A novel argument for cyclicity from the (in)visibility of infixes at morpheme junctures: Bottoms up!

This paper investigates what happens to grammatical relationships between two morphemes when an infix incidentally lands between them. Most relationships—including allomorphic and morphophonological relationships—persist, showing that these relationships are established before the intrusion of the infix. However, infixes do disrupt word-level phonology, which thus must be established after the intrusion of the infix. These findings furnish a novel argument for grammatical models that are (i) cyclic and bottom-up (or inside out) with respect to both exponence and certain (morpho)phonological operations, and (ii) noncyclic with respect to certain purely phonological processes, which are computed only at specific junctures (e.g., the word). A further argument from the interaction of infixation with object incorporation supports a post-syntactic model of morphology, such as Distributed Morphology. This work brings together earlier case studies of Hunzib (Kalin 2022b ), Nancowry (Kalin 2023 ), and Palauan (Embick 2010 ) with several more languages, including a particularly extensive new study of the Bolivian language isolate Movima (drawing on Haude 2006 , 2019 ), facilitating broader and stronger conclusions.