Our response to Chris C. was that the facts that he adduces in
favour of the OBJ/COM distinction have to do with making some
stipulations on the lexical entries on verbs, and given that
grammatical functions are not stipulated in current LFG, the OBJ/COMP
distinction cannot in principle yield an analysis of the facts. No one
has refuted this argument either.

	We did offer an alternative analysis of Chris C.'s facts as a
kind of optional bonus. Even if our analysis turns out to be
incorrect, or is vague, as Joan says, our main point remains
unrefuted. Given current LMT, the distinction between OBJ and COMP
cannot be maintained in LFG. If it turns out that in order to account
for the behaviour of `find' and `realise' we need to *stipulate* that
`find' takes a sentential COMP while `realise' takes a sentential OBJ,
we have to go back the pre-LMT LFG and assume that grammatical
functions are stipulated in subcategorisation (Question 2). Are we
ready to do this?

QUESTION 4: In addition to stipulating lexical semantics and argument
structure, is it necessary to stipulate grammatical category
information in the subcategorisation of verbs?

We suspect that the answer Chris P. gives is `Yes'. We also suspect
that he believes that we have been arguing for a `yes' answer to
question 4. Now, while we (T, A, M) are not averse to stipulating
category selection, we believe that we have not made any arguments for
this position in any of our messages.

One final remark. Chris P. says that the conclusion that we need to
abandon the distinction between OBJ and COMP will not constitute
progress in research communities outside LFG, because they have known
this all along. True, but such things are very common in disciplines
with multiple research communities. The notion `syllable' was absent
in the generative community for a long time. Did Kahn's (re)discovery
of the syllable constitute progress in phonology in general? Raising
from subject to object was taboo in the GB community for a long
time. Did the acceptance of this raising constitute progress in syntax
in general?

Mohanan, Alex, Tara