The paper describes two parsing schemes: a shallow approach based on machine learning and a cascaded finite-state parser with a hand-crafted grammar.It discusses several ways to combine them and presents evaluation results for the two individual approaches and their combination.An underspecification scheme for the output of the finite-state parser is introduced and shown to improve performance.

The paper presents an approach to ellipsis resolution in a framework of scope underspecification (Underspecified Discourse Representation Theory). It is argued that the approach improves on previous proposals to integrate ellipsis resolution and scope underspecification (Crouch, 1995; Egg et al., 2001) in that application processes like anaphora resolution do not require full disambiguation but can work directly on the underspecified representation. Furthermore it is shown that the approach presented can cope with the examples discussed by Dalrymple et al. (1991) as well as a problem noted recently by Erk and Koller (2001).