Unaccusativity in HPSG

[Howard Anthony Gregory]

I would be very happy if the information needed for unaccusativity is in the
end semantically predictable, but was under the impression that this is still
a somewhat open-ended research question; and that the Relational Grammar
arguments for the irreduceability of underlying grammatical relations (e.g.
Rosen 1984), while sidelined in much current work, have not been finally laid
to rest. If so, the approach of Pollard (1994) and others might need to be
explored further and even extended. Those that are open to this possibility
might be interested in the Stratified Feature Grammars of Johnson and Moss
(several publications including the one below - I can dig out other references
if needed). This is a typed feature structure grammar but involves attributes
(at least those for valence features / grammatical relations) being
parametrized by numbers representing successive strata. In this approach,
arguments of unaccusatives appear as unabashed underlying objects, and so on.

Regards,
Howard Gregory

David E. Johnson, Lawrence S. Moss: Some Formal Properties of Stratified
Feature Grammars. Ann. Math. Artif. Intell. 8(1-2): 133-173 (1993)

___________________________

H. A. O. Gregory
Seminar für Englische Philologie
Georg-August-Universität Göttingen
Käte-Hamburger-Weg 3
37073 Göttingen
Deutschland

E-mail: howard.gregory at phil.uni-goettingen.de
Web:    http://www.gwdg.de/~hgregor1/