This week, we’ll have an invited talk by Gloria Melessmoen (UBC) on Friday, February 21 at 3PM at Sherbrooke 680 in room 1041. The details of the talk are given below.
Title: Non-concatenative morphemes require phonologically defined strata
Abstract: What is the division of labour between different components of the grammar? Specifically, what does the phonological grammar “know” about morphosyntax and how is this knowledge used in word formation? These questions can be explored through the study of non-concatenative morphology because it provides insight into the interface between morphology and phonology, as well as the larger architecture of the grammar.
Non-concatenative morphology in the Salish language family motivates a model of the spell-out component of the grammar that integrates components of Prosodic Morphology (McCarthy & Prince 1986), Moraic Theory (Hyman 1985; Hayes 1989), Generalised Nonlinear Affixation (Bermúdez-Otero 2012; Bye & Svenonius 2012; Zimmermann 2017), and Stratal Optimality Theory (Kiparsky 2015; Bermúdez-Otero 2017). The empirical basis of this model is a comprehensive survey of reduplication across Salish. The survey is supplemented by fieldwork on two languages (ʔayʔaǰuθəm and St’át’imcets), which allows specific predictions of the model to be tested. In this talk, I focus specifically on the question of domain(s). I provide evidence that a stratum-based model is necessary to account for systematic gaps in multiple reduplication patterns: only one prosodic affix can be realised per stratum. This analysis limits multiple reduplication by number of morphemes, rather than number of segments (cf. Zimmermann 2021; Mellesmoen & Urbanczyk 2021). While the phonology must be able to differentiate between phonological content belonging to different morphemes (i.e., different strings), it does not have access to broader morphosyntactic information or structure. I provide evidence that strata in the model developed here must be defined purely on a phonological basis, without reference to morphological domains (i.e., there is no “stem”). This marks a divergence from traditional Stratal Optimality Theory, which is informed by Lexical Phonology (Kiparsky 1982; Mohanan 1982).