Talk:Atlantic–Congo languages

Nomenclature

@Wojak6, Greenwhitedino, Florian Blaschke, Kwamikagami, and Sagotreespirit: I can't escape the impression that something is wrong with the way how we present the Niger-Congo languages here. In the main article, we correctly say that the maximal version of Niger-Congo is disputed, and that "Atlantic-Congo" represents the undisputed core of "Niger-Congo". Yet, all information about major languages, internal classification and typology are in the article about the controversial construct "Niger-Congo", whereas this article about a solid language family is a pitiful stub. Even the statement about the third-largest language family by number of speakers is in the "Niger-Congo" article, not here, even though the "Atlantic-Congo" core still appears to have way more speakers than the Austronesian languages.

There are two approaches I can see to handle this:

1. We could expand this article by moving information about internal classification, typology (especially noun classes) here, only leaving the discussion about the inclusion of Mande etc. in the Niger-Congo article.
2. Alternatively, we keep Niger-Congo as the main article for the bulk of the solidly included branches (= "Atlantic-Congo"), present it as an established language family, and point out that the extent of Niger-Congo is disputed (followed by a short discussion of Mande etc.).

The first solution would be in line with Glottolog, but the second one actually better reflects the nomenclature of most Africanists. Virtually all of them use "Niger-Congo" for what they maximally see as a proven (or at least promisingly proposed) entity. Dimmendaal's Niger-Congo exludes Mande, Ubangi and Kadu; Güldemann only accepts the scope of "Atlantic-Congo" as valid, but refers to it as "Niger-Congo" (or "Niger-Kordofanian"). AFAICS, the term "Atlantic-Congo" has been mostly employed by scholars who accept the maximal version of "Niger-Congo" that includes Mande and all Kordofanian groups. Hammarström's use of "Atlantic-Congo" in Glottolog as term for the highest valid clade is quite insular, and should not be imitated here. –Austronesier (talk) 14:36, 22 March 2021 (UTC)

I recently discussed this issue with one of the premier Mande scholars, and he says that, while he accepts NC as a working hypothesis, it won't be establishable until we have reconstructions of Mande and Atlantic-Congo, neither of which we have at present. So it is true that NC is the more common clade in the lit, it is not an established language family. It is thus similar to the other two high-level clades promoted by Greenberg: NS and KS.

The problem with treating it as valid but disputed in extent is that Mande is the "Niger" part of NC. An analogy would be if it had yet to be demonstrated that the Formosan languages were AN, but we used AN rather than MP for the remainder. Or used "Australian" for PN, and discussed how it wasn't established that any of the non-PN families were related. That doesn't seem to me to be a helpful approach.

I would therefore prefer your first approach: Discuss the validity of the NC hypothesis in the NC article, and discuss the bulk of the languages -- including the characteristic noun-class system -- here. — kwami (talk) 21:47, 22 March 2021 (UTC)

My only worry is recognizability. Almost no-one calls this thing "Atlantic-Congo", except for lumpers who treat it as a subgroup of NC. The only splitter I am aware of is Hammarström. So most take the easy second approach: use the best known name, and only include what's sound, like Dimmendaal (2011) (and later works), Creissels (2014) who continue to use "Niger-Congo", but exclude Mande, Ijoid, Dogon, and Ubangian. It's not really unusual: there's Na-Dene, which remains ND even after Haida is removed; Dogon minus Bangime; Sino-Tibetan minus Kradai; or TNG, where virtually everyone has their own idea about its scope. And the lower Niger still flows through "Atlantic-Congo"-speaking lands.

But OTOH, the first approach is definitely more structured and less confusing for readers. Judging from drive-by edits, many don't make it past the infobox and the third lead sentence. A presentation as in Good (2020) (which inspired my idea for the second approach) is technically the best way to handle it, but not the most accessible one. If no-one really minds, I think we just implement the first solution. –Austronesier (talk) 14:09, 24 March 2021 (UTC)

@Kwamikagami and Austronesier: Niger-Congo classification is messy and much of it is still a mystery, like Trans-New Guinea and Sino-Tibetan. Only some branches have seen solid work done by Africanists, so we need to use what we have. Even though Hammarström is not a full-time Africanist, many of his insights are still valuable enough to be included here. Skimming through the various recent surveys on African language classification, I don't see any kind of opposition to "Atlantic-Congo" or other similar ideas, and in fact Güldemann (the editor of some African language classification chapters that Hammarström has written) and others appear to give implicit support to Atlantic-Congo in their recent publications. That being said, African linguistics is not really my area of expertise, but I can tell that this is similar to the Dene-Yeniseian / Na-Dene / Athabaskan issue where delineating specific families can get somewhat confusing. It's best to include all of these terms so that we can clarify ambiguities and mention notable proposals. — Sagotreespirit (^talk) 18:43, 27 March 2021 (UTC)

African historical linguistics is also full of "friendly lumpers," unlike Amerindian linguistics. You can propose virtually any new classification scheme of African languages and not be ostracized as a quack, as long as you're not pushing some kind of conspiracy theory about monogenesis from Semitic languages or Black Egyptians. As a result, there is highly unlike to be any kind of opposition or explicit refusal of recognition from the Africanist community regarding the "Atlantic-Congo" idea. But I do think that we need to mention the origin and history of "Atlantic-Congo" in more detail. Is it a recent Glottolog neologism, or does it have a lengthy history? — Sagotreespirit (^talk) 18:50, 27 March 2021 (UTC)

The term is 50+ years old. Williamson (1989), which was (and perhaps still is) the most widely accepted general classification of NC (Olson 2004 'An Evaluation of Niger-Congo Classification'), divides NC into Mande, Kordofanian and Atlantic-Congo. The names and branching date back at least to her 1971 classification. 'Bantu-Guinean' is an older term (not hers), but is obsolete.

Williamson (1989) says, "there is a post-Greenberg consensus that the languages in ... Atlantic-Congo form a unity, even though there is no complete agreement on either the higher or the lower nodes of the family tree."

Bennett & Sterk (1977) use 'Niger-Congo' for the current topic, but only because they retain 'Niger-Kordofanian' for what is now almost universally called 'Niger-Congo' -- that is, Bennett & Sterk's 'Niger-Kordofanian' consists of Kordofanian, Mande and "Niger-Congo proper". They use the dab "proper" repeatedly in their article, in which they "do not reject" the possibility of a relationship among Kordofanian, Mande and NC proper. 'Niger-Kordofanian' is of course now obsolete. Several speak of "core NC" for this clade. I'm not aware of sources that use 'Niger-Congo' without disambiguation for the scope of this article, whereas 'Atlantic-Congo' is used that way without disambiguation.

I see where Dimendaal says the scope of NC is unclear, and e.g. excludes Ubangian, and differentiates "genealogical NC" from "referential NC", but could you cite where he uses the term 'NC' without a dab for this smaller clade, excluding Mande? — kwami (talk) 20:04, 27 March 2021 (UTC)