AW: Increasing interest in the HPSG conference

[Carl Pollard]

Hi John,

(Aside to everyone else: in the highly unlikely event you don't know
John, I assure you he does not need to establish his scientific bona
fides.)

I should respond to this since I raised these issues.

>
  We're getting mails with the heading on how to increase interest in
  HPSG (among linguists), and the mails themselves concern topics such
  as the intuitionist interpretation of the logics underlying the
  special formalism in which linguistic descriptions, including
  abstract (theoretical) descriptions, are about.

  I submit that there simply is no LARGE community of linguists
  interested in this.  There is a small, very capable community
  organized as MOL (Mathematics of Language), and maybe HPSG work
  primarily aimed at the mathematical foundations of grammar should
  team up with them.  I've attended two of their meetings, and the
  attendees are both good and receptive.  Formal grammar is also good.
  But also small in numbers.

  ... I think HPSG focuses so much on metaissues of formalism,
  interpretation and mathematical foundations that the average working
  linguist tunes out.
>>

I'd say this recent discussion is highly atypical of what people
working in HPSG think about, write about, and give conference papers
about. The only way to get the attention of the communities focusing
on linguistic description and on computational linguistics (which is
what most HPSG practicioners do, theory hacks like me are thin on the
ground) is to produce good analyses and good programs, and get the
word out in the journals and conferences that are devoted to those
activities.  Ivan has rightly been giving that advice for years; see
also Chris Brews's recent posting. In that connection, your advice
about level of presentation is well-taken (especially by me, I'm
afraid).

But a framework-specific conference (and a framework-specific mailing
list) are two venues where issues about theoretical foundations and
future directions CAN be addressed, inter alia; and they are venues
where one get assume more familiarity with framework-specific arcana.
The recent categorial grammar conference in Montpellier managed to
fill five days and sustain a very high interest level, with talks and
papers that dealt with the full spectrum of linguistic concerns --
descriptive, computational, psycholinguistic, and logical/foundational
(and much more of that last type than are ever even dreamed of at HPSG
meetings). If anything, the concern with technical issues is an order
of magnitude MORE intense than it is in the HPSG community, but I
sensed none of the pessimism, MP-envy, or concern about how to get
attention that led up to the discussion we've been having here.
Frankly, it was an eye-opening and inspiring experience, I think the
HPSG community would benefit a lot from more contact with the CG
community, and that CG could also learn alot from HPSG (especially in
terms of scaling up the complexity of the empirical issues they
focus on). Neither framework is ever going to have mass appeal with
the math-phobic, but there's no reason why they can't intereact more
with each other. The Formal Grammar conference could provide a setting
for that to happen. (MOL, on the other hand, is really an applied math
meeting, much more focused on issues without much appeal beyond the
theory hackers.)

John, there's part of your message I don't understand. In your point two,
you start by saying

  It would also help if the HPSG field focused less on standard GB
  fare, i.e. binding theory (where HPSG & GB are quite similar),
  control, long-distance dependency, etc. (I do not exempt myself from
  this criticism).  Naturally one needs to be able to handle these
  things, but the Chomskyan query "Where are the new facts?"  is also
  quite reasonable to all of us as scientists.

but end up saying

  If HPSG cannot provide inisightful, radically innovative
  perspectives on core GB issues, or, alternatively, insightful
  analyses on other interesting syntactic issues, something is wrong.

I can think of several dfiferent points you might have intended to
make here, but I wondered if you would mind saying a bit more. You
also listed a number of innovative HPSG-based analyses that you thought
hadn't had the impact they deserved to, but I wasn't sure whether you
meant they should have been presented in a more traditional, arboreal
way, or whether you meant that, since they were NOT easily paraphrased
in more mainstream terms, they were doomed to obsurity.

One last point about semantics, which you touched on at the end. You
mentioned that a lot of formal semanticists use GB syntax. That's
true, but they also use semantic formalisms that come out of the
Montague tradition. I think it is very important for people doing
semantics inside HPSG to keep that in mind. I know, I had a lot to do
with putting HPSG on a situation semantics track 20 years ago, firm in
my belief that Barwise and Cooper were on the verge of producing the
book that would be to SS as Dowty, Wall, and Peters was to PTQ. But
the fact is, that didn't happen; as for the problems that SS attacked
Montague semantics for, more headway was has been made within the
Montague-inspired tradition than without.

Carl