Unaccusativity in HPSG

[Stefan Müller]

Hi,

> I always thought SUBJ, COMPS, and SPR were grammatical functions.

Well, yes.

SUBJ as such is not a problem, since one can say that it is the first
element with structural case on a list that represents the valence
(ARG-ST for instance), so what is represented under SUBJ follows from
something.

SUBJ is also good if we want to have a uniform way to refer to phrases.
A VP is something fully saturated (i.e. with nothing on COMPS). There is
no such motivation for separate representation of VP internal arguments,
therefore they are just the rest, i.e. COMPS (or SUBCAT in Pollard 1990
for non-finite verbs).

Subjects also differ from COMPS in terms of visibility. See Detmar's and
Ivan's forthcoming work on locality.

If one has a valence set and pointers to an `external argument' and an
`internal argument', then the value of the EA and IA feature does not
follow. EA and IA become primitives of the theory. This is what I
considered problematic.

ERG and DA are used to mark unaccusativity. So they are different from
SUBJ or Direct Object. What I wanted to say is that it is better to have
a single feature than having two of them.

Best wishes

	Stefan

@InCollection{Pollard90a,
   author      =	"Carl J. Pollard",
   title	      =	"On Head Non-Movement",
   booktitle   =	"Discontinuous Constituency",
   editor      =	"Harry Bunt and Arthur van Horck",
   address     =	"Berlin, New York",
   publisher   =	"Mouton de Gruyter",
   pages	      =	{279--305},
   note	      =	{Published version of a Ms. dated January 1990},
   year	      =	1996
}

--
Stefan Müller

Universität Bremen/Fachbereich 10      Tel: (+49) (+421) 218-8601
Postfach 33 04 40
D-28334 Bremen

http://www.stefan-müller.net

http://www.cl.uni-bremen.de/~stefan/Babel/Interaktiv/