Uan umur

The twenty-first text, and the first long one. Counted before it was written: twenty texts, 1875 words of running Amadunia, and the longest was thirty-four sentences. Everything the language had was a page.

Length is its own test. A short text can hide what a long one cannot — who is who after forty sentences with no gender and no case, whether a tense holds across a whole life, whether six sentence shapes become a drone.

Written at 300 roots, with no grammar beyond Lesson 26.

Anak lai in dom keci. Mama senang. Papa senang.
Dom in tarik keci. Es bage sub.
Anak kula, lala, kanta.
Mama sema kalima. Anak dengar.
Papa katab nama anak in libro.

Din-din go. Anak anda. Anak kimbia in bage.
Anak kan flor, pohon aur tair in asman.
Anak tanya: "Ini es ke?" Mama respon.
Anak tanya tena: "Porke?" Papa kanta.

Anak go skola. Skola in madina keci.
Mualim beri libro. Anak buka libro.
Kalima in libro muskil. Anak no faham.
Mualim sema: "Kara tena."
Anak kara tena. Sasa anak faham.
Anak katab nama ta. Anak katab kalima baru.
Anak studi numero aur historia.
Rafiki otur dekat. Rafiki genc aur hayai.
Kita kula pan una. Kita kimbia una.

Tahun-tahun go. Anak genc sasa.
Insan genc mau kan dunia.
Insan genc sema: "Mi mau go madina kabir."
Mama sedih. Papa sema: "Go."

Insan genc ambil tren. Tren anda cok hora.
Insan genc kan pahar, sungai, hutan aur bahari.
Madina kabir. Cok insan in tarik.
Insan genc naiti kamra keci. Kamra kotor.
Insan genc gasil kamra una sabun aur akua.

Insan genc naiti rabota in market.
Tacir beca fruta aur nyama.
Insan genc ambil fruta. Insan genc beca fruta.
Insan-insan lai. Insan-insan kupi.
Aksam lai. Insan genc funga market.
Pesa keci. Insan genc kula pan aur corba.
Insan genc lala. Insan genc punya mimpi.

Insan genc kan insan baru in market uan din.
Insan baru kupi legum aur telur.
Insan genc sema: "Salam." Insan sema: "Salam."
Ta-ta sema. Ta-ta kanta. Ta-ta anda in tarik.
Insan genc sema: "Mi ama yu." Insan sema: "Mi ama yu."

Ta-ta naiti dom keci in madina.
Es pencere kabir in kamra.
Ta-ta pone tabla aur du kursi in kamra.
Anak lai. Anak keci kula, lala, kanta.
Historia go tena.

Tahun-tahun go. Rambut siya sasa putih.
Anak-anak kabir sasa. Anak-anak go madina baid.
Es sukut in dom. Ta-ta otur in bage.
Ta-ta kan sol. Ta-ta ingat din-din eski.

Insan sema: "Mi ingat yu in skola."
Insan genc — no genc sasa — respon: "Mi ingat yu in market."
Ta-ta kanta tena.
Rat lai. Stela in asman.

Insan mati. Es sukut in dom tena.
Anak-anak lai. Anak-anak kan bage.
Anak keci tanya: "Kim es insan ini?"
Mama respon: "Ta es nasab yu. Ta suda hidup sini."
Anak keci kan foto. Anak keci ingat nama.

Insan mati. Kalima no mati.
Nama in libro. Historia in kalima.
Anak keci sema nama tena.

Line by line

What length found

101 sentences, 417 words, 122 roots — three times the longest text before it, and a fifth of everything the language has ever written in one page.

Who is who survived, and it should not have been obvious. The language has no gender and no case, so ta is he, she and it alike, and forty sentences separate the child from the old person it becomes. What holds it together is that the subject is named again whenever it changes — anak, insan genc, insan, ta-ta — which the language forces by never dropping a subject. The one rule that looked like a cost, no pronoun dropping, is what makes a long story readable.

The middle name was wrong for a day. The second stage was Genc — the young one — in seventeen sentences, and genc is an adjective. Nothing in the language lets an adjective stand as a noun, and letting it would cost a settled rule: an adjective predicate takes no es and a noun predicate must have one, so Ta genc would stop having one reading. It is insan genc now, a noun with an adjective after it, which is the shape the language actually has and says exactly what was meant. Found September 5, 2026 by counting adjacent word classes across the repository; every word in all seventeen sentences existed, which is why nothing had seen them. Recorded in copula.md.

The tense held by not being marked. The whole life is told in the unmarked present, and suda appears once — Ta suda hidup sini, she lived here — spoken by someone standing in the past’s garden. Marking every sentence would have been correct and unreadable. That the unmarked present carries a whole life is an open question answered by use rather than by decision, and it is worth saying out loud.

The drone is not real, and I claimed it was. Writing this page it felt like subject-verb-object over and over, so the first version of this section said so. Then it was measured, and the measurement disagrees twice.

By sentence shape — every sentence reduced to its parts of speech — this text uses 52 distinct shapes, more than any other text in the language; the next highest are story 2 and text 22 with 27 each. Its three commonest shapes cover 33% of it, which makes it the eighth least repetitive of thirty texts. And 8% of adjacent sentence pairs share a shape, fifteenth of thirty: the recipe is 28% and the poem 25%, both for good reasons.

Four of those five figures were wrong until September 6, 2026. The paragraph said 56 shapes, 31%, fifth of twenty-one and eleventh of twenty-one. Fifty-six is not a figure the repository’s own shape scanner produces at any scope — it gives 52 over the poem’s block and 53 over the whole file — and the other three went stale when text 22 and text 23 were written four days later. This is the section that exists because a claim about this text was measured and refuted; measuring it again is the same act, and it should not have taken four days. Every figure in it is derived now.

So length did not flatten the language. It did the opposite — a long text visits more situations and reaches for more shapes. The drone was in the writing of it, not in the reading, and the honest record of that is this paragraph rather than the one it replaced.

Gaps

No word for a song, and this page had been borrowing the verb. The line read Ta-ta kanta kanta eski — glossed “they sing an old song” — using kanta once as the verb and once as a noun. The dictionary glosses it to sing and nothing else, and one root does one job: the two roots that break that rule are named and this was quietly a third. The line says Ta-ta kanta tena now — they sing again — which loses the song and the age of it. Found September 6, 2026 by reducing every sentence in the material to its parts of speech and reading the shapes that occur once; the song is on the list of words the writing has asked for.

I reached for a word that does not exist, twice, without meaning to. Both drafts of this page wrote lakin — the candidate in the briefing for “but” — once at Pesa keci, lakin… and once at Kamra keci, lakin…. check.py refused both, correctly: a text invents nothing. That makes six pages that have wanted the word, and this is the first time it arrived in a draft by reflex rather than by argument.

No word for then, over eighty-eight sentences (the list). A life is a sequence and the page can only set its parts side by side. Every paragraph break in the block is doing work the grammar cannot.

Nothing for growing old. (Neither on the list nor in the checklist; it is recorded here as a judgement rather than a hole.) eski is old of things; a person is no genc, not young, which is what the text says. Whether that is a gap or a language choosing not to have one is a judgement, and it is recorded here rather than claimed as a hole.

Roots used

122 of 300.