Evolution of Human Languages

An international project on the linguistic prehistory of humanity
coordinated by the Santa Fe Institute
Areas of Research
Languages of the World: Etymological Databases
Interactive Maps
<< Home

Research Focus Area

Evolution of Human Languages

Classification Of African Languages

George Starostin, Russian State University for the Humanities

Since Joseph Greenberg's seminal work on genetic relationship between the languages of Africa, culminating in his 1963 monograph on the subject, very little progress has been done on the verification of his proposed model - elegant and promising on the surface, beset with all sorts of methodological and substantial problems on the inside. Greenberg's classification of nearly all known languages of the African continent into one huge (Niger-Kordofanian), two large (Nilo-Saharan, Afro-Asiatic) and one small (Khoisan) macrofamily has been generally accepted by Africanists as a "working model", but not so much because of its convincing verisimilitude as due to its pragmatic convenience to researchers - and because, with its huge scope and with its author's unparalleled coverage of the material, there simply was no alternative.

Today, as is the case with most of Greenberg's classification proposals, some parts of his African taxonomy stand better scrutiny than others, yet not a single one of his macrofamilies remains impenetrable to criticism. Afro-Asiatic, due to certain immaculate grammatical and lexical isoglosses than its members, fares best of all, yet grave doubts exist about its "Omotic" subbranch. Niger-Kordofanian is stable "at the core", but many of its "peripheral" branches display so many peculiarities of their own that they might as well constitute independent families. Worst of all is the situation with Nilo-Saharan and Khoisan macrofamilies, whose taxonomic reality is at present under serious doubt by most major Nilo-Saharanists and Khoisanists alike.

G. Starostin's research, originally centered around the Khoisan macrofamily, has now expanded to including data from the rest of Greenberg's macrofamilies as well (with the exclusion of Afroasiatic). The basic idea is to come up with an updated version of Greenberg's model, that would be improved in at least two respects: (a) rely on more up-to-date source materials, especially considering the wealth of new lexicographic data published over the last fifty years; (b) target, first and foremost, the "core basic lexicon" of African languages - some of the most stable items on the Swadesh wordlist, least susceptible to areal convergence (borrowings) and most convenient as "genetic markers" that are most likely to preserve traces of the original relationship pattern.

The methodological procedure employed for this purpose borrows from both the classic comparative method and Greenberg's "multilateral comparison", tempering the permissiveness of the latter with stricter demands on phonetics and semantics, as well as taking into consideration the areal distribution of the observed lexical forms and proceeding strictly "bottoms up", from intermediate low-level reconstructions to deeper levels. For Khoisan languages, such an approach has already shown that it is probably wiser to speak of two macrofamilies rather than one (although with possible further connections on a much deeper time scale). Main research is now conducted on the second most problematic of Greenberg's macrofamilies - Nilo-Saharan. All the major results are expected to be published in a large monograph within the next few years.