state of HPSG and HPSG conference

[Detmar Meurers]

 Hi,

it's a beautiful summer's day, so let me bring up some positive
points ;-) It seems to me that many of the good points that were
brought up as goals are already in place: For example, I think there
are a lot of reasons for agreeing with John's point that papers
should be written in a way that is generally accessible; one reason
that was not yet mentioned is that writing which emphasizes
theoretical issues ages very quickly - do we want to spend all of
the little time we have, on writing papers that no-one in ten years
can understand any more? Discussions of pre-theoretical
generalizations often remain relevant and accessible (even if they
admittedly don't produce many short-term stars).

But just like Anne mentioned, I think this is already the way
many/most of us in HPSG write linguistic papers: First an empirical
discussion without theoretical notions, a discussion of what
linguistic properties are relevant and the pre-theoretical
generalizations, then a final section with the formalized HPSG
analysis. There are many illustrations of this general structure -
check out the work Anne mentions or, say, Stefan Müller's work, or
my Raising Spirits paper or the joint work with Kordula De Kuthy,
etc. etc. if you wonder what I'm talking about.

A related positive point is that, as far as I can tell, this has not
gone unnoticed. I recently collected references to my work for a
performance review and was pleasantly surprised to find citations by
people working in Minimalism/GB, LFG, Optimality Theory, Dependency
Grammar, Performance Grammar, and others. Similarly, Stefan Müller's
work or that of Tibor Kiss seems to be often-cited outside of HPSG
circles - and I would think that the world is not radically
different outside of Germanic linguistics.

In conclusion, it seems to me that -- different from the atmosphere
of some of the discussion -- we are on the right track in the sense
that we need to emphasize the good things we are already doing
rather than look for a new path.  In the past 10 years, we have
gained many insights into linguistic phenomena and, generally
speaking, about when what kind of grammatical relations are
established and whether the domains they operate in are local, or
lexically extended head domains, or non-local domains. Similarly, we
understand the role and relevance of (partial) constituency, head
clusters, phrasal elements behaving like words, the
morphology-syntax interface, etc. etc. much better than before.
Apart from continuing this work, to me an interesting challenge
seems to be to integrate the syntactic work with the other modules
of linguistic competence, such as context (information structure),
which will allow us to replace some syntactic stipulations with more
explanatory accounts (I realize this is not the use of "explanatory"
in some mainstream syntax). Naturally, each of us will have their
individual research agenda like that, but it seems that many are
natural extensions of the work we've done in the HPSG architecture -
and it does not preclude adding more of an emphasis on so-far
neglected areas, such as language acquisition.

Regarding the HPSG conference, I think Florian in his posting on June
28 mentioned an important aspect: While general linguistics conference
talks allow us to present the empirical and pre-theoretical level, we
also need a forum in which one can present HPSG theories and formal
issues without having to introduce the architecture first. This is the
important role of the HPSG conference. So, as far as I see, it's not
surprising that this conference does not attract a huge audience of
general linguists - it's by design and serves an important purpose for
the HPSG community.

Regarding the general linguistics community, one thing that has often
frustrated me (even on a sunny day ;-) is that general linguistics
conferences generally select presentations based on one/two page
abstracts, and no reviewing feedback is provided. My experience is
that if an abstract does not mention mainstream buzzwords, it does not
get in. After a good handful of such experiences I've therefore
stopped sending papers to such "linguistics" conferences and
workshops. It's exactly parallel to what Chomsky pointed out in
Manufacturing Consent when he mentions that the short TV snippets
(interviews etc.) cannot be used to establish alternatives to the
mainstream opinion. So one way, I think, the situation in general
linguistics could be improved is by establishing a conference with
proper reviewing of full papers and provide reviewer feedback (which
should be possible given that it's the general practice in many other
fields).

Finally, I wanted to support another idea that was mentioned in
another form: I think that for many of us it would be a mistake to
(re)focus on the mainstream agenda and issues in order to gain more
influence there - and thereby abandon the sustainable research
agenda we've been establishing.  If the "big issues" in mainstream
syntax yielded empirically relevant fundamental insights into
language (as opposed to being a fascinating glass bead game), why is
it that such work plays virtually no role in computational
linguistics? Note that this is the case even though, after the
heydays of number-crunching, the field is integrating more and more
insights about language of the kind "which relations are established
in which domains" and "what information is encoded where" -
questions which directly correspond to work in the HPSG paradigm.

Lieben Gruß,
Detmar

--
Detmar Meurers, Assistant Professor, Dept. of Linguistics, OSU
201a Oxley Hall, 1712 Neil Avenue, Columbus OH 43210-1298, USA
http://ling.osu.edu/~dm/                 GnuPG key on web page