This paper recasts Chomsky's (2007, 2008) Feature Inheritance, arguing that the syntactic relation between phase heads and non-phase heads must be regarded as that of identity: Non-phase heads are (copies of) phase heads. From this perspective, e.g., C and T are one and the very same unit in the lexicon, a unit that can (and sometimes must) undergo a process of Internal Merge (IM) during the derivation, creating a discontinuous object (i.e., a non-trivial chain) that gives rise to the – apparent – non-phase head / phase head distinction. The analysis has several consequences. Firstly, it provides an explanation for the fact that the features of phases heads are inherited by non-phase heads (under a copy-based approach, this is simply automatic). Secondly, it accounts for the functioning of phase and non-phase heads `as a unit' in the Case-agreement systems (as argued by Chomsky 2004, Epstein et al. 2012). Thirdly, since non-phase heads are copies of phase heads, it also follows that they need not be seen as feature-less elements in the lexicon (as mere feature receptacles, cf. Richards 2007). Fourthly, this approach allows us to dispense with the idea that either C and T trigger simultaneous operations (Chomsky 2008), or that the operations triggered by T are actually driven by C (derivatively); this is welcome, since both alternatives involve a restricted (phase-based) margin of countercyclicity. Finally, the present analysis makes it possible to dispense with Feature Inheritance, understood as a brandnew, UG enriching, feature-depriving mechanism. Rejecting this commonly accepted view of Feature Inheritance takes into account the problems attributed to feature-movement in the syntax (Chomsky's 1995 Move F).
©2014 by Walter de Gruyter Berlin/Boston