The Dead Songs We Still Sing

TL;DR

• Ritual speech often survives as pure charged sound after everyday meaning dies.
• Aboriginal Australian initiation gives the cleanest bullroarer-adjacent cases: sacred songs no one could translate, sung inside male ceremonial worlds that also guarded the roaring instrument.
• The Mandan O-Kee-Pa gives the hook-suspension trail: Catlin heard a fixed prayer during suspension and could not get a translation.
• Rome, Nepal, the Trobriands, South Sulawesi, North Africa, Zoroastrianism, Buddhist dharani, and Greek magical papyri all show the same pattern.
• The scandal is simple: ritual does not need semantic understanding to survive. Sometimes not understanding is the point.

“no one can translate it”

• K. Langloh Parker, The Euahlayi Tribe (1905)

What Is A Dead Song?#

A dead song is not necessarily a song in a dead language.¹ It is worse and more interesting: a song whose social life continues after its semantic key is broken. The mouth still knows it. The body still knows when to sing it. The elders still know that it is dangerous, sacred, lucky, ancestral, or required. But the ordinary meaning has gone black.

That is the tabloid version of the story: the corpse is still singing.

Anthropology has a quieter vocabulary for the same thing: archaic ritual language, esoteric registers, nonlexical vocables, foreign-derived liturgies, dream songs, secret speech, magical names, fossilized formulae. But those tidy labels hide the hot wire at the center. A community can preserve exact sound for generations after it stops preserving exact sense.

This should not surprise anyone who has watched children chant playground rhymes, football crowds roar inherited nonsense syllables, or churchgoers recite creeds in a language they cannot parse. But the cases below are not merely “I do not know Latin.” They are stronger. They include songs said to be unintelligible even to the singers; priestly hymns scarcely understood by the priests; shamanic languages that ordinary speakers cannot follow; magical syllables whose power is said to vanish when translated; and initiation songs performed beside the oldest noisemaker in the human ritual toolkit, the bullroarer.

For the broader snakecult.net argument about ritual technologies as deep cultural fossils, this is a close cousin of the bullroarer as a marker of male initiation, the Pleiades-bullroarer memeplex, and the way language breaks Eden. The present article is the linguistic side of the same beast.

A ritual can outlive its own translation. When that happens, sound becomes the artifact: a portable fossil of authority, terror, ancestry, and divine contact.

The Bullroarer Cases Are Not Subtle#

Start with the line that provoked this whole hunt. K. Langloh Parker says she had been allowed to hear “Byamee’s Song,” a sacred Euahlayi song that only fully initiated men could sing. Parker adds that the old language of the song was already lost: the song was known, the right to sing it was guarded, but its full meaning was not understood Parker, The Euahlayi Tribe.

That is the first specimen under glass: a sacred song survives inside an initiation system after translation fails. The point is not that every singer is stupid or careless. The point is that the rite cares more about inheritance, restriction, and sound than about ordinary paraphrase. The initiated mouth is authorized to reproduce the pattern; semantic explanation is secondary.

A still sharper Australian case comes from R. H. Mathews’ account of the Birdhawal Dyerrayal initiation. Women and elders circle the ceremonial enclosure chanting a song whose meaning, Mathews says, was unintelligible even to the people singing it. Then, the next morning, the sound of Turndun is heard: Turndun is the bullroarer Mathews, “The Birdhawal Language,” ANU Press PDF. Later in the rite, bullroarers are shown, swung, and worked into the secret machinery of male initiation.

This is exactly the kind of case that should make people stop being too tidy. The dead song and the bullroarer are not two isolated museum labels. They sit in the same ceremonial voltage field: initiation, secrecy, gendered restriction, terrifying sound, and old words. If the exact song is not the bullroarer itself, it is still moving in the bullroarer’s weather.

Australian material keeps yielding variants. In western Arnhem Land, Brown and Evans describe a Marrku songset that keeps ancestral languages alive even when no fluent speakers remain available for ordinary conversation Brown and Evans, “Songs that keep ancestral languages alive”. Warlpiri initiation songs in contemporary Kurdiji practice are known in full by fewer specialists, while many participants hum or sing without command of the symbolic associations Curran, Contemporary Ritual Practice in an Aboriginal Settlement. NSW travelling songs such as Wanji-Wanji are cited in song-revival work as cases where old texts may still be sung without full understanding Hodgetts, Guthi Girrmara.

The bullroarer is not present in every one of those later cases. But Australia makes the deeper point unusually visible: old sound, sacred restriction, and initiation technology tend to cluster. The instrument that makes a voice without a throat is the perfect neighbor for a song with a voice but no translation.

The Plains Hook-Swinging Trail Is Real, But Messier Than The Meme#

The Plains trail matters because it has the image everyone remembers: men suspended by flesh, turned until they faint, praying through pain. George Catlin’s account of the Mandan O-Kee-Pa is the classic source. He describes candidates hung by splints, turned in the air, and pushed through an ordeal that he understood as a great religious ceremony Catlin, O-Kee-Pa: A Religious Ceremony. At the climax, one suspended man utters a fixed prayer or cry to the Great Spirit. Catlin says he could get no translation.

That is the hook-swinging case. It is real enough. The language problem is inside the same rite as the suspension.

But the bullroarer trail should be handled with a little violence and a little care. Catlin’s O-Kee-Pa account is full of drums, rattles, masks, animal impersonations, medicine songs, and sacred theatrical terror, but I do not find a bullroarer there. The stronger Plains-adjacent link is different. The Arapaho Language Project notes the bullroarer as an air instrument connected to wind and the Ghost Dance, sometimes used to begin singing, while eagle-bone whistles belong to the Sun Dance Arapaho Language Project, “Air Instruments”. That is not the same as saying “the Mandan hook-swinging rite used a bullroarer.” It says something cooler and more complicated: Plains and near-Plains sacred sound worlds include hook-suspension, fixed untranslated prayers, Sun Dance whistles, Ghost Dance songs, and bullroarer-linked singing, but those elements do not collapse into one neat ritual.

Frances Densmore’s Teton Sioux material adds another layer. She records Sun Dance singers and informants for whom sung prayer reaches Wakan Tanka with special force, and she discusses sacred-language opacity in the larger musical corpus Densmore, Teton Sioux Music. George A. Dorsey, James Mooney, and later Ghost Dance materials show how dream songs and intertribal ritual phrases could move rapidly across language boundaries Mooney, The Ghost-Dance Religion. Elsie Clews Parsons’ Caddo notes record Ghost Dance songs whose words were intelligible to some and unintelligible to others, often said to have been dreamed or heard from nonhuman sources Parsons, “Notes on the Caddo”.

So the answer to the original Plains hunch is this:

1. Hook-suspension plus untranslated prayer: yes, Mandan O-Kee-Pa.
2. Sun Dance piercing plus sacred song: yes, in the wider Plains ceremonial world.
3. Bullroarer plus Plains Sun Dance: not supported by the clean sources I checked.
4. Bullroarer plus Ghost Dance / singing: yes, Arapaho material puts it there.
5. Bullroarer plus secret initiation more generally in North America: yes, especially California and Northwest Coast evidence.

That is not a deflation. It is a better monster.

The Americas Have A Whole Underworld Of Half-Understood Songs#

Once you stop demanding that every case be the same rite, the Americas open up. The old sound keeps returning in different costumes.

In California, Frederick Starr says the Cahuilla eagle-dance songs were so old that the words were not understood even by the singers Starr, American Indians. That is almost too perfect: the song is not merely foreign to outsiders; the singers themselves are downstream of a vanished key.

The California ghost and Kuksu complexes are where the bullroarer gets back on stage. S. A. Barrett’s Pomo material describes ghost ceremonies restricted to initiated men, with dancers impersonating the dead Barrett, Ceremonies of the Pomo Indians. Edwin Loeb’s comparative work on tribal initiations and secret societies ties the California material to bullroarer use in secret-society settings Loeb, “Tribal Initiations and Secret Societies”. Here the relevant package is not hook-suspension. It is ghost impersonation, secrecy, initiation, and the roaring hidden object.

On the Northwest Coast, the winter ceremonial complexes intensify the same pattern. Loeb summarizes Kwakwaka’wakw evidence from Boas in which the bullroarer is whirled secretly as part of the spirit-abduction drama around novices Loeb 1929; Boas’ foundational report remains the deeper archive Boas, The Social Organization and the Secret Societies of the Kwakiutl Indians. Frances Densmore’s Nootka and Quileute Music then gives the neighboring dream-song atmosphere: songs learned in dreams, songs attributed to animals, and Klokali ceremonial material where bullroarers also appear in the broader record Densmore, Smithsonian record.

Again: not one tidy box. It is an underworld. Old song, spirit impersonation, dream acquisition, secret-society discipline, and artificial spirit-voices keep crossing each other.

Rome Had Priests Singing A Corpse Of A Hymn#

The phenomenon is not only tribal, oral, or “primitive,” in the lazy old sense. Rome had it in official costume.

The Salii, the leaping priests of Mars, sang the Carmen Saliare during archaic Roman rites. Ancient and modern summaries repeatedly stress how old and obscure the hymn had become. Smith’s classical dictionary, preserved at LacusCurtius, says the Salian songs were scarcely understood even by the priests themselves LacusCurtius, “Salii”. The Bibliotheca Augustana presentation of the Carmen Saliare likewise frames it as archaic ritual poetry already difficult in antiquity Carmen Saliare.

Rome also gives the Carmen Arvale, the hymn of the Arval Brothers. Its preserved text is famous because it can be printed, inspected, and still not easily domesticated: archaic forms, debated readings, uncertain interpretation Carmen Arvale text. The Arval hymn is not as perfect a case as the Salii, because the modern scholar’s problem is not automatically the ancient singer’s problem. But together they show a literate state’s version of the dead song: official religion preserving fossil words because oldness itself is evidence.

That should matter for how we read the Australian and American cases. The survival of opaque ritual language is not a failure of literacy. It is a general property of ritual conservatism. If the Roman state can keep a half-dead hymn on payroll, a secret society can do it in the bush.

Shamanic Languages Are Built To Go Dark#

Some dead songs are accidents of language change. Others are engineered to be difficult. Shamanic speech often sits in the middle. It is old, metaphorical, specialized, compressed, and guarded. Ordinary people may know the ritual exists and still not understand its words.

The Chintang Rai material from eastern Nepal is one of the cleanest cases. Rai describes risiwa as a ritual language of Mundum chanting: ordinary speakers cannot chant or understand it, and even experienced ritual specialists may not understand it completely in ordinary semantic terms Rai, “Mundum: A Case Study of Chintang Ritual Language”. That is not just a foreign language. It is ritual opacity as a social technology.

The Himalayan pattern repeats. Idu Mishmi Igu shamans in Arunachal Pradesh use long ritual chants in a special register that ordinary Idu speakers understand only partly; Delley and other recent authors emphasize the gap between everyday language and shamanic performance Delley, “The Igu Shamanic Tradition of the Idu Mishmi”. Tamang bombo shamans chant in archaic and obscure forms rather than ordinary colloquial Tamang Holmberg, Order in Paradox. Gurung Ghatu dance-song traditions are often described as sung in a language neither ordinary Nepali nor Gurung audiences fully understand Kailash/Wisdom Library, Ghatu study; contemporary reporting preserves the same idea of a trance-song tradition whose words are inherited more than parsed Nepali Times, “Ghatu dance of Lamjung”.

South Sulawesi adds a priestly case with a name that sounds like it came from a fantasy novel but did not: basa to ri langiq, the “language of the sky” associated with Bugis bissu ritual specialists. Recent work describes bissu as mediators able to use this divine or archaic language in ritual performance Sage Open article on bissu; other accounts stress that the language is restricted to bissu and divine communication Atlantis Press PDF. This is not necessarily “no bissu understands it.” It is a different species: the community does not understand it because it is a priestly speech-channel.

North Africa gives a foreign-derived ritual-language case. In Algerian Diwan or Gnawa-adjacent performance, Hausa-linked suites can survive in trance/healing contexts after the community no longer speaks Hausa as an ordinary language Turner, Diwan/Gnawa PDF. Makran’s Siddi communities likewise preserve African-style songs in a language of possible Swahili origin that is no longer understood locally During, “The Spirit Cults of the Makran Coast”. Here the dead song is a scar of forced movement, trade, slavery, and diaspora. The words are not meaningless; they are displaced ancestry.

The Trobriands Turn Magic Into A Tape Recorder#

The Trobriand Islands are especially important because Gunter Senft has spent decades documenting how special speech varieties work there. In Kilivila, he distinguishes ordinary language from magical speech and from biga baloma, the “language of the spirits of the dead.” In harvest and mourning songs, old linguistic forms can be preserved even when singers no longer understand the semantic content Senft, “Magic, missionaries and religion”.

This is almost the ideal laboratory for dead songs. The song is not merely a degraded text. It belongs to a theory of efficacy. Magic works because it is repeated correctly, inherited correctly, and anchored to ancestral authority.

That is why the tape-recorder metaphor is useful but incomplete. A tape recorder preserves sound without understanding it. A ritual singer preserves sound while being understood by the community as authorized, obligated, transformed, or dangerous. The human tape recorder is also a priest.

This is why the dead-song pattern matters for the diffusion and ritual-mask problem. Ritual packages can transmit forms whose original explanations mutate or vanish. The sound survives because it has acquired a function independent of paraphrase.

Magic Words Do Not Need Translation#

At the far end of the spectrum are ritual words that are not merely hard to translate. They are powerful because they resist translation.

Greek and Egyptian magical papyri swarm with voces magicae: foreign-looking names, strings of vowels, divine epithets, and syllables that do not behave like ordinary sentences. Monika Amsler’s study of voces magicae treats them as world-making ritual language rather than random filler Amsler, “Voces Magicae and Imperial / Late-Antique World-Making”. The Ephesia Grammata, a famous protective formula, became a named set of nonsense or magic words with protective force Bernabe, Oxford Academic chapter.

Iamblichus gives the philosophical version of the same instinct. In On the Mysteries, he argues that sacred foreign names should not be translated, because translation destroys the divine affinity and ritual power of the sounds Iamblichus, On the Mysteries, section 7. There is the doctrine in plain view: sound is not a vehicle for meaning; sound is a substance.

Buddhist dharani traditions keep the same nerve alive. Dharani are often chanted as Sanskritic syllabic formulas across communities that do not speak Sanskrit, and explanatory traditions often concede that full translation is impossible or beside the point 84000, Ratnaketu Dharani; Dharma Drum Mountain, dharani explanation. Byzantine chant has its own nonlexical tradition in kratemata and teretismata, extended singing on vocables rather than ordinary words Varelas, “Nonsense Syllables in Byzantine Chant Tradition”.

Vedic mantras deserve extra care. It is too crude to say “Vedic mantras are meaningless.” The better claim is that Indian tradition itself debated meaning, obscurity, and interpretation. Yaska’s Nirukta exists because Vedic words had become difficult enough to require systematic explanation, and modern discussions of Kautsa preserve the old controversy over whether some mantras have meaning at all Viswanathan, PhilPapers abstract. Frits Staal pushed the argument further, claiming that ritual can be governed by syntax and performance rather than semantics Staal, “The Meaninglessness of Ritual”. That thesis is deliberately provocative, but it belongs in this dossier because it states the terrifying possibility cleanly: ritual can run without meaning in the ordinary sense.

Zoroastrian Avestan gives the fossil-liturgical version. Avestan ceased to be ordinary speech but remained ritually necessary; reform debates later attacked prayer recitation when it became unintelligible to many performers and hearers Andres-Toledo, UNL Digital Commons; Dhalla, History of Zoroastrianism, chapter 59. Again, the point is not ignorance as failure. It is preservation as a religious act.

A Field Guide To Songs Whose Words Went Dark#

The following table is deliberately broad. It includes direct dead-song cases, specialist ritual languages, nonlexical sacred sound, and bullroarer-adjacent initiation complexes. The last column is the important one: it tells you how close the bullroarer is. This is not a purity test. It is a map of proximity.

Why The Words Die But The Song Survives#

The dead-song pattern has several engines. They can work separately, but the strongest cases stack them.

1. Exact sound becomes more important than explanation.
Magic formulas, mantras, and initiation songs are often treated as recipes. Change the sound and the ritual fails. That makes phonetic conservatism stronger than semantic updating.

2. Specialists monopolize interpretation.
If only initiated men, shamans, priests, or bissu may perform the language, ordinary comprehension is not the goal. Opacity becomes part of the office. The right person may know how to use the words even when no one can translate them into household speech.

3. Language shift breaks the bridge.
A community changes language; the song does not. This is the Trobriand, Makran, Diwan, Avestan, and many Aboriginal archival cases. The song becomes a phonetic fossil.

4. Dream origin excuses opacity.
If a song comes from a bird, a dead relative, a spirit, a mountain, or a dream, it does not need to be normal speech. North American Ghost Dance songs and Northwest Coast dream songs make this especially visible.

5. Non-understanding can increase power.
A transparent instruction belongs to the market. An opaque formula belongs to the temple. This is why Iamblichus can defend untranslatable divine names and why a bullroarer can be more frightening before you know it is only wood on a cord.

The last point is the ugly one. Ritual does not merely survive ignorance. It can feed on controlled ignorance. A hidden bullroarer works because women and novices hear a spirit. A dead song works because the singer becomes a carrier of something older than explanation.

That is why the dead song belongs beside the bullroarer cosmogenesis problem. Both are ritual technologies of invisible agency. One makes a voice without a visible body. The other makes a message without a visible meaning.

The Final Scandal: Meaning Is Not The Payload#

Modern people are trained to think language is a meaning-delivery system. Words are packages; translation opens the package; communication succeeds when the package arrives.

Ritual language laughs at that model. In these cases, the payload is not only semantic content. It is age, danger, authority, memory, group boundary, divine address, and bodily performance. The community does not merely ask, “What does this sentence mean?” It asks, “Who is allowed to say it?” “Who taught it?” “What happens if it is sung wrong?” “Who must not hear it?” “What invisible being hears it first?”

That is why dead songs keep singing. They are not failed communications. They are successful survivals.

And maybe this is the most important lesson for EToC-style thinking. Before language becomes the private voice of self-consciousness, language is also a public machine for producing altered states, social hierarchies, ancestral continuity, and sacred terror. Some of its most durable forms are not the clearest ones. They are the forms most tightly bound to ritual action.

The words die. The mouth keeps the corpse moving. The bullroarer starts to turn. Everyone understands enough to be afraid.

Footnotes#

Sources#

1. Parker, K. Langloh. The Euahlayi Tribe. 1905.
2. Mathews, R. H. “The Birdhawal Language”. Reprinted by ANU Press.
3. Catlin, George. O-Kee-Pa: A Religious Ceremony; and Other Customs of the Mandans. 1867.
4. Arapaho Language Project. “Arapaho Air Instruments”.
5. Densmore, Frances. Teton Sioux Music. Bureau of American Ethnology Bulletin 61, 1918.
6. Mooney, James. The Ghost-Dance Religion and the Sioux Outbreak of 1890. Bureau of American Ethnology, 1896.
7. Parsons, Elsie Clews. “Notes on the Caddo”. 1921.
8. Starr, Frederick. American Indians. 1898.
9. Barrett, S. A. Ceremonies of the Pomo Indians. 1917.
10. Loeb, Edwin M. “Tribal Initiations and Secret Societies”. University of California Publications in American Archaeology and Ethnology 25(3), 1929.
11. Boas, Franz. The Social Organization and the Secret Societies of the Kwakiutl Indians. 1897.
12. Densmore, Frances. Nootka and Quileute Music. Bureau of American Ethnology Bulletin 124, 1939.
13. Laade, Wolfgang. Traditional Songs of the Western Torres Straits. Folkways Records notes, 1960s.
14. British Museum. Muralag bullroarer object record.
15. Haddon, A. C. and Sidney H. Ray. Reports of the Cambridge Anthropological Expedition to Torres Straits, Vol. III: Linguistics. 1907.
16. Brown, Reuben and Nicholas Evans. “Songs that keep ancestral languages alive: A Marrku songset from western Arnhem Land”. 2017.
17. Curran, Georgia. Contemporary Ritual Practice in an Aboriginal Settlement. ANU.
18. Hodgetts, Leah. Guthi Girrmara. University of Newcastle.
19. Senft, Gunter. “Magic, missionaries and religion: Some observations from the Trobriand Islands”. 2010.
20. Rai, Novel Kishore. “Mundum: A Case Study of Chintang Ritual Language”. 2009.
21. Delley, Razzeko. “The Igu Shamanic Tradition of the Idu Mishmi”. 2023.
22. Holmberg, David H. Order in Paradox: Myth and Ritual among Nepal’s Tamang. Cornell University Press, 1989.
23. Kailash / Wisdom Library. Study of Ghatu dance-song tradition.
24. Nepali Times. “Ghatu dance of Lamjung”.
25. Nurhayati et al. “Bissu and the sacred language of South Sulawesi”. Sage Open, 2023.
26. Atlantis Press. Article on Basa To ri Langi and Bissu ritual language.
27. Turner, Richard Brent. Study of Algerian Diwan/Gnawa trance music.
28. During, Jean. “The Spirit Cults of the Makran Coast”.
29. LacusCurtius / Smith’s Dictionary. “Salii”.
30. Bibliotheca Augustana. Carmen Saliare.
31. The Latin Library. Carmen Fratrum Arvalium.
32. Amsler, Monika. “Voces Magicae and Imperial / Late-Antique World-Making”. Annuaire de l’EPHE, 2021.
33. Bernabe, Alberto. Chapter on the Ephesia Grammata. Oxford Academic.
34. Iamblichus. On the Mysteries, Section 7.
35. 84000. The Ratnaketu Dharani.
36. Dharma Drum Mountain. Explanation of dharani.
37. Varelas, Christos. “Nonsense Syllables in Byzantine Chant Tradition”.
38. Viswanathan, S. “Frits Staal, Kautsa, and the Meaning of Vedic Mantras”. PhilPapers record.
39. Staal, Frits. “The Meaninglessness of Ritual”.
40. Andres-Toledo, Miguel Angel. Zoroastrian Avestan ritual language study. University of Nebraska-Lincoln Digital Commons.
41. Dhalla, M. N. History of Zoroastrianism, Chapter 59.