Proto-Australian language
Proto-Australian is the reconstructed common ancestor of nearly all Australian Aboriginal languages. It is hypothesised to have been spoken approximately 6,000 years ago.
The Australian mainland is home to at least 250 distinct languages. Serious study of the Australian languages by trained linguists dates from the beginning of the twentieth century. A Vienna-based Catholic priest, Wilhelm Schmidt, was the first to attempt to classify Australian languages into different groups, using the limited material then available. Arthur Capell was the first to suggest one ancestral language for the entire continent, which he termed "Common Australian", in 1956.
Since the 1960s, most specialists have agreed that the majority of Australian languages belong to a single language family, named Pama-Nyungan. The Pama-Nyungan languages, which stretch across approximately seven eighths of the continent, from the rainforests of the Cape York Peninsula in the north east, across the deserts of the interior to the southwesternmost extremity of Western Australia, are believed to descend from a single ancestral language, Proto-Pama-Nyungan, spoken thousands of years ago. This language has been partly reconstructed.
The remaining languages occupy the remaining eighth of the continent, including Arnhem Land, the Kimberly region, and the Gulf Country west of the Gulf of Carpentaria. These languages have been classified into many separate families, sometimes termed "non-Pama-Nyungan" as a term of convenience. While the non-Pama-Nyungan languages are distinct from each other, and from Pama-Nyungan, there are certain similarities, for example the use of pronoun prefixes, that suggest that they, and the Pama-Nyungan languages, descend from an even older language. This language is called Proto-Australian.
The earliest attempts to reconstruct Proto-Australian focused on PN languages. More recently, comparative work on NPN languages has assisted in identifying aspects of the ancestral language, which can be assumed to be grammatically more similar to non-Pama-Nyungan languages than to the Pama-Nyungan languages. Despite the diversity of the linguistic picture, a common set of noun class prefixes, ancestral to both Pama-Nyungan and Non-Pama-Nyungan, was reconstructed using the comparative method in 2017, providing evidence for Proto-Australian. This was then extended to a book-length reconstruction of Proto-Australian, published in 2024.
History of the proposal
The Australian mainland is home to more than 250 distinct languages. While these languages are diverse, it has long been recognised that they share important phonological, grammatical and semantic similarities and are likely all genetically related.Some attempts were made in the nineteenth century to establish relations between Australian languages, but these were based on isolated wordlists and are considered "amateurish" by modern linguists.
Rigorous study of Australian languages began with the twentieth century. In 1919, Wilhelm Schmidt was the first scholar to systematically survey the languages of the whole continent. Schmidt created various groupings for Australian languages, which he did not term "families". He argued that the "southern" languages were clearly related to each other, while the "northern" languages were not. Schmidt's work, based in many cases on very limited available evidence, was "heroic" but is now considered outdated. Arthur Capell was a pioneer in classifying the typology of Australian languages, but also in proposing that all Australian languages, no matter their typological differences, belong to a single family, which he called 'Common Australian' in 1937. He subsequently suggested the importance of common verb forms in reconstructing this hypothetical ancestor. During his career Capell made important observations about the typological differences between Australian languages and how these might have come about.
Typological issues
For some time linguists have classified Australian languages into two broad typological groups. Languages of the first group are dependent-marking, relying heavily or exclusively on suffixation, and lack grammatical gender. These languages are sometimes called "suffixing" languages, and are typical of the Pama-Nyungan family, which covers approximately seven-eighths of the continent. It is generally assumed that Proto-Pama-Nyungan was of this type. The second type of languages is head marking, usually with a complex of bound pronouns prefixed to the verb, and makes use of a number of what are called noun classes or grammatical genders. These languages are commonly termed "prefixing" languages.The only "prefixing" Pama-Nyungan language is Yanyuwa; all other prefixing languages, geographically occupying the remaining eighth of the continent, have been classified into over twenty separate families.
The continent-wide classification of Australian languages into Pama-Nyungan and assorted non-Pama-Nyungan families relies particularly on the work of Kenneth L. Hale, Geoffrey O'Grady, Carl Voegelin, Florence M. Voegelin and Stephen Wurm, published between 1966 and 1977. This classification posited that all Australian languages ultimately formed a single genetic group. This classification was expressly designed as "provisional", but nonetheless became widely accepted.
A question for Australian comparative linguistics then became the nature of the relationship, if any, between the prefixing, non-Pama-Nyungan languages, and the Pama-Nyungan languages.
Pama-Nyungan, diffusion, and Dixon's position
In 1980 R. M. W. Dixon, a central figure in Australian linguistics, published The Languages of Australia, in which he asserted "all but two or three of the 200 languages of Australia can be shown to belong to one language family - the 'Australian family. Dixon provided some phonological and morphological reconstructions of what he terms "Proto-Australian" in this volume. Because the set of languages that he was using as a source was largely Pama-Nyungan languages, most linguists consider that these reconstructions are more appropriately treated as reconstructions of Proto-Pama-Nyungan. Dixon's position, meanwhile, is that there was no basis for assuming Pama-Nyungan is a genetic grouping, and the distinction between "Pama-Nyungan" and "non-Pama-Nyungan" is purely typological and not genetic.In 2002, Dixon published Australian Languages: Their Nature and Development. In this volume he revised his earlier assessment that all Australian languages could be proven to be genetically related. Instead he argues forcefully that phonological, morphological, and lexical diffusion in Australia is so pervasive and the time depths involved are so vast that it is impossible to use the comparative method to reconstruct any large-scale family relationships. He continues to argue that the idea of Pama-Nyungan as a clade is "totally without foundation".
Most other linguists reject Dixon's conclusions as to the validity of Pama-Nyungan or the comparative method in Australia. In 2004 Geoffrey O'Grady and Kenneth L. Hale wrote that Dixon's position is "extravagant and spectacularly erroneous so bizarrely faulted, and such an insult to the eminently successful practitioners of Comparative Method linguistics in Australia, that it positively demands a decisive riposte." Nicholas Evans argues that there is no evidence for Dixon's assertion that around 50% of the vocabulary of neighbouring Australian languages reflected patterns of diffusion and that morphological and phonological "evidence for Pama-Nyungan has, in fact, been slowly adding up over the last four decades".
In their 2024 reconstruction of Proto-Australian, Harvey and Mailhammer argue that "the available materials do not support" proposals for unusually widespread diffusion and "there is no evidence that any of lexical, morphological, or phonological diffusion have been of unusual frequency in Australia."
The status of the non-Pama-Nyungan languages within Australian languages
In 2003, Evans set out the possibilities for higher-order subgrouping among the Australian languages:- The "rake" model; Pama-Nyungan is one clade of Australian languages, but the over twenty separate families of non-Pama-Nyungan cannot be classified into a single larger genetic group.
- The "diffusion" model, whereby Proto-Australian was a suffixing language typologically close to Proto-Pama-Nyungan, and the prefixing languages have all acquired a head-marking, prefixing typology through their geographic proximity to each other. Evans rejects this analysis.
- The "binary" model, where essentially Pama-Nyungan and the non-Pama-Nyungan languages are two separate, equal nodes from the same tree. A problem with this model is that it does not account for the relative lack of diversity in Pama-Nyungan compared to the diversity found in non-Pama-Nyungan.
- The "Pama-Nyungan offshoot model", which Evans attributes to Geoffrey O'Grady. This is that idea that "Pama-Nyungan was a relatively recent daughter node within a larger Stammbaum containing most extant Australian languages. Of the attested Australian languages he excluded only Anindilyakwa and the Tasmanian languages from membership in a lineage descended from 'Original Australian. On this understanding, Proto-Australian is more likely to share typological characteristics with the non-Pama-Nyugan languages, while characteristic Pama-Nyungan features are more likely to be innovations. Most of the early work on Australian languages focused on Pama-Nyungan languages, but as more high-quality material has become available on non-Pama-Nyungan languages, any similarities that non-Pama-Nyungan languages share assume greater importance in reconstructing Proto-Australian.
A significant development came in 2017. Harvey and Robert Mailhammer published a list of noun class prefixes based on non-Pama-Nyungan languages that could be traced back to Proto-Australian, a dependent-marking, prefixing language. In 2024, Harvey and Mailhammer published a book-length reconstruction of Proto-Australian which included some modifications, revising their earlier analysis to claim that PA noun class markers were optional proclitics rather than compulsory prefixes.
| gloss | PA | PPN | Proto-Nyulnyulan | Proto-Tangkic | Arnhem |
| 'hand' | *maɻ | *maɻa | *maɭa | *maɭ-ta | Ngandi ku-maɭ |
| 'left hand' | *t̪aku | *caku | *t̪aku | Bininj Gunwok kun-cakku | |
| 'I, me' | *ŋaj | *ŋaj | *ŋaju | Burarra ŋaj-ppa | |
| 'spear' | *ta-n | *ɻa | *ɻa | *ɭaa | Burarra ra |
| 'scold' | *t̪u-n | *t̪u-n | *ci | *t̪uu | Bininj Gunwok tu |