complements

[Tara W Mohanan]

The issues involving OBJ vs. COMP are becoming more interesting,
particularly with Joan's last message.  However, to make sure that we are
not talking at cross purposes, a few clarifications are in order.

When we engage in debates on theoretical constructs in a framework, it is
important that we keep track of the meaning of the term acquired in the
overall context of the framework. When the contexts change, the meaning of
the term changes, and the term refers to a slightly different construct,
resulting in what Kuhn calls incommensurability. The purpose of the
following message is to point to the shifts of meaning in the use of the
term COMP.

PART I

When we first suggested that the distinction between OBJ and COMP be
abandoned, what we had in mind was that we should not use grammatical
functions to distinguish between the second arguments of "discover" in (1)
from that in (2):

(1) The children discovered [the cookies].
(2) The children dicovered [that there were cookies in the jar].

Instead of assuming that "the cookies" is an OBJ and "that there were
cookies in the jar" is a COMP, we said that both should be treated as OBJ.

Our proposal was made in the context of a framework which crucially makes
assumptions A and B.

(A) Only an NP can be an OBJ (in English).
(B)  Only an S/S-bar can be a COMP.

We hope we are not mistaken in assuming that Chris Culy abandons assumption
(A), and treats the nominal and sentential arguments in (1) and (2) as OBJ.
However, he holds that COMP is necessary to distinguish between (2) and
(4):

(3) The children found [the cookies].
(4) The children found [that there were cookies in the jar].

In claiming that the sentential argument in (2) is an OBJ but that in (4)
is a COMP, Chris abandons Assumption A. (We take it that Chris subscribes
to B). If so, the word "COMP" no longer means the same thing we started out
with. Hence, the issue itself changes a little.

PART II

Joan's last message is not about the need to make a functional distinction
between the post verbal constituents of (3) and (4) (or between (2) and
(4)), but about the need to make a functional distinction between the post
verbal constituents of (5) and (6) (examples slightly altered for overall
parallelism).

(5) John saw [a cook].
(6) John became [a cook].

She claims that it is necessary to assume that the NP "a cook" is an OBJ in
(5) but a COMP in (6). If we grant that this distinction is valid, it still
does not follow that it is valid to assume a functional distinction between
the post verbal constituent in (1) and (3) on the one hand, and the post
verbal constituent in (2) and (4) on the other. That is to say, the
assumption that "become" takes a COMP while "see" takes an OBJ does not
entail the assumption that the post verbal S of "find" is a COMP while its
post verbal NP is an OBJ.

In many grammatical descriptions of English (e.g. Quirk and Greenbaum), the
expression "a liar" in (8) is treated as a "complement" in contrast to the
the expression "a cookie jar" in (7) which is treated as an object.

(7)     The children found Bill [a cookie jar].
(8)     The children found Bill [a liar].

If we claim that "a liar" in (8) is a COMP and "a cookie jar" in (7) is an
OBJ, we are expressing the same intuition. (Joan does not use the
construction in (8) in her message, but the contrast is sharper between (7)
and (8) and we assume that we are not misrepresenting her position.)

Treating the final NPs in (6) and (8) as COMPs crucially involves
abandoning  assumption (B). The meaning of COMP has then changed in yet
another way within the framework, both with respect to previous LFG
assumptions and with respect to Chris Culy's proposal.

The claim that we (TAM) had made was that it is not necessary to
distinguish between the post verbal constituents of (1) and (2) in terms of
grammatical functions. Whether or not we need to make a functional
distinction between the third arguments of (7) and (8) is a separate issue.
Furthermore, our arguments against the distinction between OBJ and COMP
were made in the context of a framework that assumes both A and B. Clearly,
these arguments need to be re-examined if we abandon either A or B.

PART III

Until recently in LFG, the expression "a liar" in (8) as well the other
examples of COMP that Joan cites in her message were treated as X-COMP.
Thus, the clausal argument in (9) was treated as a COMP while the clausal
argument in (10) was treated as an X-COMP:

(9) John promised Bill [that he would behave].
(10) John promised Bill [to behave].

The crucial distinction between the two functions are the assumptions in
(C) and (D):

(C)     The subject of an X-COMP is functionally/obligatorily controlled.
(D)     The subject of a COMP cannot be functionally/obligatorily controlled.

The post verbal elements that Joan treats as COMPs in her message were also
treated as X-COMPs earlier.

(11) Mary looks [proud of herself].
(12) Jogging kept Susan [in a bad mood].

Our initial proposal to abandon the distinction between OBJ and COMP was
made in the context of a framework that subscribes to both (C) and (D). In
assuming that "proud of their homework" in "The children are proud of their
homework" is a COMP, we are abandoning assumption (D). If so, the issue has
moved from (E) to (F):

(E) The distinction between OBJ and COMP should be eliminated.
(F) The distinction between COMP and X-COMP should be eliminated.

If we understand her recent message correctly, Joan subscribes to (F) and
collapses COMP and X-COMP. If this is so, Joan and us (TAM) agree on one
thing:  we do not need a three-way distincton among OBJ, COMP and X-COMP.

Mohanan, Alex, Tara

----
Tara W Mohanan, Dept of English & Literature, NUS