What is the current understanding of Greenberg's classification of African languages?

Question

In a reply to the criticism of his classification of the languages of the Americas, Greenberg (1989: 107) characterized his work on African languages as follows:

[...] my classification is clearly the basis of present-day African historical linguistics. There is no alternative classification, and the few proposals cutting across my four basic groups have received no general support. All disputes have been at the level of subgroupings, some of which I had said were tentative; and indeed some of the changes were suggested by me. When one considers that even today there is no unanimity regarding Balto-Slavic as a subgroup of Indo-European (IE), similar disputes among the much less studied African languages should come as no surprise.

Is this still a more or less accurate description of the status of his classification of African languages? I.e., has his broad classification been refined and received relatively minor improvements regarding subgroupings, or have there been major revisions or serious questions about the validity of (parts of) his classification?

Greenberg, Joseph H. 1989. “Classification of American Indian Languages: A Reply to Campbell.” Language 65 (1): 107–114. http://www.jstor.org/stable/414844

Answer 1

Greenberg's four major African language families are Niger-Kordofanian, Afro-Asiatic, Nilo-Saharan and Khoisan.

When he says that "...the few proposals cutting across my four basic groups have received no general support," this would be an accurate statement today, as it was in 1989: There haven't been any well-received proposals for language families which crosscut members of Greenberg's four families. e.g., there is nothing like a proposal for a family which groups together the Chadic languages (Greenberg's Afroasiatic) and the Benue-Congo languages (Greenberg's Niger-Kordofanian).

However, when he claims that "[a]ll disputes have been at the level of subgroupings," this would not be correct today, though it may have been in 1989. Concerning the first three families (Niger-Kordofanian, Afroasiatic and Nilo-Saharan), disputes about these have usually been about subgroupings: e.g., Atantic may not be a branch of Niger-Congo, etc. However, concerning Khoisan, disputes have not been about subgroupings, but about the coherence of the whole family itself. As for Khoisan, there is a dispute about the basic coherence of the family as a whole, and most Africanists no longer believe that Khoisan is a proper genetic unit, although the term "Khoisan" is still used to refer to something like "the click languages in southern Africa". For discussion on the problems of Greenberg's Khoisan, and new proposed groupings for the languages formerly classified as Khoisan, see Gueldemann (2008).

It might be added that of Greenberg's works on language classification, that of the African languages is the only one which has been successful as far as being adopted in a very similar form by the majority of practicing linguists today. His proposals for languages of the Americas and his Indo-Pacific hypothesis have never been truly taken seriously.

Answer 2

3 of the 4 macrophylums have been streamlined. Khoi-San is now Khoe-Kwadi, Kordofanian has been removed from Niger-Congolese, and some have removed Omotic from Afro-Asiatic.