Number Markedness: Evidence from Gangalidda
By Jessica Mathie
I present data from Gangalidda (Australia) which shows that plural is more marked than dual in this language. This challenges the claim made in Harley and Ritter (2002) that dual is universally more marked than plural. Evidence that Harley and Ritter’s claim does not hold in Gangalidda is found in the distribution of dual clitics. In clauses with two non-singular arguments, the dual clitic is able to cross-reference both dual and plural entities. Assuming a privative feature geometry, this distribution is only possible if the features of the dual clitic are a subset of those present for the plural clitic (see Mathie In prep for an analysis assuming binary features). If the dual clitic had more features, as it does in Harley and Ritter’s geometry, it could not be inserted into a plural syntactic context, since it would be overspecified. I demonstrate that the Gangalidda distribution can be accounted for by the feature geometry proposed by Cowper (2005), in which plural is more marked, and I further demonstrate that Harley and Ritter’s geometry is not able to straightforwardly capture the Gangalidda facts. Section 1 outlines the feature geometries proposed by Harley and Ritter (2002) and Cowper (2005). Section 2 presents the Gangalidda data, illustrating the contextual neutralization of dual and plural clitics. Section 3 gives a featural account of the Gangalidda system, supporting the geometry in Cowper (2005).