What is Songhay's Family?

Question

What's the current thinking on genetic affiliations of Songhay?

My old texts confidentially place it as Nilo-Saharan. This places a likely proto-Songhay homeland back closer to the Nile, with its speakers following the Sahel to the Niger.

However, that entire language family appears to be controversial now. The current Wikipedia page on the subject mentions alternate theories of it being related to the Berber Languages (which would make it Afro-Asiatic), or Mande (Niger-Congo). The obvious negative, that it is its own family, doesn't seem to have been mentioned.

It seems like if it belonged to one of the three main large language families native to Africa, it ought to be fairly obvious which one it is. There shouldn't be different people separately arguing for all three.

So anyway, what is the current consensus on this? Is it really completely and utterly up in the air like WP is implying, or is there are clear majority opinion?

Answer 1

This is a very small field, where one abstaining scholar prevents reaching a consensus. I think one should discount any one-off pre-Greenbergian affiliation claims which have not been re-affirmed, since they were generally not well-founded save for lowest levels (such as "Bantu", but not e.g. "Nilotic" insofar as it excluded "Nilo-Hamitic"). Dimmendaal's "independent family" position is the methodologically conservative one (disclaim any relationship if you're not positive what the relationship is). Bender maintained that it was Nilo-Saharan; so does Blench. Heath, who has descriptively put the Songhai languages on the map, has as far as I know not expressed an opinion on the matter (presumably because weighing in would be pointless speculation, and he's more interested in the facts). Nicolai took a variant of Dimmendaal's position ("Berber creole", essentially denying that it is part of any specific existing language family. The standard bacterial-reproduction model of language family assumes a continuous splitting from a single genetic source, whereas creoles have multiple parent, so that all three of the main African phyla may have been involved in the creation of Songhai, if the creole theory is right – even if the major contribution was from Berber). Nicolai also considers the question "open". Creissels supports Delafosse's (reputed) claim to a Mande connection.

Lexical items are easily borrowed across language phyla (and there is no systematic reconstruction of the claimed proto-language). There is relatively little grammatical structure proposed to be shared between NS and Songhai, and Creissels has actually amassed more morphological evidence for a Mande connection. By way of a summary evaluation, I agree with Heath's position.