East African Coastal History (original) (raw)
REVIEW ARTICLE
EAST AFRICAN COASTAL HISTORY
BY RANDALL L. POUWELS
University of Central Arkansas
Swahili and Sabaki: A Linguistic History. By Derek Nurse and Thomas J. Hinnebusch. Edited by Thomas J. Hinnebusch, with a special addendum by Gerard Philippson. (University of California Publications in Linguistics, i2I). Berkeley and Los Angeles: University of California Press 1993. Pp. xxxii +780. 880 (ISBN 0-520-09775-0).
Shanga. The Archaeology of a Muslim Trading Community on the Coast of East Africa. By Mark Horton. (Memoirs of the British Institute of East Africa, I4). London: The British Institute in Eastern Africa, 1996. Pp. xvi +458. £75£ 75 (ISBN I-872-56609-x).
I
Nurse’s and Hinnebusch’s Swahili and Sabaki : A Linguistic History is the most comprehensive study yet done of Swahili history through linguistic analysis. It is an encyclopedic work representing many years of research by the authors and other scholars, and it focuses particularly on the emergence and evolution of the Swahili language. The massive and diverse evidence they marshal is, of course, almost entirely linguistic: as such they discuss four basal parameters of language relationship and change, namely lexis, morphology, phonology and tone. (The last two are treated together, and G. Philippson reviews the latter.)
For the historian, most of this material is difficult reading, save perhaps for the chapter on lexis, and will be wholly comprehensible only to those having some training in linguistics. However, the authors provide a chapter which synthesizes their data into a comprehensive whole and which interprets local oral traditions and archaeological information to formulate an historical hypothesis which surpasses what their linguistic evidence alone offers. It is questions about the precise relationship between languages and the people who speak them that immediately raise further issues about the applicability and interpretation of language histories like this one to ‘history’, as most people understand the term. While a detailed discussion of this issue lies outside the scope of this article, and has been dealt with elsewhere, suffice it to say that the histories of language clusters, languages and dialects adhere closely to certain facets of the history of the peoples who speak them. 1{ }^{1} Historical linguists can provide historians with valuable evidence concerning cultural relationships between peoples, the emergence of distinct cultural groups and the relative chronology of these relationships and processes. However, the principal limitation of such evidence lies in the temporal and spatial relativeness of the processes it delineates. More to the point, it would
- 1{ }^{1} For excellent discussions of these issues, see Christopher Ehret ‘Linguistics as a tool for historians,’ Hadith I, ed. B. A. Ogot (Nairobi, 1967), 119-33; or the introductory chapters in idem., Southern Nilotic History (Evanston, 1971); and D. Nurse, ‘The contributions of linguistics to the study of history in Africa’, Fournal of African History, 38 (1997), 359-91. ↩︎
be very risky to rely on linguistic evidence alone to establish the absolute dating of the processes it describes. Moreover, linguists can provide only reasonable approximations as to where language-speaking groups ancestral to modern language-speakers might have been located. Because the interpretation of Nurse and Hinnebusch ventures into such terrain, this facet of their work invites critical commentary.
Essentially, the language history Nurse and Hinnebusch present adheres closely to one that has evolved among ethnolinguists over the past two decades and which was outlined in far less detail in an earlier volume Nurse co-authored with Thomas Spear. 2{ }^{2} This model holds that the original core Swahili-speakers, the ProtoSwahili, and their nearby hinterland neighbors stemmed from remote Eastern Bantu-speakers who first had inhabited the region between the Western Rift and Lake Nyanza sometime after 1000 b.c. Their closest coastal-hinterland ancestors had been the Proto-Northeast-Coastal Bantu, who the authors and other linguists believe had appeared in the interior between the Wami and Rufiji rivers, c. 200-400 A.D. Soon, the Northeast-Coastal grouping split into several divergent language clusters, including the Proto-Sabaki-speakers. Spreading (or migrating, as Nurse and Spear believe), along the coast and coastal hinterland, the Proto-Sabakispeakers were an extremely short-lived, though relatively compact community. According to Nurse and Spear, and Nurse and Hinnebusch, they migrated northwards across the Juba River to a location associated with the legendary Shungwaya, and there began evolving into separate proto-language-speaking communities, namely the Swahili, Mijikenda, Pokomo, Elwana/Malankote and Comorians. By late in the first millennium, this process had reached a very advanced stage. The Swahili then split into two, broad, dialect-speaking groups, Northern and Southern Swahili.
This paradigm also suggests that the ancient Swahili and their Sabaki-speaking relations inherited a host of cultural traits from their ancestors that included, among other things, a wetland farming tradition that later, with expansion into drier environments, saw the addition of cereal cropping, livestock breeding, limited fishing and iron-working (pp. 306-7). Some of these practices long before had been passed to their Eastern Bantu ancestors from Southern Cushitic-speakers who inhabited the region between Lake Nyanza and the Eastern Rift. Additionally, Northeast-Coastal- and Sabaki-speakers continued having significant contact with Southern Cushitic groups who dwelt around the headwaters of the Wami River, as well as in the Taita Hills (pp. 300-I). Southern Cushitic terms concerning cereals cultivation and livestock, for example, were borrowed in Northeast-Coastal and Sabaki. Later, Pokomo and Northern Swahili-speakers continued borrowing from Southern and Eastern Cushitic groups found in the Tana River area (p. 329). Indian and Persian terms concerning kinship (e.g. bibi and bwana), coconut and mango cultivation, cotton, fishing and spices also found their way into Sabaki and Proto-Swahili, attesting to the adaptations to the coastal environment that occurred in both those stages, as well as to new cultural influences (pp. 293-7, 317).
As in the earlier Nurse and Spear volume, Nurse and Hinnebusch indicate that most Arabic cultural influences date from this later period (pp. 302-4, 314-16). This book also includes fresh attestations concerning the post-sixteenth-century history of Swahili, for example evidence of a heavy preponderance of northern cultural influences, primarily from the Lamu archipelago, on southern dialects between the seventeenth and nineteenth centuries (pp. 506-10). This was followed in the nineteenth and twentieth centuries by the pre-eminence of Kiunguja, the Zanzibar dialect, which of course became the standard form of Swahili.
- 2{ }^{2} D. Nurse and T. T. Spear, The Swahili : Reconstructing the History and Language of an African Society, 800-1500 (Philadelphia, 1985). ↩︎
Although they do not cite it, independent evidence exists to support the last three points, underscoring the credibility of their work and that of other historical linguists as crucial to understanding the history of Africa. As will be discussed in detail below, the results of work by several archaeologists, including Horton, supports the view of an early coastal civilization in which exogenous influences were of a restricted nature in an environment that remained essentially African. Moreover, even that quintessential cultural import, Islam, existed on the coast until at least the seventeenth century in adapted and internalized forms among Muslim communities that had remained fundamentally local in outlook. 3{ }^{3} When Arabization and pressures towards greater ‘orthodoxy’ - i.e. forms influenced by greater access to written sources - did become noticeably greater and more extensive after the seventeenth century, they were due to a resettlement of saintly families from Southern Arabia and the Red Sea region who brought with them greater access to such sources. They initially settled in the Lamu region, but after learning the local dialects and even translating religious works into written KiSwahili, many remigrated southwards as far as Kilwa and the Comoro Islands. They brought with them their religious charisma which helped to perpetuate the cultural prestige, hence linguistic influence, of their Lamu regional homelands. As for the later pre-eminence of the Zanzibar dialect, this is a well-known phenomenon that requires no explanation here.
Such an endorsement, however, does not mean there are no problems in this work. I am less confident than they that Hinnebusch’s and Nurse’s dialect sampling, especially of the southernmost Swahili, was broad enough to form a reliable database; I would be more comfortable if they had made a greater effort to collect language data from the coasts of southern Tanzania or northern Mozambique. They themselves discuss this problem in their first chapter, and mention a number of important, but rare or extinct, forms such as the ‘old’ Mombasa, Kilwa, Mozambique and Kerimba Island dialects. Another concern, related to the first, is the need for more definition of the precise linguistic and historical relationships between the Northeast-Coastal languages and other Bantuspeaking groups to the west (the Corridor group) and to the south of coastal Tanzania. Archaeological research in recent years has indicated networks connecting this region not only with its immediate hinterland, but much farther south, at least as far as Natal. 4{ }^{4} Even the *-ni locative ending which, according to Nurse and Hinnebusch, distinguishes Proto-Sabaki from other Northeast-Coastal Bantu languages, exists in some Nguni languages.
Strangely too, considering Hinnebusch’s previously-stated caution about interpretations which venture beyond what strictly linguistic data afford, 5{ }^{5} this book continues to predicate the existence of a Proto-Sabaki homeland north of the Juba
- 3{ }^{3} Marina Tolmacheva, ‘Toward a definition of the word Zanj’, Azania, 21 (1986), 108-10. This also is argued in R. L. Pouwels, Horn and Crescent: Cultural Change and Traditional Islam on the East African Coast, 800-1900 (Cambridge, 1987).
4{ }^{4} As, for example, R. T. Duarte, Northern Mozambique in the Swahili World (Uppsala, 1995); and P. J. J. Sinclair, J. M. F. Morais, L. Adamowicz and R. T. Duarte, ‘A perspective on archaeological research in Mozambique’, in Thurstan Shaw, P. Sinclair, B. Andah, and A. Okpoko (eds.), The Archaeology of Africa: Food, Metals and Towns (London and New York, 1993), 409-31.
5{ }^{5} T. J. Hinnebusch, ‘The Shungwaya hypothesis: a linguistic reappraisal’, in J. T. Gallagher (ed.), East African Culture History (Syracuse, 1976). Hinnebusch pointed out that the model of the emergence of the Sabaki languages from NEC, as summarized above, did not specifically support the hypothesis of a Sabaki natal area north of the Juba river at Shungwaya, as was supported by Thomas Spear and others at the time. For more about this, see below. ↩︎
river in the manner of Nurse’s and Spear’s earlier work. However, much of the evidence they provide appears actually to discredit this hypothesis. For one, they cite a close similarity between the Sabaki languages and those of the larger Northeast-Coastal cluster from which Sabaki evolved, one so close as to be ‘barely discernible’. Almost equally hard to distinguish is the separation of Proto-Sabaki and Proto-Swahili (pp. 473-6, 503). Therefore, if one agrees that the Northeast-Coastal-speaking Bantu did evolve somewhere between the Wami and Ruvu rivers, a position supported by most linguists and by the archaeological evidence of Chami and Kusimba (discussed below), something entirely different is suggested, namely that the process of evolution from Northeast-Coastal to Sabaki to Swahili was so rapid that the chances that proto-Sabaki emerged in situ as a language community hundreds of kilometers to the north of Proto-Northeast-Coastal Bantu are very slight indeed.
Another difficulty lies in identifying the source and dating of loan-words from Arabic. The widespread presence of Arabic terms in the secular Swahili vocabularies of administration, trade, housing, cuisine, clothing and nautical technology most likely came later. However, Nurse and Hinnebusch themselves point out the difficulties one encounters in trying to date religious terminology which clearly has a ‘canonical’ status. Also, it is possible that for many centuries Arabic and KiSwahili remained functionally distinct, Arabic being used only as a written ‘canonical’ language while Swahili remained the language of daily discourse.
II
Mark Horton’s long-awaited Shanga: The Archaeology of a Muslim Trading Community on the Coast of East Africa is a landmark publication, crowning years of ground-breaking research by the author. This is a work which should bring him and the British Institute in Eastern Africa a great deal of pride, for with it Horton establishes himself as one of the leading coastal archaeologists of this generation. (Of equal importance has been the work of an emerging generation of East African archaeologists, such as Abungu, Chami, Juma, Kusimba and Mutoro. 6{ }^{6} ) However, anyone who has followed his career will know that Horton’s views of the early history of the coast differ appreciably from his distinguished forebears such as Wilson, Kirkman and Chittick. He and others have advanced in telling fashion the archaeological case for the indigenous roots of coastal towns like Shanga, Manda and Kilwa, thus paralleling the equally pioneering contributions to East African historiography by linguists like Ehret, Nurse, Hinnebusch and others. The information provided by archaeology is a natural complement to historical linguistics as well as written and oral accounts, insofar as it provides reasonably precise chronological information on cultural change. Like linguistics, too, it provides important evidence about interactions between peoples and cultures. Its principal weakness lies in its reliance on material culture with which it is often difficult to associate specific peoples.
Many of Vansina’s most trenchant criticisms of archaeologists certainly would not apply to Horton’s work at Shanga. 7{ }^{7} Beginning in 1980, Horton spent six
- 6{ }^{6} See, G. H. O. Abungu and H. W. Mutoro, ‘Coast-interior settlements and social relations in the Kenya coastal hinterland’, in Thurstan Shaw et al. (eds.), The Archaeology of Africa (London and New York, 1993), 694-704; F. A. Chami, The Tanzanian Coast in the First Millennium A.D. (Uppsala, 1993); Abdulrehman M. Juma, ‘The Swahili and the Mediterranean worlds: potteries of the Late Roman period from Zanzibar’, Antiquity, 70 (1996), 148-56; Chaparuka Kusimba, personal correspondence.
7{ }^{7} J. Vansina, ‘Historians, are archaeologists your siblings?’, History in Africa, 22 (1995), 369-408. ↩︎
seasons investigating the locale, taking care beforehand to examine areas of the site beyond merely the remains of stone structures. This was accomplished by a careful site survey, numerous soundings and at least two substantial excavations outside the center of the settlement (pp. 10-13). Pains also were taken to consider the larger economic, social and cultural context of the location.
The book is arranged into twenty-two chapters beginning with an exhaustive examination of all known external and internal documentation concerning Shanga. Subsequent important chapters concern the tombs and cemeteries; excavations of the Friday mosque and the areas to the east and west of it; discussions of architecture, both domestic and public; finds of local pottery, imported pottery, glass, beads and coins; and descriptions of metalworking, the domestic economy and subsistence. The volume concludes with three chapters summarizing the chronology and development of Shanga, Swahili settlement and trade, and the place of Shanga in Swahili history. The author presents an impressive array of evidence to support his interpretations.
Perhaps of greatest importance is Horton’s discovery of the earliest mosque yet unearthed in sub-Saharan Africa, which he dates to the late eighth century. The additional find of Islamic burials roughly contemporary with the earliest mosque layers adds considerable support to his claim of a Muslim presence in East Africa at this early date. (So far, no evidence of a comparably early Islamic presence in Sudanic Africa has been corroborated, and previous work in East Africa has attributed the arrival of Islam to the eleventh century and later.) His careful study of this mosque through more than twenty stages of continual reconstruction and enlargement over six centuries also provides convincing proof of the waxing of Islam at Shanga. Horton thus documents the emergence and steady development of a tiny Muslim community of uncertain origins at Shanga into one which, by about 1000 A.D., appears to have become predominant. The construction of two new mosques in the fourteenth century, when the town flourished, suggests that the conscientious observation of Islamic practices might have been associated with mercantile activities, which may have been the case from the eighth century when evidence suggests the appearance of craft industries and imported goods almost precisely at the time of the construction of the first mosque.
Horton’s discussion of general social change and the progress of Islam at Shanga contains, however, some of the most controversial material in the volume. He is less convincing when he bases his argument about the growth of the local Muslim community to majority status partly on the rise and decline of two nearby structures which he believes (with no additional corroborative evidence) served as communal lodges for pre- or non-Islamic religious functions. This hypothesis seems to be based on very thin evidence, indeed.
Concerning the general origins and growth of Swahili civilization, Horton again presents exciting new material that contributes to major alterations of our views. First, the thorough treatment of imported ceramics provides greater precision to coastal chronology than what previously had obtained. 8{ }^{8} Also, his discovery of the early, haya-like central enclosure seems to reinforce the linguistically-based argument that associates Swahili beginnings with neighboring, Bantu-speaking societies (see above).
Here again, though, Horton’s treatment of his evidence occasionally raises additional difficulties. First, his assignment of all local pottery to the distinctive Triangular-Incised Ware ceramic tradition (hereafter TIW), which archaeologists believe was the earliest ceramic remains of coastal-hinterland peoples, datable as
- 8{ }^{8} See N. Chittick, ‘The East Coast, Madagascar and the Indian Ocean’, in J. D. Fage and R. Oliver, The Cambridge History of Africa, Vol. 3: From 1030 to I600. (Cambridge, 1977). ↩︎
far back as the fourth century A.D., appears questionable. One expects that a great deal of homogeneity might have existed in the earliest varieties (which Chami and others have discovered on the Tanzanian coast 9{ }^{9} ) but even Horton’s ‘earliest’ Shanga samples would be representative of later TIW varieties, types in which regional or exogenous characteristics would already have begun to show themselves. In fact, there does exist in Horton’s collection a good candidate for a southern coastal variety, one which seems to have been influenced by north-west Indian or Mediterranean traditions. These are the so-called Red-Slip wares found in each of the four major phases of Shanga’s development as Types 4 in phase A; 12 in phase B; 26 in phase C; and 36, 37 and 39 in phase D. Described by Horton as a ‘rare and very fine pottery’, found more commonly in the Comoro Islands, Kilwa, Madagascar and Pemba, these samples appear to be distinct since red slipping and occasional graphiting are their most distinguishing decorative attributes: i.e. all other features usually found in other TIW are absent. Furthermore, its manufacture appears to have been confined to its own fine, reddish fabric and smaller forms. The least that can be said is that archaeologists need to study these wares more closely. 10{ }^{10} Perhaps they have not done so because the pendulum concerning exogenous influences has swung too far in the opposite direction.
Readers familiar with Horton’s earlier work will find that in Shanga he seems to abandon his support for Allen’s notions about ‘pastoralist’ origins for the Swahili. Nowhere in this volume does he advocate this hypothesis to the exclusion of all others. Yet he appears to be giving in to the weight of the evidence only reluctantly. Both he and Allen, for example, have argued that the discovery of rock crystal at Shanga suggests the early existence of a ‘trade route’ linking the coast to the Rift Valley; ‘pastoralist’ domination of this route and of the coast; and, hence, ‘pastoralist’ origins for the Swahili. However, only small quantities of rock crystal were actually found at Shanga, as few as two beads in some phases, and Horton fails to mention its discovery at other coastal sites, including Kilwa. What does this suggest about the likelihood of pastoralist origins emanating from the Rift Valley? Elsewhere, he examines the faunal record as a crucial indicator of origins, cultivating or ‘pastoralist’ (pp. 393ff.). Employing this kind of terminology suggests a view that oversimplifies and overstates: i.e. that the early Swahili had to have been either ‘pastoralists’ or ‘cultivators’. More to the point, he does not explain himself when he applies these terms.
In place of an explicitly pastoralist hypothesis, in chapter 21 Horton discusses three possible models for Swahili beginnings. The first one is based on evidence presented by Abungu which associates northern coastal pottery with a ‘pastoral’ tradition that hypothetically extended from the northern coast inland along the Tana basin to the Eastern Highlands. This, of course, is the position Horton and
- 9{ }^{9} Felix A. Chami, ‘Excavations of a coastal Early Iron Age site in Kisarawe district, Tanzania’, Nyame Akuma, 30 (1988), 34-5; and ‘Limbo: early ironworking in Southeastern Tanzania’, Azania, 27 (1992), 45-52; William Fawcette and Adria LaViolette, ‘Iron-Age settlement around Mkiu, Southeastern Tanzania’, Azania, 25 (1990), 19-26; William Fawcette, Adria LaViolette and Peter Schmidt, ‘The coast and the hinterland: University of Dar es Salaam archaeological field schools, 1987-88’, Nyame Akuma, 32 (1989), 38−4638-46.
10{ }^{10} See H. T. Wright’s discussion of these wares in ‘Early communities on the island of Maore and the coasts of Madagascar’, in C. P. Kottak, J.-A. Rakotoarisoa, A. Southall and P. Verin (eds.), Madagascar, Society and History (Durham, NC, 1986), 76; and ‘Trade and politics on the eastern littoral of Africa, A.D. 800-1300’, Archaeology of Africa, 600. Two locations have been advanced as possible sources for this ceramic tradition. One is north-western India, suggested by Paul Sinclair in personal correspondence, and the other is Egypt, as suggested by Juma in ‘The Swahili and the Mediterranean worlds’. ↩︎
Allen advocated in earlier work. 11{ }^{11} On the other hand, Horton’s second and third models emphasize ceramic evidence which squares Swahili origins with the Early Iron Age tradition and the linguistic indicators of a Bantu-speaking past. All three models acknowledge that by iooo A.D., the Swahili probably were Bantu-speaking mixed cultivators and fishers. His first model, however, one in which these cultivators would have been preceded by Cushitic-speaking pastoralists at sites like Shanga (pp. 410-11), can be dismissed virtually out of hand in view of the overwhelming linguistic evidence provided by Nurse, Hinnebusch and others which indicates a Bantu-speaking past. This hypothesis is implausible and needlessly complicates the picture since its correlates would require that, by iooo A.D., these Cushitic-speakers would have had to have engaged in a language ‘shift’ to Bantu dialects on an order that rarely, if ever, occurs.
In this context, it is important to note that Bantu speech is continuous along the entire coast from Kenya to Mozambique, and Cushitic terms are not present in the core, ‘diagnostic’ Swahili lexis, as one certainly would expect had the original Swahili been Cushitic-speakers. Rather, Cushitic cultural vocabulary in Swahili is confined to items concerning cereal and livestock husbandry. This suggests strongly that the Swahili had a continuous history rooted in a Bantu-speaking past in which neighboring Cushitic-speaking communities played a contributory, yet only ancillary, part. Indeed, we might just think of these early coastal Bantuspeakers as ‘pastoralists’ if they kept a few domesticates, as farmers often do. Quite possibly, too, there could have been Cushitic-speaking pastoralists who settled the site at Shanga alongside Bantu-speakers. In fact, Horton did find evidence suggesting that the site might have seen more than one ethnic group living there. Moreover, even if we choose to think in terms of these limited usages and possibilities for the meaning of ‘pastoralists’, it must be noted that he found virtually no evidence of domestic livestock in the first two centuries at Shanga. Noting, as he and Abungu do, the possible influences of ‘Pastoral-Neolithic’ on (northern) TIW, it must be noted that the linguistic record of Southern Cushitic influences on pre-Swahili history was regional in scope, and not just local, and preceded the first settlement at Shanga by almost two millennia. If anything, Horton is casting his net neither far enough nor deep enough in looking for his pastoralists.
Finally, Horton also points out the low amount of fish in the diets of early Shangans as a possible indication of ‘fish-avoidance’, i.e. again a pastoralist-like taboo. However, this is defective logic, based not on the presence of evidence but on its absence. And contrary to Horton’s position, one would expect that if the earliest settlers at Shanga were primarily cultivators, their seafood diet would largely have been limited to protein-rich shellfish easily procured from in-shore areas. Only later, when they had evolved a more complex maritime technology would they have taken up off-shore fishing. This, of course, is what Hinnebusch’s and Nurse’s reconstructed vocabularies for Proto-Sabaki and Proto-Swahili suggest and what Horton did, in fact, find at Shanga.
II I
The two works reviewed here, along with others cited, represent the largest steps taken in over a generation towards achieving a better understanding of East African coastal history. Neither one of these volumes answers all questions, nor are they
- 11{ }^{11} M. C. Horton, ‘Early settlement of the northern Swahili coast’ (Ph.D. dissertation, Cambridge University, 1984); and ‘The Swahili corridor’, Scientific American, 257 (1987), 86-93; J. de V. Allen, Swahili Origins: Swahili Culture and the Shungswaya Phenomenon (Athens, OH, London and Nairobi, 1993), passim. ↩︎
free of problems. Based on the entirety of the work completed so far, therefore, a brief summary of the most probable model for the early history of this region seems in order here. What follows is a personal interpretation based on the strength of the evidence available. As such it is not assumed to be ‘definitive’, but rather is intended to foster further discussion and to enter a few caveats.
Sometime after about 400 в.с. some of the northernmost Eastern Bantuspeakers, inhabiting areas to the south and southeast of Lake Nyanza (termed by Ehret the 'Kati’12), began an expansion towards the east and south-east. Three groups eventually moved to the coast and its hinterland. Perhaps the earliest to arrive, the people whom Ehret calls the Upland Bantu, spoke Bantu dialects ancestral to the Thagiçu languages today spoken largely in central Kenya. To their south was the Rufiji-Ruvuma group. Quite possibly these two were the first to introduce a planting agricultural tradition, along with livestock husbandry, ironmaking and Kwale-ware pottery (Kwale has been found in Mozambique). A third, and slightly later, Bantu-speaking group who inhabited central-eastern and northeastern Tanzania around I-300 A.D. were Northeast-Coastal Bantu-speakers. They developed the distinctive pottery tradition, called Tana or TriangularIncised Ware, which most archaeologists presently associate with the Early Iron Age on the East African coast and its hinterland. Early forms of TIW appear to have included some elements from Kwale, either because Northeast-Coastalspeakers had absorbed their Upland predecessors or because they themselves had previously made Kwale ware and early TIW was transitional, as Chami contends. The Northeast-Coastal cluster soon split into the Seuta, Ruvu and Sabaki groups. The Sabaki homeland was slightly north of the Northeast-Coastal natal area, between the Pangani and Tana rivers. The Proto-Sabaki-speakers continued expanding along the coast and its hinterlands, evolving into several distinct languages and cultures, c. 600-900 A.D., when TIW was entering its ‘late’ phase. Those who inhabited the hinterland became the Mijikenda. Tana river peoples became the Pokomo and Elwana. The Swahili were the coast-dwelling Sabakispeakers, while the Comorians were closely related peoples who populated the islands from which their name derives. Placed in this context, Shanga initially was settled while this process of differentiation was still occurring, possibly explaining the early, haya-like central enclosure Horton uncovered. At that stage, these late Proto-Sabaki-speakers had not developed the later sophisticated maritime technologies, but farmed and consumed shellfish. Continuing the tradition of their Eastern Bantu ancestors, their domestic buildings were rectangular.
What about the role of foreigners and foreign ideas? As documented both at Shanga and in the reconstructed lexicons of Proto-Sabaki and Proto-Swahili, the process of continued differentiation in coastal civilization combined local adaptations to the environment as well as the selection and internalization of exogenous ways as new needs and openings, such as new food resources and commercial opportunities, presented themselves. Towns like Shanga and Manda thrived because local Africans were quick to take advantage of these. Given the frontier character of the coast, its physical and human geography was particularly rich. Contrary to Mazrui’s and Shariff’s hypothesis, 13{ }^{13} the early Swahili were almost certainly not ‘Arabs’, just as it is unlikely they were the Cushitic-speaking
12{ }^{12} 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). Besides Hinnebusch and Nurse, Horton, Chami and other sources, I am basing this reconstruction partly on Ehret’s book which is an expansion of his ‘The East African interior’, in M. El Fasi (ed.), General History of Africa, Vol. 4: Africa from the Seventh to the Eleventh Century, (Berkeley and Los Angeles, 1988), 616-43.
13{ }^{13} Alamin M. Mazrui and Ibrahim N. Shariff, The Swahili: Idiom and Identity of an African People (Trenton, NJ, 1994), 64-5.
‘pastoralists’ of Allen and Abungu. Nevertheless, there were Southern Cushiticspeaking peoples inhabiting the region near the Northeast-Coastal Bantu-speaking settlement area as well as near the northern coast, namely Ehret’s Mbuguan and Dahaloan groups. Some might even have inhabited early coastal settlements alongside their Bantu-speaking neighbors, as Horton’s findings at Shanga indicate; however, their influences were through cultural borrowings, as evinced by Southern- and Eastern-Cushitic loan words, not because they ‘were’ the early core-Swahili. By the Northeast-Coastal and Sabaki stages, the history of contacts between Eastern Bantu and Southern Cushitic communities was already a long one, as Ehret has documented. Therefore, as attested by both the linguistic and archaeological evidence, the parts played by ‘pastoralists’ and ‘Arabs’ in the history of the Swahili were important, but essentially contributory.
IV
For a long time now, the historiography of the Swahili and their neighbors has been addled by the evidence from oral tradition. Local historians have related tales of migration from Middle Eastern or Shungwayan homelands to towns and villages along the coast and near-interior. For their part, many past Western observers of coastal civilization assumed the racial superiority of ‘white’ to black peoples, and assigned the origins of coastal settlements to ‘Arab’ immigrants. 14{ }^{14} More recently, certain questionable notions concerning the use of oral tradition for reconstructing history have found their way into Swahili historiography through the ‘standard’ model explaining the origins of the Swahilis’ neighbors, the Mijikenda. Some Mijikenda, as well as some Swahili, Pokomo and Taita tell of a homeland to the north of the Juba river at a place calling Shungwaya. Because such accounts are so widely related, some researchers have believed that a literal acceptance of Shungwayan origins has been adequately ‘tested’ and proven. 15{ }^{15} Criticism from Morton, and more recently from Willis and Walsh, has been based on evidence that these traditions were unknown among many Mijikenda prior to the twentieth century. 16{ }^{16} Clearly, what those who have accepted the ‘standard’ model of Mijikenda origins have failed to appreciate is the likelihood of interdependence among their traditions, especially since they have been collected among societies that are not discrete entities. 17{ }^{17}
The accepted model of Mijikenda origins began affecting ideas about Swahili origins when Thomas Spear pointed out that many (though not all) the peoples whose oral traditions tell of Shungwaya origins speak languages of the Proto-
14{ }^{14} This, of course, was no different than it was for centuries all over the continent, as every experienced African historian knows. For beginners, however, one need look no further than the comments provided by the Portuguese sources given in G. S. P. Freeman-Grenville, The East African Coast: Select Documents (Oxford, 1962). Similar and more recent views were stated by numerous nineteenth- and twentieth-century sources from Burton to Chittick.
15{ }^{15} Thomas T. Spear, The Kaya Complex : A History of the Mijikenda Peoples of the Kenya Coast to 1900 (Nairobi, 1978), especially 16-80; and Cynthia Brantley, The Giriama and Colonial Resistance in Kenya, 1800-1920 (Berkeley and Los Angeles, 1981).
16{ }^{16} R. F. Morton, ‘The Shungwaya myth of Mijikenda origins: a problem of late nineteenth-century coastal history’, African Historical Studies, 5 (1973), 397-423; Justin Willis, Mombasa, the Swahili and the Making of the Mijikenda, (Oxford 1993), 26-34; and Martin Walsh, ‘Mijikenda origins: a review of the evidence’, Transafrican Fournal of History, 21 (1992), I- 18 .
17{ }^{17} On the problem of interdependence, see J. Vansina, Oral Tradition as History (Madison, 1985), 152-9 and 187-90. Spear published samples of these traditions in Traditions of Origin and their Interpretation: The Mijikenda of Kenya (Athens, OH, 1981).
Sabaki cluster mentioned above. Evidence of a shared ancestry among these peoples, he argued, supported the likelihood that these traditions were fundamentally true. This revised view was elaborated further by Nurse and Spear in a joint publication, and again, as discussed above, in the Nurse and Hinnebusch volume under review. 18{ }^{18} Their model held that, following the Proto-Northeast Coastal phase (which evolved in north-east Tanzania), Sabaki-speakers trekked en masse to Shungwaya, which they placed north of the Juba river. Many later remigrated southwards, populating the coast, the Tana river and coastal hinterlands, and evolved into today’s Swahili, Pokomo, Mijikenda, Elwana and Comorians.
I should mention, for my own part, that I accepted the hypothesis of a Shungwaya inasmuch as it explained Swahili traditions about the ‘Shirazi’. However, I also entertained doubts about the larger model because of its uneconomical nature and the inconsistency between it and the traditions themselves; these relate that such migrations took place after the Oromo expansion into the northern hinterland in the sixteenth century and not a full thousand years earlier when the Sabaki expansion occurred. 19{ }^{19}
One also must consider the contributions of archaeologists to this confused state of affairs. For a long while, historians had little archaeological data with which they could check the historical traditions of the Swahili or their neighbors. The first generation of archaeologists produced little to contradict such accounts since their research strategies continued to be based on the old assumptions about ‘Arab’ origins. Beginning in the 1970s and extending into the 1980s, historians like Allen, Nurse, Spear and myself started called attention to the African character of coastal civilization. Allen was especially forceful in calling for archaeological work on hinterland sites. 20{ }^{20} Meanwhile, a new generation of like-minded archaeologists was in training and preparing to respond to this call.
The earliest result of this new archaeology was Soper’s discovery of an Early Iron Age pottery called Kwale ware which appeared to have a regional distribution. In 1979, Phillipson found a slightly later ceramic type which, because it was restricted to coastal-hinterland sites, was claimed as the first identifiable Swahili pottery, variously named ‘Wenje’, ‘Tana’, and ‘Triangular Incised’ ware. 21{ }^{21} Mutoro’s discovery of late-first-millennium TIW in Mijikenda kayas meanwhile conclusively laid to rest the old hypothesis of a Proto-Sabaki homeland in Shungwaya. 22{ }^{22} Subsequent discoveries of both Kwale and TIW wares in coastal locations from southern Somalia to southern Mozambique, as well as many offshore islands, the Pare Hills, Mt. Kilimanjaro and the headwater areas of the Wami and Rufiji rivers added to the emerging picture of a region between the Pangani and Tana rivers that had been loosely unified by language and trade from about 500 A.D. onwards. 23{ }^{23}
- 18{ }^{18} Nurse and Spear, Swahili. 19{ }^{19} Pouwels, Horn and Crescent, 12-13.
20{ }^{20} For example, see James de V. Allen, ‘Swahili culture reconsidered’, Azonia, 9 (1974), 105-38; R. L. Pouwels, ‘Islam and Islamic leadership in the coastal communities of Eastern Africa, 1700 to 1914’, (Ph.D. thesis, U.C.L.A., 1979); Nurse and Spear, Swahili.
21{ }^{21} Robert Soper, ‘Iron Age sites in north-eastern Tanzania’, Azania, 2 (1967), 19-36; and D. W. Phillipson, ‘Some Iron age sites in the Lower Tana valley’, Azania, 14 (1979), 155−60155-60.
22{ }^{22} Henry Mutoro, ‘An archaeological study of the Mijikenda Kaya settlements on the hinterland Kenya coast’, (Ph.D. thesis, U.C.L.A., 1987), 221-33. Mutoro’s discovery placed the settlement of Mijikenda sites at least 600 years earlier than Spear’s model permitted.
23{ }^{23} The work of many archaeologists has contributed to this new scenario. A sampling would have to include the following: G. H. O Abungu, 'Communities on the River Tana, ↩︎
The greatest remaining problems center on establishing what relationships exists between Kwale wares and TIW, identifying who made these wares and connecting the linguistic evidence with the historical and archaeological data. Perhaps the most recent and crucial archaeological contributions that have attempted to address the first two questions have been Abungu’s and Chami’s. Abungu’s research has focused on Tana river sites; hence, he believes that, for the northern coast at least, the origins of TIW might lie in the Pastoral Neolithic wares found in Kenya’s Rift Valley. 24{ }^{24} The implication of such work is that the Swahili have the ‘pastoralist’ roots which Allen and Horton have hypothesized, rather than an origin in the Bantu-speaking communities where other historians place coastal beginnings. 25{ }^{25} Chami has countered with a carefully argued case favoring the latter perspective, basing his position on extensive research he and others have conducted on the Tanzanian coast, including Zanzibar and Pemba. 26{ }^{26} Of the two positions, the very early dates and the large quantities of the TIW on the central and northern Tanzanian coast and interior favor that area as the likeliest ‘natal’ zone for both TIW and Sabaki-speakers. This view is supported by virtually all linguists who have been active in the field over the past twenty-five years. 27{ }^{27} More problematic is Chami’s assertion that a direct link exists between Kwale and TIW, thus implying that the same people(s) were responsible for both traditions. He has identified early forms of TIW, dated to c.300−600c .300-600 A.D., which he argues include some design elements of Kwale ware. 28{ }^{28} So far, though, only one site has yielded evidence of TIW directly overlaying Kwale, whereas in all other locations TIW has not stratigraphically been found directly over Kwale. 29{ }^{29} However, Chami has established apparently beyond reasonable doubt that the ‘early’ TIW found at Kenya coastal sites and along the Tana river is, in fact, a later form which appeared after the seventh century.
The very early dating and location of TIW at so many coastal and hinterland sites seriously challenges the initial belief of some that it was strictly a Swahili tradition. Rather, it now seems to have been a somewhat earlier pottery associated primarily with farming, limited fishing technologies and ironworking. Therefore, a more likely association would be with an appreciably earlier stage of coastalhinterland regional development, most likely Bantu-speaking of the Proto-Northeast-Coastal or Proto-Sabaki phases. If one accepts the likelihood of this last proposition, it also furnishes powerful, perhaps conclusive, evidence against the standard model of far-northern Shungwayan origins for Sabaki.
- Kenya: An archaeological study of relations between the delta and river basin, 700-1890 A.D.', (Ph.D. dissertation, Cambridge University, 1989); Chami, ‘Excavations’ and ‘Limbo’; and Fawcette, et al. ‘The coast and the hinterland’.
24{ }^{24} G. H. O. Abungu, ‘Agriculture and settlement formation along the East African coast’, Azania, 29-30 (1994-5), 248-56.
25{ }^{25} Allen, Swahili Origins; Horton, ‘Early settlement’.
26{ }^{26} F. A. Chami, The Tanzanian Coast; W. Fawcette, A. LaViolette and P. Schmidt, ‘The coast and the hinterland’.
27{ }^{27} See B. Heine, Status and Use of African Lingua Francas (Munich, 1970); W. J. G. Moehlig, ‘Versuch einer historischen Gliederung der nordostlichen Bantusprachen auf lautgleichender Grundlage’, Afrika und Ubersee, 6I (1978), 175-89; and the works by Ehret, Hinnebusch and Nurse previously cited.
28{ }^{28} Chami, The Tanzanian Coast, 72-4.
29{ }^{29} See Randi Haaland, ‘Dakawa: An Early Iron Age site in the Tanzanian hinterland’, Azania, 29-30 (1994-5), 238-47. ↩︎
A few final remarks bearing on continuing East African coastal research seem to be in order. First, scholars must continue to broaden their perspective even further than they already have done. One thing has been established firmly over the past two decades: that Swahili civilization is African, not exogenous, in origin and character. Yet, like all civilizations, its uniqueness is based partly on innovation and partly on the selective incorporation and transformation of imported ideas. Therefore, research efforts should continue to interpret this civilization within the broader contexts of both East African and western Indian Ocean developments. Unquestionably, this civilization emerged as a local adaptation to these developments and the coastal environment. Yet the time has passed when entirely localized studies and interpretations suffice. Already there are indications that this early emergence had wider regional significance. Mapunda’s discovery of TIW-like pottery at Ruhuhu is an example. 30{ }^{30} Very few scholars of East African history have shown themselves willing to assume the risks of such broad interpretations, although Ehret and Phillipson are refreshing exceptions. 31{ }^{31}
Secondly, if broader interpretations are needed, more regional studies are required too. Both linguists and archaeologists working on the coast seem content to interpret the general picture largely on the basis of their personal research. Excellent as many of these have been, they leave one puzzling over some important unanswered questions. For example, where does Northeast-Coastal fit in the bigger picture? What possible linguistic and historical connections exist between Northeast-Coastal- and non-Northeast-Coastal-speaking communities, particularly between coastal groups as far south as Natal and as far west as the Corridor region? Archaeologists and historians need also to get away from oversimplified hypotheses about ‘who’ the earliest settlers in coastal sites were. What precisely is meant, for example, when some claim they were ‘pastoralists’?
Researchers need to demonstrate a willingness to create hypotheses from the most logical, economical explanations suggested by the available evidence and do away with the gratuitous guesswork that has plagued the field for a long time. Increasingly it seems, some have shown a willingness to grasp at straws and to construct radical new theses based on the thinnest of evidence. 32{ }^{32} When new information points to simpler, more likely solutions, they have been reluctant to abandon untenable positions. In the worst instances, positive assertions have been made from non-existent evidence. Therefore, scholars must exhibit greater skepticism about prevailing models and be prepared to abandon old thinking habits as new information surfaces. 33{ }^{33}
30{ }^{30} B. Mapunda, ‘Preliminary report on archaeological reconnaissance along the Ruhuhu river basin, southern Tanzania’, Nyame Akuma, 36 (1991), 32-40.
31{ }^{31} Ehret, The Classical Age; D. W. Phillipson, The Later Prehistory of Eastern and Southern Africa (London, 1977).
32{ }^{32} Allen’s Swahili Origins is the most egregious example. For a review, see R. L. Pouwels, ‘Swahili Civilization,’ Fournal of African History, 34 (1993), 518-20.
33{ }^{33} See Vansima’s remarks along these lines in ‘The power of systematic doubt’, History in Africa, I (1974); and the reply by D. Henige, ‘Gambit declined: pervasive doubt about systematic doubt’ in R. Harms et al., Paths toward the Past : African Historical Essays in Honor of Fan Vansina (Madison, 1994), 77-96.