Towards a Restricted Realization Theory
In this paper we advance arguments in favor of a vocabulary-insertion-only program for non-lexicalist realizational models of morphology, Distributed Morphology (Halle & Marantz 1993) in particular. We claim that this end can be achieved through relatively simple and well-motivated proposals. We suggest that the vast majority of non-syntactic mechanisms can be obviated by the adoption of non-terminal insertion (Radkevich 2010; Svenonius 2012; Merchant 2013; Starke 2009; Caha 2009; see Embick & Marantz 2008 for a contrary argument)– specifically post-linearization spanning. We restrict our discussion here to whether weak suppletion must be the output of phonological processes (see e.g. Marantz 1997; Embick & Halle 2005; Harley & Tubino Blanco 2013) or simply listed (see, e.g., Bermúdez-Otero 2013), and whether the containment prediction (Embick & Marantz 2008; Embick 2012) is indeed fatal for non-terminal insertion. We also propose here that Siddiqi’s (2009) feature blocking system can be co-opted for the tasks typical of impoverishment.