Onpode
Cover art for The mechanism of memory: how societies without writing maintained historical knowledge

The mechanism of memory: how societies without writing maintained historical knowledge

November 12, 2026 · 13 min

Ryan Castillo & Jordan Hale

Before we get into it — you look like you've been chewing on something since you walked in. I mean, I have been, yeah. I keep thinking about what it would feel like to be the only copy of something. Like, imagine you're one of the South Slavic guslari that Milman Parry recorded in the…

Pre-literate societies developed sophisticated systems for transmitting historical knowledge across generations without written records. Rather than relying on casual memory, these cultures embedded histories into structured linguistic and performative forms—poetry, rhythm, melody, genealogy, ritual, and repeated narrative patterns—designed to survive the demands of oral transmission.

0:0012:57
Get the next episode on History →

Follow it free — new episodes land in your feed.

Or make your own — any topic, in minutes

More Onpode episodes on History →

About this episode

Pre-literate societies developed sophisticated systems for transmitting historical knowledge across generations without written records. Rather than relying on casual memory, these cultures embedded histories into structured linguistic and performative forms—poetry, rhythm, melody, genealogy, ritual, and repeated narrative patterns—designed to survive the demands of oral transmission.

Grounded in 12 sources
Contradictions at the Heart of the Canon: Jan Vansina and the Debate over Oral Historiography in Africa, 1960–1985 | History in Africa | Cambridge Core ↗ · cambridge.org
Ideology and Oral Traditions: Listening to the Voices ‘From Below’1 | History in Africa | Cambridge Core ↗ · cambridge.org
Material mnemonic technologies (Chapter Four) ↗ · cambridge.org
Welsh Writing Before 1500 ↗ · doi.org
Kia kōrerorero tonu ai: a review of the dialogue at the interface of Indigenous oral tradition and archaeology in Aotearoa New Zealand and Oceania ↗ · doi.org
Knowledge Preservation through Oral Traditions in Ancient India ↗ · doi.org
Oral Tradition, History, and Archaeohistory of Indigenous Australia ↗ · doi.org
Genealogies and oral histories as chronological networks: interfacing whakapapa (Māori genealogies) with Gregorian calendar year archaeological radiocarbon dates ↗ · doi.org
Voices on the wind, traces in the earth: Integrating oral narrative and archaeology in Polynesian history ↗ · doi.org
Bartlett revisited: Direct comparison of repeated reproduction and serial reproduction techniques. ↗ · doi.org
Memory and Mystery: The Cultural Selection of Minimally Counterintuitive Narratives ↗ · doi.org
Memory in Oral Tradition ↗ · doi.org
Read transcript

Ryan Castillo: Before we get into it — you look like you've been chewing on something since you walked in.

Jordan Hale: I mean, I have been, yeah. I keep thinking about what it would feel like to be the only copy of something. Like, imagine you're one of the South Slavic guslari that Milman Parry recorded in the nineteen-thirties — illiterate, no notebook, no archive — and you're carrying the full length of Homer in your head. Except you're not really carrying it. That's the part that broke my brain.

Ryan Castillo: You're not carrying it — you're reconstructing it. Every performance.

Jordan Hale: Every single performance. And Parry's whole discovery — what he found in the field, what Albert Lord then formalized — is that those singers weren't reciting fixed lines. They were composing in real time using this vocabulary of metrical formulas. 'The wine-dark sea.' 'Gray-eyed Athena.' Phrases that just slot into the meter automatically. The constraint is the technology.

Ryan Castillo: Which reframes what fidelity even means in an oral system. Because the formula vocabulary is the memory — that's Lord's claim — not some decoration on top of the real thing.

Jordan Hale: And it's not just the guslari, you know — the Vedic tradition in ancient India was doing this deliberately, across millennia, with philosophy, law, ritual sequences. Before writing. Sometimes alongside it, but the oral transmission was the spine.

Ryan Castillo: Look, that's the case that actually sharpens the question for me — because when you're preserving ritual sequences and legal content orally, the stakes of distortion are real. Functional stakes. So what does it mean when that oral record diverges from documentary evidence?

Jordan Hale: That's it — that's the thing. We read the divergence as error. But what if oral-formulaic composition, the constrained reconstruction, is filtering for something specific — and the leak is actually the point? The clue isn't that the story changed. The clue is what stayed the same.

Ryan Castillo: So the divergence from written records isn't the problem to solve. It's the data.

Jordan Hale: But okay, that word 'data' — I want to make sure we're not glossing over *how* the constraint actually works, because I think that's the part that broke my intuition. Like, imagine a grandmother teaching her grandchild a family recipe entirely by having them make it twelve times with her. No written card. The method is the memory. You can't drift from it because the *doing* of it corrects you.

Ryan Castillo: And that's exactly what oral-formulaic composition is — not metaphorically. The formula is metrically pre-fitted. 'The wine-dark sea' isn't a pretty phrase; it fills a specific rhythmic slot. Deviation breaks the meter, and the meter is the error-correction.

Jordan Hale: Wait — so it's like... the constraint punishes drift automatically?

Ryan Castillo: Exactly. The guslari aren't memorizing lines — Lord's whole argument is that they're composing in performance from a traditional vocabulary of thematic units. The formula survives because it *has* to fit. What doesn't fit gets shed.

Jordan Hale: So the structural mold is the thing with fidelity, not the specific wording. Which is — I mean, that's completely backwards from how I'd assumed memory worked. We think rigidity preserves things. But Lord is saying it's the opposite — the flexibility *within* the constraint is what makes it survive.

Ryan Castillo: Think of it like a jazz musician who knows the chord changes cold. They can improvise anything — but they can't go outside the changes. The changes are fixed. That's the checksum.

Jordan Hale: And it's not one culture's trick, you know — the Norse and Welsh traditions were doing versions of this. Old Norse scaldic poetry, genealogical lore, charms — structured so tightly that deviation from the pattern is cognitively *costly*. Like, you'd feel it go wrong.

Ryan Castillo: The Welsh case is interesting precisely because fewer than twenty pre-twelfth-century manuscripts survived — meaning oral structure was doing the preservation work before writing showed up to externalize it.

Jordan Hale: Right — but here's where I think we have to push one layer further, because there's something we haven't said yet. Deviation from the formula is what gets *lost*. That's the mechanism. But that also means oral transmission isn't a neutral recording — it's selecting for what fits the structural mold.

Ryan Castillo: Which is Frederick Bartlett's finding coming in from a totally different direction. His serial reproduction studies — people retelling a story down a chain — showed details don't just disappear randomly. They rationalize. They cohere around existing cultural schemas.

Jordan Hale: No way — so Bartlett is essentially describing the same mechanism Parry and Lord found in the field, just in a lab?

Ryan Castillo: Same engine, different context. Which complicates the clean version of this — because if the formula is the error-correction, it's also the *filter*. The constraint keeps the structure alive, but the structure decides what counts as content worth keeping.

Jordan Hale: So oral tradition isn't imperfect writing — it's a different technology with a different lossy profile. And the losses aren't random noise. They're... almost diagnostic. What dropped out tells you what the community decided didn't survive the filter.

Ryan Castillo: And that diagnostic angle is actually where Bartlett gets uncomfortable — because his serial reproduction studies don't just show filtering. They show *rationalization*. The chain of people retelling the story doesn't drift toward noise. It drifts toward the existing schema. The story gets more internally coherent, not less.

Jordan Hale: Wait — it gets *tidier*?

Ryan Castillo: Tidier, yes. Which is the same mechanism as formulaic constraint — just institutionalized. Parry and Lord found it in the field. Bartlett found it in a lab. Same engine.

Jordan Hale: That means — wait, no, I want to get this right. The Vedic schools weren't just lucky to have good memorizers. They were actually... building in resistance to that rationalization pull. Like, deliberately. Because the eleven distinct mnemonic techniques — the path mnemonic, the variegated mnemonic, different methods across different schools — that's not elegance, that's redundancy architecture. If one chain drifts toward its schema, another chain catches it.

Ryan Castillo: That's the distributed archive point. No single mind holds the tradition — it's spread across performers, occasions, even physical places. The community *is* the storage medium.

Jordan Hale: Which is a totally different vulnerability profile than a written record. A manuscript burns. A distributed oral tradition — you'd have to silence an entire community simultaneously.

Ryan Castillo: Right — but the number that matters here is: how do you detect corruption in a distributed system? Written text, you compare manuscripts. Oral tradition, the error-correction mechanism is... what exactly?

Jordan Hale: Genealogy. I mean — that's the concrete case, isn't it? A lineage list isn't sentimental. It's a chronological record, a territorial claim, a social map — all structured as a chain that's easy to encode and hard to corrupt without someone noticing the break.

Ryan Castillo: Because every person in the chain is a checksum. Skip a generation and the relational logic collapses downstream.

Jordan Hale: And the audience knows the genealogy too — that's the distributed part. It's not just the performer holding it. The community holds it collectively, so deviation isn't just cognitively costly, it's... socially exposed.

Ryan Castillo: So the structural mold survives because *multiple* people would feel it break. That's genuinely more robust than I'd have assumed.

Jordan Hale: And it makes me wonder — I mean, this is maybe complicated — whether oral and written records are even competing versions of the same thing. Or whether trying to reconcile them is actually the wrong goal entirely. That part I think we need to get into, because it complicates everything we've said so far.

Ryan Castillo: The incommensurability problem. Yeah — that's the layer underneath all of this.

Jordan Hale: But incommensurable is — I mean, that's a strong word. Like, are we actually saying the Māori oral histories and the colonial written records aren't even playing the same game? Because that feels like it lets everyone off too easy.

Ryan Castillo: Not off easy. It's actually harder. Because if they're incommensurable, you can't just average them. You have to decide what question you're asking before you pick your source.

Jordan Hale: Okay but — wait, is that actually what's happening with the Icelandic sagas? Because that debate is still live, right? Whether those texts are a living oral culture, or just literate scribes imitating oral form.

Ryan Castillo: Still unresolved. And the stakes of that question are real — because if the sagas are literate reflexes of oral tradition rather than oral tradition itself, then the evidential weight collapses. You're not reading a community's memory. You're reading someone's literary performance of it.

Jordan Hale: That's — yeah. That's a completely different thing.

Ryan Castillo: And it's why the social-meaning filter matters structurally. Oral tradition isn't preserving what happened. It's preserving what the community needed to keep alive — territorial claims, cosmology, identity. The chronological detail, the administrative precision — that's not what acoustic memory selects for.

Jordan Hale: So the compression isn't random loss — it's, like, the community actively deciding what's load-bearing. Genealogy survives because it anchors who you are and where you belong. The specific date a battle happened? That's not load-bearing for a community that needs to know *why* it matters.

Ryan Castillo: And Bartlett's finding is exactly that — the chain doesn't drift toward noise, it drifts toward cultural schema. Meaning oral transmission isn't just lossy, it's *directionally* lossy. It systematically sheds what doesn't cohere with the existing frame.

Jordan Hale: Which means when Australian Aboriginal or Polynesian oral histories conflict with material records — the conflict itself is data. You're not looking at an error. You're looking at what the community weighted.

Ryan Castillo: Right — but the methodological tension isn't resolved just by saying that. When Polynesian oral histories get corroborated by archaeology, great — constrained transmission wins. But when they conflict with material records, we still don't have an agreed framework for which one to weight.

Jordan Hale: And we're mostly studying the traditions that survived to be integrated — the Aboriginal, the Māori, the Polynesian records that made it to documentation. That's what strikes me. We don't know what the filter dropped because the dropped things are gone.

Ryan Castillo: That's survivorship, yeah. But I think there's a sharper version of it — a story too complex to fit the mold is a story that disappears. That's not survivorship bias layered on top of the system. That's the system working.

Jordan Hale: Oh. So the complexity ceiling isn't a failure mode. It's — the filter *is* the technology.

Ryan Castillo: Which reframes the reconciliation question entirely. If you try to treat oral and written records as competing versions of the same event, you're assuming they were capturing the same thing. They weren't. The divergence isn't the problem. It's the actual historical datum.

Jordan Hale: So the question isn't which one is right. It's — what does each system's filter tell you about the community doing the remembering? That's a different methodology entirely and I'm not sure historians have fully landed there yet.

Ryan Castillo: And I'm not sure they can. Land there, I mean. Because the oral/literate distinction was always an analytical category — it was never a clean historical one. Real communities straddled it constantly. The guslari performing for scribes. Literate monks writing down sagas. The category keeps leaking.

Jordan Hale: Which means — I mean, the frameworks we're using to evaluate oral histories were built inside literate traditions. By people whose entire epistemic toolkit was shaped by writing. And now we're asking those frameworks to fairly assess something they were never designed to see clearly. That's... I don't know how you get out of that without deciding whose court you're in first.

Ryan Castillo: That's the specific thing I can't resolve. If you insist on reconciling an indigenous oral history with a documentary record — forcing them to testify in the same evidentiary framework — you've already made a choice. You've picked the court. And the rules of that court were written by the literate side.

Jordan Hale: And dismissing oral accounts reproduces exactly that bias. But treating them as structurally equivalent without any methodological adjustment — without accounting for what the social-meaning filter actually selects for — that's not respect either. That's just flattening a different kind of distortion onto it.

Ryan Castillo: Two kinds of truth that weren't designed to agree.

Jordan Hale: Yeah. And the question I keep turning over — the one I genuinely cannot settle — is what happens to indigenous sovereignty, to a community's historical authority, if we keep insisting on reconciliation. Because maybe what a community chose to remember differently is the most important thing. Maybe *why* they weighted what they weighted is the actual history. And reconciliation just... erases that question before it gets asked.

Ryan Castillo: No verdict on that one from me. That's genuinely open. I think we got to the edge of something real today — which is probably the right place to stop.