Linguistic Evidence for the Introduction of Ironworking in Bantu-Speaking Africa
I
Did Africans once independently invent the smelting of metals or did they obtain this technology from Europe or the Middle East? This continues to be an unresolved and hotly disputed issue, mainly because the dates for the earliest appearance of smelting in Africa south of the Sahara remain inconclusive. All the earliest sites in Western and West-Central Africa from Walalde in Senegal to the Tigidit cliffs and Termit in Niger, the firki plains south of lake Chad, Taruga, and perhaps Nsukka in Nigeria, Ghwa Kiva (Nigeria), and Doulo (Cameroon) in the Mandara mountains, Gbabiri (Ndio district) in the Central African Republic, and a few sites in Rwanda, Burundi, and Buhaya cannot be dated more closely than between 840 and 420 BCE. Greater precision is impossible because the C14 curve runs flat during these four centuries, hence all these sites yield the same date. (Alpern, Killick, Mc Eachern, Holl, Jézégou/Clist, Kanimba Misago). If the earliest "real" dates fell before 800 BCE, they would support independent invention, while later dates strengthen the case for borrowing. Still, this information does tell us that ironworking was adopted in the northern parts of West and West -Central Africa and in the region of the Great Lakes within the span of a mere four centuries.
The emergence of ironworking must have left linguistic traces in the relevant terminology irrespective of whether it spread by borrowing or by independent invention—hence historical linguistics can contribute to this debate. That approach is best tested by an examination of the relevant vocabulary in Bantu languages because the historical study of those languages is further advanced than that of any other language family in [End Page 321] Africa (Nurse/Phillipson). Moreover Bantu-speakers occupy a large portion of the continent.
II
Thirty years ago François Nsuka and Pierre de Maret (de Maret/Nsuka; Nsuka/de Maret) established that no single word relating to metallurgy in any Bantu language could confidently be attributed to Proto-Bantu (hereafter PB) and hence they argued that PB-speakers did not know how to work iron; that skill was acquired only after PB had begun to split and its daughter languages were spreading over the sub-continent that they now occupy. Nevertheless, they continued tacitly to assume that ironworking diffused only once and from a single cradle throughout most of the Bantu-speaking area and all subsequent scholars with the exception of Klein-Arendt have accepted this stance.
In an essay originally written in 1973, but updated later and published only recently, Christopher Ehret (2003) reached the same conclusion, and went on to claim that within the Bantu-speaking world ironworking first appeared in the Great Lakes region and diffused from there all over the subcontinent. He further argued that the ironworking vocabulary in the cradle region itself was probably derived from Central and/or East Sudanic languages and hence that ironworking probably diffused from the Nile valley. This essay is seriously flawed, mainly because his database is deficient, especially for languages other than Eastern Bantu, and his assertions are often not supported by the data cited.
Meanwhile, in 2004 Klein-Arendt published a massive set of words about iron working in "savanna Bantu," i.e., the Bantu languages of eastern, southern, and east-central Africa as part of an ongoing study to document all the Bantu "migrations" and "cultural" diffusions that have succeeded each other across this area from the onset until the very recent past. With regard to the advent of ironworking, he cautions that the huge lacunae that exist in the documentation of adjacent non-Bantu languages prevent us from making definitive assertions, but still concludes that there were very likely one or several initial introductions of ironworking from West Africa, as well as one or several more from Northeast Africa (Klein-Arendt 257). Still it seems to me that the addition of...
