HPSG Diagrams

[Luis Casillas]

On Mon, Dec 02, 2002 at 04:37:48PM -0800, Paul Kay wrote:

> Luis's observation that this barrier operates at the graduate student
> level, and thus might influence career choices, is disturbing if one
> wants to see the HPSG community grow.

A point of clarification: the course I was talking about is an undergrad
intro to syntax course.  At Stanford most grad students certainly have
to figure out some way of doing AVMs, but it's a particularly good
environment to do so, since you are surrounded by the day-to-day life of
the HPSG and LFG communities; you go next door and you ask somebody how
to do it.

My concern is in general for people who don't have such a resource, of
which the undergrads I mentioned were only one case.  Grad students are a
great example too, but I think in general people who are interested in
HPSG are not necessarily interested in learning an arcane typesetting
system from the the early 80s with an interminable array of third party
add ons.

This all boils down to something I've been telling Ivan for a while:
somebody should write a grant proposal and get money for a project on
computer-assisted education for syntactic theory, based on HPSG.  Then
you pay a few bright CS undergrads and pay them to help you design and
write an application that does things like the following:

1. Allow the user to define types, constraints and hierarchies with
   graphical editors.
2. Allow the user to enter tree structures in a tree editor, with nodes
   labeled by AVMs (the common notational convention in HPSG).
3. Check whether a tree satisfies the constraints in a grammar.
4. Output all these graphical representations in formats suitable for
   importing into other software.
5. Support downloading problem sets from a central server, and
   submitting answers online for automatic correction.
6. Work on something other than Unix, and not require immense amounts
   of RAM or processor power.

This kills many birds with just one rock: it would be useful both for
students doing a course based on something like Sag and Wasow's
textbook, but also for people in general as a tool one could use just
to draw AVMs and trees for one's papers, output them, check that
they license the structures you think they do, etc.  Also, I think it
would raise the bar in general for the teaching of linguistics.

--
Luis Casillas
Department of Linguistics
Stanford University
http://www.stanford.edu/~casillas/