Measured Language: Quantitative Studies of Acquisition, Assessment, and Variation

5. Subgrouping in Nusa Tenggara: The Case of Bima-Sumba

63 5 ■Subgrouping in Nusa Tenggara The Case of Bima-Sumba EMILY GASSER Yale University ■ THE AUSTRONESIAN LANGUAGE FAMILY is one of the largest and most diverse in the world, but despite extensive historical work by scholars over the past century, much of its internal structure remains poorly understood. The Bima-Sumba (Bi-Su) subgroup, consisting of roughly twenty-seven Central Malayo-Polynesian (CMP) languages (Lewis 2009), was first proposed in 1938 by the Dutch language officer S. J. Esser as part of an atlas of colonial Indonesia. Although it has since been repeatedly cited in the linguistic literature, the first paper to investigate Bima-Sumba’s validity as a subgroup was by Robert Blust (2008), who concluded that this particular collection of languages is not in fact monophyletic; that is, they cannot be traced back to a single exclusive common ancestor. Here, I return to the question of Bima-Sumba’s existence as a legitimate and cohesive subgroup, using Bayesian phylogenetic tools to reconstruct a likely family tree using lexical data. These methods have not been uncontroversial (see, e.g., Eska and Ringe 2004), but have repeatedly been shown to be accurate and useful in historical linguistics (e.g., Dunn et al. 2008; Greenhill, Drummond, and Gray 2010; Greenhill and Gray 2009) and have recently been used to investigate historical hypotheses on a number of language families (e.g., Bowern and Atkinson, in press; Gray and Atkinson 2003; Gray and Jordan 2000; Rexová, Bastin, and Frynta 2006). The analysis presented here shows that when the FloresLembata (FL) languages are included, Bima-Sumba does indeed constitute a single clade (i.e., a branch of the tree), contradicting Blust’s finding. Based on this result, I propose the recognition of a new subgroup, Bima-Sumba-Flores (BSF), encompassing all of the Bima-Sumba and Flores-Lembata languages. Subgrouping Hypotheses within CMP Figure 5.1 illustrates schematic trees that show a simplified version of the four main subgrouping hypotheses discussed below. Tree (a) represents Esser’s original classification ; (b) represents the current standard classification, as listed in the Ethnologue; (c) is Blust’s proposed classification; and (d) shows the Bima-Sumba-Flores hypothesis presented here.1 Tree (e) is a simplified excerpt of the Austronesian family tree produced by Gray, Drummond, and Greenhill (2009), which constitutes independent 64 Emily Gasser ■ Figure 5.1 Five Subgrouping Hypotheses a. c. b. d. e. 65 SubGroupinG in nuSa TEnGGara corroboration of the Bima-Sumba-Flores hypothesis. Further differences in higherlevel structure between their tree and the other modern hypotheses are discussed below. The Bima-Sumba-Flores languages belong to the Central Malayo-Polynesian branch of Austronesian. The first Bima-Sumba proposal appeared in 1938, when the Dutch colonial government in Indonesia published their Atlas van Tropisch Nederland. This volume included Esser’s categorization of a number of the languages of Indonesia into nineteen implicitly genetic groupings: seventeen Austronesian and two non-Austronesian (Esser 1938). The languages currently recognized as part of the CMP were divided by Esser into the Sula-Bacan, Ambon-Timor, and BimaSumba groups. It is unclear how Esser arrived at his classification, as he presents no supporting evidence for his decisions. His accounting of the languages of Indonesia was far from exhaustive, and his list of the Bima-Sumba languages includes six of the twenty-seven classified as such in the Ethnologue (Lewis 2009). Several conflicting subgrouping hypotheses regarding the Sula-Bacan and Ambon-Timor groups have been proposed since Esser (e.g., Blust 1981; Campbell 2004; Collins 1983; Dyen 1965; Hughes 1987; Mills 1991). As these proposed groupings are not mutually compatible, in this analysis I sample from the subgroups of CMP as listed in the Ethnologue, though the source of these groupings is not clear and they may well fail to hold up to further investigation. Nevertheless, sampling from this classification ensures wide genetic and geographical coverage in the languages used here. Investigations of Bima-Sumba Bima-Sumba, however, has not received such scrutiny. Bima-Sumba has been generally accepted as a classification for Bima2 , Savu, and the languages of Sumba and western Flores (e.g., Forth 1988; Klamer 1994; Musgrave 2008), but little evidence has been presented to support its existence. Blust (2008), using the comparative method, was the first to examine Bima-Sumba. He compares four Bi-Su languages representative of the geographic span of the family and fails to find sufficient numbers of exclusively shared...