Spatial and Planning Models of ASL Classifier Predicates for Machine Translation
Matt Huenerfauth · 2004
A difficult aspect of the translation of English text into American Sign Language (ASL) animation has been the production of ASL phenomena called “classifier predicates. ” The complex way these expressions use the 3D space around the signer challenges traditional MT approaches. This paper presents new models for classifier predicates based on a 3D spatial representation and an animation planning formalism that facilitate a translation approach and are compatible with current linguistic accounts of these phenomena. This design can be incorporated into a multi-path architecture to build English-to-ASL MT systems capable of producing classifier predicates.