Filler-gap mismatches

Wed May 9 01:23:40 UTC 2001

Hi Yehuda,

> the LFG and HPSG approaches, so I'll leave that to Ron :) But it seems to
> me that representing everything as feature structures (AVM) is a claim
> that, while the specific features may be different, all these different
> dimensions are essentially the same.

I really don't think anything substantive is at stake here. In LFG, the
models, are tuples: <Cstruc,Fstruc,Phistruc,....>. And LFG specifies what each
of the various strucs are like and how they correspond. In HPSG, the models
are (for the moment) feature structures, directed graphs of a particular
kind. An HPSG model for an LFG-theory would look like this:

       PHISTRUC__>Phistruc
      /
sign o--CSTRUC-->Cstruc
     |\
     | \__FSTRUC__>Fstruc
      \
       ...

HPSG specifies what each of the various strucs is like and constraints placed
on signs affect the possible correspondences across strucs.

> LFG divides f-structure attributes into grammatical functions and features.
> Grammatical functions are "functional" in the sense that they specify the
> purpose(s) of various elements in the syntax. Some elements in the syntax
> function to express the arguments of predicates (SUBJ, OBJ, the OBL family
> of functions, etc.), some function as adjuncts, and some to express
> relationships between the clause (or elements thereof) and larger syntactic
> units (SUBJ) and discourse units (FOCUS, TOPIC).

and some function as NPs, others as PPs. Function is a pretty general notion...

> Features provide the content of the functions (the value in the AVM). The
> PRED feature ties the argument-taking properties together and represents
> the general meaningfulness of elements in syntax (not their meaning, which
> is semantics, but the fact of their meaningfulness, which is relevant for
> how things function in the syntax). In a sense, the PRED is the f-structure
> "head

I'm sorry. You lost me there. And since `head' is a notion native to
c-structure (cf. Bresnan's discussion of Lyons in several papers in the 70s),
wouldn't head-like behavior in f-structure be better accounted for by a theory
that unified c- and f-structure?

>  Other features, such as PERS, NUM, and GEND, provide other aspects
> of the content of the functional elements of a sentence. Features like CASE
> (and PCASE, if it really is distinct from CASE) help determine the function
> of the element of which they are a part.

This is something I've always been perplexed by. PERS, NUM and GEND are
used in many languages to encode distinctions that are grounded in
semantics (aggregate-->plural, male-->masc, etc) but which have become
conventionalized in certain ways that make a purely semantic analysis
impossible. We need a notion of `semantic' and we need a notion of
`conventionalized'. But why do we need a notion of `f-structure' to deal with
person, number and gender distinctions?

Likewise, traditional grammar speaks of case functions: the `dative of
purpose', the `dative of interest', and so forth, but this is not what LFG
encodes in its CASE value, which seems rather to correspond to the `formal'
(i.e. `non-functional') side of case. I think many HPSGers are confused about
why the dichotomy between c-structure and f-structure is so clear to LFGers.
Perhaps this is related to the fact that people in the HPSG community tend to
have been more involved in semantic research than LFG people (I know there are
exceptions in both directions, please nobody take offense). For this reason,
the PRED feature also seems redundant to many HPSGers. I think this is because
they find it sufficient to construct such relations in the semantic analysis
(Ackerman and Webelhuth are of course to be singled out here as the people who
have really tried to confront this issue).  LFGers, on the other hand, seem
not to be bothered by any potential redundancy here, as the (very interesting)
work on LFG semantics tends to be something that is so modular that a
syntactician can safely ignore it, if (s)he is so disposed. That seems less
true of work in HPSG, though this is only an impression.

> constituents or f-structure units. There also may be mismatches between
> c-structure features and f-structure features. For example, in languages
> with differential object marking we might want to say that "unmarked"
> objects are Caseless in c-structure and have the feature [CASE ACC] in
> f-structure.

This seems to imply that there are two CASE features: one in c-structure;
one in f-structure. Is that really what you mean?

Carl:
> >I hope not and think not. What I hope and think will happen is that
> >further evidence will resolve conflicts between the two frameworks and
> >cause them to both evolve into things that are more similar to each
> >other. I cannot buy the view that there are two (or more) essentially
> >static total packages and we are waiting to see which total package is
> >best.

I emphatically agree.

> enough known for there to be a common understanding of their import. The
> filler-gap mismatches which the subject line of this thread refers to are a
> case in point. They are old news for LFG, and not considered a problem.
> However, they are apparently (at the very least) not as well known in HPSG,
> and I would imagine even less known in derivational circles. So I don't see
> that we are yet ready to converge. In the meantime, the various frameworks
> *do* make different claims and different predictions, and it is worth

I agree with Bob Levine's response to you about this. Note, in addition,
that the `assure' facts Bob discusses were actually discovered by Kayne,
a renowned `derivationalist'. In fact, most of the mismatch facts
we're discussing here were unearthed by transformationalists. I also wanted
to add that, at least in cases like this, `frameworks' don't make predictions.
Particular theories, some of which might be formulated in more than one
framework, make predictions. Getting clear on this is important for any
discussion that tries to cross framework boundaries...

All best,
Ivan

-------------------------------------------------------------------------
Ivan A. Sag
Professor of Linguistics
Director: Symbolic Systems Program (2000-2001)

Email: sag at csli.stanford.edu
WWW: http://www-csli.stanford.edu/~sag/sag.html

Dept. of Linguistics              CSLI - Ventura Hall
Fax:   650-723-5666               Fax: 650-725-2166
Office: MJH 040B                  Office: Cordura 228
Phone: 650-725-2323               Phone: 650-723-2876
-------------------------------------------------------------------------
Stanford University - Stanford, CA  94305  USA
-------------------------------------------------------------------------