information

[Ivan A. Sag]

Dear Achit,

>   I would be grateful if someone can respond to my
> following questions:

I'll try.

> 1-What are the equivalent of these concepts of GPSG in
> HPSG:
>
>   - LP rules (how can I express the order constraint
>     between constituents in HPSG)

GPSG-style LP Rules have been used in HPSG, but there has been considerable
work developing a new framework for LP rules called `Linearization Theory',
which allows the words of the daughters of an ID rule (= HPSG schema) to be
`shuffled' instead of just concatenated. The pioneers here are Mike Reape and
Andreas Kathol, though others have also contributed in significant ways.
If you've never seen this kind of analysis, you might start with Kathol's
webpage or his book:

Kathol, Andreas. 2000. {\it Linear Syntax}. Oxford: OUP.

Others will no doubt have more papers to suggest....

>   - FFP:  Foot Feature Principle

This is roughly equivalent to the NONLOCAL Feature Principle of Pollard
and Sag 1994, though Pollard and Yoo 99:

@Article{pollard-yoo99,
  author =      "Carl Pollard and Eun Jung Yoo",
  title =       {A Unified Theory of Scope for
    Quantifiers and {\it Wh\/}-Phrases},
  journal =     {Journal of Linguistics},
  year =        "1998",
  volume =      "34",
  pages = "415--445"
}

reject the existence of any uniform principle governing the inheritance of
specifications for nonlocal features.  And in Ginzburg and Sag 2000, the
NONLOCAL Feature Principle is reduced to the Generalized Head Feature
Principle.

> 2. Can I introduce a new ID Schémata  when applying
>    HPSG to certain languages for example arabic
>    language,or must I use only the 06 ID schematas
>    and only them ?

Those 6 schemata are just the ones that Pollard and Sag found useful in
analyzing English in 1994 (well, 1991, actually, but never mind...:-)).
Neither of us would limit ourselves to those schemata now. Why should you?
(Why should you even if we did...?)

> 3. I want have more information about the
>    "generalized HFP" in the current HPSG version?

The GHFP is an idea explored in Ginzburg and Sag 2000. It relies crucially on
default constraints. This is not the GB community (nor any of its
derivatives), hence there is no `current HPSG version'. Defaults remain
controversial and there is ongoing discussion between roughly two groups of
HPSG researchers: those who use defaults and those who don't. At the recent
HPSG meeting in Trondheim there was some very interesting discussion about
this.  Olivier Bonami's excellent paper on French morphology, for example,
used default principles and, as Detmar Meurers observed, the results he
achieved simply stand as an elegant systematization of a complex body of
data; a theorist who wants HPSG to be free of default constraints may well
try to develop an alternative. I think the discussions in Trondheim laid the
groundwork for much improved discussion across the varying viewpoints. And I
fully expect someone to work out an alternative to Ginzburg and Sag's GHFP
that avoids defaults (actually, the appendix to G&S already shows one view
of what such an account might look like...).

> 4. If we make a comparison between GPSG and HPSG what
> would be the main differences?

I believe the default logic used to formalize GPSG was fundamentally flawed.
The notions of `privileged' instantiation, `overriding' defaults, etc. were
not worked out properly (there is more recent work on defaults that
MIGHT fill this gap...). Also, since (1) NLs are trans-context-free
(cf. Culy, Shieber, Manaster-Ramer, Kac, and others -- people argue over who
showed what and who showed it first...)  and (2) GPSG was meant to be weakly
equivalent to CFGs, I guess this is a rare case of a linguistic theory
actually being falsified. HPSG has explored any number of new ideas: types,
type constaints, a highly articulated view of lexical organization,
inheritance hierarchies (not just for lexicons), linearization theory, a
careful distinction between descriptions and models, and other things that
will be pointed out by others, I'm sure...

> 5- What would be the state of affairs that we can make
> regarding the applying of HPSG to many languages?

HPSG descriptions now exist for a number of languages, with more being
undertaken all the time (at least I hope that's really true). There is work on
at least English, German, Japanese, Korean, French, Dutch, Polish, Bulgarian,
Spanish, Portuguese, Romanian, Welsh, Greek, Scandinavian (various),
Greenlandic, Hungarian, Hebrew, Austronesian (various), Mandarin, Lai,
Luiseno, Potawatomi, Salish, Dogon, Kwa, and Australian (various). Well, this
is just a partial list (please no one get mad at me for not mentioning their
language...) and there's much more work done on some languages than others.

I hope this is at least somewhat helpful...

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
-------------------------------------------------------------------------


>          Thanks in advance,
>
>                               Best regards,
>                               Achit AM.
>
>
>
> __________________________________________________
> Do You Yahoo!?
> Make international calls for as low as $.04/minute with Yahoo! Messenger
> http://phonecard.yahoo.com/