12h45
morphemes in a language; however, having two reduplicative morphemes does not entail the grammaticality of MR. Languages vary with respect to whether MR is permitted, as well as whether there are additional restrictions on reduplicant size that do not exist outside of MR (see Zimmermann 2021’s typology). Some languages have “variable MR” grammars: while MR is attested in the language, certain combinations are avoided without semantic motivation (Mellesmoen & Urbanczyk 2020). Variable MR languages require a phonological account that can predict the attested combinations while simultaneously ruling out the unattested combinations. Analyses of MR that treat reduplication as involving prosodic affixation and fission have accounted for avoidance of MR by using Integrity to limit the overall number of copied segments (Zimmermann 2021), or by proposing a restriction on Gen that only allows for 1:2 mappings between input and output (Mellesmoen & Urbanczyk 2021). In this talk, I show that neither approach can account for variable MR grammars found in Salish languages. I propose that fission is limited by morpheme: segments associated with one morpheme in the input may maximally associate with prosodic units from one other morpheme. This approach predicts that MR is possible if (i) each type of reduplication happens at a separate stratum (Stratal OT – see Kiparsky 2015 and references therein), or (ii) if at least one type of reduplication is not triggered by a prosodic affix. I show that these predictions hold through case studies of St’át’imcets (Lillooet) and ʔayʔaǰuθəm (Comox-Sliammon).”
