NOTES ET DOCUMENTS HAIG DER-HOUSSIKIAN University of Florida The Evidence for Niger-Congo Hypothesis The purpose of this paper is to evaluate the evidence for Niger-Congo hypothesis with reference to the position of Bantu presented by Greenberg first in the Southwestern Journal of Anthropology 1949-50) then in his Studies in African Linguistic Classification 1955 and most recently in The Languages of Africa 1963 1966) The evidence is evaluated in terms of historical precedents comparative linguistic methodology and the particular claims of the hypothesis itself An evaluation specially within the context of recent and current historical research is necessary because most current statements concerning African lin guistic relationships assume the Niger-Congo framework take that framework as given and proceed from there hypothesis is undoubtedly very strong one But it is based on no stronger or more specific and less subjective evidence than earlier references to Bantu and Sudanie relationships by Johnston 1919-1922 and Westermann 1927) major characteristic of Bantu classificatory literature from Comparative Grammar of South African Languages in 1862 to publica tion of 1963 has been the absence in varying degrees of clear distinction between typologically and genetically based classifications In spite of refer ence to method as primarily typological because the latter paid no attention to any factor other than the presence or absence of noun prefixes Greenberg 1966 Greenberg nevertheless belongs in camp The difference between these two men is not so much matter of methodology as it is one of approach Meinhof did not discover the relationships of the Bantu languages by way of his correspondences and reconstructions He assumed or reiterated the relationships based on and 1891 typological observations and went on from there In the case of Greenberg however certain typological criteria are admittedly used for positing hypotheses of genetic relation ship When resemblances can be assembled which are recurrent in many languages which extend over vast and widely separated geographical areas and which encom pass elements with morphological functions pronouns and the m.ost stable parts of the vocabulary then common origin is the only adequate explanatory hypothesis Greenberg 1966 2) Those recurrent resemblances refer to nothing other than the trait of Niger- Congo morphology which provides the main material for comparison the system of noun classification by pair of affixes one for the singular another for the plural Greenberg 1966 9) In his introduction to the first of two volumes of his Comparative Study of the Bantu and Semi-Bantu Languages Johnston makes the following statement