Reduplication in Distributed Morphology

Abstract

The two extant approaches to reduplication in Distributed Morphology (DM) are: (i) the readjustment approach, where reduplication is claimed to result from a readjustment operation on some stem triggered by a (typically null) affix; and (ii) the affixation approach, where reduplication is claimed to result from the insertion of a special type of Vocabulary Item (i.e. a reduplicative affix–“reduplicant” or “Red”) which gets inserted into a syntactic node in order to discharge some morphosyntactic feature(s), but which receives its own phonological content from some other stem (i.e. its “base”) in the output. This paper argues from phonologically-conditioned allomorphy pertaining to base-dependence, as in the case of durative reduplication in Tawala, that the latter approach best accounts for a necessary distinction between “reduplicants” and “bases” as different types of morphemes which display different phonological effects, including “the emergence of the unmarked” effects, in many languages. I also defend a blended model of DM which incorporates a constraint-based Correspondence Theoretic vision of Phonological Form. In this model the syntax builds morphological structure as per standard DM assumptions, which in turn leads to local and cyclic restrictions on allomorph selection, as argued in Embick (2010). I argue contra Embick (2010), however, that the phonology must be an essential part of the grammar in order to account for surface form-oriented (or “output-centered”) prosodic morphology such as reduplication and mora affixation. In this model, the output of Morphological Structure serves as an input into PF, which I construe as Optimality Theoretic tableaux as in Correspondence Theory, thus accounting for surface-oriented phonological copying effects like base-dependence.