As mentioned under 1.1., African linguistic diversity is much less than that exhibited by other
language areas. There are several possible explanations for this. The first might be that extant
classifications underestimate actual diversity. Ever since Greenberg (1963), the Africanist scientific community has by and large accepted his
classification of African languages in four large phyla : Niger-Congo (aka Congo-Kordofan or Niger-
Kordofan), Nilo-Saharan, Afro-asiatic and Khoisan. This classification was admittedly far from
revolutionary in its broad outlines, since the Khoisan phylum was more or less identical with the click
language group recognised at least since Bleek, Afro-asiatic was equivalent to the old Hamito-Semitic
family, to which Greenberg boldly added not only the "Chado-Hamitic" languages of Westermann
(long felt to be related to Hamito-Semitic, e.g. by the French semiticist M. Cohen), but also the
Chadic ("non-Hamitic") languages that Westermann wanted to keep separate from the others on
account of their lack of grammatical gender opposition. The other two phyla "Niger-Congo" and "Nilo-Saharan" were largely based on Westermann's
West- and Ostsudansprachen, with a few modifications, like moving the Songhai group (Mali, Niger)
into Nilo-Saharan, but most notably the provocative inclusion of Bantu languages as a sub-branch of
the Benue-Congo family within Niger-Congo (although Greenberg himself acknowledged that
Westermann tacitly supported this interpretation, while the French traditional Africanist school -
Homburger, Delafosse, etc. - considered Bantu and "Sudanic" languages as the two branches of a
"Negro-African" phylum). There does not seem currently to be any doubt left as to the inclusion of Bantu into Niger-Congo,
nor indeed about the cohesion of the latter, albeit with a great many disagreements about its internal
structure. The unity of the other three phyla however is still disputed, least of all Afro-asiatic, where
only the place of Omotic (South-Western Ethiopia), or indeed its inclusion within the phylum is still
open to queries, at the very least for some of the component languages. But most Khoisan specialists
do not now regard all click languages as being related : the inclusion of Hadza (Tanzania) is generally
rejected, whereas of the other sub-groups the only one to form an accepted genealogical grouping is
Khoe (which admittedly contains the largest number of languages), all other languages being potential
isolates.
But the most disputable phylum appears to be Nilo-Saharan : even setting apart the case of
Songhai, whose inclusion is rejected by many specialists, it is notable that constitution of the phylum
by Greenberg took place fairly late: in his original classification, published in the early '50's, he only
recognized an "Eastern Sudanic" group, included along with "Central Sudanic" and two smaller
language groups, Kunama and Berta, within a larger "Macro Sudanic" or "Chari-Nile" phylum. Only
in 1963 was the decision taken to join "Macro Sudanic" with "Central Saharan" (Kanuri, Teda,
Zaghawa), Songhai and three smaller groups of the Ethiopia-Sudan region (Maban, Fur and Koman)
which had been left isolated in the first version.
One of the most telling proofs of the problematic status of "Nilo-Saharan" is that two recent
attempts at reconstructing "proto-Nilo-Saharan" (Bender 1996 and Ehret 2001) end up with two very
different - in fact incompatible - internal classifications of the phylum. Even "Eastern Sudanic" which
should presumably prove most resistant to restructuring does not escape entirely unscathed:
Greenberg's "Teuso", (nowadays more generally called Kuliak, a remnant language group in eastern
Uganda) is taken by Bender outside of the"Eastern Sudanic" family altogether, whereas Ehret firmly
retains it (in fact many contemporary researchers would consider Kuliak an isolate).
If we now turn to those languages which most specialists would hesitate to classify or declare
outright to be isolates, we can propose the following list (cf. Blench, Maho, etc.)
- Hadza (Tanzania): Khoisan for Greenberg
- Sandawe (Tanzania): Khoisan for Greenberg
- Kwadi (Angola): Khoisan for Greenberg
- Jalaa (Nigeria): not mentioned by Greenberg, possibly Niger-Congo
- Laal (Chad): not mentioned by Greenberg, possibly Niger-Congo
- Kujarge (Chad): not mentioned by Greenberg
- Ongota (Ethiopia): not mentioned by Greenberg, possibly Afro-asiatic
- Shabo (Ethiopia): not mentioned by Greenberg, possibly Nilo-Saharan
- Gomba (Ethiopia): not mentioned by Greenberg, possibly Afro-asiatic
- Bangi-me (Mali): not mentioned by Greenberg
- Pre (Ivory Coast): not mentioned by Greenberg, probably Niger-Congo
- Irimba (Gabon): not mentioned by Greenberg
As mentioned above, the study of linguistic isolates should prove of particular interest in view of
establishing the existence of a former diversity in the African continent. By 'linguistic isolates', should
not only be meant those 12-odd languages impossible to classify with any certainty - which might of
course be partly due to defective information - but also the original stratum of languages which can be
classified - albeit with considerable hesitation as to their exact hierarchical position. A notable
example might be that of Dahalo, an undoubtedly Cushitic remnant language of coastal Kenya, which
has in its lexicon almost 100 lexical items containing clicks - these being of course unknown in other
Cushitic languages and unrelated to other click languages of Eastern and Southern Africa. One should
also keep in mind in this respect, the particular lexicon identified by Bahuchet in various pygmy
languages.
The first task then will consist in collecting and collating data on all known language isolates in
Africa as explained in 1.3. All our international collaborators, who are specialists in several of these
languages and have accumulated considerable data on them (e.g. Gueldemann for Khoisan, Mous on
Afro-asiatic, etc.) will naturally be involved in this task, but we will also contact other scholars with
first-hand experience of the other languages, in order to insure as broad a coverage as possible. This
joint cooperative effort will be carried on in the first year of the project, at the end of which an
international conference should come up with the new proposed classification. |