Languages in the same family descend from a shared ancestor, which is why they tend to share vocabulary roots and sentence structure. That relatedness is the single best predictor of whether machine translation between two languages will read naturally — and where it will quietly fail.
Indo-European is the largest family here, with 85 languages and roughly 4.9 billion speakers.
At a glance
| Family | Languages | Approx. speakers |
|---|---|---|
| Indo-European | 85 | 4.9 billion |
| Niger-Congo | 36 | 412 million |
| Austronesian | 35 | 673 million |
| Turkic | 13 | 184 million |
| Afro-Asiatic | 12 | 615 million |
| Sino-Tibetan | 12 | 175 million |
| Uralic | 7 | 20 million |
| Nilo-Saharan | 6 | 16 million |
| Dravidian | 5 | 277 million |
| Austroasiatic | 4 | 116 million |
| Eskimo-Aleut | 3 | — |
| Mayan | 3 | — |
| Mongolic | 2 | 5 million |
| Northeast Caucasian | 2 | 2 million |
| Japonic | 1 | 128 million |
| Tai-Kadai | 1 | 3 million |
| Kra-Dai | 1 | 65 million |
| Tupian | 1 | 7 million |
| Quechuan | 1 | 10 million |
| Northwest Caucasian | 1 | — |
What relatedness does to translation
Between related languages — Spanish and Portuguese, Dutch and German — machine translation has cognates to lean on and comparable clause structure, so output usually reads naturally. The trap is false friends: words that look shared but have drifted apart. Portuguese esquisito means “strange”, not “exquisite”. These are more common between relatives, not less.
Between unrelated families there are almost no cognates, so idioms and fixed expressions rarely survive a literal mapping. Short, plain sentences translate markedly better than figurative ones — which is worth knowing before you paste in a paragraph of marketing copy.
Indo-European
85 languages · roughly 4.9 billion speakers
EnglishHindiSpanishPortugueseBengaliUrduRussianPunjabi GurmukhiPunjabi PakistaniPersianGermanMarathiFrenchItalianPashtoGujaratiPolishBhojpuriUkrainianAwadhiOdiaMaithiliSindhiDutchRomanianKurdish KurmanjiNepaliSinhalaAssameseDariGreekMarwadiSerbianCzechSwedishKurdish SoraniPortuguese PortugalBulgarianBaluchiTajikAlbanianAfrikaansArmenianDanishCroatianNorwegianSlovakSicilianRomaniCatalanGeorgianLombardLithuanianHunsrikDogriBelarusianBosnianMacedonianSlovenianKonkaniGalicianVenetianLatvianYiddishOccitanLimburgishSilesianCorsicanFrisianIcelandicIrishLatinLuxembourgishScots GaelicWelshDhivehiSanskritManxPapiamentoFaroeseBretonFriulianLatgalianLigurianOssetian
Niger-Congo
36 languages · roughly 412 million speakers
SwahiliYorubaIgboFulaniLingalaChichewaKinyarwandaShonaBambaraZuluRundiTwiKitubaXhosaTswanaTshilubaSesothoLugandaTivBembaDyulaKikongoSepediEweTsongaBaouleFonTumbukaGaKigaSwatiNdauVendaNdebele SouthKrioDombe
Austronesian
35 languages · roughly 673 million speakers
MalayTagalogJavaneseIndonesianSundaneseFilipinoMalagasyCebuanoMadureseIlocanoHiligaynonMinangBikolAcehneseBalineseWarayBetawiPangasinanKapampanganBatak TobaMakassarIbanHawaiianMaoriSamoanJawiFijianBatak KaroBatak SimalungunChamorroChuukeseMarshalleseTahitianTetumTongan
Turkic
13 languages · roughly 184 million speakers
TurkishUzbekAzerbaijaniKazakhUyghurKyrgyzTatarTurkmenBashkirChuvashCrimean TatarTuvanYakut
Afro-Asiatic
12 languages · roughly 615 million speakers
ArabicHausaEgyptian ArabicOromoAmharicTamazight TifinaghSomaliHebrewTigrinyaTamazightAfarMaltese
Sino-Tibetan
12 languages · roughly 175 million speakers
CantoneseChinese TraditionalMyanmar BurmeseMeiteilon ManipuriTibetanMizoJingpoKokborokChineseHakha ChinDzongkhaNepalbhasa Newari
Uralic
7 languages · roughly 20 million speakers
Nilo-Saharan
6 languages · roughly 16 million speakers
Dravidian
5 languages · roughly 277 million speakers
Austroasiatic
4 languages · roughly 116 million speakers
Eskimo-Aleut
3 languages
Mayan
3 languages
Mongolic
2 languages · roughly 5 million speakers
Northeast Caucasian
2 languages · roughly 2 million speakers
Japonic
1 language · roughly 128 million speakers
Tai-Kadai
1 language · roughly 3 million speakers
Kra-Dai
1 language · roughly 65 million speakers
Tupian
1 language · roughly 7 million speakers
Quechuan
1 language · roughly 10 million speakers
Northwest Caucasian
1 language
Related
Method and limitations
Family assignments come from each language's own description. Where an entry names only a branch — “part of the Romance language group” — it is mapped to its parent family.
Disputed groupings are deliberately not forced. Korean is described in its entry as a probable isolate that some linguists place in Altaic; rather than pick a side, it is left unclassified here, as are Basque, Esperanto and other isolates and constructed languages. Of 248 languages, 231 resolve to a family.
Speaker totals are the sum of the individual figures we hold, which exist for 190 languages. They count first- and second-language speakers inconsistently across sources, so treat them as orders of magnitude rather than precise counts. Families are ordered by number of languages, not by speakers.