Wednesday, 15 January 2014

Pama-Nyungan Expansion - Bellwood's Account

Peter Bellwood's First Migrants (Wiley-Blackwell 2013), a book on prehistoric migration, has plenty of sections devoted to proposed early- and mid-Holocene language family dispersals, including, of course, Indo-European, Afroasiatic, Austroasiatic, and Austronesian.  But Bellwood also talks about Pama-Nyungan, an indigenous Australian family that I discussed in an earlier post on this site.  Pama-Nyungan is a good example of a language family that spread over a wide area without the help of plant or animal domestication, and so it runs counter to Bellwood's earlier view of language families spreading primarily due to the adoption of agriculture by the protolanguage-speaking population (known as the 'farming/language dispersal hypothesis').  Bellwood gives a good account of the Pama-Nyungan expansion, clear evidence of his disavowal of strong forms of the farming/language dispersal hypothesis since the publication of his First Farmers nearly ten years ago.

File:Macro-Pama-Nyungan languages.png
Distribution of the Pama-Nyungan languages (yellow).  Non-Pama-Nyungan in other colours.  As I say below, it matches the distribution of backed unifacial stone blades.  h/t Wiki.
While I disagree with Bellwood's conclusions about Indo-European and some other families, his discussion of Pama-Nyungan has several good features:
  • a) Bellwood accepts the view of the majority of Australian linguists that Pama-Nyungan is a real, genetically-related language family, descended from a common ancestor; this view was challenged by R. M. W. Dixon, the maverick Australian linguist, but is now the consensus.
  • b) Bellwood also accepts that the family is of a relatively recent origin, and isn't, as Dixon and others claimed, a disparate phylum dating backed tens of thousands of years.  Bellwood places its origin in the mid-Holocene, c.4000 BP (approx. 2000 BCE), as did Pat McConvell, and others.
  • c) Bellwood also associates it with the spread of backed unifacial blades (stone blades intended to be attached to handles and used for a wide variety of purposes) across most of continental Australia in the mid-Holocene.  There is a strongly overlapping distribution of backed unifacials at archaeological sites and Pama-Nyungan languages, and backed unifacials are unknown in the areas in which non-Pama-Nyungan indigenous languages were once spoken.  I discussed this in my post on the topic, and it forms the core of my own argument about Pama-Nyungan.
Bellwood's view has another feature about which I am undecided, and that is that:
  • d) he associates backed unifacials, and therefore implicitly Pama-Nyungan, with the Toalian (or Toalean) archaeological culture of mid-Holocene South Sulawesi, Indonesia.

I'll discuss d) first.  The Toalean is an archaeological culture of South Sulawesi best known from the site of Maros, excavated by Ian Glover in the 1960s.  It has been described by David Bulbeck et al here, in a paper on the wonderful website of OXIS, run by Ian Caldwell.  It was a pre-Neolithic (and therefore almost certainly pre-Austronesian) archaeological culture that I'm familiar with primarily due to the presence of arrowheads in the assemblage, the so-called 'Maros points', conventionally dated c.5500-3500 BP (approx. 3500-1500 BCE).  (I have a research proposal on archery in eastern Indonesia that I'm hoping to get funding for, so this is familiar territory for me.)
Maros points in the British Museum, excavated by Ian Glover.  I think these are the very ones on display in the Southeast Asia section.  h/t The British Museum.
Bellwood associates the Toalean with Australia on the basis of the shared technology of backed unifacials, which do not appear elsewhere in southeastern Eurasia or Australasia at this time.  Sulawesi is a long way from Australia, and the technology doesn't appear anywhere en route between the two, but Bellwood notes that later fishermen and teripang (sea cucumber) catchers from Sulawesi made journeys to Australia in the nineteenth century without bothering with Timor or Flores, and therefore claims that the distribution isn't so strange.

File:Holothuroidea (Sea cucumber feeding).jpg
All the way from Sulawesi to Australia for this.  h/t Wiki.
I'm not sure whether I accept Bellwood's analogy with teripang fishermen, and I'm not sure I agree with the idea that the Toalean and Australian backed unifacials are related.  Certainly the presence of arrowheads in later Toalean assemblages speaks against an Australian connection.  The bow was not known in prehistoric Australia; Bellwood has confirmed to me that he believes the arrowheads are a late-Toalean technology and, according to his view of the contentious dating of Toalean, post-date his proposed Australian connection.  I suppose it is, therefore, a plausible connection.

The technologies are similar and the distance is not insurmountable, so what we need is more research.  Specifically, we need some ancient DNA ('aDNA') from recovered Toalean specimens to compare with Australian aDNA from mid-Holocene sites - this is probably the only viable research option, given that the pre-Austronesian languages of South Sulawesi have entirely disappeared from the earth.  Ancient DNA is difficult, though, and given that the native inhabitants of Sulawesi probably had a lot in common, genetically, with indigenous Australians, it would be hard to show even if good samples could be recovered, which isn't assured.  Bellwood concedes this, and I think he's right to raise the possibility of a Sulawesian connection.*

Now for a), b), and c).

a) and b) I'll leave to the linguists.  As a layperson when it comes to the details of ancient Australian languages, I can only go with the consensus.  The consensus is that Dixon was wrong, and that Pama-Nyungan is a language family with a single ancestral language, proto-Pama-Nyungan, which had a recent origin in the mid-Holocene.

Moreover, the distribution argument is compelling, and makes sense of Pama-Nyungan's time-depth: we may conclude that Pama-Nyungan and backed unifacial blades are related and dispersed together.  What this means is that if we can make sense of the distribution of backed unifacial technology in Australia, we can probably make sense of the dispersal of Pama-Nyungan, as the two seem to be related beyond reasonable doubt.

The earlier Evans-McConvell model of Pama-Nyungan explained the expansion of both the language family and the stone tools by proposing that, in the mid-Holocene, favourable climate made resource exploitation easier (especially foods from wild cycads), and that the people most affected by this used the surplus generated to hold large celebrations that could help incorporate new members into their groups.  These people were supposed to be the speakers of proto-Pama-Nyungan, who used backed unifacials to better exploit the abundance.  Bellwood takes the model as given, although he doesn't explicitly endorse it (he's more interested in the Toalean connection).

The problem with this is that it is now known to be wrong, as Peter Hiscock shows in his Archaeology of Ancient Australia.  The large-scale cycad exploitation actually dates to over ten thousand years prior to the mid-Holocene, and probably had nothing to do with Pama-Nyungan; certainly there was no explosive growth in resource exploitation at the time Pama-Nyungan probably expanded.  More importantly, the climate was at its worst in the mid-Holocene, with extremely dry conditions prevalent across the continent.  Backed unifacials therefore probably came about as an efficient means of overcoming the problem of harvesting scarce resources in a dessicated environment, and the tools probably spread so quickly across Australia due to the need for even small populations to spread themselves thinly in order to exploit the available resources.

Hiscock considers this the most reasonable argument about backed unifacial technology (without mentioning the Pama-Nyungan connection), and, given the correlation between Pama-Nyungan and backed unifacials, it must also be the ultimate explanation of Pama-Nyungan.  While Bellwood says nothing that invalidates this view, it is curious that he doesn't mention it, despite citing The Archaeology of Ancient Australia in his account (albeit about a different topic).  I find this strange, and assume it means that the archaeologists and the linguists need to have a chat about backed unifacials and Pama-Nyungan, because they aren't reading from the same hymn sheet.

Despite this absence, Bellwood's account is surprisingly good.  I say 'surprisingly', because some of Bellwood's views take unorthodox positions on language that swim strongly against the mainstream, but in this case, most of what he says is part of the consensus, and, as ever, it's compellingly-argued and based on recent and interesting research.  I already recommended it in my 2013 review, but I'll recommend it here again: First Migrants is a good book about human migration in prehistory, and covers a lot of obscure subjects in interesting ways, including Pama-Nyungan.  Where the mainstream disagrees with him, at least Bellwood proposes interesting and possibly worthwhile interpretations, although I do feel he could pay more attention to the linguistic arguments than he does at present.

*We need aDNA because some people have claimed that Pama-Nyungan/the dingo/backed unifacials were transmitted to Australia by migrants from India.  This is on the basis of a single study showing that modern indigenous Australian populations sometimes show the presence of genotypes from South India.  Without aDNA, no realistic case can be made (and in any case, the dingo's DNA is clearly Southeast Asian).


  1. I haven’t seen ‘First Migrants’, but I’m looking forward to reading it. However, I have to take issue with your characterization of RMW Dixon.

    He is still alive, and it’s misleading to call him a ‘maverick’. On the contrary, he has been very influential in language typology and field linguistics – witness the string of OUP volumes co-edited with A. Aikhenvald and his 3-volume ‘Basic Linguistic Theory’. In ‘The Rise and Fall of Languages’, Dixon presents the Pama-Nyungan hypothesis as based on glottochronology and lexicostatistics, methods that any intro historical linguistics textbook will tell you have been discredited. There’s an accepted division of linguistic areas into spread zones (Europe, island SEA/Oceania) and residual zones (the Caucasus, Papua New Guinea, Australia). Dixon explains how family trees work much better in the former, while factors like diffusion make them problematic in the latter.

    Johanna Nichols’ ‘Linguistic Diversity in Space and Time’ is also worth reading about these issues. Archeologists have intervened in these matters and been proved wrong in the past – witness Renfrew’s ‘Archeology and Language’.

  2. Both Aikhenvald and Dixon are somewhat outside the mainstream, and both present controversial personal hypotheses as uncontested. They're not on the fringe, and they argue their positions well, of course. But Dixon's 'Rise and Fall' doesn't exactly present the state-of-the-art, and it is far more controversial than the idea of Pama-Nyungan genetic relatedness. 'Maverick' is right.

    'Late' isn't, though, so thank you for the correction.

  3. Maybe this will help you: "YOLNGY SEA RIGHTS IN MANBUYNGA GA RULYAPA (Arafura Sea)" Ian McIntosh.

    There's a part about "north-east Arnhem Land Aborigines" who had contact with people from Eastern Indonesia and the relation with a certain Moluccan island.

    Goodluck with your research and funding on Eastern Indonesia.


You can post anonymously if you really want to, but I would appreciate it if you could provide some means of identifying who you are, if only for the purpose of knowing who has written what.