Coverage
What each language has so far
Word counts, phrase counts, how much of the core vocabulary is covered, and how much of it can be heard. The gaps are listed alongside the totals.
- 4,712
- Words checked by a person
- 0
- Recordings, from 0 voices
- 1,199
- Still need a definition
- 358
- Still need a speaker
Where the languages are
County by county, which language communities are mapped there and how much of each one exists so far. Hover or tap a county to inspect it.
Shaded by the best covered local language. Swahili and English are listed where they are spoken but do not set the shade, or every county would look the same.
Taita-Taveta
coast · 65.3% of core meanings
Taita and Taveta communities shape the county’s linguistic identity.
- Taita65.3% covered
158 words, 15 phrases
- Taveta64.9% covered
157 words, 15 phrases
- SwahiliLingua franca
712 words, 0 phrases
The thinnest language mapped here, with 157 words.
Every language
Sorted by size. The last two columns are the ones that matter most, because a word nobody has said out loud cannot teach a machine anything.
| Language | Words | Phrases | Core meanings | Recordings | Voices |
|---|---|---|---|---|---|
| Swahili | 712 | 0 | 53.8% | None | None |
| Dholuo | 198 | 20 | 73.8% | None | None |
| Luhya | 191 | 17 | 72% | None | None |
| Kalenjin | 189 | 16 | 68.9% | None | None |
| Kenyan Hindustani | 161 | 16 | 64.9% | None | None |
| Taita | 158 | 15 | 65.3% | None | None |
| Duruma | 157 | 16 | 65.3% | None | None |
| Taveta | 157 | 15 | 64.9% | None | None |
| Chonyi | 157 | 16 | 65.8% | None | None |
| Giriama | 157 | 16 | 64.9% | None | None |
| Digo | 157 | 16 | 65.3% | None | None |
| Kamba | 153 | 15 | 55.1% | None | None |
| Borana Oromo | 151 | 13 | 64% | None | None |
| Gusii | 149 | 12 | 62.2% | None | None |
| Logooli | 149 | 12 | 63.1% | None | None |
| Kikuyu | 144 | 14 | 45.8% | None | None |
| Bukusu | 140 | 13 | 55.1% | None | None |
| Somali | 137 | 14 | 57.8% | None | None |
| Sheng | 127 | 10 | 48% | None | None |
| Nandi | 118 | 11 | 48.9% | None | None |
| Kipsigis | 118 | 14 | 48.9% | None | None |
| Maasai | 115 | 16 | 47.6% | None | None |
| Turkana | 97 | 11 | 40% | None | None |
| Samburu | 91 | 15 | 36% | None | None |
| Embu | 87 | 11 | 36% | None | None |
| Meru | 87 | 11 | 36% | None | None |
| Sabaot | 68 | 14 | 27.1% | None | None |
| Pokot | 54 | 10 | 21.3% | None | None |
| Orma | 52 | 19 | 18.7% | None | None |
| Rabai | 50 | 16 | 19.6% | None | None |
| Kuria | 49 | 14 | 19.1% | None | None |
| Suba | 48 | 15 | 19.6% | None | None |
| Olutsotso | 45 | 15 | 14.7% | None | None |
| Pokomo | 42 | 12 | 17.8% | None | None |
| Rendille | 25 | 10 | 10.7% | None | None |
| Tugen | 22 | 12 | 7.6% | None | None |
Thinnest first
The languages furthest from covering the core vocabulary. An hour spent on one of these moves the corpus further than an hour spent anywhere else.
Not one of these languages can be heard yet
Words on a page are half the work. Speech recognition, text to speech and every device built on them need recorded voices, and those cannot be collected later from speakers who are no longer here.