Doubly conditioned phonology in Cophonologies by Phase

Phonological alternations can be unconditioned, applying uniformly across a language, no matter the context.

They can also be specific to particular morphological environments, like English velar softening (k–>s) before some /ɪ/-initial suffixes (-ism, -ity) but not others (-ish, -ing). Numerous frameworks have been proposed to model morphologically conditioned phonology: Exception features (Chomsky and Halle 1968), Lexical Morphology and Phonology (Kiparsky 1982), Stratal OT (Kiparsky 2000, 2008), Indexed Constraint Theory (Ito and Mester 1995, 1999; Pater 2010), Cophonology Theory (Orgun 1996; Inkelas 1998; Inkelas and Zoll 2005, 2007).

In this talk I present data from two understudied languages, Sacapultec (Mayan) and Guébie (Kru), showing that phonological alternations can not only by triggered by the presence of a single morpheme, but they can also be doubly morphologically triggered. That is, there are phonological alternations that only occur when two morphological triggers are present simultaneously. I show how this double conditioning can be modeled in Cophonologies by Phase, a framework which combines phonological evaluation at syntactic phase boundaries with morpheme-specific phonological adjustments using weighted constraints. Alternative models struggle to account for the double conditioning and/or the locality of these effects.