The First Person of Sahul

TL;DR

• Surveying 300 languages, Schapper, San Roque and Hendery (2016) found that Australia and New Guinea name tree, firewood and fire in ways the rest of the world rarely does, and called Sahul “a large diffusion area”.
• Australian languages say “I” with ŋa-: about four in five of them, against one in thirteen languages worldwide (ASJP). New Guinea’s highland family says na.
• On the coast facing Saibai, at the old land bridge, the Pahoturi River languages say ŋəna, from Proto-Pahoturi *ŋa-na (Ross 2005, p. 39), and their ancestor sounded Australian (Evans 2019). Ŋa- is the best candidate for a word of Proto-Sahul.

I am not aware that anyone has tried to explore the latter hypothesis further. — Andrew Pawley, on a distant link between New Guinea and Australian languages, Papuan Pasts (2005), p. 6

One word for tree, firewood and fire #

In Duna, a language of the Papua New Guinea highlands, rowa means “tree”. It also means “firewood” and “fire”, and it stretches to cover matches and cigarette lighters. Across the Torres Strait on Cape York, thuma in Wik-Ngathan and yugu in Guugu Yimithirr carry similar meanings (Burenhult et al. 2017, p. 456).

English keeps these apart, and so do most languages. The CLICS⁴ database collects documented cases of one word doing two jobs across some 3,400 language varieties. It lists nineteen languages that call fire and tree by a single name. Seventeen are in Sahul: fourteen in New Guinea, from Kalam and Huli in the highlands to Wambon in the southern lowlands, and three in Australia, Kuuku-Ya’u on Cape York (yuku), Yorta Yorta on the Murray (bitja) and Kurrama in the Pilbara (karla). The other two are in Amazonia.

The pattern is not a shared word. Rowa, yugu, bitja and karla have nothing in common in their sounds. What they share is a way of carving up the world, in which the living tree, the dead wood and the flame are one thing seen at three moments. A rare way of dividing experience that covers a whole region is the mark of people who were once in contact, or who came from the same source.

What did Schapper, San Roque and Hendery find? #

Antoinette Schapper wrote the grammar of Bunaq, a Papuan language of Timor, and works across the Papuan and Austronesian languages of Wallacea and New Guinea. Lila San Roque wrote the grammar of Duna, the language of rowa. Rachel Hendery, a digital-humanities linguist at Western Sydney University, mapped the data. Together they treated Australia and New Guinea as one sample.

Their chapter, “Tree, firewood and fire in the languages of Sahul”, surveys 300 languages, 217 Papuan and 83 Australian (Schapper, San Roque and Hendery 2016). Earlier writers had noticed the link between tree and fire in each region separately. The three authors were the first to treat it as one phenomenon across both. They coded every language for which of the three meanings share a word and plotted the result on a single map.

The commonest pattern turned out to be firewood and fire under one word, with the tree kept apart, correcting the older impression that tree and fire were the usual pair. CLICS shows it on both sides of the Strait: Gurindji and Walmajarri warlu, Nyangumarta wika, Kija and Bunuba in the Australian north and west, and Asmat yisa on New Guinea’s south-west coast (CLICS⁴). The patterns present in Sahul, the authors conclude, are rare worldwide. Sahul is “a large diffusion area”, and they ask Papuanists and Australianists to study it together.

That is a claim about thought: the two halves of the old continent share a habit of mind. The question it raises is whether anything harder than a habit survives, a form with its sounds intact. The best place to look is the one word every language must have.

Who has seen Sahul as one country of speech? #

The idea is old, and it was dropped rather than refuted.

Andrew Pawley, the leading historian of the Trans-New Guinea family, wrote the sentence at the head of this essay about Foley’s evidence: nobody had followed it up (Pawley 2005, p. 6). The fields had split. Australianists and Papuanists trained apart and published apart, and a comparison that the diffusionists of the early twentieth century had taken for granted became nobody’s job. The same thing happened to the ritual evidence, as I describe in One Sahul, and to archaeology, as I describe in Australian Archaeologists Can’t Handle the Truth.

Only Trombetti set the two halves’ first persons side by side, and his Papuan witness was well chosen. Dabu, the source of his nana “I”, is an old name for Agob, a language of the Pahoturi River (Glottolog). The one scholar who compared the first persons of Sahul picked, without knowing it, the shore where the evidence is clearest.

How do Australians say “I”? #

Across Australia the first person begins with the velar nasal, the sound at the end of English sing, written ng in Aboriginal orthographies. Warlpiri says ngaju, Gamilaraay ngaya, Pitjantjatjara ngayulu. The northern families outside Pama-Nyungan put the same syllable on the verb as a prefix: Kune nga-, Wardaman nga-, Maung nga- (Evans 2003, p. 19). Mark Harvey and Robert Mailhammer, who have reconstructed Proto-Australian, the ancestor of nearly every language on the continent, give its “I” as *ŋaj (Harvey and Mailhammer 2024).

This is unusual. The ASJP database holds basic word lists for over five thousand languages. In it, about four Australian languages in five begin “I” with ŋ. Worldwide the figure is about one in thirteen, and much of that comes from one family, Sino-Tibetan, whose “I” is also ŋa. In Africa and the Americas an ŋ- “I” is close to absent, and across New Guinea and the islands around it fewer than one language in twenty has one.

New Guinea’s largest family, Trans-New Guinea, spread through the highlands between 6,000 and 10,000 years ago with the “I” *na, made with an alveolar n (Ross 2005, p. 29; Pawley 2005, p. 101).¹ Hundreds of its descendants keep it: Melpa na, Enga namba, Ipili ni. The two halves of Sahul draw a clean line, ŋa to the south and na to the north.

Except at the old bridge.

Download the data (CSV)

Where does ŋa- cross the Torres Strait? #

Saibai Island lies a few kilometres off the coast of Papua New Guinea, in Australian waters. Its people speak Kalaw Lagaw Ya, a Pama-Nyungan language, and their “I” is ngay (Mitchell 2015, pp. 344, 357).

On the mainland opposite, along the Pahoturi River, live speakers of a small Papuan family: Ende, Em, Agob, Kawam, Taeme and Idi. Their “I” is ŋəna or ŋən (Lindsey 2019, p. 234). Malcolm Ross, surveying the pronouns of every Papuan family, reconstructed the family’s ancestral “I” as *ŋa-na (Ross 2005, p. 39). Ende also uses one word, yu, for both fire and firewood (CLICS⁴), the commonest Sahul pattern. Further west, on the south coast of Indonesian Papua, the two Bulaka River languages were recorded with “I” as Yelmek ŋöl and Maklew ŋöllo (Ross 2005, p. 39).²

This is the stretch of New Guinea that faced Australia across dry land, and Nicholas Evans, reconstructing its sound systems, found it was once Australian. Proto-Pahoturi and Proto-Yam, he writes, “look much more Australian-like than their contemporary descendants”, with the velar nasals, retroflex consonants and rich sets of liquids typical of Australia. Over the millennia since, they converged with their Papuan neighbours, once the waters of the Torres Strait cut them off from the Australian sound world (Evans 2019, p. 4). Evans leaves open whether this reflects shared descent or ancient contact. Either way, this shore belonged to the Australian side of Sahul, and its “I” is the Australian one.

Isn’t a nasal “I” just chance? #

The velar ŋ is a rare way to say “I”. Outside Australia and Sino-Tibetan it is uncommon, it is absent from the Trans-New Guinea reconstruction, and it is scarce in New Guinea. In Sahul it turns up where you would look for it: across Australia, and in New Guinea at the one place that touched Australia, in a family whose ancestor Evans independently found to be Australian in its sounds.

The plain n is another matter. Harald Hammarström has shown that about a quarter of the world’s languages begin “I” with n, and about half of Papuan languages do (Hammarström 2012, p. 442). That is why the comparison here is between ŋa and ŋa, not between any two nasals.

The Torres Strait was never one community. Its eastern islands speak Meriam Mir, a Papuan language whose “I” is ka (Piper 1989, p. 68). The shared “I” is not everywhere along the crossing. It is on the western shore, opposite Saibai.

When was Sahul last one country? #

The modern “I” of Saibai arrived with Pama-Nyungan, after the sea. The Pahoturi “I” sits in a family that Evans places in the Australian sound world before the sea, and that drifted from it only once the Strait opened. A pronoun can cross water. A whole sound system does not, and the Pahoturi shore had both. The most economical reading is that ŋa- was there while the bridge stood.

If so, it is older than Proto-Australian as currently dated, and Australia’s *ŋaj and Pahoturi’s *ŋa-na drew on the same older source. That source is the last language community of Sahul. Australia and New Guinea have been apart for 8,000 years. On the shore facing Saibai, people still say “I” the way that community did.

What does the first person tell us about the self? #

What follows is my own reading. The argument above stands without it.

I have argued for years that pronouns track the self. In The Unreasonable Effectiveness of Pronouns I proposed that we have had a word for “I” for as long as we have known there was an I to name, so the spread of the first person follows the spread of self-awareness. On that view the young dates in the table are what the theory expects: two great sweeps of the first person in the early Holocene, na in the New Guinea highlands and ŋa across Australia, meeting at the edge of the drowning plain.

Sahul also shows how an “I” can travel. In Central Australia the first person was taught at initiation. Southern Loritja novices gave up the everyday ngaiulu for ngankulu, the respectful “I” of the Western Desert, as I describe in I Man Now. A pronoun is supposed to be the least borrowable word in a language. In Sahul it was handed over in the rites that made boys into persons, and those rites crossed the water too: the bullroarer cults of the Sepik and of Arnhem Land share their origin story and closely matching sacred names, as I argue in Nggwal: The Cult That Crossed Sahul, and the bullroarer itself came south out of New Guinea, as I argue in Always From the North. The people came too: a man buried at Cairns before Europeans settled there was one-eighth Papuan, and the geneticists who could date that ancestry have not (The Test They Wouldn’t Run). Whether ŋa- crossed the land or the water, it travelled with the people who taught one another what it means to say “I”.

Sources #

1. Bouckaert, Remco R., Claire Bowern, and Quentin D. Atkinson. “The origin and expansion of Pama–Nyungan languages across Australia.” Nature Ecology & Evolution 2 (2018): 741–749.
2. Burenhult, Niclas, Clair Hill, Juliette Huber, Saskia van Putten, Konrad Rybka, and Lila San Roque. “Forests: the cross-linguistic perspective.” Geographica Helvetica 72 (2017): 455–464.
3. Carroll, Matthew J. The Ngkolmpu Language, with Special Reference to Distributed Exponence. PhD thesis, Australian National University, 2017.
4. Evans, Nicholas. “Introduction: Comparative non-Pama-Nyungan and Australian historical linguistics.” In Evans (ed.), The Non-Pama-Nyungan Languages of Northern Australia, 3–25. Pacific Linguistics 552, 2003.
5. Evans, Nicholas. “Australia and New Guinea: Sundered hemi-continents of sound.” Proceedings of the 19th International Congress of Phonetic Sciences, Melbourne, 2019.
6. Hammarström, Harald. “Pronouns and the (preliminary) classification of Papuan languages.” Language & Linguistics in Melanesia, special issue 2012, part II: 428–539.
7. Harvey, Mark, and Robert Mailhammer. Proto-Australian: Reconstruction of a Common Ancestor Language. Berlin: De Gruyter Mouton, 2024.
8. Harvey, Mark, and Robert Mailhammer. “New evidence confirms our Indigenous languages have a common source, but how they spread remains a mystery.” The Conversation, 29 January 2025.
9. Lindsey, Kate L. Ghost Elements in Ende Phonology. PhD dissertation, Stanford University, 2019. Appendix A.
10. Mitchell, Rod. “Ngalmun Lagaw Yangukudu: The language of our homeland.” Memoirs of the Queensland Museum – Culture 8(1) (2015): 323–446.
11. Morris, John. “Trombetti, Gatti and the birth of the Indo-Pacific concept.” Mother Tongue 11 (2006).
12. Pawley, Andrew. “Introduction to the chapters on historical linguistics” and “The chequered career of the Trans New Guinea hypothesis.” In Pawley, Attenborough, Golson and Hide (eds.), Papuan Pasts, 1–14 and 67–107. Pacific Linguistics 572, 2005.
13. Piper, Nick. A Sketch Grammar of Meryam Mir. MA thesis, Australian National University, 1989.
14. Reesink, Ger, Ruth Singer, and Michael Dunn. “Explaining the linguistic diversity of Sahul using population models.” PLoS Biology 7(11) (2009): e1000241.
15. Reeves, Jessica M., et al. “The sedimentary record of palaeoenvironments and sea-level change in the Gulf of Carpentaria, Australia, through the last glacial cycle.” Quaternary International 183 (2008): 3–22.
16. Ross, Malcolm. “Pronouns as a preliminary diagnostic for grouping Papuan languages.” In Papuan Pasts, 15–65. Pacific Linguistics 572, 2005.
17. Schapper, Antoinette, Lila San Roque, and Rachel Hendery. “Tree, firewood and fire in the languages of Sahul.” In Päivi Juvonen and Maria Koptjevskaja-Tamm (eds.), The Lexical Typology of Semantic Shifts, 355–422. Berlin: De Gruyter Mouton, 2016.
18. Tjuka, Annika, Robert Forkel, Christoph Rzymski, and Johann-Mattis List. CLICS⁴: Database of Cross-Linguistic Colexifications. 2026. Edges FIRE–TREE and FIRE–FIREWOOD.
19. Wichmann, Søren, Eric W. Holman, and Cecil H. Brown (eds.). The ASJP Database, version 21. Concept “I”.
20. Greenhill, Simon J. “TransNewGuinea.org: An online database of New Guinea languages.” PLoS ONE 10(10) (2015): e0141563. Map data via the Lexibank edition.
21. Bowern, Claire. Pama-Nyungan wordlists, Lexibank edition. Map data.
22. Hammarström, Harald, Robert Forkel, Martin Haspelmath, and Sebastian Bank. Glottolog 5.4. Language coordinates and the Agob (Dabu) entry.
23. NOAA National Centers for Environmental Information. ETOPO1 Global Relief Model, via ERDDAP. Map land and shelf.
24. Wurm, Stephen A. “Possible wider connections of Papuan languages: Papuan and Australian.” In New Guinea Area Languages and Language Study, vol. 1, 925–934. Pacific Linguistics C-38, 1975.