Douglas Boone
Richard L. Watson
editors
May 1996
Summer Institute of Linguistics
Sudan Branch, Box 44456,
Nairobi, Kenya
Limited circulation of
full report: February 1992, May
1996
Limited circulation of
condensed report: June 1996
© SIL-Sudan, 1996
This language survey of the Moru-Ma'di language family in Sudan, Uganda and Zaïre is the result of much cooperation.
The survey team wishes to acknowledge the following, to whom they owe sincere thanks:
First mention goes to SIL's Survey Coordinator, Ted Bergman, for his leadership, especially in initiating the survey and finding funding.
We are very grateful for the financial assistance given by the Pew Foundation.
We extend our gratitude to the Avokaya, Kaliko, Ma'di, and Moru language committees, without whose help and guidance this survey would not have been possible, and also to individuals too numerous to mention.
Our thanks to Dr. Livingstone Walusimbi of the Department of Languages and Linguistics of Makerere University for advice and assistance.
Thanks to Mr Jeroham P. Kaddu, General Secretary of the Bible Society of Uganda, for his practical assistance.
Thanks to the Mission Aviation Fellowship (MAF) for arranging a visit to the Moyo area and providing flights and accommodations in Kampala, and to the Lutheran World Federation (LWF) for accommodations in Moyo.
We are also grateful to the Moyo Ma'di Language Committee for providing valuable linguistic and Scripture-in-Use information.
We are also most appreciative to the missionaries of Kuluva station of Africa Inland Mission (AIM) and of the Roman Catholic missions at Ediofe (Arua town) and Lodonga (Aringa County) for their hospitality and kind help with transportation and locating language helpers.
A word of thanks also to the District Executive Secretary at Arua for providing transportation and administrative and demographic information.
The Linguistic surveyors in Zaïre worked under the auspices and care of CECA 20 (Communauté Evangélique au Centre de l'Afrique), and all survey travel in Zaïre was carried out under an Ordre de Mission.
Our thanks, therefore, to the Administrative Leaders of the CECA Bureau in Bunia for their kindness and eager cooperation.
Thank you to the people, pastors and missionaries of CECA 20 in the areas where the survey was conducted. Their hospitality was warm and their generous sharing of wisdom provided insight that the short snapshot of a survey trip is incapable of giving.
Thank you to Tasile Maraka Filipo whose ear for tone was indispensible.
Our gratitude also extends to MAF whose reliable flight service enabled the involvement of more personnel at the strategic times and places.
Thank you to the Commissaires de Zone of Faradje and Watsa for kindly sharing demographic information.
Thank you to the many kind chiefs of groupements and collectivités who not only allowed us to do the research but kindly helped us.
Finally, our thanks to the Lord God for putting the following team together, and for providing the daily strength and divinely ordered events, without which this project would not have been done.
Lugbara, Logo, Kaliko, Avokaya, Moru, and Ma'di (with Luluba) are a chain of closely related languages, each with a number of dialects.
"Lugbara, Moru, Ma'di, and Avokaya each have literature either published or in preparation for publication. However, these will not cover the needs of all the people involved. Desire for Scriptures in their own vernaculars is so strong that some people are wanting to initiate translations of their own. This could produce an unnecessary number of translations of doubtful quality unless all the translation work can be coordinated and help be given in planning the optimum number of translations which will cover all the language groups."
There are language committees in most of these groups. One common concern is in the area of orthography. In Kaliko and Logo, there is considerable interest in developing a suitable orthography. Even in languages with published materials there is a need for improvement of existing orthographies: it is said that the primary reason that the Ma'di New Testament is not used is because it is so hard to read. Another issue is the need to choose or confirm a 'reference dialect' of each language for developing a standard written form. Materials (including translated Scripture) published in these reference dialects will encourage the development of these languages with the ultimate benefit of facilitating communication and fostering greater unity. (It should be emphasized that usage of a standard written form of a language will not replace spoken variations among those who use it.)
A Moru-Ma'di Survey planning meeting was held in Nairobi in March 1988. Present were Ted Bergman (Africa Area Task Assessment Coordinator for SIL), Dick Watson (Moru-Ma'di Survey Project Coordinator), Ursula Wiesemann and Constance Kutsch Lojenga (Linguistics Consultants with SIL), Dr. Livingstone Walusimbi (Chairman of the Department of Linguistics at Makarere University in Kampala), Carol Borreson, Douglas Boone and Vern Hein. This group drafted the following statement of purpose 17 March 1988:
The purpose of the Moru-Ma'di Chain Survey is to determine the relationships between speech varieties within the Moru-Ma'di subgroup of Central Sudanic languages.... 'Relationships' will be defined by:
(1) degree of lexical and grammatical similarity
(2) degree of intelligibility
(3) sociolinguistic dynamics
The purpose of the survey is also to discover the widest possible extent of effective written communication.
The following are the GOALS of the project:
A. TRANSLATION To suggest the best means of ensuring that every speaker of one of these speech varieties has access to Scripture which s/he can understand. The survey should provide the information which will enable the church and Bible translation agencies:
1. To select speech varieties for translation so that every group will accept and be able to use at least one.
2. To prevent the production of redundant translations.
With respect to translation activities, our aim is to meet translation needs efficiently, more quickly and with a wiser use of personnel and tools than would otherwise be possible.
B. LINGUISTICS Our linguistic goals are twofold:
1. To provide comparable linguistic and sociolinguistic data for all Moru-Ma'di speech varieties.
2. To serve as a model for future surveys among other multi-dialect language chains.
Our aim is to be able to compare any Moru-Ma'di speech variety with any other; for example, the 'Ogambi dialect of Logo' with the closest 'Avokaya dialects', or the 'Wa'di dialect of Moru' with its closest 'Avokaya dialects'.
C. LANGUAGE DEVELOPMENT To suggest guideline options for the language planning in the regions where these speech varieties are spoken.
1.To provide the information which each language committee needs in order to choose the best reference dialect for developing a standard written form of their language.
2.To determine which dialects could be covered by each proposed standard.
3.To provide a basis for a standard Moru-Ma'di orthography, or as few variant orthographies as possible.
Our aim of "bringing groups together" is in accordance with the objective of UNESCO as stated in Les langues communautaires africaines et leur utilisation dans l'enseignement et l'alphabétisation (UNESCO, Bureau régional d'éducation pour l'Afrique, 1985, p. 13):
Les Etats qui partagent [des langues communautaires] pourraient coopérer de manière très étroite à leur developpement.
("States which have regional languages in common could work together on developing them.")
This report presents all research results considered relevant for the making of recommendations for language development and translation.
A summary report in French for the Zairean church has already been prepared. The present report, therefore, assumes a certain linguistic background on the part of its readers, but not a previous acquaintance with the educational, religious and political systems of Sudan, Uganda, and Zaïre. The authors of the report felt that these should be explained so that the reader can understand the situational factors of which linguistics is only one.
The concept of linguistic clusters is useful in a language chain like Moru-Ma'di where 'language' boundaries are unclear. Traditional divisions and group names based as much on ethnicity as linguistic similarity are understood to represent a classification of Moru-Ma'di speech varieties into languages. These accepted terms (Avokaya, Kaliko, Logo, Lugbara, Moru, and Ma'di) are useful for organizing our data, but we call them 'dialect clusters' to indicate our reservation of judgment as to where the true 'language' boundaries lie.
Chapter 2 is a technical overview of the survey tools and how they were used.
Chapter 3 presents the results of the survey using the above tools. Section 3.1 describes the relationships between the neighboring clusters, while 3.2 describes relationships of dialects or speech varieties within a given cluster.
Chapter 4 moves from the descriptive to the prescriptive by explaining the recommendations of the Moru-Ma'di Survey team based on the information presented in Chapter 3.
In Chapter 5 we evaluate the survey in the light of our original goals.
Finally one finds the references and bibliography as well as numerous appendices.
The following three linguistic tools were used. Each is reproduced in an appendix.
A simple measure of the relatedness of two speech forms is the number of similar words they share. Suppose, for example, that we are told that of one hundred commonly-used words sampled from a speech form called Lugbara, ninety are similar to words used to mean the same thing in a speech form called Ma'di, whereas only twenty of those Lugbara words are similar to the words meaning the same thing in a speech form called Kebu. We say that there is 90% lexical similarity between Lugbara and Ma'di, and only 20% lexical similarity between Lugbara and Kebu. It seems reasonable to conclude that Lugbara and Ma'di are more closely related than Lugbara and Kebu.
To take another example, suppose there are three groups of people called Ma'di. These people speak dialects called 'Burulo, Lokai, and Okollo. Of 160 'Burulo words sampled, 157 sound like the corresponding Lokai word; that is, the words are exactly the same (for example, the word mí means 'eye' in both 'Burulo and Lokai) or there are small differences in pronunciation (for example, the 'Burulo word for 'to die' is dà, while the Lokai word is drà). The corresponding Okollo words are elicited from someone who speaks that speech form, and it is found that 120 of these words are like the 'Burulo words. Though all three speech forms are called 'Ma'di', the lexical similarity figures of 98% 'Burulo-Lokai and 75% 'Burulo-Okollo are evidence that 'Burulo and Lokai are dialects of a single language, whereas Okollo may or may not be a dialect of the same language. The dialects with more words in common are generally more closely related.
For the Moru-Ma'di dialect chain survey, a 170-item elicitation list was used (See Appendix 9). Native speakers of each speech form were read the list, one item at a time, and asked the word used in their dialect for each item. For many speech forms, information was elicited from several individuals, resulting in multiple lists; at other times, a group of people were interviewed at the same time. When time for the full list was not available, only 100 items were elicited; these shorter lists give less information but help to double-check the accuracy of longer lists.
The responses were written phonetically and in many cases tone was marked as well. Thus when two lists were compared, allowances did not need to be made for differing orthographies. The lists were compared with two measures in mind: lexical similarity and phonetic similarity.
The count of lexical similarity is simply the proportion of 'similar' words in the two speech forms, calculated by dividing the number of words judged similar by the total number of words compared. Thus, if for a given pair of languages, 153 words out of 161 are judged similar, the two languages are said to be 95% lexically similar, since 153/161 = 0.95.
The count of phonetic similarity is less easily determined. 'similar' words are compared for phonetic differences; if, for example, it has been decided that the 'Burulo word dà is similar to the Lokai word drà, we note that these have the same vowel and the same tone, but different consonants (d vs. dr). Another 'similar' pair consists of mí and mí, that is, the 'Burulo word and the Lokai word are identical. For each pair of corresponding phones (sounds), a 'degrees of difference' value is assigned. Sounds considered identical have zero degrees of difference. Extremely similar, but distinct, sounds, such as b and p have one degree of difference; less similar sounds, such as o and e have two degrees of difference, and so on.…
Then, for a pair of lists, the degrees of difference between phones (sounds) in words counted as lexically similar are added up and divided by the number of comparable phones. The measure of phonetic similarity of two lists is then the average number of degrees of phonetic difference found for each 100 pairs of comparable phones. The higher the measure, the less phonetic resemblance between the two lists.
Note that the 'ratio of phonetic degrees of difference' is a relative measure, since it depends on how one assigns 'degrees of difference values'. A person could use a very coarse measure (0 for the same phone, 1 for a difference) or a very fine one (where two phones could differ from between 0 and 9 degrees of difference). Using a coarse measure for near-identical lists would yield a very low ratio of degrees of difference; using a fine measure might even yield a ratio of more than one hundred! (Since it can only be calculated for words counted as lexically similar, the ratio also depends on one's judgments of lexical similarity. A conservative policy— counting forms to be lexically similar only when highly phonetically similar— would yield a lower lexical similarity percentage and lower ratio of degrees of difference than a liberal policy— counting as lexically similar any forms which could conceivably be cognate, irrespective of the amount and complexity of phonetic mutations needed to explain the cognate relationship.)
In cases where two lists appear to be equally similar to a given list ('X') on the lexical level, their relative degrees of phonetic similarity to list 'X' may suggest that one of the corresponding speech forms is more closely 'related' to speech form 'X'.
Again, as an example, suppose that the Lugbara, Ma'di and Kebu words for 'fire' are aci, asi, and kisi, respectively, and that it has been decided that these are 'similar'.
There is only one degree of phonetic difference between the Lugbara word and the Ma'di word, because the vowels are identical and there is one degree of phonetic difference between s and c (which represents the ts sound). For a list of many similar words, this pair would add one to the total number of degrees of phonetic difference between the data and three to the total number of comparable phones.
In contrast, there are four degrees of phonetic difference between the Lugbara word and the Kebu word: 1 for 'no consonant' vs. k, 2 for a vs. i, 1 for c vs. s, and 0 for i vs. i. This adds four to the total number of degrees of phonetic difference between the data and four to the total number of comparable phones.
For a list of one hundred words, Lugbara and Ma'di might show a total number of degrees of phonetic difference of 150 for 450 comparable phones in 84 words, while Lugbara and Kebu might have 57 total degrees of difference for 120 comparable phones in 20 words. To compare the amount of phonetic variation, we calculate the ratio of degrees of phonetic difference per hundred comparable phones. For Lugbara-Ma'di, this is (150/450)x100, or 33; for Lugbara-Kebu, it is (57/120)x100, or 48. Thus we note that not only does Kebu exhibit weak lexical similarity with Lugbara, but even when the Kebu word resembles the Lugbara word, it tends to be less similar to the Lugbara word than the Ma'di word is.
Inter-language intelligibility does not only depend on how similar the words are in the two languages. For the speakers of two related speech forms to understand each other (without having to learn each other's language) the speech forms must have a similar grammatical structure. Realizing this fact, we attempted to compare low-level grammatical features by means of a thirty-five-item list of phrases, clauses, and questions. The list (see Appendix 9) consists of two major parts: verbal research (the first twenty items) and research of the noun phrase (the remaining fifteen items). The verbal research was considered the most important, so that in cases where only some of the data could be collected, the first twenty items were elicited; where possible, however, the full list was elicited.
No quantitative procedure has been developed to measure how great a hindrance to comprehension a given grammatical difference would be. Thus, this report only makes simple qualitative statements about dialectal grammatical differences and alternative possibilities within a dialect or cluster. These are taken into account together with conclusions about inter-language comprehension made from lexical comparisons, intelligibility testing and sociolinguistic observations.
The purpose of the sociolinguistic questionnaire is to gather an adequate sampling of information about language and dialect use and attitudes relevant to determining needs for separate language/dialect development versus standardization within a language area. (Strictly speaking, the questionnaire was actually an interview schedule since the questions were posed orally and answers noted by the researcher. A true research questionnaire would be answered in writing by the person being questioned.)
Beginning with the results of wordlist analyses, speakers of a dialect or dialect area are questioned in the area of language (and dialect) use and attitudes with reference to languages and dialects surrounding them having greater than 70% lexical similarity. There is value in random sampling and also in sampling special interest groups. The main questionnaire is intended for a broad cross section by random sampling (see Appendix 9). The supplements contain specialized questions for church leaders, educators, and government officials. In order to refer to the significant dialects of an area (in question 17), the researcher elicits dialect names according to 'folk perception' and sees how those relate to the results of the wordlist analyses.
The questions cover five areas: general information, language use, attitudes towards language use, dialect use, and attitudes towards dialect use. It was necessary to take great care in selecting wording in English, French, Arabic, and Bangala because the terms 'language' and 'dialect' are not used in the same way everywhere.
This general sociolinguistic questionnaire (SLQ) was supplemented by two others. By means of formal interview we probed language use and attitudes both in churches and in schools. See Appendix 9 for examples of both the church leadership questionnaire and the Primary School questionnaire.
Wordlists, phrase lists, and sociolinguistic information were gathered in Juba, Sudan between March and June 1988.
Dick Watson gathered a small amount of information on the Ma'di of Moyo and Adjumani during a short trip into Uganda in October 1988 or thereabouts. Some data on Ugandan Ma'di were gathered in Juba at the same time as data on Sudanese Ma'di. Data (wordlists, phrase lists, and sociolinguistic information) on Lugbara and the southern 'Ma'di' dialects were collected by Douglas Boone and Louis Otika in November 1989.
Five multi-day trips were taken in 1988; other data were elicited at Aba, where Douglas Boone and Vern Hein were based. All survey activities, on site at Aba or 'on the road' were organized through local CECA church leaders. The CECA workers who traveled with the surveyors provided invaluable guidance.
The surveyors took part in language committee meetings among the Kaliko-Omi (July 30- Aug. 5 at Adja) and the Logo (June 8-21 at Todro). The bulk of Kaliko-Omi and Logo data were collected as a formal part of these meetings.
Help by Constance Kutsch Lojenga (on wordlists) and Ulla Persson (for Kaliko) supplemented the work of Boone and Hein in Zaïre.
Wordlists were elicited in Sudan from individuals representing the various speech varieties of Moru, Ma'di (Sudanese dialects and that of Moyo, Uganda), Kaliko (Sudanese dialects), as well as one complete list in Ojila, the main Sudanese dialect of Avokaya. Some of the lists consisted of only 100 items, but full 170-item lists were collected for all major dialects. The Moru data are known to have been collected carefully, with all phonetic distinctions noted and all lists double-checked with the person from whom the list was first collected. Tone was noted on the Kaliko lists (and the Avokaya-Ojila list) but not for Moru, nor for Ma'di.
The wordlists gathered in Uganda representing Moyo Ma'di as spoken in Moyo and in Adjumani corroborate the Moyo and Adjumani lists taken in Juba, Sudan. In 1989, eight wordlists of 200 items each were taken, representing the various Ugandan Lugbara subdialects (essentially one per county), as well the Oguko and Okollo dialects of Ma'di. Tone was noted on these lists. (The additional items were collected in order to compare results with those of the survey done for the Languages of Uganda project in 1968.) In some subdialects, two or even three speakers were available to give data; on one occasion, four lists were taken simultaneously, much as was done for the Logo data (see below).
Lexical data in five Logo dialects and in Kaliko-Omi were collected using a distinctive 'group approach' by which speakers of each dialect came to a consensus concerning what word was wanted but any differences in tone, pronunciation, or word choice, even among speakers of the same dialect, were also noted. An advantage of this approach was that ambiguous items on the elicitation list were made clear in the course of the thorough discussion.
Other wordlists were collected from individuals representing some of the varieties of each language. Some of the Kaliko lists were 100-item lists, and lists were not collected in all Lugbara dialects. Tone was always marked.
Tasile, a Logo man skilled in tone differentiation, was usually present when wordlists were collected. His familiarity, not only with Logo, but also with other Zairean Moru-Ma'di speech varieties, allowed the surveyors to regularize the phonetic and tonetic transcription of data collected with his aid. He also helped clarify certain semantic nuances.
Phrase lists were gathered in three Ma'di dialects (including Moyo Ma'di, a Ugandan dialect) from five speakers of these dialects in individual interviews. Data for Kaliko and Avokaya were elicited in a similar way.
Eight phrase lists were gathered in Uganda in 1989, in five Lugbara (sub)dialects, in the two 'southern Ma'di' subdialects, and in Moyo Ma'di. Usually, the same person or persons who offered wordlist data also gave phrase list data.
As was the case for the wordlists, the group approach was used to collect phrase lists at Todro and at Adja. Again, the search for consensus helped ensure that the surveyors elicited what they wanted to elicit while turning up instances of free variation. Two individual phrase lists were taken for Logo-Bari and another for Kaliko-Ma'di.
Sociolinguistic data were often elicited conversationally rather than by means of a question-by-question interview. The researcher spent some time chatting with people about their families, conditions, and language in general. The survey team tried to question wisely in each location, discarding irrelevant questions and questions to which the answers were found to be predictable, and following up on others which proved to be more significant. That is, the interview schedule was not followed in strict order, sometimes questions were omitted, and notes were made of information offered though not explicitly elicited. No claim is made for randomness or representativeness of sampling.
Extensive SLQs were done among Ma'di residents of Juba, mainly refugees. There was no way to sample randomly as one had to interview the people at hand. However, among those interviewed were men and women, young and old, of each social and ethnic group. Other data were gathered among the Kaliko, and two SLQs were done in Avokaya.
In 1989, eleven SLQs were conducted in Uganda, with the same people who provided the lexical and grammatical data. Due to time constraints, this often meant getting information from only one speaker of each subdialect.
In Zaïre, although a good number of SLQs were conducted, particularly in Logo and Kaliko-Ma'di, much of our quality information was obtained by 'bumping into' good sources of information as opposed to formally sitting down to do an SLQ.
A word of explanation is in order concerning reliance on anecdotal data and on the lack of random samples. While planning, the survey team was guided by the statement of purpose and by cost-benefit principles. As in some cases SLQ questions with predictable answers were omitted, so in choosing population we settled for fewer actual SLQs than a sound random sampling would necessitate, and in fewer places. In places we did visit, however, we did seek the typical cross-section of male/female, young/middle age/elder samples. The minimal amount we could have learned by more samples in each of these categories was deemed not to be worth the cost involved (miles, hours).
Avokaya, Kaliko, Logo, Lugbara, Luluba, Ma'di and Moru are the agreed upon members of the Moru-Ma'di subgroup of Central Sudanic Languages within the Nilo-Saharan language family (Greenberg 1966).
Tucker and Bryan (1966) and Caprile and Thomas (unidentified work, 1968, cited by J.-P. Caprile in Barreteau, ed. 1978, pp. 240, 242) have Moru-Ma'di as a sub-group within Moru-Mangbetu of the Central Sudanic group, where the other three sub-groups are Mangbetu, Mangbutu-Efe, and Lendu. Caprile and Thomas group the rest of Central Sudanic as 'sara-Bongo-Baguirmi'. Tucker and Bryan leave unresolved the question of whether the 'larger units' of Bongo-Bagirmi (including Kresh) should be considered two branches of one language group. The listing of seven Moru-Ma'di languages is ours; often Luluba is listed as a dialect of Ma'di and Lugbara is listed as two languages. As the term 'Moru-Ma'di' is well established, the present survey uses the same name.
Speakers of these languages live in the area where the borders of Zaïre, Sudan and Uganda converge (see Summary Chart of subgroups and Maps in Appendix 1). This area in the center of Africa has been in turmoil for centuries; the succession of waves of immigrants and the drawing of national boundaries has contributed to linguistic diversity.
An example of the resultant sociolinguistic complexity is seen in the case of John, a 25-year-old farmer near Didi, Zaïre. John was born in Yendu, Sudan of Logo parents from Kitambala, Zaïre. He grew up speaking Logo-ti at home as well as Kaliko (a Sudanese dialect), Arabic, English and some Kakwa. Driven from his schoolteaching job in Sudan by the war, he has had to polish the Bangala he knew slightly in Sudan. His wife is a Zairean Kaliko-Ma'di and this is the language of their home.
Before reporting the relationships between each neighboring pair of Moru-Ma'di speech forms, it may be helpful to sketch relationships across the whole chain. The matrix summarizes the lexical similarity of ten representative Moru-Ma'di dialects:
| Moru-Miza | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| 69 | Avokaya of Sudan | ||||||||
| 65 | 76 | Logo (five main dials.) | |||||||
| 67 | 78 | 86 | Kaliko-Ma'di-Didi | ||||||
| 63 | 71 | 72 | 87 | Kaliko-Sud-East | |||||
| 63 | 73 | 78 | 91 | 88 | Kaliko-Omi | ||||
| 62 | 70 | 69 | 78 | 82 | 82 | Ugandan Lugbara-composite | |||
| 61 | 65 | 65 | 71 | 74 | 73 | 81 | Ugandan Ma'di-composite | ||
| 62 | 63 | 62 | 69 | 73 | 71 | 81 | 99 | Ma'di - Lokai | |
| 64 | 61 | 62 | 67 | 71 | 67 | 73 | 81 | 80 | Lulubo - est. |
Most of these figures are based on a comparison of about 160 lexical items. The estimated percentages for Lulubo are based on a comparison of about 140 lexical items, and there were about 170 items used in comparing Lugbara and Ma'di dialects.
These figures are a statistical estimate of true lexical similarity among all common words in these dialects, based on a sample of fewer than two hundred words. The estimates are subject to sampling error. For figures between 75% and 95%, the estimated possible error is 3-5 percentage points; for figures between 60% and 75%, the possible error is 6-8 percentage points.
The greatest similarity is observed between dialects of the same language (Ma'di and Kaliko), and between dialects spoken in geographically contiguous territories, especially within the same country (Logo and Kaliko-Ma'di, Kaliko-Omi and Lugbara, Lugbara and Ma'di, Ma'di and Lulubo, Avokaya and Logo). Moru and Avokaya are not immediate neighbors, which is probably the reason that their lexical similarity figure is not remarkably high.
Although there is a 'chaining' effect (no simple tree diagram could faithfully summarize these data due to the gradations in the figures across the whole language chain), the figures for non-neighboring languages tend to be less than 75%. Given this degree of dissimilarity in vocabulary, and considering the unlikelihood of widespread contact between people who live so far apart, it would seem reasonable to concentrate our efforts of description on pairs or triples of neighboring dialect clusters (languages) and on relationships among dialects of the same language, rather than every possible pair of speech forms or on all the speech forms at once.
For a snapshot of relative phonetic closeness (or rather distance), we present this matrix of degrees of phonetic difference:
| Moru-Miza | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| 32 | Avokaya of Sudan | ||||||||
| 29 | 20 | Logo (five main dials.) | |||||||
| 31 | 21 | 23 | Kaliko-Ma'di-Didi | ||||||
| 32 | 23 | 28 | 16 | Kaliko-Sud-East | |||||
| 30 | 24 | 24 | 13 | 19 | Kaliko-Omi | ||||
| 35 | 28 | 26 | 21 | 21 | 25 | Ugandan Lugbara-composite | |||
| 34 | 31 | 34 | 30 | 31 | 33 | 32 | Ugandan Ma'di-composite | ||
| 41 | 39 | 41 | 36 | 39 | 37 | 36 | 17 | Ma'di - Lokai | |
| 40 | 42 | 41 | |||||||