Cophonologies by phase

Introduction This paper proposes a model of the syntax-phonology interface which is compatible with a central assumption of mainstream syntax, that phonology and morphology are interpreted from syntactic structures, and mainstream phonology, which models phonological processes by ranked or weighted constraints. The central idea is to combine Cophonology Theory (Ito & Mester 1995, Anttila 2002, Inkelas & Zoll 2005) with PhaseTheory (Chomsky 2001, Abels 2012, Boskovic 2014), which allows cophonologies to scope over the spelled-out chunks of syntax. Cophonologies by phase Inkelas (2014) demonstrates that morphologically conditioned phonology and process morphology involve the same operations at several levels. As process morphology is easily modeled with constraint rankings rather than morphemes, Inkelas’s generalization provides a direct argument that both process morphology and morphologically conditioned phonology are due to cophonologies, the association of constraint rankings with particular morphemes. We propose a model of cophonologies where constraint rankings are associated with Vocabulary Items in Distributed Morphology (Halle and Marantz 1994). In particular, we propose that a VI maximally consists of the association of morphosyntactic features with three phonological components: featural content (F ), consisting of tonal or segmental features, prosodic content (P), which situates the VI in the prosodic hierarchy, and a subranking of constraints (R) which serve to provide a partial override to a default master constraint ranking (or which combine in a weighted constraint model). F , P or R can be null for a particular VI. Consider a hypothetical verbalizing suffix -ga, which, with its host, corresponds to a prosodic word and which is associated with the constraint-ranking B≫A: