This is a development release of a TAG parser that can be used to parse
using the XTAG English grammar. It is still a preliminary release --
trying to use it immediately as part of a runtime system is probably
unrealistic. This short description will tell you what capabilities the
parser currently includes and what functionality it currently lacks.

To use this release you will need to understand various formal aspects
of Tree Adjoining Grammars, especially derivation trees. Familiarity
with the XTAG English grammar and its components will be helpful also.

You can get more information about the XTAG English grammar including a
technical report explaining the formalism and other aspects of the
grammar at the following URL:

http://www.cis.upenn.edu/~xtag/

What this package contains:
---------------------------

* It is a general TAG parser written entirely in ANSI C. It has been
  tested on SunOS 5.7 and on Linux 2.x. See INSTALL.notes for other
  requirements.

* It uses a chart based head-corner parsing algorithm. The parser
  can also be used as a probabilistic parser when parameter values are
  available.

* It has a small memory footprint and should run reasonably on systems
  with 64MB of memory for sentences with less than 25 words (depending
  on grammar and lexicon size).

* This release comes with the complete stable version of the XTAG
  English grammar.

* It comes with a Tcl/Tk derived/derivation tree viewer as well as a
  viewer for the grammar trees/families. 

What file to read next:
-----------------------

Read the README file for location and other information about this
package.

Read the HOWTO file for instructions on running the parser and using
the elementary tree viewer.

Caveats:
--------

Many aspects of the codebase will keep changing in future versions. Any
additions or suggestions are welcome. Until a better system is in
place, any changes you want to contribute should be sent to me as patch
files by email (mailto: anoop at linc.cis.upenn.edu). The code is
licensed under the GPL (the licence is included with this package in a
file called COPYING in the current directory).

Installation should not be too difficult for people who have experience
in installing other software packages. However, I cannot promise any
support for problems you might face in installing the parser. If you
are having problems installing, then first try the precompiled binary
versions of the software available from the same ftp site as the source
package. (ftp://ftp.cis.upenn.edu/pub/xtag/lem/)

Anoop Sarkar
anoop at linc.cis.upenn.edu