1I imagine being a historian of early Africa feels much like being a detective searching for the next lead in an unsolved mystery. The work involved is at once challenging and rewarding to the point of compulsion. Historians committed to recovering early history in Africa often have limited, if any, written documents to turn to for writing those histories. Because of the dearth of written data, it is important for us to explain the historical methods we use in our work. This chapter is centered on explaining the methods, presumptions, and pathways of questioning that I used in writing a history of central-east Tanzania. Its objective is to set up an overarching framework for the reader. The chapter is laid out in a number of sections. The first part introduces the primary methodological approach applied throughout the work. The subsequent section foregrounds the relationship of the Niger-Congo and Bantu language families to Bantu-speaking people in central-east Tanzania. That discussion leads into an assessment of prior lines of historical inquiry centered on central-east Tanzania. It argues that they have contributed to fostering a "model minority" mold that commonly glosses over what preceded and supported the lauded achievements attributed to Swahili history. It ends by detailing the foremost themes and questions underpinning this investigation.


Words of History, History of Words

2A good thing about using words for historical reconstruction is that they do not die easily. People cling to language because they know without reservation, without need for theoretical rumination, without self-reflection, that language is the guardian of culture, knowledge, and experience.

3Languages, for all their complexities of morphology, phonology, syntax, and the like, embody in their lexicons, their vocabularies, the fundamental minutiae of what people who speak and spoke those languages think and thought about. That is because languages in their spoken forms represent the tool people use to communicate to others the physical things and abstract notions that matter to them. And when things matter to people, and this of course applies to more than language, they usually do not give them up unless there are real-world reasons that compel them to, or make it advantageous for them to do so. But even when such conditions are met and change occurs, people often leave their marks by modifying, consciously or not, the new things they are forced or choose to adopt, making them their own over time.

4Through the study of language change, we stand to learn a lot about its past speakers. The essential foundations for using language evidence to recover history are established by what is called the comparative method of historical linguistics. Applying its techniques allows the reconstruction of the phonological history of a language family, and the reconstruction of the phonological history of the family in turn provides the critical apparatus for elucidating the histories of the individual words of the languages of the family. With this apparatus in place, the history of words spoken in a language are examined to discern if they persist from a language spoken by the speakers of earlier ancestral forms of the language; if their meanings were changed or had new nuance of meaning added to them over the intervening historical time span; if they were given up/traded in for new words that may have been introduced to a language during a period of interaction with a distinct community of language speakers, and more. Such queries allow historians access to the histories of words and things that mattered to people in earlier periods of history. Moreover, this method does not depend on access to prerecorded texts, which makes it an ideal tool for writing histories of preliterate societies.

5The comparative method of historical investigation of languages is more than two centuries old. The discipline emerged from scholars who were interested in understanding the linguistic relatedness of languages belonging to the hypothesized Indo-European family of languages. Over the past two centuries, their initial interests have developed into a body of methods and knowledge about the history of proto-Indo-European, a language spoken more than six thousand years ago, which has diverged since that time into more than four hundred modern-day spoken languages. Contemporaneous with the development of Indo-European studies were linguistic researches centered on the recovery of Native American cultural history. And following their leads, Africanist linguists and historians, more than any other specialists, have tapped into the wealth that language reconstruction offers for the recovery of early African language and history. not depend on access to prerecordedUsing linguistic methods tailored to suit the objective of historical recovery, historians have pushed, for example, to understand the way that Bantu-speaking descendants, over millennia, have migrated into the greater portion of the southern third of the African continent.1 Just as importantly, recovering early Bantu social, economic, and political histories has taken center stage in their research priorities.

6Two historians, Christopher Ehret and Jan Vansina, are pioneers in the cutting-edge use of interdisciplinary methods in the study of early African history. They have dedicated a significant portion of their lives to researching and writing on topics related to Africa's early past using historical linguistic methods as a primary prism. Among their seminal contributions on early African history are Ehret's An East African Classical Age and Vansina's Paths in the Rainforest.2 Following in their footsteps are historians Christine Ahmed-Saidi, Kairn Klieman, and David L. Schoenbrun, who in their contributions to early Bantu history make innovative use of historical-linguistic methods to write histories covering distinct regions of the African continent.3 Their collective body of research continues to bring forth well-substantiated, rich counter-narratives to entrenched rhetoric that touts as fact the perception that histories of oral-based African societies are irretrievable. Without their commitment, abstractions of a "primitive" or "pre-historical" Africa would prevail without critique. Instead, they continue to develop the viable and constructive ways to remedy what could be dismissed as a gaping hole in knowledge owing to a "data dilemma." The work they do has blazed trails and improved strategies for recapturing histories that once were considered to be out of the reach of scholars at best and unimportant at worst.


Comparative Historical Linguistics: One Way Back

7The methods of historical linguistics are useful for reconstructing histories of all people who lived in oral societies or in societies where printed historical records were not preserved. The methods involved in language reconstruction rest on two important axioms. One is that words are texts. The other is that transformations in language and history always intersect. As a result all spoken languages—living languages—cradle remnants of histories belonging to the people who spoke them at different periods in the past. Their spoken words provide an avenue for reconstructing the histories of those peoples. This is possible because words holding a place in a language's lexicon derive from one of three possible scenarios, none of which is more important than another, but each of which reveals information about a language's—and therefore its speakers'—history. In one development, words trace back to an, earlier, related spoken language. In other instances, words may have been taken up from another language, what linguists sometimes term transferred, loaned, or adopted words. Finally, speakers of a particular language may have invented new words. These are called word innovations. Such innovations commonly develop in one of two ways. They may be rooted in a word already in the lexicon. This regularly happens by deriving a noun from a verb already present in the language. Another innovation can include adding new meaning to an existing word.

8Each of these circumstances directly reflects information about historical circumstances relevant to those language-speakers' experience. Consider, for example, a word argued to be an innovation at a particular point in a language's history. Such a development suggests that those language speakers may have coined it to communicate a new idea or material invention, or as an alternative way of identifying a new practice. Another possibility is that an innovation may represent a shift in the meaning of an older word previously in speakers' lexis, which is sometimes prompted by new developments in social structure or technologies. If a word is transferred into a language, speakers may have transferred it to name a product or practice they acquired from a distinct society. At other times, speakers may replace words for something they already named in their society with a loaned word from a distinct language because they consider the people from whom they learned the word to be prestigious.

9These represent only a few examples of the sort of questions historians who reconstruct words ask when assessing the historical significance and implications of word histories. For historians, recovering the words spoken by extant-language speakers' proto-language ancestors is akin to archaeologists discovering ancient, rare artifacts. Reconstructed words reveal to us aspects of proto-language speakers' material culture, technologies, and ideas. Borrowed words reveal the patterns of cross-cultural contacts and the movements of things, customs, and ideas from society to society.

10In order for words to function as useful texts for historians, word etymologies should be reconstructed within a time-depth framework. Understanding the sequence of change in words is the element that permits historians to write history from them. Recovering the chronology of linguistic change rests in part on determining the genetic relationships of the languages germane to the study. This is what is introduced in the next section.


Language and Genetic Relationships

11The steps involved in determining language relationships are not obscure or complicated to learn, but because of the deliberate and methodical way in which shards of information must be pieced together and analyzed before historical accounts can be written using them, it does help if the researcher is tenacious. Notwithstanding the need for fortitude, when the method is applied with determination and thoroughness the fruits of its labor can be both abundant and invaluable because they allow access to historical times that would otherwise remain invisible.

12Claiming that languages are genetically related means to say that at some point in the past they descended from a common language. That language is referred to as their "proto-language" or ancestral language. The biological process of mitotic division is a helpful metaphor to illustrate the way language divergence occurs. When a "mother cell" divides into what are termed its "daughter cells," each daughter cell embodies the genetic makeup of its mother. At the end of mitotic processes the mother cell ceases to exist; it evolves into or becomes its daughters. Similarly, when a proto-language diverges, it evolves into or becomes its daughter languages. And each descendent language embodies the linguistic features of its proto-language. New daughter cells and new daughter languages go through changes that result from experiences over the course of their lives. But, nevertheless, they retain qualities inherited from their ancestors. Because of this, daughter cells and daughter languages can be analyzed individually and compared to one another to try to discern attributes that characterized their ancestor. To do this, cells' chromosomes can be dissected and scrutinized. When it comes to languages, similar analysis can be undertaken on its corpus of lexis, syntax, phonology, phonetics, and morphology.

13The basic classification of the Ruvu languages and their membership in the Northeast-Coastal Bantu group of the Kaskazi branch of the Mashariki languages has been previously laid out in a variety of studies.4 So the broad foundations already exist. But to uncover the history of Ruvu peoples it is necessary first to confirm and refine details of internal relationships within the Ruvu subgroup. Lexicostatistical methods comprise the initial approach. This methodology rests upon the fundamental principle that the closer the genetic relationship among languages is, the higher the retention of common cognate words in their shared core vocabulary.

14 Cross Reference:
To apply this method, investigators must compile comprehensive lists of "core" vocabulary from each language in question. Core vocabularies include words that are usually found in the languages of people across geographic place and culture. They are distinguished from what is called "general vocabulary" because they are theoretically more resistant to change in meaning, to deletion, and to transfer. Stated differently, they are words that people are less likely to replace with others, even if they are presented with a new lexical option. They include such words as those for certain body parts (for example, leg, nose, eye), fundamental human actions (for example, to sleep, to stand, to sit, to come, to go), personal pronouns (for example, I, you), some flora and fauna (for example, tree bark, tree, animal), and some especially salient phenomenon of the natural world (for example, rain, cloud, water, star, moon). There exists in several variant forms a standard list of one hundred core vocabulary meanings commonly used in this kind of exercise. The comparative lists collected in this fashion for Ruvu languages appear in the Appendix.

15To initiate the lexicostatistic exercise, however, one must first determine whether the words claimed to be inherited from a proto-language are indeed shared retentions of particular old cognates, that is, if they truly stem from a common root that shared a common meaning. Verifying the sound correspondence rules for the present-day languages is the way to do this. Regular sound correspondence helps us determine whether two phonologically similar words are in fact cognates, loans, or chance resemblances. The method used to do this is based on theory pertaining to the regularity of sound shifts within languages. On the whole, it is accepted that all languages attest regular and definable rules with regard to sound change. To use an Indo-European family example, we know that English father, foot, fish and Latin pater, ped-, pisc- are cognates because they show regular sound correspondences throughout. The most obvious individual correspondence is in the word-initial /f/ and /p/, which indeed represent regular outcomes respectively in English and Latin of the proto-Indo-European consonant *p. Proto-Indo-European *p regularly shifted to *f in English (this is expressed as /p/>/f/), but in Latin that did not happen. Instead proto-Indo-European *p remained *p in Latin. Hence in all cognates shared between Latin and English and directly inherited by each from their proto-Indo-European ancestor, we expect to see this correspondence in the word initial position. The vowels and the other consonants in these pairs of words also show regular sound correspondence, although it would take us far afield to show all the evidence for these here.

16Using the reconstructed systems of regular sound correspondences among related languages as a starting point, one is able to determine if words are cognate. This is confirmed when words exhibit regular sound shifts and when they retain the proto-language meaning. In the case of the Bantu languages overall, and therefore for the Ruvu subgroup of Bantu, the basic structure of the comparative historical reconstruction has been long established, although additional details particular to the phonological histories of the Ruvu languages will be noted where relevant to the historical arguments. Once the proto-language phonological system and descendant language sound shifts are identified, the words that are cognate and those that are not can be systematically identified, and the total number of cognates shared by each language pairing can be counted up and calculated as a percentage of the whole one hundred-meaning list. This is where the basic principle applies. The higher the shared cognation between particular languages is, the closer their relationship. To demonstrate this process for our purposes, we turn to the Ruvu data itself.

17It should be noted that these data do not form the sole basis for the internal subclassification of the Ruvu languages followed here. Other scholars have adduced a variety of evidence, both phonological and grammatical, in previous studies.5 In addition, lexical innovation in the lexicons of culture, as well as in word borrowing, which emerge in the discussions of subsequent chapters, fall into patterns that provide further confirmation of the classification and thus the linguistic stratigraphy of Ruvu history presented below.


Proto-Ruvu and Its Daughter Languages

18The work of generating the corpus of lexical evidence essential to writing early Ruvu history centered on the languages spoken in central-east Tanzania belonging to the Ruvu subgroup of languages. The primary study was carried out over a period of approximately fifteen months between 1997 and 1999. During that time I worked among indigenous speakers of ten extant Ruvu languages: Doe, Gogo, Kaguru, Kami, Kwere, Kutu, Luguru, Sagala, Vidunda, and Zalamo. In each instance, I collected both core and general vocabulary lists. From the outset, the study had some unavoidable limitations because only the histories of a finite number of words are scrutinized in this book. That said, they do reflect the words that naturally flowed from the stories Ruvu-speaking people elected to share with me when they generously gave of their time to work with me. For all the acknowledged gaps that are contained within its pages, my position is that this book is only the beginning of a story that should be continued.

19 Table 1:
Ruvu Shared Cognation Percentages
The core vocabulary collected firsthand by this researcher from forty-three Ruvu speakers between 1998 and 1999 provides the data for calculating the percentages of cognate retention between each pairing of the ten languages. Table 1 compiles these findings.

20In this table the Ruvu languages are arranged according to their relative ranges of shared cognation into proposed genetic subgroups that evolved after the proto-Ruvu period. Following established procedures, wherever it is possible to do so, the languages having the highest rates of cognation with each other are positioned next to each other along the diagonal.

21The language subgroups proposed were determined by considering two factors. The first was the highest rates of cognation between individual pairs of languages. The second was the relative coherency of ranges of cognation held by pairs or groups of languages relative to those pairs and groups more distantly related to them. By considering both of these features the established groups are not supported solely by rates of cognation based on their individual shared retention rates with another language; they are corroborated by being of consistently similar distance of relationship to other subgroups of languages within the wider Ruvu group. Where there are instances of inconsistency—namely, evidence contrary to the general patterning within subgroups, a phenomenon termed "skewing"—they are indicated by orange boldface type on the table.

22Indications of skewing do not pose a problem for the proposed subgroups because cognate rates represent the accumulation of random changes within languages over long time spans. So instead of representing a potential dilemma, skewed cognation rates are either statistical outliers or they are rates skewed by word borrowings that spread from one language to another at an early historical point, before sound changes that would have revealed them as borrowings took place. In the latter instance they are important clues that on-the-ground historical developments may have contributed to atypical rates of core vocabulary retention or replacement. For instance, skewed figures often alert the researcher to considerable historical contacts between distinct groups of language speakers, which would highlight the need to pay particular attention to other evidence, such as borrowings in culture vocabulary, that may indicate these contacts. Such histories normally arise because the earlier speech communities shared borders during the eras when they were diverging from their common proto-language. Or it may be the case that various language groups were economic partners or that they intermarried substantially in earlier times, among other kinds of close interaction. Hence, what might be at first thought to be "kinks" or errors in the data reveal potentially important additional features of histories that should be probed further, particularly when their general vocabularies are analyzed.

23With shared cognate retention rates among languages determined and organized on a table representative of their genetic relationships, it is useful to organize them in a way that is illustrative of the history of the divergence of the proto-society, in this case proto-Ruvu, into its descendant intermediary speech communities and subsequent recent languages. This can be depicted using a tree model, also called a dendrogram. It is something that looks very much like the type of genealogical tree one might construct to record a personal family history.

24But before that is done, it is important to address some of the reservations that have been expressed about using dendrogram/tree models to represent language divergence. A major concern is that the use of linear lines to represent the divergence of a proto-language into its daughters implies that languages diverge suddenly, as if they splinter off of the proto-language. However, that is not how languages diverge. As noted in the mitotic analogy, language divergence is a process, one that happens over a long period of time. Languages remain mutually intelligible dialects long before they become distinct languages.

25 Figure 1:
Proto-Ruvu Divergence and Internal Group Contact
Another concern related to the use of dendrograms is that it is often thought that they imply that once languages have diverged they are isolated entities. But that, too, is not accurate. Language speakers often share borders, and speakers can be multilingual, or have any number of things that keep them in close historical interactions with speakers of other languages. To take into account the evidence of upward skewing already noted in table 1, the dendrogram model offered in figure 1 includes horizontal arrows between various descent lines on the tree. They are meant to be suggestive of periods of suspected heavy contact that lead to skewed cognation rates after the proto-language began to diverge, but they do not attempt to quantify the influence of those relationships.


Glottochronology: Estimating When Languages Diverged

26The dendrogram provides a constructive visual representation of the pathways of language divergence, but the application of an additional linguistic tool, glottochronology, provides a useful way for historians to hypothesize relative approximations for the likely dating of language divergence over time. Glottochronology-based dating methods were developed using evidence from a great variety of languages and language families. The empirical findings consistently show that the accumulation of lexical replacement in core vocabulary tends to distribute around the same median figures for defined spans of time regardless of the language under investigation or region of the world. For example, after a 2,000-year period of divergence of two languages out of their common mother language, their percentage of cognates still retained in common in the one hundred-meaning list can be expected to range around a median figure of 55%. Or if, for instance, the extant daughter languages of a particular proto-language are compared and it is found that their pairwise cognate retention counts range around a median of 74 or 75 core vocabulary words in common out of the original one hundred, then it is possible to roughly estimate that their proto-language ancestor began diverging about one thousand years ago. Any particular count could be less or more than the median, but will be within a relatively few percentage points of it.

27Just as there are debates waged on the use of dendrogram illustrations, there are also uncertainties surrounding the use of glottochronology. The reason for the criticism appears to stem largely from a mix-up about what the tool of glottochronology claims to do. Ehret's view "is that both the detractors of glottochronology and many of its early supporters misunderstood it as a phenomenon involving regular and predictable change. But in fact what is involved is the accumulation of many changes that are individually random and unpredictable."6 In other words, glottochronology never suggests that all languages substitute core vocabulary at a fixed rate, or that they substitute the same words. At a theoretical level, it presumes that every language and its changes are unique. When that is understood, then it is possible to contemplate what the positive possibilities are for applying the method, even if the dating results are rough projections.

28 Establishing the subclassification of a language group amounts to the creation of a relative chronology of the succession of divergences by which the languages of the group evolved out of their common proto-language. Each line of language descent depicted in this way constitutes a linguistic stratigraphy. What glottochronology adds to this is the ability to assign very rough dating estimates to the stratigraphy.7 Having these estimates allows us to compare them with conclusions drawn from archaeological research. Klieman has highlighted the importance and value of using correlative approaches, wherein linguistic and archeological data are compared wherever possible because, " . . . neither historical linguistics nor archeological dating techniques provide finely calibrated chronologies."8 The stance assumed in this book is that we must make honest use of our available tools of analysis lest early history be deemed, yet again, inaccessible.9 Table 2 provides the standard median estimates for core vocabulary retention over time used in this book.

Table 2: Glottochronology-Derived Median Dating Estimates10

29 Applying these estimates to the shared cognation rates established in table 1, which were used to construct the dendrogram in figure 1, provides a way to estimate the probable approximate timing of proto-Ruvu divergence into its descendant languages. What follows is a narrative explanation of those results. In chapter 2 these results are compared with estimated dates yielded by archeologists.


Language Classification and Glottochronology

30Based on rates of shared cognation represented in table 1 and the underlying principles surrounding language divergence and emergence, a three-way split of Ruvu languages into West Ruvu, Kagulu (a single language), and East Ruvu is proposed. West Ruvu forms a rather straightforward group with the three members, Sagala, Vidunda, and Gogo. Their shared cognation range of 74%–78.5% is distinctly higher than their scores with any outside language group. With this rate of cognation it is likely that the proto-West Ruvu language began diverging into Vidunda-Sagala and Gogo as long ago as the early part of the second millennium, probably by around the eleventh or twelfth century CE. The divergence of Vidunda and Sagala would have happened only slightly later, based on their shared rate of 81%.

31Of the three groups that diverged from proto-Ruvu, the East Ruvu group evolved the most descendant languages. It represents a classic illustration of the process involved in "dialect chaining." Specifically, it represents the course by which languages evolve first into dialects and then into languages in their own right, while their speakers maintain a contiguous chain of contact in geographical space. The narrowest subgroup within the East Ruvu branch is Southeast Ruvu, comprised of Zalamo and Kutu. They share a distinctly high 93% cognation rate with each other, and in turn share a cognation range of 73.5%–89.5% with the rest of the East Ruvu group. The 93% cognate rate for Southeast Ruvu suggests that their divergence into two distinct dialects may date roughly to as late as 1700 CE.

32The meaning of the other cognation rates, those comparing Southeast Ruvu with the remaining East Ruvu languages, Doe, Kwere, Kami, and Lugulu, are more challenging to interpret because there are concurrent historical dynamics at work. First, there is rather clear evidence of a straightforward dialect divergence in the tight range of cognation shared among the Central-East Ruvu subgroup—consisting of three divisions, Kami, the Southeast Ruvu cluster of Kutu and Zalamo, and Kwere—at 84.5%–88.5%. This range makes a good case for postulating a proto-Central-East Ruvu society dating to around the fifteenth century, out of which Kami, proto-Southeast Ruvu, and Kwere dialects then emerged. But what can be said about proto-Central-East Ruvu's relationship to Doe and Lugulu? Two ranges of figures characterize the data. Doe's range with Central-East Ruvu runs 81%–83.5%, with a particularly high figure of 89.5% cognation with Kwere. Lugulu has, for all practical purposes, the same range, 79.5%–85.5% with Central-East Ruvu. Lugulu and Doe's cognation with each other, however, falls significantly lower, at 73.5%.

33This pattern suggests that something other than seamless language divergence transpired. This is where it is likely that an important historical dynamic may have been at play. Given that language evolution does not occur in a vacuum, and that contact between speech communities continues during the formation of dialect networks such as these, it is probable that ongoing heavy inter-linguistic contact and ongoing word transfers characterized their relationship. In fact, the usual historical significance of anomalously high figures is that strong contact continued after divergence between the particular descendant communities involved. Based on this, it is likely that some innovated words and meanings have been adopted from one language into the other, and, what is more, that because they are so recently diverged and so phonologically similar, such adoptions may be virtually impossible to identify as transfers. This phenomenon is very common, particularly for languages that are within a period of divergence of fewer than 1,500 years. This particular percentage pattern, of two distinct cognation ranges, is typical when a proto-language diverges into a chain of daughter communities. The higher range indicates the daughter communities adjacent to each other in the chain, while the lower range likely represents the daughter dialects that did not share adjacent borders. In this instance, proto-Central-East Ruvu, because of its higher figures with both Doe and Lugulu, would have been centrally located. Doe and Lugulu must have been at the ends of the chain, with proto-Central-East Ruvu between them. This distribution probably reflects a long-term stability of population groups, because their modern-day descendants still live in much the same geographical relationship to each other.

34But what about the one significantly higher figure of Doe with Kwere? This would suggest that they have shared strong and more recent contacts. The somewhat higher percentage that Lugulu shares with Kami, at 85.5%, reflects a long-term history of contact, although one perhaps not as influential as the linguistic intermingling between Doe and Kwere speakers.

35Finally, the proto-East Ruvu language must be dated based on the low figure of 73.5%, the one shared by the languages at the ends of the dialect chain, Lugulu and Doe. The usual pattern in instances like this one is for the percentages of cognation between the geographically most distant members of the chain to be those that most closely reflect the true time depth within the chain. Based on this principle, proto-East Ruvu probably comprised one language with little or no dialect differences among its speakers at about the tenth century.

36Now, what about proto-Ruvu's third descendant branch? Earlier it was proposed that Kagulu stands on its own as a direct offshoot of proto-Ruvu, rather than belonging to either proto-West or proto-East Ruvu. What may seem a potential problem is explained by dialect chaining. As we have seen, in a dialect chain composed of three primary groups, it is common for a central member to have closer scores with its neighbors on each side than those groups at the ends of the dialect chain would have with each other. And this is precisely what is found in this case. Kagulu's range of cognation with the East Ruvu languages runs from 69% to 75%, while it shares a range of 61%–74% with West Ruvu languages. These percentages place Kagulu at approximately equal time-depth distance from both proto-West and proto-East Ruvu. The core range of cognation percentages of the West Ruvu languages with East Ruvu, as is characteristic of languages whose common ancestors were spoken at the ends of the chain, is a step lower, at 59%–69%. These percentages, in other words, support the inference that the proto-Ruvu language at first diverged into a chain of three daughter dialects, with proto-East Ruvu in the east, ancestral Kagulu in the middle, and proto-West Ruvu to the west.

37Three individual percentages, of Sagala with Lugulu and Kami, at 75% and 71%, respectively, and of Vidunda with Lugulu at 70%, run distinctly higher than the rest of the figures comparing West Ruvu with East Ruvu languages. Because the Sagala speech territory in recent centuries has been close to those of the Lugulu and Kami, while Vidunda adjoins Lugulu at the southwest, these higher figures can be taken to reflect close cultural relations dating since the split of the proto-West Ruvu society, that is, after the eleventh century.

38In general, as we have already noted, the low range of figures in a dialect chain most nearly reflects the actual beginnings of divergence within the group. In this case, the proto-Ruvu data, with a range of 59%–69% and a median around 65%, suggest that sometime during the sixth century CE, the proto-Ruvu language began to diverge into its three daughter dialects, and it is thus from this approximate period that original proto-Ruvu community descendant speakers began to emerge on the central-east-Tanzanian landscape.


Ruvu Patterns of Divergence and Settlement: A Schematic Representation

39Having established the cognate retention ranges among the Ruvu languages and interpreted the significance of these ranges for the Ruvu periods of divergence, historical mappings of the proposed proto-language communities and their present-day daughter languages can now be generated. What follows below are two mapping sequences. The first is a schematic illustration of language divergence. The second shows the Ruvu and other relevant languages plotted on physical maps of Tanzania. In the schematic representation, languages that can be argued to have maintained more closely tied interrelations based on shared cognation rates are positioned adjacent to each other. Placed farther apart are those languages whose cognation rates suggest they have had less or no contact after the proto period. The second mapping sequence depicts the most plausible places Ruvu language communities moved into and out of as they diverged within the central-east-Tanzania hinter-coastal zone in the successive eras since the proto-Ruvu period. Examining the second map sequence permits one to envisage the early regional groupings of Ruvu communities as they evolved into the recognizable ethnolinguistic communities of recent centuries. As we will see in chapter 2, the social transformations and interactions Ruvu descendants experienced in those eras likely influenced some of those movements.

40 It is important to note that in the following map series, the size, shape, and color of the symbols used for illustrative purposes are in no way meant to be suggestive of population size, pathways, or rate of movement and contact. The color patterning conveys which languages derive from the proto-West Ruvu, proto-East Ruvu, and Kagulu subgroups. Wherever solid lines are used they are intended to give only a broad sense of the hypothesized areas in which populations may have moved and to suggest a somewhat general sense of the areas in which those communities may have settled. In the instances where symbols overlap, the intent is to illustrate a degree of relative contact that may have occurred between groups without suggesting any absolute scale.


Schemata 1: Proto-Ruvu Language Divergence Sequence

41 c. 600-900 CE Proto-Ruvu divergence into three primary branches—proto-West Ruvu, proto-East Ruvu, and Kagulu—is represented in the sequence between 700–900 CE. Based on the evidence of its upwardly skewed cognate counts with both the East and West Ruvu groups, the ancestral Kagulu appear to have held a central geographical position with ongoing contacts transpiring between them and the East Ruvu on one side and the West Ruvu on the other. In the scheme, Kagulu is thus positioned between proto-West Ruvu and proto-East Ruvu to convey the proposed dialect chain.

42 c. 1000-1400 CE In the subsequent phase, c. 1000 to 1400 CE, proto-East Ruvu diverged into Doe, Lugulu, and proto-Central-East Ruvu. Independently, proto-West Ruvu diverged into Gogo, Sagala, and Vidunda. Note that Doe is positioned more closely to Kagulu, and that Sagala and Vidunda are placed nearer to Lugulu than either is to Gogo. Again, based on shared cognation, this representation attempts to highlight languages whose speakers likely participated in a higher degree of contact with particular linguistic neighbors.

43 c. 1500-1700 CE The representation of c. 1500 CE to 1700 CE portrays the developing dialect chains that came into being as Kwere, Kami, and proto-Southeast Ruvu diverged from their proto-Central-East Ruvu ancestor. The West-Ruvu languages experienced no further divergence.

44 c. 1700-1800 CE By 1700 CE, proto-Southeast Ruvu began to evolve into Zalamo and Kutu dialects. The increased group clustering and overlap draws attention to their patterns of shared higher rates of cognation with some languages as compared to others, which likely relates to ongoing interactions since that period. For example, Kami and Lugulu are positioned to represent their closer contacts with Sagala than with Gogo. Doe and Kwere are also positioned closely together to account for their high-end-of-the-range cognate percentages with each other. For similar reasons, Zalamo and Kutu are positioned closer to Kwere and Kami than they are to Doe and Lugulu or to Kagulu or the West Ruvu group.

Full Schemata 1

Mapping Ruvu Divergence on the Tanzania Landscape

45In order to hypothesize the movement of people and their languages in physical space, it is helpful to use a theory called the "principle of fewest moves." This involves the application of a familiar scientific principle that stipulates that the soundest explanatory model is the one that accounts most parsimoniously for the evidence. As human communities ebb and flow, leading eventually in some instances to community expansion across geographical space, they usually do it in relatively small groups and at a gradual pace, and only rarely, especially before recent times, in movements characterized by large, sudden mass migration. Applied to language divergence, the principle suggests that "whichever history requires the fewest shifts of ethnic and language boundaries to explain the modern-day ethnic and speech territories is the best and most probable solution."11 Employing this tenet gives us a way to postulate the core settlement area for the proto-Ruvu language and to suggest the subsequent movements that its daughter-language speakers took that resulted in their covering large regions of the central-east Tanzania hinter-coastal zones in which Ruvu descendant speakers now live.

46 In addition to the maps illustrating proto-Ruvu divergence, there are four maps appended to the sequence that predate proto-Ruvu emergence in central-east Tanzania. These are included to provide a sense of relative relationship among proto-Ruvu-descended languages and their linguistic siblings as well as their relationship with their common ancestral language. They are introduced here but will be central to the discussion in chapter 2. Maps 1 and 2 depict the region in the eras prior to the emergence of proto-Northeast-Coastal Bantu. Map 3 illustrates the historical period when Ruvu's earliest direct ancestral language, proto-Northeast-Coastal Bantu, was first spoken in central-east Tanzania. The fourth map shows the period when proto-Ruvu's ancestral proto-language, proto-Wami, diverged from proto-Northeast Coastal Bantu.12 Also found within the maps are the names of non-NECB languages noteworthy in various historical periods because they played a part in Ruvu history. It is aspects of those developments that are discussed throughout the remaining chapters.

Map 1: Azania Bantu, c. 200 BCE-1 CE Map 2: Azania and Upland Bantu, c. 1-200 CE
Map 3: Proto-Northeast Coastal Bantu, c. 300 CE Map 4: Northeast Coastal Bantu Language Groups, c. 500-600 CE

47When conceptualizing maps using the principle of fewest moves one works in reverse-chronological order to postulate the historical position of language communities. However, the text illustrations are herein arranged in chronological order so as to represent the way language divergence played out within central-east Tanzania from early to later times. To illustrate the principle of fewest moves as applied in positioning the proposed language communities, refer to the two map slideshows. These are arranged in both chronological and reverse-chronological order.

Map Slideshow 1:
c. 200 BCE to 1700 CE
Map Slideshow 2:
c. 1700 CE to 200 BCE

48 Map 5: Northeast Coastal Bantu Language Groups, c. 600 CE In the period 300–600 CE proto-Northeast-Coastal Bantu likely began a sequence of divergences. The first of these resulted in ancestral Asu speakers settling into the South Pare Mountains and in proto-Sabaki speech communities spreading north of the Pangani River, along the narrow stretches of the Kenyan coast. Proto-Wami speech communities then emerged among extant Northeast-Coastal Bantu communities, who made their homes within the area bordered by the Ruvu and Pangani Rivers. Then in an ensuing stage of divergence, the proto-Wami community evolved into two speech communities, proto-Seuta and proto-Ruvu. The most probable proto-Ruvu homeland was in an area that lay between the lower Wami and Rufiji Rivers. The proto-Seuta zones would have lain to the north of the proto-Ruvu in the area between the lower Wami and Pangani Rivers.

49 Map 6: Northeast Coastal Bantu Language Groups, c. 800-900 CE The proto-Seuta language did not begin to diverge until about 1000 CE. But its sister language, proto-Ruvu, started to diverge into nascent language groups between the seventh and eighth centuries CE. Today, existing Ruvu languages trace their ancestry back to those three primary branches. The cognation-retention ranges of Kagulu with the descendant languages of both the proto-West and East Ruvu speech communities suggest that the three initially formed a chain of language communities spread along an area between the Wami and Rufiji Rivers. The main direction of expansion leading to the proto-Ruvu divergence would have been westward into drier interior areas, away from the most desirable and first-settled immediate hinter-coastal zones.

50Map 6 includes an additional proto-Ruvu-descended language community whose likely homeland locale has not been previously discussed. This proposed branch represents a posited fourth proto-Ruvu daughter language. We suggest this because the proto-Rufiji-Ruvuma-descended languages (proto-Rufiji-Ruvuma is a linguistic sibling of PNECB) spoken along the lower Rufiji River and to its south in the coastal hinterland attests word transfers whose etymologies are rooted in Ruvu history. Ruvu speakers living in proximity to them and probably intermingling with them were the apparent donors of those lexical items into Rufiji-Ruvuma languages. Adhering to the principle of fewest moves, the proposed, now extinct, South Ruvu group from which these loanwords came would have comprised communities that settled south of the Rufiji River, along the coast and next to the river itself.

51 Following the proto-Ruvu split into three or four daughter languages, a subsequent era of branching involving both the proto-West and the proto-East Ruvu societies took place during the early centuries of the second millennium, roughly 1100–1400 CE. The proto-West Ruvu language evolved into the ancestral Gogo and a short-lived proto-Vidunda-Sagala community, which soon thereafter diverged into distinct Vidunda and Sagala speech communities. The proto-East Ruvu society gradually evolved into the ancestral Lugulu, proto-Central-East Ruvu, and ancestral Doe groups. West-Ruvu-speaking people at that time would have moved west into new lands and interacted with Njombe-speaking peoples who lived in or, more probably, were moving north into that region in those centuries. In the second half of the second millennium, the proto-Central-East Ruvu evolved its daughters, Kwere, Kami, and proto-Southeast Ruvu. As they expanded, some would have encountered non-Bantu Parakuyu Maasai speakers living in the region. Then, within the next few centuries, c. 1700, the most recent members of the East-Ruvu language group, the Zalamo and Kutu, emerged. This final map shows the approximate settlement areas of the ten present-day languages comprising the Ruvu branch of the Northeast-Coastal Bantu subgroup.

Map 7: Ruvu Peoples and Neighbors, c. 1400 CE Map 8: Ruvu Peoples and Neighbors, c. 1500 CE Map 9: Ruvu Peoples and Neighbors, c. 1700 CE

52The mapping sequences provide the most immediate way to illustrate the probable approximate earlier geographical positions of the genetically related Ruvu relatives as well as their more distantly-related Bantu relatives and various of their genetically unrelated but historically relevant linguistic neighbors. But perhaps just as importantly, mapping the various branches of Ruvu subgroups and its neighbors through time provides a canvas upon which the histories that are reconstructed from the general language data can be written. This is because each time a word's etymology is reconstructed to a proto-language period, it can be argued that its associated ideas or material meaning comprised an attribute of the society of the particular proto-language speakers. That reconstruction in turn creates an opportunity for claiming in a very general sense, based on historical evidence, the prevalence of a community cultural feature.13 In this way the historian who uses language opens avenues into the rich data that comparative linguistics combined with ethnography provides for reconstructing early history.

53The next section of the chapter introduces the deeper historical context that underpins the proto-Ruvu period. Its emphasis is not on establishing the genetic relationships of languages, but on highlighting the point that the Ruvu languages are but one subgroup of many languages rooted in much deeper histories going back to the proto-Mashariki and often to the still much earlier proto-Bantu and proto-Niger-Congo eras. The goal is to highlight the relevance of recognizing that when word etymologies are reconstructed for words spoken by contemporary Ruvu speakers throughout this work, it is often the case that elements of the cultures of much earlier ancestral language communities are revealed because those words commonly represent inherited material culture, ideas, and the like that once characterized communities of long, long ago.


The Niger-Congo Family and Its Bantu Branch

54The genetic relationships of the Ruvu subgroup to Bantu as a whole and of Bantu to Niger-Congo are important because much of this work makes arguments based on a history of shared words and meanings across languages. Language is more than just an aspect of culture; it represents "the collective memory bank of a people's experience in history."14 Moreover, because the common historical descent of the languages that form this "memory bank" go back to these distant common ancestries, the myriad communities that speak the various descendant languages have transported numerous enduring, early shared cultural expressions, both abstract and material, throughout what can be termed Niger-Congo and Bantu diasporas.15 Because of this, one of the purposes here is to uncover, whenever they appear in the evidence, the cultural continuities stemming from the shared historical background of the Ruvu languages—and the societies that speak them—with other societies sharing this historical ancestry. This approach is inspired by the goal of insuring that the reader is aware of the deep-seated continuities that Bantu-speaking people in central-east Tanzania preserved along with the changes in that heritage that their own particular historical experiences inspired. By having an understanding of what characterized pre-Ruvu societies, we gain a better perspective on what was different and new in developments that took place in the past 1,400 years among the Ruvu.

55Of the four language families (sometimes referred to as phyla) indigenous to Africa, the Niger-Congo language family contains the largest number of extant languages on the continent. It is, in fact, the largest known language family in the world. Of the estimated 2,035 languages spoken on the continent of Africa, a recent study estimated the Niger-Congo family comprises 1,436 languages, with a conservative estimate of 360 million speakers.16 Like other language families indigenous to the continent—Afrasan (Afroasiatic), Nilo-Saharan, and Khoisan—the Niger-Congo language family long, long ago comprised communities of speakers who spoke a common ancestral language, proto-Niger-Congo. It is thought that the proto-Niger-Congo speakers lived in the inland areas of the Niger River by approximately 10,000 BCE.17 Over time the proto-Niger-Congo community diverged into a number of distant communities who settled in disparate regions. As these communities multiplied and expanded over wider and wider territories, their linguistic differences increased. The proto-Niger-Congo language diverged into many mutually-unintelligible daughter languages.18 Over the long term of thousands of years, this pattern of speech community divergence and language emergence repeated many times. The original daughter languages of proto-Niger-Congo underwent further divergence, giving rise to their own daughter languages—and in so doing became intermediate proto-languages—and these daughter languages underwent further periods of divergence of their own. This recurrent process over the long term brought into existence a deeply diverse family of hundreds of languages spoken by a great diversity of societies.19

56In other words, over the past ten thousand or more years, between the proto-Niger-Congo period and the present, the Niger-Congo language family evolved many languages that were in their own right ancestral to others. They were proto-languages. While the precise genetic relationships, that is—how each of the descendant languages are related to one another—is a matter of lingering debate, there is consensus that the Niger-Congo family comprises an overarching genetic unit.20 Of the many proto-languages evolved from Niger-Congo and their later descendants, none has garnered more scholarly attention than that of the Bantu branch.

57A great deal of energy and effort has gone into sub-classifying Bantu languages. Such efforts have spawned debates that range from whether or not a language is Bantu at all, to the particulars of correctly sub-classifying its more than five hundred descendant languages and dialects.21 While each debate is worthy of attention in its own right as a means of understanding the genealogical relationships of the language family, as in the Niger-Congo case, there is general consensus that a proto-Bantu community took shape during the fourth millennium BCE.

58Applying the principle of fewest moves to Niger-Congo descendants researchers propose that proto-Bantu speakers set up their communities somewhere in the regions of the Sanaga and Nyong Rivers of modern-day southern Cameroon as many as 5,000 years after their earliest Niger-Congo-speaking ancestors lived. Over the course of subsequent centuries, Bantu communities diverged along water pathways shaped by extensive river mosaics that took them south and southeastward. The creation of a number of Bantu language subgroups resulted from their movement. By the second millennium BCE, one Bantu speech community, the ancestral Savanna-Bantu society, came into being in the rainforest covered regions where the Kwa and Congo Rivers converge. While some of those communities remained close to that region, by the middle of the second millennium BCE, a number of Savanna-Bantu descendant communities made their way eastward, probably along the Kasai and Sankuru Rivers. Of eventual wide-ranging impact for much of the eastern and southern continent was one easterly lying branch of Savanna-Bantu descendants, the proto-Mashariki communities, who by 1000 BCE made their homes near the Western Rift just west of the African Great Lakes. Their presence in the region ushered in a time of economic and cultural transformation over the following five centuries in areas around the Western Rift zone. Then, during the last few centuries BCE, and lasting into the fifth century CE, Mashariki-Bantu descendants engaged in a relatively swift expansion that took them as far east as the Indian Ocean, north to Uganda and Kenya, and south eventually as far as Kwazulu-Natal.22

59By the middle of the first millennium BCE, the proto-Mashariki Bantu had diverged into what became two Mashariki Bantu subgroups, Kusi and Kaskazi. The proto-Kusi language then evolved into at least four descendant language subgroups—the Nyasa, Makua, Shona-Sala, and Southeast Bantu—whose communities reside today largely in the southeastern and far southern areas of the African continent. Proto-Kusi's sister language, proto-Kaskazi, eventually diverged into a number of subgroups of its own. They include, among others, Lakes Bantu, Takama, Upland Bantu, Njombe, Northeast Coastal, Langi, Chaga-Dabida, Thagiicu, Rufiji-Ruvuma, Kilombero, and Rukwa.23

60 Map 3: Proto-Northeast Coastal Bantu, c. 300 CE Among the several early Mashariki-descended communities whose languages came to be spoken in Tanzania, it was the proto-Northeast-Coastal Bantu society's language that was directly ancestral to the Ruvu languages, as was noted earlier. Of the four primary Northeast Coastal Bantu subgroups—Seuta, Ruvu, Asu, and Sabaki—the Sabaki group has so far garnered the most scholarly attention. The history of the Swahili language (a Sabaki descendant) and Swahili societies grabbed the interest of scholars early on because of the economic, political, technological, and religious transformations that the early Swahili instigated as mediators of the encounter of hinterland East Africans with transoceanic commercial systems of the Indian Ocean. Because of disproportionate interest in Swahili history vis-à-vis other Northeast-Coastal Bantu language descendants, scholars have considered Swahili history as significant while largely glossing over other NECB histories, even though much of NECB history preceded Swahili emergence and in contemporaneous times paralleled and supported Swahili accomplishments.

61In the following discussion, the unevenness of the historiography is addressed. I maintain that the evident unevenness is not an attempt to ignore overtly what transpired before and in tandem with Swahili history, but rather that it is the result of an unfortunate tendency to assume that the Swahili achievements are occurred because of their historic associations with people and places external to the continent. As a consequence, not much attention is paid to interior, continentally based historical developments and people that may have brokered Swahili achievements.


A Case of Swahili Exceptionalism

62There is no questioning that ancestral-Swahili-speaking people living along the central regions of East Africa's hinter-coast succeeded in developing transoceanic networks to rival any of their time. Similar to histories of Dynastic Egypt, of the Mali Empire, and of the Kingdom of Kongo, the Swahili city-states represent in the eyes of many people one of a handful of Africa's "model minority" civilizations of the past. They have been used to demonstrate to naysayers that Africa has had its share of political and economic successes. That historians have revelled in such research is not surprising, particularly when one considers that the discipline of African history was founded in tandem with the dismantling of European occupation of the African continent. Viewed from that perspective, it is easily understood why the focus on recovering African political and economic histories in a time of emerging nationhood made good sense. People with a concern for the continent's future looked for evidence of indigenous, effective governance and economies that they rightly speculated were part of Africa's pre-colonial past. For general supporters of Africa, the strength of the African past was at once a source of pride as well as a potential reservoir of nation-building strategies.

63Beyond that, proving that Africa had played a major role on the world stage carried an additional advantage. It meant that the continent and its people could stake their claim and place in the decades-old idea of progress and civilization that had excluded them by definition. One might say that Africans, and their external supporters, may have felt pressured to sort out Africa's economic and political histories for the sake of their future in a world where nation-states and urban life defined the essence of modernity.

64Archaeologists were among the first to recover evidence of technological achievements likely instigated by coastal-dwelling people in Tanzania who erected impressive towns. And on top of that, there was later archaeological evidence that suggested people living and working along the Swahili Coast had been persistent agents in transoceanic networks connecting Africa with contemporaneous European and Asian societies, economies, and politics since as early as the late first millennium CE. Following up on that sort of evidence, the history-centered scholarship of the last few decades has turned increasing attention to the approximate1600-mile stretch of Kenyan and Tanzanian coastline, with the furthest reaches inclusive of areas in Somalia and Mozambique.

65Since the 1980s, the predominant focus of those histories has been realigned significantly, turning away from early presumptions that Swahili accomplishments were the consequence of relationships instigated and dominated by Arab and Persian foreigners. In the 1990s, research efforts have given voice to the importance of indigenous, African agents responsible for Swahili achievements. While this shift has been an important turn, it still falls short of a holistic approach to the historical events that unfolded along the coast, because to this point this new path of inquiry about the Swahili past remains focused predominantly on the narrow coastal corridor and its intercontinental partners. Consequently, the homelands and histories of people who lived beyond the narrow corridor of the "Swahili Coast" have been inattentively referred to and historically glossed as the hinterland. As a result, historians have, for the most part, settled on a Swahili periodization of Indian-Ocean and East-African-coastal history that for all practical purposes begins in the ninth century CE.

66With new evidence continually brought to bear because of the work done by determined archaeologists, the coastal economic and political achievements are supported by mounting evidence that East Africa was indeed home to historical events of worldwide consequence. At the same time, the disparity in the historiography that has developed, however subtle or unintentional, maintains the impression that those coastal histories begin and end in the tight stretches of the coastal "Swahili Corridor." The outcome has been that the principal, popular narrative characterizing coastal history is largely synonymous with "Swahili history and the hinterland."24

67More than fifty years after the first African nations achieved independence we should ask why historians' eyes remain preoccupied with "progress," which has largely emphasized the history of societies through the analytical lenses of politics and economics. To be fair, however, the predisposition to center on "progress" is not particular to African history. Instead, it is the result of a general bias fostered by the idea that what is believed historically worthy are those societies that have "progressed" to the point that they are characterized as tantamount with "civilization." But why is this so?

68Perhaps an answer is partially found in the opening of Vansina's recent How Societies Are Born: Governance in West Central Africa Before 1600.25 He poses forthrightly the question that begs asking, "Why focus on governance?" In his view, "local communities did not unite and form a single, fully fledged community until common overarching institutions of governance were formed." And this is important because, "common governance creates and maintains societies, which, in turn, form the matrix in which much else flows."26 The implication is that government is the key to understanding societies. 27

69Vansina's response is telling because it underscores a rationale underpinning many historical studies, namely, that government sustains the flow of progress. But, is this correct? Is this too narrow a premise? The presumption of the central importance of governance and politics subtly, if unintentionally, overlooks the possibility that "progress" might be the cumulative effect of intersecting, intermingling, and long-enduring sociocultural minutiae from which governing ideas and institutions flow, or that all together different sets of circumstances are involved.

70Though I do not suggest that politics and economics are unimportant in history, I do question whether this lens, which seems pervasive in studies of Africa generally, is the most fruitful approach for recovering early African history. Echoing a point made by Jean and John Comaroff, it seems that such emphases fail to entertain the possibility, that "history involves a sedimentation of micropractices into macroprocesses, a prosaic rather than a portentous affair in which events mark rather than make the flow of existence."28 In an attempt to try out an alternative approach, primary among the questions I ask is whether religion might have been the realm of cultural expression that sustained Ruvu societies.


Religion in Ruvu History

71The catalyst for this line of inquiry was inspired by an opinion shared more than three decades ago by a group of Africanist scholars who emphatically advised that to understand politics and economics, as well as other sociocultural institutions in Africa, it was critically important to investigate religious history.29 As they proposed, " . . . it is impossible to abstract religious history from political or economic or social history without distortion, as it should also be impossible to abstract political history without similar distortion."30 The problem is not that government, economy, or religion are more or less important in societies, but rather that there is a tendency to define and view them as divisible aspects of polyvalent societies.

72To his credit, Vansina recognizes and addresses the function and relevance of religion in How Societies Are Born, as well as his corpus of scholarship on Africa more generally. But in my reading, he assesses it from the standpoint of religion's being one of those matters that "flows" because of government. In so doing, he brings to the forefront a paradox with which scholars often deal when addressing the intersection of abstract ideas with that of their physical expression, what Vansina calls the relationship between cognition and reality.31 As anthropologists and sociologists have debated over the life of their disciplines, speaking to these intersections becomes a chicken-and-egg problem of whether social structure influences religious ideas and meaning—in this case—or religious ideas and meanings breathe life into social institutions. Going down that intellectual path that would take us into complex discussions about such theoretical frameworks as functionalism, structuralism, Marxism, intellectualism, and postmodernism, among others.32 Although I do think it is important to acknowledge these scholarly approaches because they are important in eventual theory-building, at the same time too much of the focus on the paradigmatic aspect of the debate seems, at least to a point, to neglect interest in the growth of knowledge in favor of the interest of engaging in scholarly debate. While conjecture is important as a beginning, it cannot move far without more complete understandings of the historical circumstances, events, and ideas that would facilitate solid theory-making. In other words, it seems unlikely at this point that better informed theories are possible until we know more about the details of Africa's early history.33 After all, "both physical and cognitive realities are equally valid at the ontological level."34 When all is said and done each is historically important, no matter which came first—on that we should be able to agree.

73 What I would like to do is to present this book as an offering toward early history "progress-making" built around the proposition that Bantu religion—which is defined here as an understanding that spirit forces have a place in cognitive and physical realities—is essential to recovering Bantu history. This is because Bantu people, like people generally, conceptualized their worldviews from the foundational way they explained the cosmos. The corollary of this is that Bantu people, like their Niger-Congo ancestors, imagined the corporeal and spiritual worlds as interlinked realms. Spirits influenced the sphere of the physical as well as the reverse.35 Their ideas about religion, their abstract reasoning, their cultural practices, and their social institutions were based on and expressed from the perspective of a worldview that assumed such truths. Beyond this, dealing with the vicissitudes of existence—from very local levels of interaction, whether individual or collective, local or global, to strategies aimed at wielding far-flung power, taken here as "the means by which people are believed to gain access and control over people and resources through mastery of transformation processes [emphases added],"36—for these people and their societies commonly involved spirit forces.37

74At the same time, while effectuating power may have involved spiritual forces, these were not the only constituents that mattered. Rather, we should view them as core ingredients in a complex worldview that guided people in the negotiation of various means of achieving power. From disease prevention to environmental rejuvenation to social order, it was up to people to configure the appropriate line of power-making to solve, or, better yet, prevent problems. Across Bantu societies, I suggest, then, that it was taken for granted—it was epistemological—that in creating social institutions, from matriclans to centralized kingdoms, and in defining social and political roles, from birth attendants (midwives) to diviner-doctors, from chiefs to rainmakers, spirits mattered because they played a part in accessing and making power.

75It is thus in the spirit of what a cohort of Africanist scholars suggested more than three decades ago that I am curious about what can develop if we center religion as not incidental to any institution—at least for now—but rather as a dominant prism through which early Bantu history is viewed. Furthermore, I am interested in bringing to bear a set of particular questions throughout the work. How did deep-seated locally interpreted religious knowledge contribute to shaping the social institutions that Bantu descendants imagined and created? And, is it possible that all social institutions—both ethereal and physical—engaged in continual dialectic and discursive exchanges? So in the end it does not matter which came first; rather it opens up the opportunity to center our attention on matters of historical recovery.

76But in order to open up to such alternative approaches, there must be a conscious step away from a familiar worldview in which there is a dichotomy between a secular world separate from that deemed spiritual or religious.38 This is necessary because, as far as the available historical evidence suggests, such realms of separation were not conceptualized in early Niger-Congo or Bantu ideological frameworks. While physical space may have been bounded, ethereal forces were not. This point is important because it confronts a common assumption that is too often telescoped into the past. Moreover, and for a slightly different reason, it is important because there may be detractors who question as essentialist in nature the intentions of people who propose that religion may be a core element in Bantu worldviews. While this may be a legitimate concern when it comes to such matters, it is important to be cautious about jumping to conclusions.39

77The belief in ancestors or spirits as omnipresent historical agents has opened up the door for admonitions of well-intended people who have been charged with perpetuating depictions of a "primitive Africa." I propose that we need take an alternative point of view because there is a real reticence that can result from the fear that one may be accused of propaganda-making, which can silence the sharing of evidence we have in hand—namely, the considerable testimony indicative of the centrality of the realm of the ethereal in Bantu history. For that reason, in my view, it must be made clear that the suggestion that religion should not be everywhere embodies the biases of modern ideas of "progress" and "world-religions paradigms." These constructs since the late eighteenth century have maintained that "legitimate" religions are theological and intellectual matters, of which African religions, by definition, could have no part because they are deemed inherently "primitive." But, "[i]f we allow that historical consciousness and representation may have taken very different forms from those of the West, people everywhere turn out to have had history all along."40 Buying into the presumption that religion should be expressed in the "West's" way silences the recovery of history in Africa.


Challenges in Writing the History of Religion in Africa

78In my view, the "primitive religion" narrative and other parallel phraseology embodies two streams of sometimes contentious but nonetheless entwined ideas about Africa that ultimately and often insidiously permit ineffective and negligent approaches to historical studies of religion and history in Africa. The first is the idea of progress, a framework refined by late eighteenth-century French philosophers, which is premised on what philosopher Michel Foucault identifies as a "modern style of reasoning."41 It then became entrenched in developing eighteenth- and nineteenth-century racist notions that touted the intrinsic, biological exceptionality of people of presumed European/Caucasoid genetic origins. Over time, particularly in the twentieth and twenty-first centuries, it has been recast, though with underlying similar notions about race, as those things identified with the "West" or "civilization." Inherent in this presupposition is the idea that progress is the privilege of Europeans' descendants, out of which biased ideas about Africans and others are fostered and accepted as truth.

79A second trope misdirecting the study of religion in Africa is the "world-religions" paradigm, which is not unrelated to the progress model. The world-religions paradigm is shrouded within the modern-reasoning paradigm. It thus represents a problematic issue warranting particular attention because its predominant beneficiaries are those religions that convey evidence of historical relevance and progress as defined by an often-biased intellectual community. Fundamentally, the construct exists because of Judaism, Christianity, and Islam (Abrahamic religions). In effect its "world" status qualifier demands the need for those religions unworthy of dwelling within its definition, i.e., non-world religions. In historian Elizabeth Isichei's view, the rationale for the "world-religions" nomenclature rests on the presumption that they are "historical" religions, while its antithesis, "traditional religions," are not.42

80The typical expression used currently to refer to religions indigenous to Africa is "African Traditional Religion," or some near equivalent.43 What is particularly striking is that presumptions about the nature of religion in Africa as being essentially ahistorical are embedded in such qualifiers. Consequently, African societies, religion, and knowledge are assessed from a viewpoint that is often construed as evidence of societies that are "traditional" in the sense that they are inefficient, unchanging, and, at worst, "anti-progress"! Remedying this demands that people who want to contribute to dismantling such erroneous ideas must get to the root of the "problem of religion" in their historical studies of religion and society. But to do that, it is necessary to examine the ways in which histories require an understanding of the local knowledge—epistemology—that establishes the context of its rationale and expression.

81While many scholars have challenged ideas of a static African historical past, those challenges have largely addressed secular matters—economics and politics—and not those often implicated in the realm of the religious, metaphysical, supernatural, or "primitive" religion. And while on the one hand it may be said that knowledgeable people reject ideas of a static temporal past in Africa outright, this does not appear to inherently transfer to or challenge predominant thinking about Africa's indigenous religions and their implication for history. While this may be expected, it is problematic because as Ranger and Kimambo suggest, "it is impossible to abstract religious history from political or economic or social history" without delivering a warped interpretation of the overall picture of history.44 The dynamism in African religious history is not done any good by trying to define it in terms of western paradigms that belong to functionalist, structuralist, or intellecualist approaches to understanding social histories. And it would be at this point unfruitful to argue that African social institutions were absolutely guided by religious ideas or that all people intellectualized their worlds in the same way. At the same time, we should not regard these as impossibilities. Being mindful of broad theoretical possibilities permits us to take the best of what scholars have had to offer in the development of approaches to religious history in Africa. But we also must have an eye on the theoretical models that emerge from episteme anchored in African societies.45

82What we lack are inquiries into the deep histories of religion, knowledge, and society in Africa. The challenge is to begin to fill in the profound gap in the depth of our historical knowledge. So the intention here is not merely to expose entrenched bias, but rather to contribute to a "building upon" trajectory that leans on some of the well intentioned work of pioneer scholars who have been concerned with how best to reconstruct African history and religion's place in it. To be sure, it is important that our work on local histories, regardless of geographic breadth, strives to play a part in the construction of early histories. Those histories represent the epistemologically rooted contexts in which later people lived. To move in this direction, we turn to the ways that Ruvu people shaped their societies and thus shaped history.


Note 1: W. H. I. Bleek, A Comparative Grammar of South African Languages, Part 1, Phonology (London: Trübner, 1862); J. H. Greenberg, Essays in Linguistics (Chicago: University of Chicago Press, 1957); Joseph Harold Greenberg, The Languages of Africa (Bloomington: Indiana University Press, 1963). back

Note 2: Christopher Ehret, An African Classical Age: Eastern and Southern Africa in World History, 1000 B.C. to A.D. 400 (Charlottesville: University Press of Virginia, 1998); Jan Vansina, Paths in the Rainforest: Toward a History of Political Tradition in Equatorial Africa (Madison: The University of Wisconsin Press, 1990). back

Note 3: Kairn A. Klieman, "The Pygmies Were Our Compass": Bantu and Batwa in the History of West Central Africa, Early Times to c. 1900 (Portsmouth, N.H.: Heinemann, 2003); Christine Saidi, Nacimbusa: Mothers and Daughters in East-Central African History, 500 BCE to 1800 CE (New York: University of Rochester Press, forthcoming); David L. Schoenbrun, A Green Place, A Good Place: Agrarian Change, Gender, and Social Identity in the Great Lakes Region to the 15th Century (Portsmouth, N.H.: Heinemann, 1998). back

Note 4: Ehret, African Classical Age, 36, 186–88; T. Hinnebusch, D. Nurse, and M. Mould, eds., Studies in the Classification of Eastern Bantu Languages (Hamburg: Helmut Buske Verlag, 1981); G. Philippson and D. Nurse, "The Northeastern Bantu Languages of Tanzania and Kenya," 3. back

Note 5: Hinnebusch, Nurse, and Mould, eds., Studies in the Classification of Eastern Bantu Languages; Derek Nurse and Thomas J. Hinnebusch, Swahili and Sabaki. back

Note 6: Christopher Ehret, "Language and History," in African Languages: An Introduction, ed. B. Heine and D. Nurse (Cambridge: Cambridge University Press, 2000), 287. back

Note 7: See Sheila Embleton, Statistics in Historical Linguistics (Bochum: Studienverlag Brockmeyer, 1986) and S. Starostin, "Comparative-Historical Linguistics and Lexicostatistics," in Time Depth in Historical Linguistics, Vol. 2, ed. A. McMahon C. Renfrew and L. Trask (Cambridge: McDonald Institute for Archaeological Research, 2000). back

Note 8: Klieman, "The Pygmies Were Our Compass," xxvii–xxix. back

Note 9: The evidence supporting the continued value of glottochronology is most convincingly seen when again and again there are correlations found between archaeology and linguistic based conclusions. See, for instance, Stanley H. Ambrose, "Archaeology and Linguistic Reconstructions of History in East Africa," in The Archaeological and Linguistic Reconstruction of African History, ed. Christopher Ehret and Merrick Posnansky (Los Angeles: University of California Press, 1982); Felix Chami, The Tanzanian Coast in the First Millennium AD: An Archaeology of the Iron-Working, Farming Communities (Uppsala: Societas Archaeologica Upsaliensis, 1994); Christopher Ehret and Merrick Posnansky, eds., The Archaeological and Linguistic Reconstruction of African History (Berkeley: University of California Press, 1982). back

Note 10: Christopher Ehret, "Testing the Expectations of Glottochronology against the Correlations of Language and Archaeology in Africa," in Time Depth in Historical Linguistics, Vol. 2, ed. A. McMahon C. Renfrew and L. Trask (Cambridge: McDonald Institute for Archaeological Research, 2000), 395. back

Note 11: Christopher Ehret, "Bantu Expansions: Re-envisioning a central problem of early African History," The International Journal of African Historical Studies 34, no. 1 (2001): 37. back

Note 12: For information on prior language classifications referenced in reconstructing these linguistic periods, see Ehret, African Classical Age; Tom Güldemann and Rainer Vossen, "Khoisan" in African Languages: An Introduction, ed. Bernd Heine and Derek Nurse (Cambridge: Cambridge University Press, 2000); Hinnebusch, Nurse, and Mould, eds., Studies in the Classification of Eastern Bantu Languages; Derek Nurse, Description of Sample Bantu Languages of Tanzania, vol. 5 no. 1, African Languages/Langues Africaines (London: International African Institute, 1979); Nurse and Hinnebusch, Swahili and Sabaki; Philippson and Nurse, "The Northeastern Bantu Languages of Tanzania and Kenya." back

Note 13: For a useful discussion of just what a proto-word represents for the writing of history and as an aspect of one's research design, see Vansina, Paths in the Rainforest, 32–3. back

Note 14: Ngugi wa Thiong'o, Decolonising the Mind: The Politics of Language in African Literature (Portsmouth: Heinemann, 1981), 15. back

Note 15: Ibid., 4–13. back

Note 16: The estimate for Afrasan (Afroasiatic) is 371; Nilo-Saharan, 196; and Khoisan, 35. See Kay Williamson and Roger Blench, "Niger-Congo," in African Languages: An Introduction, ed. Bernd Heine and Derek Nurse (Cambridge: Cambridge University Press, 2000), 2, 11. back

Note 17: Christopher Ehret, The Civilizations of Africa: A History to 1800 (Charlottesville: University Press of Virginia, 2002), 37. back

Note 18: As those speech communities diverged their languages evolved first into dialects of the proto-Niger-Congo language. For a period these would have been mutually intelligible. This is analogous to the history of the common language ancestor once spoken by the ancestors of the Spanish, French, and Italian, for an Indo-European example. Secondly, the fact that they were new languages does not preclude people from being multilingual, which they likely were. back

Note 19: While no complete genetic reconstruction has been completed at this point, I am resting this in part on the perspectives of Williamson and Blench, who maintain, "given the extensive sound and meaning in particular grammatical elements as well as in basic vocabulary, and the complex changes they have undergone, it is inconceivable that this could be due to chance and extremely unlikely that it is due to borrowing. Moreover, there are many words of basic vocabulary which recur throughout Niger-Congo, often in modified form after undergoing sound changes." See Williamson and Blench, "Niger-Congo," 13. back

Note 20: Ibid., 11. back

Note 21: See, among others, Ehret, "Bantu Expansions"; Greenberg, Essays in Linguistics; Malcolm Guthrie, Comparative Bantu: An Introduction to the Comparative Linguistics and Prehistory of the Bantu Languages, Vol. 3 and 4 (Farnborough: Gregg, 1970); Kairn A. Klieman, "Hunters and Farmers of the Western Equatorial Rainforest Economy and Society, 3000 B.C. to A.D. 1880" (Ph.D. thesis, University of California Los Angeles, 1997); J. Vansina, "New Linguistic Evidence and 'the Bantu Expansion,'" The Journal of African History 36, no. 2 (1995): 173–95. back

Note 22: Ehret, Civilizations of Africa, 113–5. back

Note 23: For detailed coverage on Bantu sub-classification in these regions, see Hinnebusch, Nurse, and Mould, eds., Studies in the Classification of Eastern Bantu Languages; Saidi, Nacimbusa; Catherine Cymone Fourshey, "Agriculture, Ecology, Kinship and Gender: A Social and Economic History of Tanzania's Corridor 500 BC to 1900 AD" (Ph.D. thesis, University of California Los Angeles, 2002); Philippson and Nurse, "The Northeastern Bantu Languages of Tanzania and Kenya." back

Note 24: In 2000, historian Thomas Spear reported that in the previous fifteen years "some 270 books, theses, and articles" centered on Swahili history. He also commented, "This degree of scholarly activity is as welcome as it is unusual in African history. . . . " See Thomas Spear, "Early Swahili History Reconsidered," The International Journal of African Historical Studies 33, no. 2 (2000). back

Note 25: Though there are other historical works I could draw upon for examples, I turn to Vansina's because he is among the most influential historians of early African history. back

Note 26: Jan Vansina, How Societies Are Born: Governance in West Central Africa Before 1600 (Charlottesville: University of Virginia Press, 2004), 2. back

Note 27: Ibid., 3. back

Note 28: John Comaroff and Jean Comaroff, Ethnography and the Historical Imagination (Boulder, Colo.: Westview Press, 1992), 30. back

Note 29: The setting for the production of the edited volume was in part fueled because of problematic works newly published at the time, which assumed that Africa either had no true history of religion or, alternatively, that it had religions synonymous with ideas of monotheism commonly associated with twentieth-century Christianity. The latter was an attempt to make a claim that African religions were fundamentally akin to Christianity and therefore just as progressive. Ranger and Kimambo, eds., Historical Study of African Religion. For examples of the works they were addressing, see John S. Mbiti, African Religions and Philosophy, 2nd. (Portsmouth, N.H.: Heinemann, 1989); Geoffrey Parrinder, Religion in Africa (New York: Praeger Publishers, 1969). For works that were centered on ideas of an African history that included religion, see E. E. Evans-Pritchard, Theories of Primitive Religion (Oxford: Clarendon Press, 1965); Okot p'Bitek, African Religions in Western Scholarship (Nairobi, Kenya: East African Literature Bureau, 1970). back

Note 30: Ranger and Kimambo, eds., Historical Study of African Religion, 2. back

Note 31: Vansina, Paths in the Rainforest, 71–73. back

Note 32: For a short synthesis of the debates, see André Droogers, "From Waste-Making to Recycling: A Plea for an Eclectic Use of Models in the Study of Religious Change," in Theoretical Explorations in African Religion, ed. Wim van Binsbergen and Matthew Schoffeleers (Boston: KPI Limited, 1985); Harold W. Turner, "The Way Forward in the Religious Study of African Primal Religions." back

Note 33: Comaroff and Comaroff, Ethnography and the Historical Imagination, xi. back

Note 34: Vansina, Paths in the Rainforest, 72. back

Note 35: James W. Fernandez, Bwiti: An Ethnography of the Religious Imagination in Africa (Princeton: Princeton University Press, 1982), 39. back

Note 36: Eugenia W. Herbert, Iron, Gender, and Power: Rituals of Transformation in African Society (Bloomington: Indiana University Press, 1993), 2. back

Note 37: Comaroff and Comaroff, Ethnography and the Historical Imagination, 23–28. back

Note 38: V. Y. Mudimbe and B. Jewsiewicki, "Africans' Memories and Contemporary History of Africa," History and Theory 32, no. 4 (1993): 5. back

Note 39: Comaroff and Comaroff, Ethnography and the Historical Imagination, 9. back

Note 40: Ibid., 5. back

Note 41: Jacques, "From Savages and Barbarians to Primitives." back

Note 42: Elizabeth Isichei, The Religious Traditions of Africa: A History (Conn.: Praeger, 2004), 3–7. back

Note 43: See, for example, E. O. Babalola, Traditional Religion, Islam, and Christianity: Patterns of Interaction (Ile-Ife, Nigeria: Olajide Printing Works, 1992); E.O. Babalola, "The Reality of African Traditional Religion: A Yoruba Case Study," The Nigerian Journal of Theology 6 (1991): 50–63; A. Ekwunife, "The Priesthood in African Traditional Religion," Bigard Theological Studies 19 (1999): 5–39; S.N. Ezeanya, "Women in African Traditional Religion," Orita 10 (1976): 105–21; J. Lorand Matory, "Surpassing 'Survival': On the Urbanity of 'Traditional Religion' in the Afro-Atlantic World," The Black Scholar 30, no. 3–4 (2000): 36; Geoffrey Parrinder, African Traditional Religion (Westport, Conn.: Greenwood Press, 1976); Wade Clark Roof, "Traditional Religion in Contemporary Society: A Theory of Local-Cosmopolitan Plausibility," American Sociological Review 41, no. 2 (1976): 195–208; Shaw, "The Invention of 'African Traditional Religion,'" 339–353; Dominique Zahan, The Religion, Spirituality, and Thought of Traditional Africa (Chicago: University of Chicago Press, 1979). back

Note 44: Ranger and Kimambo, eds., Historical Study of African Religion, 2. back

Note 45: Droogers, "From Waste-Making to Recycling: A Plea for an Eclectic Use of Models in the Study of Religious Change," 101–114. back