Generated from the material a learner reads — the twenty-six lessons, the thirty texts and the phrasebook — counting every word inside an Amadunia sentence and ignoring the vocabulary tables, which teach a word rather than use it. 6632 words of running Amadunia, and 291 of the 300 roots appear in one.
The other nine appear in no sentence of more than one word anywhere: bai, bas, hi, mersi, ok, pardon, plis, ya — and madad, which is held back on purpose until its class is decided. The eight are interjections and are used alone, which the language grants: a single constituent may stand as an utterance. This page counts runs of two words or more, so a one-word utterance is invisible to it, and until September 6, 2026 the count read every one of the 300 because the phrasebook’s list of twenty words — a row of vocabulary separated by dots — was being read as two ten-word sentences. Excluding it took twenty words off the total and left the eight with nothing, which is the truth about them and not a gap.
check.py regenerates this page and fails if it disagrees, so
the numbers follow the corpus rather than the other way round.
The shape of it
| Share of all running words | |
|---|---|
| first 10 | 31% |
| first 25 | 50% |
| first 50 | 65% |
| first 100 | 81% |
| first 150 | 90% |
| first 200 | 95% |
| first 300 | 100% |
Ten roots carry a third of everything written. That is steeper than a natural language and it is what a small dictionary looks like from the inside: the grammar words — mi, yu, ta, es, in, ini, suda, no, aur — do most of the work, and the nouns take turns.
For a learner the practical reading is the second row: twenty-five roots cover half of everything. They are the first twenty-five below.
The forty commonest
| Root | Meaning | Uses | Share | |
|---|---|---|---|---|
| 1 | mi | I, me | 664 | 10.0% |
| 2 | yu | you — one form for everyone | 240 | 3.6% |
| 3 | ta | he, she, it — no gender | 215 | 3.2% |
| 4 | in | at, in, on | 195 | 2.9% |
| 5 | anak | child | 151 | 2.3% |
| 6 | es | is, am, are — before a noun predicate only; see grammar/copula.md | 134 | 2.0% |
| 7 | lai | to come | 123 | 1.9% |
| 8 | dom | house, home | 120 | 1.8% |
| 9 | go | to go | 116 | 1.7% |
| 10 | no | no; not — before the predicate, see grammar/negation.md | 115 | 1.7% |
| 11 | ini | this — after the noun, last in the phrase | 109 | 1.6% |
| 12 | suda | past marker (before the verb) | 107 | 1.6% |
| 13 | aur | and | 104 | 1.6% |
| 14 | hao | good | 101 | 1.5% |
| 15 | mau | to want | 99 | 1.5% |
| 16 | ca | tea | 88 | 1.3% |
| 17 | sema | to say, to speak | 82 | 1.2% |
| 18 | kita | we, including you — see grammar/pronouns.md | 73 | 1.1% |
| 19 | keci | small | 72 | 1.1% |
| 20 | kula | to eat | 72 | 1.1% |
| 21 | rafiki | friend | 70 | 1.1% |
| 22 | kabir | big | 69 | 1.0% |
| 23 | saufa | future marker (before the verb) | 58 | 0.9% |
| 24 | sol | sun | 58 | 0.9% |
| 25 | dari | from | 56 | 0.8% |
| 26 | insan | person, human | 55 | 0.8% |
| 27 | kan | to see | 55 | 0.8% |
| 28 | punya | to have | 54 | 0.8% |
| 29 | pan | bread, food | 50 | 0.8% |
| 30 | libro | book | 47 | 0.7% |
| 31 | din | day | 46 | 0.7% |
| 32 | lebi | more | 45 | 0.7% |
| 33 | tarik | road, way | 44 | 0.7% |
| 34 | market | market, shop | 42 | 0.6% |
| 35 | akua | water | 41 | 0.6% |
| 36 | sasa | now | 41 | 0.6% |
| 37 | sini | here | 40 | 0.6% |
| 38 | itu | that | 39 | 0.6% |
| 39 | por | to, for | 39 | 0.6% |
| 40 | mama | mother | 38 | 0.6% |
Used exactly once
1 roots appear in a single sentence in the whole corpus:
salam
The list read nine until September 6, 2026 — bai, bas, hi, madad, mersi, ok, pardon, plis, ya — and every one of those nine has moved to zero rather than one, because their single appearance was the twenty-word list this page had been counting as two sentences. Nothing about them changed; what changed is that they are no longer credited with a sentence nobody wrote. salam is the only root left that is used once and used in a sentence.
It was 37 roots long two texts ago. Text 9 was written from it and took eighteen, one of which needed a second attempt because a word alone between two commas is named rather than used. What survived that was all of one kind — peace, art, history, law, nature, number, chance, to die, to try, a thousand — and eleven texts about tea, hospitals, letters and errands had not reached a single one. Abstract words need an argument, not a scene, so text 10 was written as one, and took all ten.