Predicate Clefts in LFG

[Seth Alfred Cable]

Dear Joan (and other LFG-Listers),

The construction I have in mind is indeed the "Verbal Cleft" discussed in
Koopman's thesis (which she calls the "Wh-type of V-movement").  Although
the details vary between languages, it has the following abstract form:
(i) the focus/topic field contains a verbal root, possibly bearing
infinitival or nominalized morphology; (ii) the main clause contains a
copy of the verbal root in the focus field.  An example from Haitian would
be the following:

Se manje Jean manje sel.
   eat   John eat   salt
(roughly) "It's EATING John did to the salt"

One of the most confounding puzzles which this construction presents to
movement-based accounts of long-distance dependencies is that, although
the relationship between the initial V and the V in the main clause
satisfies all the criteria for A'-movement, there is NO TRACE anywhere in
the structure.  Movement-based accounts of this structure must play a
game of "find the missing gap": Koopman proposes that the lower V is a
kind of resumptive pronoun, others propose that the cleft is derived from
a doubled-V construction, another idea is that this is a kind of
"do-support" to prevent tense morphology from being stranded.  Each of
the many proposals is often reasonable for the particular languages
under the author's consideration, but fails to have any cross-linguistic
validity.

When we take a step back from the melee, though, one can't help but feel
that the problem might well be that our "criteria for A'-movement"
really don't have anything to do with "displacement" at all.  We first
discovered these these properties of "island-sensitivity" and "licensing
of parasitic gaps" in the context of "wh-movement" of NPs.  But, suppose
that history was different; suppose that English had a predicate cleft and
we observed these properties in predicate clefts FIRST.  I think a
natural view of the phenomenon might have been (might be) something like
the following:

(a) Both the gap in NP-displacement and the V in a predicate cleft bear a
feature which says something akin to "Look out!  There's something
identical to me higher in the tree which bears the discourse feature
TOPIC/FOCUS".

(b) The exact content of this feature, the kind of higher tree-positions
it can "point to" are constrained by universal grammer.  The nature of
those constraints derive the island sensitivities.

(c) The reason why the feature appears on an NP "gap" and not on an NP,
but appears on a V and not a "V-gap" is that (i) for a particular
catagory, the feature seeks out the "least marked"/"most featureless"
member of that catagory; (ii) the catagory NP has NP "gap" (no real
phonological, semantic or syntactic features), but V doesn't have any
"gap"; there is no "verbal trace".  VP has no "pronominal elements" at
all; thus, for verbs, this feature seeks out only full Vs.

This rough grammatical picture, of course, fits very, very nicely within
LFG (and for more subtle reasons, I think it does better in LFG than in
HPSG).  So, I thought that I should get in touch with what researchers
working within the LFG tradition have independently developed regarding
predicate clefts.  There may be aspects to the construction that might
actually force a very different analysis within this framework from what
I am imagining above.  More than anything, though, I need to understand
and build on the work of those who have studied these phenomena within the
conceptual structure of LFG.

Thus, even work on the more standard clefts and pseudoclefts might be of
relevance here, especially since "predicate clefts" seem to take on a
'biclausal structure' in many Kwa/Kru languages.

Please, any information anyone out there has would be greatly greatly
appreciated.

Best,

Seth.


On Thu, 13 Nov 2003, Joan Bresnan wrote:

> Hi Seth.
>
> In my textbook, pp. 247-8, there is a brief discussion and analysis of a possibly relevant
> Icelandic construction which I call "focussed complement extraposition" (`Sigga's opinion is
> that self lacks talent'--due to Joan Maling).  These constructions show connectivity effects
> which I capture with f-structure sharing, and would be relevant to similar connectivity effects
> in pseudo-cleft constructions, I think.
>
> If, however, by "predicate clefts" you mean the kind of  West African construction discussed
> by Hilda Koopman  in her book on verb raising, the above discussion would not be helpful.
>
> Perhaps you can explain a little more about your focus of interest?
>
> Joan
>
> On Mon, 10 Nov 2003 20:17:28 -0500, Seth Cable <scable at MIT.EDU> wrote:
>
> >Hello,
> >
> >I'm working on the syntax of predicate clefts, and I would be very
> >interested to know the work that has been done on them in the LFG tradition.
> > What might be some good works in which the predicate cleft has been given
> >an LFG treatment?
> >
> >Best,
> >
> >Seth.
>