Australian aboriginal life is almost nightmarishly difficult to find out about. There are a few popular books on the subject, but these tend to treat indigenous Australians as noble savages or portray ancient Australia as a place with little meaningful cultural variation. The traditional image of Australian aborigines boils down to little more than living in deserts, foraging for food with boomerangs, killing kangaroos with spears, and maybe having the odd dance. As there are few accurate popular works dealing with the actuality of Australian life prior to Europeanisation, it is unlikely that this popular image will change. Even the academic works are hard to get hold of or prohibitively expensive, and even if you live in Australia it is easier to find books on the prehistory of Indo-European than it is to find out about Pama-Nyungan (see below).
But in fact there are plenty of fascinating aspects of indigenous Australian life that deserve greater appreciation. It seems as if it has always been quite a diverse place in terms of human culture. Although all Australians prior to the arrival of Europeans and Makassarese in the late eighteenth century relied on foraging, there are lots of different kinds of foraging strategies and lots of ways in which foragers can eke out a living. We shouldn't for a moment think that Australians were uniform across the seven-and-a-half-million square kilometres of the island-continent or their (probably) 50,000 years of human history. In fact, a high level of diversity appears to have been present from the beginning, with riverine, lacustrine, montane, and marine environments all seeing some form of exploitation at human hands by 40,000 year BP (although the desert environments we most closely associate with Australia's first people were almost certainly uninhabited until much later on). Add in Tasmania and the Torres Strait and we're suddenly confronted with an even greater diversity of attitudes, influences, and environments (including the famous taboo on eating fish among indigenous Tasmanians).
So Australia has always been diverse, with moth hunters and palm-nibblers and lake fishermen. But the mid-Holocene, around 4,000 years ago, is when things get really interesting, at least from my perspective, with the development of a tradition of making backed unifacial blades that swiftly spread across the continent.
You may have noticed that I have an especially strong interest in the intersection between archaeology and linguistics, and that I am interested in the spread of language families across the earth. You may also be aware that I am somewhat critical of the theory that language families inevitably spread with agriculture, the so-called language/farming dispersal hypothesis, whose principal proponents are Peter Bellwood and Colin Renfrew. It is true in some instances that language families correlate with or are enabled by the spread of agriculture, as is probably true of Austroasiatic in mainland southeast Asia, of Sino-Tibetan (or Tibeto-Burman) in China and the Tibetan Plateau, and of most of the Niger-Congo expansions in Africa (the latter is somewhat controversial, as Roger Blench has suggested that the initial expansion of Niger-Congo was under way before agriculture came to the fore in West Africa - but I digress). It is probably not true in other instances: Altaic, Indo-European, and Iroquoian seem to have relied on other mechanisms. Uto-Aztecan is a trickier one, and much debated (pfft, as if Indo-European isn't...).
It's also clear that languages can spread in other ways: English is not a global language because of the spread of agricultural techniques, at least not primarily, and the Romans bought or commandeered so much wheat from Egypt and Spain that they barely developed any agricultural innovations of their own despite spreading their language across Europe and north Africa. What matters is that people in other groups have a reason to start speaking the new language, or that the group whose language is being propagated has sufficient strength to annihilate and replace the original population - not whether they have a special grain or tuber. The correlation with agriculture is guaranteed to be weak.
The best example against the language/farming dispersal hypothesis comes from Australia. Pama-Nyungan languages once covered about 7/8 of Australia's territory, including the entire east, west, south, and centre. Pama-Nyungan
is a well-supported genetic entity and its languages show close
relationships to one another, demonstrating that this is a family that
spread relatively recently - in the mid-Holocene, perhaps, at around
the same time as Indo-European. Only Tasmania and Arnhemland preserved non-Pama-Nyungan languages, and even that latter swampy, tropical land was penetrated by Yolngu, a Pama-Nyungan language on the northern tip.
Modern Pama-Nyungan speakers are also much more likely to carry the HTLV-1 retro-virus than non-Pama-Nyungan speakers, and as this is inherited, at least in part, we are probably looking at the expansion of a group of people - or at the very least we can say that the speakers of Pama-Nyungan languages share a high degree of inherited material. This means that the expansion must have been more like true migration than advection, although of course the reality of it was probably quite complicated (as it is with any language family). Just so you're in no doubt, this means that the Pama-Nyungan expansion is attested to by both linguistic and biogenetic evidence.
Now, Australians didn't develop agriculture before Europeans arrived, so we can't link the expansion to the domestication of any grains or animals or whatever else. This is one of the few language family expansions that we can say right off the bat had absolutely no connection to agriculture. The trouble is, the model proposed by linguists is a little old and as far as I can tell, it doesn't fit perfectly with the archaeological evidence as we now have it.
The theory proposed by linguists Nicholas Evans and Patrick McConvell goes like this: In the mid-Holocene, there was a major intensification of resource exploitation. Cycads and other seed-bearing plants were harvested in intensive ways with the growth of a delayed-return economy across much of the continent. This is linked to the development of backed unifacial blades and points, which is certainly useful for archaeologists investigating the spread of the family as these are very durable entities. The overlap between unifacials and Pama-Nyungan is almost perfect, and unifacial blades are absent archaeologically in almost all non-Pama-Nyungan-speaking territory.
So, they claim, Pama-Nyungan speakers may have had an economic advantage even when moving into densely-populated areas, like riverine central Queensland (which was far from sparsely populated in the mid-Holocene), albeit one not quite comparable to the advantage bestowed by agricultural production, as a result of their superior tools and their greater surplus. This in turn provided the ability to hold massed gatherings for ritual purposes, enhancing the prestige of the group and thereby leading to the adoption of Pama-Nyungan languages by non-Pama-Nyungan-speaking peoples.
Evans and McConvell also link Pama-Nyungan to a complex of outward-looking, inclusive, social traits. They believe that Pama-Nyungan speakers emphasised patrilineal descent, in contrast to their non-Pama-Nyungan neighbours; that they were more likely to engage in group exogamy, marrying people in tribes other than their own; and that they preferentially linked themselves and others in ritual gatherings and song cycles, emphasising links across the landscape and deep into the past. (This is incidentally similar to the manner by which Arawakan is thought to have propagated itself in South America.)
The Urheimat of proto-Pama-Nyungan is hypothesised to have been in Queensland between 5-3,000 years ago, with the initial expansion occurring down the east of Australia over subsequent millennia. 2-1,000 years ago, a further expansion occurred, this time into the desert and the west of the continent. The recent expansion into this area is well-attested by the linguistic data, especially the similarities across the languages of desert Australia (including such widespread languages as the Western Desert language).
Anyway, that's the Evans and McConvell model. Unfortunately - unfortunate because it's such a neat and elegant model - the archaeological situation isn't quite right. While Pama-Nyungan languages certainly correlate with unifacial blades, the intensive exploitation of resources reported for the mid-Holocene appears to have been an illusion - or, rather, its sudden appearance is an illusion, with the intensive exploitation of seed-bearing plants and the likely growth of ritual gatherings occurring many thousands of years earlier in the Pleistocene. Delayed-return economies were presumably not new in Australia and the Pama-Nyungans would have had no necessary advantage if they had focused on intensive exploitation of cycads or similar products.
Given the correlation with unifacial blades, another model of Pama-Nyungan can be proposed. We could even say that any model that accounts for the spread of unifacial blades also accounts for the spread of Pama-Nyungan due to the obvious overlap in both space and time between the two phenomena. Fortunately, we have such a model, one proposed by archaeologist Peter Hiscock. This is essentially a risk reduction model: backed unifacial blades, which are exceptionally versatile, were developed and used by native Australians as they provided a means of reducing the risk inherent in subsistence foraging.
This was likely linked to climate change. Precipitation in Australia was lowest between 4000 and 2000 years ago (ie, the mid-Holocene) as a result of globally changing weather patterns (including the development of what we now call the El Niño effect). Drier and more variable climatic conditions came to dominate. This made foraging especially difficult and subsistence must have been threatened, with resources more widely scattered and much less predictable. Backed unifacial blades - about as versatile as a Swiss army knife in the right hands, capable of butchering animals, scraping vegetables, you name it - conferred a distinct advantage on those who used them. They weren't projectile points but rather all-round tools for gleaning a living from the land. They were also economical tools, easily and efficiently knapped in large quantities. There appears to have been a trend of cost reduction in producing blades at this time, regardless of what kind they were. Blades of all kinds (not just unifacials) from across the continent show signs of repeated re-sharpening. This independently testifies to the need for economy and efficiency in tool design in the mid-Holocene.
It seems that far from Pama-Nyungan spreading due to the development of a more outward-looking culture related to the intensive exploitation of cycads and the consequent growth of ritual gatherings, it was instead caused by a less edifying motive: the need for survival in a land swiftly becoming harsher, tougher, and less predictable. This would also make sense of the fact that Pama-Nyungan languages correlate strongly with identifiable biogenetic features, assuming the backed unifacials were spread across the landscape by a migration of Pama-Nyungan speakers.
Either way, in Pama-Nyungan we have one of the only uncontested examples of a language family's expansion occurring outside of an agricultural or pastoral setting. It shows that non-agricultural expansions are possible, and that connecting agricultural spread to language family is not always appropriate - certainly not inevitably appropriate. We don't need to connect Austronesian to the first introduction of agriculture to island southeast Asia and we don't need to connect Indo-European to Neolithic Thessaly or Mehrgarh. Agriculture clearly isn't the only motive for the spread of a language family nor its adoption by other groups of people, and instead we should think about language adoption in terms of plausible reasons for action, and about the migrations and diffusions of human beings in terms of the same.
Post-scriptum: It is possible that Evans' and McConvell's views have changed since the publication of their article in Archaeology and Language II in 1998, but it is difficult to find out more without owning a copy of their more recent text on the subject, Archaeology and Linguistics: Aboriginal Australia in Global Context (which, as far as I can tell, corroborates the model outlined above, although I can't absolutely verify that without another trip to the Bodleian library).
The following texts have been most useful to me in understanding the issue:
Hiscock, Peter. 2006. 'Blunt and to the point: changing technological strategies in Holocene Australia'. In I. Lilley (ed.). Archaeology of Oceania: Australia and the Pacific islands. Oxford: Blackwell.
---------- 2008. The archaeology of ancient Australia. Routledge: Abingdon.
McConvell, Patrick. 1996. 'Backtracking to Babel: the chronology of Pama-Nyungan expansion in Australia'. Archaeology in Oceania 31: 125-144.
McConvell, Patrick and Evans, Nicholas. 1998. 'The enigma of Pama-Nyungan expansion in Australia'. In Blench, Roger, and Spriggs, Matthew. Archaeology and language II: archaeological data and linguistic hypotheses. Routledge: Abingdon.