Dick Hudson wrote:
> 
> >Incidentally, as far as I can see, the theory that can most trivially be
> >translated into WG is HPSG. Does anyone else see the same similarities?
> ## Another interesting point. Off the top of my head I can think of the
> following points of similarity and difference:
> SIMILARITIES
> s1. Both recognise heads and are 'head-driven'.
> s2. Both use default inheritance (though I can't remember whether HPSG now
> recognises it throughout the grammar or just in the lexicon).
> s3. Both are monostratal (assuming we can give this notion some content!!).
> s4. Both allow 're-entrance' - i.e. two attributes sharing the same value.
> s5. Both make a strict ID/LP separation - i.e. word-order rules are
> separate from subcategorisation etc.
> s6. (Perhaps marginal in this context.) Both recognise the need to involve
> social categories in the grammar.
> 
> DIFFERENCES
> d1. HPSG distinguishes rules from lexical items (doesn't it?).

I think that this is a hard call. P&S sneak a lot of rules into lexical
items (or at least that's my feeling). However, some people like to
think
of HPSG as a modular theory. I guess they feel that there is no
continuum
between words and rules.


> d2. HPSG has phrase structure.
> d3. HPSG allows ordered lists as attribute-values. (I once toyed with these
> in a WG 'valency structure', so I wouldn't expect to win this battle.)
> d4. Related to d1: HPSG doesn't seem to accommodate non-canonical
> constructions easily, because all structures have to be sanctioned by a
> general rule.

In general, I think d4 is true. But as Nik points out, Sag is moving
towards the notion of grammatical construction (see Sag's JL paper).
This apparently dissappoints people like Borsley.

Joe
__________________________________________________________
Home page: http://lingua.fil.ub.es/~hilferty/homepage.html



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%