RedTyp: A Database of Reduplication with Computational Models
Hossep Dolatian, Jeffrey Heinz · ScholarWorks@UMassAmherst (University of Massachusetts Amherst) · 2019
Reduplication is a theoretically and typologically well-studied phenomenon, but there is no database of reduplication patterns which include explicit computational models. This paper introduces RedTyp, an SQL database which provides a computational resource that can be used by both theoretical and computational linguists who work on reduplication. It catalogs 138 reduplicative morphemes across 91 languages, which are modeled with 57 distinct finite-state machines. The finite-state machines are 2-way transducers, which provide an explicit, compact, and convenient representation for reduplication patterns, and which arguably capture the linguistic generalizations more directly than the more commonly used 1-way transducers for modeling natural language morphophonology.