GG (June 2007)
This is the June 2007 release of GG, an HPSG for German developed
at the DFKI Language Technology Lab,
Saarbrücken, Germany, as part of the DELPH-IN consortium. The grammar was first published in June 2006.
Development was funded by the German Ministry for Education, Science,
Research and Technology ("Bundesministerium für Bildung und Forschung
(BMBF)"), as part of the projects DISCO (ITW 9002 0), Verbmobil (01
701 V0), Whiteboard (01 IW 002), Quetal (01 IW C02), and Collate (01
IN A01 B). The grammar is currently developed by Berthold Crysmann.
The grammar is released under the LGPLLR, a version of the LGPL adapted to
linguistic resources. See the above link or LICENSE for details.
Online Demo
You can try out the grammar online and browse the analyses at http://gg.dfki.de/demo/gg.
Download
The grammar is currently available as a
.tar.gz archive. Soon, it will be moved to a subversion
(version control) server hosted at DFKI.
The November 2007 version, which inter alia supports punctuation is available from here.
Installation
The grammar is intended to be used with systems that can run TDL-style
grammars, such as the LKB (Copestake 2001) or Pet (Callmeier 2000),
and can be used for parsing (LKB & Pet) and generation (LKB).
LKB
In the LKB, you can load the grammar via the script file lkb/script.
We encourage the use of the lexical database (lexdb) by Benjamin
Waldron, which is distributed as part of the LKB. See the DELPH-IN
Wiki for installation instructions. I recommend the following settings
for *lexdb-params*: ((:dbname "gg") (:mode "gg")), to disable sortal
restrictions.
Pet
The grammar files come with a Makefile to facilitate installation:
Run `make' to compile .grm files for the Pet runtime parser cheap(1).
Make sure the `flop(1)' preprocessor is in your PATH.
Documentation
Documentation for the grammar can be found in the
literature. Publications particularly relevant to this version of the
grammar are:
Müller & Kasper (2000)
Crysmann (2003)
Crysmann (2005)
The grammar uses revised German spelling conventions, at least as far as
the ss/ß rules are concerned. Support for the old spelling conventions
will be added in the future. The grammar and the lexicon are encoded
in UTF-8.
KNOWN ISSUES:
Owing to a bug in Allegro CLIM, direct input of 8bit characters from
the LKB GUI is not possible. As a workaround, you can use either TeX
or HTML syntax, or invoke the parser from the Lisp prompt
'(do-parse-tty "Am Anfang war das Wort.")'.
Alternatively, you can use the Trollet GUI by Pavel Mihalyov, which
fully supports UTF-8.
See TODO for grammar related known issues.
Berthold Crysmann, Fefor, Norway, 17th June 2006
References:
@InCollection{mueller_s-kasper_w00,
author = {Stefan M{\"u}ller and Walter Kasper},
title = {{HPSG} Analysis of {German} },
booktitle = {Verbmobil: Foundations of Speech-to-Speech Translation},
pages = {238--253},
publisher = {Springer},
year = 2000,
editor = {Wolfgang Wahlster},
address = {Berlin}
}
@Article{crysmann_b04rlc,
author = {Berthold Crysmann},
title = {Relative Clause Extraposition in {German}: An Efficient and Portable Implementation},
journal = {Research on Language and Computation},
volume = {3},
number = {1},
pages = {61--82},
year = {2005}
}
@InProceedings{crysmann_b03ranlp,
author = {Berthold Crysmann},
title = {On the Efficient Implementation of {German} Verb Placement in {HPSG}},
booktitle = {Proceedings of RANLP 2003},
address = {Borovets, Bulgaria},
pages = {112--116},
year = 2003
}
@InProceedings{crysmann_b05hpsg,
author = {Berthold Crysmann},
title = {Syncretism in {German}: a Unified Approach to Underspecification, Indeterminacy, and likeness of Case},
booktitle = {Proceedings of the 12th Int'l Conference on Head-driven Phrase Structure Grammar (HPSG), Aug 22--24, Lisbon},
year = {to appear},
address = {Stanford},
publisher = {CSLI publications}
}