No Word for It

The internet loves an untranslatable word. Lists of them circulate like postcards: words for the light through leaves, for the ache of a missing person, each offered as a small romance about some other language's soul. This page is not that. It is the checked version of the question, built from seven layers of this ground that each collected a real corpus and counted: Catullus's excrucior, Villon's antan, the Gītā's adhikāra, the Hebrew Bible's ḥesed, the Analects' 仁, the Qur'an's one absent alif, and the Lord's Prayer's epiousios.

Watch what the translators actually did, corpus by corpus, and the sentimental question ("what does the word mean?") gives way to an operational one: what are the moves? Across seven unrelated source traditions and four hundred years of English hands, the same five keep recurring.

IShatter: one word into many

The commonest move. The word is too big for any one English word, so the record breaks it into pieces and spends a different piece each time. The King James Bible, finding nothing large enough for ḥesed (the covenant-keeping love the Hebrew Bible turns on, in the book 251 times by that layer's own recount), shattered it into what its page calls "about a dozen" English words. Eleven distinct renderings stand in the member's verified grid, across the word's two senses:

The Gītā's adhikāra shatters more quietly. Eight public-domain English versions, 1785 to 1897 (of the more than three hundred English translations that exist), render the one word with five distinct English words, two of them doubled:

And when the missing word is a torture, the shatter shows its teeth. Catullus ends his fourteen-word couplet on excrucior: a first-person passive with a cross buried inside it. Ten renderings by nine public-domain hands scatter it into torture, torment, agony, the rack, the pains of hell, and merely "vext". Not one writes "crucified". Exactly one, Kelly's plain prose of 1854, keeps the passive.

IIFlatten: many into one

The opposite move, and the proof that the count measures the translator, not the word. Long before the King James shattered ḥesed, the Septuagint had flattened it: one Greek word, ἔλεος, "pity", covering (by a published lexical count the member page cites and is careful to say it did not recompute) about 87 percent of the word's occurrences. Every "mercy" in the English psalms descends from that single early choice.

Lyall did the same to 仁, deliberately. The Analects layer counts the graph 109 times across 59 passages, and its three-hand gallery shows Lyall rendering it love, every time, and letting the sentences strain. A count of one is still a decision about a word with no equivalent; it just spends the whole budget in one place.

IIICoin: make the word

The rarest move. In 1870 Dante Gabriel Rossetti, needing a refrain-sized English word for Villon's antan (which meant, exactly, "last year"), made one: yester-year, printed hyphen and all. The member layer verifies the claim against the dictionary itself: Merriam-Webster dates the English word's first known use to 1870, this poem. The other public-domain hand, Payne's of 1878, stayed literal with "last year's snow"; the living language kept Rossetti's coin.

The move even has an ancient precedent inside these seven: the source can coin too. Origen, the first person we have examining epiousios, said the Evangelists appear to have invented it. Some words arrive in the record already new.

IVFork in one hand

Sometimes a single translator cannot settle it, and the record catches the indecision verbatim. Jerome, translating the same Greek word in the same prayer in the same Vulgate, wrote supersubstantialem in Matthew and cotidianum ("daily") in Luke. Legge rendered 仁 six different ways inside one 1893 book:

Cranstoun did it to Catullus in miniature: unable to settle excrucior, he printed two complete versions in 1867, one ending on the rack, the other importing "the pains of hell". The fork in one hand is the untranslatable word caught in the act: not a disagreement between schools or centuries, but one mind, holding both, refusing to lose either.

VThe fork is upstream

And sometimes the divergence is not the translator's at all. The Qur'an's opening sura holds a word that is two words: مَٰلِك / مَلِك, Owner or King, one long vowel apart, and both readings canonical. Of the classical Seven readers, two (plus Yaʿqūb) read Owner; five read King. The four public-domain English Qur'ans split exactly along that fault:

The extreme case closes the walk. Epiousios, the word behind "our daily bread", occurs in the whole surviving record of ancient Greek essentially nowhere else; four rival derivations are still live; the member page's verdict is that the meaning was lost before the ink was dry. Five public-domain English Bibles line up two against three, and the split is not about Greek at all:

What the count counts

Seven corpora, seven tallies, and the temptation is to line them up: is ḥesed at eleven "more untranslatable" than antan at two? The data itself forbids the ranking, and the refusal is the honest finding. The corpora run from two hands to ten; they were bounded by what is out of copyright, not by what exists; and the same word lands on opposite ends depending on who is counting: ḥesed is a one-word flatten in the Septuagint and an eleven-way shatter in the King James. 仁 is six words in Legge and one in Lyall. The count is a real, checkable observable of a collected corpus. It is not a scale the words share.

The move belongs to the translator, not the word. the one sentence this portal adds; everything above it is lifted

That is the portal's whole addition: a sorting of seven verified records into five recurring moves, and the observation, forced by ḥesed and 仁 carrying opposite counts in different hands, that which move a word gets is a fact about the hand that held it. No law forces the choice. One move entered the dictionary. Two of the seven words were forked before English ever arrived. And in four hundred years of trying, nobody has been wrong, exactly: as the Catullus layer puts it, every hand chooses what to lose.