How This Corpus Handles Evidence

Almost everything that goes wrong in writing about Yorùbá tradition goes wrong in the same way: three different kinds of statement get written in the same voice, so the reader cannot tell which is which. "Odùduwà descended from the sky at Ilé-Ifẹ̀," "radiocarbon dates from Ifẹ̀ place substantial occupation from around the eighth century," and "the natives worship idols and are governed by superstition" are all sentences about Ifẹ̀, and they have nothing whatever in common as evidence. This file sets out how this corpus keeps them apart.

The three registers

Register one: what the tradition says of itself

This covers ìtàn (narrative, history, the account of how something came to be), oríkì (praise poetry attached to a person, lineage, town or òrìṣà), ẹsẹ Ifá (the verses of the Ifá corpus), proverbs, songs, and the accounts given by priests, elders and title-holders about their own institutions.

These are primary sources of the first importance and they are not neutral. Yorùbá oral genres are performed by people with positions to defend. An oríkì praising a lineage is produced by and for that lineage. A dynastic ìtàn establishing that a ruling house descends from Odùduwà is a claim to legitimacy made in a competitive field where other houses make other claims. The Ifá corpus is transmitted by babaláwo whose authority the corpus itself underwrites. None of this makes the material unreliable; it makes it evidence of a particular kind, which historians handle by asking who is performing, for whom, and against which rival account.

When this corpus reports material in this register it says so explicitly: "the tradition holds," "the ìtàn of Ilé-Ifẹ̀ gives," "in the Odù Ọ̀sá Méjì it is said." That phrasing is doing real work and is not a hedge.

Register two: what historians and archaeologists document

This covers excavation and its dating, linguistic reconstruction, art-historical and materials analysis, the reconstruction of trade from goods and residues, documentary records where they exist, and the whole apparatus of academic history.

It has its own limits, and they should be stated. Radiocarbon gives a date range for a sample, not a date for an event, and the relationship between the two requires an argument. Linguistic reconstruction gives relative sequence and rough time-depth, not calendar years. Excavation in Yorubaland is thin relative to the size of the region and heavily weighted toward a few sites, above all Ilé-Ifẹ̀, so absence of evidence is very often absence of digging. Where a claim in this register rests on a single excavation or a single scholar, this corpus says so.

The most important recent work in this register for Yorùbá deep history is Akínwùmí Ògúndìran's The Yorùbá: A New History, which combines archaeology, historical linguistics, environmental science and oral tradition into a single account and traces Yorùbá origins to the region of the Niger-Benue confluence with subsequent expansion driven in part by climate change 1. It is cited throughout the history section.

Register three: what colonial and missionary sources claimed

This covers the Church Missionary Society journals, travellers' accounts, colonial administrative reports, and the early ethnography written by missionaries and officials.

These sources are indispensable and they are systematically distorted. They are indispensable because for much of the nineteenth century they are the only continuous written record of Yorùbá life, and because the CMS journals in particular were kept largely by Yorùbá agents who were themselves the first literate generation of Yorùbá people, which makes them a Yorùbá source as well as a mission source 2. They are distorted because their authors were engaged in an argument. A missionary describing an òrìṣà shrine was describing something he was there to abolish, in a vocabulary supplied by his theology, for readers in England whose donations funded the mission. J. D. Y. Peel's reading of these archives establishes both facts at once: the journals are extraordinarily rich, and their authors "saw heathenism as an absence or vacuum, rather than as something with durability, style, and an ethos of its own" 2

This corpus cites colonial-era sources freely and always labels them. Where such a source is the origin of a claim still in circulation, the claim is traced back to it and the claim's context is given. The file on the demonization is largely an exercise in exactly this.

Why the separation matters

Blending register one with register two produces the celebratory error: tradition restated as established history, so that "Odùduwà came from the east" becomes evidence of a migration from Arabia or Egypt. Blending register three with register two produces the hostile error: missionary characterisation restated as ethnographic fact, so that "the Yorùbá worship idols and have no notion of God" enters the reference literature and stays there for a century. Both errors have the same shape. Keeping the registers apart is the single most useful discipline available.

The confidence markers

Every file carries confidence: high | medium | contested in its frontmatter, and the marker refers to the file's central claims rather than to every sentence in it.

High means the claim is supported by multiple independent sources of good quality and is not seriously disputed by working scholars. Yorùbá is a tone language with three phonemic tones. The Ifá corpus is organised into 256 Odù. Ifá was inscribed on the UNESCO Representative List in 2008 3. These are high-confidence claims.

Medium means the claim is supported but rests on a thinner base, or is an interpretation on which reasonable scholars could differ, or involves reconstruction from indirect evidence. Most of the history before the nineteenth century sits here. So does most translation of philosophical vocabulary, because rendering àṣẹ or ìwà into English involves interpretive choices that the English word then conceals.

Contested means scholars actively disagree, and the file names the positions rather than choosing between them. Whether Yorùbá religion is properly described as monotheistic is contested: E. Bọ́lájí Ìdòwú argued in 1962 for what he called diffused monotheism, in which the òrìṣà are conceptualisations of attributes of Ọlọ́dùmarè, and that reading has been criticised on the ground that it imports a Christian theological frame into a system that did not organise itself that way 4. The dating of the Ifẹ̀ sculptural corpus is contested. The historicity of specific dynastic sequences is contested. Where you see the marker, expect the file to lay out the disagreement rather than resolve it.

A note on what confidence markers are not. They are not a measure of how important something is or how strongly anyone believes it. A religious claim held with total conviction by millions of people is not thereby a high-confidence historical claim, and saying so is not a comment on the claim's religious standing. The marker describes evidentiary support and nothing else.

Initiate-held knowledge, and why it stays a gap

A substantial part of Yorùbá religious knowledge is not public. It is transmitted within initiation, under conditions that include restriction on who may receive it and on what may be repeated outside. This applies to portions of the Ifá corpus, to the interior content of particular òrìṣà cults, to Ògbóni and Òṣùgbó procedure, to elements of masquerade practice, and to a good deal of the pharmacological and ritual knowledge held by herbalists and priests.

This corpus does not attempt to supply that material, for three reasons.

The first is straightforwardly evidentiary. Where knowledge is restricted, what circulates in public is unreliable by construction. Published "secret" material is typically either a partial account from an informant who was not authorised to give it, a reconstruction by an outsider, or an invention. A corpus that repeated it would be presenting low-grade material at the exact points where readers most want certainty, which is the worst possible place to be inaccurate.

The second is that the restriction is itself a fact about the system and should be described rather than circumvented. That Ifá training takes years, that Odù are learned in a sequence, that certain material is given only at certain stages, that some things are said only inside a grove: these are structural features of how the knowledge is organised and transmitted, and they tell you something real about the tradition. Reporting the structure is more informative than smuggling out fragments of the content.

The third is that the alternative is speculation, and speculation in this domain has an established habit of becoming citation. A guess published in one book is footnoted in the next and is a fact by the third.

So where this corpus reaches the boundary, it says so in that form: this material is initiate-held, it is not published, what is publicly attested is the following, and beyond that the corpus does not go. Files where this is the central limitation are marked confidence: contested.

The translation rule

Any Yorùbá text quoted anywhere in this corpus, whether a proverb, an oríkì, an ẹsẹ Ifá, a song or a prayer, is presented in three layers: the original with full diacritics, a literal gloss showing what the Yorùbá actually says including the metaphor before it is smoothed, and an idiomatic English rendering with the translator named. Then a note on what the translation loses.

This exists because translation in this field is where most of the damage was done and where a great deal of it remains. The clearest case is the mission-era rendering of Èṣù as Satan in the Yorùbá Bible, a translator's decision made in the mid-nineteenth century with consequences that are still shaping how Yorùbá people understand their own inheritance 5. That decision is not visible to a reader of the Yorùbá Bible; it looks like a fact about Yorùbá religion rather than a choice about English. Showing the layers is how such choices are made visible.

Where published translations differ materially, this corpus shows both and names the translators. Where a mission-era translation is visibly shaped by the translator's theology, that is stated and what the Yorùbá actually says is shown. Sources are never silently corrected.

What this corpus does when it does not know

It says so, and it says what would settle the question. That formulation appears throughout. It is more useful than a confident sentence, and it is the only honest thing to write when the evidence is not there.