SLRK Articles Menu
Strategy Leader Resource Kit Home

Language, Tribe and Ethnic Clusters:
Analyzing the Ndengereko Cluster in Tanzania

Dr. Orville Boyd Jenkins

I am uncertain of the status or identity of the peoples called Matumbi and Machinga in Tanzania.  A Bible translator says that all these peoples speak forms of the same language and have very similar marriage, burial and other rites.  The Matumbi are part of a larger cluster of peoples, referred to in some sources as the Greater Ndengereko.  He also says the Nghwele, Kwere, Kami and Kutu are probably the same group.  He includes the Rufiji people in this group.

In regard to speech form (language), the names Nghwele and Kwere refer to the same form of speech.  The standard reference on languages is the Ethnologue.  The new edition of the Ethnologue has retired the former language name Nghwele (previous language code NHE), saying that the code has been merged with Kwere language (language code cwe).  There is no separate reference now to any language or dialect under the name Nghwele.  Rather, the Ethnologue notes "The Kwere call their language 'Nghwele'."

This report in the Ethnologue is based on an analysis of speech forms identified and analyzed.  If there is a separate ethnic group calling themselves Nghwele who consider themselves a separate tribe of people, their language is the same as that of the people calling themselves Kwere, who call their own language Nghwele.  It appears redundant to have a separate entry for both names.  It appears that Nghwele should be considered an alternate name for Kwere.  Different databases will likely handle this case in different ways.

In the Ethnologue, I note that in Ethnologue Edition 13 (1995) Matumbi was listed as a separate language.  Current Edition 15 still has it, but last population is from 1978.  Matumbi is the name of a separate language, but the name is also used for the cluster of related languages that includes Matumbi, Ndengereko and others.

In understanding this matter adequately, it turns out that three clusters of Bantu languages are involved:  MATUMBI, YAO (Machinga, Mwera, etc)etc.) and ZIGULA-ZARAMO (Kwere-Nghwele, Kutu, Kami, etc.)]

Further Question Comment:
This Bible translator understands their historical relationships in this way:
When they moved into the coastal area, the Rufiji settled around the river and took that name, the Matumbi moved further south and settled on a series of small hills ("matumbi" means mountain in their language) and the Ndengereko stayed north of the river and lived only along the coastal plain area

Further Answer:
This would explain their similarity as well as their current distinction as separate peoples (tribes).  This reflects the comon migration and naming patern of the Bantu peopoels.  They often name themselves after the places where they live.  For more on naming patterns, see section below: Language and Ethnicity in Bantu Naming Patterns.

There is growing agreement that the Ndengereko and Rufiji languages and peoples are the same.  A recent proposed update of the Bantu language family by Jouni Filip Maho lists these two as dialects of one language under the language name of Ndengereko. lists the two peoples as segments of one ethnicity.

At this time, all classifications appear to consider the Matumbi language different enough from Ndengereko-Rufiji to continue to list Matumbi as a separate language, rather than a dialect of Ndengereko. As further information comes forth from the field linguists, their reports will help clarify this.  See Maho on this also.

The Ethnologue shows these languages as a cluster, related to Yao, Makhuwa and others.  The Ethnologue gives this cluster the name Matumbi (which is also the name of one of the languages in the cluster).

It is also important to keep in mind that what is in focus is language information.  For ethnic classification we may find this helpful, but must look for additional clues, and independent information not necessary to classify speech forms.

Ndengereko Sub-Cluster
The phrase in the original question "the Greater Ndengereko" would then seem to refer to the closely related dialects/languages of Ndengereko, Rufiji, and possibly the Matumbi, depending on interpretation and acceptance of new information proposed by field linguistis of the Pioneer Bible Translators.

Ndengereko would then continue to be one of the languages in the Matumbi group, with Rufiji as one of the two dialects.  Later verification might indicate a need to move Matumbi to Ndengereko, to be reported as a dialect rather than a separate language.

The underlying question is how to FORMALLY classify those languages (and the related ethnicities).  Note that names linguists may use for classifying similar languages are largely formal and arbitrary.  They may have to choose a common name, which is not actually used by the peoples involved for all the related groups.

Narrower Bantu Identification
The tendency in Bantu language usages is to identify at the family or lineage level.  It depends on what level of identity they are answering.

For instance, along the Kenya coast, the related tribes of the "mijikenda" (Nine Villages) may variously identify themselves as WaGiriama ("Giriama People," the name of the largest group by which up-country people know them. Asked further for a more precise local identity, they would answer by their individual "sept" or "tribe" names: Duruma, Chonyi, Jibana, etc.

Several of these individual "tribes" of the Mijikenda can understand each other; their speech is classified as dialects of Giriama.  Others cannot hear others, and their speech is classified by linguists as separate languages.

The Matumbi Language Group
Here are all the individual languages of this group, with their Registry of Peoples (ROP) codes:
 Matumbi       [mgw]    ROP code 106409
 Mbunga        [mgy]     ROP code 106496
 Ndendeule    [dne]      ROP code 107157
 Ndengereko [ndg]      ROP code 107158
 Ngindo         [nnq]      ROP code 107270
 Nindi            [nxi]        ROP code 107157 (included in Ndendeule)
 Rufiji             [rui]        ROP code 108434


If these identified speech forms are actually distinct forms of speech, it would be possible to consider the speakers as separate ethnicities.  However, if linguists are discovering that some (or all) of these names actually refer to the same varieties of speech, then the language and people entries and their codes likely need to be updated.

Ethnic databases can organize their ethnic entities to account for what they are finding in the field research. The existing published codes (which necessarily lag behind the findings in the field) can be assigned as they fit.  Annotation memos should explain what has been done in the updating process.  This provides a tracking history for later cross-checks, both within the database, and for use by other databases in correlating and exchanging information.

The individual names of these tribes or languages, as listed in the current lists, would have developed in their current locations or at points along their migration, related to settlement and separation from larger groups.

They all share a common origin linguistically.  The term "Bantu" is a linguistic classification term.  The Bantu (or Bantu-speaking) groups are classified according to language similarity, which usually indicates common migration patterns.

We can think of it as a mother group back at some point in history, from which all the current known groups arose over time.  Just like any family that grows apart over generations as children and their families move and they lose contact.  They may even move to a new land or continent and learn a different language.

That is, there were no Matumbi until a group settled around the hills in a certain area, and others moved on.  This group came to be known as the "Hill People." Others moved on to gain a name or give a name of their clan to a River, a possible origin for the Rufiji as a separate clan-tribe, etc.

Most of the "tribe" or clan names are either places or ancestors or points of separation from the parent group.

There is a well-known example of this in Kenya: Coastal Bantu clans migrated northward and settled generally in Shungwaya.  They were driven south by Oromo groups, and settled along the coast at different places at different points in their history:  Thus Mijikenda, Swahili and some others share a memory of Shungwaya, and are sometimes called the Shungwaya peoples.  The names and current locations came later.

Levels of Identification
Whether they are separate "peoples" is a technical formal distinction.

For shared literature, the boundaries are sometimes narrower than for oral communication or sharing of communication and customs.  So translation definitions may be narrower.  So literacy agencies and Bible translators might find more strategic divisions am among them.

This is the focus of the linguistic databases classified in the Ethnologue.  The need for separate literature classifies a speech form as a separate language with its own unique language code.  The status of the speakers of any particular speech form in a database will depend on the purpose of the database.

Communication Access and Decision-Making
For access strategy, you have different factors, so sometimes or other unreached access strategy databases have broader entities.  It depends on how easily concepts and social change can spread internally.  Sometimes other social factors limit this even if they can understand each other, so you need smaller separate "peoples."

For communication strategy, the primary question would be:  where do you find barriers where culture exchange, new ideas and social change seem to stop flowing?

Sometimes the discrete ethnolinguistic boundaries may be broader than the formal translation listings, because the people have a strong sense of shared identity and still exchange culture, intermarry, share social institutions, etc.  The similarity of the worldview, and their self-identity concepts are often the deciding factor.  Most databases are a compromise of various factors, so agencies disagree on their listings.

Ethnicity and Language
Differing purposes and focuses for each ethnic or linguistic database will lead to different ways of accounting for identified groups.  Some entities that are given separate status and codes in the Registry of Peoples (ROP) may be clustered under one entry in some databases.  On the other hand, sometimes the ROP will consider as one ethnic entity what another database accounts for as separate smaller ethnic entities.

Language databases commonly fail to correspond with ethnic databases.  There is only about a 50% correlation of language to ethnicity worldwide.

Some databases are focused on communication strategies at a local level.  For example, see or Joshua Project.  An even more specific classification is found at Global Recordings Network, which focuses on oral forms, and often has separate materials village to village.

If you compare Joshua Project and, you will find that their entities  differ somewhat.  Both use the codes provided by the ROP for their entities but they don't always agree on the entities, or how they should be divided or assigned the available codes.

Both have entities for which they do not show any ROP code.  These ROP codes enable various database to link, compare and exchange their data more reliably than by name, which is notorious for introducing errors.

Click Here For More on Determining Ethnicity.

Another component expanding our circle of investigation is the name Machinga, also used in most sources to designate another language and people.  There is one peopel called teh Machinga who live in almost the same location as the Matumbi?

The language of the Machinga people is related to Yao (who live primarily in Mozambique), and classified in that group, not the Matumbi group.  I surmise from this that they were in a swirl of separate Bantu migration that swung up from further south and wound up near the Matumbi.

Machinga speech is identified in two broad groups, all related to Yao.  Some forms are classified as a separate language in the broader Yao cluster.  Machinga is also listed as a dialect in the Yao language itself.  The same names commonly occur like this in various places for groups of people or forms of speech.

Older information giving the meaning of "machinga" as "mountains" has been confirmed in recent extensive field investigations by a Dutch linguist, Pieter Kraal (Leiden, 2005).  There may be more than one group of people or language known by this name.  Reports are unclear whether any group of people actually use this name as a self-name.

Whether this is a name actually used by a group of people themselves is a separate question from whether the term indicates a valid form of speech or ethnic identity.  This meaning of the word machinga, however, is different from the meaning uncovered by the Pioneer Bible translators, related to Swahili "hawker"  This meaning may give us a clearer insight into the origin and relationship of the Machinga as a separate people.  This might also indicate a different grouping related to the Yao further south in Mozambique that might identify themselves more specifically by the name "Machinga."

A Branching Stream
They had already been separated from the Matumbi-Ndengereko group farther back in history (probably several generations, or even centuries), for long enough that the Yao-related speech and the Matumbi-related speech was already so different they could not tell it was related.  They retained their own speech, having no motivation to relate strongly to the Matumbi.  Their language has likely not grown closer to the Matumbi speech, even though they are neighbours.

A comparable example is the German-speaking and Italian-speaking Swiss in neighbouring cities.  So the Matumbi speech would be more similar to Ndengereko, their closer cousins in the same migration, than to the speech of the Machinga, with whom connections are beyond communal memory.

A colleague of mine near this area reports that the Bible translator he knows there confirms the Ethnologue report that the Kwere call their language Nghwele.

Language and Ethnicity in Bantu Naming Patterns
It may be helpful here to note how Bantu-speaking peoples commonly refer to their language.  A group will normally refer to their speech by whatever name they call themselves, using a prefix to designate speech.

For instance, the Swahili call themselves Wa-swahili (The Swahili People) and their speech Ki-swahili (speech or language of the Swahili).  Smaller groups within the broad Swahili group of peoples call themselves by more specific names:  In Lamu, they are A-amu (The Lamu People, called "wa-amu" in Standard Swahili) and their speech is Ki-amu (the speech of the Amu).

Likewise, in the language of the Tswana group of peoples, the people are Ba-tswana (the Tswana People), while the speech is Se-tswana (language of the Tswana People).  Individual "tribes" of the Tswana (also called a "tribe") call themselves and their language(s) by the equivalent forms of their own clan or tribal names:  Tlokwa – Batlotwa, Setlokwa, etc.

The Ragoli people of Western Kenya call themselves Ma-ragoli (Ragoli People), while they refer to their language as Lu-ragoli (Speech of the Ragoli).  This pattern is consistent across all the hundreds of Bantu languages all over Africa.  A prefix on the basic root just indicates reference to the speech of a group of people known by that name.

This may be the people of a village, region, clan, "tribe," or other grouping.  It just depends on the focus and view at what level.

Now, we turn to our question about the Kwere-Nghwele cluster of people. It is helpful to note that in current classifications, the Kwere (Nghwele) language is a member of the Zigula-Zaramo cluster.  This cluster is a different grouping of the Central Bantu languages (Cluster G20 in Central G) from the Matumbi cluster which includes Ndengereko speech (Cluster P10 in Central P).

People Name and Language Name
It is significant that the Kwere call their language by a different name, namely Nghwele.  This tells us something happened in their history.  Maybe the merging of two peoples, perhaps previously related, perhaps one moving in to the area of an already-settled earlier group.

Perhaps the difference in name for people and language indicates the breakoff of one group (Kwere) from a larger group (called Nghwele), and the breakoff group continued to call their language by the previous name of the larger parent group.  Perhaps the Kwere are the parent group, or perhaps the Kwere and other related groups all were part of a larger parent group (Nghwele) whose sub-groups gradually came to call themselves by other names.

The only source I have found so far that reports the use of Nghwele as the ethnic self-name of the group called Kwere is the Pioneer Bible Translators.  Pioneer does have a commendable history of comparative linguistic field work and analysis in the language and peoples of the region.

It is possible that the new field analysis information indicates the Kwere language, and perhaps others in the Zigula-Zaramo cluster, should now be moved to the Central P group, maybe as additional members of the Matumbi-Ndengereko cluster.  We would need to await the formal re-classification of the Comparative Linguists.  Alternatively, we might gain access to their field information and analyze it for clues to ethnic relations and worldviews.

The linguistic information itself is not definitive in determining ethnicity.  But linguistic information can often provide clues to ethnic identity.  This could give us information by which to evaluate how closely the group's self-identity and worldview concepts relate to members of the Matumbi language group.

Maybe the name Nghwele comes from a place name from some early era in their migration.  Maybe it is the name of some important ancestor.  More local-level probing in the worldview and oral history of the Kwere and the related peoples will yield some answers to these standard questions.

It would be helpfully significant if it were discovered that other Matumbi groups also used the name Nghwele (Nghwere) in some way, for their language, for a certain clan, as a generational age-grade name, etc.

This leads to four concluding points:

1.  It is significant that the Kwere call their language by an different name than Kwere.  We may not now know what this name/word means, or where it came from, but it does indicate some variation within the Kwere group and their relation to the broader group that speak similar languages, by whatever name we list them.

2.  The name may come from various sources after which peoples or smaller groupings within one people or cluster might be known:  a previous settlement are, ancestor, great historical event, parent ethnic group in their history, etc.

3. The fact that growing evidence is now indicating that all the peoples of this group can understand one another to some extent indicates a close historical and likely genetic relationship.  So the names and how they are used should be further investigated.  Various ones of these names may be only alternative names for the same cluster or parts of it.

4.  One factor is the level of mutual intelligibility between the speech of various smaller groupings of these peoples.  Where differences occur, are they due to borrowing from different languages among different sub-groups?  Or the variant words all derived from different words in the shared historical speech?

Information I have found so far indicates the consensus is that the names Nghwele and Kwere refer to the same ethnolinguistic people.  Without further information, I tend to think they should be considered one ethnolinguistic group.  In the Registry of Peoples (ROP) I have made a notation related to this but will keep on the lookout for further information for some time before retiring the Nghwele name and code. The language code will now be the same for the two entries.

I would like to know if there is a separate group within the broader Kwere community that call themselves as well as their language Nghwele.  This would be a factor in determining whether there was any basis for some separate entry for the name Nghwele.

A contact in the Rufiji area reports that a Bible translator he was in discussions with indicates he cannot find anyone that actually calls themselves by the name Nghwele.  This translator is the primary researcher for his translation agency.  I have not met him personally, but am told he has spent considerable time around Kilwa and Lindi.  He is now researching in the Kami, Kutu, Kwere and Nghwele areas.

Practical Ethnolinguistic Boundaries
For practical purposes, the level of mutual intelligibility – that is, the level to which individuals of different sub-groups can understand each other while speaking their own form of language – would be a primary factor in deciding where ethnic boundaries should be defined.

We depend here on linguists on the scene.  This is where investigations by translators become crucial.  Linguists on site wrestling with comparisons of similarities and differences are the ones who will know how similar the various forms of speech, and thus what names represent what speech when certain persons are referring to those forms of speech.

What the current investigations are finding should be primary factors on which we rely about how to account for the variations among these neighbouring and related groups in people.  We decide what we can based on what we know at any one time, and revise as more becomes clear.

A critical deciding factor is usually which other speakers or groups of speakers any individual or group says they themselves can communicate with.

Strategy Groupings
Tallies on how many "people groups" are unengaged is determined by how the entities are designated in the field strategy view.  So there has been some confusion over whether certain entities should be really considered primary ethnic "people" or grouped with others.

This is often a formal decision related to classification and strategy.  What we are rally looking at is "how to account for this particular community."  How this is best done depends on the purposes of the database, and the intentions and goals of the agency.

People First, Name Later
We sometimes (in western analytical models) approach it as "who are the <name> people group" – this really contains an assumption that the name is an entity in itself, and is correctly referring to some actual objective entity.

A better question is "What does this name REFER to" and so the context of WHO used it WHEN and WHERE for WHAT purpose in the listing.

Three Levels of Consideration for Classification
Another factor is that strategy divisions may be different (different factors to take into account, including geography) than the formal classification groupings of peoples.  For instance, the CPPI database of the International Mission Board is meant to be a strategy-oriented database, but the people classifications may be different from the groupings that best facilitate strategy planning and access.  This is basically a field and team call in the CPPI.

A third level of considerations then relate to administration and logistic support for TEAMS in the field.  Each of these three, of course, interrelate, but considerations at one of these three levels is different, so team support and management should not usually be a consideration in how the people listings are determined, since that has to relate to standard conventions in classification across academic and agency disciplines.

Strategy communication considerations are closer to the top-level considerations, since they deal with worldview and language communication.  But access strategy may require different groupings that do not change the classification factors.

Related Articles:
For More on Ethnicity by this author
Language Continuums and People Clusters:  Adjusting Ethnic Entities in Reference to Language Research Updates

Related on the Internet:
A Grammar of Makonde (Chinnima, Tanzania), by Pieter J Kraal (Oct 2005) Leiden University Dissertations Online
A Referential Classification of the Bantu Languages
Bantu P Group Languages Matumbi-Makonde-Ndengereko-Yao and Related
     Links to Many Online Publications by Major Linguists
Bantu P Group Languages Ethnologue
Global Recordings Network
Joshua Project
Makonde and Related Peoples and Languages Pertinent information found in the 15-page Introduction
Registry of Peoples (ROP)

Relevant Language Groups:
Matumbi (Ndengereko) Language Group
Matumbi Language Ethnologue
Mbunga Language Ethnologue
Ndendeule Language Ethnologue
Kami Language Ethnologue
Ngindo Language Ethnologue
Nindi Language Ethnologue
Rufiji Language Ethnologue
Zigula-Zaramo Language Group Ethnologue
Kami Language Ethnologue
Kutu Language Ethnologue
Luguru Language Ethnologue
Kwere Language Ethnologue
Zaramo Language Ethnologue
Zigula Language Ethnologue
Yao (Makonde) Language Group Ethnologue
Machinga Language Ethnologue
Makonde Language Ethnologue
     There is a Maraba dialect of Makonde; different from the Maraba (Makwe) language also in Tanzania and Mozambique (code ymk)
Mwera (Nyanza) Language Ethnologue
Mwera Language Ethnologue
Ndone Hamba Language Ethnologue
Yao Language Ethnologue
Makwe (Maraba) Language Ethnologue
     The language is called Maraba in Tanzania; in the Swahili family; different from the Maraba dialect of Makonde (kde)


Topic arose in an Internet chat with a researcher in East Africa, 24 May 2006, and later exchanges with others.
Written as an article over May, June and July 2006
Article finalized and posted on SLRK 22 August 2006
Last edited 4 August 2010

Orville Boyd Jenkins, EdD, PhD
Copyright © 2006 Orville Boyd Jenkins
Permission granted for free download and transmission for personal or educational use.  Other rights reserved.

SLRK Articles Menu
Strategy Leader Resource Kit Home

filename:  langtribendengereko.html