Reading Güldemann's recent ~400-page handbook review on the classification of the languages of Africa; so far seems like fair treatment & in decent detail. Already the introduction has good underlining of basic methodological points like
- where substantial shared material exists, one should default to a relationship and not propose mass loaning / creolization / "mixing" without further arguments
- correspondence sets without proper reconstruction can be almost as important for tracing external relationships as fully reconstructed proto-forms
- not every language has numerals and thus not every relationship involves them either
- lack of proper documentation is an obstacle for proper classification
Finished reading after a week and a half, and my recommendation stands! A well balanced review: makes the basics of the field clear for newcomers, does not stray into either speculation or "it hasn't been done so it can't be done" skepticism, some minor engagement with primary data like a few key vocabulary items identified as Niger-Congo membership markers, often marks out areas where more work is definitely needed.
The "domains" of Khoisan (13 pages) and Niger-Congo (128 pages) are discussed in the best detail I think. The latter especially makes it clear how, between the various perhaps spurious units, many are only really spurious as subgroups but their members belonging in the family overall still looks likely, e.g. Adamawa, Ubangian.
To throw out one follow-up idea, this also suggests to me interesting new possibilities on the origin of Niger-Congo. The traditional classification of the family "starts" in the west, with Mande, Dogon and Atlantic branching early and splits then proceeds eastwards / clockwise around Gulf of Guinea towards Bantu; if so it would be very mysterious how some really early branches would have ended up in Kordofan instead. However if Mande & Dogon are now excluded and the components of Atlantic cannot be definitely shown to be early splits, and the Adamawa / Ubangi area also abounds in small language groups only weakly established as NC members — maybe it's instead there in the northeast that we have some early-splitting branches, with later areal convergence producing some superficial appearence of large subgroups. And from a homeland somewhere in this direction, it would be then already more reasonable to then find a few other early-diverging branches in Sudan. (Or indeed even further; Güldemann takes even the very speculative concept of "Niger-Saharan" surprizingly seriously and, with full Nilo-Saharan basically obsolete now, there would be in principle openings also there for even more distant relatives to the family. Getting ahead of things here though.)
The treatment of the also large Nilo-Saharan "domain" ends up necessarily less in-depth (75 pp.) due to almost everything about this being still in the air. There are good reviews especially about the state of the large subunits Central Sudanic (solid, needs more reconstruction especially in its eastern end) and Eastern Sudanic (several likely connections in there, the idea fully panning out is not yet a given). Many of the other small units are left basically at mention that they exist and may not be documented too well. (George) Starostin's interesting basic-lexicon survey from some years back gets no mention unfortunately, even though he has many of the same conclusions as Güldemann, and though he does not generally shy away from citing also only preliminarily circulated manuscripts if particularly relevant.
Afroasiatic is also covered in less detail (38 pp.), now for somewhat opposite reasons: it's fairly solidly established and has proportionally lots of specialist literature on the overall family already, so there's less about overall classification to even review. The history of its overall assembly is probably good for context but, unlike in the other cases, maybe doesn't have too much to contribute to understanding the current situation. Finer attention to open questions within Chadic, Cushitic and Omotic would have been nice, but perhaps there is indeed a better place and better writer for those elsewhere.
The title of the paper still is "Historical linguistics and genealogical language classification in Africa" though, and it might have been nice to underline better that former does not end once the latter has been established. Classification clearly takes front seat here and a couple issues of historical linguistics (e.g. any proto-cultural reconstruction, engagement with neighboring disciplines like archeology and genetics) get sidelined entirely. Maybe it's for the better, across a whole continent there would be easily some additional 400 pages to be written on each these!; don't get your hopes too high.
The whopping 83 pages of bibliography lastly seems like a very good starting point for branching more into particular issues, although obviously also this is only the peak of the iceberg.
