Assessing Complexity Results in Feature Theories

Abstract

In this paper, we assess the complexity results of formalisms that describe the feature theories used in computational linguistics. We show that from these complexity results no immediate conclusions can be drawn about the complexity of the recognition problem of unification grammars using these feature theories. On the one hand, the complexity of feature theories does not provide an upper bound for the complexity of such unification grammars. On the other hand, the complexity of feature theories need not provide a lower bound. Therefore, we argue for formalisms that describe actual unification grammars instead of feature theories. Thus the complexity results of these formalisms judge upon the hardness of unification grammars in computational linguistics.