A Case of Intensive Lexical Diffusion: Arnhem Land, Australia

Abstract

The languages of Arhem Land, Australia, form a contiguous block and Sprachbund, despite a sharp genetic boundary. Thus in Ngandi and Ritharngu, which belong to different major genetic divisions, about half the total vocabulary is shared, virtually all because of recent diffusion. 'Core' vocabulary and highly inflected stems (verbs) are among the borrowed items, but linguistic structural factors affect the extent of diffusion. Extensive over-all borrowing is attributed to local demographic and cultural features: small size of ethnolinguistic group, absence of strong tribal organization, and extensive cross-linguistic marriage.* 1. BACKGROUND. In eastern Arnhem Land, along the coast of the Gulf of Carpentaria, several ethnolinguistic groups (linguistic 'tribes') were in close contact with each other prior to the recent intrusion of whites. I contend that some of these languages came into contact at a fairly recent time (let us say a thousand years ago), and that lexical diffusion among them has reached a level not reported elsewhere for stable multilingual zones. The plausibility of this claim is increased by my previous documentation (Heath 1978a) of extensive phonological and grammatical diffusion involving the same languages. In addition to structural convergence, involving inherited materials, I showed that direct borrowing of many bound affixes had occurred: noun-class prefix systems, negative suffix, denominative verbalizing suffix, pre-inflectional thematizing verbal suffix, comitative affixes (with nouns and verbs), several case suffixes (ergative-instrumental, genitive-dative, ablative, instrumental), dyadic dual suffix with kin terms, and various others. Before proceeding, I will review the evidence for the genetic relationships posited among the languages. The crucial division is between the Yuulngu group, bottled up in northeastern Arnhem Land, and the so-called 'prefixing' languages which surround the Yuu. group to the south and west. The prefixing languages, or at least those involved in this paper, are a genetic group, but the fine-grained subgrouping has not been worked out. The Yuu. group consists of approximately ten languages; and except for one or two Yuu. offshoots in the west which do not concern us, the group is internally quite homogeneous. That the Yuu. languages are closely related has been obvious on lexical and grammatical grounds to all observers, and some of the 'languages' grade into each other in dialect-chain fashion. Adjoining Yuu. languages are approximately as similar to each other as Spanish to Portuguese, or German to Dutch. The wider genetic connections of the Yuu. languages are not to the adjoining prefixing languages, but rather to languages far to the south-in the southern