Languages
Coverage by language
Ethiopia has more than 80 languages. The catalogue lists 100 entries, every language Glottolog records for the country, of which 91 are living or endangered. The remainder go beyond the familiar "more than 80": 6 extinct languages, 2 pidgins, and a few whose classification or status Glottolog does not record. Full method.
Languages with a model or dataset
Ethiopian languages with a model or dataset from this community, and what each has. Papers are indexed works matching the language name; a ≈ marks a count that over-matches.
| Language | Family | Models | Datasets | Papers |
|---|---|---|---|---|
OromoAfaan Oromooorm |
Cushitic | 7 | 1 | 270 |
Amharicአማርኛamh |
Semitic | 22 | 9 | 1248 |
SomaliAf-Soomaalisom |
Cushitic | 6 | 2 | 193≈ |
Tigrinyaትግርኛtir |
Semitic | 7 | 3 | 119 |
Wolayttawal |
Ta-Ne-Omotic | 1 | 0 | 26 |
Geezግዕዝgez |
Semitic | 5 | 0 | not recorded |
6 of 100 catalogue entries. Ge'ez is liturgical rather than spoken, and Wolaytta's coverage is a single tokenizer, so the smaller counts here should be read with care.
Languages with research but no released resources
Languages that have a research literature but no open model or dataset. These are possible next targets.
-
Sidamo
sid5 papers≈ -
Afar
aar33 papers≈ -
Kafa
kbr7 papers≈ -
Gedeo
drs1 paper≈ -
Silt'e
stv3 papers≈ -
Nuer
nus6 papers≈ -
Kambaata
ktb20 papers≈ -
Awngi
awn10 papers -
Berta
wti34 papers≈
Languages with no model, dataset or indexed paper
The ten with the largest recorded speaker populations. Those figures are not shown at present, because the available sources are not good enough to publish; the ordering still uses them.
-
Sebat Bet Gurage
sgwSemitic -
Bench
bcqTa-Ne-Omotic -
Alaba-K'abeena
alwCushitic -
Aari
aiwSouth Omotic -
Saho
ssyCushitic -
Ethiopian Sign Language
ethSign Language -
Konso
kxcCushitic -
Northern Gumuz
gukGumuz -
Xamtanga
xanCushitic -
Mesqan
mvzSemitic
85 of the 100 languages in the catalogue have nothing built and nothing published. See the full catalogue.
Coverage counts work built for a language. A hundred-language model that lists Amharic among its tags is not counted here; see the note on multilingual work for why that distinction is kept.