ALS presentation has the slides for my recent talk at the Australian Linguistic Society’s Annual Conference.

Abstract

Tangkic and Pama-Nyungan: Sister or Subgroup?

This paper examines the status of Tangkic and its relationship to Pama-Nyungan. While early classifications (e.g. O’Grady, Wurm & Hale 1966) placed Tangkic as one of the primary subgroups of Pama-Nyungan on primarily typological grounds, since Evans (1990; 1995) and Blake (1990) it has been customary to treat Tangkic as non-Pama-Nyungan, albeit a close relative (the closest family apart from Garrwan; though see Harvey (2009) for discussion). Evans’ basis for classification was the pronominal system, with the reconstructions of Tangkic forms being rather different from those typically reconstructed for Pama-Nyungan (e.g. by Blake 1990). That in itself might not be evidence against shared genetic relationship however, since, after all, pronouns do change, and other Pama-Nyungan subgroups (e.g. Yolŋu, Karnic) are reconstructed with at least some forms other than those reconstructed to Proto-Pama-Nyungan (Schebeck 2001; Bowern 1998).

Bouckaert, Bowern, and Atkinson (2018) include Tangkic among the groups discussed in their phylogeny of Pama-Nyungan. In their tree, Tangkic is a subgroup within Pama-Nyungan, not a sister to the Pama-Nyungan family. This classification is based on sparce lexical cognates, but ones which might a priori be thought to be indicative of shared genetic relationship. We might therefore say that Tangkic is a “sparce evidence” question – that is, there are sufficient differences between the reconstructed pronominal systems to cast doubt on the genetic affiliation. Lexical data is historical distrusted in Australia (see e.g. Alpher 2004). The best we might say at this point is that the claim is “not proven”.

In order to evaluate a claim like “Tangkic is part of Pama-Nyungan”, we need to evaluate how likely words are to be jointly descended from a common Proto-language. We need to consider not just individual lexical items and key cognate, but the distribution of inheritances across the lexicon. For example, what is the distribution of identified loanwords: do matches cluster in semantic fields that are associated with borrowing? do matches cluster in vocabulary which is known to be frequently borrowed elsewhere? What is the relationship between cognate forms, frequency, and stability? Highly stable items are more likely to be cognate than less stable items if the relationship is genealogical. Old relationships should be reflected by slowly changing words If the matches are mostly in words known from elsewhere to change quickly, that is evidence against relationship (though not necessarily evidence for borrowing).

Phylogenetically, we can look at the distributions of candidates across subgroups. If words are genuinely old, they should not just be found in the closest subgroups (geographically) to the candidate language/subgroup. Because retention is not diagnostic of subgroupiness, retentions should be more scattered across the family than innovations: that is, if arguing for relationship purely on shared retention, the retained items would not be expected to cluster in a single subgroup. Conversely, if the items appear to cluster with a single group or geographical area, that is indicative of loans.

The evidence for Tangkic here is mixed. On the one hand, the lexical cognates do not cluster in particular subgroups of Pama-Nyungan. There are only 12 clear cognates in the list used but they are spread across the family. In contrast, cognates between Garrwan and other Pama-Nyungan languages tend to cluster in adjacent subgroups. Conversely, the standard of evidence used to exclude Tangkic, if applied elsewhere, would see Pama-Nyungan fractured into several different groups. For example, Western Torres exhibits a similar degree of cognacy to Tangkic, as do several other subgroups. Tangkic’s non-singular first person pronouns have a stem ŋarr-; that form is also found in singulars in Proto-Yolŋu.

References
Alpher, Barry. 2004. Pama-Nyungan: phonological reconstruction and status as a phylogenetic group. In Claire Bowern & Harold Koch (eds.), Australian Languages: Classification and the Comparative Method, 93–126. Amsterdam: John Benjamins.
Blake, Barry J. 1990. Languages of the Queensland/Northern Territory border: Updating the classification. In Peter Austin, R. M. W. Dixon, Tom Dutton & Isobel White (eds.), Language and history: Essays in honour of Luise A. Hercus (Pacific Linguistics C–116), 49–66. Canberra: Dept. of Linguistics, Research School of Pacific Studies, Australian National University.
Bouckaert, Remco R., Claire Bowern & Quentin D. Atkinson. 2018. The origin and expansion of Pama–Nyungan languages across Australia. Nature ecology & evolution 1. https://doi.org/10/cmg9.
Bowern, Claire. 1998. The case of Proto-Karnic: morphological change and reconstruction in the nominal and pronominal system of Proto-Karnic (Lake Eyre Basin). Canberra: Australian National University.
Evans, Nicholas. 1990. The Minkin language of the Bourketown region. In Geoffrey N. O’Grady & Darrell Tryon (eds.), Studies in Comparative Pama-Nyungan, 173–207. Canberra: Pacific Linguistics.
Evans, Nicholas. 1995. A grammar of Kayardild. Mouton de Gruyter.
Harvey, Mark. 2009. The genetic status of Garrwan. Australian Journal of Linguistics 29(2). 195–244.
O’Grady, G. N., S. A Wurm & K. L. Hale. 1966. Aboriginal languages of Australia (a preliminary classification). Victoria, BC: Department of Linguistics, University of Victoria.
Schebeck, Bernhard. 2001. Dialect and social groupings in northeast Arnheim [i.e. Arnhem] Land (LINCOM Studies in Australian Languages 7). (Ed.) R. M. W. Dixon. Muenchen: LINCOM Europa.