Niger-Congo Reconstruction: contributions to Paris 2004

Home

This page is intended to be a contribution to the debate about Niger-Congo reconstruction. The Santa Fe Institute is sponsoring a joint workshop with the LLACAN (CNRS), Paris on Niger-Congo Reconstruction Today. The conference will take place 12-16 October 2004, at CNRS, Paris, France. I am posting some papers here relevant but also commenting on a few issues.

1. You can download a paper that organises all Westermann’s Sudanic roots into an alphabetical series and marks the subgroups of Niger-Congo in which they are found. I also have columns for the groups he didn’t recognise. For example, I checked the Ijoid with Kay Williamson so we can say whether these roots are found there or not. I think Westermann’s observations clearly show the relevance of postulating a Mande-Congo node in Niger-Congo reconstruction.

2. I did a paper some time back that proposed some evidence linking Kordofanian with the rest of Niger-Congo, given that many of Greenberg’s original proposals don’t hold up. Kay Williamson put forward some similar glosses in her paper for the 1997 Leipzig Congress. Many of these, however, do not show Kordofanian is specifically Niger-Congo; ‘two’, three’, ‘five’ etc. turn up widely in Nilo-Saharan. My paper tries to remedy this problem although it is unfinished in places. Comment very welcome.

3. While compiling this I became concerned with some methodological issues. One of these concerns selectivity in choosing cognate roots. A family like Atlantic is so lexically diverse that you can pick and choose cognates to fit almost any case. ‘Elephant’ is a good example. I always vaguely accepted that Fulfulde nyiiwa must be cognate with the widespread -nyi roots in Benue-Congo and that ‘elephant’ should be reconstructed to Atlantic-Congo. But why should it? There are hardly any other cognates in Atlantic, which is anyway extremely internally diverse and –nyi is not such an uncommon segment after all. This is not the case with Mande, which has many glosses that can be tracked across the family. Perhaps we need to demand higher standards in tables of cognates.

4. When you are compiling sets of cognates in Niger-Congo families, one thing it is difficult not to notice is that Ijoid is actually the most deviant group. There seem to be many glosses that simply don’t have Ijoid cognates. I don’t know what to make of this. Did Ijoid break off very early, or was it influenced by unknown languages and replaced a number of key lexemes? Dogonic is also a problem, of course, but partly because the documentation is so weak. The Hochstetler study lists 18 lects most of which would count as languages on a lexicostatistic count, and one of which, Bãngere, is unlikely to be Dogon at all.

5. There’s something fairly inadequate about the characterisation of Kadu. If is Nilo-Saharan, how come it has a number of Niger-Congo roots, including some not in the other Kordofanian languages? I have posted my paper for the Nilo-Saharan meeting in Khartoum as a contribution to this issue.

6. I have also posted Kay Williamson’s Proto-Ijoid reconstructions, which makes available a mass of new data on this neglected group.

Title	Content	Status	Download
Westermann re-analysed		Unpublished	Westermann
Kordofanian	Glosses linking Kordofanian with Niger-Congo	Unpublished	Kordofanian
Kadu		Conference paper	Kadu
Kay Williamson; Proto-Ijoid reconstruction	Proto-Ijoid reconstruction	Unpublished	Ijoid

Top of the Document