The evidence linking Dogon to the Niger–Congo family is weak, and their place within the family, assuming they do belong, is not clear. Various theories have been proposed, placing them in Gur, Mande, or as an independent branch, the last now being the preferred approach. The Dogon languages show no remnants of the noun class system characteristic of much of Niger–Congo, leading linguists to conclude that they likely diverged from Niger–Congo very early. Roger Blench comments, and: The Bamana and Fula languages have exerted significant influence on Dogon, due to their close cultural and geographical ties. Blench suggests that Bangime and Dogon languages may have a substratum from a "missing" branch of Nilo-Saharan that had split off relatively early from Proto-Nilo-Saharan, and tentatively calls that branch "Plateau".
Languages
The Dogon consider themselves a single ethnic group, but recognise that their languages are different. In Dogon cosmology, Dogon constitutes six of the twelve languages of the world. Jamsay is thought to be the original Dogon language, but the Dogon "recognise a myriad of tiny distinctions even between parts of villages and sometimes individuals, and strive to preserve these". The best-studied Dogon language is the escarpment language Toro So of Sanga, due to Marcel Griaule's studies there and because Toro So was selected as one of thirteen national languages of Mali. It is mutually intelligible with other escarpment varieties. However, the plains languages—Tene Ka, Tomo Ka, and Jamsay, which are not intelligible with Toro so—have more speakers, and Jamsay and Tommo so are most conservative linguistically. Calame-Griaule appears to have been the first to work out the various varieties of Dogon. Calame-Griaule classified the languages as follows, with accommodation given for languages which have since been discovered, or have since been shown to be mutually intelligible. The two standard languages are asterisked.
Douyon and Blench report an additional variety, which is as yet unclassified:
Ana Tiŋa.
Blench noted that the plural suffix on nouns suggests that Budu is closest to Mombo, so it has been tentatively included as West Dogon above. He also notes that Walo–Kumbe is lexically similar to Naŋa; Hochstetler suspects it may be Naŋa. The similarities between these languages may be shared with Yanda. These are all extremely poorly known.
Pre-Dogon language
, formerly considered a divergent branch of Dogon, turns out not to be Dogon at all, and is possibly a language isolate. Blench believes that it is a remnant of the pre-Dogon languages of the area; the Dogon appear to have been in the area for many thousands of years. Additionally, Blench suggests that there is a Nilo-Saharan substratum in the Dogon languages, with the Nilo-Saharan substrate being a currently extinct branch of Nilo-Saharan that Blench tentatively refers to as "Plateau."
