A last Resort Approach to (Number) Feature Inheritance1

Abstract

AbstractIn this paper, we assume Chomsky's (2005, 2007, 2008) feature inheritance framework as an attempt to explain the correlation between the presence of C and agreement/finiteness on T in English. We argue, however, against one of its critical assumptions, namely, that T is a head deprived of φ‐features (i.e., a mere feature receptacle as put by Richards 2007). The proposal of this paper is three‐fold. First, based on the possibility of ellipsis in different infinitival contexts, we argue that there are two types of T heads in English: the one used in ECM contexts, containing no φ‐features, and the other occurring in finite/control/raising clauses and for‐infinitives, which contains (at least) a person feature. Second, even though finite clauses, for‐infinitives, and control clauses are all headed by C, only the T head in finite clauses displays full agreement. We show that maintaining the feature inheritance hypothesis in the face of this challenge leads us to further distinguish two types of C heads in English: only the C head in for‐infinitives is φ‐complete, whereas that in finite/control clauses hosts only an inherent number feature. Finally, after reviewing the problems with existing motivations for Chomsky's feature inheritance hypothesis, we advance a novel approach to feature inheritance in which only the number feature is inherited as a last resort strategy to form a φ‐complete head for Case valuation in a phase.