Implications of Katu Nominalization for Infixation Typology

In the Mon-Khmer language Katu (as spoken in the Lao P.D.R.; Costello 1998), nominalization is marked with a variety of forms at the leftmost periphery of the root word, and occasionally as a phonologically reduced infix, /-r-/, appearing in the coda of the initial syllable of the nominalized word. This paper will analyze the Katu case as an example of prosodically conditioned allomorphy, resultant in OT from a surface merger of underlying root and affix segmentism. It will be shown that segmental default-to-opposite, a phenomenon predicted but unattested in Katu, must be ruled out in OT grammars with the Edge Proximity Condition, a strengthening of the Morphology/Phonology Strict Concordance Condition (McCarthy and Prince, 1995). The resulting theory makes a strong prediction: that infixation may only occur where an affix is dominated by an edge-bound prosodic constituent. Challenges posed by cases of prosodic subcategorization, as exemplified in English expletive infixation, will be considered, along with an alternative prohibition on morpheme dislocation, the Subcategorization Non-violability hypothesis of Yu (2003).