COLING 1992, The Fifteenth International Conference on Cognitive Linguistics An Alternative to Deep Case for Representing Relational Information

No one has come up with a completely satisfactory set of deep cases relations (or thematic relations). The underlying reason is that any finite set of case relations can capture only some of the generalizations desired. I propose instead a feature-space representation of relational information, where the axes are such things as degree of responsibility, degree of activity, and degree of affectedness. The role of a participant in an event can then be described as a point in this space, allowing more accurate representation of relational information. The domain of validity of each relevant linguistic generalizations corresponds to a prototype-centered region in the space. This proposal is easy to implement.