Revisiting sign language classifiers: (How) do classifiers classify?

Abstract

Sign language classifiers combine conventional, non-conventional, arbitrary, and iconic elements. While other similarly complex constructions, such as ideophones and gestures, have been analyzed for their at-issueness, classifiers have not. In this paper, we introduce a new type of classifier in HMSL, which we call size-and-class, along with findings on the projection properties of SAS classifiers in LSF. Strikingly, we find that classifiers exhibit the opposite behavior to these other hybrid objects in the way they contribute meaning. To account for the observed difference, we identify three possible hypotheses: i) an iconicity-based hypothesis, ii) a structure-based hypothesis, and iii) a prominence-based hypothesis. We find evidence in favor of the latter two, but call for future work targeting the at-issueness of different types of classifier to further determine whether structure or prominence, or perhaps both, determine at-issueness.