Morphological perspectives on the Georgian verb

is paper explores in detail Georgian verbal agreement/tense morphology and aims at uncovering regularities and natural classes in this paradigm, using in turn the tools of Distributed Morphology and Channel eory, or articial learning. AŸer reviewing part of the morphological literature on the topic of a›x blocking in transitive verbs in the present, I turn to a decomposition of all non-perfect person/tense combinations. I ošer a preliminary analysis in the Distributed Morphology framework, allowing to establish meaningform regularities in the paradigm. Admitting di›culties linked to the latter framework (many operations are needed and seem to be ad hoc), I turn to a Channel (or Accessibility-based) analysis that does away with these operations. It is not evident that the latter account is radically superior, though, since many manipulations not mentioned in Keine’s (2012) seminal paper seem to be necessary in this special case, calling for an extension/discussion of this model. e last part deals with a dišerent approach, that of inžectional learning, where a›xes are selected by an articial learner in a serialist optimality setting. e latter approach reveals interesting ndings, notably with regard to a formmeaning pair not previously considered as forming a natural class, namely the -s su›x of Georgian found not only in the third singular, but also the third plural.