Subgrouping in Nusa Tenggara: The case of Bima-Sumba

Emily Gasser, Yale University

Abstract

An important moment in the classification of the languages of Indonesia came in 1938, when the Dutch government published its Atlas van tropisch Nederland. Included in this atlas was a map, compiled by the colonial language officer S. J. Esser, showing the locations of many of the languages of the archipelago and breaking them down into nineteen groups. Some of these groupings have since been borne out by evidence, while others have been shown to be incorrect. According to Blust (2008), the Bima-Sumba subgroup has since been accepted in the linguistic literature, but no evidence has been presented to support its existence. Blust argues, based on shared innovations, fossilized morphology, and phonological irregularities in a set of four representative languages that while there likely is a Hawu-Sumba subgroup, Bima should only be included in a higher-order group with other nearby languages as well.

I reevaluate Blust’s claims about the Bima-Sumba subgroup using computational methods. Wordlists representing 17 Bima-Sumba languages and three related Flores-Lembata languages were coded for cognacy and run through the Splitstree, TraitLab, and Beast programs, which use Bayesian methods to produce a likely family tree. Based on the resulting phylogenies, I argue for the existence of a single Bima-Sumba group divided into two major subgroups: one consisting of the languages of Flores, including the traditionally separate Flores-Lembata languages, and another consisting of the languages of Hawu and Sumba.