Promoting interest in HPSG

[Emily M. Bender]

Dear Tibor (and others),

> As you have already seen and will see in this reply, I mostly hold
> positions which differ from yours. This shouldn't obscure that I
> appreciate that you started the debate.

And I appreciate your continuing it :)

> I am actually not a person to think in 'MP or us'.

I think I should state at this point that I strongly favor having
a diversity of frameworks available.  All frameworks suggest different
questions to ask, and we are the richer for working from different
points of view, provided that we also know how to talk to each
other.

> > But, I'm not sure what you mean by "lack of precise structural
> > predictions", or how this pertains to the current question.
>
> I find it somewhat irritating that any analysis of the verbal complex is
> fine. (Just an illustration, many further examples come to mind.)

Asking that the framework/theory/formalism constrain in advance
the range of possible analyses so that certain ones are clearly
"wrong" requires that we have solved several large problems before
we begin the linguistic analysis.  It seems to me a better plan
of attack (or at least a valid one) to develop (and test against
testsuites and corpora) large grammar fragments, and then ask
what generalizations can be drawn across them.  On the other hand,
I also recognize that there is a complementary approach (perhaps
the one that you take?) wherein linguists put forth strong claims
about possible grammars and then attempt to falsify or support
them by building grammars which are constrained by those claims.
What I would ask, though, is that the claims be clearly delineated
from the formalism so that they can be tested, improved, and revised
in an incremental fashion.

> If you claim (to the best of my knowledge, nobody does that, I just use it
> as an illustration) that we have a total of 4.000 different Principles A
> each of which does the work for one of the 4.000 still existing languages,
> you do not make any prediction.

The scenario that I imagine is developing a Principle A for one
language, and then trying it on the next one, and discovering that it
needs modification etc.  If, when one was done working through all the
languages in ones sample (preferably a balanced sample representative
of the world's language families), one found that there were in fact
about 32 different Principles A which could be 'reduced' to a single
Principle A with 5 binary parameters (say), that would be a very
interesting result.  The next question would then be to ask what it is
about a) human cognition b) functional pressures on language change or
c) the problem of communicating complex ideas in a sequence of sounds
that might explain this relatively constrained range of possibilities?

A colleague who asked to remain anonymous sent the following, which
I think also relates well to this discussion:

>I think that for (3) to happen [more interest in HPSG --EB], there
>would need to be more of a focus in the HPSG community on typological
>issues.  My own view (as someone perhaps straddling the border, being
>familiar with HPSG, but not really doing any work in it) is that work
>in this tradition tends to focus on single languages (overwhelmingly
>Indo-European, at that---a related, but separate issue) without much
>consequence for larger typological issues.  I feel like the focus of
>linguistics more broadly has shifted to typology (some theoretical
>work in (OT-)LFG has made this shift, as has work in chomskyan syntax
>since the 1980s).  I don't feel that the HPSG community has really
>made the shift.  (E.g., I don't get the sense that if I read some
>random paper in HPSG in language X that data from language Y will
>have much bearing on the analysis, potentially counterexemplifying
>claims).

This is an interesting different perspective on the issue of what
is a "restrictive" theory.  I think that it would be valuable to
have a theory of how languages relate to each other which would enable
research on one to inform research on others.  In my opinion, the
source of that theory should be external to the formalism.

Emily