Sudanic Africa

Text and sources archive: 1

The Nuba Mountains:
Who Spoke What in 1976?

by Herman Bell

A Communication presented to the Third Conference on Language in Sudan
(Language Situation in Sudan)
at the
Institute of African and Asian Studies
University of Khartoum
3-5 December 1995

Being a study of the published results from a major project of the
Institute of African and Asian Studies:
the Language Survey of the Nuba Mountains

This report will focus on a 1976 language survey of the Nuba Mountains. The area is noted for its extreme linguistic diversity. It is located in the geographical centre of the Sudan 200 km west of the White Nile (10 deg.- 12 deg.30'.N, 29 deg.- 31deg.E). The survey will be compared with a parallel study of the Ingassana Mountains in the eastern Sudan (11 deg.20'.N, 34 deg.E).

In early 1976 the Language Survey of the Sudan was conducted in the Nuba Mountains within an area of 50 000 km2. The field work was accomplished by staff members of the Teachers' Training Institute in Dilling under the direction of a team from the Institute of African and Asian Studies of the University of Khartoum. They surveyed 29 localities throughout the Nuba Mountains. A similar sample (# 30) was collected in the Ingassana Mountains. The results from each locality were published individually in 1978 and 1979.

Results from the individual samples have long needed to be compared and assessed as a group. Already in 1979 a report was planned to give comprehensive coverage to the results of the survey (see item A - 4 in the appendix: List of Publications below). The present study is only one step in that direction. The eventual goal will be an integration of the results with a view to providing adequate answers to the questions posed by the survey.

There were seven main topics of inquiry. This report will deal only with the first topic: Languages Spoken.

  1. Languages Spoken: What was the language profile of each community? What were the patterns of multilingualism? To what extent did people use Arabic as a lingua franca?
The remaining six topics raise issues which are integrally related to the information in Languages Spoken. The remaining topics are summarized below.
  1. Languages Needed for Communication: What languages were needed to reach the whole community? Who did not speak Arabic? Old women? Young children?
  2. Inter-ethnic Patterns of Communications: How many people spoke the language of another ethnic group?
  3. Language Dynamics: What evidence was there that certain languages were declining in use or being abandoned?
  4. Context of Languages: Which languages tended to be used more frequently at home? Which were used more frequently in the market place?
  5. Literacy: What proportion of the respondents could read Arabic or other languages?
  6. Languages and Education:

Magnitude of Survey

The investigators recorded a total of 8453 respondents; 352 of these were in the single sample from Ingassana and the remaining 8101 appeared in the 29 Nuba samples, i.e. an average of 279 respondents per sample in the Nuba Mountains.

The Nuba Mountain survey detected 70 languages; 38 of these were spoken by 2% or more of a sample. In the Ingassana sample only 2 languages were attested in addition to Arabic; all three of these were spoken by more of 2% of the sample.

The results from Ingassana were not strictly comparable with those of the Nuba Mountains. The Ingassana investigator concentrated on rural settlements which were ethnically Ingassana, but the Nuba Mountain investigators were trained to follow procedures which were intended to achieve a representative coverage for the whole geographical area.

Languages Spoken: Monolingualism and Multilingualism

Most of the respondents in the survey spoke at least two languages. This was the normal pattern in the Nuba Mountains and also in Ingassana.

In only 3 of the 30 samples was there a substantial majority of people who spoke only one language. These 3 localities on the northeastern periphery of the Nuba Mountains (### 11, 19 and 22) were inhabited predominately by Arabic monolinguals. The largest proportion of monolinguals in a non-Arabic language was recorded in the Ingassana Hills where one-third of sample # 30 spoke Ingassana alone. There was a similar situation in two localities of the southern Nuba Mountains where over one quarter of sample # 2 spoke Krongo alone and just over one quarter of sample # 1 spoke Masakin alone. Elsewhere the proportion of monolinguals was progressively lower. In localities # 13 and # 30 Arabic monolinguals accounted for only 1% of the sample.

In the vast majority of cases, when a person spoke two or more languages, one of them was Arabic. Occasionally there were individuals who spoke two languages, but not Arabic. These individuals appeared in one quarter of the Nuba Mountain samples and in Ingassana (### 1, 9, 13, 20, 23, 25, 26 and 30), but nowhere did they represent more that 2% of the sample. Only local languages were involved. There did not appear to be any serious rival to Arabic as a lingua franca.

There were many examples of persons speaking three languages. Arabic was always one of the three. The other two were normally African languages, although in two cases the third language was English (# 11 and # 17). Every sample except # 6 yielded evidence suggesting multilingualism in three languages (trilingualism).

As used here, the term 'multilingualism' indicates only that a person has claimed to speak several languages. It does not suggest the degree of proficiency. Patterns of multilingualism are displayed below in the appendix of information sheets entitled '1976 SAMPLE OF LOCALITY'.

Multilingualism has normally been displayed here in terms of three major languages. In one sample only two major languages were available (# 22). Occasionally four languages were present in large numbers and called for a display in terms of four major languages (### 13b, 14b and 21b). In another sample (# 4) the category 'speakers of Arabic' conceals a number of bilinguals in Arabic and Talassa. It might have been preferable to display Talassa there as a fourth major language. Often the category 'speakers of Arabic' conceals a small number of bilinguals who spoke something that was not a major language of the sample. Furthermore, the charts do not indicate extreme situations such as five languages or more being spoken by one individual. Usually, however, the pie chart with three major languages approximates the actual situation and is relatively easy to understand as a graphic display. Multilingualism must be conveyed in a form which is not confusing.

Two-fifths of the localities showed a substantial degree of trilingualism, 'substantial' being defined here as 5% of the sample or more (### 3, 5, 8, 9, 12, 13, 14, 17, 23, 25, 27 and 28). Trilingualism rose to 16% in sample 23. An astonishing 65% of sample # 5 was identified as trilingual; however, this must be viewed as a questionable result. It was widely out of line with other results and it appeared in a sample which received relatively low marks in a quality control check by Mohamed Yousif Sid Ahmed. On the other hand, the two Nubian languages involved are very closely related ­ Dilling and Dair (Ghulfan). These languages were spoken by a trilingual group representing 14% of sample # 3. Dilling is not so closely related to Nyimang, but these two languages were also spoken together by sizeable trilingual groups (# 23 - 16% and # 27 - 9%). Even with an unrelated language, Hausa, Dilling occurred in a sizeable trilingual group (# 28 - 8%). Could it be said that Dilling was functioning as a minor lingua franca in the shadow of Arabic? If so, this phenomenon may be restricted to areas around the town of Dilling. In an outlying rural area the Wali dialect of Dilling was only involved in a low degree of trilingualism (# 21 - 1% and # 24 - 3%).

Dair and Dilling were not described here as mutually intelligible dialects, but they are certainly very closely related languages. Multilingualism involving Dair and Dilling together would not be surprising. It is more remarkable when multilingualism involves languages which are unrelated or only remotely related to each other as in samples ### 8, 9, 13, 14, 17 and 28. Sample # 17 reveals 10% trilingualism in three unrelated languages: Arabic, English and Koalib.

The pie chart presented below for the locality Keiga (# 14) shows a remarkable and complex pattern of multilingualism involving unrelated or only remotely related languages.

The most remarkable feature here was the 3% multilingualism in four different languages (quadrilingualism): Arabic (Ar), Keiga (Kg), Jirru (Jr) and Dabri (Dr = Dair). Arabic was spoken everywhere. Otherwise every possible combination of these four languages was attested in patterns of trilingualism and bilingualism. Arabic was not genetically related to any of the other languages. Jirru was distantly related to Dabri (Dair). Keiga was so remotely related to Jirru and Dabri (Dair) that this relationship was not generally recognized by linguists at the time of the survey in 1976. A contrasting display of this sample in terms of only three major languages may be seen in the appendix on the information sheet entitled '1976 SAMPLE OF LOCALITY' # 14a.

Map 1: Spoken Arabic

Arabic was prominent in every one of the 30 samples. However, it was less prominent in Ingassana than it was in the Nuba Mountains. Only two-thirds of the Ingassana respondents spoke Arabic (67%). In the Nuba Mountains the two localities with the lowest proportion of Arabic speech were # 2 Krongo: Arabic 72% and # 1 Masakin: Arabic 73%. Elsewhere use of Arabic ranged between 80 and 100%. On Map 1 this information is displayed for all Nuba Mountain localities of the survey together with a special inset for Ingassana.

Map 2: Arabic - Spoken Often

Map 1 by itself could be misleading. No attempt was made there to distinguish between varieties of spoken Arabic and different levels of skill in speaking Arabic. A respondent who spoke a few words of Arabic might be given as much weight as another who was fluent. A partial corrective to this situation was provided by the following survey questions:
Do you speak ..... (this language) 'often' ....?
or Do you speak ..... (this language) 'seldom' ....?
The responses to these questions provided clues to how heavily a particular language was used. The results for each of the 30 samples appear in the appendix below on the information sheets designated as '1976 SAMPLE OF LOCALITY'. For instance, sheet # 30 reveals that 67% of the sample from Ingassana spoke Arabic, but only 22% spoke it 'often'.

In many localities, e.g. ## 1, 2 and 30, there was a relatively high proportion of people who spoke the principal non-Arabic language 'often', but a much lower proportion who spoke Arabic 'often'. This would be consistent with the use of Arabic as an occasional lingua franca. It was the non-Arabic language that was used heavily. In other localities, e.g. ## 17 and 19 the opposite pattern may be observed. Here, Arabic had a relatively high proportion of people who spoke it 'often' and the proportion for the principal non-Arabic language was lower. Although this could be variously explained, it would be consistent with a shift away from a Nuba Mountain language into Arabic. This explanation will need be tested by reference to data collected under topic 4 of the survey: Language Dynamics; however, that falls outside the scope of the present report.

Responses to 'often' and 'seldom' were not expected to yield incontrovertible results. The distinction between 'often' and 'seldom' was clearly subject to various interpretations. Also, if an investigator neglected to record different responses to this question, his results would be seriously distorted. This may account for the divergent results in sample # 8 where everyone was ticked as speaking Arabic 'often', although a more carefully executed sample (# 9) in the same area indicated that only 65% spoke Arabic 'often'.

Map 2 is restricted to those who claimed to speak Arabic 'often'. Its purpose is to balance an exaggerated view of the prominence of Arabic speech which was suggested by Map 1.

Map 3 and Profile of Multiplicity: Number of Primary Languages in Each Sample

The proximity of a large number of different languages to each other is an outstanding feature of the Nuba Mountains. This may be observed on Map 3 which gives a geographical display of the number of primary languages in each locality, 'primary' being defined here as 'spoken by at least 5% of their sample'. There were never less than two primary languages in a sample and sometimes as many as 6. The average was 3.9 primary languages per sample.

The first two sheets in the appendix entitled Profile of Multiplicity provide an enumeration of all the languages in each sample and the relative number of speakers for each language. There was an average of 12.6 languages per sample, but the number of languages varied widely from one locality to another. As many as 29 languages were recorded in # 1 and 28 languages in # 17. On the other hand, only 5 languages each were attested in #7, # 22 and #27 and only 3 in the Ingassana Mountains (# 30).

Questions arise. Does the presence of a large number of languages promote a more rapid shift to the lingua franca, Arabic? Conversely, does the presence of a small number of languages support the maintenance of a local language?

Superficially, there seems to be conflicting evidence from the three localities where speakers of Arabic were outnumbered by speakers of a local language, Masakin (# 1), Krongo (# 2) and Ingassana (# 30). In Ingassana a small number of languages were reported -- only 3. However, a particularly large number of languages were present in the other two localities: Krongo -- 17 and Masakin -- 29, which was the maximum number of languages found in any of the samples. Thus, Masakin provides an extreme contrast to the pattern observed for Ingassana.

A full answer to these questions lies outside the scope of this report, but the number of languages in each sample will need to be considered together with data on varying rates of shift to Arabic assembled under topic 4 of the survey: Language Dynamics.

Map 4: Language Families in the Nuba Mountains

A shift to the lingua franca may be facilitated by exposure to Arabic and also by loan words from Arabic in local languages. However, the pattern of language relationships here could be argued not to facilitate shift. Arabic bears no obvious genetic relationship to the indigenous languages of the Nuba Mountains. The Hausa language of West Africa does have a remote relationship with Arabic, but the similarities are too faint to be of any practical value in facilitating a shift to Arabic.

One of the groups with no genetic relationship to Arabic is the Kordofanian group found in the eastern Nuba Mountains. The Kordofanian languages, such as Masakin and Koalib, have a distant relationship with other languages of Africa. Kordofanian noun classes have a recognizable affinity with the noun classes of the Bantu languages of central and southern Africa and also with the noun classes of Fulfulde and Wolof from West Africa. All of these languages have been classified together in a vast trans-African phylum known as Niger-Kordofanian.

A second group of languages appears in the northwestern and central parts of the Nuba Mountains. Languages such as Dilling, Nyimang, Temein and Daju have been classified as Nilo-Saharan, another vast phylum which includes the majority of Sudanese languages such as Nile Nubian, Fur, Dinka and Ingassana as well as others as far away as Songhai in Mali.

In 1976, at the time of the survey, Roland Stevenson held the opinion that the southwestern Nuba Mountain languages known as the Kadugli-Krongo group were related to the Kordofanian languages. Afterwards, he changed his opinion, and it is now widely accepted that Kadugli-Krongo is affiliated to the Nilo-Saharan phylum rather than to Niger- Kordofanian.

Map 4 shows roughly where these various language groups are to be found in the Nuba Mountains with (1) Niger-Kordofanian languages in the east, (2a) Nilo-Saharan languages in the northwestern and central areas and (2b) the Kadugli-Krongo group in the southwest. Arabic represents a completely different phylum of languages variously known as (3) Afro-Asiatic or Hamito-Semitic, and yet it is Arabic which serves over the whole area as the lingua franca.

Map 5: Languages from Countries West of the Sudan

The Nuba Mountains lie across an ancient east-west route for traders and for West African pilgrims on their way to Makkah. The survey recorded a number of languages associated with countries to the west of the Sudan. There were at least traces of these languages in more than two-thirds of the samples. They were absent only in a group of rural Nuba Mountain localities ### 6, 10, 12, 14, 21, 22, 26 and 27 and were also absent in the Ingassana sample (# 30). One-fifth of all samples attested West African languages as primary languages, i.e. spoken by at least 5% of the total (### 3, 15, 18, 19, 20 and 28). More than one-third of sample 28 spoke Hausa. More than half the respondents in samples 15 and 18 spoke either Hausa or Borno (Kanuri). The integration of people of West African origin into Sudanese society is related to the issue of their access to national services, and especially to education, see survey topic 7: Languages and Education. The occurrence of languages from countries west of the Sudan is displayed in Map 5.

Map 6: Spoken English

There were at least traces of English in 26 of the 30 samples. English was absent only in ### 4, 13, 14 and 30. One-fifth of all samples attested English as being used as a primary language, i.e. by at least 5% of the total (### 5, 6, 8, 11, 15 and 17). In two of these localities English has been referred to as one of the three major languages of the sample (# 11: English 9% and # 17: English 13%), but it ranked far below Arabic, since 99-100% of the respondents in these samples spoke Arabic. A geographical display of English use is provided by on Map 6.

Language and Dialect:

In the years around 1976 the survey was able to draw upon the expertise of Roland Stevenson who was a specialist on the languages of the Nuba Mountains. In most cases the survey followed Stevenson's opinions on what might be called distinct languages and what might only be called mutually intelligible dialects of a single language. Stevenson was fully aware of the complexity of the problem and of the difficulty (perhaps impossibility) of reaching satisfactory conclusions.

However, in at least one part of the Nuba Mountains, classification of the languages/dialects seems to have followed local usage rather than Stevenson's opinion. Stevenson considered Koalib and Otoro to be mutually intelligible dialects of a single language. In the survey Koalib and Otoro were generally treated as separate 'languages'.

In fact, Stevenson's earlier publications had referred to Koalib and Otoro as though they were separate entities, probably reflecting local points of view.

The published Sample of Locality # 20: Shwai, Otoro demonstrates the problem. When Koalib and Otoro are considered dialects of a single language, data will be displayed for three languages as in information sheet # 20a in the appendix below. When Koalib and Otoro are treated as separate 'languages', data will be displayed for four languages as in information sheet # 20b below. Even though # 20b represents the form in which the original sample was published in 1979, # 20a is more consistent with the generally preferred treatment of language and dialect data in the Nuba Mountain survey.

Laro has been treated as a separate language in this report, but after further investigation it may well be classified as a dialect of Koalib.

This is a perennial problem. Inconsistencies are inevitable given the fuzziness of the criteria and the state of our knowledge. However, in general, the Nuba Mountain survey aimed to classify its data for analysis in terms of languages rather than dialects.

Considerable effort was given to solving the problem of assigning unusual dialect names to the appropriate language. Field investigators were trained to identify a language by filling in a list of ten diagnostic words when an enigmatic name was given by the respondents. In spite of precautions, there were still problems. The name 'Moro' in this survey probably designates two neighbouring, but distinct Kordofanian languages: (1) 'Moro Hills' closely related to Masakin and (2) 'Moro' more closely related to Tira and Koalib.

Brief Acknowledgement

The survey of the Nuba Mountains was only one phase of the Language Survey of the Sudan. The Language Survey has been a continuing project at the Institute of African and Asian Studies of the University of Khartoum since 1972. I am grateful to the Institute and to the Ford Foundation for their substantial support for the Language Survey of the Sudan. I am also grateful to the University of Bergen for having recently granted me research time to return to this project.

The Nuba Mountain phase benefited particularly from the guidance of Yusuf Fadl Hasan, Sayyid Hamid Hurreiz and Yusuf al-Khalifa Abu Bakr. Scholars who published work in the early stages of the Language Survey are cited below in the appendix on publications. The Nuba Mountain survey followed an earlier phase conducted by Björn Jernudd and Ushari Ahmad Mahmud. When the 1976 results were being prepared for initial publication, a great deal of help was provided by Lilith Haynes. I am grateful to Diego Valle of Bergen for his technical assistance with the maps in the present study. To mention some contributors and not others, however, is unfair to the large team of dedicated students and scholars from inside the Sudan and from abroad. A full acknowledgement of the members of the large team involved has already been published in the Sample of Locality Series (see appendix on publications) and will be published again, it is hoped, when a more extensive analysis of the results of this survey is completed.


This report represents an early step towards the final analysis, but it is not a conclusion. The vast amount of labour contributed by members of the team to the Nuba Mountain survey calls for an appropriate conclusion. Even though everyone in the project was heavily involved with other duties, most of them devoted themselves to the project with diligence. Many serious obstacles arose and had to be overcome. It was a great achievement for the team to have produced the initial publications of 1979, which are the basis of the present work on the Nuba Mountains. Linguistically the Nuba Mountains are one of the most diverse and complex regions on earth and have been characterized as the Caucasus of Africa. The region has recently been subjected to exceptional pressures for change. It is a potential testing ground for contemporary concerns on culturally sensitive development and on the absorption of traditional areas into a global system. Both theoretical and practical considerations call for an adequate conclusion to the project.

Organization of the Data Display


Map 1: Spoken Arabic
Map 2: Arabic - Spoken Often
Map 3: Number of Primary Languages in Each Sample
Map 4: Language Families in the Nuba Mountains
Map 5: Languages from Countries West of the Sudan
Map 6: Spoken English


The data sheets supporting this report are arranged below as follows:

I. Profile of Multiplicity
II. Information Sheets on Each Sample (# 1 - # 30) Entitled '1976 SAMPLE OF LOCALITY'
III. Languages of Survey & Abbreviations
IV. List of Publications of the Language Survey of the Sudan as of 1979

Home Index Forward

Responsible for these Web pages is Knut S. Vikør. Last updated 5.11.98.