Language Precedes World

In the summer of 637 the armies met at Mag Rath. The Annals of Ulster record it in a line ¹ — attestation, not story. No speeches, no casualties, no meaning. The battle has its own page on this site, and that page keeps three layers apart: what the annals attest, what the saga remembers, and what the story is made to mean. Keep them apart here too.

What the annals do not record is what the poets recorded. Auraicept na nÉces — the Scholars’ Primer, the handbook of the filid — opens by explaining why it came to be written at all, and the explanation is a casualty list of a peculiar kind. It names four glorious events of that battle. Two of them are men.

Suibhne is the king who broke. He goes out of the battle into the trees, into geltacht — which is not a nickname but a condition, with its own entry in the lexicon and its own vocabulary of terror, flight, feathers, raw food, and visions. ³ What he does in the branches is compose. This site is named for him.

Cenn Fáelad mac Ailella took a sword-blow to the head in the same battle. He was carried to the house of Bricín at Tuaim Drecain, at the meeting of three roads, where three schools stood together: Latin letters, Irish law, Irish poetry. The blow, the tradition says, struck out his inchinn dermait — his brain of forgetting. Thereafter whatever he heard rehearsed in the three schools by day he held by night, and he wrote it down. ⁴

Notice what the Primer has done. It has taken one battle and produced two makers from it, and it assigns both afflictions to the same cause: an excess of poetry. One man loses his reason and gains song. The other loses his forgetting and gains grammar. And the second man’s work is the book that makes the claim this essay is about.

Ask the Primer where Irish came from and it does not answer the way a grammar answers.

After the dispersal from Babel, Fenius Farsaid keeps a school. Seventy-two disciples go out to the seventy-two nations and learn the seventy-two tongues — the schema is Isidore’s, and the Primer’s compilers had him in front of them. ⁷ After ten years they return, and at their own request a language is assembled out of what they have brought back, so that the speech should belong to nobody but themselves.

The technical term is bérla tóbaide: the selected language, the language cut out. The Primer asks itself why the phrase should be used of Gaelic and answers that it was selected from every language, and that for every obscure sound of every language a place was found in Gaelic, owing to its comprehensiveness beyond every speech. ⁵

Two things follow, and the second is the one this essay needs.

The first is a limit, and it must be stated before the parallel is drawn. Selection is not creation from nothing, and the text is scrupulous about this. It puts the objection to itself — did Gaelic not exist before it was selected? — and answers plainly that it did indeed, for otherwise the seventy-two languages are not found. What is being claimed is engineering: a tongue cut from the best of all the others and regulated, chosen for its harmony and its comprehensiveness. ⁶

The second is the ordering. Having made the language, the tradition then supplies it with a people and a history. The language, Goídelc, takes its name from one Gaedel; the people, the Goídil, take theirs from another. ⁸ And the descent that carries those people to Ireland is the Lebor Gabála, the national origin-legend, the book that tells the Gaels where they came from. The sequence is not decorative. Language, then a people to speak it, then a history for that people to have happened in.

That this is an ideological document and not merely a technical one is settled ground, and worth saying so the essay does not appear to be discovering it. ¹⁰ A learned class is asserting the standing of its own vernacular. What interests this room is the shape of the assertion: not our language is ancient, which is the ordinary claim, but our language was made, and the world followed.

In June 1955 an Oxford professor of English is asked by his American publisher to explain himself, and does. The invention of languages, he tells them, is the foundation. The stories were made to provide a world for the languages rather than the other way about. To him a name comes first, and the story follows. ¹¹

That is a private letter, and private letters are where writers flatter their own methods. But he said it in public too, in the foreword to the second edition of The Lord of the Rings: the Silmarillion material was primarily linguistic in inspiration, and was begun in order to provide the necessary background of history that the Elvish tongues needed in order to exist. ¹²

He had already done it once in the smallest possible compass. As an undergraduate, reading the Exeter Book, he came on a word in the Advent Lyrics that he could not entirely gloss — earendel — in a line that carries a second word he would need later: ¹³

Two seed-words of a mythology sitting in one sentence of Old English verse, neither of them his: the name he could not fully account for, and middangeard, the middle enclosure, the earth between. In September 1914 he wrote a poem about the figure the word seemed to imply. Everything after — the languages, the legendarium, the world — grows from the far side of that line. He did not invent Middle-earth. He found it in a dative phrase and translated it.

The full case belongs to the Tolkien study, where the languages and the scholarship are set out at length. What this essay needs from him is only the ordering, and the fact that he stated it plainly, more than once, without being asked to defend it.

Here the site’s own argument begins, and the change of register should be visible.

The obvious objection to setting these two beside each other is that the Irish claim is weaker than it looks. Fenius’s school does not invent a language; it selects one, and the text says so outright. Tolkien, on the usual account, invented his outright. Two different acts, and the parallel is a coincidence of rhetoric.

I do not think that objection survives contact with how Tolkien actually worked.

Quenya’s phonology is Finnish. Sindarin was built with Welsh-like phonotactics and a mutation system founded on the same lenition of voiced stops after vowels that Welsh performs. Khuzdul stands on triconsonantal roots after the Semitic model, primarily Hebrew. The dwarf-names came out of the Dvergatal; the speech of Dale is rendered by Old Norse and the speech of the Mark by Mercian Old English. He had Latin, Greek, Gothic, Old and Middle English, Old Norse, Finnish and Welsh, and out of what he had read he cut two nearly complete languages and a dozen more in outline.

That is Fenius’s method exactly. Send out to the seventy-two tongues, take what each does best, and cut a language from the harvest. Neither tradition claims creation from nothing. Both describe a learned class performing comparative selection on the languages available to it and calling the result a made thing — and both then discover that a made language is uninhabitable until somebody supplies the speakers.

The difference is one of candour rather than of kind. The Primer says the language it assembled is the one actually spoken in Ireland. Tolkien claimed only that his was spoken in a country he had also to build. The medieval version is the bolder claim by a considerable margin.

The strongest thing about this parallel is that it cannot be a borrowing.

There is no evidence that Tolkien knew the Auraicept, or the dinnshenchas, or the Irish grammatical tradition at any point in his life. His Irish was weak. He worked at it and never mastered it, and said the language was unattractive to him — this from a man who taught medieval Welsh and could read it for pleasure. Where he took Irish material he took narrative: the Lebor Gabála behind the early Noldor, the immram behind his own St Brendan poem. He did not read the grammarians. ¹⁴

He is also on record warning against precisely the kind of parallel-hunting this essay could decay into. “Celtic” of any sort, he said in English and Welsh, is a magic bag into which anything may be put and out of which almost anything may come. ¹⁵ That warning is taken here as binding, which is why this movement exists at all.

So: not influence. Convergence — and that is the stronger claim, not the weaker one.

An idea inherited tells you about the person who inherited it. An idea arrived at twice, by two learned classes with no contact, in different centuries, working on different languages for different reasons, tells you something about the idea. A poets’ school in medieval Ireland and a philologist in mid-century Oxford reached the same order of operations: make the language, then build the place where it could have been spoken. Neither was eccentric. They were both doing something the mind evidently does when it takes language seriously enough.

Until now the Word-Hoard has been defined by what it is not. Four wings hold four matters; this room holds none of them. Tolkien’s centre of gravity is Germanic, there is no Germanic wing, and inventing one would break the Tetramorph — so he is kept here rather than filed. All of that is a true account of how the room came to exist and a poor account of what it is for. A room defined negatively is a leftover.

The claim above gives it a charter. The Tetramorph divides what is read. The Word-Hoard holds what one reads with — and holds the argument that the instrument is not always downstream of the matter. Sometimes the instrument comes first, and the matter is what has to be built so that the instrument can be used at all.

That is worth keeping a room for, and it is why the room is not an annexe to the Germanic material. Its charter is Irish. The name over the door is Old English, taken from Beowulf, and the modern witness inside could not read Irish — and those two facts are the argument rather than an embarrassment to it. A claim that turns up in two traditions that never met is not the property of either.

It also rhymes with the office this whole site is named for. The bard’s dilemma is that a word given no durable body does not last: the poems in the tree vanish into the air unless a second maker writes them down. Fenius’s school and the Oxford philologist had the inverse problem. A made language with no speakers, no history and no literature is not a word waiting for a body; it is a body waiting for a world. So they built the world. And in the cosmology Tolkien eventually wrote for his, the making is done by singing ¹⁶ — which is the bardic claim restated as theology: the word does not describe the thing, it makes it.