Evaluating context unification for semantic underspecification

Author: Koller, Alexander

Editor:

Context unification has been proposed as a powerful formalism for the uniform and underspecified treatment of scope ambiguities and parallelism. We develop an algorithm for context unification that alleviates the enormous computational and overgeneration problems of earlier algorithms. It is incomplete for full context unification, but finds all linguistically relevant solutions. We apply an implementation of the new algorithm to evaluate context unification as a tool for semantic analysis.