The Linguistic Settlement of the Nuba Mountains*

by Robin Thelwall, The New University of Ulster and Thilo C. Schadeberg, Rijksuniversiteit te Leiden
From: Sprache und Geschichte in Afrika 5 (1983) 219-231

I. Introduction
The linguistic complexity of the Nuba Mountains (Southern Kordofan province, Sudan) has been known for over seventy years (Seligman 1910/11). Recent research into the detailed relationships of the forty-odd languages of this region has made it possible to speculate, with a lesser degree of uncertainty, on the patterns or migration and the time sequence implied by present distribution. Hypothesizing about the past situation in the Nuba Mountains involves balancing logical possibilities with circumstantial evidence.

An apparently elementary problem dogs the discussion at its outset: the name Nuba and its relatianship to Nubia. Arkell (1955.177-78) still provides the most succinct discussion:

"The earliest occurrence of the name Nubia or Nuba is in the Greek writer Eratosthenes c. 200 BC, who mentions the Nuba as being on the west of the Nile 'as far as the bends of the river'. This should mean as far as the Dongola Reach. The name of the Nuba apparently comes, like so many other tribal names in the Sudan (Berti, Berta, Burgu, etc-) from a word in their own language which means 'slaves'; and it is not impossible that the ancient Egyptian word nub for 'gold' arose from the fact that this metal came to them first from their southern neighbours whom they looked on as slaves.

The name Nuba today usually implies an inhabitant of the Nuba Mountains in southern Kordofan inhabited by remnants of peoples of varying language and race, the majority of whom are now negroid. The Nuba Mountains were probably so called after the Brown race of Nubian-speaking immigrants from the steppe country further north, from which they were displaced by nomad Arabs about the 14C AD. The name 'Kordofan' for this steppe country probably comes from a Nubian word kurta meaning 'man' -"

It should be noted that even today people in the Nuba Mountains use "Kordofan" to refer to El Obeid and the country round it and not to refer to the Mountains.

The ethnically and linguistically fragmented situation indicates that the Nuba Mountains have in all probability served as a retreat area. This may have happened at various times in history for basically two reasons. The first would be climatological: the desiccation of the Sahara has certainly time and again impelled people to migrate in search of more abundant water, either to remaining rivers and lakes, or just more generally southwards. In a more close-up perspective the driving forces are on the whole politico-economical. Looking at the map it is not difficult to find the areas of power concentrations from which people might have found it expedient to seek refuge. To the north, the Sahel empires have succeeded each other for centuries. The period of intensive slave-raids over a century ago was a severe threat and a bitter experience for the southern neighbours of Kordofan (i.e., El Obeid). South of the Nuba Mountains, the large and compact area occupied by Dinka and Nuer speakers also has the appearance of a relatively recent centre of expansion. Therefore, we should not overlook the possibility that some present-day inhabitants of the Nuba Mountains came there from the south.

2. The Evidence
The ten language groups established by the MacDiarmids (1931) can serve us well as a point of departure. They are lexicostatistically definable as having an internal cohesion of not less than 45%, based on a 100-word list. The groups and their internal distances (rounded off to multiples of five) are:

1. Heiban 45%
2. Narrow Talodi 65%
3. Tegem (=Lafofa) -
4. Rashad 55%
5. Katla 50%
6. Kadugli 60%
7. Nyimang 60%
8. Temein 60%
9. Daju 60%
10. Nubian 85%

Stevenson (1956-57) summarizes all significant research on Nuba Mountain language to that date, and is also based on his immense collection of manuscript data, largely unpublished, though summary extracts form the main basis for the relevant sections in Tucker and Bryan (1956, 1966). Those publications also contain detailed maps of distribution on which our Map 1 is based. Thelwall (1978, 1981a,b) and Schadeberg (1981a,b) provide subclassifications of most language groups, which will be summarized further on.

The first eight groups listed above are confined to the Nuba Mountains; Daju and Nubian are the only ones that have close connections outside. Our argumentation makes critical use of such information. However, the fact that Nyimang, Temein, Daju and Nubian have all been classified - together with Nilotic and several other language groups - as Eastern Sudanic has no consequence in our present context. Different branches of Eastern Sudanic are very distant from each other; they generally share less than 20% in lexicostatistic terms. Since there are no indications that the Nuba Mountains were the original home of Eastern Sudanic such distant genetic links are judged to be unconnected with the appearance of those four language groups in the Nuba Mountains.

The affiliation of Kadugli is presently open (see Schadebert 1981c). Again, nothing in our argumentation depends on remote possible links with Nilo-Saharan. On the other hand, the fact that groups (1) through (5) may be classified as Kordofanian, and that all Kordofanian languages are spoken exclusively within the Nuba Mountains is certainly relevant. Our hypotheses about the relative chronology of the influx of the various groups are based on these three types of clues about each language group: (i) internal diversity, (ii) immediate external genetic links, and (iii) geographical distribution. Naturally, other evidence such as loanwords and historical traditions should be taken into account as they become available.

3. Hill Nubian
Nubian is a language group which presently spreads from Darfur to the Nile (see Map 2). The most prudent interpretation of our lexicostatistical data (Thelwall 1978, 1981a) leads to the subclassification shown in figure 1.

Figure 1: Subclassification of Nubian

We can propose with some confidence that the centre of gravity and hence centre of dispersion lies outside the Nuba Mountains, perhaps in the Darfur-Kordofan boundary region. We have documents in Old Nubian from the Nile region south of the first cataract (Aswan) dating from perhaps the 7C AD and good reason for relating the Old Nubian of these documents more closely to present-day Nobiin (Mahas) than to any other present variety of Nubian. Behrens (1981) has even proposed a date as much as two thousand years earlier for the presence of Nubian speakers in this area, based on proposed Nubian loans in Egyptian. We also know that the late medieval Nubian kingdom of Alwa was in control of the Nile south of Soba (10 km south of Khartoum on the Blue Nile). Place name evidence (based on Survey Department maps made from observations at the beginning of this century) shows that Nubian names occur on the Nile and nearby as far south as Kosti, but how far back this presence goes is an open question. We can be confident it predates the Arabization of the Gezira, and so is probably not less than 500 years, but it may weIl be over 2000.

The indications mentioned above would point to a movement of pre-Nobiin speakers across the Bayuda desert to the Nile as long ago as 2000 years, and perhaps a parallel movement to the Nile around Kosti of pre-Dongolawi speakers, with their subsequent spread north to later link up with the Nobiin. The tribal map of the present day marks "Nubawi" over the area east and northeast of El Obeid, and it may be that Nubian speech has only disappeared in this area in the last hundred years (see Bell 1973).

We may thus assume a large zone from Darfur in the West to the Nile in the East in which Nubian was present during a long period. From this area some Nubians must have come to settle on the northern Nuba Mountains. Whether this occurred due to pressure from Arab nomads as Arkell (1955) proposes, or whether an earlier date should be assumed is not clear. The relative closeness of the Hi1l Nubian dialects to each other does not suggest the presence of isolated Nubian communities in these hills for several millennia.

Likewise, the linguistic subclassification of Nubian lends no support to the belief that Hill Nubians are refugees from the former Christian Nubian empire on the Nile. This theory derives some of its popularity from the fact that it appears to lend part of the age-old prestige of the Nile civilizations to present-day hill dwellers. The presence of Christian words for the days of the week in Hill Nubian could be explained by assuming past contacts between Christian Nubia and the Hill Nubians.

4. Daju
For the Daju also we have good linguistic evidence and scanty but cogent historical tradition. Languages of the Daju group are presently spoken in Wadai, Darfur, the Nuba Mountains and Bahr el Ghazal (see Map 2). The subclassification in Figure 2 emerges from the work of Thelwall (1978, 1981a,b).

Figure 2: Subclassification of Daju

Thus, the focal point of Daju prehistory appears to lie in Darfur. It is known that the Daju ruled in Darfur before the Fur and Tunjur, i.e. perhaps as early as the 14C AD. The centre of their kingdom seems to have been the southern part of the Marra range, but there are traces of them much further north. The Meidob Nubians have a clan that is supposed to be descended from Daju slaves. They may of course represent groups that were assimilated into the Nubians much earlier than the rise of Daju power, when the Nubians were almost certainly dominant in the area where the Birgid are today, which is next to the Daju. But both possibilities should be considered.

The explanation of the presence of three TJaju groups in the Nuba Mountains is affected by the linguistic subgrouping, which shows a clear division between Shatt and Liguri on the one hand and the remainder of Daju including Lagawa on the other. The Lagawa, and their close linguistic kin the Nyolge or Nyalgulgule of Bahr el Ghazal seem clearly to have moved to their present positions from southeast Darfur, most likely as a result of the decline of the Daju kingdom, i.e., after the rise of the Tunjur perhaps in the 14C. (For a succinct summary of Darfur history at this period see Balfour-Paul 1955 and O'I'ahey and Spaulding 1974).

The Shatt and Liguri, however, because of a number of distinct lexical, phonological and grammatical features (see Thelwall 1981 b), must be assumed to have separated from the rest of Daju much earlier than the Lagawa. Furthermore, their considerable distance from each other leads us to propose a migration into the Nuba Mountains predating not only the Lagawa (and Nyolge) but probably also the Nubian arrival in this area.

5. Nyimang; Temein; Kadugli
Nyimang and Temein are two small language families, each consisting of two or three languages, all spoken exclusively in the Nuba Mountains. They have been classified as two (out of ten) branches of Eastern Sudanic. Genetic relationships within Eastern Sudanic are too distant - and too uncertain - as to permit any inferences about migration at the time depth with which we are here concerned. We can only note that both groups show an internal divergence of about 60% (see Figure 3). The data are taken from Thelwall (1981a).

Figure 3: Subclassification of Nyimang and Temein

The Kadugli group shows a very similar internal diversification. It consists of at least six languages (and several more dialects), and it is also not represented outside the Nuba Mountains. The subclassification in Figure 4 is based on a preliminary lexicostatistic hierarchical cluster analysis of nine wordlists (92 items). The proposed tree holds equally for all clustering methods including Nearest and Furthest Neighbour; the absolute height of branching follows the Branch Average figures.

Figure 4: Subclassification of Kadugli

Note that Stevenson's division into Eastern, Central and Western Kadugli is only in part borne out by this calculation; in particular, his Eastern division consisting of Keiga, Kamdang and Kanga/Kufa appears to be non-coherent.

AIl three groups here discussed have a reasonably compact distribution within the Nuba Mountains: Kadugli along the southwestern edge, Temein to the West, and Nyimang to the north. This suggests outside origins and immigration from these respective directions. Assuming that equal internal diversity corresponds to some roughly consistent time depth we may argue that at this particular time in history conditions prevailed in the Nuba Mountains which resulted in population scattering and reduced inter-group communication. As it is more likely that such conditions originated outside the refuge area we may further speculate that migration to the Nuba Mountains and diversification occurred in close historical union.

6. Kordofanian
This leaves us with the five groups representing four branches of Kordofanian (see Figure 5).¹

Figure 5: The major branches of Kordofanian

The outside relations of Kordofanian are too distant to be relevant in the present context. The whole Kordofanian language family is located within the Nuba Mountains where it occupies the most central and most widespread geographical position (see Map 1). There appears to be a continuous history of branching, beginning with a (presently assumed) four-way split into Katla, Heiban, Talodi and Rashad. This primary split must have preceded the subsequent split of Talodi into Tegem and Narrow Talodi (25%). On the basis of this evidence it is clearly indicated that the development of Kordofanian occurred in the Nuba Mountains, and that Kordofanian has the longest linguistic history in this area.

The resulting hypothesis regarding the relative chronology of the linguistic settlement of the Nuba Mountains is then:
1. Kordofanian
2. Nyimang; Temein; Kadugli
3. Daju I: Shatt, Liguri
4. Hill Nubian
5. Daju II: Lagawa

Map 1: Language distribution in the Nuba Mountains


Map 2: The distribution of the Nubian and Daju language groups


* The present article arose out of a paper by the first author and a critical reply to it by the second author, both presented at the Second Nilo-Saharan Linguistics Conference at Coleraine, Northern Ireland in July, 1983.
¹ The internal cohesion of Kordofanian as well as its relatedness to Niger-Congo are presently almost universally accepted. A number of noun class prefixes of Kordofanian have clear cognates in several branches of Niger-Congo, thus providing evidence for the genetic relatedness of the two language families. (These two statements were part of my contribution "Kordofanisch" to Lexikon der Afrikanistik, ed. by H. Jungraithmayr and W.J.G. Möhlig, Berlin 1983, though they do not appear in full in the published version. TCS.)

Arkell, A.J. 1955. A History of the Sudan from the Earliest Times to 1821. London.
- Balfour-Paul, H.G. 1955. History and Antiquities of Darfur. (Museum Pamphlet 3.) Khartoum.
- Behrens, Peter. 1981. 'C-Group-Sprache-Nubisch-Tu Bedawiye', Sprache und Geschichte in Afrika 3:17- 49.
- Bell, Herman. 1973. 'An extinct Nubian language in Kordofan', Sudan Notes and Records 54: 73-80.
- MacDiarmid, P.A. and D.N. 1931. 'The languages of the Nuba Mountains', Sudan Notes and Records 14:149-62.
- O'Fahey, R.S., and J.L. Spaulding. 1974. Kingdoms of the Sudan. London.
- Schadeberg, Thilo C. 1981a. A Survey of Kordofanian. Vol. 1: The Heiban Group. (Sprache und Geschichte in Afrika, Beiheft 1.) Hamburg.
- Schadeberg, Thilo C. 1981b. A Survey of Kordofanian. Vol. 2: The Talodi Group. (Sprache und Geschichte in Afrika, Beiheft 2.) Hamburg.
- Schadeberg, Thilo C. 1981c. 'The classification of the Kadugli language group', in Nilo-Saharan, ed. by Th.C. Schadeberg and M.L. Bender, pp. 291-305. Dordrecht.
- Seligman(n), Brenda Z. 1910/11. 'Note on the language of the Nubas of Southern Kordofan', Zeitschrift fur Kolonialsprachen 1:167-88.
- Stevenson, Roland C. 1956-57. 'A survey of the phonetics and grammatical structure of the Nuba Mountain languages, with particular reference to Otoro, Katcha and Nyimang', Afrika und Ubersee 40:73-84, 93-115; 41:27-65, 117-52, 171-96.
- Thelwall, Robin. 1978. 'Lexicostatistical relations between Nubian, Daju and Dinka', in Etudes Nubiennes: Colloque de ChantiLLy, 2-6 Juillet 1975, pp. 265-86.
- Thelwall, Robin. 1981a. 'Lexicostatistical subgrouping and lexical reconstruction of the Daju group', in Nilo-Saharan, ed. by Th.C. Schadeberg and M.L. Bender, pp. 167-84. Dordrecht.
-The Daju Language Group: Systematic Phonetics, Lexicostatistics and Lexical Reconstruction. D. Phil. diss., New university of Ulster. Coleraine.
- Tucker, A.N., and M.A. Bryan. 1956. The Non-Bantu Languages of North-Eastern Africa. London.
- Tucker, A.N., and M.A. Bryan. 1966. Linguistic Analyses: The Non-Bantu Languages of North-Eastern Africa. London