default inheritance

[kay at cogsci.berkeley.edu]

I probably am not really entitled to an opinion in this discussion, but
here goes anyway. As I understand things, unless defaults are really
abbreviations in an underlyingly monotonic system, as I take them to be in
current HPSG, they involve overrides. The prototype has, say, feature [A
+] and some related non-prototype case inherits everything from the
prototype except that it has feature [A -]. Thus, [A +] -> [A -].  The
problem with this sort of procedure is that unless you have a theory
tightly restricting which features can be overridden under which
conditions it takes only one override to make a silk purse out of a sow's
ear. [KIND SILK.PURSE] -> [KIND SOW'S.EAR].

Paul

On Tue, 23 Oct 2001, Dick Hudson wrote:

> Dear HPSG list,
> I was pleased to see Luis Casillas's message about default
> unification/inheritance, and I very much hope it will provoke some
> discussion on this list. It seems to me one of the most important issues in
> linguistic theory.  Construction grammarians are divided for/against
> defaults; Word Grammar is definitely and very strongly for, and so is
> Cognitive Grammar; LFG doesn't seem to recognise it, nor in general do
> theories that use Optimality; HPSG recognises it only in the type hierarchy.
>
> Like Luis, I would like to know whether the only objections are
> computational - the need to keep checking for overriding facts and the time
> this takes in a serial computer system.
>
> Dick Hudson
>
> Richard (= Dick) Hudson
>
> Phonetics and Linguistics, University College London,
> Gower Street, London WC1E  6BT.
> +44(0)20 7679 3152; fax +44(0)20 7383 4108;
> http://www.phon.ucl.ac.uk/home/dick/home.htm
>


__________________________________________________________
 Paul Kay                      Department of Linguistics
 kay at cogsci.berkeley.edu       University of California
 www.icsi.berkeley.edu/~kay    Berkeley, CA 94720, USA