Beyond constituency tests: A reply to Osborne

Abstract

Timothy Osborne argues that phrase structure grammars (PSGs) postulate unnecessarily complex structures, and that Dependency Grammar (DG) is to be preferred on grounds of simplicity (1:1 word-to-node ratio) and empirical adequacy (capturing the results of constituency tests). In this reply, I argue that, while some of Osborne’s criticisms of PSGs are justified, there are both empirical and theoretical problems with his major claims. In particular, his version of DG is too restrictive with respect to certain constituency facts (modified nouns, verbal phrases), and what it gains in simplicity qua number of nodes, it loses in requiring a more complex interface between syntax and other linguistic components (phonology, semantics). I argue that Mirror Theory, a framework that is in a sense intermediate between DG and PSGs, answers Osborne’s justified criticisms while not suffering from the problems of his version of DG.