Speaking with Substance Methods of Language and

0 downloads 0 Views 2MB Size Report
innovations or borrowing styles of speaking, particularly loanwords, from other ...... From an older word for a kind of long blade, the inhabitants of the southern ...... case to demonstrate “a continuum of emphasis on wild game across time and.
SPRINGER BRIEFS IN ARCHAEOLOGY CONTRIBUTIONS FROM AFRICA

Kathryn M. de Luna Jeffrey B. Fleisher

Speaking with Substance Methods of Language and Materials in African History

SpringerBriefs in Archaeology Contributions from Africa

Series Editor Ann B. Stahl University of Victoria Victoria, BC, Canada

[email protected]

SpringerBriefs in Archaeology presents concise summaries of cutting-edge research and practical applications in all aspects of Archaeology. Featuring compact volumes of 75 to 125 pages, the series covers a range of content from professional to academic. Typical topics might include: • • • • •

A snapshot of a hot or emerging topic A contextual literature review A timely report of state-of-the art analytical techniques An in-depth case study A presentation of core concepts that students must understand in order to make independent contributions.

Briefs allow authors to present their ideas and readers to absorb them with minimal time investment. Briefs are published as part of Springer’s eBook collection, with millions of users worldwide. In addition, Briefsare available for individual print and electronic purchase. Briefs are characterized by fast, global electronic dissemination, standard publishing contracts, easy-to-use manuscript preparation and formatting guidelines, and expedited production schedules. We aim for publication 8-12 weeks after acceptance. Both solicited and unsolicited manuscripts are considered for publication in this series. Briefs can also arise from the scale up of a planned chapter. Instead of simply contributing to an edited volume, the author gets an authored book with the space necessary to provide more data, fundamentals and background on the subject, methodology, future outlook, etc. SpringerBriefs in Archaeology contains a distinct subseries focusing on Archaeological Heritage Management edited by Douglas Comer, Helaine Silverman and Friedrich Lüth, in conjunction with the International Committee on Archaeological Heritage Management (ICAHM). The series will address critical contemporary problems and illustrate exemplary work in archaeological heritage management in countries around the globe. The series will take a broad view of the concepts of archaeology, heritage, and management in accordance with ICAHM’s mandate itself. More information about this series at http://www.springer.com/series/13523

[email protected]

Kathryn M. de Luna  •  Jeffrey B. Fleisher

Speaking with Substance Methods of Language and Materials in African History

[email protected]

Kathryn M. de Luna Department of History Georgetown University Washington, DC, USA

Jeffrey B. Fleisher Department of Anthropology Rice University Houston, TX, USA

ISSN 1861-6623     ISSN 2192-4910 (electronic) SpringerBriefs in Archaeology Contributions from Africa ISBN 978-3-319-91034-5    ISBN 978-3-319-91036-9 (eBook) https://doi.org/10.1007/978-3-319-91036-9 Library of Congress Control Number: 2018942615 © The Author(s), under exclusive licence to Springer International Publishing AG, part of Springer Nature 2019 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Printed on acid-free paper This Springer imprint is published by the registered company Springer International Publishing AG part of Springer Nature. The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

[email protected]

In memory of Jan Vansina, who saw human stories in all forms of evidence

[email protected]

Acknowledgments

This book could not have been written without a multitude of conversations, some of which have been ongoing for decades. Befitting a book on interdisciplinary research methods, we blend our list of interlocutors and teachers, for each of us gained from the others’ mentors and conversationalists and many such colleagues have supported us both. The genesis of our conversations was in the planning of a conference that we organized with Susan McIntosh at Rice University in 2011, entitled “Thinking Across the African Past: Archaeological, Linguistic and Genetic Research on Precolonial African History.” The conference and its participants (published as a special issue of the African Archaeological Review in 2012) stimulated numerous discussions. Over the several years it took to write this book, we benefitted from conversations about method with numerous colleagues: Jan Vansina, Susan McIntosh, Susan Kus, Joe Miller, Chris Ehret, Rhiannon Stephens, Neil Kodesh, Kairn Klieman, John McNeill, Tim Newfield, Mary Prendergast, David Anthony, Scott MacEachern, Pete Robertshaw, and members of the Spring 2018 HopkinsGeorgetown African history workshop. Conversations and collaborations with Stephanie Wynne-Jones and Matthew Pawlowicz influenced this book more than they (or likely we) realize. We appreciate masterful index list produced by Trish Patel. The book was transformed by the insightful remarks of our two anonymous reviewers and the series editor, Ann Stahl. Others taught us the value of interdisciplinary training long before we put pen to paper: Tim Earle, Chap Kusimba, Andrew Reid, Joe Miller, and James Deetz. Adria LaViolette and David Schoenbrun ensured that we asked questions about evidence and methods in the first place; we are grateful for their guidance. We appreciate the financial support of the projects that serve as case studies in the pages that follow: Fulbright-Hays, Rice University History Department, Rice University Humanities Research Center, Social Science Research Institute at Rice, Georgetown Office of the Provost, National Science Foundation (research on Pemba with A. LaViolette; BCS0138319 and INT 9906345), and the Chittick Fund of the British Academy (Ceramics and Society Project with S. Wynne-Jones). Colleagues at the Department of History at the University of Zambia and the Livingstone vii

[email protected]

viii

Acknowledgments

Museum as well as Brave Mushabati and the scores of Zambians who patiently engaged peculiar questions about insiku while hosting their questioners ensured a collaborative, collegial research experience during linguistic and, eventually, archaeological work in Zambia. We thank our research collaborators at the Department of Archives, Museums, and Antiquities on Zanzibar, the University of Dar es Salaam (in particular, Prof. Bertram B.B.  Mapunda), and the many communities that supported research on Pemba Island and beyond.

[email protected]

A Note on Orthography

When discussing linguistic data in this volume, we follow the orthography of the published work under consideration in each chapter. Differences in representing the high, close vowels are of greatest concern. In Chaps. 2 and 3, we follow the Bantu Lexical Reconstructions database. Here, the Bantu seven-vowel system is represented as i ɩ e a o ʊ u. In Chaps. 4 and 5, we follow Nurse and Hinnebusch (1993), who mark the high, close front (i) and back (u) vowels as į and ų, respectively. Bolded terms used throughout the text are defined in the glossary that comprises Appendix A.

ix

[email protected]

Contents

1 Introduction��������������������������������������������������������������������������������������������   1 A Supplemental Approach to Interdisciplinarity ������������������������������������   3 How Archaeologists and Historical Linguists Generate Information About the Past ������������������������������������������������������������   5 Connecting Archaeological and Linguistic Evidence: Three Approaches ��������������������������������������������������������������������������   20 A Supplemental Approach to Archaeology and  Historical Linguistics����������������������������������������������������������������������   27 Scope, Organization, and History of the Volume������������������������������������   29 2 The Politics of Food Collection in South Central Africa��������������������   31 Methods and Evidence����������������������������������������������������������������������������   31 Summing Up��������������������������������������������������������������������������������������������   46 3 Comments, Dialogue and Supplemental Reading: South Central Africa����������������������������������������������������������������������������������������   47 Commentary��������������������������������������������������������������������������������������������   47 Dialogue��������������������������������������������������������������������������������������������������   58 Question One ��������������������������������������������������������������������������������������   58 Question Two ��������������������������������������������������������������������������������������   60 Question Three������������������������������������������������������������������������������������   63 Question Four��������������������������������������������������������������������������������������   65 Question Five ��������������������������������������������������������������������������������������   66 Supplemental Reading: Rethinking the Scale of Household History in South Central Africa������������������������������������������������������   67 4 When Did Feasting Emerge on the Eastern African Coast? New Perspectives from Historical Linguistics and Archaeology����������������   75 Introduction����������������������������������������������������������������������������������������������   75 Archaeology and Linguistics in Eastern Africa ��������������������������������������   76 Swahili Historical Linguistics and Archaeology ������������������������������������   78

xi

[email protected]

xii

Contents

Background on Food and Politics in Ancient Eastern Africa������������������   81 The Ritual Politics of Feasting on the Swahili Coast������������������������������   82 Rethinking the Feast: Historical Linguistics and Archaeology in Dialogue��������������������������������������������������������������������������������������   86 Competitive Feasting Before AD 1000?��������������������������������������������������   90 Conclusion ����������������������������������������������������������������������������������������������   93 5 Comments, Dialogue, and Supplemental Reading: Eastern African Coast������������������������������������������������������������������������������������������   95 Commentary��������������������������������������������������������������������������������������������   95 Dialogue��������������������������������������������������������������������������������������������������  106 Question One ��������������������������������������������������������������������������������������  106 Question Two ��������������������������������������������������������������������������������������  107 Question Three������������������������������������������������������������������������������������  113 Question Four��������������������������������������������������������������������������������������  115 Supplemental Reading: Speaking Well and the Development of Feasting��������������������������������������������������������������������������������������  118 Afterthoughts on Dating and Methods: Linguistic and Material Culture Change ������������������������������������������������������������������������������  124 6 Conclusion����������������������������������������������������������������������������������������������  127 Learning the Methods and Reading the Evidence ����������������������������������  128 Toward a Supplemental Approach to Material and Speech ��������������������  129 Speaking with Substance ������������������������������������������������������������������������  133 Appendix A: A Glossary for Non-specialists������������������������������������������������  135 Bibliography ��������������������������������������������������������������������������������������������������  141 Index����������������������������������������������������������������������������������������������������������������  153

[email protected]

About the Authors

Kathryn  M.  de  Luna  is an historian of early Central Africa and Associate Professor of African History at Georgetown University. She is the author of Collecting Food, Cultivating People: Subsistence and Society in Central Africa (Yale, 2016) and co-editor with Ericka Albaugh of Tracing Language Movement in Africa (Oxford, 2018). Her current research focuses on the history of pyrotechnologies and the senses in early Central Africa. Jeffrey  B.  Fleisher  is an archaeologist of complex societies specializing on the Swahili coast of eastern Africa. His current research at the medieval site of Songo Mnara in Tanzania focuses on the use of open space with Swahili towns; he has worked to integrate historical linguistic data with archaeological materials from this site in order to interpret spatial activities. His previous research on Pemba Island, Tanzania, focused on the ritual politics of consumption, addressing feasting practices and emergence of social power.

xiii

[email protected]

List of Figures

Fig. 2.1 Historical geography of Botatwe language divergences before AD 1250 Fig. 2.2 Historical geography of Botatwe language divergences after AD 1250 Fig. 4.1 Eastern African coast, with sites mentioned in the text Fig. 4.2 Congregational mosque at Chwaka, Pemba Island, Tanzania; top from Pearce (1920); bottom after excavations in 2004 Fig. 4.3 Sherds of Siraf (left) and Sasanian–Islamic jars (right) from Tumbe, eighth to ninth century AD Fig. 4.4 ETT assemblage forms, including (a) globular jars, (b) necked jars, (c) open bowls, (d) closed bowls, and (e) carinated bowls, illustrated by Matilde Grimaldi Fig. 4.5 Sherds of sgraffiato (top left) and celadon (top right) bowls, from Chwaka, thirteenth century AD; profiles of sgraffiato bowls. (Adapted from Horton 1996) Fig. 4.6 Graphited red-slipped bowls: rim and body sherds (top); reconstructed bowl from rim and base sherds (bottom); all from Bandarikuu on Pemba Island, eleventh century AD Fig. 4.7 Red paint/burnished and graphited bowl from Tumbe, Pemba Island, eighth to ninth century AD (left, exterior; right, interior)

33 34 77 83 84 85 86 87 90

xv

[email protected]

List of Tables

Table 2.1 Outline classification of Botatwe languages Table 2.2 Ratios of wild to domesticate fauna in the collections from major regional sites in South Central Africa Table 4.1 Approximate date of emergence of reconstructed protolanguages and dialect clusters of the Swahili coast

32 37 79

xvii

[email protected]

Chapter 1

Introduction

Scholars of Africa’s early pasts have a long history of sustained interdisciplinary collaboration. Historians, linguists, archaeologists, anthropologists, and other disciplinary specialists on Africa are particularly well-versed in the basic methods of other disciplines, and interdisciplinary narratives of Africa’s early pasts are read widely across disciplines compared to researchers in other world regions. It is not uncommon for an historian specializing in twentieth-century Tanzania to teach narratives of precolonial Africa produced by archaeologists and historical-linguists. Likewise, archaeologists of Africa commonly teach about the use of historical linguistics and oral traditions as means to reconstruct the precolonial past. Thus, the audience—scholarly and student—of early African history is much wider than in other world regions, with a profound effect on interdisciplinary research methods. After a hiatus of interdisciplinary research in the 1980s and 1990s, the twenty-­ first century has seen a resurgence of projects that rely on evidence generated from multiple disciplines (Blench and MacDonald 1999; Bostoen et al. 2015; de Filippo et  al. 2012; Delius and Schoeman 2010; Fleisher et  al. 2012; Grollemund et  al. 2015; Mulaudzi et al. 2010; Pakendorf et al. 2011; Skoglund et al. 2017; Stahl 2009, 2010; Stahl and LaViolette 2009; Swanepoel et  al. 2008; Tishkoff et  al. 2009). Inspired by how the converging interests of Africanist archaeologists, linguists, historians, and geneticists enrich what we know about life in early African communities, we feel that conversations about methods must remain central to the enterprise of interdisciplinary scholarship. Debates about the appropriate use of data and the limits of their interpretation within and across disciplines may appear convoluted or just plain confusing to a practitioner specializing in another methodology or world region (Robertshaw 2000; Vansina 1995). This can be so because debates focus on either highly specialized knowledge or particular stances toward data, knowledge, and the past that do not translate across disciplines. For this reason, we have much to gain from exposing and studying how scholars learn to use each other’s data. This volume explores the process of interdisciplinary scholarship and learning, following the lead of others who have drawn attention to the “fractures that inhibit © The Author(s), under exclusive licence to Springer International Publishing AG, part of Springer Nature 2019 K. M. de Luna, J. B. Fleisher, Speaking with Substance, SpringerBriefs in Archaeology, https://doi.org/10.1007/978-3-319-91036-9_1

[email protected]

1

2

1 Introduction

interdisciplinary communication” (Stahl et al. 2004:83) in an effort to chart a path forward. We document the interaction between two specialists of early African history—an archaeologist and an historian using linguistic evidence—around two case studies. The volume is an exposé of the process, rather than the result, of talking across disciplines. Our aim is to reveal the debates, surprises, errors, and confusion that precede and influence the production of the tidy interdisciplinary reconstructions of early history common in Africa and other world regions. The process of debate and collaboration across disciplinary boundaries lies at the center of this volume while the usual outcome of collaboration—interdisciplinary syntheses—lie at the periphery. By shifting the conventional focus of scholarly writing to the process of interdisciplinary inquiry rather than its product, we hope to accomplish four goals. First, we seek to teach the basic principles of our respective disciplines. We offer a short explanation of our methods in the introduction as a frame for more specific discussions of method in later chapters, which focus on case studies. Second, we seek to expose the pitfalls and opportunities of working with another specialist with similar goals, but a vastly different method and source-base. The two case studies in this volume document our experiences of learning to think and see through the lens of an unfamiliar discipline without becoming a specialist in that field. In the process, we explore the variety of ways in which scholars can link the archaeological and linguistic records, which is our third goal. Finally, we highlight the contribution of interdisciplinary approaches to Africa’s past to similar efforts in other world regions. Here, we take a radical stance for some specialists, particularly non-Africanists. Drawing on Ann Stahl’s (2001) work on interdisciplinary approaches to lived history, we argue that partial, indirect archaeological and linguistic evidence of meaning-­making put into supplemental tension (rather than reconciled) offers an important method for the challenge of connecting the archaeological and linguistic records. In the remainder of the introduction, we address each of these goals. We begin with a brief explanation of Stahl’s supplemental approach to historical texts and material objects in her effort to recover lived history. However, we argue that applying this approach to the combination of historical linguistic and archaeological data reveals a different set of tensions, silences, and epistemological stances than those Stahl outlines for her historical and archaeological sources. These differences have important implications for key debates about the methods used to access the past through archaeological and linguistic evidence. Understanding these implications requires a basic familiarity with the methods of archaeology and historical linguistics. For those who do not claim expertise in both disciplines, we offer readers a short introduction to the foundational principles of each method in the second section of the introduction. We follow this section with a review of the different methods by which scholars have sought to connect archaeological and linguistic data to develop interdisciplinary arguments about the past. In this third section, we draw on case studies from Africa as well as influential or provocative studies from other world regions. In the fourth section, we return to the implications of the supplemental

[email protected]

A Supplemental Approach to Interdisciplinarity

3

approach to historical linguistic and archaeological sources for debates about their use in interdisciplinary research. We close the introduction with an outline of the contents and format of the rest of the volume.

A Supplemental Approach to Interdisciplinarity Stahl’s (2001) exploration of interdisciplinary epistemology and methodology focuses on historical, anthropological, and archaeological sources. Nevertheless, it provides a useful conceptual framework for our approach to archaeological and historical linguistic sources. Following Trouillot (1995), one of Stahl’s major concerns in writing historical anthropology is the way that different data sources each create particular “mentions and silences.” Silences can “enter the process of historical production at multiple moments” including through the making of “sources, archives, narratives, and history” (2001:1; see also Trouillot 1995). Thus, any interdisciplinary project must examine the way that different approaches not only create possibilities of ‘mentions’—ways of accessing and producing the past—but also silences. Historians and archaeologists must attend to and investigate these silences, which can serve to erase particular actors, ideas, or relations. In Stahl’s approach, one way to attend to the particularities of various sources and approaches is to consider them as supplemental rather than additive. Additive approaches simply seek to add data from one data set to another, with little thought of the ways these may intersect and/or interact. Her notion of the “supplement” comes from Dirks (1996:34–6, via Derrida): “a supplement is something that is added as if external to the thing itself, but its necessity paradoxically proclaims the essential inadequacy of the original.” Thus, in working to create histories that synthesize various sources, Stahl argues (2001:15) that we should hold them in tension: Conceiving of anthropological, historical, and archaeological perspectives, questions, and evidence as supplemental, rather than additive, places them in productive tension, enabling us to see the possibilities and limits of their distinct forms of knowledge….such a perspective also highlights the overlapping yet distinct processes involved in the production of mentions and silences within each discipline.

Through this productive tension, Stahl observes a contrast between historical and archaeological sources, noting that historical sources rely on oral and textual resources while archaeological sources rely on material resources. This contrast reveals that these sources undergird different forms of Trouillot’s (1995:2, 29) “historicity”: historicity 1 (sociohistorical process, or “what happened”) and historicity 2 (historical narratives, or “what is said to have happened”). As Trouillot (1995:29) describes, historicity 1 can be the “traces” of “what happened” (what he calls the “materiality of sociohistorical process”) and thus invokes archaeological data and sources, in contrast to historicity 2, which includes historical sources and narratives. Working between historical and archaeological sources and treating

[email protected]

4

1 Introduction

them as supplemental, therefore, highlights their dialectical relationship and the ways that their alternative historicity might allow for a more nuanced understanding of the lived history of African peoples. Stahl’s distinction between sources based on language (historical, oral historical, and anthropological sources illuminating historicity 2) and sources based on “the material residues of life” (archaeological sources illuminating historicity 1) is a powerful one that advances interdisciplinary reconstructions based on archaeological and historical materials. This distinction between the meaning produced by language (historicity 2) and the meaning produced through practice (historicity 1) is important to her supplemental approach. The meaning-making of historical sources, Stahl argues, comes from viewing societies “through the lens of their own epistemologies” (2001:16), while in archaeology “material objects are endowed with meaning through practice, and…play a role in forging, transforming, and reproducing meaning.” Our approach to interdisciplinary historical research builds upon Stahl’s insights, using them to probe the relationship between a different pair of historical sources: archaeological and historical linguistic, including the reconstruction of words and their meanings. Both the supplemental approach and the relationship between forms of historicity and sources rooted in language take a decidedly different shape when applied to the pairing of archaeological and historical linguistic evidence. While Stahl’s supplemental approach assures that the contrasts between sources are recognized and productively used, a point to which we will return throughout the volume, we must also address what we see as greater similarities between linguistic and material residues, revealing a different set of tensions, contrasts, and connections between archaeology and historical linguistics. Historical linguistic sources were not part of Stahl’s source base for her interdisciplinary approach, and thus we need to introduce historical linguistics into the argument that she has already begun. The scale of the pasts recovered through language history spans hundreds of years and geographies that often exceed the sites and regions that regularly frame archaeological research. The histories of words’ changing meanings result from moments of creativity, exchange, and debate between individuals, but that granular narrative resolution can only be apprehended through careful, informed historical imagination. We cannot know those conversations as their participants might have narrated them because, in this case, the scale of the linguistic evidence is too small— individual words and even lexicons afford only partial glimpses of the ideas mobilized in such episodes. Rather, word histories reveal the success of such episodes of innovation. The invention of words might result from intentional commentary of the sort associated with historicity 2, as one speaker or a small group of speakers purposefully draws out connections between the known and the new, perhaps by proposing a new ways to use old terms. Or, the invention of words can be spontaneous—a joke, a joyous exclamation, a fearful reaction to an imminent threat, the unintentional outcome of episodes of “what they did in the world” associated with historicity 1. But such episodes of creativity are only preserved in the linguistic record when other speakers of the language accept them by using them, transforming one individual’s act into the everyday doings of the broader public, into the everyday

[email protected]

How Archaeologists and Historical Linguists Generate Information About the Past

5

practices through which meanings—of words, of actions, of materials—are continuously reconstituted and refashioned. In this way, the different scales and resolutions of historical linguistic and archaeological evidence actually lead to significant similarities in the access they provide to the active constitution of social and cultural life (rather than a commentary or assessment of it). But there are also significant differences between archaeological and historical linguistic sources. The most frequently cited difference is that archaeological materials, like textual sources, are direct: they were produced by the individuals whose lived experiences we seek. Historical linguistic evidence is indirect: it is the product of upstreaming (Vansina 1990), of explaining the patterns of the present—whether the distribution of related languages, the meanings of their words, or the similarities of their sounds—as outcomes of the past. We will explore these similarities and differences in the context of the case studies, returning to them in the conclusion. Stahl’s approach to working interdisciplinarily thus informs our efforts to work between archaeology and historical linguistics. Like Stahl, we emphasize a supplemental approach to archaeology and historical linguistics, which allows us to exploit the possibilities of these distinct types of data, cognizant of their limitations. And yet, we also hope to demonstrate in the case studies that follow that archaeological and historical linguistic approaches provide similar access to data that reveals evidence of historical social practices, of lived history. A supplemental approach to working with these sources demands that we examine not only how they may offer similar or different types of historical understanding, but also the way they were produced. The methods of our disciplines are thus crucial as well, as they determine the shape and quality of the data. We offer here short descriptions of the methodologies used by archaeologists and historical linguists. These are not meant to be exhaustive treatments but rather general introductions. Readers seeking further detail should consult titles listed in the Methods section of the bibliography. These sections serve to document certain issues that are common to both fields, related to dating, scale, samples/sampling, preservation, and the nature of our access to the lived experiences of past people.

 ow Archaeologists and Historical Linguists Generate H Information About the Past In what follows, we describe the basic methodologies of archaeology and historical linguistics, attempting to characterize what we believe are important methodological concerns for a supplemental approach. There are many aspects of these methodologies that resonate between the fields, including concerns with scale, dating, and the processes by which data persist into the present. For some of these, such as dating, we see similar issues, for example the challenges in constructing firm chronologies. For other issues, such as scale, there are real distinctions between the qualities of the data produced: the spatial scale of historical linguistic research is often greater than that of archaeological research, far exceeding the largest ‘regions’

[email protected]

6

1 Introduction

in which most archaeologists work. Understanding such distinctions in methods and the data they produce is crucial to working supplementally with archaeology and historical linguistics. Archaeological Evidence  Archaeologists work with a wide range of material evidence to answer questions about the past. These include artifacts, features, and other environmental and ecological data. Artifacts are the most commonly known category and include such items as ceramics and lithics (stone tools)—these are materials made or altered by humans.1 Features are less moveable items like houses and graves, as well as landscape features like field system, roads, and pits. Other sources of archaeological data include animal bones, plant remains, pollen, phytoliths (silca bodies that form in plants), soil chemicals, and a host of other data types that help to interpret the past when they are found in the context of an archaeological site or alongside other evidence of human activity. Archaeological data are often distinguished from other forms of historical evidence in not intentionally recording past activity, approximating what Trouillot (1995) called historicity 1, or ‘what happened in the past,’ contrasted with historicity 2, ‘what is said to have happened.’ McIntosh (2005:58) likens this distinction to that found in courtrooms between testimony and evidence. Although testimony often represents first-person accounting of an event, it is a narrative created with intention. Material evidence, such as fingerprints or clothing threads, have no intention—they are left behind; they can be found, but require interpretation. Deetz (1996:212, 259) emphasized this distinction, arguing that material culture offers a degree of objectivity not afforded by historical sources: “material culture…holds the promise of being more democratic and less self-conscious in its creation than any other body of historical material.” Although this may be true, archaeological evidence also requires a great deal of processing, analysis, and interpretation in order for archaeologists to understand its historical consequences. How archaeologists find this evidence—and what evidence they seek— depends on the types of questions they ask, which is the first step in the archaeological process. In popular imagination, archaeologists spend their days digging slowly, with small brushes, revealing small fragments of the past. This, however, is just one possible stage in the archaeological process, the stage that involves fieldwork. But before any excavations occur, archaeologists must formulate questions and develop methodologies that allow the possibility of answering them. Where they choose to conduct field research follows from these questions and methods, as the archaeologist needs to have confidence that the site or region will offer the possibility of providing the evidence to answer the questions posed. Only after questions have been posed, methodologies developed, and the archaeologist has gained an understanding of previous archaeological research as well as the region’s geography, geology, and ecology, do archaeologists begin field research.

 Bolded terms in the text are defined in Appendix A, a glossary for non-specialists.

1

[email protected]

How Archaeologists and Historical Linguists Generate Information About the Past

7

Most research projects will begin with some type of archaeological survey and/ or testing, before excavations are carried out. These include a range of techniques such as systematic pedestrian surveys (with clear sampling strategies), geophysical surveys to look beneath the ground, geoarchaeological testing to understand the nature of soils and how and why they have changed, and small test excavations. With the knowledge gained from these activities, archaeologists might carry out larger area excavations, to reveal deep stratigraphies and/or spatially extensive areas such as houses or settlement layouts. In addition to the collection of artifacts and the recording of features, archaeologists also take soil samples for the study of charred plants, phytoliths, and soil chemistry. And the work extends far beyond surveys and excavations, as all materials need to be processed, analyzed and classified. This includes washing and sorting materials, but also recording qualitative and quantitative variables such as the shape, color and composition of ceramics, glass, and metals. Archaeological Approaches to Space and Time  All archaeological research is explicitly spatial, and the recording of the spatial extent of archaeological remains is a key feature of research. As presented above, archaeological research works at different spatial scales, extending from small scale (that of the artifact/feature) to larger loci of human activity (the site), to larger collections of sites and other signatures of human activity such as road or field systems (the region). The primary work of the archaeologist is thus to record the spatial context of archaeological materials at these different scales—artifacts and features within sites, and sites and other traces of human activity in regions. Thus, the context of archaeological materials is perhaps the most important aspect of archaeological research. Although the analysis of objects and features is a key part of archaeological research, archaeologists must know where and with what they are found in order to interpret the past; as McIntosh (2005:67) succinctly states: “it is the context, rather than the artifact itself, that provides the greatest wealth of information for interpretation of the past.” In excavations, the contextual data recorded will include where artifacts/features were found (in three dimensions), as well as what other objects were found in association, and in what type of material the archaeological materials were found (matrix). In surveys, archaeologists clearly identify the parameters of the region they investigated (survey universe), the scales at which they carried out their surveys (survey units), and how they looked for archaeological materials (survey procedures: survey teams, inspection of ground surface, subsurface testing). In both excavations and surveys, archaeologists are of course interested in the objects and sites found—a particular type of ceramic vessel, or a site with evidence of unique type of production—but to draw on these to interpret the past requires that they be situated in their particular contexts. When examined in context, artifacts and other archaeological materials are part of larger assemblages made up of diverse materials and associations. For example, a particular ceramic vessel found in an excavation will be analyzed individually for its properties (e.g. its shape, materials, decorations), but the interpretation of that vessel requires that it be set within an

[email protected]

8

1 Introduction

assemblage that includes other materials found with it as well its location within an excavations. Thus, a particular ceramic vessel would most likely be part of a larger assemblage of other pottery, and it is from this larger dataset that an archaeologist would build an understanding of pottery use in that time and place. Likewise, the location of a particular ceramic vessel is crucial to interpreting activity. Was it located in a house, a public building, or in a trash pit? Was it intentionally buried in the foundation of a structure? Did it contain residues of food or other materials? Such contextual clues provide insight into dimensions of use and meaning. Formation Processes and the Durability of the Archaeological Record  Thus far, the importance of understanding archaeological context has been framed around an assumption that the materials found by archaeologists are in a meaningful context, one directly conducive to historical interpretation—for example, a ceramic vessel with starch residue set in a house floor might be interpreted as evidence of cooking or storage. However, such assumptions and determinations can only be made after archaeologists have considered the depositional processes that created the context in which archaeological materials are discovered. Archaeologists guard against the assumption that all materials are found in the exact context in which they were used; this is often referred to as the ‘Pompeii Premise,’ referencing the famous Italian site that was rapidly buried in ash, thus freezing all objects and people in place. Although such sites do exist they are the extreme exception rather than the rule. Since the 1970s, archaeologists have actively investigated the formation processes of archaeological sites themselves (Schiffer 1976). This means that they use an understanding of natural and human activities to assess the context of archaeological remains. One determination that they must make is the type of deposition that they are examining. Materials that are deposited in the precise location in which they were generated are often called de facto refuse—these are materials that are left behind at abandonment, and approximate the Pompeii Premise. Most archaeological materials, however, are either primary or secondary refuse—these are materials found in deposits that represent either the primary discard of those objects (throwing material in a trash pit that is quickly covered and subsequently undisturbed) or materials that suggest some evidence that the materials were subsequently disturbed/used/ redeposited once again. There are a number of different ways that archaeological deposits could have been altered in the past, and archaeologists make every effort to assess the ways that the archaeological record was altered through time. Natural process, like erosion, freezing and thawing of soils, and disturbances from roots, rodents and bugs can have a significant effect on archaeological deposits. All of these processes can alter the context and association of materials in an archaeological site. Human activities also alter archaeological deposits—archaeological materials can be moved and redeposited, reused for other purposes, and transformed by activities like trampling and farming. Archaeologists must also assess whether their assemblages are representative of past materials and activities. Such an assessment requires archaeologists to recognize that they only ever have a sample of material from the past and thus must

[email protected]

How Archaeologists and Historical Linguists Generate Information About the Past

9

c­ onsider how well that sample serves their research goals. In devising a methodology to investigate a site or region, archaeologists will establish a clear sampling strategy. All surveys and excavations are a sample because archaeologists are not able to dig up an entire village or city, nor is it likely that they will be able to survey every square mile of a region. Thus, archaeologists attempt to retrieve a sample (of artifacts, houses, sites) that represents the broader patterns of society. There are many ways that a sample may be considered problematic or inadequate. Small samples of artifacts—especially materials like ceramics and lithics—may be inadequate to understand the full range of activities that these materials represent. For example, at a Swahili stonetown site, samples of ceramics from a range of contexts should be sought—from trash pits of elite and non-elite houses, as well as materials in primary and secondary context. One benefit of materials like ceramics and stone is that they are relatively resilient and remain generally unaltered in archaeological deposits. However, other materials like animal bone can be greatly affected by soil pH and chemistry and thus faunal assemblages might be missing certain types of small or easily degraded bones. Without thinking about the effect of this on a faunal sample, the interpretation of diet and consumption may be skewed toward those bones that are present. Although there is no precise measure of what can be considered an adequate sample of material, archaeologists must discuss the possible limitations of their samples and how they attempted to mitigate these issues or add caveats to their interpretations. Thus the assessment of the context of archaeological materials and samples is perhaps one of the most difficult and important activities that an archaeologist must do, and it requires careful attention to stratigraphy, soils, and the assemblages of artifacts found. Therefore, when archaeologists are excavating and recovering artifacts and features, they are not only attempting to record things in the context recovered, but they are also simultaneously assessing whether the context in which they find archaeological materials is primary, secondary, or even further displaced in time or space, and whether their sample is representative of the activities they aim to investigate. In reporting their findings, archaeologists need to be explicit about how archaeological deposits were transformed through time and bring this discussion into the way their interpretations are formulated. Archaeological Temporality  The strength of archaeological research is its ability to build arguments about changes over the long term, including time scales that include hundreds and thousands of years. There are many ways that archaeologists build chronological frameworks for their data, including both relative and absolute dating techniques. Relative techniques include stratigraphy, in which the order of materials in layers can be used as means to determine what is older and younger, and seriation, where the form and style of materials can serve as a basis for relative ordering. Absolute techniques allow for archaeological materials to be assigned a calendrical date and include techniques like radiocarbon (14C) dating and dendrochronology (tree ring dating). Although absolute dating techniques have served to transform archaeological interpretations by allowing for more precision in dating archaeological materials, archaeologists must use and present them appropriately.

[email protected]

10

1 Introduction

Because radiocarbon dating is the most common technique used by archaeologists, we focus on it. Archaeologists can only apply radiocarbon dating to objects that contain the requisite amount of carbon (e.g., charcoal, bone, shell). Therefore, archaeologists will often use samples of charcoal that they find in association with the materials they aim to date (e.g., ceramics). This can be a relatively clear cut association—charcoal from a hearth to date the final use of it—but often the archaeologist must examine formation processes to establish that the material being sampled for dating is directly associated with the material or feature that the archaeologist hopes to date. For example, charcoal associated with a builder’s trench may be used to date the period of construction, but charcoal found in a house hearth will likely date to the period of abandonment of the structure. All absolute dating techniques have limitations. For example, radiocarbon dating requires a minimum amount of material to date, can only be applied on materials up to 50,000 years before present (bp), and will vary based on the material sampled. All of these factors (and more) must be understood when submitting materials for analysis and the results used for interpretation. Radiocarbon dates must also be calibrated (through simple computer programs, eg, OxCal, Bronk Ramsey 2009, 2017) since it is now recognized that the concentration of 14C in the atmosphere varied over time, with implications for the precision of radiocarbon dating (Bronk Ramsey et  al. 2006). When radiocarbon was first developed and used, this was not recognized and so older archaeological reports often contain uncalibrated dates. Radiocarbon calibration can alter dates dramatically, which adds further complications in how radiocarbon dates are presented and interpreted. Today, there is another way that archaeologists manipulate radiocarbon dates in an effort to increase their ‘precision’, through Bayesian modelling. This type of modelling uses additional information, such as stratigraphic and contextual data, or other dating techniques, to generate estimates that can be more precise than those offered through simple calibration. Such modelling must also be reported fully to assess the quality of the chronological estimates it produces (Bayliss 2015). There are only a few applications of this in African archaeology (e.g., Loftus et al. 2016) but this will change quickly. Perhaps the most common way that radiocarbon dates are misused is in their presentation: the results of radiocarbon dates are ranges rather than precise calendrical dates, and they must be presented as such (e.g., AD 1200–1400). Using only the earliest (e.g., AD 1200), the latest (AD 1400) or the center (AD 1300) date in that range is not appropriate—the ranges are not periods of occupation or use, but ranges in which there is a significant chance that the material was used and deposited. Archaeologists can present ranges in either one or two sigma confidence intervals—one sigma means that there is a 67% chance that the dated material falls somewhere in that date range; two sigma means that there is a 95% chance. The latter is always a wider time frame, but provides the most statistically reliable date range. Making Sense of the Archaeological Record  The discussion thus far has focused on how archaeologists go about conceptualizing their methods, how they carry them

[email protected]

How Archaeologists and Historical Linguists Generate Information About the Past

11

out, and how they assess the quality and integrity of the data they have collected. It is imperative that archaeologists provide a discussion of their methods and approach augmented by regional maps, excavated sections and plans, and a description of artifact categories (and how they constructed them). They must also fully account for how they constructed their chronological estimates and how they assigned particular artifacts, features and sites to different periods. Such detail, which is often presented in an excavation report or site monograph (e.g., Horton 1996; McIntosh 1995), is crucial for assessing the reliability of the archaeologist’s interpretations. One of the primary ways that archaeologists make sense of the archaeological record is through the application of analogy. Analogic reasoning governs the way that particular objects might be interpreted as ‘tools’ (analogies based on form or ‘formal analogies’), as well as how regional site patterns can be interpreted as representing ‘states,’ other complex socio-political formations, or hunter-gatherer economies (sometimes called ‘relational analogies’). Archaeologists work in  a broadly comparative framework, using analogies from different cultures, times, and places to build arguments about the emergence of particular types of social forms, power, and agency. However, one particular form of analogic reasoning—the direct historical approach—restricts analogies to those from historically-related communities, especially where there is a demonstrated continuity between the past and present (Steward 1942; Stahl 1994). Wylie (1989) and Stahl (1993; 2001) have described how archaeologists must attend to both source- and subject-side concerns when working with analogies. Source-side concerns are those that focus on how an analogy is chosen: was it based on perceived similarities in economic organization or ecological context, and does that provide the most useful criteria? Subject-side concerns are those that examine how archaeologists apply analogies to archaeological data, to evaluate their effectiveness in revealing patterns in the past. Such concerns are tantamount in archaeological reasoning; it is often in these steps that problematic assumptions are introduced. For example, Stahl (2001:37) discusses the ways that colonial models of ‘native’ people were often written into seemingly objective ethnographies, and thus the use of these ethnographies to probe pre-colonial social formations must therefore “interrogate the categories and knowledge claims of archives as they relate to practice and power.” Wylie (1989; see also Stahl 2001) offers one important way to do this: through “tacking,” moving back and forth between the subject and source to explore not only points of convergence but divergence as well. In this way, analogies are used not only to reveal patterns that existed in the analogic source material, but also expose those that emerge from the subject material, as “disjunctures or contradictions” (Stahl 2001:39). Archaeological interpretations are often written after the fieldwork is complete, assemblages are analyzed, and chronologies constructed. These range from: basic explanations of the chronology of an artifact or site; functional explanations of a technology or settlement; interpretations of social, political, economic or religious activities of a community or society; or insights into how people constructed meaningful identities, social relationships, and communities through their engagement

[email protected]

12

1 Introduction

with material objects. Through all these, we can see how archaeologists are not just interested in ‘where’ and ‘when’ questions but also ‘how’ and ‘why.’ Interest in the latter characterizes anthropological archaeology, marking a shift from so-called culture historical, to processual and later to postprocessual approaches (Johnson 2010). Suffice to say that archaeologists have learned much from each of these approaches, and rather than simply rejecting one approach for the next, have built upon them. Although some archaeologists would eschew postprocessual approaches today, the understanding that humans use material objects in the constitution of their social and cultural lives is a key insight that places the material record front and center in building a rich, lived history. These insights may involve daily practices akin to Bourdieu’s (1977) habitus, the culturally-learned/situated habits, tastes and skills that people use to live and act in the world, but also public, ideological, or cosmological statements. The construction of monuments and their arrangements may offer narratives that particular groups or individuals attempt to convey, akin to historicity 2; Ashmore and Sabloff (2002) provide a clear example with Maya civic plans, arguing that they present cohesive visions of Maya cosmological and political order. And yet, within the context of those monuments, archaeologists can recover material traces of how people lived within them, how they built around them, deposited their trash near them, defaced, altered, or destroyed them: in this way, the contextual associations of archaeological data offer evidence for historicity 1. Historical Linguistic Evidence  In the African context, historical linguists use two kinds of ‘raw’ data. First, they study recorded instances of speech in extant languages, often from their own fieldwork. This fieldwork involves collecting examples of ‘natural’ speech and interviewing speakers to elicit wordlists, translations, examples of particular phonological, lexical, grammatical or other linguistic features in different sentence constructions, and so on. Second, historical linguists study documents that record information about and in African languages (dictionaries, grammars, maps with local place names, published letters, newspapers, etc.). The assumption guiding historical linguists’ research is the idea that languages (however defined) change over time as speakers learn to communicate in particular social contexts, adapting their efforts in the face of changing goals and circumstances. Over very long spans of time, these changes can result in the development of speaking styles that differ sufficiently from previous styles as to merit designation as a new language. New languages develop from older ones in a slow process. The innovations or changes that differentiate styles of speech (by region, gender, occupation, etc.) slowly aggregate until we can identify the emergence of a new language, which has diverged from its linguistic ancestor, or protolanguage, in a process like cell division (mitosis). Language divergence results from internal innovations or borrowing styles of speaking, particularly loanwords, from other speech communities (groups of people speaking a common language). Similarities between languages belie historical connections between them, whether these historical connections are the result of descent from a common linguistic ancestor or contact and borrowing across language boundaries in the past.

[email protected]

How Archaeologists and Historical Linguists Generate Information About the Past

13

To sum up: historical linguists identify (1) genetic relationships between modern languages and among their protolanguages; (2) changes unique to specific generations of protolanguages; and (3) innovations borrowed across linguistic boundaries. To track historical patterns of inheritance, innovation, and contact through time, historical linguists compare across very large numbers of languages fairly small units of linguistic features, such as—but not limited to—phonology (e.g. the pronunciation of consonants and vowels or tonal patterns), lexis (individual words but also assemblages of words, whether lexicons relating to an activity like iron-­ smelting or collections of synonyms, antonyms, or homophones), and morphology (meaningful units from which words are ‘built’, such as noun class prefixes, verbal affixes, etc.). A firm classification or ‘family tree’ of the genetic relationships between modern languages and their historical antecedents (for an example, see Fig. 2.1) is essential to historical linguists’ efforts to distinguish between common linguistic features resulting from inheritance from a shared, ancestral protolanguage and common linguistic features resulting from contact and borrowing between protolanguages in the past. Historical linguists begin the task of classifying languages—establishing their genetic relationships—by identifying a group of languages that are hypothesized to be related because they appear similar in grammar, lexis, morphology, phonology, and so on. There are several methods available for developing preliminary classifications; lexicostatistics (counting cognate rates within core vocabulary, which is very basic vocabulary thought to be resistant to change), phylogenetic modeling, and near-neighbor network analysis have been applied in the African context (e.g. Bastin et al. 1999; Grollemund 2012; Holden and Gray 2006; see also Marten 2006). However, the comparative historical linguistic method remains an important part of both the generation of the data analyzed in such methods (recognizing cognates, for example) and in confirming the subgroups proposed through such methods. In the comparative method, historical linguists compare the linguistic features of the hypothesized relatives as well as languages spoken nearby and languages hypothesized to be more distantly related. This comparison allows the historical linguist to trace the boundaries of the genetic group and its internal subdivisions, called branches or subgroups. Subgroups are identified by innovations in phonology, morphology, grammar, lexis (internal innovations and loanwords), and other features that can be proven to be unique to the group in question but absent in (proto) languages beyond it. A single innovation is a poor foundation for a genetic group, but many different kinds of innovations—in lexis, in morphology, in grammar, and in phonology—unique to that subgroup constitutes a solid body of evidence for its validity as a genetic group. It should be noted that some languages have been particularly resistant to classification, either because they are unusually conservative or unusually innovative. The comparative method illuminates the historical development of features of languages, but such developments were contingent on the actions of speakers. For this reason, linguistic evidence can tell us about human history. For the sake of simplicity, we might divide the historical evidence developed through historical linguists’ methods into two categories. Historical information embedded in the

[email protected]

14

1 Introduction

classification addresses the development of the language family as a totality, as a process to be explained. With the classification, all linguistic data are brought to bear on the same question: the historical development of the language family. But smaller units of linguistic data, particularly lexis, illuminate developments well beyond the narrow focus on language history precisely because speakers use language to conduct everyday life. Scholars working in different regions, disciplines, and collaborative groups have different stances toward the two kinds of histories these categories of linguistic data illuminate. This is a matter of epistemology that impinges on debates about methods, and we will return to these points at the end of the chapter. First, we need to know more about the kinds of narratives these data support. Classifications: Populations in Time and Space  Inherent to the classification of a family of languages is a story of settlement, movement, contact, and learning by speech communities. Methods to locate protolanguages and their speakers in time and space assume that human mobility was a fundamental reason that languages diverged, at least in earlier periods of human history. Sometimes, as in the case with the settlement of Polynesia, migration cut off all communication between speakers in the linguistic homeland and those who moved away. In the case of the Bantu languages of Africa, speakers of a protolanguage generally expanded at their frontiers, such that speech at the geographical periphery underwent different changes (perhaps also more rapidly) than speech at the core, usually as a result of contact with neighboring communities, new environments, and so on. Eventually, these changes rendered speech at the old linguistic core and at the periphery mutually unintelligible. Of course, this idealized model does  not capture the messiness of such processes in practice, especially in the context of widespread multilingualism and high degrees of mobility from and toward the core. Nevertheless, scholars have very approximate ways of estimating the dating and geographies of language change. There is a relative chronology inherent to the classification itself, but the divergences of particular protolanguages can be assigned very approximate absolute dates through glottochronology, a much-debated method that generates approximate dates of divergence by assessing cognate rates against a presumed rate of change in basic vocabulary.2 The period during which protolanguages were spoken can also be dated through direct associations with the archaeological record (de Luna 2012a; Vansina 2004). Protolanguages can be located in space through the principle of least moves, the principle of greatest linguistic diversity, direct associations with the archeological record, or all three (Crowley and Bowern 2010; de Luna 2012a; Dimmendaal 2011; Ehret 2010; Vansina 2004).  The debates turn on whether changes in core vocabulary average out to near-constant rates of change over very long periods of time. Surprisingly, even strident opponents of glottochronology advance alternative methods following the same universalizing logic with respect to rates of language change. See comments in Chap. 5, note 3. For debates on the method in the African context, compare Vansina (1990, 2004) to Ehret (2000, 2011). 2

[email protected]

How Archaeologists and Historical Linguists Generate Information About the Past

15

After dating and locating speech communities associated with the protolanguages of a classification, we can ‘read’ classifications as the linguistic outcomes of earlier periods of human mobility and contact. As evidence of demographic history, classifications suggest that the lines linking generations of language change reveal the geography and pacing of the human mobility, interaction, and settlement that undergirds language change. Scholars working in all regions of the world have used just this sort of historical evidence to make arguments about regional prehistories, often connecting the linguistic and archaeological (and sometimes genetic) records in their interpretations, as we explore below. Words and Things  The words, reconstructed to particular protolanguages, provide another form of historical evidence. The technical process of reconstructing the form (sound shape) and antiquity of a word is fairly straightforward. A word’s phonological shape and distribution in extant languages determine whether it was present in the lexicons of protolanguages within the classification. A word’s phonological shape and distribution also tell us which of three historical processes is responsible for its presence in a protolexicon: inheritance, internal innovation, or borrowing from another language. Deviance from the regular sound changes definitive of protolanguages (and subgroups) indicates borrowing; adherence to regular sound changes signals inheritance or innovation (distinguished through the distribution of the word in the languages of a subgroup or family). Some words, however, cannot be reconstructed because they do not contain any of the sounds that underwent change in a particular group of related languages. Like archaeologists, historical linguists must also think about their evidence as a sample. The words and things approach assumes that if a word can be reconstructed to a protolanguage, the speakers of that protolanguage likely made use of the object or idea named by the reconstructed word. The fact that words are formed from even smaller linguistic units—of sound, morphology, meaning, and so on—matters a great deal in the reconstruction of protolexicons. Starred forms or protoforms— the words reconstructed to particular protolanguages—can only be proposed when the historical development of the smaller units that comprise words are carefully studied, a process that can take weeks or months for a single word, especially in larger or poorly documented language families. The collection of sounds that comprised the word are reconstructed on the basis of sound change rules and the clutch of most likely meanings to be associated with the word are reconstructed on the basis of derivation, morphology, and comparative semantic analysis (see below). Historical linguists represent protoforms in a manner identical or similar to this one: *-pɩ ́-, “to be burnt, hot, cooked, ripe” (Bastin et  al. 2002, Main 24913). The star marks the form as a proposed protoform. In the Bantu context, accents mark tone and, when they are nouns, reconstructions are usually followed by a parenthetical reference to the noun classes in which the root was used with the meaning under  Following convention, citations to individual lexical reconstructions in databases and compendia, such as Bastin et al. (2002) and Guthrie (1967–1971), will include the identification number of the reconstruction, rather than the page number. 3

[email protected]

16

1 Introduction

discussion. In this sample, the dashes represent the location of prefix and suffix, respectively, which are added in the process of conjugating the verb root. Social Context and Meaning  How do we know what ‘words’ (as collections of sounds that can be reliably reconstructed into the fairly distant past with some measure of accuracy) meant to speakers and listeners who used them in the deep past? There is a continuum to how scholars approach this important question. On one end, there are scholars who reject the very possibility that the meaning of reconstructed words can be known for earlier periods because meanings do not follow the kinds of patterned rules that govern other aspects of language change studied through the comparative method, such as sound change. Other scholars take a middle ground, accepting meanings attached to reconstructed words for materials that can be confirmed through other, independent historical records, such as the archaeological record. At the other end of the spectrum are scholars who seek out methods and evidence to illuminate the most likely social contexts in which words’ meanings were generated, contested, and put into the workings of everyday life. Necessarily, these three stances toward the meanings of reconstructed words reflect different stances toward evidence, method, and knowing the past. Here, we describe some of the strategies used by scholars in the second and third groups, who seek to draw as near as possible to the most likely meanings people in the past associated with clusters of sounds (words) reconstructed through sound change rules.4 These strategies set the stage for our discussion of methods for linking archaeological and linguistic evidence (section “Connecting Archaeological and Linguistic Evidence: Three Approaches”) and their relation to the supplemental approach to interdisciplinary research advocated by Stahl (section “A Supplemental Approach to Archaeology and Historical Linguistics”). The problem of meaning can, of course, be approached as a matter of fact. As such, the meaning of reconstructed words in earlier periods can be ‘proven’ by the recovery of evidence of the named object (a new arrowhead style or kind of building material) or the named practice (a new pottery firing technique or control of reproduction through a novel marriage practice with its attendant genetic traces) in another, independent archive, such as archaeology, genetics, soil science, and so forth. Success in the independent confirmation of meanings proposed for protoforms should inspire greater confidence in meanings proposed for protoforms relating to domains of life that cannot be similarly tested. But the ‘confirmation’ of the existence of a practice, technology, or object gives us little sense of the social context of its invention or contestations over its use and value—matters of great concern in the humanistic social sciences and humanities. Moreover, misfits,  Significantly, many debates about the validity of methods used to reconstruct the meaning of protoforms stem from the limits of older forms of lexical reconstruction and ignore more recent linguistic theories supporting methods of semantic reconstruction. Debates about Indo-European protoforms are particularly good examples of the limits of lexical reconstruction and the potential of the methods of semantic reconstruction undertaken in other world regions (e.g., Fleisch and Stephens 2016). 4

[email protected]

How Archaeologists and Historical Linguists Generate Information About the Past

17

contradictions, and silences within and between records are particularly revealing, a point we discuss in the chapters that follow. This brings us, then, to methods for discerning the social context of words’ use and the impact of that context on the creation and contestation of meaning. Although rarely compared, historians using textual sources and historians using language evidence share the strategy of comparison as they seek to know what their historical subjects meant in the traces we recover of their interactions. In trying to understand the practice of marriage, the nature of affection, or the meaning of work, historians using both textual and linguistic sources seek out patterns that reveal a shared core of understanding and a range of alternatives developed as people negotiated the possibilities of their time and space. Most historians seek out this common core and range of possibilities by comparing as many kinds of documentary references as possible within the confines of the various archives relevant to their study, often paying particular attention to the pragmatics (context-specific usage patterns) associated with different genres of text (letters, district commissioners’ reports, census records) as well as matters of identity (gender, class, race) and locality. Historical linguists do the same, pouring through proverbs, ethnographies, life histories, dictionaries, grammars, field interviews, and so forth but they do so across multiple— even hundreds—of languages and dialects, producing hundreds or thousands of individual datapoints, each assessed for the context of its historical production: for example, the influence of Christian ideals on missionary dictionaries or the impact of more recent population movements on evidence of cross-linguistic contact.5 At the most general level, the core meaning of a reconstructed word in the African context is discerned by analyzing an archive made up of the hundreds or thousands of data  points (dictionary glosses, sample sentences, proverbs, stories, etc.) collected in modern evidence (usually from the nineteenth century onwards in the Bantu-speaking African regions) within tens or hundreds of languages. The sheer volume of data that conform to broad patterns of use and meaning should give the skeptic pause. When reconstructing a lexicon, each word receives this pain-staking treatment. And, some parts of the lexicon might also generate opportunities to confirm the antiquity of meanings through independent historical records, lending greater credence to the proposed meanings of those non-material word reconstructions in the lexicon. This initial assessment of patterns to discern core meanings within large volumes of data constitutes an analysis of the distributions of meanings attached to collections of sounds, the antiquity of the latter of which can be verified through an analysis of the phonology. Yet, the antiquity of the word (meaning and sound pair) is hypothesized as (but not proven to be) the most likely cause of its pattern of distribution across the vast volume of data (consider Stephens’ (2016) application of Geeraerts’ prototype theory). It is, for example, more likely that the meaning “tree” adhered to the proto-Bantu *-tí than that the meaning “tree” was invented hundreds of times independently for the same cluster of sounds, or that it rapidly  Historians using documentary evidence also track such influences: Hanks (2010), Landau (2010), Peterson (2004). 5

[email protected]

18

1 Introduction

spread in the recent past, after all the languages using this demonstrably ancient sound cluster emerged as extant languages. This effort to historicize distribution and meaning is best buttressed by attention to the process by which the word itself was created. Words’ construction carries information about the contests and contexts surrounding words’ invention. The morphology that speakers mobilized to derive new words from common roots, the range of meanings they tied to the root from which they developed multiple words, and the layers of meaning attached to each of those words over time all contain historical information that illuminate the social contexts of debates about knowledge and activity. How a word was ‘built’ from morphological features—its derivation and etymology—teaches us about the conceptual leaps speakers made when they invented a word for a new idea, object or practice. For example, the ancient Bantu root word for “spirit” (usually ancestral spirit) derives from the verb “to extinguish,” revealing layers of meaning that mark a conceptual tie that does not exist in the English word for “spirit,” which derives from the Latin spirare for “breathe.” But words are also connected through their meanings to other words developed from completely different roots. Synonyms and antonyms are good examples of this phenomenon. Words may be connected to words developed from completely different roots through practice, thereby constituting a specialized lexicon. For each word in these specialized lexicons, we can trace out the historical information embedded in morphology, the semantic ranges of roots, and changing meanings to produce a far more complex body of historical information about past practices; studying the individual words as part of a lexicon offers a way to firm up—but never fully, factually confirm— non-material aspects of semantic domains. To sum up: we may apprehend the layers of meaning speakers attached to words over time and, thereby, illuminate how inherited terms were kept relevant in changing historical circumstances. The words and things approach is underpinned by far more evidence than the mere glosses attached to cognates in dictionaries and wordlists. Ethnohistorical data such as proverbs, oral traditions, and riddles provide examples of words’ use in context (pragmatics). Patterns in word use across language boundaries are vital for identifying: (1) the range of meanings of ancient root words; (2) the most likely—but not the absolute—history of semantic change accounting for diverse meanings observed today; and (3) the social, political, and economic contexts in which words were used in the past. Social Context & the Ethnographic Record  Like archaeologists, historians using linguistic evidence to study the past make use of more recent and even contemporary ethnographic data. As noted above, the ethnographic archive is a rich trove of information about words’ use in context (pragmatics), which is an important resource for reconstructing the historical development of protoforms’ meanings (Stephens and Fleisch 2016). In this way, ethnography is an important resource for thinking about the social contexts of meaning-making. Yet, historians and archaeologists differ in their use of the ethnographic record, even as both study ethnographic evidence for patterned similarities.

[email protected]

How Archaeologists and Historical Linguists Generate Information About the Past

19

In the case of formal or relational analogies, similarities across case studies facilitate archaeologists’ analogical thinking about the structuring of economic practice or political authority in different times and places. For archaeologists, the objects of the archaeological record and those of the ethnographic record are similar (in form, function, production, etc.) to the archaeologists’ practiced eye (and, quite often, their interlocutors!) and this is sufficient validation for the use of ethnographic analogy, whether an analogy between a recent and more distant past in the same region or an analogy comparing across regions. In contrast, historians are interested in patterned similarity resulting from common historical contexts, an interest shared by archaeologists using the direct historical approach. To note that communities in both Çatalköyük (central Turkey) in the seventh century BC (Hodder 2006) and societies living on the Batoka Plateau (central Zambia) in the late first millennium AD used similar strategies of incorporating the bones of wild fauna into houses as a means to domesticate ‘wild’ spaces is an acceptable observation in archaeology because it foregrounds the way that different societies solved similar problems in different periods and places (see Chap. 3, note 2). However, it is not as valuable an observation to the historian because the reasons for the similar strategies do not share a common historical cause; indeed, these societies were never in contact. Thus, historians using ethnographic data alongside linguistic evidence bring in patterns of similarity in ethnographic data as evidence if and when such patterns might be proposed as the outcome of a common historical process. What is unique about historians’ use of ethnographic evidence is the comparative treatment (Sapir, 1916) of a regional ethnographic record. Because words naming the ethnographic objects, actions, titles, etc. in the regional ethnographic record and those words’ protoforms are historically related to one another, historians use the regional ethnographic record in a comparative manner. Just as historians build up reconstructed lexicons, they also build up patterns of similarity in the regional ethnographic record describing the lifeways of the many modern-day speech communities whose ancestral speech communities’ history is sought. Read through this comparative lens, the regional ethnographic record offers rich evidence of redundant bundles of objects, actions, actors, and places associated with one another in the performance and contestation of the practices and institutions historians study, such as divining, bridewealth, cattle-loaning, or kingship (Schoenbrun 1998). These redundant bundles are revealed through comparison and invite historical explanation. In the process of reconstructing lexicons relating to such practices and institutions, bundles of associated objects, actions, actors, and places are often recognizable in the semantic histories of words. The durability of such bundled associations does not justify projecting descriptions of the ethnographic record into the past. Rather, they invite us to attend to the processes of change that made such associations durable in the first place—changes visible in the patterns of difference across the ethnographic record of a region and in patterns of differentiation in lexicons and the semantic domains of their constitutive vocabularies. Interdisciplinarity and Comparative Historical Linguistics  If thinking about words’ historical meanings requires thinking about social context, we are best

[email protected]

20

1 Introduction

served by learning about such historical contexts from as many records as possible. Non-linguistic archives are essential to the art of apprehending semantic histories. In a warning against the words and things method, Paul Heggarty suggests that there is no method inherent to historical linguistics6 that would keep a scholar from reconstructing to the early Germanic protoform of the modern word “mouse” the meaning of the object used to direct the cursor on a computer screen (2015: 608–609; see also Heggarty and Renfrew 2014). Of course, no one with knowledge of the history of technology or the archaeological record corresponding to the linguistic period and location of early Germanic speech would posit such a meaning! What this rather amusing example highlights, then, is the fact that efforts to reconstruct meaning are already interdisciplinary. They require: (1) the forms of historical thinking about the creation and curation of sources that are second nature in archaeology and history; (2) theories from other branches of linguistics as well as philosophy and anthropology; and (3) knowledge of datasets that stand outside those recovered by any one discipline. This is certainly also the case with archaeological approaches to meaning (see above). This important parallel has implications for a supplementary approach to archaeological and historical linguistic datasets. We will return to the strategies underpinning a supplementary approach to interdisciplinarity at the end of the chapter. But first, we must consider the methods by which distinct historical datasets can be connected in the first place.

 onnecting Archaeological and Linguistic Evidence: C Three Approaches If archaeological and linguistic data are complementary, the magnifying power of drawing the two datasets together depends on the methods by which they are simultaneously brought to bear on a common research question. The epistemologies shaping questions, research, and interpretation within and between the disciplines of Linguistics, Anthropology, and History, among others, have a profound but underappreciated effect on debates about the reliability of various methods of connecting such independent historical records. Generally, archaeologists, historians, and linguists have forged connections between the archaeological and historical linguistic records by identifying correlations, common processes, and direct associations. These three approaches are not mutually exclusive: scholars of early history often leverage more than one approach to explore evidence of human activity at different resolutions. Here, we introduce methods of connection through case studies from Africa, Europe, Asia, and the Americas. In some cases, the methods underpinning the case studies were explicitly in conversation through disciplinary and interdisciplinary debates. In other cases, lack of conversation has shaped the research agenda in ways that offer new opportunities and methods for interdisciplinary research.  Linguistic theory itself has resolved many of these debates, in no small part by rejecting the kind of lexical reconstruction Heggarty parodies. Consider strategies put forward in Fleisch and Stephens (2016), Ortman (2012), Schoenbrun (2016). 6

[email protected]

Connecting Archaeological and Linguistic Evidence: Three Approaches

21

Correlation  We begin in Africa, where, in the context of decolonization in the 1960s, scholars sought to produce a usable precolonial past for emerging African nations. The result was a productive relationship between linguists, historians, and archaeologists who leveraged linguistic, oral historical, and archaeological evidence to recover African histories predating the arrival of Europeans. The problem of the Bantu expansion—the process by which some 500 or more related languages came to be spoken across the southern third of the continent—constituted a major, interdisciplinary research problem at mid-century. It matched the magnitude of the narratives produced about other world regions, suggesting that Africans, too, had produced historical change on a continental scale and that they had done so before European colonization. In the mid-twentieth century, interdisciplinary research on the Bantu Expansion took off as archaeologists and linguists developed methods for identifying and reconstructing the pace and direction of the spread of their respective units of study—material cultural traditions and protolanguages—in order to reconstruct the earliest demographic history of the continent. Historians, linguists, and archaeologists worked to synthesize the archaeological and linguistic evidence, smoothing out misfits and disjointed timelines. Indeed, the most famous such synthesis—that of Roland Oliver (1966)—remains one of the ten most-cited articles ever published in The Journal of African History. These syntheses were developed through the correlation of the archaeological and linguistic records: scholars identified overlaps in the geography of linguistic expansion and the approximate geographical spread and radiocarbon dates of a material cultural tradition to trace the pacing and direction of the diffusion of languages, populations, and material cultural traditions (see contributions to Ehret and Posnansky 1982 for examples of this approach). Such diffusions were hypothesized to arise from common processes affecting both records, such as the spread of agriculture, domesticated goats and sheep, and metallurgy (Oliver 1966; Phillipson 1976), the so-called “Bantu toolkit.” This work adopted ideas about the revolutionary character of subsistence and technological change as explored in the Eurasian context with increasing frequency after World War II. Such narratives of the Bantu Expansion compared African prehistory to other world regions, explicitly incorporating Africa into global historical narratives first proposed by Childe (1936; see also de Luna 2016; Klieman 2003). Later, the emphasis on subsistence change in this scholarship would inspire Colin Renfrew’s influential work on Indo-European origins (see below). By the 1970s and 1980s, however, scholars debated the simplistic association Bantu Expansion research made between languages, cultures, and populations and questioned the idea that archaeologists could identify a single path of diffusion for aspects of material culture associated with Bantu languages (Gramly 1978; Kuper and van Leynseele 1978; Lwango-Lunyiigo 1976; Möhlig 1989; Schmidt 1975). Pessimism, then, defined most archaeologists’ and linguists’ attitudes toward interdisciplinary research on the Bantu Expansion problem throughout the last two decades of the century and into the next (Robertson and Bradley 2000; Eggert 2005). Yet correlation has remained common in work at a narrower, regional scale, though archaeologists, historians, and linguists proposing such correlations usually show great care not to blindly assert a simplistic equation of language with population and

[email protected]

22

1 Introduction

material culture (e.g. Ehret 1998, 2000, 2011; Gonzales 2009; Denbow 1986, 1990, 2014; Herbert and Huffman 1993; Huffman 1989; Huffman and Herbert 1994/1995; Klieman 2003; Saidi 2010; Schoenbrun 1998; cf. Ehret 2001). To some degree, this approach underlies the next two. Common Regional Processes  The study of Indo-European and Bantu origins developed together, with methods generated and tested in the Indo-European context applied to the African context and vice versa. Scholarship on both world regions faced similar critiques in the 1970s and 1980s about methods of correlating archaeological and linguistic data. In the Indo-European case, Colin Renfrew’s interventions stand out as particularly influential. With the publication of Archaeology and Language in 1987, Renfrew took a strong position against the simplistic way that language and material culture had been correlated—equating a culture with a people and a people with a language (Renfrew 1988:438). He argued that these types of approaches were a “travesty of archaeological interpretation” and a misalignment of “culture” with assemblages of similar artifacts. His approach proposed to abandon efforts to correlate particular protolanguages with particular archaeological signatures and instead link them up through archaeological evidence of processes of economic, social and demographic change. This was a shift in method from correlating archaeological and linguistic heuristics (material cultural traditions and protolanguages) to identifying common processes shaping the independent historical records, though the two methods had been combined in earlier work in Africa and Polynesia, as cited by Renfrew (1987:277–84). As with other regions of the world, Renfrew’s research on archaeology and linguistics was aimed primarily at answering a question of origins. His extended hypothesis, what has become known as the Anatolian or the first-farming/language-­ dispersal hypothesis, argues that the original speakers of Indo-European were found in eastern Anatolia in the 6th and 7th millennium BC. Like Childe some 50 years earlier (and undoubtedly influenced by him), Renfrew saw the adoption of farming as the “major process which undoubtedly radically affected the whole of Europe” (1987:146), a continental-wide transformation that would have also had dramatic linguistic effects. Renfrew’s arguments about the spread of Indo-European languages and cultures relied on two primary models: wave of advance and elite dominance. In the former, groups of farmers moved short distances in search of new farmland, thus slowly, generationally, spreading out into Europe. The latter model argues that, through military technology and organization, the speakers of Indo-European were able to dominate existing populations and ultimately impose their language on them. These two models are part of what Renfrew (1987:121) considered “the primary processes by which languages come to be spoken in a particular region.” Renfrew was quite selective about the types of linguistic analysis that should be used to reconstruct Indo-European origins; he was particularly dismissive of any efforts to reconstruct protolexicon (what he called “linguistic paleontology;” 1987:77–86). In dismissing the efficacy of reconstructing protolexicon to help reconstruct IndoEuropean origins, Renfew dismissed an entire body of data that not only could be

[email protected]

Connecting Archaeological and Linguistic Evidence: Three Approaches

23

used to triangulate against his own reconstructions, but also dismissed efforts to reconstruct the structural features of phonology and morphology (Baldi 1988, p. 447), a point to which we will return below. Renfrew’s focus on common processes had a profound effect on a generation of scholars working in a variety of world regions, including Africa (e.g. Bellwood 2005; Diamond 1997; Grollemund et al. 2015). Here, we want to focus on one case study at length because it reveals the degree to which correlation between the location and dating of archaeological and linguistic datasets continues to underlie research that has developed from Renfrew’s insights into the role of common processes in shaping the archaeological and linguistic records. David Beresford-Jones, an archaeologist, and Paul Heggarty, a linguist, have both collaborated with Renfrew and adopted his approach in their efforts to identify the “real-world driving forces” responsible for the expansions of the Quechua and Aymara across the Andes (Heggarty and Beresford-Jones 2010:164). As in many other parts of the world, early attempts to integrate evidence from the archaeological and linguistic records of the Andes correlated material cultures, language groups, and populations, an approach that stemmed from the assumption that such features of past societies overlapped in a tidy fashion. In contrast to earlier approaches and in keeping with the approach of Renfrew, Heggarty and Beresford-Jones insist that links between the distinct historical records of language and material culture must be made through “driving forces.” As they explain (Beresford-Jones and Heggarty 2011:358; italics in the original): ...our founding principle is the linguistic ‘fact of life’ that language expansions do not ‘just happen’; rather, they happen only for those very same reasons of real human demography and sociocultural context that archaeology seeks to describe through its own, independent data: the material cultural record. It follows, too, that great language dispersals must have been driven by real-world processes of commensurate scale…This principle is one we can make use of to identify correspondences between archaeological and linguistic patterns on three levels: chronology, geography, and above all causation—or in other words, when, where, and why?

Thus, Heggarty and Beresford-Jones seek historical processes of adequate chronological and geographical scale to be likely drivers of language expansions. Heggarty and Beresford-Jones’ research focuses on the language dispersals of Amyara and Quechua (Beresford-Jones and Heggarty 2012a, b; Heggarty and Beresford-Jones 2010). Correlating time and place between units in the material cultural and linguistic records remains a vital part of their method. They posit links between the horizons of traditional Andean chronology, which many (but not all) archaeologists identify as periods of significant geographical diffusion of people and ideas. These horizons also represent the expansions of the Quechua and Aymara language families across space. The Chavín Early Horizon and Wari Middle Horizon exhibit the expansive material cultures that best fit with the expectations of the spread of languages across space, given the current geographic distribution of extant Quechua and Aymara languages and toponyms hinting at linguistic substrata. Aware that the chronologies of language expansion are difficult to pin down, Beresford-­ Jones and Heggarty use the degree of internal diversity associated with stages of

[email protected]

24

1 Introduction

Indo-European expansion dated through textual sources; they apply the same approximate dates for stages with the same approximate measure of internal diversity within the Andean language families. They also argue that Aymara spread first because they identify an Aymaran substrate to Quechua (Beresford-Jones and Heggarty 2011:369–70). The correlations hinge on the observation that the chronological and geographic patterns of language and material culture fit best if Aymaran expansion is associated with the Chavín Early Horizon and Quechua with the Wari Middle Horizon. But the correlation of the linguistic and archaeological records is only the first step in identifying the ‘driving force’ that can best account for the expansion of the Quechua and Aymara families. If the scales of geography and chronology are satisfied in the correlation between Aymaran / Chavín Early Horizon and Quechua / Wari Middle Horizon, the next step, according to Heggarty and Beresford-Jones (2010), is to identify a cause for the simultaneous expansion of language and material culture (see also Beresford-Jones and Heggerty 2011). It is here that the archaeological record comes to the fore as the source of evidence for causal processes. In the case of the Andes, Heggarty and Beresford-Jones (2010) argue, the driving force of language dispersals was not simply the introduction of agriculture, as has been argued in other regions of the world (Bellwood and Renfrew 2002). Rather, they posit a two-step process. Although agriculture had long been practiced in the diverse ecologies of the Andes, it was not until the Chavín Early Horizon that maize became ubiquitous in the archaeobotanical record. Many factors contributed to the spread of languages, Heggarty and Beresford-Jones (2010) admit, but the adoption of an extensive form of agriculture facilitated expansion across the landscape in part through the (assumed) dynamics of population increase in the context of successful cereal cultivation. They attribute the dispersal of Quechua to the military expansion of the Wari state (Beresford-Jones and Heggarty 2012b). This summary of Heggarty and Beresford-Jones’ collaboration cannot capture its complexities. In the process of laying out these arguments, they make significant contributions to debates within Andean archaeology and linguistics. These include discussions of: the nature of particular archaeological horizons and the relative impact of ideology and material concerns in their respective expansions; the associations between historic speech communities and material cultures; and the history of agriculture. They also lay out methodological arguments about the relative value of language family trees and dialectic continua in modeling the expansion of languages in the past and, following Renfrew, they reject glottochronology and linguistic palaeography. To reiterate what they see as their methodological contribution, BeresfordJones and Heggarty (2012a:12) “seek to correlate language family dispersals with major expansive processes in the material cultural record.” They insist that a synthesis of human prehistory is best attempted through correlations between language families and material cultures that overlap in time, space, and, most importantly, in the commensurability of scales of change across records. Such commensurability is assumed to be caused by the same common processes—causes, as Beresford-Jones and Heggarty explain, that are “coherent above all in the commensurate scale of their impact” in each record (Beresford-Jones and Heggarty 2012b:80).

[email protected]

Connecting Archaeological and Linguistic Evidence: Three Approaches

25

The main contribution of Renfrew’s approach and its application by other researchers is the insistence that scholars focus on historical processes—what people did in the past. Yet, this approach has also been critiqued. It tends to highlight processes undergirding diffusion and struggles to attend to histories in situ. The incessant hunt for ‘fitting’ causes and ‘ultimate’ driving forces glosses over contingency and the particularities of specific historical contexts (for example, consider critiques raised in comments on Heggarty and Beresford-Jones 2010:182–7; Lau 2012; Ramos 2011). There is circularity in the reasoning by which correlations and causation are identified and also a potential for feedback between the records such that the demands of one are used to identify the ‘correct’ interpretations of another because one record (archaeology) is given primacy when it comes to evidence of what happened in the past (consider critiques in Beresford-Jones and Heggarty 2012b:77, 79). Significantly, these critiques can be addressed by incorporating lexical reconstructions into the linguistic dataset, a third approach. Words, Things, and Practice  Renfrew and others advocated for a deeper reading of the material cultural record, probing it for evidence of the kinds of large-scale processes they advanced as explanations for the diffusion of material cultures and languages. Yet, the linguistic record also contains evidence of such common processes in the lexicons developed to name and discuss them, opening a third method by which archaeological and linguistic evidence can be connected. When lexical evidence is incorporated into historical reconstructions, scholars are able to generate direct associations between the archaeological and linguistic records. Briefly, this approach links the earliest linguistic and material cultural evidence for a particular tool, activity, or practice in a region. This approach has been leveraged against a number of methodological and scholarly problems. In the African context, Jan Vansina (2004) promoted this method as a way of dating protolanguages by associating the earliest, radiocarbon-dated archaeological evidence of cattle with the first protolanguage to have vocabulary relating to cattle within the same geographic zone. In the Eurasian context, David Anthony (2007) used this method to contest Renfrew’s proposed Indo-European homeland as well as the dating, direction, and pacing of the diffusion of Indo-European languages and their speakers. Like the approach to shared regional processes in Beresford-Jones and Heggarty’s work on the Andes, this method also begins with the correlation of a speech community and material culture corresponding in time and place to its proposed homeland. In contrast, however, it has the potential to build up greater confidence in the links proposed between early forms of speech and ancient material cultures by the sheer density of proposed direct associations and the specificity and rarity of the objects and activities associated across independent historical records (de Luna 2012a; Kirch and Green 2001). In cases like Polynesia, where scholars are tracking the diffusion of languages and material cultural practices across a ‘blank slate,’ such direct associations are the most compelling. Over the past 50 years, research on the extensive island chains of Polynesia has been thought of as a ‘natural laboratory’ for studying the dispersal of populations, languages, and material cultures because the islands were first colonized in the last

[email protected]

26

1 Introduction

two millennia. Moreover, as island communities, they retain a good degree of linguistic diversity and integrity. The best published account that uses historical linguistics and archaeology is that of Kirch and Green (2001) who propose a “triangulation method” that seeks to compare independent evidence from historical linguistics, archaeology, comparative ethnography, biological anthropology, and oral traditions. They note: the analytical power of the triangulation method—and the robustness of the historical reconstructions derived from it—only holds, however if one treats each data source separately, respecting the relevant subdisciplinary methods, inferences, and conclusions as they are developed independently, based exclusively on the evidence from that field (Kirch and Green 2001:43, emphasis original).

They locate Ancestral Polynesia in Western Polynesia, “a geographically diverse island world” that includes Fiji, Samoa, and Tonga. As they explain, the confidence that researchers have in this ‘homeland’ is quite different from homeland debates in the Indo-European literature—this is because of the remarkable convergence of linguistic, archaeological, and comparative ethnographic data (Kirch and Green 2001:99), a convergence built up, in part, through direct associations. Significantly, even studies that focused exclusively on the ‘words and things’ approach have come to similar conclusions (Pawley and Green 1971; Pawley and Ross 1993). The bulk of their study, however, is a detailed description of the Ancestral Polynesian world, including subsistence, food preparation and cuisine, material culture, social and political organization and gods, ancestors, seasons and rituals. In these chapters, they draw on a robust archaeological literature, a vast lexical database (POLLEX), and a rich ethnographic literature. It is clear that Kirch and Green view the reconstruction of Ancestral Polynesian society not as an end in itself, but rather the foundation on which other studies can emerge (Kirch and Green 2001:30): reconstruction of the ancestral, structural base from which all later Polynesian societies and cultures arguably descended is an essential step that must be accomplished before more sophisticated analyses of cultural divergence and evolution can proceed.

By working with semantic reconstructions to not only place a people in a time and place, but also to understand how society was organized and functioned, their work is distinguished from the Indo-European and Andean cases. In this way, their work is truly an historical anthropology: it is the foundation on which scholars might build stories of change over time in social life. Kirch and Green’s attention to practice in the Polynesian case paralleled changes in interdisciplinary research in the African context, particularly in research on Bantu-speaking regions. Despite the hiatus of interdisciplinary research on the scale of the entire Bantu Expansion, the disciplinary research agendas of some scholars working on early Africa independently shifted in surprisingly parallel fashion away from the correlations of streams of historical evidence and large-scale processes explaining diffusion towards questions about social life. Beginning as early as the 1970s, but gaining momentum from the late 1990s, historians took up the comparative historical linguistic method to investigate early social history (de Luna 2016; Ehret 1971, 1998 among many others; Ehret and Posnansky 1982; Gonzales 2009;

[email protected]

A Supplemental Approach to Archaeology and Historical Linguistics

27

Klieman 2003; Saidi 2010; Schoenbrun 1998; Stephens 2013; Vansina 1978, 1990, 2004). Historians brought to the table a different set of research concerns than linguists using the same data. Although the history of languages and language families and the relative significance of both diffusion and in situ processes featured in their scholarship, historians were ultimately concerned with lived history (Stahl 2001), with what people did in the past. This dovetailed with similar shifts toward local history and social life in archaeology, opening the possibility for a new kind of interdisciplinary collaboration.

 Supplemental Approach to Archaeology and Historical A Linguistics The central chapters of this volume explore the potential contributions of the supplemental approach to the kinds of histories we can generate and the methods by which we do so. This requires attention to more abstract questions about how particular epistemological stances toward knowing the past impinge on problems of method, particularly those used to connect the linguistic and archaeological records. In the preceding section, we introduced three general methods by which scholars have connected the linguistic and archaeological records in their efforts to identify and explain processes of historical change. This diversity of approaches stems from the goals of the researchers. Are they are interested in large-scale global historical processes (such as the transition to agriculture or the development of language families) or small-scale historical processes (such as the politics of localized transformations in food consumption or the development of novel marriage practices facilitating the expansion of a particular population)? Research questions and interdisciplinary methods also stem from particular epistemological stances toward knowing the past, which, in turn, shape debates about the validity of the ‘words and things’ method in historical linguistics (described above) and, to a much lesser degree, certain postprocessual approaches in archaeology.7 It bears stating explicitly that we are proponents of using the full suite of methods and theories available for generating data and developing historical interpretations because we seek and practice a contextualized approach to the past. As noted above, some scholars who similarly combine archaeological and linguistic methods and datasets adhere to a strict historical empiricism. With respect to early human history (particularly histories of periods and places without textual archives), this stance is often associated with the use of linguistic and archaeological evidence in publications in scientific journals and with studies focused on large-scale processes, such as the Andes case study above. With respect to linguistic evidence, this stance either rejects the very possibility of semantic reconstruction for protoforms or it  In archaeology, symbolic and structural approaches were once criticized for positing interpretations that could not be verified (Earle and Preucel 1987); most recently, phenomenological, symmetrical, and new materialist approaches have garnered similar criticisms (Brück 2005). 7

[email protected]

28

1 Introduction

accepts only reconstructed words with meanings that can be independently verified. As a result, this strict empirical approach often silences as simply ‘unknowable’ certain classes of linguistic evidence, such as lexis. Empiricism has long been questioned in those fields focused solely on the study of the past, namely in archaeology and history. These disciplines have developed a subtle skepticism of empirical approaches as a means to accommodate the ways that various historical processes—whether the construction of archives, the vagaries of preservation, or the concerns of scholars—impinge upon our understanding of the past. A shift away from strict empiricism toward a more contextualized approach has characterized archaeology and history since at least the mid-twentieth century. To be sure, there is a paradox that emerges from empiricists’ insistence on using only independently verifiable datasets. In explaining the ‘how’ and ‘why’ of past processes, such as the transition to agriculture, these studies require the attribution of our own twentieth- and twenty-first-century logics, theories, and models of historical change, thereby eliding contextualizing evidence of how historical actors understood their own worlds and, on a more granular scale, how they may have acted in ways that did not at all follow or accord with such external, explanatory logics.8 In this way, privileging strict empiricism at the stage of assessing research methods available to produce historical data paradoxically requires a decided lack of empiricism if historical change is to be explained. Of course, we all depend on models and theories to interpret our evidence. But a strict empirical approach seeking only verifiable data ultimately obstructs our ability to hold a range of sources in supplemental tension and is, itself, a product of late-nineteenth-century ideas about the nature of knowledge and history. In this volume, we advocate for a supplemental approach to archaeology and historical linguistics, which allows us to consider the possibilities and limitations of these distinct types of data, without rejecting them as unviable if they are difficult or impossible to fully verify. Both material objects and word histories provide indirect access to the interior lives of historical subjects. Words literally speak of non-­material domains of life, but reconstructed lexis and semantics are indirect testimonies of the past. Although archaeologists recover direct evidence of the past, the very materiality of that evidence means that it offers indirect insight into non-­material domains of life. In each case, scholars must apply methods and theories to this indirect evidence and marshal it toward an interpretation of the full range of lived history. By braiding together direct and indirect testimonies of the past, diverse methods, and a range of theoretical approaches, we can keep in our sights the contingency, agency, and lived history of both past communities and, significantly, the scholarly communities to which we belong. Such efforts should be presented in a way that makes clear when the data attests to descriptive reconstructions of the past that we can verify through other datasets and when they attest to interpretative reconstructions about the motivations and  In addition to the examples cited above, consider recent debates in the African context about the role of environmental change in determining, to use the researchers’ verb choice, the pace and direction of the expansion of Bantu languages (Grollemund et al. 2015). 8

[email protected]

Scope, Organization, and History of the Volume

29

values of past peoples developed in productive tension with a range of verifiable datasets. Significantly, the supplemental approach allows us to strengthen confidence in interpretations developed from indirect evidence, thereby addressing concerns with the ‘verifiability’ of such data. When two disciplinary archives independently support converging interpretations of practices, experiences, and ideas, we can have greater confidence in those interpretations while allowing for the particular strengths and weaknesses of the data, the methods that generate those data, and the theoretical approaches undergirding the interpretation of the data. As much as interdisciplinary collaboration has and will probably continue to focus on alignment, consensus, confirmation, direct association, and convergence, points of contradiction and disjuncture are equally fascinating. Such ill-fitting evidence offers a rich vein of information about processes of change. Do changes that inspire transformation in only the linguistic or archaeological record to the exclusion of changes in the other indicate past contestations around efforts to change the status quo? Do such limited transformations indicate stalemate, unresolved negotiation, or incomplete change? Or, are such limited transformations indicating ways of acting in the world that were more about speech or the material world and less about the other field of communication and action? Do they suggest delayed adoption of change or adoption by certain segments of the population and not others? Disjunctures and contradictions are poorly suited to answering questions about origins and diffusion—concerns that have long dominated studies linking archaeological and linguistic evidence—because they inhibit the discovery of the ‘right’ answer. But, they are also rich with information for those interested in contingency, failure, and foreclosed paths, in the turbulence of life as it was lived in the ancient past.

Scope, Organization, and History of the Volume This book is an extended conversation between an archaeologist of the eastern African Swahili coast and an historian who uses linguistic evidence to reconstruct the past of south central Africa. Using our previous independent research projects as case studies, we expose our processes of interdisciplinary learning and collaboration as they unfold from the point of data analysis. We work backward, from tidy narratives to the nitty-gritty mis-readings and confusions obscured by those polished historical reconstructions. Each case study comprises two chapters. The first chapter is an attempt by one of the authors to use her own and the data from the other discipline to construct an interdisciplinary synthesis. The second chapter critiques that synthetic narrative in three steps. It opens with a formal critical commentary by the specialist in the discipline from which the author’s synthesis seeks to incorporate published evidence. At the end of the commentary, we then document the conversations we had as we sought to understand the critique and suggestions. It is in this dialog that the learning process is exposed. Finally, the critique chapter

[email protected]

30

1 Introduction

closes with a short narrative, a second attempt that takes a supplemental approach to the silences produced by our disciplines and methods, which are often given mention through the language of the other discipline and method. These narratives exploit the misfits, gaps, and incommensurability of our two datasets while also showcasing their unique strengths. It is easier to use the book, we think, if we share how it was written. We wrote the book in response to our frustrations with trying to work in regions where we did not benefit from the sort of interdisciplinary collaborations that we encourage here. We wanted to understand how a non-specialist might meet the goal of reading another discipline’s data for evidence of lived history rather than merely mine the other data for confirmation of historical processes observed in one’s own dataset. Our first manuscript consisted of our current case studies, the commentaries that follow, and our dialogues. But the peer review process forced us to take seriously what revisions might develop from what we learned in our interactions in the commentary and dialogue. It is important for the reader to understand that we wrote these supplemental readings of the archives much later than the other chapters, which, aside from some minor changes, we have generally left in their original states. In this way, the reader can better track what we learned in the process of collaborating on each case study. These supplemental readings required a new introduction and conclusion explaining what we learned through the process of combining archaeological and linguistic historical evidence with greater sensitivity to their unique qualities. In the first case study, de Luna traces the impact of food collection on the political culture of Botatwe-speaking food producers in south central Africa, c. AD 750– 1350. In the second case study, Fleisher explores the emergence of feasting on the Swahili coast, and its relationship to the negotiation of power within coastal towns, c. AD 700–1500. We hope that this documentation of interaction and learning exposes the unique interpretative power of our different forms of historical evidence as well as the methodological confusions and misinterpretations of non-specialists that we all too often hide in interdisciplinary scholarly writing.

[email protected]

Chapter 2

The Politics of Food Collection in South Central Africa

The project from which the following case study emerged (de Luna 2016) began with a question about the relationship between food collection and food production after the transition to cereal-based agriculture in south central Africa. I wanted to understand how food collection practices practiced by agrarian societies supported social, political, and economic innovations. Why did farmers innovate the tools, techniques, and names of practitioners relating to certain kinds of food collection activities and not others? What did these kinds of food collection have in common that made them a coherent category of human endeavor? What were the social meanings of efforts in these activities and how did these meanings shift over time as the political and economic context of south central Africa changed?

Methods and Evidence To answer these questions, I drew on both the linguistic and archaeological records. With respect to language evidence, I focused on vocabulary relating to hunting, fishing, and gathering within ancestral Botatwe languages, a subgroup of the larger Bantu language family. Linguistic evidence indicates that Botatwe speech communities were mixed agriculturalists by the mid-first millennium. The historical development of the Botatwe family is illustrated in Table 2.1. Figures 2.1 and 2.2 illustrate the linguistic geography developed from the classification using the methods described in the introduction. With respect to archaeological evidence, I used publications about regional excavations and surveys for two purposes. First, I sought out direct associations between the linguistic and archaeological records (see Chap. 1 section “Connecting Archaeological and Linguistic Evidence” for more information on direct associations). The details of this effort are published elsewhere (de Luna 2012a), but

© The Author(s), under exclusive licence to Springer International Publishing AG, part of Springer Nature 2019 K. M. de Luna, J. B. Fleisher, Speaking with Substance, SpringerBriefs in Archaeology, https://doi.org/10.1007/978-3-319-91036-9_2

[email protected]

31

32

2  The Politics of Food Collection in South Central Africa

Table 2.1  Outline classification of Botatwe languages Botatwe (57–71% [100–900]; 64% median [500])  I. Greater eastern Botatwe (63–74% [500–1000]; 68.5% median [750])   (a) Central eastern Botatwe (70–77% [800–1100]; 73.5% median [950]    (i) Kafue (78–81% [1200–1300]; 79.5% median [1250])     1. Ila     2. Tonga     3. Sala     4. Lenje    (ii) Falls (91% [1700])     1. Toka     2. Leya    (iii) Lundwe   (b) Soli  II. Western Botatwe (76–81% [1100–1300]; 78.5% median [1200])   (a) Zambezi hook (83% [1400])    (i) Shanjo    (ii) Fwe   (b) Machili (84–58% [1400–1450]; 84.5% median [1425])    (i) Mbalangwe    (ii) Subiya    (iii) Totela Source: de Luna (2010) Extant languages are underlined. Names of proposed subgroups (corresponding to proposed protolanguages) are in bold font. Cognation rate ranges are in parentheses following the name of the branch whose internal relationships they describe. Medians of the plotted bell curves of cognation rates are also included. Approximate dates (developed through glottochronology) for the beginning of a protolanguage’s divergence follow the cognate rate information in brackets. Thus, the cognate rates among the Central Eastern Botatwe subgroup fall into the range of 70–77% with a median of 73.5%. This proposed protolanguage developed in the last half of the first millennium as Greater Eastern Botatwe diverged and dissolved at the close of the first millennium or perhaps into the early second millennium. Words reconstructed to Central Eastern Botatwe illuminate this period

underlie the proposed connections between particular protolanguages and regional archaeologies in the narrative that follows. Second, I used the archaeological record as an additional, independent record attesting to the historical development of subsistence technologies relating both food collection and production. The Narrative1  We pick up our story in the late first millennium, when speakers of Central Eastern Botatwe, the inheritors of a diverse, flexible, encompassing, and largely undifferentiated subsistence system, settled the middle Kafue region and began to experiment with using certain kinds of subsistence practices to mark social difference. As a result of this experimentation, some kinds of food collection activi The paragraphs in this section reproduce in whole or part selections from the following published works: de Luna (2012a, b), (2015). 1

[email protected]

Methods and Evidence

33

Fig. 2.1  Historical geography of Botatwe language divergences before 1250 CE. This illustration maps the first few divergences of the Botatwe family, plotting those divergences outlined in Table 2.1 in space. The map is read north to south, the direction of the spread of this language family as determined by the known over-arching geography of the Bantu Expansion. The darker the protolanguage name, the earlier in time it is proposed to have existed. This figure can be used with Fig. 2.2 to trace the application of the principles of least moves and greatest linguistic diversity in identifying the approximate location of protolanguages

ties—particularly hunting and fishing with spears—came to be understood as distinct from the agrarian economy and eventually formed the foundation for new kinds of political innovation in central Africa (de Luna 2016; cf. Finlayson and Warren 2010 for a European example). The Politics of Spearcraft: Making Social Difference through Subsistence, c. AD 750–1250  Proto-Central Eastern Botatwe speakers settled across the different microenvironments of the middle Kafue region in the last centuries of the first millennium to mitigate the threats that an increasingly dry and cool climate posed to cereal agriculture. These shifting climate conditions and settlement strategies encouraged great innovation in hunting and fishing. As Central Eastern Botatwe communities and their linguistic descendants, speakers of proto-Kafue, participated in a regional revolution in the technologies of spear hunting, rapid current fishing, and metallurgy, they invented a new category of landscape—the bush—and cultivated a distinction between food production and particular forms of food collection. This distinction was transformative because it created a novel path to celebrity, friendship, and ancestorhood based on knowledge of the bush, a politics of talent and technology that resisted the centralization of political and ritual authority around the agricultural economy. We know of this revolution in subsistence practices from the linguistic and archaeological records.

[email protected]

34

2  The Politics of Food Collection in South Central Africa

Fig. 2.2  Historical geography of Botatwe language divergences after 1250 CE. This illustration maps the last divergences of the Botatwe family, plotting those divergences outlined in Table 2.1 in space, ending with the early twentieth century locations of extant Botatwe languages. This figure can be used with Fig. 2.1 to trace the application of the principles of least moves and greatest linguistic diversity in identifying the approximate location of protolanguages. Thus, the proposed location of proto-Kafue lies at the approximate center of the greatest diversity of the locations of modern-day languages that developed from it (Lenje, Ila, Sala, Plateau Tonga / Valley Tonga) with the least moves necessary to account for those locations in a majority of the modern-day Kafue languages

Hunters speaking proto Central Eastern Botatwe and their proto Kafue-speaking descendants invented an elaborate lexicon to talk about spearcraft from the mid-­ eighth to the mid-thirteenth century, a lexicon that suggests these hunters radically transformed both their hunting technologies and the role of hunting in the local economy. Evidence for major changes in the strategies, technologies, and politics of spearcraft begins with the development of new tools. Proto Central Eastern Botatwe speakers and their linguistic descendants, speakers of proto Kafue, developed vocabulary for new types and parts of spears. For example, an ancient Bantu word for spear, *-túmò, was reborrowed at this time in the form *-sʊ́mò, probably because it now carried an iron point. Central Eastern Botatwe villagers followed their neighbors in applying an ancient Bantu term for “stone”, *-tádè, to “iron, iron ore, and iron bloom,” but they also converted the root into noun class 9/10 to produce *(ɩ)ntale, novel joints of iron taping or wire. Kafue marksmen a few centuries later adopted the *lusákò, a new spearshaft. Hunters and smiths collaborated to develop at least two new spear points between the mideighth and mid-thirteenth centuries: *(ɩ)mpùla, “spear point” and *-bèjɩ, a “barbless point used on both spears and arrows” developed from the inherited verb, *-bàij-, “to carve.” The *(ɩ)mpùla was a throwing spear with a short, barbless point

[email protected]

Methods and Evidence

35

and perhaps an iron cone hafted onto the spear butt. The iron cone probably facilitated digging, but it would also have served as a counter-weight to improve balance and trajectory when the spear was thrown. The large barbless point of the *(ɩ)mpùla was well suited to hunting larger game without damaging the hide, as a barbed point is likely to do. A spearman who used these increasingly specialized points needed to know what animals he expected to hunt before he went into the bush, which required detailed knowledge of environmental conditions and the seasonal and daily habits of game. This efflorescence in spearcraft was supported by the development of an intensive smelting and smithing culture known from the archaeological record and dating from the ninth through the eleventh centuries southeast of the Kafue floodplains, near modern-day Lusaka, and in the Zambezi Valley (Phillipson 1968; Vogel 1971a, b, 1975). Indeed, archaeological evidence from woodlands sites in the middle Kafue region, the homeland of Central Eastern Botatwe and, later, Kafue speech communities, also show evidence of a local iron smelting tradition, with the greatest period of intensity beginning in the late ninth century and continuing at least through the tenth century, overlapping with the transition between the Central Eastern Botatwe and Kafue periods (Derricourt 1985:66, 72). As part of the innovation in spearcraft, Central Eastern Botatwe speakers re-­ conceptualized the social meaning of skill in the kinds of hunting and fishing undertaken with spears. From an older word for a kind of long blade, the inhabitants of the southern savannas, from northeast Angola to the middle Kafue, developed a new noun, *-pàdʊ́, to name a “celebrated, skilled hunter/spearman.” Skill in hunting with spears was nothing new, but the category of person who might be celebrated for it was an innovation in the social landscape. The honorific ultimately derived from an ancient Bantu verb *-pá, “to give” with the extensive verbal affix (Bastin et al. 2002, ID 23442; Guthrie 1967–1971, C.S. 1404). The morphology of the term described the actions of such a talented, celebrated person: “giving” his weapon toward his quarry or the generous distribution of the fruits of his labors to his community of supporters. To describe what was new about the kind of work undertaken by *-pàdʊ́, Central Eastern Botatwe speakers invented a new verb, *-wèz-, to talk about hunting or fishing with spears, possibly throwing spears. This word was used alongside the inherited term *-gúɩm-, “to hunt with bows and arrows.” By the turn of the first millennium, then, Botatwe communities living near the abundant game of the Kafue floodplains distinguished between at least two forms of hunting: archery and spearcraft, with the latter including fishing with spears. In Ila, the best documented Botatwe language, kuweza also refers to the hunt for economic success and its use to launch and manage one’s social position, precisely the process that could produce a status like *-pàdʊ́.3 Indeed, the most likely source  Following convention, citations to individual lexical reconstructions in databases and compendia, such as Bastin et al. (2002) and Guthrie (1967–1971), will include the identification number of the reconstruction, rather than the page number. 3  As Fielder notes, “the Ila themselves describe a man’s economic struggle, his seeking for wealth and power, as his buwezhi, his hunting (kuweza—to hunt, pursue)” (1979:624). 2

[email protected]

36

2  The Politics of Food Collection in South Central Africa

root is *-gèd-, a widespread Bantu verb with a broad range of meanings. As a verb, the root commonly glosses as “to try,” “to think,” or “to measure,” but in noun form, it commonly means “wisdom” or even “guile” in central African Bantu languages (Bastin et al. 2002 ID 1345; Guthrie 1967–1971, C.S. 797). It is a root combined with suffixes to derive verbs with meanings such as “to liken” in some east Bantu languages and “to imitate” in some Bantu languages in central Africa. When Central Eastern Botatwe speakers developed the verb *-wèz- in the last quarter of the first millennium to talk about a new form of hunting and fishing with spears, they sought to describe the social claims implicit in command of this particular kind of technological knowledge. Again, the rich documentation of the Ila language helps us to imagine the social stakes of knowledge of spearcraft. In the causative tense in Ila, kuwezya not only means “to cause to hunt,” it also means “to cause to help,” calling to mind the social networks of dependency in which hunters both sought to be entangled and undoubtedly sometimes found themselves ensnared. Muwezele, derived from the applicative of kuweza, “to hunt on behalf of,” glosses elegantly as “a popular person.” In a reduplicated form, Ila speakers use the verb kuweza-weza to talk about making light of difficulties, reminding us that hunting has a long history as a path to more prosperous social ties and welcomed material wealth. To be sure, reduplication of the causative, kuwezya-wezya, glosses with the expected “to hunt a little,” but it also means “to deride” or “make light of,” demonstrating the contingencies of hunters’ aspirations (Fowler 2000:493, 761; see also de Luna 2013; Tuden 1962:33). Not all hunters’ efforts were successful and even strenuous effort might not protect against ridicule in the face of failure. Nor did success ensure others’ acceptance of the social status to which a hunter aspired. Ila speakers took an inherited word for a form of hunting with spears and developed out of that word a range of meanings conjuring up the lifelong pursuit of social status, the obligation of assistance, and the belittlement awaiting unsuccessful attempts at providing help and performing the status to which one aspired. Central Eastern Botatwe homesteaders developed the verb *-wèz- in the last quarter of the first millennium to talk about a new form of hunting with spears that was bound up with the social ambition of hunters, their kin, and their households as well as their social obligations. Linguistic evidence for shifts in hunting practices in the faunal records of sites in the middle Kafue region and on the Batoka Plateau can be compared to archaeological evidence for transformations in hunting. The archaeological record of the middle Kafue and Batoka Plateau regions dating from the mid-eighth to mid-thirteenth centuries boasts rich innovation in hunting technologies. Ironworking flourished, with new forms of points, ferrules, and end-spikes. The regional faunal record shifts toward higher mortality rates of gregarious antelope species, hippos, and elephants, all species hunted by groups armed with spears. Moreover, fully-gown adult animals dominate mortality curves. Archaeologists interpret these faunal assemblages as indicative of the emergence of skillful communal hunting practices (de Luna 2012a; Derricourt 1985; Fagan 1967; Fagan et  al. 1969; see also references in Table 2.2).

[email protected]

Methods and Evidence

37

Table 2.2  Ratio of wild to domestic faunal remains at major regional sites in South Central Africa Site Kalala (Kafue)

Pre-­700 CE Phases 1–2 300 CE W: 82.2% (46) D; 17.8% (10) Total = 56

700–1300 CE Phases 3-9A 700–1100 CE W: 95.1% (2460) D: 4.9% (127) Total = 2587

Musa (Kafue)

Sebanzi (northern Batoka)

Kangila (northern Batoka)

1300–1700 CE Recent Phases 9B-9D 1400–1500 CE W: 80.4% (444) D: 19.6% (108) Total = 552 1600– 1700 CE: W: 80.2% (425) D: 19.8% (105) Total = 530

All levels 800–1280 CE W: 24.5% (60) D: 75.5% (185) Total = 245 Level 3 post-cranial W: 28% (16) D: 72% (41) Total = 57 Level 2 post-cranial W: 35.1% (25) D: 64.9% (46) Total = 71 Level 1 post-cranial W: 39% (27) D: 61% (42) Total = 69 Level 3 teeth W: 34.5% D: 65.5% Level 2 teeth W: 0 D: 100% Level 1 teeth W: 4% D: 96%

Notes and source Analysis of this sample combines buffalo and ox so some data designated as domesticated may be wild. Derricourt (1985) Analysis of this sample combines buffalo and ox so data designated as domesticated may be wild. Derricourt (1985) Fagan and Phillipson (1965)

Fagan (1967)

(continued)

[email protected]

2  The Politics of Food Collection in South Central Africa

38 Table 2.2 (continued) Site Matobo (southern Batoka / valley escarpment)

Ingombe Ilede (southern Batoka)

Pre-­700 CE Pre-­700 CE: W: 51% D: 49%

700–1300 CE 1300–1700 CE Recent 700 CE–1100 CE: W: 29% D: 71%

800–1300 CE Level 7 W: 96% (23) D: 4% (1) Total = 24 Level 6 W: 70% (21) D: 30% (9) Total = 30 Level 5 W: 75.5% (34) D: 24.5% (11) Total = 45 Level 4 W: 75.4% (92) D: 24.6% (30) Total = 122 Level 3 W: 77.4% (113) D: 22.6% (33) Total = 146 Level 2 W: 75.9 (91) D: 24.1% (29) Total = 120

Level 1 W: 49.5% (60) D: 50.8% (62) Total = 122

Notes and source Data for site as a whole: Site W: 38.7% (70) Site D: 61.3% (111) Site total = 181 Watt (1980) Fagan et al. (1969)

(continued)

[email protected]

Methods and Evidence

39

Table 2.2 (continued) Site Isamu Pati A (southern Batoka)

Pre-­700 CE

Kalundu (southern Batoka)

Level 8 W: 51.4% (18) D: 48.6% (17) Total = 35 Level 7 W: 85.8% (18) D: 14.2% (3) Total = 21 Level 6 W: 66.7% (10) D: 33.3% (5) Total = 15

700–1300 CE Level 8 700 CE W: 89.5% (17) D: 10.5% (2) Total = 19 Level 7 W: 62% (26) D: 38% (16) Total = 42 Level 6 W: 57.5% (19) D: 42.5% (14) Total = 33 Level 5 W: 56.2% (36) D: 43.8% (2) Total = 64 Level 4 W: 39.5% (19) D: 60.5% (29) Total = 48 Level 3 W: 47.2% (26) D: 52.8% (29) Total = 55 Level 2 1200 CE W: 13% (3) D: 87% (40) Total = 46 Level 5 W: 81% (21) D: 19% (5) Total = 26 Level 4 W: 42.5% (17) D: 57.5% (23) Total = 40 Level 3 W: 34.4% (11) D: 65.6% (21) Total = 32 Level 2 W: 35.1% (13) D: 64.9% (24) Total = 37

1300–1700 CE Recent Level 1 W: 18.2% (4) D: 81.8% (18) Total = 22

Level 1 W: 54% (13) D: 46% (11) Total = 24

Notes and source Fagan (1967)

Fagan (1967)

Reprinted with permission from de Luna (2012a) Where available, stratigraphic level and date indicated. All dates calibrated via OxCal to two-­ sigma confidence level

[email protected]

40

2  The Politics of Food Collection in South Central Africa

Despite these similarities, there are also localized differences in the faunal records of middle Kafue and Batoka sites (Table 2.2). The long sequence of innovation in spearcraft and communal hunting reconstructed to the Proto-Central Eastern Botatwe and Proto-Kafue occupation of the middle Kafue closely corresponds to the records of faunal and iron remains recovered from the Kafue Middle Iron Age (Derricourt 1985). The Middle Iron Age can be divided into early (eighth century), transitional (ninth century), and late (tenth century onward) phases based on stratigraphy and the cluster analysis of pottery. Within the Kafue, wild species have always dominated faunal collections, but the more detailed data published for Kalala Island shows a consistently higher number of gregarious species and those typically hunted with spears (zebra, hippopotamus, antelope, etc.) from the transitional MIA (ninth century) onwards (Derricourt 1985:211). These dates closely align with the proposed emergence of Proto-Central Eastern Botatwe around the middle of the eighth century and coincide with the majority of innovations in the vocabulary of spearcraft discussed above. On the Batoka Plateau, however, sites like Isamu Pati, Ingombe Ilede, and Matobo demonstrate a slowly decreasing ratio of wild to domesticated species until the thirteenth and fourteenth centuries. After this date, there is a sharp decline in wild species in the Isamu Pati assemblage. At Kalundu, the same trend of a slow reduction in wild species is punctuated by sharp upward spikes between excavation levels eight and seven, six and five, and two and one. In contrast, at Kangila, the only site for which we have data demonstrating change over time along the northern edge of the plateau, post-cranial remains demonstrate a modest increase in the percentage of wild species over time, while an analysis of the few teeth recovered suggest the opposite trend. These localized differences reflect a continuum of emphasis on wild game across time and space. Significantly, the region and period of highest investment in the archaeological record—the Middle Iron Age of the Kafue—corresponds to the spatial and temporal framework of the proposed Proto-Central Eastern Botatwe and Proto-Kafue speech communities who innovated the hunting vocabulary discussed above. Butchery patterns within late first millennium and early second millennium assemblages from the middle Kafue and Batoka Plateau regions exhibit a shift from an earlier pattern in which entire carcasses are represented in the assemblage to specific portions of the carcass being brought back to the village, skewing the portions represented in the faunal record. Beginning in the Central Eastern Botatwe period and continuing through the Kafue period, hunters changed how they processed carcasses at the site of the kill. Three possible scenarios, or a combination thereof, can account for changes in the faunal record, but all three indicate that hunting and meat had taken on a new status. Hunters may have been hunting further from the homestead, eating some of the meat in hunting camps, drying other portions, returning with only preferred bones. However, this long-distance hunting requires great investment of men’s time for very little return in calories. Moreover, all the species represented in the faunal record were likely abundant in the nearby Kafue Flats, particularly in the dry season. Alternatively, hunters may have selectively, wastefully butchered game when it was killed, particularly if innovations in

[email protected]

Methods and Evidence

41

hunting technologies increased the number and frequency of kills. Indeed, if animals were killed in ever-greater numbers but only selectively butchered, hunting was no longer a vital source of protein for ordinary people. It had a new value outside of nutrition. Or, perhaps hunters butchered their kills in a way that followed local conventions of carcass division after communal hunts (discussed below), though we have no indication of where other carcass portions were deposited. In each scenario, the development of novel butchery and consumption practices demonstrate that hunting and meat had a value beyond nutrition for Central Eastern Botatwe and Kafue hunters and their households, kin, and neighbors. Likewise, the development of words to broadly delineate hunters into categories of talented and ordinary and to link spearcraft to social ambitions suggests that hunting was no longer a general pursuit strategically employed by farmers. Rather, some hunting and fishing, particularly with spears, was part of the work of distinguishing oneself and developing dependencies, of acting on the full suite of economic and social aspirations Ila speakers came to associate with the verb *-wèz-. That talent in spearcraft had social stakes was marked in one final innovation dating to the centuries around the turn of the millennium: the invention of a new landscape, “the bush”, *-sókwe, in which skilled hunters now worked. Significantly, spearmen were not the only figures to work in the shadow of the bush. The archaeological record suggests the invention of this new landscape category coincided with a regional shift in the geography of smelting as smelts were relocated from the village to uninhabited areas beyond the village (Bisson 1976; Fagan 1967:88–91; Fagan et  al. 1969:43; Inskeep 1962:152–53, 155–56, 1978:113–17; Killick 1990:76–77; Robertson 2001:262; Vogel 1971a:39, b:39–40, 1975:66). Not surprisingly, this shift also coincided with the invention of a new category of local notable, a rich, wealthy person, *-vʊbi, named from an older word for bellows, a tool central to the work of metallurgy (de Luna 2015). In modern Botatwe languages, isokwe is the quintessential “bush”; it is an open space with the scrub vegetation that indicates that it is neither settled, nor cultivated (Fowler 2000:129, 242; Mukanzubo Kalinda Institute 2011:861, 970). For those Botatwe communities about which we have a rich ethnographic literature, the bush is a key concept for understanding the social, economic, and ritual dimensions of the landscape because it is firmly associated with metaphysical forces implicated in acts of transformation, such as hunting, initiation, smelting, and, in other central African societies, investiture. Like other Bantu-speaking societies, the bush for Botatwe speakers is a place imbued with a transformative power of great social potential and possible danger homologous with the power of human fertility (Herbert 1993). The invention of *-sókwe elaborated on an older, more widespread verb, *-còk-, “to incite,” which, itself, derives from an ancient Bantu term that glosses as “to poke in, put in, prick with a point, hide, ram in” (Bastin et al. 2002 ID 644–5; Guthrie 1967–1971, C.S. 371). Contemporary attestations in the best documented Botatwe languages reveal a complicated network of meanings tying together ideas about “provoking,” “inciting,” and “stabbing” with “being first,” “establishing,” or “originating.” This semantic range suggests that speakers imagined acts of creation in the

[email protected]

42

2  The Politics of Food Collection in South Central Africa

bush to require provocation by poking or spearing. Modern attestations make explicit the link to hunting. As a transitive verb, kusoka means “to examine trap(s)” and “to start a quarrel” while kusokoma means “to poke, as with a spear in a burrow, or in an animal’s chest to kill it” (Fowler 2000:652–53). When Central Eastern Botatwe, Kafue, and neighboring communities named the open bush around them with the passive tense of the verb, *-còk-, they imagined this landscape to be a place of potential creation, a place that was literally “the poked, the prodded, the hidden, the entered” place. It was a setting for and ingredient in acts of origination incited by the plunging action of spears thrust into quarry stalked, netted, or trapped in the bush, the pricking of furnaces with bellows’ points or the rhythmic push of air goading the flames of the smelt. The bush, *-sókwe, was a place of great potential power that could be incited and activated by those spearmen and smelters capable of prodding such generative forces towards acts of creation and social significance. Hunting as a Tool of Social Integration, c. AD 950–1250  The status earned by *-pàdʊ́, celebrated hunters, who could master the aspirational strategies of *-wèz-, hunting and fishing with spears, and the excitable life forces of *-sókwe did not go unchallenged. But the politics of reputation in bushcraft was not rejected by later generations, either. Rather, the dawn of the second millennium saw Kafue-speaking men and women innovate on the inherited body of ideas associated with spearcraft to meet the new challenge of settling in lands on the Batoka Plateau already inhabited by speakers of a different set of languages and crafters of different material cultures. In this way, spearcraft was both an object changed by new social and political aspirations in the early second millennium and a tool to effect social and political change. To understand the significance of the value of hunting as a large-scale practice of social integration, it is helpful to understand the smaller-scale practices innovated at the same time (de Luna 2016). Importantly, this was a period of major transformations in kinship, as illuminated by a new word, *-kóba, borrowed across the region to name matri-groups and indicating the initial shift from bilateral to unilateral kinship reckoning. At the same time, Kafue speakers innovated terms for specific kinds of kin, including the matrilineal uncle, *-tʃɩCɩa, a figure with great responsibility for and over his sisters’ children. This period also saw the use of two names for the cohorts of cousins involved in emerging virilocal cross-cousin marriage practices. The first, *-kwésʊ, named parallel cousins, or matrikin of one’ own (classificatory) generation, with a gloss as “those from our place,” indicating virilocal marriage patterns. The second, *-kwàshɩ, named cross-cousins or in-laws from one’s own ­generation, and literally glossed as “the seizers,” indicating the potentially violent nature of Kafue expansion out from the middle Kafue area, north toward the Lukanga swamps and south across the Batoka Plateau and into the Zambezi Valley. As fraught as the adoption of these new social ties may have been, Kafue speakers’ vision of what a family looked like and how its members should behave came to dominate the social imagination of all residents of the region in the early second millennium.

[email protected]

Methods and Evidence

43

It was in this changing demographic context that members of the short-lived Proto-Kafue speech community used their ancestors’ great innovation in spearcraft to recast the work of communal hunting in the first few centuries of the second millennium. Like earlier innovations in spearcraft and the celebration of hunters, communal hunting was nothing new, but the development of new words to talk about it suggests the innovation of novel techniques and values associated with communal hunting. The reconstructed vocabulary referring to the tools and organization of group hunting is rich, like the cluster of innovations relating to spearcraft. Three words help us to understand what Proto-Kafue communal spear hunting might have looked like. The first, *cìlà, derives from a word for a very large hunting net, *-kìdà, borrowed by Proto-Kafue speakers from speakers of western Bantu languages to the north and west (Vansina 1986, 1990). When Proto-Kafue speakers borrowed this word, they adopted it into noun class seven without adding a noun class prefix and used it to talk about a communal hunt. Communal hunts were probably large enterprises; in the early twentieth century, they drew scores and even hundreds of participants and attracted enough attention from colonial conservation officers to be banned by the middle of the twentieth century (Ansell 1955; Boyle 1958; Colson 1958:32, 1959:106, 2006:104–05; Haller 2007; Livingstone 1865:311; Rennie 1982; Reynolds 1968:57–58; Scudder 1962:198–99; Siamwiza 2007:251; Smith and Dale 1920, vol. 1:153–55). While beaters chased animals toward an entrapment (a bog, a line of hunters holding a large net, a long pitfall, a fire), other hunters waited, prepared to spear the animals at the entrapment, *-pàndo, another Proto-Kafue innovation that has come to mean “dry season communal hunt” in both Tonga and Lenje. In addition to using *-pàndo, Proto-Kafue speaking battue hunting parties also encircled game. Commencing in two facing lines at some distance apart, one line of hunters would advance, beating the animals forward as both lines curved tighter and tighter, encircling the animals. As the hunters closed in, they stabbed their prey. While all modern-day Kafue languages refer to this form of hunting as “kuoba banyama”, it is impossible reconstruct the entire phrase. It is only with the noun *ibalo, “circle of hunters in a battue hunt” deriving from *-bada “ring,” that we have a common root that explicitly refers to battue hunting by encirclement, typical of precolonial elephant hunting. Chila offered opportunities to engage with those communities speaking other Bantu languages, as the very borrowing of the term indicates. Kafue speakers might well have sought to include communities among whom some Kafue families had begun to build homes and marry. The timing of this new emphasis on communal hunting in the context of establishing new settlements and new kinship and marriage practices across linguistic and material cultural boundaries suggests that one of the ways immigrant Kafue speakers met the social and ritual imperatives of settling in the occupied lands of the Batoka plateau was through massive communal hunts, which offered opportunities to cooperate and generated in very large quantities prized resources like meat and skins that could be shared. Proto-Kafue speakers’ new emphasis on communal hunting unfolded in the context of continued innovation in the technologies and techniques of spearcraft described above. The invention of the *-bèjɩ carving spear point in this period suggests that butchery practices and

[email protected]

44

2  The Politics of Food Collection in South Central Africa

meat distribution may have become a more formal and perhaps more ritualized process. Faunal remains from the Kafue and Batoka region in the late first and early second millennium demonstrate clear preferences for particular carcass parts (Derricourt 1985:208–15; Fagan 1967:70–82; Fagan et al. 1969:254–55). The traditions of butchery accompanying these hunts may have followed particular alliances forged across communities such that participants from one region, clan, or speech community traditionally took the foreleg, while those from another group took the cranium, and so forth. The politics of reputations tied to knowledge about the bush developed by Central Eastern Botatwe speakers and conserved by Kafue speakers faced a challenge with the move onto the Batoka Plateau: while reputations could travel, they now traveled through a landscape populated by other local notables, infamous hunters, and wealthy smelters. Those who drew on the knowledge and power of their own ancestors to ensure the success of their activities in the bush began to work in a landscape studded by the graves of others’ ancestors and controlled by their willful spirits (cf. Klieman 2003). For the earliest Kafue families to live among strangers, it was not at all certain how ancestors and heroes engaged with non-kin and newcomers. Chila was a political tool used to establish in a very public setting those shared hunting heroes who would become the ancestor spirits protecting future generations in return for the commemoration of their own skills, feats, and fame. To be sure, shared heroes may well have smoothed over the inconvenient problem of lack of access to firstcomers’ ancestral spirits for the earliest Kafue settlers participating in chila and settling among and even marrying into foreign families on the Batoka Plateau. Chila allowed a wider public to see and participate in the violent and competitive aspects of bushcraft. For some, chila opened a social space for collaboration across social and linguistic boundaries. Yet others who attended a chila hosted by a Kafue-­ speaking community or observed recently settled neighbors mounting a chila hunt near their villages might well have watched with some trepidation as large groups of spearmen, beaters, carriers, and children put on a show that, in essence, exhibited masterful coordination in the corralling and killing of swiftly fleeing animals. These were technical skills transferrable to warfare, a message that may have been particularly clear to communities who felt that middle Kafue ways of life were being forced upon them by Kafue-speaking immigrants and those who welcomed them to settle on the Batoka Plateau. Ultimately, chila opened the path to the status of *-pàdʊ́ to more young men, changing the scale of the politics governing the tangible and intangible products of hunters’ pursuits; it also democratized access to the socially and politically potent products of hunters’ pursuits. Kafue hunters in the early second millennium—whether low-skill beaters joining the festivities of a neighborhood chila or expert *-pàdʊ́—increasingly used their knowledge of spearcraft to target species of antelope with strict preferences for wetland margins. These species are among the gregarious quarry so well represented in the regional faunal record. Kafue speakers innovated new names for such antelopes, a creativity that speaks to the importance of antelopes to Kafue speakers.

[email protected]

Methods and Evidence

45

For example, Kafue speakers integrated into their vocabulary and environmental knowledge a name for sitatunga, *-jóbé, reconstructed to proto-Kafue as *- zóbé. They also adopted a regional term for Kobus lechwe, *n(y)anjá, that was circulating across the wooded savannas from the Okavango River through the Zambezi hook and into the Sabi-speaking areas east of Botatwe communities. Similarly, the broad savanna form *mbàbàlá was likely borrowed by Kafue communities from Kusi speakers and used to talk about bushbuck, a species inhabiting areas of well-watered thick brush and forest near permanent water, like the Kafue River and its tributaries. The phonology of *n(y)alʊvwɩ, a new term for reedbuck, secures its status as a Kafue innovation. This term is a compound word combining an old feminine pre-­ prefix (with the regional form n[y]a-) meaning “mother of” or “female” with the noun class 10 prefix and an old, inherited proto-Savanna term for arrow, *-gúí, a root that was later used in some languages of the Kafue subgroup to name communal hunts and game drives with fire. Antelope like those Kafue speakers named in the early second millennium were the quarry of trappers, independent hunters and chila. The addition of a feminine pre-prefix to antelope names seems to have been something of a linguistic fashion in Kafue speech about certain kinds of antelope. In addition to *n(y)alʊvwɩ, proto-Kafue speakers added the feminine prefix to the borrowed term for kobus lechwe, *n(y)anjá, and an old inherited word for duiker, *-sɩ́á, now *n(y)asɩ́á. Though many neighboring communities also used these latter two roots, Kafue people were the only speakers to add gendered prefixes to them. Although it may have originated as part of an argot defining *-pàdʊ́ who were in the know, this lingo was certainly common parlance only a generation or two later. Where the ethnographic record is richest, we learn of the practice of hunters giving wives and lovers the skins of lechwe and reedbuck to fulfill the obligation of men to provide women and even younger girls to whom they are betrothed with clothing, especially skins for aprons (Colson 1960:20, 110–11; Smith and Dale 1920, vol. 1:96, 155). The feminine prefixes attached to particular species might indicate both the preferred species for making women’s aprons and the antiquity of this gendered form of gift giving. It may well be that the produce of hunters’ efforts afforded them a way to meet the mundane material needs of their female dependents and affirm social ties of responsibility and obligation with an exemplary and perhaps even coveted kind of apron leather. The gifting of women’s aprons to wives, would-be wives, and lovers specified an existing or future sexual relationship in the ethnographic record cited above, which adds another layer of meaning to such gifted skins. When stalked and hunted, rather than snared along field margins or game paths, these species were acquired in the uninhabited bush, *-sókwe. When Kafue hunters began to mark with feminine gender the names of antelopes whose hides could be worked into supple leather and cut into aprons donned by favored wives and skirting the hips of desired lovers, they may have summoned up in the minds of their listeners ideas about the virility and productivity associated with the bush and with its technicians (de Luna 2015). Chila communal hunts may have been a way to more intensively exploit and also open up access to the kinds of virility associated with the bush and evoked by its products.

[email protected]

46

2  The Politics of Food Collection in South Central Africa

Summing Up If the earliest Botatwe communities created a subsistence system that integrated a range of strategies, their linguistic descendants worked hard to distinguish work undertaken in the bush from the labor of the fields and steams near villages. The distinctive value of spearcraft was not a foregone conclusion to the practitioners who invented a novel path to singularity, fame, friendship, and ancestorhood based on their knowledge of the bush. Unfolding in the lands that bridge central and southern Africa, these innovations in the politics of technology and talent resisted the centralization of power around the agricultural economy at a time when neighboring societies were just beginning to experiment with consolidating authority around royals and chiefs, experiments that were foundational to the establishment of the famed precolonial kingdoms and chiefdoms of central and southern Africa. Subsequent generations of Botatwe farmers further developed the social and political significance of bushcraft, revealing the dynamism of subsistence strategies usually assumed to be static and peripheral. Novel forms of communal hunting were at the heart of a demographic revolution on the Batoka Plateau in the early second millennium, facilitating the incorporation of non-Kafue speakers into Kafue families. Kafue peoples extending their settlements into lands that were already occupied by other communities saw in their inherited practices of bushcraft the opportunity to build communities. But this required recasting the social scale of hunting, changing it from a pathway for individual success into a viable strategy of social integration and intimidation they named chila. Innovations in the meaning of antelope hides and the scales and the sociality of hunting with spears occurred at the same time that Kafue speakers were inventing new ways to build ties of marriage, kinship, and friendship. Because the strategies, tools, meanings and communities associated with hunting changed over the centuries in south central Africa, its capacity to support novel political endeavors was far more durable.

[email protected]

Chapter 3

Comments, Dialogue and Supplemental Reading: South Central Africa

Commentary One of the remarkable feats of this paper is your effort to get inside the archaeological data and use the patterns found in that data to compare, contrast, and explore the questions and issues that you pose. Your effort is novel in the way that you have attempted to corral and synthesize the archaeological data, and to use your assessment of it to build an historical narrative that weaves together archaeological and historical linguistic data. You are explicit about how archaeological data is to be deployed—“direct associations between the linguistic and archaeological records” through particular classes of material culture and words—as you attempt to construct a regional overview of “the archaeological record as an additional, independent record attesting to the historical development of subsistence technologies relating both food collection and production.” In these two efforts, one that relies on particular material object types and the other that seeks patterns across time and space, you demonstrate a mastery over the archaeological material. My comments on this essay are aimed at exploring how you use the archaeological data and I focus mainly on a critical assessment of different types of archaeological data and how they are employed and presented, and where the limitations of those data exist. In this way, these comments capture how an archaeologist might read the efforts of a non-archaeologist using archaeological data. I also indicate places that the archaeology might offer insights that you have not explored. These latter efforts, however, are attenuated by the very significant limitations of the extant archaeological datasets, and this is the first topic that I cover—the problems associated with the way archaeology in your region of study has been recorded and reported. I then explore a series of archaeological issues that must be considered as you interpret archaeological data from a report. These include issues of sample size, stratigraphy, and dating. Finally, I sketch out an argument about the need for thinking about contextual data in archaeological assemblages and how a historian might © The Author(s), under exclusive licence to Springer International Publishing AG, part of Springer Nature 2019 K. M. de Luna, J. B. Fleisher, Speaking with Substance, SpringerBriefs in Archaeology, https://doi.org/10.1007/978-3-319-91036-9_3

[email protected]

47

48

3  Comments, Dialogue and Supplemental Reading: South Central Africa

deploy archaeological data. Your essay works mainly at two scales—at the site and regional level—and thinking contextually (if possible with the data on hand) might be a useful way to engage more deeply with the questions you seek to address. Challenges of Archaeological Reporting  In putting archaeological data in conversation with historical linguistic data, you have clearly struggled with a ­complicated and relatively sparse archaeological literature on south central Africa. This is no easy feat for a non-archaeologist. First, the data you use was written by archaeologists generally for archaeologists—these are detailed reports on surveys and excavations (e.g., Derricourt 1985; Fagan and Phillipson 1965), from a time when archaeology was just beginning in earnest in Zambia. Granted, certain publications were aimed a broader audience (e.g., Fagan 1967) and archaeologists who were engaged in research on the Bantu Expansion saw themselves as part of a multi-­ disciplinary effort to understand this phenomenon. However, in re-reading this literature to write this commentary, I was struck that many of the publications were focused on reportage and the dissemination of the findings from archaeological research. These are useful in that they present raw data in many forms, including descriptive and tabular form. However, moving from publication to publication, there was little consistency in how data were presented and which variables were highlighted. For example, the faunal data discussed by Fagan (1967) for Isamu Pate and Kalundu is presented in variety of ways for each site: presence/absence, age distributions, meat weight distributions, but there is no comprehensive accounting of the material by species, by level or phase. In contrast, Derricourt (1985) presents a detailed accounting of faunal NISP (Number of Identified Specimens) by period but not by excavation unit. Without a full accounting, or the ability to map the materials across space, it would be difficult to use these data for new analyses; even for an archaeologist, it would take a great amount of work to bring such disparately reported data together in a comprehensive way. The other challenge with these publications is that, because they represent early reporting on archaeology in a region, they offer interpretations of their datasets that are extremely provisional. At times, therefore, such publications overstate the types of interpretative claims that can be drawn from their data, even if presented in a preliminary way (a point I take up below in the section on faunal remains). The data come from very limited surveys and excavations at only a handful of archaeological sites and thus represent limited coverage of a complex and variable region. Finally, because of the age of many of the reports, they do not have the benefit of insights from decades of study on particular types of artifacts, including ceramics, faunal assemblages, metals, and other materials. When an archaeologist reads such literature, they do so with an understanding that the data presented in older publications requires some critical assessment and cannot be taken at face value. Dealing with Excavated Assemblages: Problems and Concerns  In this section, I offer a critique of how you have deployed some of the archaeological data from the Batoka Plateau. My aim is to provide some guidance on how to better evaluate and

[email protected]

Commentary

49

use previously excavated material in a region with such limited knowledge of the archaeological record and inconsistency in its reporting. In Chap. 1, I discussed some of the ways that archaeologists must evaluate the quality of their data through sampling and formation processes. Here, I apply those techniques to faunal assemblages because they loom large in your chapter, but I also briefly discuss ceramics. The way you make use of faunal assemblages from archaeological sites helps to illustrate what needs to be considered when drawing together data from a variety of sites that are individually reported. There are some qualities of the data that one should examine before re-deploying them and these include: how the data are reported; the sample size of the data from each site; the preservation of materials; and finally the context in which the materials were found. As I have noted, there are multiple ways that data can be reported and it is (unfortunately) rare for archaeologists to offer a full account of the data. For faunal data, the most common form of reporting is the Number of Identified Specimens (NISP). These counts can include different bones or fragments of bones from a single animal or multiple animals. This value may be used to compare the dominance of particular species but, due to the fragmentary nature of most archaeological faunal assemblages, it will also likely include multiple bones from a single individual. There are other ways to examine a faunal assemblage, such as Minimum Number of Individuals (MNI), a statistic that is based on specific, sided bones such that two left femurs of a goat in an assemblage would equal an MNI of ‘2’, but a left and a right femur, would produce an MNI of ‘1’ (Lyman 1994). Finally, zooarchaeologists calculate weight as well, including the weight of the material, as a way to measure its abundance in the assemblage, as well as more complex calculations that estimate the biomass of animals represented in the assemblage—the dietary contributions of different animals to a community (Casteel 1978; Reitz et al. 1987; Reitz and Wing 2008). When using simple weight measures, one must be remember that larger animals weigh more than smaller ones, and therefore such interpretations can be biased toward larger animals. There is of course no one approach that offers the ‘best’ evidence, but it is useful to know how and why faunal analysts choose one over the other. It is important to know that measures such as MNI will exclude many of the bones in the assemblage as it reduces the assemblage to bones that accurately assess a minimum value of individual species; as such, this is a very different count than NISP which typically reflects the full analyzed sample. Finally, there are always bones in an assemblage that cannot be identified, or cannot be identified to species—the latter are often assigned more general categories based on size class (Mammal I, II, etc.). Faunal analysts should report the numbers of these ‘unidentifiable’ fragments as this will also help assess the overall fragmentation of the assemblage. I discussed issues of sample size, preservation, and context in the introduction, and these apply especially when considering faunal assemblages. Questions that need to be asked include: Is the sample representative of the patterns that the investigator hopes to discuss? Are there any problems with the sample—are there fragmentation or preservation issues that have degraded the assemblage overall or

[email protected]

50

3  Comments, Dialogue and Supplemental Reading: South Central Africa

perhaps affected particular parts of the assemblage? For example, if there were a reliance on small animals whose bones would have been more subject to decomposition in the soils or affected by recovery bias (e.g., screen size), are they underrepresented in the assemblage? How might processing of bone in the past affect identification to zoological taxa in the present; were some bones subjected to greater fragmentation (e.g, breaking for marrow processing), making them harder to identify by reference to their morphological features? Were the materials found in a general site midden (such as a communal trash pit), a particular type of feature (a pit or hearth?), or are the materials from more general deposition layers broadly associated with occupation? Is there any evidence of differential preservation between contexts or feature types? There might not be a way to answer these questions with any confidence, but it is important to indicate whether the excavator self-reported any challenges with the assemblage, and/or that such issues have been considered when using previously excavated data. Many of these issues can be exacerbated when attempting to compare across faunal assemblages, especially when an analyst is attempting to explore differences in faunal assemblages as meaningful. Describing changes through time in a faunal assemblage from a single, stratified site is less problematic in that it can be argued that preservation issues would remain consistent within a site, assuming that consistent excavation strategies were applied throughout. However, if different teams used different recovery strategies (such wet sieving, finer screens, or flotation), there may be significant differences between the assemblages. Once one begins comparing across sites, one must attend to the issues of preservation, sample size, excavation and recovery strategies, and reporting, to be sure differences between assemblages are due to past human practices and not to these other factors. Two examples from your chapter illustrate where a consideration of these factors matters. Two important publications that include the results from excavations are Derricourt (1985) on research carried out at sites near the Iteshteshi Dam (especially the Kalala Island site) and Fagan’s (1967) early work further south, at Isamu Pati and Kalundu. The Kalala Island faunal assemblage is relatively large and diverse (n = 11,809), but only approximately 20% (n = 2322) of these bones were identified to species or general mammal categories, with the rest unidentified. This is not problematic and most zooarchaeological assemblages are dominated by unidentified specimens. The identified specimens includes a wide range of species, from small rodents and fish to larger mammals, and Derricourt reports them fully, by phase, in Table A7.5 and Appendix 7. This suggests that there are not likely overwhelming preservation issues with this assemblage, and this is further supported by his consideration of the effect of the soils on the faunal remains, as well as other factors that may have biased the sample (collection procedures, etc.). Additionally, Derricourt’s excavations included units situated over a large area, including deposits in and around houses, and thus, taken as a whole, the assemblage is likely broadly representative of the fauna consumed at the site. However, the other assemblages, especially when broken down into individual levels, are less robust, and therefore care must be taken as to how much interpretation can be drawn from them. For example, the assemblage from Isamu Pati seems

[email protected]

Commentary

51

to be quite large based on the scale of excavations (see Fagan 1967:Fig. 16). But, there is no complete accounting of the assemblage; one table that compares wild and domestic faunal remains lists only 387 fragments identified (Table II in the Appendix), a small sample size given the scale of the excavations. There is also little discussion of how preservation issues might have affected the deposits at Isamu Pati and there is no discussion of how bones were collected and their state of preservation at collection. Given how little we know about the assemblage, I remain skeptical about the insights drawn from the data sets as reported. Two examples illustrate how and why an archaeologist might be skeptical of the interpretations you offer. First is the data presented for the ratio of wild versus domestic fauna. You draw from published faunal data to demonstrate a “slowly decreasing ratio of wild to domesticated species until the thirteenth and fourteenth centuries.” Additionally, detailed evidence from Kalundu is cited, showing “the same trend of a slow reduction in wild species…punctuated by sharp upward spikes between” particular levels. I focus on the Kalundu data since it is used as a specific case to demonstrate “a continuum of emphasis on wild game across time and space…” It is problematic to discuss “upward spikes” in wild to domestic species between levels with this data because the sample sizes are so small; the assemblage in most levels includes 20 bones or less. Although there is no hard and fast rule as to what constitutes a minimum sample size for faunal assemblages, one must use caution in extrapolating from small samples, or at least recognize their limitations. It is also important to think about what the excavator’s ‘levels’ are indicating— are these simply arbitrary levels that the excavator imposed, or are they meaningful stratigraphic or temporal units? Levels also have varying volumes based on their extent or thickness, and thus many archaeologists present their faunal data as densities—the NISP per volume (NISP/m3)—which allows for a normalized way of comparing across contexts. Therefore, while it is possible to recognize that the faunal data includes both wild and domestic animals,1 the argument for patterning through time is made problematic by both sample size issues and lack of clarity on what the levels mean. Although the tabular presentation of so many contexts and the ratios of wild to domestic seem to create an argument for such a pattern, one must look at and evaluate the quality of the data for any one site. If any of these sites suffer from sample size issues, it might be better to exclude them rather than weaken the case that you are trying to make. Another example where a lack of understanding of the quality of the dataset as well as issues of sample size might impact the ability to draw conclusions is the argument you put forward about changing butchery patterns. You argue that: Butchery patterns within late first millennium and early second millennium assemblages from the middle Kafue and Batoka Plateau regions exhibit a shift from an earlier pattern in which entire carcasses are represented in the assemblage to specific portions of the carcass being brought back to the village.  There are, of course, challenges in distinguishing between wild and domesticated animals based on faunal remains. This can be exacerbated by fragments not identified to species, which are often reported in a broader taxonomic category (eg, bovids), which can be remains of both wild and domestic animals. 1

[email protected]

52

3  Comments, Dialogue and Supplemental Reading: South Central Africa

This argument seems to be based on Fagan’s analysis of the distribution of animal body parts from Isamu Pati (Fagan 1967:83, Fig. 29). Fagan argues that “it is also probable that game meat was comparatively abundant, since there is careful and systematic selection of those body parts which were brought back to the mound” (1967:82). However, in this case we do not know the sample size used to create these frequencies; using other tables as a guide, however, suggests that it was small. This argument is then tempered when he offers a number of alternative reasons as to why this distribution occurs, including “complicated ceremonial and sociological limitations which affect the dismembering and distribution of slaughtered stock or game” and “differential preservation.” This is the place where one hopes that the archaeologist will raise additional information to either support or exclude certain alternative hypotheses, but in this case Fagan does not. Some of this discussion equally applies to ceramic assemblages, for example, thinking about whether there is an adequate sample when comparing between contexts or assemblages. But there are different issues to consider when using published ceramic data. In general, one must understand how ceramics were analyzed to produce the descriptions presented. As you well know from your own critiques (de Luna 2012a), the description and presentation of ceramics in Zambia, and in the Batoka Plateau was seen as key to understanding the Bantu Expansions, and thus there was much effort put into descriptive analyses (Huffman 1989; Katanekwa 2016; Phillipson 1977). Over the years, there has been a proliferation of different types and ceramic classification systems. Though you might want me to offer up a simple summary of this complex data set, I cannot do that here because it is likely not possible to make all of these different classification systems compatible. There are some studies that document the analysis process (such as Derricourt 1985), and these are the type of datasets that should be most useful to you. However, the difficulty arises when attempting to compare ceramic material between sites. Unless the same analysis system was used for both assemblages, it is not possible to compare them systematically. Additionally, some archaeologists simply ascribe their ceramics to classes described at other sites, without carrying out their own analysis. This type of comparison can result in recognizing similarities between assemblages, but also risks masking important differences. One remedy to such problems is to reanalyze assemblages of ceramics from multiple sites into a single classification system—this is what Stephanie Wynne-Jones and I did to understand the variability between Early Tana Tradition ceramic assemblages along the eastern African coast (Fleisher and Wynne-Jones 2011). Assemblages that had long been regarded as uniform were shown to have important differences alongside the well-recognized similarities. Dating and Periodization: How to Use Radiocarbon Dating Results and To Think about ‘Levels’ and Stratigraphy  I turn next to some concerns about your use of periodization established by Fagan (1967) and Derricourt (1985) for the Kafue region and how you use those periods to examine change over time. The dating of materials via radiocarbon samples is of course very useful and important to archaeology. As highlighted in Chap. 1, there are several issues that we must attend

[email protected]

Commentary

53

to when working with radiocarbon dates, including the need to calibrate radiocarbon dating results and be aware of reporting conventions. I know from your published work (de Luna 2012a) that you have calibrated all published, uncalibrated dates, so what follows is not a critique but a methodological explanation for others attempting to use published radiocarbon dates. All radiocarbon dates must be calibrated to account for deviations in the amount of CO2 in the atmosphere through time. Today, most lab results will be returned with both the measured date (years before 1950) and some version of a calibrated date (applying different calibration curves). Easy-to-use programs are available to translate those dates reported as uncalibrated into calibrated ones. Importantly, however, many early reports such as Fagan (1967) and Fagan and Phillipson (1965) were published before the need to calibrate was understood. These publications report the measured date only. Next, calibrated radiocarbon results are offered at two probabilities—one and two sigma. At one sigma, there is a 68% probability that the dated material falls within the range offered. At two sigma there is a 95% probability that that the material dates within the range offered. Therefore, even though the one sigma probability often considerably narrows the date range, most archaeologists would be hesitant to use this range as a basis for arguments since it leaves a large probability of being incorrect. Two sigma probabilities offer the greatest confidence in dating the site, even if they often represent longer periods of time. Finally, both the measured date and the calibrated date should be reported as a date range—the range that extends plus/minus from the reported date. Even archaeologists are often guilty of misreading or misapplying this type of data when they assign a single date to radiocarbon results, choosing the median as a reference date. Just because the date is at the midpoint, it does not offer a greater probability of precision—the probability discussed above (one, two sigma) applies to the full date range. When reporting this range, it is important to note a calibrated date by putting ‘Cal AD or ‘Cal BC’ before the date (or, alternatively ‘Cal CE’ or ‘Cal BCE’), as well as noting if it is a one or two sigma probability. As discussed, Derricourt’s periodization is one of the most sophisticated for the region: radiocarbon dates from a number of his sites are used to construct an overall chronology for the phasing of excavations and the evolution of material classes. He establishes a tri-partite system—early, middle and recent—for the Iron Age as follows: EIA: Fourth to eighth centuries AD MIA: Eighth to eleventh centuries AD RI: Fourteenth to nineteenth century AD This contrasts with the more general periodization offered by Philipson and others that refer simply to an Early and Later Iron Age. With the issues presented above in mind, let’s consider the way that Derricourt uses radiocarbon dates to establish chronological control. The radiocarbon dates from Kalala Island, the site with the most radiocarbon samples, are used as a foundation for all of his chronological reconstructions and therefore merit careful atten-

[email protected]

54

3  Comments, Dialogue and Supplemental Reading: South Central Africa

tion. In his Appendix 2 (1985:136) he lists all the uncalibrated radiocarbon results, and offers a single calendrical date, which is the midpoint of the +/− distribution. On its face, this is problematic, for the reasons sketched above. However, in the text (1985:65–6) he lists calibrated dates at both the one- and two-sigma levels. However, in the discussion that follows, he generally argues that the median date from the derived range is the mostly likely date, even though the ranges are extremely wide and include multiple centuries. For example, he (1985:66) argues that “we can therefore say the ‘most probable’ date for the floruit of phase 3 was A.D. 755, for phase 4 A.D.785; for phase 5a A.D. 815; for phase 6 A.D. 880; for phase 5b a date shortly after A.D. 815 would seem valid.” While he sees these dates as tightly clustered, he offers no other information as to why the ‘most probable’ date is from the median of the range. It is just as possible, based on the radiocarbon dates reported, that these deposits date to from the 7th to 11th centuries, if we look at the calibrated, 2-sigma date range. Therefore, what is actually a very wide date range indicated by the radiocarbon dates is telescoped into two centuries with no explanation as to why. This sense of a constrained chronology is then read into the occupation history of a series of degraded house floors in units C, D, E, F, H, I, J and L to suggest that they were all occupied between AD 755–880. This chronology is also used to interpret the cluster analysis of pottery presented later in the volume. All of this suggests that the short chronology and its basis for the Middle Iron Age needs to be reconsidered, given that the two sigma range for a number of the dates extend from AD 675–1035, a much broader set of dates than indicated by the arguments presented. How do these dating issues affect the way that you use the archaeological data for your linguistic reconstructions? Although I know that you calibrated all uncalibrated dates in published materials, you nonetheless rely on Derricourt’s periodization. I hope I have shown why one might be skeptical of this periodization, based on the way he used radiocarbon dates in constructing it. Reading Context: an Archaeology of Deposit, House, and Village  An understanding of the context of archaeological materials is important in thinking about how we envision artifact assemblages as ‘representative,’ for example of household activities or economic practice among other possibilities. Archaeological context can also offer insight into the spatial patterning of human activity. Such a scale might be represented in the distribution of materials across a site—either focused on a particular type of context, like houses and yards, or more generally as a pattern of material across an entire settlement. This scale often includes associations of artifacts and features from a single or related contexts. I highlight this scale because I think there is much that could be done with it when thinking about archaeology and historical linguistic data supplementally. In general, researchers make correlations at either the scale of the individual artifact (words for certain types of spears in your example) or at the much larger scale of language communities (in which case archaeologists often fall into the trap of correlating these with ceramic distributions). The contextual data that would allow us to be more creative in addressing a wider range of scales is (unfortunately) rarely available in African archaeology because so many projects rely on small, test excavations that do not provide the

[email protected]

Commentary

55

necessary scale to document contextually related assemblages of materials. The utility of contextual data is that it offers a way of putting assemblages of material culture into patterns of use, and thus could provide a link to some of the meaning derived from clusters of word histories to actual social action in the past. This is a key insight that I gleaned from working on this project with you: that historical linguists can reconstruct lexicons around particular topics. You convincingly capture such a scale when talking about the location of smelting activities; archaeologically, these would be known through a collection of materials such slag, tuyeres, and possibly fragments of furnaces and associated charcoal deposits. Here, you argue, “smelts were relocated from the village to uninhabited areas beyond the village” around the turn of the new millennium. The data you present is extensive (at least in terms of the number of citations) and seems to indicate that there is a robust pattern drawn from regional and site-based studies that support this conclusion (although seeing some of the actual data would have been useful as well). This is perhaps one of the most convincing deployments of archaeological data in service of your arguments—part of its strength, I would argue, is that it offers contextual data with firm associations of artifacts in time and space. Several datasets in the material that you already draw on might also  prove useful in this regard; I turn to these now to offer some idea of their possibilities but also (again) the potential pitfalls. The most interesting dataset comes from Fagan’s (1967) description of the large-­ scale excavations he carried out at Isamu Pati. The extent of coverage is extraordinary, even in terms of the excavations carried out in the last couple of decades. His excavations covered hundreds of square feet, revealing a large range of features in situ, including house plans, grain bins, pits, and burials. Because there are such limited numbers of each type of feature, however, one must be careful not to assume that these are timeless patterns. Nevertheless, the spatial arrangement of such features—and the associated material assemblages—offer some of the most interesting data on how material assemblages were contextually related. The best-preserved structure, Floor IP/F/1  in CC (Fagan 1967:46, Fig.  11), includes the full plan of a house and associated yard, with good indications for the location of the door, hearth, internal storage, grindstones, and grain bins. Here, we can begin to understand distinctions between internal and external activities, with cooking inside, and grinding and grain storage outside. The assemblage of faunal remains found in association with the burned and collapsed house might offer some indication food remains related to the use-life of the house. From the text, it is difficult to know if the house was abandoned and then used as a dumping ground afterward, or if the house burned while still occupied and was then abandoned. With evidence of the latter scenario, one might be able to think about the faunal assemblage as an indication of the types of food brought back to the house and eaten there. This scenario would offer an actual domestic deposit that could to speak to ancient decisions made about how hunted animals were processed and what parts were brought back to the settlements, providing a firmer material basis to the butchery patterns you suggested and I discussed above. The house assemblage from IP/F/1 is

[email protected]

56

3  Comments, Dialogue and Supplemental Reading: South Central Africa

exceptional, to be sure, but if Fagan had reported assemblages of material that were associated with other likely house deposits (Floors 9, 19, and 20, p. 49) one might actually start to understand the relationship between archaeological assemblages and household consumption and/or ritual depositions, among many other issues. These are the types of patterns that archaeologists should work to build up, and they could also be important datasets around which a lexicon of the house, household activities, gender distinctions, ritual, and/or the social space of kinship could be used supplementally. There are other ways that this archaeological data might speak to your linguistic data and historical concerns; some of these are available in the archaeological record of your region but would require you to reinvestigate the published sources to collate that data. Because there is household level data from some sites, you could attempt to investigate how the development of houses and the activities associated with them (some possibilities discussed above) could provide an important contrast with developing notions of the bush. Looking at faunal assemblages related to particular houses and contexts could also help to reveal practices in houses and villages that can be linked to the development of spearcraft. For example, if you could isolate particular household assemblages, you might be able to determine if certain houses were associated more closely with hunting (higher densities of hunted animals?), and others more associated with domestic animals. This could serve as a supplemental dataset to think about the “social stakes of the knowledge of spearcraft.” The evidence of communal burial and its association with the central kraal also provides some tantalizing indications of village-based practices that might be akin to those communal hunting practices that you see as associated with social integration. While some of the datasets from published work may not be robust enough to make these interpretations, future work could be guided by thinking through the kind of data needed. Other intriguing features from Isamu Pati include a series of pits, dug into the levels below the floor of the houses as well as a set of pits dug into the lowest and sterile layers of the site. The former offer evidence of what might be called a ‘structured deposit’ (Richards and Thomas 1984; Gifford-Gonzalez 2014; Garrow 2012), comprised of intentionally placed and buried materials. Beneath the floors of both Floors 1 and 2 were such pits “filled with animal bones and wood ash which were dug under and sealed by the hut floor” (Fagan 1967:47). Both pits contained the bones of young wild animals (Fagan 1967:48–9, emphasis added): One is immediately struck by the considerable proportion of immature animals in these two caches of bones. The completeness of the skulls, even if the brain has been removed...suggests that the heads were buried deliberately before the floors were laid, and that the pits fulfilled some other purpose than that of rubbish pits.

Here then are archaeological features—well-buried, described and dated—that indicate the importance of burying juvenile wild animals as possible foundation deposits in the floors of houses. How might such an act resonate with the types of consumption you describe? Although we should be careful not to assume these are

[email protected]

Commentary

57

‘ritual’ depositions (Pollard 2008) we can assume, based on their context, that they were intentional. And the possibility of their ritual associations might be drawn out further by examining the distribution of the types of bones found in the pits and the type of consumption that they indicate; this is a process that Stahl (2008) has used to great effect in research in Ghana. The pits located in the lowest levels of the site, dug into the subsoil, offer insight into depositional practices from the earliest dates of the site. Fagan (1967:51–3) suggests that, based on the materials associated with these pits that they may have been used for storage, perhaps for vegetables or fruit. These features offer a beginning point to think about different types of storage at the site. Because there is evidence of both pits and above-ground grain bins in the archaeological record (we know from Floor 1 that grain bins were located outside the structure), you could also examine if there was a shift from storage in pits in the early periods of the settlement to storage in grain bins later. The final class of contextual data from Isamu Pati suggests a sense of the organization of features and structures across the site (Fagan 1967:53–7). These data offer evidence on the layout of the community, with houses located on two discrete high points separated by a ‘central area’ that was possibly used as a cattle kraal. However, this central area also contained nine burials, and thus it might have also served as a cemetery. At this scale we also see the differential distribution of animal bones (Fagan 1967:56, Fig. 16), with the greatest density surrounding the houses and the smallest numbers related to the central area. These data provide a sense—albeit limited—of how the village was inhabited and used, and allows one to begin thinking about the movement of people within and through it. Was the central area both a kraal and cemetery? If so, can we think about a linkage between domestic animals and memory practices of people and ancestors since they were physically associated in space? Taking into account the time when these individuals were buried, might this have been a moment when it became important to associate kin with activities related to domesticated animals? And how might this offer a supplemental narrative to those you offer about the importance of hunting wild animals? Some additional questions that an archaeologist might address would require renewed excavations and analysis. For example, one could ask where different types of foods were stored in a village, using faunal assemblages with better contextual associations and perhaps residue analysis on storage vessels. One direction might to reconstruct a lexicon around storage, pits, and forms of deposition. Were different types of foods—such as wild or domestic—processed, stored, and/or eaten in particular spaces, or managed in different ways in the village/house? If we were able to map not just the distribution of ‘bone fragments’ across the site, but rather understand what types of animals were found where, might we see other patterns of animal use and discard? Are there particular parts of the settlement where wild animals are being processed and are those the same as where domestic animals are processed and intentionally deposited? The data from Kalala Island would seem to be full of similar promise as that from Isamu Pati. Derricourt (1985) excavated eight 4 x 4 m units clustered together at the center of the site. He reports features from these units and shows the location

[email protected]

58

3  Comments, Dialogue and Supplemental Reading: South Central Africa

of likely collapsed houses in Figs. A4.5-A4.8. These data indicate similar types of house deposits as at Isamu Pati and, because Derricourt reports more completely the faunal material, it would seem possible to link faunal assemblages with particular houses, which could help to understand how people used different types of animals—wild and/or domestic—in their domestic space. However, a challenge with the Kalala Island data is that it is reported by phase and not by unit. Because of this, it is impossible to understand how the archaeological data related contextually or to see differential distributions of material across space. By reporting all data by phase, Derricourt has privileged the chronological changes in the assemblage over the spatial distinctions that might be seen within single phases. What if we were able to use the data from the Kalala Island excavations to examine the distribution of faunal remains across the spatial extent of the site for any single period? This would not only provide important contextual data, but would also facilitate comparison with Isamu Pati.

Dialogue Question One de Luna:  What struck me as the key take-away from your response is just how difficult it is to work with earlier archaeological publications. On the one hand, they cannot support the insights archaeologists can now develop from new techniques and from asking new questions. You provided compelling examples of what more we might wish we knew and how follow-up study or future research might reveal relevant evidence, with powerful ways of resonating with the topics I’ve studied through language. But, I am stuck with a published record that I now recognize as much more flawed than I had realized, beyond the fact that it lacks the sensitivities and methods of newer research. Fagan’s publications, for example, helped to shape the reporting practices that would become standard, but his volumes on the Batoka Plateau sites are still missing that “full accounting” of the faunal assemblage. To be honest, before reading your commentary on my essay and your explanations here, I would not have known how to determine whether or not it was a ‘full accounting.’ So, then, what do I do with older styles of reporting? Fleisher:  By pointing out the limitations of the data, my hope is not for you or others to give up on using the published work. Rather, I would hope that future assessments with such published data would be more honest about what it can and cannot do for a particular interpretation. Spell out the limitations, mention where the data has issues, offer caveats. I think that historians, in an effort to create ‘the narrative,’ often leave out such hand wringing, pushing it into the footnotes (Stahl et al. 2004).

[email protected]

Dialogue

59

For archaeologists this is the most honest way forward and allows the reader to decide on the strengths and weaknesses of an argument. de Luna:  To be more specific, I can see that we both face issues with preservation. What kinds of problems might impact the preservation of a faunal assemblage and how might a non-specialist approach the record with a critical eye if such issues are not reported? What, in other words, does an ideal presentation of the faunal assemblage look like? Fleisher:  The short answer to your question is that the best data to use would be that reported in a monograph or the appendix of an article. Such data would provide a fuller accounting of the analyzed fauna and provide an overview of the assemblage. Ideally, one would have access to the spreadsheets and/or databases from the original analysis, but these are rarely available; even the appendices in many publications may not document all the data that was recorded. There is a growing recognition that archaeologists must become better about publishing their data, or at least making it publically available. New digital platforms, such as tDAR (The Digital Archaeological Record, https://www.tdar.org/) and the Archaeology Data Service (http://archaeologydataservice.ac.uk/) offer places for long-term data storage, while sites like Open Context (https://opencontext.org/) offer new modes for publishing archaeology. An example from my own work is the Coastal Ceramics Project, which resulted in a peer-reviewed publication (Fleisher and Wynne-Jones 2011) that did not contain all the data; these were published in a searchable database with the Archaeology Data Service (http://archaeologydataservice.ac.uk/ archives/view/coastalceramics_2012/index.cfm), where the full database could also be downloaded. In terms of preservation issues, you want to see if the archaeologist discusses them; if they do not, then that might be a red flag and they may be moving uncritically from the data to the interpretation without considering the mediating formation processes. The challenge for a non-archaeologist using reported faunal data is that the data is often already manipulated—sorted into particular groups according to a particular variable—to make an argument for that particular article. If that argument suits what you are trying to say than it is not a problem. But if you would like to manipulate the data to see other possible patterns, then you need the raw data, in tables, and a clear sense of its limitations. In general, then, any high quality faunal report should include the following: an overview of the assemblage, a discussion of how the material was analyzed, a discussion of preservation and taphonomic issues, a full accounting of taxonomic representation by unit and level. Good examples of such reports from sub-Saharan Africa can be found in the project monographs from excavations in Mali and Senegal (MacDonald 1995:291–318; MacDonald and MacDonald 1995:311–334).

[email protected]

60

3  Comments, Dialogue and Supplemental Reading: South Central Africa

Question Two de Luna:  Let me pose a “nuts and bolts” question. You discuss different ways of representing the faunal assemblage, including NISP (Number of Identified Specimens), MNI (Minimum Number of Individuals), as well as weight measures and you imply that they tell us different kinds of information about the relationship between people and animals in the past. What kinds of historical questions are best answered by which type of numerical representation of the faunal assemblage? And, what other information might an archaeologist recover to explore the relationship between people and animals—butchery marks and tools? Hide processing tools? Charred meat on the insides of pots or even pots forms that reveal cooking techniques for meat? Can you give examples of rich studies that have probed this kind of material about the relationship between animals and people to get at the issues of meaning and practice on the African continent (or beyond)? Fleisher:  This is a very difficult question to answer since I do not think that particular types of “numerical representations of the faunal assemblage” can be connected directly to historical questions. I am not sure that I would say that different ways of representing faunal remains “tell us different kinds of information about the relationship between people and animals.” Different measures, however, may be more useful to compare across assemblages (using density measures as opposed to NISP alone; Logan and Stahl 2017), and help to ameliorate some issues that arise when NISP values are compared between contexts or sites. What these indices do is to offer different ways of describing the assemblage and these may be brought to bear on particular questions. For example, you are interested in examining differences between sites in terms of the percentage of effort residents placed on hunting wild animals and keeping domestic stock. For such a broad-based comparison, you could start with the simplest measure, NISP—just a count of the number of bones that are identified to a particular taxon—or (even better) to calculate the densities of material in different contexts (if you have access to data on both NISP and excavated volume). However, these measures can potentially overstate the importance of a particular taxon, since some animals might have more bones than others or bones subject to greater fragmentation than others; some might preserve better at a particular site; or the faunal assemblage might contain a single, complete skeleton of one individual. This is where the MNI measure helps to compare across assemblages—the hundreds of bones of the last example might now be narrowed to an MNI of 1. MNI values and density measures (NISP/m3) are also more useful for intrasite comparisons, examining distinctions between different parts of a site or perhaps different periods at a site. The challenge with MNI, though, is that the assemblage must not be too fragmented that you cannot determine elements or sides for a species. For all of these measures, however, and in thinking through the interpretation of variability in any artifact class, one must think about all the possible causes for the patterning that have been identified. Understanding how very different actions or activities might create similar types of archaeological patterns,

[email protected]

Dialogue

61

referred to as ‘equifinality,’ is a great concern for archaeologists, but especially zooarchaeologists and archaeobotanical specialists. In working with animal bones, for example, the data of NISPs, MNIs, and densities can be used to interpret social acts and patterns, but they also might be the result of natural processes that altered or transformed the assemblage post-deposition. An honest accounting of how different processes may have affected an assemblage should, and why the interpretation offered is the best fit for it, should be a part of any presentation of faunal data. There are many other variables that archaeologists  can record. These include modifications to the bones such as burning, cut marks, rodent gnawing, and abrasions, as well as evidence for breakage—did breakage occur when the bones were fresh or later, when they were dry? Burning can occur from all sorts of activities, such as cooking (roasting) but could also occur as part of the burning of trash. Being able to decipher the ‘meaning’ of burning would require contextual information— was the bone deposited near a hearth? Or was it found among other burned artifacts? These contextual clues might allow an archaeologist to make an interpretation. Finally, analysts can code not only the species and element of the animal but sometimes also the approximate age at death. Such data might help an archaeologist think about issues of seasonality and when certain animals were being hunted. With these data, you might begin to think about particular sites as places that were inhabited primarily by hunters at a particular time of the year, and the associated assemblage at such a site might help to think about what practices were happening in such a locale. In terms of examples that offer a “rich analysis … about the relationship between animals and people” with implications for “issues of meaning and practice,” there are few (but see Stahl 2008; Antonites et al. 2016a, b). So much of zooarchaeology is tied up in reporting the assemblage composition, the types of activities that created it (hunting and trapping, domestic stock keeping), and the way animals were used, distributed, managed, eaten, and discarded that it is hard find examples that get at issues of meaning. One very promising direction is what Russell (2012) has called a ‘social zooarchaeology.’ Her approach (and others like it) seek to move “beyond protein and calories” to examine how animals “fulfill a variety of roles—as pets, symbols, wealth, objects of feasting, and sacrifice, and so on” (Russell 2012:7). Although this is precisely the type of direction you seek, it is still a small but growing part of zooarchaeology, in Africa especially. As Antonites et al. (2016a: 347) note, South African zooarchaeologists are primarily focused on using faunal data for palaeoenvironmental reconstructions, and understandings of the “introduction and spread of domesticates in the region and…herding strategies and bloodlines.” While these same zooarchaeologists sought to “address the social aspects of human-­animal interactions” there were no articles in that special issue that do that (although Antonites et al. 2016b do look at the contextual meaning of worked bone artifacts). Let me turn briefly to literature on fish bones from eastern African coastal sites, both because I know it well and because it has been used in interesting ways. One of the earliest published zooarchaeological analyses was Horton and Mudida’s (1993) analysis of fish bones and shells from Shanga. This analysis was aimed at

[email protected]

62

3  Comments, Dialogue and Supplemental Reading: South Central Africa

investigating Swahili ‘origins’ and the debate (at the time) as to whether ancestors of the early Swahili were coastal farmers/fishers or pastoralists. In that article, they find the data ambiguous: the earliest levels of Shanga contained both small amounts of fish/shellfish remains and domesticates, leaving open the question. However, it was argued then that changing patterns of faunal remains at coastal sites, as well as differences between contexts within a particular site, might offer insight when larger and more samples were acquired. In a later publication (Mudida and Horton 1996, as part of the Shanga monograph), a similar, somewhat ambiguous conclusion is reached, although Mudida and Horton argue that the faunal evidence might provide evidence of a “multiethnic society” (Horton 1996:411), given that different types of fauna were found in different quadrants of the site (Mudida and Horton 1996:392). In the southwest and southeast of the site, the assemblages were dominated by chicken, sheepgoat, and cattle; these assemblages also included camel, dugong and fowl. The assemblages from the northwest part of the site had higher quantities of sheep/ goat and chicken, much less cattle, and no camel. This area also contained many more fish bones than the others. Mudida and Horton (1996:392) have suggested that these assemblages indicate that “separate communities of economic groups” were living at the site—perhaps a mixture of pastoralist, hunting, and farming/ fishing communities. The most recent work on fish bones is that by Quintana Morales (2013; see also  Quintana Morales and Horton 2014). Her work has benefited from much larger assemblages from a number of different coastal sites. These data have allowed her to compare aquatic subsistence strategies along the coast, demonstrating the variable ways that coastal communities exploited their coastal niches, showing distinct differences in the way offshore and nearshore communities fished and collected. What this work does is to start to use faunal assemblages to reconstruct social practices—types of fishing strategies—as well as distinctions between fish consumption practices across particular sites. Through this, we can begin to understand the nature of human-animal interactions, and how those were used to create meaningful distinctions in social practice. In the case of the Swahili, there might be certain types of fish that were used in elite consumption settings, perhaps feasts, while others were consumed more commonly throughout the community. This is similar to how I described household assemblages and their possible meanings above, and thus might resonate with your efforts to think about the implications of the privileging of skill among hunters on the Batoka Plateau. How would an understanding of how fished (or hunted) animals were distributed and used in a settlement help you to think about the way hunting created status among your historical actors? de Luna:  You won’t be surprised that I was uncomfortable with the idea that bone distribution tells us about ethnic quarters at a site or economic or occupational specialization. Language evidence could contribute to our efforts to understand how and why meat consumption sustained practices of inclusion and exclusion without

[email protected]

Dialogue

63

assuming that ethnicity is in play (though, we should also be careful not to reject such earlier forms of identity politics). Through language evidence, we can trace connections between changing patterns of meat consumption and other lexicons (of subsistence, settlement patterns, or cooking, for example), all the while attuned to semantic extensions developed to make and contest strategies of inclusion and exclusion (a species associated with joking relationships; a carcass segment associated with tribute; totemic species, and so forth). In this particular case, perhaps we see a specialization among cooks who sought to distinguish themselves (and those who consumed their food) from others in the household or the community. At this scale, we are talking about something rather different than an identity as encompassing as ethnicity. In any case, it would be interesting to put the two kinds of data together around these issues. I found the distribution of concentrations of waterbuck at certain areas of Tsodilo and in certain quarters at Bosutswe fascinating as I had many words for waterbuck in the same period in the middle Kafue. While the archaeologists posited this same argument about ethnicity and trade, I wondered about broader regional innovations in cuisine, cultivated taste, and aspiration as these animals were tied to the story I was writing about social aspiration and bushcraft (de Luna 2016; Denbow 1999, 2011; Denbow et  al. 2008; Wilmsen 2011; Wilmsen et al. 2009). Fleisher:  But the key here is not just finding enough faunal material or drawing together regional assemblages, but rather having access to large enough samples of data from a variety of different contexts to think about how animals were being used within a settlement. This is not to dismiss the importance of unique deposits and how these might provide information about emerging practices, but we can only know these are ‘unique’ by placing them in a broader set of contexts and assemblages. I have also worried about how certain forms of animal remains have been uncritically associated with particular ‘ethnicities’ and that is why work like that of Quintana Morales is so important, because it builds out from increasingly robust datasets that allow us to see distinctive patterns within and between settlements. We need think carefully about the meaning of patterned variation in archaeological assemblages and not assume that it emerged from a reified sense of ‘ethnic’ differences.

Question Three de Luna:  Dating seems to be place where we share concerns about representation. You explain that “archaeologists and others often assign a single date to radiocarbon results, choosing the median as a reference date.” As an archaeologist, why do you think some colleagues focus on medians rather than ranges? I can’t help but wonder why both of our fields need to work with ranges, but tend to mask or simplify those ranges to meet the expectations of audiences. Is this hold-over from an

[email protected]

64

3  Comments, Dialogue and Supplemental Reading: South Central Africa

approach to the past that privileges events and dated texts in order to foreground sequence and causation (rather than coproduction and inter-relationality)? Is this about the ­primacy we give to the “first” or “earliest” in a quest for origins (of urbanism, of agriculture), often to be able to identify who, exactly, was responsible for such innovation? Oddly, such a tendency might have further muddled attempts at correlation undertaken by folks like Nurse and Spear (1985) or Nurse and Hinnebusch (1993) as discussed in your chapter. This kind of chronological interval (measuring with a range of centuries) is a way in which our datasets are actually very well-matched. Fleisher:  I have a relatively short answer to this: I think that archaeologists would like for their radiocarbon dates to be more meaningful than they are, and they are often willing (it seems) to throw caution to the wind when presenting radiocarbon results to make a particular case. This is what is happening when archaeologists report the median date from a range, especially if it fits a particular interpretation that has emerged from other data sources. I also think that we need to do a better job in teaching students how to present and critically assess radiocarbon dates. I’m not sure this has to do with the privileging of events and dated texts, but rather is more likely the result of confirmation bias, where the archaeologist ends up bending the radiocarbon data to fit a preconceived argument. In the very near future, we will also need to reckon more often with the application of Bayesian modelling to radiocarbon dates, as this will surely become more common. Such approaches will allow for more fine-grained chronologies, but will require similar critical engagement with the quality and basis of such models (Bayliss 2015). de Luna:  You see the narrowing of the range as a strategy to seriate processes, sites, changes in the service of arguments that require a much finer chronological scale than the dating data alone allows. Still, it is a shame that the culture of academic writing is to make an argument, rather than make many arguments. For me, this still comes down to the fact that, uniquely, within our two disciplines, we are each in conversation with colleagues (historians and anthropologists) who work on much finer, confirmable time scales because they work on much more recent periods. Fleisher:  One other problem you face is that radiocarbon dates processed in earlier decades had much larger error ranges than those calculated today with AMS dating. So, for example, the dates that Fagan reported from the 1960s have very large ranges, often +/− one hundred years or more, while AMS dates can return ranges that are +/− 20 or 40 years. This means that over time archaeologists have had access to smaller date ranges even from their radiocarbon dating results. A greater challenge today is how archaeologists report calibration results, since these often do not fall into tidy date ranges but can also indicate a variety of possible ranges.

[email protected]

Dialogue

65

Question Four de Luna:  I learned a great deal from your comments on context. It inspired me to think about the kinds of vocabulary that might better match the kinds of contexts archaeologists can recover. Needless to say, this way of thinking with archaeological evidence foregrounds the lexicons of space and movement (through space). This discussion also brought up a shared concern with motive and meaning. How do archaeologists think about motivation? This question hit me as I thought about your insistence on the distinction between an “intentional” deposition and a ritual deposition of pits filled with the bones of juvenile wild animals under foundations. Fleisher:  There are many directions to go in answering this: the turn toward meaning in early symbolic and structural archaeologies (Hodder 1982, what has become known in the shorthand as ‘postprocessual archaeology,’ discussed briefly in Chap. 1); research on agency in archaeology (Dobres and Robb 2000) and how to think about the agency of past actors through material remains; studies of memory and materiality (Mills and Walker 2008a); and finally the interventions on ritual (Swenson 2014) and everyday life (Robin 2013) in archaeology. However, what you are referring to is what I indicated in my comments, and that is the study of depositional histories (Joyce and Pollard 2010), and a somewhat complicated literature on ‘structured deposits’ (Richards and Thomas 1984). This terminology refers to some early efforts to name and describe depositional practices that were seen as intentional and most often related to ritual acts. These may have been the deposits of certain animals, materials, or ancestors in places and contexts that mattered. Thus, rather than simply finding unintentionally deposited debris of past people, archaeologists sought to identify contexts that were believed to be buried with care and meaning. There have been many criticisms of this approach, however, in that it assumes that some deposits are ‘unstructured’ (Joyce and Pollard 2010; Pollard 2008). As Mills and Walker (2008b, p. 9) note “all deposits are structured in some way” and it can be argued that it is the task of the archaeologist to understand how they are structured. A very interesting example of work that pushes in these directions is Ann Stahl’s (2008) study of shrines and sacrificial practices in AD 1400–1900 Ghana. Without going through her entire argument, she attempts to track through time the way that certain animals and objects became ritually important at certain moments in time but not others: in this example, she adopted a strategy in which she ‘followed the bones’ of animals that were incorporated into formal shrine contexts (dogs and pythons). Her example showed not only how dogs moved in and out of use as sacrificial animals, but also how they may have moved “out from shrines…as they are butchered and dispersed.” By tracking this, Stahl is able to “gain insight into how the flow of body parts through a settlement potentially created, maintained, or perhaps disrupted social relationships.” An important finding of this type of research is that the effort put into determining distinct ritual and nonritual contexts is likely

[email protected]

66

3  Comments, Dialogue and Supplemental Reading: South Central Africa

missing the important ways that ritual acts are “embedded in daily life” (Stahl 2008:182; see also Bell 1992). This moves the discussion of intentionality out from what might seem like more ‘secure’ contexts like shrines and other ‘ritual’ deposits into the deposits that result from everyday acts.

Question Five de Luna:  I feel more confident after your discussion of dating about the commensurability the chronological scales of our datasets. But I am very uncomfortable now about finding ways to match our geographical scales. Part of the issue is that historical linguistics speaks to larger, less well-defined geographies. Archaeology is very site-specific, even with robust surveys protocols covering a large survey universe. How do you think archaeologists can scale up to the regional level you see me trying to work with in this paper? If the published record can be so problematic, is this an impossible task until a) reporting is standardized and b) a dense enough record of archaeology exists in a region? As I reflected on this problem, I grew all the more pessimistic when I realized that archaeology tends to focus on particular regions of Africa. Archaeologists often work in one or two areas for their entire careers and the regions of densest archaeological activity do not overlap with the regions of densest historical linguistic study. Fleisher:  In this area, I share your pessimism. Technically, there are no limits to the scale of archaeological research and an archaeologist could start with the region that encompasses one of the branches of your language families and propose an approach to study the archaeology of that region. Beyond the fact that such a project would require dozens of lifetimes to even make a dent, there are a number of trends and practices in sub-Saharan Africa that decrease the likelihood that archaeological data at this scale will be produced. First, as you know there is extremely poor regional coverage across much of the continent. Even regions in which there is a sense that we know a lot, like eastern Africa, are extremely spotty and poorly covered. Here is one surprising example: there has never been any systematic survey coverage on Zanzibar Island! And of course the amount of area surveyed in mainland Tanzania is extremely small. There are many, many reasons why coverage is limited: there are very few African archaeologists compared to other world regions; there are many areas of the continent that are either extremely inaccessible or too dangerous; there are very limited funding sources for archaeological research and only a few African countries offer the type of internal funding that would allow for long-term survey archaeology; the quality of archaeological survey is generally not very good, as many surveys are unsystematic or poorly reported and documented; there is much work focused on culture history, a type of approach that tends to favor small-scale excavations at sites and only limited survey coverage; and finally, there are few outlets for archaeologists to report their data fully. This last point has to do with a more general shift in archaeological publication. Currently, most archaeologists publish in peer-reviewed journals, which tend to

[email protected]

Supplemental Reading: Rethinking the Scale of Household History in South Central Africa

67

focus on argument-focused work rather than reportage. The reporting of data was once contained in  departmental reports or monograph series, but these are very rarely supported these days. The monograph series that do exist now ask the author to fund publication, which is normally a very expensive proposition. I have already discussed the new possibilities of making data publically available in digital archives, or of publishing data online; such resources will hopefully encourage researchers to make these data available for future analyses. These resources will hopefully be used by scholars who are subject to reporting requirements from their funding agencies: for example, the National Science Foundation in the United States now requires that grant applications have a clear ‘data management plan’ that describes how the data will be made publically available. We can also see the beginnings of more open source material with the recently published open source textbook on African archaeology (Livingstone-Smith et  al. 2017), an impressive and welcome addition. All of this is not to say that it is impossible to imagine an archaeological program that was scaled to the work that you do with historical linguistics. However, for such a program to truly connect with the regions that your work defines, it would have to be strategic and coordinated closely with historical linguists to think about how to build an adequate sample of your region such that their data might be representative of those regions. Just thinking about archaeological survey, an archaeologist could cover only a small portion of the region looking for sites and other archaeological materials, and if those sample areas were placed to address particular questions and with knowledge of the probabilities of where sites were located, then that sample might be considered meaningful. What I’m saying is that although there are many reasons why such work might be unlikely, it is not impossible. de Luna:  It strikes me that my efforts to work with the archaeological record focused on trying to scale that data up, with results that were problematic, as you have pointed out. But your comments have inspired me to try to scale down the narratives illuminated through language evidence. I may well be able to scale archaeological evidence up in some instances, such as the patterns I recognized in the changing location of smelting, but paying attention to the contexts of features and artifacts should illuminate iterations of processes I see at the regional scale in the linguistic record.

 upplemental Reading: Rethinking the Scale of Household S History in South Central Africa The vast difference between the geographical scale of our evidence was a particular challenge for de Luna’s efforts to work with an uneven regional archaeological record. The most richly contextualized archaeological evidence of how hunting articulated with social practice was buried in small pits under the floors of two houses abandoned sometime between the late tenth and late thirteenth centuries. These two houses were, themselves, among the last built after centuries of

[email protected]

68

3  Comments, Dialogue and Supplemental Reading: South Central Africa

occupation at a single site on the Batoka Plateau: Isamu Pati. The richest linguistic evidence spoke to the hunting practices of a region that was focused on the middle Kafue in the last quarter of the first millennium but, by the end of the thirteenth century, spanned a region about the size of the U.S. State of Alabama, from the middle Zambezi in the south to the middle Kafue and beyond. Following the richest, highest-quality published archaeological evidence of human practice discussed by Fleisher necessarily focused our attention on Isamu Pati and, therefore, the innovations of Kafue speakers, some of whom settled among the long-occupied Batoka mound sites in the last decades of the first millennium and the first centuries of the second. Fleisher saw challenges in de Luna’s efforts to scale up with respect to the sampling strategies and reporting practices of different archaeologists. Without the skills to undertake a restudy of regional materials, there are only certain kinds of archaeological evidence unaffected by the problems of sample size and reporting. Therefore, there are only certain kinds of archaeological evidence that can easily be compared in the published record at the geographical scale of linguistic evidence (e.g. regional changes in the locations of smelting). Yet, the archaeological record best illuminates practice when at the scale of the many interrelationships within a single site and its hinterland. Some such relationships reach a level of specificity connecting a very small collection of objects and features, such as a pit, its contents, and a house floor. De Luna saw in Fleisher’s persistent focus on domestic spaces the difficultly of recovering archaeological evidence for activities undertaken beyond the village, much less at a scale that might match the historical linguistic geography. We ended our dialogue wondering how or even whether one might be able to analyze the published archaeological record of an entire linguistic region with the careful attention to meaningful practices preserved in the tightly focused, carefully parsed reading of relationships between artifacts, features, and contexts that Fleisher advocated. But it was just this challenge of geographic scale that opened the opportunities of the supplemental approach. It was specifically in trying to weave the two historical records together in a loose, flexible fabric, keeping the gaps and overlaps simultaneously in sight, that we were able to tell a story about a local iteration of a broader process of social interaction and integration. Significantly, the innovative practices described in the narrative that follows were developed by two households in a specific community of the Batoka Plateau, as evidenced at Isamu Pati, in the process of their participation in a much more wide-spread regional phenomenon: the expansion of Kafue speakers and Kafue practices onto the Batoka Plateau. Other linguistic and archaeological studies have approached similar demographic transitions by tracking pottery styles, subsistence technologies, or linguistic strata visible in modern vocabulary. Indeed, we know this period of interaction and subsequent integration on the Batoka Plateau from multiple sources (summarized in de Luna 2012a, 2016). Although we know the general outcome, we know very little about the diversity of strategies deployed in the actual negotiations of such regional changes at the local level.

[email protected]

Supplemental Reading: Rethinking the Scale of Household History in South Central Africa

69

The power of the supplemental approach deployed here is that it allows us to connect the two records, keeping in play simultaneously the limits and riches of their geographic scales in order to tell a specific, local story about a regional process. In this way, the specificity of the archaeological record sheds new light on more broadly dispersed linguistic innovations just as those innovations highlight new meanings for the intentional acts of residents of two Isamu Pati households. These two records intersect in the story of the pits carved out and filled with the skulls and teeth of goats, cattle, klipspringer, warthog, *n(y)alʊvwɩ (reedbuck), and, most prominently, *n(y)asɩ́á (duiker). What follows is an evidence-based historical narrative that emerges from a supplemental strategy. Building the Foundations of Households on the Batoka  In the late first millennium or early second millennium, an aspiring householder set out to build a new home on the southern rim of the long-standing village of Isamu Pati, a home archaeologists now reference as IP/F/2 (Fagan 1967:45–49). The work began with a small pit, into which was deposited the split skull of a female duiker, whose brain had been extracted and perhaps consumed over conversation about the impending construction project. If words accompanied these deliberate actions of the pit-filler, with this skull, the new proto-Kafue style of adding the feminine “mother of” prefix applied quite literally for this was, indeed, the skull of a female duiker and one old enough to be a mother. Two additional duiker skull bones joined the first, along with the mandible of a young, immature one, only a half-year old. Even if not female, they were still named as *n(y)asɩ́á. An upper and lower jaw of a reedbuck, *n(y) alʊvwɩ, was added to the pit, along with fragments of bones from several other animals, but the pit-filler ensured that even the majority of these fragments came from the carcasses of the shy, independent, elusive *n(y)asɩ́á. The pit was filled and smoothed over and the work of building the first layers of the mud floor began. The outline of the house took shape, placing the now-covered pit of wild antelope bones—primarily duiker bones—near the entry, a short stride from the doorway and slightly to the left. Poles were struck into the wet floor, encircling it, and grass and branches were woven between the poles, to be plastered with mud, probably by women in the family. The house was about four or five strides across and the homesteader likely finished it with a grass roof supported by light wooden rafters. The preparations for this house-raising surely began days and perhaps weeks or months in advance. Moving house was a common enough occurrence, though not everyone had the resources and connections to build a daga house on the edge of the central kraal at Isamu Pati and regularly coat its floors with dung. This effort was carefully managed. The mud, grass, and wood were collected, dried, mixed, or bundled. And a hunter—perhaps, himself, a member of the household—had bagged a particular collection of game, yielding skulls, jaw, and other bone fragments to surrender to the household, perhaps as a gift or an exchange. The duiker and reedbuck may have been hunted with neighbors, cousins, in-laws during a chila hunt or, perhaps, by a lone hunter. Each bone deposited in the pit likely carried memories for the hunter(s), the members of the new house, and their relatives and neighbors: memories of particular chases, spear-thrusts, gifts, and meals, all laid within the

[email protected]

70

3  Comments, Dialogue and Supplemental Reading: South Central Africa

cache buried near the doorway, slightly behind the screen of the wall and secured beneath the house floor. Eventually the house was complete and a wife tended is hearth, cooking for her spouse, children, in-laws, and relatives. Now and again, she and her relatives and guests likely directed a moment’s thought toward the aspirations and anticipated effects associated with the cluster of bones buried near the doorway. After all, this effort was the first of its kind and, therefore, bore the burden of careful planning and continued assessment. No other house floor excavated between the middle Kafue and middle Zambezi shows evidence of a similar, intentional deposit interred beneath it before the one at IP/F/2. This was, it seems, an experiment that mobilized in a new way a more wide-spread set of ideas about the unique power of particular antelope species (those renamed with a feminine prefix) in human efforts to secure households’ well-being and success. Sometime later, the house was burned—accidentally or purposefully. But, another house, IP/F/1, was built on the same site in the time it took 35 inches of material to accumulate at the site—perhaps a generation or two later. The new house-builder knew of and employed the strategy of the bone-filled pit, crafting the second one at Isamu Pati and the only other yet discovered between the middle Kafue and middle Zambezi Rivers. The second pit deposit affirmed the significance of the contents of its predecessor: again, the bulk of the deposited material consisted of the skull bones of wild fauna, including duiker and reedbuck (“2 adult duiker mandibles, 1 maxilla of a very young duiker, 1 fragmentary maxilla of reedbuck, 1 very young klipspringer mandible, 1 bird bone and 2 warthog molars”).2 But the deposit also contained “1 fragmentary frontal lobe of a calf [and] 2 maxillae and 1 mandible of 2 goats (both immature),” thereby incorporating domesticated animals and placing an even greater emphasis on juvenile animals. Once again, as the house floor and walls were put into place, the doorway was oriented such that the pit sat beneath the floor at a location about two paces in and slightly to the left. The parallels with the pit dug beneath IP/F/2 are remarkable; together these intentional deposits attest to a local strategy associated with the creation of new houses, and by extension, perhaps new households and families, calling to mind possible concerns with fertility, accessing ancestors, and lineage and village politics. The deposits also tell us that wild and, later, domesticated animals were associated with the concerns and aspirations addressed by the deposits. They tell us that young animals were more significant than old and that bones from the skull could best do the work for which the pit was crafted. The persistent species—reedbuck and, above all, duiker—were best suited to this work: whether because they were the most efficient to catch or held symbolic meaning is difficult to assess from this record. It is also difficult to know why the earlier pit contained a very few bones from domesticated animals—goat and cattle—when these species were just as common in the rest of the faunal record at the time both houses were constructed. Finally,  Compare to Rosemary Joyce’s (2008) use of the concept ‘citation’ as a means of performing connectivity across time in practices that resulted in structured deposits. Joyce’s evidence, however, is far richer; therefore, the interations of practice she studies can be read as citations to the past. 2

[email protected]

Supplemental Reading: Rethinking the Scale of Household History in South Central Africa

71

the archaeological record tells us very little about why this strategy was employed in the first place—and, for that matter, in the second instance. Making Kin with Antelope: Rethinking Household-Founding on the Batoka  The historical geography of the regional linguistic record suggests that at least some residents of lands near and perhaps including the village of Isamu Pati spoke Proto-Kafue, a language with a projected homeland in the middle Kafue region to the north at the time the pits were constructed. By about the mid-thirteenth century, Kafue languages were the only ones spoken in the area around Isamu Pati. This process of integration and replacement occurred at the same time that social and subsistence innovations dating to the proto-Kafue era were developed in domains of life reflected in the pits beneath IP/F/1 and IP/F/2: hunting and creating households.3 From the linguistic record, we know that the process of spreading Kafue ways of life onto the Batoka and down into the Zambezi Valley involved new practices of reckoning kinship through the mother’s side only (matriclans and/or matrilineages; here we use “matrigroup”). During this era, speakers developed vocabulary naming new kinds of marriage practices, which, ideally, involved cohorts of classificatory cross-cousins (de Luna 2016). More specifically, young, Kafue-speaking brides traveled to establish households in the villages of their husbands (virilocality), extending the geographical reach of their own matrigroups and ensuring that the matrigroup’s children had close ties to their father’s kin, even if father’s kin were no longer considered relatives of a father’s children. In this way, a matrigroup could “plant” her daughters and her daughters’ sons in new places (daughters’ daughters would move away at marriage). Daughter sons’ would remain in their father’s village, have children with a wife from some other village (children who belonged to another matrigroup), and, hopefully, live an accomplished life and become an ancestor interred in the soil of their father’s place, thereby extending the reach of the matrigroup’s efficacious ancestors and, thereby, its claims to land. This process of integration was also implicated in the doings of daily life, including large-scale collaborative neighborhood hunts, *cìlà. The aspirational quality of hunting, *-wèz-, and the transformative powers associated with its landscape, *-sókwe (the bush), had been established by Kafue speakers’ ancestors. Communities used *cìlà as a method of social integration that facilitated the establishment of *-pàdʊ́, shared neighborhood heroes who would, in quite public ways, become famed, influential and—quite significantly—common ancestors for their matrigroup. Large-scale hunts also created a stage for the demonstration the kinds of skills used in raids for brides and cattle. Significantly, we learn that among all the animals hunted using skills and tools developed in the centuries before and after the  In an earlier draft of this supplemental reading of the linguistic and archaeological evidence of the Batoka Plateau, Fleisher observed a ‘paired contrast’ and suggested a reference to Hodder’s (2006) research on Çatalköyük on the basis of parallel processes of bringing wild animals skulls into the house in order to manage transitions in household composition and new balances between wild and domesticated foods. 3

[email protected]

72

3  Comments, Dialogue and Supplemental Reading: South Central Africa

turn of the millennium (in the eras of Central Eastern Botatwe and Kafue speech), it was during the Kafue period that three kinds of antelope were gendered female through the application of a prefix signifying a gloss near to “mother of”: *n(y)anjá (Kobus lechwe), *n(y)alʊvwɩ (reedbuck), and *n(y)asɩ́á (duiker). This gendering suggests that these species in particular may have been significant in new practices of gendered gift-giving and new understandings about fertility, marriage, and social reproduction. The dominant presence of *n(y)alʊvwɩ (reedbuck) and *n(y)asɩ́á (duiker) bones in the pits beneath the pair of houses built at Isamu Pati independently evokes a similar conceptual association between these particular species and the reproductive work undertaken in domestic spaces posited through the addition of the feminine prefix to their names. In this way, the pits attest to the (limited?) spread of these associations onto the Batoka Plateau. As Isamu Pati residents slowly adopted the newly introduced marriage practices and kinship ties supporting matrilineality, the daughters of the community left to establish their first married homes and to raise their children. And new women came to Isamu Pati, perhaps with very different ideas about how to establish a home with the kinds of protections ensuring fertility, success, and well-being. When they were Kafue speakers, these foreign wives may have demanded that ideas from their home villages in the middle Kafue be manifest in the construction of new houses and households. After all, these protections were not yet in evidence in the lands into which Kafue brides moved. Perhaps the first experimental pit under IP/F/2 was dug and filled by the kin of a foreign Kafue bride, as a way of inscribing and establishing their lineage and its values at the doorway of this new outpost of the matrigroup. For Kafue-speaking brides, these uncertain efforts deployed symbols of home, but those symbols worked differently and carried new meanings in foreign contexts. The most obvious is the fact that there were no *n(y)anjá (Kobus lechwe) on the Batoka. Thus, the deposits could not include all species gendered female in Kafue-­ era speech. But, perhaps this mattered little to foreign Kafue brides and their families. Those species available near Isamu Pati could still do meaningful work. The reproductive habits of the species of focus in the pits,*n(y)alʊvwɩ (reedbuck) and *n(y)asɩ́á (duiker), could easily call to fruition the ambitions of the Kafue families who sent their daughters to marry and raise children at Isamu Pati. Duiker ewes circulate independently in their territory, briefly accompanied by their young early in the offspring’s life. Reedbuck ewes leave the territory of their family group to partner, creating permanent pairings and small family groups. Anyone could see that the habits of the antelopes gendered female by Kafue speakers mirrored the kinds of reproductive practices with which Kafue and non-Kafue speaking families were experimenting on the Batoka Plateau and elsewhere in the region. The Supplemental Scales of Iteration and Process  Powerfully, taking a supplemental approach to the unique scales of archaeological and linguistic data allows us to track the initiation and abandonment of a highly localized experiment with much more wide-spread ideas and practices known for the region through the linguistic record. But, this supplemental reading of the archaeological and historical linguistic

[email protected]

Supplemental Reading: Rethinking the Scale of Household History in South Central Africa

73

records raises new questions. Were the bones interred beneath the houses of Isamu Pati householders who were experimenting with Kafue marriage practices on the Batoka Plateau buried as reminders (or exhortations) of idealized marriage and family practices? Were these practices initiated by families better connected beyond Isamu Pati, families with the means to support more robust homes adjacent to the source of Isamu Pati wealth (the central kraal)? Did the second act of pit creation incorporate domesticates to better reflect the sources of wealth that sustained successful families? Were these practices initiated during the fretful time where it was difficult to undertake the idealized virilocal marriage between cross-cousins in places without kin? After two or three generations, when outposts of middle Kafue matrigroups were well-established on the Batoka, perhaps such efforts to assert and assure the value of these forms of marriage were no longer necessary. This supplemental reading of the records emerged as a result of the writing of this volume, our dialogue, and the insistence of an anonymous peer reviewer on the value of attending to synthesis in addition to exposing the process of interdisciplinary learning. Without Fleisher bringing to light the unusual pits buried beneath IP/F/1 and IP/F/2, de Luna would not have noticed the significance of the specific species buried in the intentionally created pits beneath these house floors. And, without de Luna’s knowledge of the broader changes in kinship, marriage, and hunting practices, it would have been difficult to posit an interpretation of the meaning of the specific, limited species represented in a rare, highly-localized, and ephemeral practice, the impact of which we might narrow down to two or three generations, perhaps even of a single family. It is only through a supplemental approach—allowing our unique datasets to speak to their strengths and by accommodating their weaknesses—that we are able to recover this unique story. This is an extraordinary narrative moment in the history of the matrilineal belt in south central Africa because this fleeting story teaches us how two households used inventive, unique strategies to engage with the wider regional process of the establishment of the matrilineal belt. The linguistic and archaeological records independently attest to the coproduction of new ideas about families on the one hand and, on the other, ties binding homesteads and the bush constituted through the successful journeys of hunters, the compositions interred by pit-diggers against the very real possibility of failure, and the stooped backs of hopeful house-builders laying foundations—material, social, and conceptual—anew. This story traces the way a small group of aspiring Isamu Pati householders generated new forms of household by trying out novel material practices in the context of new ideas about kinship, marriage, and settlement. At other sites in the lands between the middle Kafue and middle Zambezi, men and women developed different tactics in the changing demographic context in which our two Isamu Pati households similarly helped kin, married, raised children, and honored ancestors. Here, in the absence of documents, the supplemental approach allows us to recover a small tale of tentative, hopeful efforts—a narrative with the kind of specificity and scale regularly recovered through archives, but for communities who left no written record of their hopes, fears, and strategies in rapidly changing circumstances.

[email protected]

Chapter 4

When Did Feasting Emerge on the Eastern African Coast? New Perspectives from Historical Linguistics and Archaeology

Introduction In this chapter I document a process in which my engagement with historical linguistics serves to challenge previous arguments about ceramics and feasting on the eastern African coast. I draw extensively on previously-published materials on feasting (Fleisher 2010), ceramics (Fleisher 2003; Fleisher and Wynne-Jones 2011), and my engagement with the historical linguistics of KiSwahili (Fleisher and Wynne-Jones 2012). In approaching the chapter for this book, I knew that I wanted to revisit one of my earlier studies, as a way to assess how my original work might have been transformed through an engagement with historical linguistics. As such, I began examining word innovations from historical linguistic reconstructions (Nurse and Hinnebusch 1993; Ray 2014) and came upon a set of reconstructed Proto-Swahili (seventh–ninth centuries AD) words that were related broadly to competition, ability, and power. I was surprised to find these words innovated during these centuries since I had previously postulated that a set of practices related to power and competitive feasting emerged forcefully on the eastern African coast only after AD 1000. After thinking through the complexities of archaeological and historical linguistic chronologies, I went back to the archaeology to think about how this discovery might challenge my previous arguments. What I found was that my data, in fact, offered some tantalizing evidence that I may have miscast the period in which empowering feasts emerged; some less obvious ceramic evidence, now bathed in the new light of historical linguistic word innovations, might help point to nascent forms of competitive politics on the coast, centuries before I had previously hypothesized. This paper attempts to chart this process of discovery and offer an initial argument about empowering feasts during the first millennium AD. I begin by describing the argument that I made about the emergence of particular forms of feasting on the eastern African coast, linked to newly-established forms of political practice. I originally (Fleisher 2010) constructed this argument through an © The Author(s), under exclusive licence to Springer International Publishing AG, part of Springer Nature 2019 K. M. de Luna, J. B. Fleisher, Speaking with Substance, SpringerBriefs in Archaeology, https://doi.org/10.1007/978-3-319-91036-9_4

[email protected]

75

76

4  When Did Feasting Emerge on the Eastern African Coast? New Perspectives…

assessment of ceramic data (both local and imported) as well as ethnographic analogies drawn from the broader literature on feasting (Dietler 2001) and the regional histories and ethnographies from the Swahili coast (e.g., Glassman 1995; Hollis 1900). I then examine historical linguistic data about collections of word innovations related to Proto-Swahili (PSW), particularly those that involve words for ­discussion, disparagement, and power. The historical linguistic reconstructions allow for the documentation of a particular set of social practices that emerged during the last quarter of the first millennium AD which include establishing social distinction through competition and verbal jousting. I argue that such practices would have been particularly salient in the period after AD 1000, when feasting practices emerged (based on the ceramic evidence). The linguistic evidence alerts us to the possibility that these practices had a deeper geneaology, and leads me to reconsider the object biographies that I sketched elsewhere (Fleisher 2010). WynneJones’ (2015:157; also 2016) notion of “biographies of practice” is particularly useful here; she argues that a “biography of practice focuses on…activities themselves, into which objects were entwined in various ways at different times.” In this chapter, I explore how archaeological and historical linguistic data can be generative of a biography of practice, both offering insight into practices as well as objects.

Archaeology and Linguistics in Eastern Africa Over the past 50 years, the archaeology of the eastern African Swahili coast has successfully documented the complex history of hundreds of coastal towns and villages that were not only linked to each other in interdependent and hierarchical relationships, but deeply entwined with relationships to communities far into the interior and across the Indian Ocean rim (Fig.  4.1; Horton and Middleton 2000; LaViolette 2008). We have an increasingly clear picture of how Swahili urban centers emerged early in the second millennium AD. The roots of most of these towns extend to large and small villages that were first settled between AD 600–800. The inhabitants of these villages were farmers, fishers, and kept some domestic animals. These communities were part of a sphere of interaction that extended from southern Somalia straight down the coast to Mozambique, and extended hundreds of kilometers inland, up river valleys (like the Tana River in Kenya and the Rufiji River in Tanzania), as well up into coastal mountain ranges and savannahs. These connections across eastern Africa were important to coastal villages, where local residents began trading with merchants that arrived from ports in the Persian Gulf and Red Sea. After AD 1000, there are significant transformations in coastal locales: some villages expanded dramatically, with the first evidence of buildings built with more permanent coral stone; many of these coral buildings were mosques, offering evidence of the growing influence of Islam in the daily life of coastal towns. The practice of Islam likely emerged slowly, over the course of the final centuries of the first millennium, with eastern African communities converting to Islam incrementally (Horton 1991). At the end of the first century of the second millennium, there was

[email protected]

Archaeology and Linguistics in Eastern Africa

Fig. 4.1  Map of the Eastern African coast, with sites mentioned in the text

[email protected]

77

78

4  When Did Feasting Emerge on the Eastern African Coast? New Perspectives…

growing evidence of hierarchy within many towns, with some residents now building domestic structures in coral, alongside a larger population that lived in earth and thatch houses (Wright 1993). All of these changes were tied to the growing importance of local and long-distance trade to towns, evident in the diversity of trade goods found in towns, but also in the development of architectural features that tied Swahili towns to Indian Ocean styles and aesthetics. By the thirteenth and fourteenth centuries AD, a handful of towns had become wealthier and likely more powerful than others, including places such as Mogadishu, Manda, Shanga, Mombasa, Kilwa, and Sofala. Dozens of other towns existed betwixt and between these larger centers and contained many of similar features, albeit on a smaller scale. By these centuries, many towns were ruled by powerful sultans or groups of elite families. These individuals, in addition to wealthy merchants, managed the flow of goods into and out of Swahili towns, the activity that formed the basis for most of the towns’ wealth.

Swahili Historical Linguistics and Archaeology I have discussed in detail elsewhere (Fleisher and Wynne-Jones 2012:183–86) the history of the relationship between archaeology and historical linguistics on the Swahili coast. As documented there, the important work of archaeologists, historical linguists, and historians was crucial in challenging and overturning the thesis that the Swahili had a colonial, Middle Eastern foundation. In the 1980s, collaborations between historical linguists and historians (Nurse and Spear 1985) that drew heavily on archaeological research (Horton 1986, 1987) “provided crucial data necessary to dismantle the colonial thesis of early Swahili society” (Fleisher and Wynne-Jones 2012:183). During this time, I have argued, “there seemed to be hope that the future of the Swahili past would be written between these important disciplines.” And yet the work that emerged in the 1990s did not follow this path, and instead focused on ‘origins’ questions that were ascendant in both fields. Archaeological research offered detailed studies focused on the origins of Islam and the Swahili people, ones that made little reference to historical linguistics (Horton 1996; Chami 1994). Historical linguistic studies (Nurse and Hinnebusch 1993) sought to reconstruct the language history, drawing upon archaeological research for chronological assistance, but little else. Thus, by the late 1980s and 1990s, researchers in both fields worked separately toward a chronological reconstruction of coastal languages (Nurse and Hinnebusch 1993) and material culture (Chami 1994, 1998; Horton 1996), aiming to chart out the settlement of the eastern African coast. There are significant challenges to correlating these data, involving the limitations of both archaeological and linguistic data. For the archaeology, this involves the way materials have been collected and interpreted as part of discrete “traditions” and “wares;” for linguistics, the limitations have to do with the Swahili language itself (discussed below) and the aim of the lexical reconstructions.

[email protected]

Swahili Historical Linguistics and Archaeology Table 4.1  Approximate date of emergence of reconstructed protolanguages and dialect clusters of the Swahili coast

Protolanguage Proto northeast coast (PNEC) Proto-Sabaki (PSA) Proto-Swahili (PSW) Northern dialects (ND) Southern dialects (SD)

79 Date AD 1 AD500 AD600–700 AD800–1200 ?AD800–1200a

From Nurse and Hinnebusch (1993) a Nurse and Hinnebusch (1993, p. 513) do not provide dates for SD due to the lack of archaeological research published at the time of publication; however, research since that time on the northern Tanzanian coast and islands has indicated dates similar to those for ND, perhaps even earlier.

The landmark study of the Swahili language (Nurse and Hinnebusch 1993) details transformations in its phonology, lexis, and morphology. Nurse and Hinnebusch argue that the current dialects of Swahili, broadly categorized into Northern and Southern dialect clusters (ND, SD), derived from a protolanguage, Proto-Swahili (PSW, Table 4.1). Swahili is one of the Sabaki languages, a set of six related languages spoken on the coast, including Mwani, Elwana, Pokomo, Mijikenda, and Comorian. Nurse and Hinnebusch reconstructed words for Proto-­ Sabaki (PSA), from which these languages emerged. Proto-Sabaki was one of four languages in the Northeast Coast Bantu group, which also includes Seuta, Ruvu, and Pare. Words were also reconstructed to this protolanguage, Proto-Northeast Coast Bantu (PNEC). This linguistic work referenced archaeological materials, using them as support for the proposed chronology and language development (Table 4.1). The work, however, was intended to be a linguistic history and not a history of the people that spoke Swahili, and thus did not seek to correlate these data sets in any systematic way. There are challenges in working with Swahili historical linguistic material in general and with Nurse and Hinnebusch’s data in particular. For one, despite the fact that Swahili is one of the best-documented Bantu languages, Swahili is relatively challenging for a historical linguist in that it emerges in an area with a high degree of linguistic similarity, thus making detailed word histories difficult (Nurse and Hinnebusch 1993:34). Additionally, the areal extent of Swahili is represented by many dialects, of which a number of key ones are no longer commonly spoken; this is due to the dominance of the dialect from Zanzibar, in the nineteenth and early twentieth centuries, and subsequently Standard Swahili (Nurse and Hinnebusch 1993:32). These features of the linguistic history mean that there is often not enough phonological change to reconstruct word histories that could be used to both understand past ideas, meanings, or things (and how they changed), especially for the periods related to the development of coastal towns, or even the settlement of the coastal corridor more broadly, from the seventh century onward. Finally, the nature of Nurse and Hinnebusch’s study—a linguistic history—does not necessarily lend

[email protected]

80

4  When Did Feasting Emerge on the Eastern African Coast? New Perspectives…

itself to detailed historical reconstructions. This is because the words that were reconstructed by them were those that provided insight into the history of the language but do not provide the density of words and ethnographic data that historians normally demand to write social histories (de Luna 2016; de Luna et  al. 2012; Schoenbrun 1998; Vansina 1990). Thus, in terms of the word histories at my disposal, I can work with what I have, not what I would want. This means that the words at my disposal for reconstructing historical changes are not really a ‘sample’—they are simply the words that Nurse and Hinnebusch chose to reconstruct, for the purposes of their linguistic history. This is the same problem that historians have with the data of archaeologists, I presume, in that archaeologists collect and interpret data in the service of particular questions, and historians might want to use the data that emerges to address other questions. Despite the limitations of these data sets, a number of historians have attempted to correlate archaeological and linguistic data to describe the outlines of early coastal history. These include Spear’s (2000) overview of early Swahili history, which attempts to correlate ceramic traditions and protolanguages (see also Nurse 1997:379). Gonzalez (2009:51–58) has attempted a more fine-grained correlation between linguistics and archaeology in “hinter-coastal Tanzania.” The problems with this type of reconstruction are legion. Most crucially, it relies on a problematic assumption that ceramic styles are related to language use. However, in an effort to make Chami’s work directly comparable with linguistics, Gonzalez (2009) problematically glosses issues of chronology and sample size. The recent contribution by Ray (2014) offers a unique and important step forward in using historical linguistics to write deeper, precolonial history. Ray’s dissertation covers an enormous span of time, but focuses tightly on the development of Swahili identity and the rituals that were associated with its construction. However, his argument about rituals focuses almost exclusively on the practices of reconciliation and group consensus. In every case, the development of identities and the integrity of the ‘social’ or ‘community’ is understood as a process of working through differences or distinctions. In this way, identity relies on the ability of groups to come to consensus or agreement, and the practices that he identifies are akin to leveling mechanisms in a functional model. The discussion below will offer a contrast by highlighting how certain forms of social practice, such as feasting, were settings in which power became possible, but not necessarily settings in which it would be confidently demonstrated. What I hope to show in what follows is that archaeologists can think more broadly about how language evidence might provide a social context in which to situate the use of material objects. Such an approach is ultimately quite different from the practice of relating archaeological and linguistic findings through direct associations alone, i.e., finding the physical object in the ground that had been reconstructed in the language, what we have called ‘correlation’ in Chap. 1. This may be useful in efforts to date the presence or absence of things, but it offers little in the way of historical understanding and the reconstruction of the practices and

[email protected]

Background on Food and Politics in Ancient Eastern Africa

81

contingencies of past people (see section on “Word, Things, and Practice” in Chap. 1). My approach here attempts a word and things approach that focuses on social practice. As will become clear, my experience working between material objects and historical linguistic reconstructions creates more questions than answers. However, in recognizing that both material objects and word histories provide indirect access to the interior lives of historical subjects I hope to construct an enriched historical understanding of these key periods of the Swahili past.

Background on Food and Politics in Ancient Eastern Africa There is currently a surge in research on food-related topics on the Swahili coast and the Indian Ocean world (Fuller et  al. 2011; Boivin et  al. 2014; Walshaw 2015). Much of this work is focused on questions of origins—when did certain crops and animals first become prominent and from where did they come? Such research has taught us much about continental interconnections and the complex movements of Asian crops, Near Eastern animals, and the early movement of African domesticates moving in the other direction. Although such questions are useful and important chronometric data, they tend to lack a focus on social context—they are narratives of time and tempo rather than social and cultural meaning and use. Without an understanding of context (both archaeological and social), such accounts often discuss the movement and adoption of plants in ways that seem to forget that it was people in history who moved, adopted, managed, incorporated and transformed them. There are exceptions to this literature, including my own previous research on the ritual politics of feasting (Fleisher 2010) and Walshaw’s (2010, 2015) paleobotanical work that examines the both the timing/origins as well as the social context in which foods became meaningful. Walshaw documented a shift in the cereal diet on Pemba Island from African millets to Asian rice. At sites occupied before AD 1000, plant remains located include primarily pearl millet, with small amounts of legumes, coconut, and rice. After AD 1000, Walshaw describes how “the economy had shifted toward rice, cotton, and coconut consumption” (2010:142), with rice occurring in all paleobotanical samples. Pearl millet, once the dominant grain, was only the fourth most common seed taxon in post-AD 1000 deposits. Walshaw has argued that this dietary change should be understood in concert with changes in other artifact classes, as well as social and religious changes that occurred on the island during these centuries. In particular, she (Walshaw 2010:149) argues that rice may have been “instrumental in feasting to secure political influence by Pemba’s elite.” That Pemban farmers chose to become rice farmers even though it was a much riskier agriculture than pearl millet possibly suggests the “rich social rewards associated with consuming rice, including the identification with, and practice of, Islam” (Walshaw 2010:150). Walshaw’s work on the social implications of dietary changes was supported by my

[email protected]

82

4  When Did Feasting Emerge on the Eastern African Coast? New Perspectives…

own research on Pemba, and my initial arguments on feasting and social change (Fleisher 2003), to which I now turn.

The Ritual Politics of Feasting on the Swahili Coast My research on feasting has attempted to link changes in ceramics assemblages and political forms, and to explore how forms of consumption were involved in the process of political negotiation (Fleisher 2010). As I have argued, “feasts were not merely reflections of power, but arenas within which power and authority became possible” (Fleisher 2010:200). I drew on Dietler’s (2001:76–85) distinction between ‘empowering’ and ‘patron-role’ feasts. The former are feasts in which individuals attempt to establish power while the latter feasts are contexts in which institutionalized power is signified or demonstrated. We can see evidence for patron-role feasts in the historical record, with Ibn Battuta’s visit to Mogadishu and Kilwa; an additional example would be the reputation of one of Kilwa’s most famous Sultans, who was known as ‘Abu al-Mawahib,’ ‘the Giver of Gifts’ (Freeman-Grenville 1962). In these cases, generosity was crucial to the effective rule of a Sultan. Empowering feasts are a bit different; they can include feasts in which there is little competition (such as feasts on market days) but also “rituals within which rank is negotiated and established, as in the series of feasts that mark progression through specific ranks or age-grades” (Dietler 2001:76). As such, these ritual activities can be socially transformative, allowing for the reproduction and constitution of relations of power. I underscore the transformative element of these feasts because it opens up the possibility that feasts fail, and that power can shift (Smith 2015). I argue (2010:202) that “this practice-oriented approach offers the possibility that feasts may not achieve their political ends, or that they may seek to propose desired relations of power rather than those already established.” This aspect of empowering feasts offers a challenge to the way that Ray (2014) emphasizes equilibrium and functional integration in the construction of identity in the Mombasa region. I focused on feasting on Pemba Island, moving backward from the fifteenth century, and seeing in the adornments of the fifteenth-century, Great Mosque at Chwaka (Fig. 4.2) evidence of the links between feasting/food and the maintenance of power. The imported bowls affixed to the walls surrounding the mihrab at this mosque may offer evidence of the public display of the feast, “visual reminders of the feast, as well as displays that invoked parts of these ritual performances, what might be called the display of consumption or completion” (2010:214). These displays, however, were likely related to patron-role feasts, evidence of the power to feed a community or the ability to entwine daily religious practice with the offering up of food. What was the historical basis for this monumental display? From what and where did this practice emerge and is there archaeological data that can be used to explore its emergence?

[email protected]

The Ritual Politics of Feasting on the Swahili Coast

83

Fig. 4.2 Congregational mosque at Chwaka, Pemba Island, Tanzania; top from Pearce (1920); bottom after excavations in 2004. The mihrab was poorly restored in the 1980s and thus now appears different from its original design

Reaching back to AD 700, I documented ceramic assemblages that were associated with first millennium settlements on Pemba Island and, in particular, settlements that were inhabited in the vicinity of what would become Chwaka. There are a number of first millennium sites in the region around Chwaka. The most prominent is Tumbe, a sprawling 20+ hectare settlement that, although it was a large village, was tied to the Indian Ocean trade system (Fleisher and LaViolette 2013). Evidence from Tumbe includes the remains of earthen houses and evidence for the

[email protected]

84

4  When Did Feasting Emerge on the Eastern African Coast? New Perspectives…

Fig. 4.3  These are pottery sherds from the Persian Gulf found on excavations at the site of Tumbe in northern Pemba Island, Tanzania. They date to the eighth to ninth century AD; to the left are sherds from Siraf storage jars, what would have been large, unglazed jars; to the right is a base sherd of a Sasanian–Islamic jar, also a large storage vessel

local production of beads and iron, but included in all household assemblages are many sherds of imported pottery, jars and bowls from the Persian Gulf. In both the imported and local ceramic assemblages, the dominant vessels were jars--large, deep containers with restricted necks (Fig. 4.3). Among the imports these included Siraf storage jars and Sasanian Islamic jars, and for locally-produced wares the oft-­recognized jars with incised and punctated decorations on the neck that are part of the Early Tana Tradition (ETT, Fleisher and Wynne-Jones 2011), but often referred to as Triangular Incised Ware (TIW, Chami 1994). Although the triangle inscriptions are carried out exclusively on necked jars, these vessels are just one of the many different morphological types that were part of coastal ceramic assemblages between AD 600–950 (Fleisher and Wynne-Jones 2011). These assemblages also included classes of vessels that might be considered more utilitarian such as spherical pots and simple open bowls, but also small numbers of decorated bowls (Fig. 4.4). Most of my focus in my original paper on feasting was on the ceramic changes that took place after AD 1000. On Pemba, there was a particularly dramatic change in the composition of ceramic assemblages: after AD 1000, most of the imported vessels were large bowls, including most prominently, various types of sgraffiato bowls from the Persian Gulf and celedon bowls from China (Fig. 4.5). The sgraffiato bowls were large vessels (up to 36 cm in diameter) with elaborate incised and chromatic decorations on the interior; the celedon bowls also contained many floral

[email protected]

The Ritual Politics of Feasting on the Swahili Coast

85

Fig. 4.4  This figure shows the full range of vessel types commonly found at late first millennium AD sites on the eastern African coast. This pottery is part of the Early Tana Tradition and includes five different vessel types: (a) globular jars, (b) necked jars, (c) open bowls, (d) closed bowls, and (e) carinated bowls, illustrated by Matilde Grimaldi

or other decorative reliefs within the characteristic pale green glaze. I argued (2010:206) that these bowls “must have been imported in response to local demand since they were not useful containers for shipping.” And, in this way, the great increases in numbers of sgraffiato bowls can be seen as an expansion of the category of vessels that once just included white-glaze bowls (Fleisher 2010:206): This dramatic shift alone—from jars containing useful food items, to large bowls—might be enough to suggest that imported ceramics were being used in novel ways in local contexts. The rapid increase in imported serving vessels might be the first indication that more communal and publicly-oriented consumption was taking place in emergent Swahili towns...It is possible, then, that the social acts associated with imported bowls were carried out by those attempting to convert their privileged trade relations into local symbolic capital, akin to what I have described previously as empowering feasts.

There was, however, an equally compelling shift in the local ceramic assemblage composition. During the eleventh century, a new ceramic form emerged at a number of archaeological sites; these were finely crafted large bowls burnished with red hematite, and some with graphited decorations (Fig. 4.6). Such bowls were larger and more elaborately decorated than those of previous centuries. And yet they drew upon decorative elements established in previous centuries—burnished red paint and graphited surface treatments—all the while emulating aspects of the imported bowls. For example, in both the sgraffiato and the locally-produced red-painted bowls, the decoration extends throughout the interior up to and over the rim, stopping just below the rim on the exterior.

[email protected]

86

4  When Did Feasting Emerge on the Eastern African Coast? New Perspectives…

Fig. 4.5  Figure showing different types of imported bowls found in eastern Africa, including sgraffiato bowls (picture, top left and profiles at the bottom, latter adapted from Horton 1996), and celadon rims (top right). The sherds are from thirteenth- to fifteenth-century AD deposits at the site of Chwaka in northern Pemba Island, Tanzania

 ethinking the Feast: Historical Linguistics and Archaeology R in Dialogue It is here that I want to depart from the arguments in my original paper. In that paper, I argued that the emergence of these large, finely-potted vessels, alongside the increasing imports of sgraffiato and other large bowls, “signal[ed] the earliest acts of empowering feasts where commensal hospitality offered the possibility of acquiring and maintaining forms of symbolic capital.” These vessels indicated, I argued, a “significant change in the way people were consuming food...the large vessels indicate communal acts of consumption, as one bowl would contain food for a number of individuals who would sit in a circle around it” (2010, p. 207).

[email protected]

Rethinking the Feast: Historical Linguistics and Archaeology in Dialogue

87

Fig. 4.6  This figure shows drawings and profiles of sherds of graphited red-slipped bowls from the site of Bandarikuu in northern Pemba Island. All date to the eleventh century AD and some show the pattern of graphited pencil lines extending from the interior base to just below the rim. The top row shows rim and body sherds; the bottom is a reconstructed bowl from rim and base sherds, showing the overall shape of the most common type of bowl

In my original essay, I drew upon Glassman’s (1995) description of feasting practices among nineteenth-century Swahili. In his study, feasting was used competitively as “a way to negotiate a changing economic world in which new sources of global power were ascendant” (Glassman 1995). In conjunction with Dietler’s (2001) definitions of feasting modes, this material was evocative because it provided an example of an historical moment when feasting became a central social practice that was linked to political and economic changes. Although the feasting practices were likely quite different in the eleventh century, there were similarities in the shift in consumption practices to include larger groups of people and greater efforts at public display.1  I use an ethnographic model/analogy derived from Dietler’s (2001) overview of feasting modes and draw on nineteenth-century Swahili historiography to help think about earlier periods on the coast. While familiar to archaeologists, this analogic strategy might be problematic for historians 1

[email protected]

88

4  When Did Feasting Emerge on the Eastern African Coast? New Perspectives…

I went into the historical linguistic data with an eye toward investigating words and word innovations that might offer some insight into the social practices of feasting that I was seeing emerge so clearly in the post-AD 1000 ceramic assemblages. Looking at the word innovations associated with the Northern and Southern Dialects, the language classes that should have been most closely associated with this period, I did not see much that was relevant; there were no word innovations that either correlated well with or contradicted my 2010 arguments. Broadening my search to earlier periods, however, forced me to rethink the historical development of these social practices. I expanded my search to earlier periods, to world innovations from Northeast Coast (NEC), Proto-Sabaki (PSA) and Proto-Swahili (PSW). In these innovations, I could see the way that new plants and animals were added to the repertoire of coastal economies, how crops became increasingly important (*nkonde, ‘cultivated area’; *lucaga, ‘raised grain store’; *lucelo ‘winnowing’; all PSA, Nurse and Hinnesbusch 1993:292) and how fishing skills seem to have broadened while remaining a key part of the productive economy. However, I was most struck by the rich language that emerged amongst speakers of PSW that seemed to describe both the creation of formalized sets of joking and mocking relationships, but also a vocabulary that describes a terminology that might indicate increasing competition. This included a collection of words for two different assessments of people: words of disparagement, scorn and ridicule, and words that described appropriate ways to talk, take action, and have power. I see the development of these as linked in that they are both conceptual and descriptive. On the one hand, the word *-puluz- means ‘be foolish’, ‘talk much, say little’ and on the other, *-kalamuk- means ‘be quick witted’ and ‘sharp’ as well as *-zųngumz-, ‘to chat’ (Nurse and Hinnebusch 1993:). In just this pairing, one can see two assessments of speech: one which recognizes skill in wit and language, and the other which disparages one for talking too much. The innovations during this period also include a rich set of words that are disparagements: *-jinga (‘foolish’), *-pumbal- (‘be stupid’), *-zųzų (‘foolish’; Nurse and Hinnebusch 1993:294–95). There are also a number of innovations that refer to joking or mocking, including *-zųm- (‘mock’) and *-Tanil- (‘joke, be familiar’). The latter term refers to a very specific form of social relationship that is well-­ documented in the ethnographic literature on eastern Africa: the joking relationship (Moreau 1941, 1944; Middleton 1992:110–12). This rich vocabulary describes both the creation of formalized sets of joking and mocking relationships, but also a terminology that might indicate a competition. During this period, forms of assessment were innovated that seemed to indicate the emergence of the public recognition of people who were successful in public speaking and performance, and those that were not. This surprised me, since, based on the linguistic reconstructions, these innovations seem to pre-date the period in which I argued empowering feasts were who could (rightly) point out that those instances of nineteenth-century feasting described by Glassman were related to a very particular historical moment. Rather than reading back the ethnographic or recent historical record, I tack between models (Wylie 1989), iterations, and the data to construct my arguments.

[email protected]

Rethinking the Feast: Historical Linguistics and Archaeology in Dialogue

89

emerging, the types of social contexts in which such forms of speech and terminology would have been salient. This sent me back to the archaeological data to think about how I was tying the emergence of a social practice to the dominance of a form of consumption. And it forced me to reconsider where and why I saw empowering feasts emerge only after AD 1000. At first, I was a bit surprised and partially discouraged, since these innovations seem to pre-date the period in which I argued empowering feasts were emerging. If, as I had argued in 2010, empowering feasts became a crucial social practice in the period after AD 1000, then one would expect that these practices would also be accompanied by a new and/or transformed language to describe and explain these practices. With what I perceived to be a misfit between the archaeological practices/ materials and the lexicon of the reconstructed words, I initially began thinking through how the linguists might have misunderstood or miscalibrated the chronology of their linguistic analysis. I had written about this briefly in my dissertation (Fleisher 2003:387); there, I attempted to challenge the way that linguists and historians had correlated the distribution of Early Tana Tradition ceramics with the spread of Proto-Swahili. My arguments relied on the somewhat casual way that Nurse and Hinnebusch (1993) had established the chronology of their linguistic periods— based on a narrative that brought together oral traditions (the Shungwaya migration narrative), archaeology (the widespread distribution of similarly-­decorated pottery along the coast with ETT), and the linguistic innovations they reconstructed. I had argued that perhaps linking the spread of PSW with eighth- to tenth-century ETT pottery was problematic, and might be better linked to a slightly later period, during the tenth and eleventh centuries AD. Although the end of the first millennium was undoubtedly a time of great social and cultural interaction along the coast and hinterland, the late tenth and early eleventh centuries witnessed the consolidation and development of places that would become Swahili towns (Fleisher et  al. 2014). What Nurse and Spear (1985) and Nurse and Hinnebusch (1993) both assumed was that the development of Swahili towns was a long, unbroken, evolutionary process that stretched from the eighth to fifteenth century. Greater emphasis on a more robust range of settlements, including those that were abandoned by AD 1000, has challenged this narrative, and allowed coastal researchers to see the variety in both regional developments, but also the way that Swahili town developments emerged in fits and starts (Fleisher and LaViolette 2013; Fleisher et al. 2015). My effort to challenge the linguistic chronology, however, was essentially an argument by assertion and I had no real way of assessing the validity of it from the linguistic data itself. Recently, however, Ray (2014) has done a great service for KiSwahili historical linguistics and computed the statistics derived from glottochronology and lexicostatistics. These data show clearly that Nurse and Hinnebusch’s original chronology is generally correct and thus firmly links the linguistic developments of Proto-Swahili to the period around AD 800. This effectively ended my proposed chronological critique, and thus I dove back into the archaeological materials to examine the ceramic patterns that led me to my previous arguments. In particular, I began thinking about the full assemblage of ETT ceramics associated with sites from the last quarter of the first millennium.

[email protected]

90

4  When Did Feasting Emerge on the Eastern African Coast? New Perspectives…

Fig. 4.7  This sherd is a red painted, burnished and graphited bowl from the site of Tumbe on Pemba Island and comes from eigth–ninth-century AD levels. The left image is of the exterior, showing the red painted surface and a band of graphite just below the rim, the right image is the interior, showing the same

Competitive Feasting Before AD 1000? An important data set is the analysis carried out by the Coastal Ceramics Project (Fleisher and Wynne-Jones 2011; Wynne-Jones and Fleisher 2013). This study of ETT ceramics was an effort to reclassify and analyze previously-excavated ceramic assemblages from a wide range of sites. The sample included different site types (large and small) as well as a variety of spatial locations (coastal/island/inland and a set of sites stretching from Manda in the north to Chibuene in the south, Fig. 4.1). That study offered a more robust description of ceramic assemblage composition for the period from AD 700–950. Five vessel types were identified (Fig. 4.4), including jars and bowls, and the frequencies of these different vessel classes were compared between assemblages.2 A notable finding was that all assemblages contained some percentage of open bowls, a vessel class often overlooked in the wake of typological work of Chami (1994) and others who have focused almost exclusively on necked jars with incised decorations. Three points emerged from studying the bowls (1) although they occur in all of the assemblages studied, their frequency differs substantially between sites; (2) the more general ‘bowl’ class includes a subset of decorated bowls, more finely potted, with burnished red paint and graphite (Fig. 4.7); and (3) like the similarity found between the dominant forms of ETT jars and imported storage vessels (described above), these fine bowls can be seen as related to a rarer imported vessel form, white-glaze bowls. Whiteglaze bowls formed a small percentage of the imported pottery and are quite small in size in comparison to other vessels. Whiteglaze bowls often had a  The Coastal Ceramics Project examined primarily rim and body sherds from 8 sites, analyzing over 2000 sherds. The frequencies included here on vessel shape and type are from counts of rim sherds (see Methodology section, Fleisher and Wynne-Jones 2011:257–259). Sherds were mended where possible, to reduce redundancies from a single vessel. There are, however, many limitations to this data, since it was excavated by at least six different researchers, each of whom applied criteria on what material to curate. 2

[email protected]

Competitive Feasting Before AD 1000?

91

‘splashed’ decoration at the rim in yellow, green, and blue. The remarkable concordance between the local and imported pottery assemblages in terms of vessel shapes and assemblage composition can be understood in one of two ways. The first suggests that coastal people sought particular types of imported ceramics; they had particular categories of vessels that were of interest and those were the vessels for which they traded. The second is that, as imported vessels were adopted into local forms of consumption, those vessel forms became inspirations for the types of vessels that were produced locally. This is obviously the distinction between external inspiration and/or selective adoption of material objects, and the reality might have been somewhere between these two positions. The locally made bowls that were part of the ETT materials were a substantial component of the assemblage at Manda, Unguja Ukuu, Tumbe, Bandarikuu, and Chibuene; only at Dakawa, Kimimba, and Shanga were bowl types significantly less than a quarter of the assemblage. These latter sites might be compared with Chami’s (1994) inland assemblages at Misasa, Mpiji, Kiwangwa, and Masuguru in central Tanzania, where bowls were always less than 20% of the assemblage. Fleisher and Wynne-Jones (2011:274) have argued that: such difference may have to do with differing consumption patterns and connections to Indian Ocean trade at these different types of sites. In comparing sites that were small village settlements during these centuries, such as Kimimba and Shanga, to those that were more demonstrably connected to Indian Ocean trade and traders, we can see that there were greater percentages of necked jars at the smaller sites and a related smaller percentage of bowls. The ratio of necked jar/bowls at Kimimba (81%/14%) was similar to that of Shanga (91%/6%), while the ratios at sites like Tumbe (66%/24%), Manda (48%/ 35%) and Unguja Ukuu (70%/28%) showed the presence of many more bowl forms.

Additionally, among ‘bowls’ was a subset of vessels decorated with red paint and graphite. Graphited red-burnished vessels have been reported previously at Manda (Chittick 1984:118–23), Kilwa (Chittick 1974:322–24), Unguja Ukuu (Juma 2004:91), Pemba (Fleisher 2003:248–49) and Shanga (Horton 1996: 253), and there are some indications that such decorative elements and forms are linked to developments in the Comoros where they comprise a much larger part of the ceramic assemblages (Wright 1984:35–40). However, they are unevenly distributed among the sites that were re-examined (Fleisher and Wynne-Jones 2011): although painted and graphite bowls were found in all the studied assemblages, they only form a substantial percentage of vessels at three sites: Manda (13%), Tumbe (14%), and Unguja Ukuu (15%). At these three sites, painted and graphited bowls were either very common or the dominant bowl class in the assemblage (at Manda 37%, Tumbe 60%, Unguja Ukuu 50%). In contrast, such bowls were only present in small numbers at other sites such as Dakawa (3%), Chibuene (6.6%), and Shanga (.04%). What do all these numbers mean? They suggest that certain ceramic vessel types were more common at sites that were well-recognized for their intensive long-­ distance trade connections. Wynne-Jones and I suggested that this uneven distribution of ceramics “seems to represent another coastal sphere of ceramic distribution found at many of the sites that were most closely linked to Indian Ocean trade at this time. As used in practice, they may indicate forms of consumption of nascent elites,

[email protected]

92

4  When Did Feasting Emerge on the Eastern African Coast? New Perspectives…

and thus be related to emerging notions of prestige” (2011: 275, emphasis added). I highlight the last sentence because it offers an important bridge to begin thinking about the implications of the word innovations I have presented, words related to this particular period of coastal history. I have previously highlighted ‘empowering’ feasts as a form that might be akin to those that were being practiced in the period just after AD 1000. And I placed emphasis on the competitive element of these feasts as I thought about the emergence of powerful people and places during these centuries (Fleisher 2010: 202). These were the centuries in which most Swahili towns either expanded considerably or were founded. Glassman’s description of eighteenth- and nineteenth-century feasts on the northern Tanzanian coast provides some evocative images of feasting practices from periods within which power was in flux. He describes how “feasting frequently took on an openly competitive form, as rival[s] sought to humiliate one another through aggressive exchanges of hospitality” (1995:58). Feasts were not simply acts to build social cohesion or belonging, they were contexts in which people could be challenged, ridiculed and humiliated. In the collection of words that were innovated during this period, I see resonances with these types of social actions. Most obviously, there are the words of disparagement—foolish, stupid—but there is also the word *-puluz- which means ‘be foolish’ but also ‘talk much, say little’ (Nurse and Hinnebusch 1993:294); this is a term that would cut to the quick in the verbal jousting that occurred at competitive feasts. The term describes a ­person who attempts to convince but fails. On the side of success, there are now more neutral terms for the types of discourse that might occur in a public feast (Nurse and Hinnebusch 1993:294): ‘chatting’ (*-zųngumz-) and to ‘be quick witted, sharp’ (*-kalamuk-). We can see in these new terms the possibility of two sides of a contested discourse, the type of jostling that would occur in a competitive feast. It is also important that word innovations during this period also mark the establishment of social distinctions (Nurse and Hinnebusch 1993:295)—words for ‘slave’ (*mtumwa) and ‘free born person’ (*mulungwana) offer insight into not only the exchange of people but also the hardening of social distinction. The innovation of the word *-Wez- to mean ‘be able’ and ‘be powerful’ (Nurse and Hinnebusch 1993:294) indicates that the power of certain people was attributed to their actions and their abilities. To be able to host and carry off a competitive feast was one way that individuals could become ‘empowered.’ The link between ability and power is extremely evocative when thinking about the possibility of the emergence of empowering feasting activities. I have previously described the distinction between empowering and patron-­ client feasts—this is the difference between a feast that has the ability to establish power and one that simply signifies or demonstrates it. I originally linked empowering feasts only to the period after AD 1000 because I saw in this period the social process of political emergence and consolidation; it is during the eleventh and twelfeth centuries that towns increased wealth through overseas exchange, but also built the first stone structures. And yet it is difficult to see the process of negotiation in these data, or the realities of failure that must have accompanied some of these empowering feasts. However, during the end of the first millennium there were

[email protected]

Conclusion

93

many clear failures: many of the most important settlements during this period were either greatly reduced or fully abandoned (Fleisher et al. 2015). For example, both Tumbe and Unguja Ukuu were largely abandoned after approximately 900–950 (Fleisher and LaViolette 2013; Juma 2004). In the century that followed, new towns were founded, often in close proximity to the old ones. And in these towns, although the ceramic assemblages showed remarkable continuities, there was also now evidence of intensified forms of consumption, as I described in 2010.

Conclusion What I have attempted to describe here is the process of working back and forth between the data of archaeology and historical linguistics, what Wylie (1989) has called “tacking.” In this effort, I have tried not to privilege one research domain over another, a problem that we identified in Chap. 1 with the different approaches to connecting archaeological and linguistic datasets, in particular those that we call ‘correlation’ and ‘common regional processes.’ The greatest limitation to my study has to do with issues of preservation and reconstruction. The issue of preservation, whether of words or things, can lead to the significant gaps in both datasets. The problems of preservation surely effect what can be reconstructed, but the choices made by archaeologists and historical linguistics also shape what gets reconstructed—what lexicons are chosen or what regions, sites, and contexts are excavated. In tacking between archaeological and linguistic data, I hope that I have shown how it might be possible to think more broadly about how the relationship between them might be generative of new questions and insights (sensu Wylie 1989; Stahl 2001). As described, my original study argued for the emergence of empowering feasts by working between data from the eleventh–twelveth centuries and that from the fifteenth century. The symbolic meaning of the bowls in the mosque at Chwaka became clearer as I was able to describe a biography of feasting practices stretching back to the eleventh century from the fifteenth century. This is what Wynne-Jones (2015; see also 2016) would refer to as a biography of practice (although I did not recognize it as such back in 2010). Although it was always possible to draw a line from the small numbers of red painted and graphited bowls in the ETT assemblages to those later post-AD 1000 assemblages where they dominate, I was hesitant to extend the argument for feasting back to sites dating to AD 700–950. This was partially due to the smaller numbers and smaller sizes of bowls in the pre-AD 1000 assemblages, and other types of data that seemed to suggest relative equality among the residents of even large villages in pre-AD 1000 settlements. For example, although there is evidence of intensive production of shell beads at Tumbe, the production itself appears to have been organized at the household scale, with relatively even participation across the site (Flexner et al. 2008; Fleisher and LaViolette 2013). I was also resistant to extending the practice back to the first millennium because I was convinced that it was in the post-AD 1000 period that competitive political acts

[email protected]

94

4  When Did Feasting Emerge on the Eastern African Coast? New Perspectives…

emerged—I believed that these were the centuries in which politics and power were negotiated and established, the foundation on which the political organizations of Swahili stonetowns were built. What the analysis here suggests is that empowering feasting practices might, in fact, extend deeper into the past, to the very beginnings of coastal settlement. The linguistic data has offered insights into past practices that were not available through archaeological data sets. Situating these practices within and among the archaeological data set of the AD 700–900 period offers a number of possible implications for how to think about feasting practice on the Swahili coast. First, the social distinctions that become so visible after AD 1000 may have already been in process by AD 800. Second, the archaeological data that indicates relative equality among the residents of pre-AD 1000 settlements might be masking emergent inequalities. It is clear that this discussion has created many more questions than answers, but it is through a working between the data sets of archaeology and linguistics—seeing their disjunctures as well as correlations—that new questions can be raised and, in time, possibly addressed through further work in these fields. Finally, I would argue that this chapter highlights that the biography of feasting practices suggests it as one of the most enduring social practices on the eastern African coast (WynneJones 2010). Investigating and documenting its historical development and the social practices that sustain it offer an important way to write a truly social history of the coast.

[email protected]

Chapter 5

Comments, Dialogue, and Supplemental Reading: Eastern African Coast

Commentary This is a bold narrative that brings linguistic evidence to bear on archaeological problems. The story told here is different from other writings by archaeologists incorporating linguistic data because you focus on words more than protolanguages, mining reconstructed vocabulary for the same hints about meaning and practice that inspire your reading of the ceramic record. My comments here will focus on three major points relating to dating, geography, and the wider range of historical data embedded in the words reconstructed to protolanguages. Dating and Correlations  The argument of this chapter turns on the discovery of a misfit between the dates of the emergence of a new kind of large open feasting bowl (some imported and others manufactured locally) and the dates of the emergence of a lexicon that describes intense verbal jousting, both of which are interpreted as evidence of intensifying social competition. More specifically, the emergence of large open bowls that were likely employed in public feasts comes several hundred years after the innovation of new words that describe the social censure of attempts to speak (powerfully, one presumes). The earlier innovation of such vocabulary inspired a rethinking of approximately contemporaneous ceramic assemblages dating to c. AD 700–950. Beyond the ubiquitous, well-studied ETT jars, you study a smaller quantity of small open bowls, both imported (white-glazed) and local (graphited red-burnished), which are the precursors to the larger feasting bowls that become common after c. AD 1000. The importance of dating to the arguments demands attention. Dating protolanguages is one of the most contentious matters in the application of the methods of comparative historical linguistics to the study of early African history. Debates rage about what glottochronology actually measures and whether it is even valid. I’ll leave aside debates about the validity of the glottochronology for now, as the

© The Author(s), under exclusive licence to Springer International Publishing AG, part of Springer Nature 2019 K. M. de Luna, J. B. Fleisher, Speaking with Substance, SpringerBriefs in Archaeology, https://doi.org/10.1007/978-3-319-91036-9_5

[email protected]

95

96

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

positions may not be irreconcilable (de Luna 2012a). Here, I’ll focus on what the dates developed from glottochronology offer. First, glottochronologically derived dates extrapolate the date of the divergence of a protolanguage from a median (or the central range of the bell curve of) a group of individual measurements (cognate rates) of degrees of similarity among pairs of languages hypothesized to belong to the group that developed from that protolanguage. It is important to realize, then, that dates developed from glottochronology tell us when protolanguages had begun to diverge into the next linguistic generation. Such dates are best represented by a range of cognate rates and, therefore, a range of glottochronologically derived dates because we are dating the beginning of a process of change from a dialect to a new language rather than a moment of time when the protolanguage had reached a (measureable) degree of internal differentiation rendering it no longer mutually intelligible across its speech community. And, of course, such a shift is a contingent process; dates should be treated as an approximation. If we return to Ehret’s glottochronologically derived dates, a range of 68–74% similarity for proto-Swahili indicates that Proto-Swahili began to diverge into the northern and southern speech clusters between late 8th and early 9th centuries (Ehret 2011:116). This date could be checked against the rates in Nurse and Hinnebusch (1993) which would need to be “cleaned” of skews from contact and isolation and then re-calculated, which may be what Ray did in his thesis (2014). The time period during which lexical innovations reconstructed to protolanguages were actually spoken—insofar as such heuristic devices indicate the “real” experiences of speakers—must be dated to the time period between the divergence that produced a protolanguage and that which dissolved it. In the case of Swahili, words reconstructed as proto-Swahili innovations were developed between the divergence of proto-Sabaki in the sixth or seventh century and the divergence of proto-Swahili around the late eighth to early ninth century. These particular dates place the Proto-Swahili innovations discussed here in an even earlier period than the small, open, white-glazed and graphite red-burnished open bowls associated with those terms above. Having identified these misfits, I want to emphasize that it is even more tenuous to apply glottochronology to the languages and dialects of the Swahili-Sabaki group than to other language groups in Africa because of the long history of contact and borrowing between some parts of the Swahili-Sabaki group. In addition to contact and borrowing, it seems that entire dialects may have been lost in recent centuries (for example, during moments of political competition between towns in the Lamu archipelago; Nurse and Hinnebusch 1993:8). As your own research reveals, processes like the dramatic demographic shifts on Pemba (LaViolette and Fleisher 2009) and the frequent collapse (and eclipse) of prominent towns suggests that similarly dramatic linguistic events, like the complete loss of dialects and languages, may have unfolded in the past. All of these changes in the linguistic record shape the accuracy of the classification and the dates developed from it. Other methods to date the divergences of Swahili rely on correlating pottery traditions and protolanguage speech communities in time and space (e.g. Ehret 2011:113–16; Nurse and Spear 1985). Like you, I am uncomfortable with this approach because it assumes that people who lived in the past treated both lan-

[email protected]

Commentary

97

guages and pots as markers of identity. Perhaps some did in some historical contexts, but folks living in the past did not share the same needs their researchers share—locating members of a recognizable community (speakers of a ­protolanguages, makers of a particular form of pottery) in time and space. They, too, were likely concerned with boundaries (and their transgression), but our historical subjects had far more information than we can discern about the contingent processes by which identities were made in specific, changing situations. I agree that you should be suspicious of this approach. Close overlaps in the historical geography of protolanguages and pottery traditions constitute an historical problem: did such communities come to use language and pottery as markers of identity and, if so, how and why did that practice of marking identity change over time? We don’t want to confuse the research methods by which we learn about the past with the historical problems we study. Like many of the scholars discussed in Chap. 1, some of the scholars you cite also use direct associations with the earliest archaeological remains of a practice, species, etc. to date protolanguages. For example, Ray (2014:67–8) claims that some linguists associate the earliest archaeological evidence for camel husbandry on the Lamu archipelago c. AD 800 with the adoption of a word for camel, a word that can be reconstructed as far back as proto-Swahili, in order to argue that proto-­ Swahili was spoken c. 800. Such one-to-one associations are problematic when they rely on a single association between the first word from something new (like camels) and the earliest evidence for it the archaeological record. As you well know, the earliest evidence of a new object or practice is always subject to revision in archaeology. We might, however, seek out a number of associations, preferably across many domains of life, from animal husbandry to ceramic decoration, cooking techniques, and hunting tools. These warnings actually raise a much more interesting question: between which kinds of vocabulary and material cultural remains do we expect to develop direct associations and why? To state the obvious, we will more likely be able to develop direct associations between the tools, agents, and objects of activities (like hunting antelope with spears in large communal hunts) than to develop direct associations in non-material domains of life like ridicule, praise, or verbal jousting. You searched for words about the social activities that you imagined were relevant to feasting rather than words about the act of feasting itself because you already assumed that feasts were sites of competition. In other words, you skipped an interpretative step that might have been a fruitful avenue of inquiry. The link between feasting and social competition, perhaps by means of the verbal jousting described in the words you discuss, is in evidence in the ethnographic record and the nineteenth-century historical record discussed by Glassman (1995). This gets to the role of ethnographic analogy in archaeological methods (as distinct from comparative ethnography undertaken by historians using linguistic evidence). My point is that the link between feasting and social competition is not yet in evidence in the earlier linguistic material and is inferred from the archaeological material. Words about feasting and communal eating—its verbs, its instruments and tools, its participants—might well have brought you into the domain of verbal competition, depending on their

[email protected]

98

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

derivations. To be sure, lexicons of other fairly common activities, like farming in east central Africa, spirit mediumship in the Great Lakes region, or hunting in south central Africa, were used to develop words about authority, ambition, and other aspirational politics (Gonzales 2009; Schoenbrun 1998; de Luna 2016). The lexicon of feasting itself might reveal similar patterns, which might, from there, link up to the words about ridicule, speech, and social competition that interest you here. Having said that, I know you were also limited by the vocabulary available in Nurse and Hinnebusch (1993) and that this is why you started with terms suggestive of social competition and speech. What does all this worry over dating mean for the story that you want to tell? As a result of an overlapping chronology for the invention of words describing socially sanctioned (or ridiculed) acts of speech and small, open bowls that are predecessors to later, larger open bowls associated (by what method, exactly—ethnographic analogy?) with feasting, you saw possible evidence of the kinds of social performances that might well have taken place in the social contexts in which food from the small open bowls was consumed. If those words were invented some centuries earlier, does the argument crumble? I think not, but it does shift the causal relationship between verbal competition and recognition and its social and material contexts. If the divergence of proto-Swahili dates to the late eighth or early ninth century, the kinds of social commentary on aspirational (and failed) speech for which you identified word innovations predate your archaeological evidence for the early fits and starts of town life on the coast. Rather than an argument about their co-­ production, we instead might wonder how the social and material contexts of such speech changed over time because we know the words were conserved—they remained vital aspects of the speech of later protolanguages. Perhaps the kinds of social competition illuminated through word histories were elaborated upon in the material record through new dining styles that brought people (strangers? kin? traders?) together around small but beautifully decorated open bowls that demonstrated the sophistication and wide-ranging trade connections of the host for the eager eye and (perhaps also) the sharp tongue of the discerning guest. Aspirational and empty speech were attempted and ridiculed in the streets between houses that now sat much more closely together—perhaps in the close quarters of new towns, there was much more need for honeyed commentary as well as better opportunities to listen in on one’s neighbor’s talk before turning to another neighbor to describe and judge its value. There is a story here of continuity, but it is a continuity that was undertaken in rapidly changing material and social contexts. The words invented by proto-­ Swahili speakers were powerful social and political tools that remained in use during those tumultuous years of the late first millennium as coastal residents experimented with other ways to claim or demonstrate their authority in the many towns that were attempted and abandoned as Swahili urban life was slowly built up in fits and starts of experimentation across a wide swath of the coastline. The successes and failures of attempts at creating new forms of urban living at the coasts may well have created the uneven geographies of contact and estrangement that eventually led to the divergence of proto-Swahili.

[email protected]

Commentary

99

The recognition that the dates associated with proto-Swahili in the publications cited refer to divergence and the significance of changing settlement patterns to the last centuries of the first millennium (the era of the forms of speech definitive of the northern and southern dialect clusters) necessarily brings us to the problem of space, both the choreography around objects and the distribution of objects across territories. Choreography and Geography  Given the significance of, first, the dating of proto-Swahili and, second, the geographical distribution of white-glaze and graphited red-burnished pottery to the argument: where do you think proto-Swahili was spoken? Do you posit that the geographical distribution of the speech community overlapped with the entire suite of sites associated with a late first millennium Swahili material culture? Do we have a historical geography of Northeast coastal, proto-Sabaki, and proto-Swahili that has not been influenced by our knowledge of the ceramic evidence? How does the revision to the dating of proto-Swahili speech described above change the nature of the geographies we associate with it or with the two clusters of speech, northern and southern, which developed out of the divergence of proto-Swahili? Finally, can we get more historical geographical evidence from linguistic data beyond the likely “homeland” of a protolanguage? Just as archaeologists manipulate the ceramic record of a region (in addition to other forms of evidence) to draw out the boundaries between communities who shared more in common with each other than they did with neighbors nearby, so, too, do historical linguists produce such heuristics in the form of protolanguages and their speech communities. And, as your work with the Coastal Ceramics Project demonstrates, archaeologists can also track particular kinds of jars, bowls, and techniques to demonstrate the uneven geographies and intensities of contact, wealth, and status. Historical linguists can also trouble the boundaries of protolanguages and speech communities to demonstrate their porous nature. Of most use to archaeologists, historical linguists can track the distribution of individual words as evidence of contact or the skews in cognate rates as evidence of much longer-standing, more intense connections (or lack thereof). Here, I want to highlight some of the ways that linguistic evidence might be brought to bear on the distributions you trace through the ceramic record. In a fascinating way, open bowls, large and small, imported and locally manufactured, stand as hubs of social interaction in your essay. From their uneven distribution to their round form, they carry all the markers of making insiders and outsiders. We can easily imagine the way that ownership could signify membership to an elite group, especially because the distribution of such bowls was so uneven across space. But even the very circular form of those early small, round bowls appeals to the imagination, for insiders could sit around such bowls, making eye contact, widening their circle to include latecomers who belong, turning their backs on outsiders, huddling their bodies down toward the bowl and toward each other to sup, blocking others’ ability to eavesdrop through the barrier formed by diners’ backs. As feasting bowls grew, was access to elite status democratized or did the social context of attempting to achieve such status become a more public event? I can imagine the

[email protected]

100

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

power of reconstructed words that describe the accouterments and bodily habits of such dining as they changed over time. But, in their absence, we can push instead on what we can learn by thinking about linguistic evidence through the lens of the geography you develop based on the distribution of white-glazed and graphited red-­ burnished small open bowls. I’ll begin by briefly summarizing your arguments. Between the seventh and mid-­ ninth centuries, most coastal (and hinterland?) communities used pottery best represented by the ubiquitous necked jars definitive of the Early Tana Tradition. Residents of most sites also made use of small open bowls, but such bowls made up a higher proportion of the ceramic assemblage at some sites, such as Manda (Lamu Archipelago), Unguja Ukuu (Zanzibar Island), Tumbe (Pemba Island), Bandarikuu (Pemba Island), and Chibuene (Mozambique coast). The sites where bowls were a substantial proportion of the ceramic assemblage are interpreted as places better connected to Indian Ocean trade in contrast to sites with a lower percentage of bowls, such as Dakawa (central Tanzanian coastal hinterland), Kimimba (Pemba Island), and Shanga (Lamu Archipelago), which were much smaller villages at the time and were less connected to Indian Ocean trade. Within the sites with a higher proportion of bowls, inhabitants of a few sites used bowls that were much finer. Potters brushed these bowls with a range of red paints and also made use of graphited decoration techniques. They were, it seems, crafting bowls that were similar to rarer white-glazed imported bowls. These sites include those on the Comoros Islands as well as Unguja Ukuu (60% of bowls were the finer wares), Tumbe (50%), Manda (37%), and, to a much lesser degree Chibuene (6.6%) and Dakawa (3%), and, barely of significance, Shanga (0.04%). Interestingly, with the exception of Manda, many sites with a higher proportion of open bowls within the assemblages you reanalyzed correspond with the known location of dialects of the southern cluster that broke off from proto-Swahili in the corresponding period. Leaving aside the very small percent attested at Shanga, we can make the same statement about the distribution of those very fine open bowls exhibiting red paints and graphite decoration. Might the processes that led to the establishment of the southern dialect cluster also account for the distribution of small open bowls? If so, how do we account for the odd place of Manda? Interestingly, in their rich linguistic study of the Swahili-Sabaki group, Nurse and Hinnebusch (1993:7) have this to say about the languages historically associated with the Lamu Archipelago more generally and Manda in particular: There is a question in the Lamu Archipelago of what is dialect, or simply accent... it is possible that historically there were more dialects in this area than today. During historical disputes between Lamu and Pate, the population of Manda Island was totally dispersed, and Manda speech is said by some local people to have been distinct from both Amu and Pate [two dialects of the Lamu Archipelago].

The ceramic record indicates great contact among some of the sites in regions associated with the early emergence of the southern dialect cluster. Were the residents of Manda in contact with speakers of the early iteration of southern Swahili dialects and did such contact shape how they spoke, leading eventually to the emergence of

[email protected]

Commentary

101

a unique Manda dialect that was lost as the population of the island dispersed? We cannot study this lost dialect to answer this question, but archaeologists are able to tease out further evidence of the possibly quite unique status of Manda with its possible stronger ties to the south. Similarly, we might well wonder about the ties far to the south, to the Comoros Islands and to Chibuene. Chibuene was an important early trade town in southern Africa with connections that stretched as far as the Tsodilo Hills in northwest Botswana at the same time that residents were using small open bowls and, less frequently, the finer wares more common at Unguja Ukuu or Tumbe. The languages associated with the histories of these regions do not belong to the Swahili group, which makes it much easier to track loanwords from the southern Swahili dialect cluster into Comoros and the southeastern Bantu languages associated with coastal Mozambique. Indeed, Nurse and Hinnebusch provide such evidence of contact and borrowing between the Comoros and the early cluster of the southern dialects (1993:133–45, 442, 484–85).1 But, they hypothesize that this contact was the result of adjacent residence on the coast in the late first millennium, after which speakers of Comorian languages emigrated to the islands (1993:512–13). This argument is based on the surprising number of shared innovations between Comorian languages and proto-Swahili and the southern dialect cluster, but a late migration to the islands is not the only explanation of such borrowing if there is other evidence for frequent contact between Comorian languages and both proto-Swahili and the dialect clusters that diverged from it. This is a problem that would benefit from a careful search for direct associations between the linguistic and archaeological records of the Comoros and different parts of the coast and a better understanding of the geography and chronology of trade between the islands and the mainland. On Innovation and Inheritance  Having discussed how linguistic evidence can speak to the common concerns of dating and geography shared by historical linguists and archaeologists, we can now consider what further information is embedded in the reconstructed words you discuss in the essay. The reconstructed words themselves certainly tell us something about the past, but they both contain and may be connected to far more information. This may well be like an archaeologist who has recovered a glass bead. On the one hand, the bead itself yields information about its production. Through analysis of the bead’s chemical composition, form, and so on, we can know about its provenance, the pyro-techniques used in its production, its likely age, and so forth. On the other hand, we also know far more about the bead when we consider the context in which it was recovered—the spaces and objects with which it was associated. Was it one among a cache of thousands of similar (or different) beads? Was deposition intentional—in a burial, perhaps— or unintentional—an object accidentally left behind in the busy day to day life of its wearer? Similarly, reconstructed words contain information about their cre Also compare the higher cognate rates between Comorian languages and those of northern dialects of Swahili to those between the Comorian languages and those of southern dialects of Swahili (Nurse and Hinnebusch 1993:274). 1

[email protected]

102

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

ation. The morphology of a word tells us about the process of derivation and this information is essential to understanding the lives of the speakers who invented the word. We can glean further historical information by considering reconstructed words as part of different kinds of groups of words that offer further context: lexicons, synonyms and antonyms, homophonic and polysemic clusters, other words derived from the same root, ancestral forms to the many attestations of the reconstruction in modern day languages and dialects, and so forth. In this final section, I’ll discuss a few of the ways that you can get more from the words you discuss in the essay. Turning to the gloss you offer for *-puluz-, we come to an important aspect in reporting reconstructions. Nurse and Hinnebusch offer the meaning to “be stupid” (1993:294), but you offer the meaning “be stupid, talk much, say little.” Two points are important here. The first is that the meaning “talk much, say little” is the meaning associated with the attestation of the verb in Mwiini and it was probably collected during fieldwork and is reported as it is a slightly different meaning than the one reconstructed to the proto-form *-puluz-. And this is not the only root where you make this minor error, but let me explain the implications using the example of *-puluz-. Nurse and Hinnebusch are suggesting that in the absence of other connotations of the verb in modern-day speech that are similar to the use of the verb by Mwiini speakers, the meaning is best rendered as “be stupid” but the semantic range was stretched to accommodate the meaning the verb came to have for Mwiini speakers, for whom quantities of meaningless speech was the epitome of stupidity. The link between speech and stupidity, in other words, is not in evidence in the other languages and, Nurse and Hinnebusch therefore hypothesize, was not part of the original meaning of the verb root *-puluz-. Now, they may well be wrong, but we have three ways to investigate that: first, we can think about how the verb was produced; second, we can seek out further attestations revealing links similar to those in Mwiini between the verb root and ineffectual attempts at speaking; third, through a more thorough reconstruction of words related to *-puluz-, whether synonyms, antonyms, polysemic clusters, the wider lexicon of speaking (well or poorly), and so forth. In that broader context, we might learn that public judgments about successes and failures at demonstrating knowledge were always rendered through speech and discussed through the lens of judgments of speech. I’ll return to this problem and this root again below. You are already exploiting all the words you can find in Nurse and Hinnebusch that relate to the social dynamics of recognition and judgment, particularly in the domain of speech. In other words, you are already thinking through the lens of a lexicon developed at a particular moment in time. Here, I want to shift to suggest some ways of thinking about the unique qualities of the reconstructions that interest you in order to better draw out the connections between them and also the social and historical dynamics of their moment of creation, for these dynamics were frozen in the elements used to create the words. In most cases, archaeologists will need to ask historical linguists about the possible derivations of reconstructions that interest them, but I’ll try to point out a few “tricks of the trade” in the examples below.

[email protected]

Commentary

103

I’ll begin with an interesting but fairly simple example. The innovation Nurse and Hinnebusch reconstruct as *-Wez- appears to be a semantic innovation. They reconstruct the meaning of this root as to “be able” and they find a potential source of this root in the Arabic words ´ezz, “power” and ´azza “be powerful” (1993:294). But, in the absence of any reconstructed tone (Swahili, as you know, is not tonal), we might similarly wonder if the older Bantu roots *-gèd- and *-gèdį- “to try” (but also in many languages near the Swahili coast as “to measure” and “to think”) might be the sources of a loan into proto-Swahili (Bastin et al. 2002, ID 1345; see also Guthrie 1967–1971, C.S. 797). I came to this idea because I work closely with a verb spoken of as kuweza in many Botatwe languages that derives from this same verb (Chap. 2, above; see also de Luna 2016) and the near phonology made me think about the possible connections of the root verb to Nurse and Hinnebusch’s reconstruction. The meaning Nurse and Hinnebusch reconstruct to *-Wez-, namely to “be able,” would seem to support as equally likely sources both this Bantu root and the Arabic root for “power.” To be sure, polyglots who spoke first millennium forms of both Arabic and Swahili might have heard both languages’ roots in play as they spoke the set of sounds, *-Wez-, associated with ability. We can imagine this word in use in many contexts, such as trade negotiations in markets, in commercial journeys inland, under the verandahs of older-styled homes or, some centuries later, in the outer and inner porches of the earliest stone houses in emerging coastal towns. In the contexts of trying for the sale and the purchase, of measuring (out of ship holds? jars with triangle incised decorations on their necks? a heavy purse?), of demonstrating the ability to complete the trade, we can imagine how one demonstrated being an able, powerful person in the uncertain, intensifying commercial world of the coast in the mid to late first millennium. Trying to be able may well have been a persistent state for many on the coast well into the second millennium, for the social and economic contexts of measuring and trying to be able were constantly in flux with the rise and collapse of commercial networks, industrial ateliers, and the material symbols of success and ability. This is, indeed, a powerful root for your story. When I saw your use of the reconstruction *-kalamuk-, I was struck by the fact that Nurse and Hinnebusch actually indicate the source root of this term, −kali, “sharp”, which derives from the early Bantu reconstruction *-kád-, to “be bitter; be sour; be sharp; be fierce,” easily looked up under the English gloss “sharp” in the massive online compendium of Bantu Lexical Reconstructions or, BLR-32 (Bastin et al. 2002, main 1657; see also Guthrie 1967–1971, C.S. 978). With your knowledge of KiSwahili and a quick skim through Thilo Schadeberg’s fabulous chapter on “derivation” in The Bantu Languages, a handbook some linguists affectionately call “the Bantu Bible”, you might recognize some of the morphology in play in the creation of *-kalamuk- (Schadeberg 2003). Here we see the root plus two verb extensions: the positional -am- and the separative –uk-. Thinking about the puzzle pieces brought together by the proto-Swahili speakers who developed this new verb 2  This database is available online: http://www.africamuseum.be/collections/browsecollections/ humansciences/blr

[email protected]

104

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

to talk about people able to “be quick witted” we see that innovative speakers were using the positional infix to fix or freeze the position of the root verb and the separative to emphasize the instransitivity of the verb, its ability to describe a state of being (rather than an action directed toward an object), and probably also the intensive and repetitive character of that state (developed from the many experiences that made it such a “state” of being) for the actor discussed with the new verb *-kalamuk-. In other words, to be “quick-witted” was to be sharp or fierce, not in a way directed at a particular goal or object, but in a sense that suggested an augmented sharpness that was iterative—or, frequently performed and observed by those describing the subject of *-kalamuk-. There is an element here of performance, observation, and judgment that calls up the very social contexts that inspired your writing about feasting. I also wondered about the root from which speakers created the similarly complicated verb *-zųngumz-, to which Nurse and Hinnebusch reconstruct the meaning “chat, etc.” (1993:294). I’d be curious about the kinds of activities to which “etc.” refers. But, I also wondered about the /z/ in the beginning of the term, especially as it precedes a high, closed, back vowel that was eventually lost in a sound change characterizing many Bantu languages. These Bantu languages (and protolanguages) inherited a reduction from the original 7-vowel system of early Bantu language history to a 5-vowel system. The loss of the two closed vowels impacted the pronunciation of certain kinds of constants in inherited word roots that had contained the now-lost closed vowels; this sound change is called Bantu spirantization. I thought that the /z/ at the beginning of the root might be an outcome of spirantization, but a quick glance through the phonology section of Nurse and Hinnebusch reveals that /z/ is not the outcome of any of the candidate consonants that would have been affected by the 7–5 vowel shift. Knowing that /z/ is often the outcome of other kinds of sound shifts beyond spirantization (and the outcome of spirantization in other languages), I wondered again about possible source roots of this verb. It would require a very careful analysis of neighboring languages and their diachronic phonology (particularly the outcome of spirantization in those neighboring languages), but I wonder if the source of this verb about chatting might be an older eastern Bantu verb, *-bų́ng-, “to wrap up” (Bastin et al. 2002, main 384; Guthrie 1967– 1971, C.S. 234). The verb stem is again combined with the positional extension paired this time with the causative to mean in a more literal translation something like to “cause to be fixed in a wrapped up position.” English similarly shares metaphors for speech that suggest that talk can have the effect of holding others. Speakers can enthrall (from Norse and carried into Old English with the meaning “slave”), or captivate, and a listener may be enraptured (from Latin raptus, “seized”). Although we could only test this very tentative derivation through the careful study of many more Swahili dialects and east African Bantu languages, it may be that the speech discussed with the proto-Swahili verb *-zųngumz- described the kind of talk that immobilized others, wrapped them up. Was this a powerful kind of talk? Did it become so banal as to connote something like the less meaningful English speech activity to “chat”? If so, what forms of social sanction changed a once powerful way

[email protected]

Commentary

105

of speaking—one that could hold the listener as though she was wrapped up (by obligation? by astonishment?)—into mere chit-chat? Was this talk rendered powerless by the emergence of new forms of speech (and their social contexts—public feasts, perhaps?) while the talk of less worthy speakers continued to be called *-zųngumz-? If this derivation holds, it indicates a loanword from hinterland or inland communities, rendering the transformation of this form of speech from enrapturing talk to mere chit-chat all the more powerful a piece of historical evidence. It suggests that Swahili speakers passed judgment on forms of speech held to be powerful in the hinterland. Given the semantic associations between witchcraft and binding or tying people, this loan may also shed light on the history of witchcraft. My last example is similarly constrained by the lack of the “raw” data I would need to do a proper analysis. But, once again, I was struck by the odd “shape” of the root *-puluz-, discussed above. Was the final /z/ again indicating a causative extension? I looked up the form *-pud- in the BLR-3 online. A root reconstructed to a fairly widespread group of east Bantu languages caught my eye: *-púdik- “hear; listen in silence [sic]” (Bastin et  al. 2002, main 2626; Guthrie 1967–1971, C.S. 1589a). The two roots have different verbal extensions and there is no reconstruction for *-pud-. But, we might try to infer a core meaning for the verb root that was changed through the addition of verbal extensions. The reconstructed verb root *-púdik- was created by adding –ik-, which is either the impositive or the homophonous neuter. The first connotes both the idea of causing the verb to happen but fixes it in a location… “to put (sth.) into some position” (Schadeberg 2003:74). In this case, the subject of the verb *-púdik- was caused to be (put) into the position of listening in silence. Or, with the neuter, the connotation is somewhat different. Indeed, the common association of the neuter extension with verbs describing experience leads me to think this the more likely extension. In any case, the neuter would suggest that the subject of the verb *-púdik- “is potentially or factually affected by the action expressed by the verb… no agent is implied, and it is typically impossible to express the agent” (Schadeberg 2003:75). Here, the verb *-púdik- describes the state of experience of the listener, who is the subject of the action verb—perhaps to “be audible, to make a sound, to speak”?—such that the subject is experiencing the audibility only because s/he is in a state of being affected by the audible sound. In contrast, *-puluz- may well be the causative. Here, the agent is front and center, for s/he is causing the action verb: s/he is the source of the sound that might create in another person the experience of hearing or listening in silence. In other words, *-puluz- seems to describe the action of the author of a kind of speech or sound-­ making that is incessant because others can only hear—they listen in silence. Here, then, I’ve posited a link between the Mwiini attestation of the verb, which carries the meaning of “talk much, say little” and the proto-Swahili meaning reconstructed by Nurse and Hinnebusch, to “be stupid.” The person who did *-puluz- was the cause, the source of a kind of speech that rendered others mute and exemplified being stupid. If further research bears out this etymology, you can feel far more confident in exploiting the connection between speech and the proto-Swahili root *-puluz-.

[email protected]

106

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

Having now demonstrated some ways that further analysis and research might push the linguistic data to do more for your argument, I’ll close with the suggestion that the adjustment I’ve proposed to the “fit” with dates now means that you can comb through Nurse and Hinnebusch’s reconstructions for the dialect clusters that emerged as proto-Swahili diverged in the late first millennium. You may find evidence of loans with Comorian languages that mirror the diffusion of graphite red-­ burnished small open bowls. In the southern cluster alone you will find many innovations that may well provide ideas about the social contexts of eating around such small, open bowls, including words for scorn, shame, the foolish, the wise, local chiefs, family members, madness, market (days), many words related to elephants and slaves (likely exports), a sweet partly fermented beer that might have been consumed at communal meals, and migration as well as the acts of wanting and electing. This is a rich corpus for the attempts, successes, and failures you identify on the Swahili coast in the last centuries of the first millennium.

Dialogue Question One Fleisher:  I appreciate the explanation about how time periods are related to reconstructed protolanguages and the difficult critique of how I have linked Proto-Swahili word innovations to the AD 800–1000 period. However, I’m wondering if you could talk more about how you came to reject my chronology? What I’m asking for is a bit more about the process of how you came to this critique. de Luna:  I began with an assumption we probably share: despite the problems associated with archaeological dating techniques, they are better than those associated with dating protolanguages. But your argument turned on the reliability of the dates associated with protolanguages, so I knew I would need to take a look at this issue. Non-specialists are often confused by the way that we report dates in our illustrations of classifications. When there is a date—even when it is marked as an approximate date for the divergence of a protolanguage—people often “read” this date as the approximate time when the protolanguage was spoken. Setting aside the question of whether protolanguages were ever spoken as such, I wanted to be sure that you (or your sources for the dates) had not made this mistake. Divergence dates are, as I mention, the period when a protolanguage begins to diverge or die out. To put a positive spin on it, they are the dates around which the next linguistic generation begins to emerge. Therefore, anything reconstructed to a protolanguage tells us about the time period between the divergence dates. Thus, protolanguages were “spoken” during the periods between the dates or date ranges assigned as the beginning of the divergence of a protolanguage.

[email protected]

Dialogue

107

Question Two Fleisher:  What your comments revealed is the confusion I ran into when trying to read between a number of different historical linguists and historians, namely the work of Nurse and Hinnesbusch, Spear, and Ehret. While I tried to work between these different authors, assuming that there was a common chronological thread that could be drawn out, it appears that their chronological reconstructions are incommensurable. Can you talk more about why this would be so? Aren’t they all working from the same data? de Luna:  What you may not have noticed is that your sources (Nurse and Hinnebusch 1993; Nurse and Spear 1985; Ray 2014; Spear 2000) use two different methods of dating. To make matters more confusing, they then sometimes cite each other. These two different dating methods actually rely on different datasets and they highlight a very important divide between historians using linguistic evidence to write about the past and linguists using linguistic evidence to write about the past. As discussed in the introduction, these two communities of scholars are often assumed to share the same methods and goals. For the most part, this is quite true. But, because we are in two different disciplines, we are also governed by very different standards of evidence of what constitutes a research question, of the delicate balance between ends and means, of comfort with different degrees of “knowability” in our evidence. One place to see some of the fault-lines is in debates about dating. This may have to do with the fact that historians, like archaeologists, work with absolute dates while linguists can think about language change over time without facing the imperative to ground their relative chronology of linguistic change in absolute calendar dates. Let me begin with glottochronology, which is a particularly contentious dating method. Most linguists reject this method (e.g. Dimmendaal 2011). You may have noticed that many of the Africanist historians whose scholarship you know make use of it, usually quite carefully, even warily. For example, Jan Vansina used the method in Paths in the Rainforests (1990), but rejected it in his later study of west central Africa (2004). Many of us fall somewhere in the middle, accepting the method in much the same way that we accept lexicostatistics as a shorthand method that produces a tentative classification that requires revision through more robust techniques (cf. de Luna 2012a, b, 2016; Ehret 2000, 2011; Gonzales 2009; Klieman 2003; Saidi 2010; Schoenbrun 1998; Stephens 2013). Like lexicostatistics, glottochronology can be revised or refined. Here, the second dating method comes to the fore: connecting the linguistic and archaeological records, which you do in this essay on the basis of overlapping dates, dates which you think are coming from the linguistic record. We have already discussed in the introduction the variety of ways historians and linguists try to connect to the archaeological record, often for the express purpose of dating protolanguages:

[email protected]

108

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

1. correlating an overlap in the approximate location and glottochronologically-­ derived dates of a protolanguage and the approximate geographical spread and 14 C dates of a material cultural tradition (e.g. Ehret 1998, 2011; Gonzales 2009; Klieman 2003; Saidi 2010; Schoenbrun 1998); 2. seeking commensurate archaeological evidence in a region in which a protolanguage was thought to be spoken for processes (like the adoption of intensive maize cultivation or the development of a new scale of social interaction) that seem to have commensurate scales with the scales of demographic or political change visible in the linguistic record (Heggarty and Beresford-Jones 2010; Schoenbrun 1998; Stephens 2013); 3. identifying the first archaeological evidence for a new technology for which one has been able to reconstruct a word within a particular protolanguage (like the wheel in proto-Indo-European) and using the date of that innovation to date and sometimes to locate the protolanguage (Anthony 2007; Vansina 2004). We can build off this method of direct association by reconstructing a larger number of words relating to a variety of domains of life (cattle keeping, hunting, cooking methods, house styles, etc.) and seeking as many direct associations as possible with archaeological evidence for the tools, activities, and  practices for which words have been reconstructed in the region where the protolanguage to which they have been reconstructed is thought to have been spoken (de Luna 2012b; Kirch and Green 2001). In most of these approaches, the approximate location of the protolanguage has been identified through linguistic methods by considering the area of greatest linguistic diversity and principle of least moves from the current locations of extant members of a language family back to the likeliest location of their common ancestral language. In some cases, this location is very difficult to discern (e.g. proto-Indo-European). Let’s return to your Swahili case study. You might now notice that Nurse and Hinnebusch (1993) are reluctant to use glottochronology as a dating method. Indeed, they know well the difficulties of using lexicostatistics to classify coastal languages because, historically, there has been a lot of contact and borrowing in these languages. This convergence can skew the measurements of relatedness (the “cognate rates” of lexicostatistics) and, therefore, the dates developed through glottochronology. Nurse and Hinnebusch (1993:21–22) describe the kinds of dating methods used by historical linguistics in other world regions and their applicability to the coastal languages in their study: In principle, linguists wishing to lay out the external [absolute as opposed to relative chronologies?] history of a language may draw on certain well-known resources, including written records in that language, the traditions of the people about their history, and the accounts of outsiders about language and people. In practice, for NEC [Northeast Coastal] and Sabaki as groupings, all these are nonexistent, and for Swahili they are of limited value... In sum, if we had to rely on these three sorts of sources, our knowledge of the history of the Swahili and their language would be very restricted, and our knowledge of Sabaki or NEC nonexistent.

[email protected]

Dialogue

109

Instead, they argue (1993:22) that a “fourth method allows us better insights”: By juxtaposing, on the one hand, the geographical area covered by extant language groups with, on the other, the geographical extent of various historical cultures revealed by archaeology, we can sometimes see an almost one-to-one correlation between a language area and a culture area.

Armed with this method (#1 in the list above), Nurse and Hinnebusch (1993:22) make a number of claims about the dates of the nodes of their classification. For example, they explain that archaeologists have observed a widespread “distinctive and homogeneous culture” along the coast, which developed organically out of its earliest roots, which can be dated to roughly 800. There is an almost one-to-one geographical relationship between the sites carrying this culture and the area where Swahili is today spoken and, as far as we can know, has always been spoken. Thus we take it that this geographical area has long been the Swahili-speaking area and that communities likely to have been using an early form of Swahili can be tentatively attributed to its earliest period, that is, around 800.

Later, they observe (1993:23) a date for Proto-Northeast Coastal “based on the same juxtaposition of archaeological and linguistic distributions.” More specifically: the first Bantu-speaking communities are likely to have entered East Africa at a point near southwest Lake Victoria in the closing centuries B.C., and it would have required some time for the ancestors of the early PNEC to move toward the east coast. The earliest archaeological attestations of Early Iron Age people on or near the coast are from the 1st century AD.

At the end of the volume (1993:490), they state more clearly that: [t]here is no direct general historical evidence for the existence, the date, or the possible locale of a PNEC community...Archaeological evidence suggests that Bantu communities first entered East Africa in the closing centuries B.C. and had reached the coast by at least the 1st century A.D. Although there is no positive evidence that these communities included the PNEC, we assume that PNEC speakers were in fact among them.

Nurse and Hinnebusch also explain that they “assume an equation of ‘Early Iron Age’ and ‘Bantu-speaking’” (1993:23). In these discussions of dating, Nurse and Hinnebusch (1993:21) explain that they are summarizing the dating schema developed in Nurse and Spear’s The Swahili (1985) where this same approach to correlation is used to date protolanguages. But, it is worth considering something else that is going on with their dating: they are proposing “best guesses” that relate the degree of similarity they see between nodes in their classification as a way of thinking about the speed of linguistic change (in other words, the timing of divergences). And they are doing so without using lexicostatistics and glottochronology.3 Nurse and Hinnebusch (1993:22) note that  Paul Heggarty (2007:323–25; see also Heggarty and Beresford-Jones 2010) does something very similar in his work by comparing his lexicostatistical rates with those of Romance languages in order to estimate the pace of language change and, therefore, the dates of divergences. Like most linguists, he rejects glottochronology on the grounds that it presumes a standard rate of change. Then, with what feels to this reader like a slight of hand, he then applies the rate of change of Romance languages (dated through documents) to languages of the Andes, thereby normalizing the pace of change in Romance languages by applying them to the pace of language change in Andean language history. 3

[email protected]

110

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

c­ orrelation between the geographies of Swahili speech and “Swahili” material culture (Tana/TIW, according to Nurse and Spear [1985]) suggests something of a diaspora because these geographies cover such a long portion of the coast, from Somalia to central Mozambique. “This implies a diaspora and the breakup of the PSW [Proto-­ Swahili] community, which would have existed for a period up to that date” (Nurse and Hinnebusch, 1993:22; emphasis mine). This point is important, because they are suggesting that such a dispersed geography would have defied the best attempts to maintain a mutually-intelligible form of proto-Swahili speech. I take this to imply that they are now dating the beginning of the divergence of PSW to c. AD 800. But I am not certain from the text. Nurse and Hinnebusch (1993:22–3) continue: If we then consider that the extent of the linguistic differences between Swahili and the other Sabaki languages hardly exceed that between the various Swahili coastal dialects, and if we also consider that the amount of innovation clearly attributable to PSW is small...then it seems likely that the PSW period was short, and that PSA [Proto-Sabaki] preceded it by only a short time, perhaps a century or two. In linguistic terms, PSW and PSA must have been quite similar, differing only in minor respects. Moreover, since the linguistic differences between Sabaki and NEC languages are relatively small, PNEC is unlikely to have anteceded PSA by more than a few centuries. An approximate date around or slightly later than 1 A.D. would seem reasonable for PNEC, perhaps five hundred years later for PSA, shortly after that for PSW.

I should be clear that we are not told if these dates are for the emergence of these protolanguages or their divergences, but I think in this case that they are implying the former rather than the latter. Later, Spear (2000) uses these dates (which, you will recall, rely on correlations elaborated in Nurse and Spear [1985]) as a starting point to explore the Sabaki “gap.” Again, correlations based on new ceramic typologies seem to fill the gap. Spear finds useful the arguments of archaeologists who reconcile Kwale/Early Iron Ware with Tana/TIW by demonstrating that the latter developed out of the former. For Spear, this helps to explain the gap between the association of PNEC with Kwale and of PSW and Tana/TIW. I suspect that there is some debate among archaeologists about the claim that Tana/TIW develops out of Kwale/EIW. Here we see how the desire to “fit” the evidence together might be getting in the way of the wonderful historical evidence embedded in misfits, a point that I bring up in my commentary above. Let me pivot here to your other sources. You also cite Ray (2014) who cites Ehret (2011). Ray reanalyzes some of Nurse and Hinnebusch’s data and also uses the dates for the three major stages of the NEC family developed by Ehret.4 So, let me focus on Ehret, who also uses the cognate rates in Nurse and Hinnebusch (1993). But, he has ‘cleaned’ the cognate rates, taking out skewed rates resulting from intense contact and borrowing between languages or from unusually conservative languages (Ehret 2011:113–4). Such skewed cognate rates will impact the dates  Ray reclassifies Sabaki using the cognate rates in Nurse and Hinnebusch (1993) but there are some inconsistencies. For example, he proposes a cognate range of 54–63% for proto-Sabaki. He also has a subgroup of Sabaki, which he calls Middle Sabaki and for which he offers a cognate range of 57–59, but this range is nearly identical to and falls within the range for the parent group, Sabaki (Ray 2014:383–4). This is highly unlikely and might be a typo or some other kind of error. 4

[email protected]

Dialogue

111

developed through glottochronology for the divergence of the protolanguage common to the languages of the subgroup. Ehret has explained this method elsewhere (1971), but you may find the explanation in a more recent publication (2000) and the application of Ehret’s method in Klieman’s Ph.D. thesis (1998) to be more helpful. Ehret dates the cognate ranges for different stages of NEC by applying the method of glottochronology to this “processed” data (2011:105–08, 113–16). He then compares the glottochronologically-derived dates for the beginning of the divergence of different protolanguages to the dates of major ceramic traditions. I’ve elaborated on these dates in my commentary. Allow me to recall that Ehret finds correlations between, for example, PNEC and the earliest Tana sites and PSA and the spread of Tana to the Pangani and Lamu region (Ehret 2011:113–16; see also Ehret 1998). For Ehret, the continued spread of Tana is related to the expansion and eventual divergence of Proto-Swahili. Now, you’ve probably already noticed that this does not agree with Nurse and Spear (1985), Nurse and Hinnebusch (1993), or Spear (2000), for whom Kwale is associated with PNEC. To return to your question, I hope it is clear that different authors are dating the divergences of protolanguages through different means, but a fit with the archaeology is almost always part of the approach. Having exhaustively reviewed the literature you cite, I find myself equally confused. If I were to undertake a project on the coast, I would begin anew with the cognate counts, clean them of skews, and then develop dates using glottochronology. I would then reconstruct words for the kinds of activities that seem to define different archaeological periods in different areas of the coast (for example, the shell bead-making you studied in Northern Pemba). I would look for as many direct associations between the archaeological and linguistic evidence as possible (method #3 above), but I would also seek misfits for what they can teach us about processes of boundary-crossing. Of course, many Swahili dialects have now been lost and it may be too late to accurately reconstruct enough vocabulary to the most recent nodes of the NEC classification. Fleisher:  There are a couple of things that I would like to know more about. One is about the diaspora/breakup of Proto-Swahili and your suggestion that scholars are arguing that “such a dispersed geography would have defied the best attempts to maintain a mutually-intelligible form of proto-Swahili speech.” This leaves me somewhat confused, since what follows proto-Swahili are the northern and southern dialects. Would these all have been relatively mutually-intelligible? I guess what I’m asking is how much would the ‘dispersed geography’ have led to mutual-­ unintelligibility given the significant and ongoing connections up and down the coastal corridor? I’m also very surprised at how common it is for historians and historical linguists to try and match protolanguages with particular ceramic types. I thought we were somewhat past that type of interpretation. The most problematic aspects of this type of work, especially with attempted correlations between Tana/TIW and a particular protolanguage is that we don’t really have any idea how (or even if) this ceramic style ‘spreads.’ The dating of these sites does not really allow for much sense of

[email protected]

112

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

which sites have temporal priority, despite claims by Chami (1994) that the sites on the southern Tanzanian coast are earliest. de Luna:  Your question about mutual-intelligibility cuts to the quick of one of the challenges of classification. I read Nurse and Hinnebusch as saying that the spread of proto-Swahili along the coast created a dialect chain effect—a situation in which speech at the far ends of that expansive speech community changed so as to be mutually unintelligible while those at the center were more similar. But this doesn’t take into account the movement up and down that coast that you see in the archaeological data. Mobility is “built into” the methods and assumptions through which we think about language differentiation over time, though we rarely study it as a topic in its own right. Nurse and Hinnebusch’s data could be read for evidence of this mobility—of links, contacts. Indeed they allude to some examples, which I’ve pointed out in my commentary above. But, to return to your question: the “moment” (which would not have been a “moment” at all, but a long process) at which point the northern cluster and the southern cluster were no longer mutually intelligible matters because it is part of the process of the divergence of proto-Swahili into the Northern and Southern dialect clusters. But it is a process. Your question then gets at an important point: to some degree, a Swahili speaker can understand far northern and far southern dialects, just as a Spanish speaker can often understand Portuguese. We have here a divergence that is still in process at the time of the study (and is even now reversing with the standardization of Swahili). Nurse and Hinnebusch recognize enough difference (in phonology and lexis, for example) to identify the Northern and Southern dialect clusters as a unique stage that follows the divergence of proto-Swahili in the historical development of the NEC languages. The line between a dialect and a language is rather arbitrary (as is the very status of a language as a bounded, measurable entity, some socio-linguists would argue; see Makoni and Pennycook 2005 for an interesting overview). The dialect clusters (North and South) would have certainly been internally mutually intelligible, but Nurse and Hinnebusch are suggesting that proto-Swahili would have undergone so much internal differentiation that it no longer meet the somewhat arbitrary definition of a coherent protolanguage because the dialects of the Northern cluster and those of the Southern cluster had much more in common internally than either clusters’ dialects had in common with dialects of the other cluster, including those dialects in the old center that belonged to opposite clusters and were in closest geographic proximity. I think historians find it useful to draw on the archaeological record and the correlation of pottery styles and speech communities as a quick way to make these links. But, we are not trained in the methods by which ceramic typologies are developed and I’m not sure that we are in a position to be able to assess which ceramic styles do or do not “spread.” This may be similar to how difficult the debates about Bantu classifications may seem to archaeologists. And it speaks to a central issue in interdisciplinary work: specialists critique non-specialists’ incorrect use of data but rarely find ways to teach each other appropriate uses of such data and the debates surrounding the methods by which it is produced.

[email protected]

Dialogue

113

Question Three Fleisher:  Your comments on geography are useful, especially if I begin to think about the connections to Northern and Southern dialects of Swahili and the uneven distribution of certain ceramic forms and decorations. It does get me thinking about some of the potential issues with the ceramic assemblages, in particular, sample size issues. Looking back on the paper now, after our initial conversation, can you better see these issues? Is there any equivalent in historical linguistics—are there sample issues that one must consider when basing historical interpretations on linguistics? de Luna:  I must admit that as I began to plot out the locations for the assemblages that you and Stephanie Wynne-Jones had reanalyzed that I was concerned that most sherds seemed to be from either very famous sites in the north and south or from sites where you have conducted fieldwork. When we talked about this database, you explained how difficult it was to gain access to the ceramic assemblages of additional sites and how few ceramics you analyzed as representative of some of the sites you included. As a result, it is quite possible that the patterns I suggest are not viable at all because your ceramic analysis is, as it turns out, quite limited, though provocative. Moreover, after looking more closely at the dating methods used by the sources you have at your disposal, I am not comfortable with the uncertainty surrounding proposed dates for PNEC, PSA, PSW, or the northern and southern clusters. All told, I would not be surprised if the geographies I tried to tease out in my commentary from what I knew from the linguistic dates and what conclusions you shared from your ceramic database prove to be incorrect. But putting the two records side-by-side raises questions that could be explored. This brings us back to the problem of the geographic scales of our different forms of historical evidence, especially how archaeology scales up to the regional level of linguistics and how linguistics scales down to the site or assemblage scale of archaeology. This is a problem we need to think about much more carefully if archaeologists and historical linguists continue use each other’s conclusions. It is probably one of the reasons that correlation has been such a common method of linking the two datasets. What kinds of evidence might we each bring to bear on the different scales that characterize our evidence? Fleisher:  Although I think there are ways that archaeology could offer larger scales (there is most definitely a lack of high quality, large-scale survey projects in ­sub-­Saharan Africa), this doesn’t concern me in the same way as you. This is partially because archaeologists themselves have thought about how and when you can bring the regional together with the local and the limitations and pitfalls in such data sources. I also think that by simply understanding the different scales at which our data work, we can avoid some of the pitfalls and confusions that have plagued previous work.

[email protected]

114

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

de Luna:  Might this have something to do with the way archaeologists and linguists have thought about our respective regions? For example, the ceramic record of the Swahili region is far more robust. But, also, the coast has long been thought of as a regional and material culture unit in a way that central Zambia has not. Indeed, you (and others) seem to push back against this, trying to disaggregate the larger unit of the Swahili coast to point out the different smaller regional histories within it. It might be helpful, then, to learn more about how archaeologists scale up to the regional level. What debates animated the development of methods to scale up? Fleisher:  I think that today archaeologists most often start at the regional; archaeological survey is not just about finding sites to excavate, but rather assessing the full range of site types and hopefully understanding the settlement system. With this data in hand, archaeologists can than choose particular, representative places to investigate through excavations and other methods, places that will help to answer questions posed. There are extensive debates on how to organize (and carry out effectively) archaeological surveys. This literature might be tedious for a non-­ archaeologist, but it emphasizes the clear articulation of a survey strategy that considers issues that hopefully strengthen archaeological samples (Plog et  al. 1978; Banning 2002). Pottery is obviously an important set of data, but any type of data can provide information on regional scale interactions. In your area, the problem is not with the pottery itself but rather with how it has been analyzed and reported. With so many different analyses that come to so many different interpretations, it is difficult to know what is truly similar and/or different—it is one of the reasons I avoided trying to work with it in my commentary chapter. de Luna:  It may well be that “archaeologists almost always start at the regional,” but it is not clear to a non-specialist how survey and excavation can, together, through sampling the areas to survey and excavate, lead to an understanding of the settlement pattern and history of an area commensurate with a ‘region’ as defined by language groups. When I think about the relevant region for the eastern Botatwe branch of the fairly small Botatwe family (just thirteen languages), the geographic setting of the story of that branch covers an area of about 50,000 square miles, which is about the size of U.S. state of Alabama. Is this the scale of ‘region’ about which archaeologists are producing data in other areas of the world? Fleisher:  The short answer is ‘no.’ Archaeological ‘regions’ will never be commensurate with the regions defined by language groups. However, I think an important step forward in collaborations between archaeologists and historical linguists would be a clear-eyed assessment of what types of regions could or should frame their research questions, what is possible, and how these different regional approaches can be used supplementally.

[email protected]

Dialogue

115

Question Four Fleisher:  Finally, I want to ask you about the word innovations. I find your exploration of the four word roots fascinating, and I can see a number of techniques that I could use to make similar types of interpretations. However, I was struck in this section by something that has occurred to me before when reading historical linguistic analysis written by historians: although the reconstruction of word histories is based on linguistic methods, there is also a kind of art to this process, a historical sensibility that defies technique alone. The drawing out of historical narrative from word histories requires, I think, a particular playfulness and interpretation. Do you agree with this? This is perhaps something that comes with a long familiarity with how words change and the way meanings adhere to them, but I’m wondering if you could reflect on this. What I’m saying is that, even if I were able to do the technical work that you did in the final section, I’m not sure I would get to the meaning-­ making you did. de Luna:  Archaeologists also play with data and interpretation, sometimes in narrative form and often with great creativity. Rod McIntosh’s historical imaginings come to mind (1998). But, your idea of an historians’ sensibility probably operates on a number of levels relating to the fact that historians working with historical linguistic data inhabit a peculiar zone between several fundamental intra- and interdisciplinary divides. I’ll elaborate with a few impressions, but these are by no means conventions. As I first read this question, I thought about the position of history astride the divide between the humanities on the one hand and the social sciences on the other. Of course this is a false divide, but historians are more often thought to bridge it than, for example, linguists. Narrative is very important to historians and this emphasis on narrative might be part of what is in play in your comment (cf. Schoenbrun 2012). On one level, perhaps you are seeing this divide of the humanities and social sciences in the difference between how historians and linguists use word histories to write about the past. We use the same methods (for the most part), but we diverge when it comes to what kinds of vocabulary interest us because, understandably, we have different goals. Let me speak in generalizations, if only to make the point. To my mind, historians want to put recognizable people—individuals with desires, foibles, and mistakes—at the heart of stories that take unexpected turns, narratives imbued with the contingencies and surprising outcomes of clearly motivated decisions or completely benign, even banal actions. Although this is an over-­ generalization, I feel that linguists working with historical linguistic data seem to work to solve particular problems—or to get as close to solving them as is possible: what was the subsistence system of the earliest Bantu communities? What is the productivity of particular verbal affixes in the earliest Bantu protolanguages? These

[email protected]

116

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

problems seem to turn on getting the evidence right to get the past right. It seems to me to be a more factual approach to history, a point discussed in Chap. 1. Historians deal in facts and causal relationships, to be sure, but we also peddle in historical possibilities, imaginings, lost opportunities, and failures. We privilege understanding what life was like in the past as well as explaining how or why change occurred because we see these as related efforts. We are not obliged to test models, isolate “driving” causes, or refine theories of change as much as we seek to understand the full range of human experience, which is necessarily both subjective and about subjectivity. While many historians are deeply engaged with studying many individual choices in order to explain the multiple and unexpected causes of larger patterns of change modeled by other disciplines, we also have the freedom to study lost causes and fringe communities for what they can teach us about a particular moment in time. Therefore, I think most historians would reject the idea that we will ever get the story entirely “right” or that there is even a single story to tell. The narrative sensibility you see in historians’ use of this data, then, may be related to historians’ interest in stories over models and the relationship between historical argument and historical description. In addition to the importance of narrative to humanistic research, historians using linguistic evidence work in ways that that are different from the majority of their colleagues in the discipline of history. For example, we work in periods earlier than most of our colleagues, for most historians work on the last 200 years. And, among those who share our focus on earlier periods with us, historians using linguistic evidence stand out in three more ways. In most contexts of African history, we do not have access to the traditional tools of the historians’ trade: documents. It is now much more common in the discipline of history to work with a great many more types of historical sources, of course, but most historians do so alongside information gleaned from the documentary archive. And, unlike archaeologists, we work with evidence that is indirect, as discussed in Chap. 1. Finally, our evidence speaks to three scales of historical narrative—in terms of chronology, geography, and sociality—that are much larger than those of most of our fellow scholars. Each of these differences, I think, shape how an historian working with linguistic evidence brings multiple sensitivities to bear on knowing the past: a sensitivity to contingency, to the humanity of our subjects, and to the limits of our indirect, large-­ scale evidence and the possibilities it offers to know the past in very different ways than those available with more conventional sources. This is a challenge, but it also allows for informed historical imagination, as you have suggested, and this is vital for scaling our stories back down to the human experience. I can’t know that attendees at a feast sat in a circle and ate around one of the bowls you recovered on the coast, but there is a good chance that they did so. The particular position of historians using linguistic evidence vis à vis other historians also highlights concerns and challenges shared with archaeologists. For example, with respect to the challenge of studying earlier periods, I have found linguists (and some historians of modern Africa) to be less comfortable when I reconstruct vocabulary relating to emotions or belief than many archaeologists, for whom even the most tenuous evidence of non-material life is welcome. Anyone

[email protected]

Dialogue

117

who works in the past (including linguists working on language history) necessarily works at one step removed from easily verifiable data (even if we acknowledge the gaps between researchers of the present and their interlocutors). The gap widens, depending on how far into the past you work, what the archives and other sources are like for your period of study, and how well-represented the communities you study are within your sources. This is a challenge with which archaeologists interested in non-material life can empathize because they, too, work with scant archives that speak only indirectly to the research questions, even if the character of the indirectness is qualitatively different than that of linguistic evidence for the deep past. Like archaeologists, historians are comfortable with certain kinds of “unknowability” because we accept that we are always working with an incomplete archive and that future historians will revise the stories we reconstruct. But, post-processual archaeologists and historians using linguistic evidence share a further sensitivity to the degrees of knowability involved in studying the past. It is worth restating that the quick questions I ask with and of the data in the Swahili case are akin to the kind of information you can get from a couple of STPs. Those attempts at play are teasers, openings for further research that will require the full suite of data collection and analysis that is masked by the narrative product of my “historical sensibility.” The narrative sensibility that you appreciate might distract from the kinds of information that other scholars want before or alongside such story-telling; perhaps it is more easily accepted or appreciated by a colleague who also works on the scant datasets of earlier pasts. Fleisher:  Of course we’ve talked much in the past about the difference between historians and linguists who use linguistic evidence, but I was fascinated to hear about how a historian might ‘play’ with a word history and thus generate some hypotheses to be tested through other investigations and research. There is a corollary to this, I think, in archaeology, as we will often talk through interpretations at the trowel’s edge (Hodder and Berggren 2003) and record these observations and interpretations on our forms and in notebooks. Such ideas are always subject to evaluation, however, as we process our finds and examine stratigraphy more fully. I appreciate the discussion of where historical writing differs from that of archaeology. Even in drafting this book, we have discussed how historians’ privileging of ‘narrative’ over ‘the data and methods’ serves to push the latter below the footnote line while archaeologists perhaps spend a bit too much time worrying about the data and methods and can thus lose sight of the narrative. I wonder how much of these writing conventions serve to make it difficult to read and digest each other’s material? I am also fascinated with your discussion of the distinctions you draw between the work you do and other historical linguists, and especially the contrast you make between “getting it right” versus offering “possibilities, imaginings, lost opportunities and failures.” The contrast between work that emphasizes certainty and a completion of sorts—“I have determined the subsistence pattern of this society”— and that that probes contingency is incredibly important. As you note, it is of course susceptible to challenges of “being wrong” but I think we are both comfortable with

[email protected]

118

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

that prospect, if it means going beyond banal description (that can, of course, be equally wrong). I feel very strongly about the necessity of exploring failures, successes, and abeyances; the built-in emphasis on continuity in the stratigraphic readings of archaeologists often works against this. My hope is that my work on feasting in eastern Africa can help to capture some of the way that such social processes not only opened possibilities (‘the emergence of social complexity’), but also provided a path to spectacular failures as well as abeyances of practices that were no longer efficacious. One of my key takeaways from this discussion is that my efforts to work with the historical linguistic data were perhaps too focused on what we have been calling direct association—I was hoping for some confirmation or context in which to place the evidence of feasting that I had previously interpreted through material assemblages. As you have rightly pointed out, the words that I assembled are, in fact, a lexicon on verbal skill that are not necessarily associated with feasting in the period on which they were innovated. Taking stock of this lexicon in its particular time period could provide insights into how people living in first millennium AD towns and villages were experimenting with ways to distinguish themselves.

 upplemental Reading: Speaking Well and the Development S of Feasting One result of de Luna’s engagement with Fleisher’s work is that it challenges him to consider the somewhat narrow view that he used to approach the linguistic data, a view that privileges the connection between ceramic forms, decorative elements, and the social practice of feasting. When Fleisher initially approached the linguistic record, he did so in the hopes of finding a direct association with the social practices of feasting that he had reconstructed through archaeological data (certain types of vessels, cooking practices). But what he found, instead, was a lexicon that seemed to resonate with those social practices, albeit in earlier centuries. The words that were found to be useful (those of disparagement and competition) offered a possible way to confirm earlier interpretations and to challenge the chronological primacy accorded to the post AD 1000 period. And yet, what we have learned is to not to meld so closely together the word innovations and material innovations. Fleisher’s efforts to see in the word innovations of verbal jousting and forms of disparagement earlier versions of feasting may, in fact, be serving to read (inappropriately) the material record of feasting from the post-AD 1000 period into earlier times. Attempting to find ceramics that linked directly to the lexicon of verbal jousting might have had the effect of obscuring what was happening in these centuries. Instead, the insights from the commentary and dialogue offer a way of considering a broader set of practices and values that involve the strategies of skilled speech. These reveal the development of particular forms of talk, action, and ways of being that emerged before AD 1000, a lexicon that might offer key insights into how people were attempting to manage and live in their rapidly changing world through

[email protected]

Supplemental Reading: Speaking Well and the Development of Feasting

119

new material engagements and practices. This supplemental reading provides a greater sense of the contingencies of practices in Swahili villages and nascent towns during the late first millennium, and the way that new practices were being attempted, routinized, and codified. Rather than attempting to push back the practice of feasting into the late first millennium AD (through an association with ceramics, lexicon, and a knowledge of post-AD 1000 feasting), it is possible to work supplementally between historical linguistic reconstructions and archaeology. With the few word reconstructions by Nurse and Hinnebusch (1993) contextualized with provisional interpretations of diachronic semantics by de Luna and the broader sense of a lexicon concerning verbal jousting, we can begin to see how forms of talk, practices, abilities, and powers were developing during this transformative time, offering insight into new types of settlements, trade relationships, and consumption evident in the archaeological record. The emergence of these lexical resources in the pre­AD 1000 world thus helps to situate the material transformations in the post-AD 1000 world as feasting becomes a new arena within which the skill of speaking well would have become particularly salient. In this way, feasting becomes an iteration or a strategy in a much larger regional story about the politics of speaking well on the east African coast. Forms of Coastal Authority, in Words and Settlements  In order to begin thinking about the pre-AD 1000 period, we need to understand different forms of authority that are believed to have developed along the coastal corridor after AD 1000. A number of researchers have argued that two forms of authority and city governance developed along the coast (Horton and Middleton 2000; Allen 1993; LaViolette and Fleisher 2005), distinguishing between ‘town governments’ and ‘towns with kings’ (Horton and Middleton 2000:157–72; see also Allen 1993), the former a more corporate form and the latter more autocratic and hierarchical. LaViolette and Fleisher (2005:340) see these distinctions represented in the physical layout of the town itself: Towns with large, dense stone sectors surrounded by small wattle-and-daub neighborhoods may have been structured around a more horizontally differentiated power base, with numerous leaders (merchant, religious, political) vying for position of ‘first among equals.’ In contrast, towns with few elite structures surrounded by extensive wattle-and-daub may have been organized around a small-scale, ranked hierarchy, whose leader ruled through custom and consent.

Such distinctions are found along different parts of the eastern African coast: towns with dense stone sectors tend to be found along the central and northern Kenyan coast while those without are found along the Tanzanian coast and points further south (although there are exceptions, e.g., Songo Mnara). There is thus a growing sense of the distinctions between Swahili towns, as our knowledge of particular towns and regions come into greater focus (Fleisher 2018). Part of Fleisher’s work with LaViolette on Pemba Island (e.g., LaViolette and Fleisher 2005, 2009) has been to document the particular city forms there, and to recognize the way they differ from other coastal towns. In research on Chwaka, in northern Pemba Island, its town plan (monumental coral mosques and tombs, but

[email protected]

120

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

housing primarily of earth and thatch) is contrasted with that of contemporary towns in the Lamu archipelago which were built extensively of coral architecture. In Chwaka there is a closer affinity to the more centralized power exhibited by a town like Kilwa (which is also dominated by monumental mosques and palaces rather than more general stone houses) and thus resonances of the distinctions between the ‘town governments’ in the Lamu archipelago and the ‘towns with kings’ at sites from Tanzania south. It is in these southern towns that Fleisher forged his arguments about the social practice of feasting, and how it became a crucial context in which power was articulated and negotiated. The linguistic data offers a possible way of correlating northern and southern coastal towns with the development of northern and southern dialect clusters. De Luna (personal comm.) commented that certain word innovations in northern and southern dialect (as reconstructed by Nurse and Hinnebusch 1993) may suggest that southern dialect communities were more hierarchically organized, and that northern dialect communities contained a richer culture of imitation and performance. Southern dialect word innovations include *šaha (‘local chief’), *-teua (elect/ appoint), *mjuzi (wise person), *-jivuna (boast), and a range of words for older family members (*baba: ‘father,’ *babu: ‘grandfather’); northern dialect word innovations include words related to performance (*-taba: ‘imitate,’ *-tamka: ‘pronounce’), the lexicon of shame and jealousy, other words (like *-randa: ‘resemble’) that could indicate a greater amount of verbal governance. This comparison is tentative, partial, and reliant on the words that Nurse and Hinnebusch (1993) generated. And yet it points to how a supplemental approach can be generative; through this supplemental approach, we seek to bring into closer proximity the broad characterizations of periods generated by independent historical records, attending to resonances without expecting full confirmation across records or even precise overlaps in specific practices or objects. What is important about these types of relationships between historical linguistic data, town plans, and material classes is that it allows for a supplemental approach to a topic that has long been dominated by twentieth-century ethnographic data; both Horton and Middleton (2000) and Allen (1993) build their arguments from such data. Verbal Jousting, Innovation, and Contexts of Social Interaction  In the Swahili case study, words were chosen that were thought to be directly associated with the practices of feasting, those that might speak to qualities and abilities that would be key to individuals participating in feasts, but also the qualities of those that might describe individuals who failed or had been vanquished in competitive feast events. It should be remembered, however, that the words available to us are those that were reconstructed by Nurse and Hinnebusch (1993) in the course of their historical linguistic study, and thus represents a sample generated for other research questions. If historical linguistic reconstruction were focused on our particular questions and issues, we could investigate a rich and extensive lexicon related to these issues. Through de Luna’s commentary, however, we can begin to understand how the words available to us provide a lexicon of the qualities, strategies, and abilities of speaking well as they were developing in the pre-AD 1000 coastal world.

[email protected]

Supplemental Reading: Speaking Well and the Development of Feasting

121

The emergence of these words (*-kalamuk-; *-Wez-; *-puluz-; *-zųngumz-) during the late first millennium would have occurred in settlements that were becoming more connected to people and places from off the coast, with visitors from the ports of the Indian Ocean rim. At this time, there were hundreds (and more likely thousands) of small settlements stretching from the continental interior to the coast and islands. Most of these were small farming and/or fishing communities, but they were part of a large, interconnected eastern African world, marked by common use of Early Tana Tradition pottery. Archaeological research in inland areas makes clear that, at this time, the coast was firmly part of this broader connection of continental settlements (Kusimba and Kusimba 2005; Helm 2000; Walz 2010). However, there is evidence that during this period, there were growing distinctions at certain coastal settlements, particularly those on the islands, which mark the increased interaction with these coastal residents and overseas traders. Sites like Manda, Tumbe, and Unguja Ukuu attest to this, with higher rates of imported pottery than other similarly dated sites and, as discussed in Chap. 4, larger numbers of red painted and graphited bowls. This is also, notably, the period when we have the some of the earliest possible evidence of Islamic practice in certain settlements (like Shanga). Archaeologists have struggled with how to think about social distinctions during this period; in Fleisher’s research on Pemba Island, he has tried to use production and household evidence to interpret what social distinctions might have been in place. With this data, he argued that a site like Tumbe was a ‘large village’ in which there was evidence of increased forms of production (shell beads) but no indications of social hierarchy. The word *-kalamuk- allows us to think ourselves into settlements like Tumbe. The reconstruction of *-kalamuk- suggests that a lexicon was developing among proto-Swahili speakers for individuals who were thought to be quick witted, with an element that such attributes was akin to being “sharp or fierce.” De Luna notes that “to be quick witted was to be sharp or fierce, not in a way directed at a particular goal or object, but in a sense that suggested an augmented sharpness that was iterative—or, frequently performed and observed by those describing the subject of *-kalamuk-.” One can imagine new contexts in these settlements in which to be quick-witted and fierce was a growing asset, hosting and interacting with foreign merchants, in which new forms of social gathering may have been developed. Such social settings might be those in which the small red painted and graphited bowls were used, perhaps in the serving of food to visitors as they hosted them. This was surely a remarkable time, one where new types of people were coming to coastal settlements, coastal residents initiated forms of trade and production, and new forms of religious practices were carried out and observed. We can think of many new forms of iterative behaviors that would involve “performance, observation, and judgment:” visitors conducting Islamic prayers; residents beginning to produce items beyond the needs of the community; households hosting guests from afar. All of these would have been contexts in which new forms of ‘quick witted and sharp people’ would have been observed and noted. This discussion also might indicate new contexts in which performance and observation were becoming more common, in settlements that were denser with

[email protected]

122

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

people. As we have noted, the most prominent places on the coast during these centuries had larger and denser populations than others—Tumbe, for example, may have sprawled over 20 hectares, while an average farming village covered one hectare or less. This was a significant change for eastern African settlements, and thus places where people were learning to live with larger numbers and a greater diversity of people. Rituals in which observation was important can be associated with the first mosques, where people would be seen traveling to and from the mosque, as well as within it. We have also posited that the form of decoration on bowls in the eleventh century AD—decorated with graphite on the interior surface—might indicate a particular proxemics of feasting. Fleisher (2010:213) has argued that interior decorations on bowls used in public consumption would have emphasized the act of consumption itself, as the decoration is slowly revealed through the course of the feast. Similarly, de Luna’s discussion of *-Wez-, which Nurse and Hinnebusch gloss as “be able, be powerful” helps to put this word into action by suggesting its use in context, capturing beautifully the uncertainty and possibilities of the time in which it emerged: trade negotiations in markets, in commercial journeys inland, under the verandahs of older-­ styled homes or, some centuries later, in the outer and inner porches of the earliest stone houses in emerging coastal towns. In the contexts of trying for the sale and the purchase, of measuring (out of ship holds? jars with triangle incised decorations on their necks? a heavy purse?), of demonstrating the ability to complete the trade, we can imagine how one demonstrated being an able, powerful person in the uncertain, intensifying commercial world of the coast in the mid to late first millennium.

In these new types of densely-settled towns, with new forms of contact, trade, and social interaction, we can see the way that a new language of ability, trying, and power became a central or even essential topic of conversation. There is a sharp contrast to *-kalamuk- and *-Wez- in words such as *-puluzand *-zųngumz- which indicate the wrong kinds of speech and performance. As de Luna explains: “*-puluz- seems to describe the action of the author of a kind of speech or sound-making that is incessant because others can only hear—they listen in silence…[t]he person who did *-puluz- was the cause, the source of a kind of speech that rendered others mute and exemplified being stupid.” In this case, the development of this quality—*-puluz—represents a form of engagement that was increasingly seen as foolish, when it was most necessary to learn how to engage and be quick-witted, rather than just droning on incessantly. Such actions would have been learned in practice, in these growing large villages, as engagement and interconnections were valued over pronouncements. In the archaeology of the Swahili coast, researchers have often focused on reconstructing when and how social hierarchies emerged on the coast—what periods, through what resources, and in what places (Wright 1993; Kusimba 1999; Horton and Middleton 2000). Yet, in all of this research, little attention has been paid to the discursive contexts of such changes, and how qualities of people that emerge in one period might become important to strategies of power in another. In this case, we are drawn to the idea that some reconstructed words might be providing an important

[email protected]

Supplemental Reading: Speaking Well and the Development of Feasting

123

foundation for the types of personal qualities that were needed for certain kinds of individuals to emerge as powerful in later centuries, and those qualities that became associated with foolishness and increasingly marginalized positions. Reconstructing a possible word history for *-zųngumz- (which Nurse and Hinnebusch (1993) gloss as ‘chat’), de Luna suggests that it might have developed from a word meaning “to wrap up,” and thus possibly “described the kind of talk that immobilized others, wrapped them up.” She goes on to ask: “Was this a powerful kind of talk? Did it become so banal as to connote something like the less meaningful English speech activity to ‘chat’?…was this talk rendered powerless by the emergence of new forms of speech (and their social contexts—public feasts, perhaps?) while the talk of less worthy speakers continued to be called *-zųngumz-?” Such questions and reconstructions point to the possible changing nature of forms of talk—from those that once ‘wrapped up’ listeners to that which was now mere ‘chat.’ Such changing meanings alert us to the ways that certain forms of practice that were once enthralling or novel, when they become routinized, may become less powerful and ‘immobilizing.’ In both this word and the last, one can see the emergence of new meanings through iterative practice. And this should force archaeologists to think about how their material remains may indicate the endurance of practices (see Wynne-Jones 2016), but that the meaning of them can change dramatically. Such a shift, we have argued, can be seen in later centuries in Swahili towns in the shift from empowering to patron role feasts—from feasts that were competitive and constitutive of power to those in which power was simply represented. In both, the serving of food was key, and the material remains may be similar, but the emergence of patron role feasts—put on by those in power—likely had the effect of making empowering feasts either obsolete or ‘banal.’ There is some archaeological evidence of this, in that large bowls become a common feature of all ceramic assemblages during the period when the most institutionalized forms of power emerge (twelfth to fifteenth centuries), suggesting that these forms of consumption were now commonplace across all walks of life. We can now begin to see glimpses of the qualities and resources that kinds of individuals began to develop in large villages of the pre-AD 1000 period, and the role that material objects may have played in new forms of talk and action. Working supplementally, the convergence of dialect divergences (northern and southern), forms of governance (‘town governments’ and ‘towns with kings’), and the differential distribution of particular ceramics (red-painted and graphited bowls) provides the basis for a new interpretation of the pre-AD 1000 period and the way this shaped the significant changes that followed in the second millennium AD. In closing, we return to the regional, coastal scale, but now focused on the pre­AD 1000 period. De Luna prompts us to think about linkages between parts of the coast, highlighting the dialect differences in northern and southern Swahili. She posits a possible correlation between the presence of the fine small painted and graphited open bowls and the southern dialect cluster (SD). This observation raises a number of interesting questions: de Luna asks if the processes that “led to the establishment of the southern dialect cluster also account for the distribution of small open bowls” and, if so, how do we account for the similarity of ceramics at

[email protected]

124

5  Comments, Dialogue, and Supplemental Reading: Eastern African Coast

Manda, a site firmly in the northern dialect cluster? She also raises questions about the languages that make up the Comorian cluster and notes the interesting relationship that these have with both proto-Swahili and the southern dialect cluster. Although not discussed fully in Chap. 4, the ceramic assemblages of eighth- and ninth-century Comoro Island sites are of great interest here. These sites contain red paint/burnished and graphited bowls (Wright 1984) in quantities that exceed those at the sites studied by the Coastal Ceramics Project, while versions of other forms and decoration (jars with incised decorations) are quite different in the size and elaboration of the decoration. Wright (1984:55–6) notes the presence of red-­ burnished graphited bowls at other coastal sites and suggests this indicates connections between them. Does the distribution of these ceramics, predominantly on sites from Pemba, Zanzibar, Kilwa, and the Comoros, indicate a correlation between this archaeological marker and the boundaries of the southern dialect cluster? The language data say ‘no,’ since Comorian by this time has emerged as a distinct language. This evidence thus suggests ceramic evidence that crosses language boundaries, a challenge to the common way that ceramics, language, and ‘culture’ are equated. Although contact between those speaking southern dialect Swahili and Comorian is not surprising, the degree of interaction in terms of common practices requires additional investigation that can be tackled with renewed work on ceramics and other material objects, as well as research on loanwords that were shared between them.

 fterthoughts on Dating and Methods: Linguistic A and Material Culture Change One of the most surprising things that we learned in this process was what we would call the lack of fit—or discrepancy—between the rates of change between the archaeological and linguistic data. By this, we mean it is extremely challenging to link the archaeological and linguistic records for the period AD 1000–1500 because the material record goes through a number of significant transformations within that period while innovations are offered only for the divergence of northern and southern dialects. For example, there are clear distinctions in the material record that can be associated with the following periods: AD 1000–1100, AD 1100– 1300, and AD 1300–1500. In each of these time periods, there are transformations in classes of material objects—ceramics, imported goods, architectural styles and production-­related materials—and yet all of these are covered by the innovations linked to northern and southern dialects. Because of this, there are no ways of positing tidy links between word innovations and some of the most dramatic changes on the coast (the development of stone-built architecture, the shifting trade patterns from the Persian Gulf to the Red Sea, increasing thread production/decreasing shell bead production).

[email protected]

Afterthoughts on Dating and Methods: Linguistic and Material Culture Change

125

A couple of examples from the historical linguistic innovations can help to illustrate this: a number of innovations seem to offer the promise of providing useful fodder for direct associations between archaeological and linguistic developments including *kiungu (Northern Dialect, ‘house of more than one storey’), *thanu (northern dialect, ‘potter’s kiln’), *buru (northern dialect, ‘maize’), *(i)duru (northern dialect, ‘stockade’), *usanga (southern dialect, ‘bead’), and *ulezi (southern dialect, ‘millet’). Taking the first word, *kiungu: this is surely a reference to houses made of coral rag, and we know that they are first built after the twelfth century AD, and are likely built of two stories shortly after that, but only in a few places. This direct association does offer some confirmation of the chronological linkage between northern dialect and the period between AD 1000–1500, but beyond this the linkage there is not much to be learned, especially because it was during these centuries themselves that coral architecture emerges, develops, and expands from mosques and tombs, to houses and other structures. This lack of fit can be seen again in the innovation of the word *buru for maize, likely innovated at the very end of this chronological period, with the intrusion of the Portuguese at the end of the fifthteenth century AD. Other innovations, such as *usanga and *ulezi are challenging in other ways, since there is archaeological evidence of beads and millet from the very earliest periods of Swahili settlement, in the seventh century AD. Frustratingly, these examples offer the possibility of direct associations, but serve to highlight the lack of fit between these datasets as they exist today: if Northern and Southern Dialects began to emerge by the ninth or tenth centuries AD, many of these innovations are associated with a time centuries after that divergence. As de Luna reminds us, however, they offer the promise of future research in which more robust lexicons could be reconstructed on particular subjects, especially those that might resonate with archaeological records.

[email protected]

Chapter 6

Conclusion

In introducing this book, we set out four goals: (1) teach non-specialists some of the basic principles of our methods as we engaged in the process of teaching each other; (2) expose the process of working with another specialist with similar goals, but a vastly different method and source-base, namely archaeology and historical linguistics; (3) explore common ways archaeological and linguistic evidence have been used together and the underlying assumptions of such efforts; and finally (4) highlight how such interdisciplinary work in Africa offers a contribution beyond the continent. We found in the process of writing and, especially so in revising, that we had to reconcile the kinds of stories we wanted to tell about the African past with debates shaping the common strategies used in interdisciplinary scholarship. We came to recognize that the kinds of stories we sought focused on the lived histories of people, which stood in stark contrast to the kinds of histories reconstructed with similar methods for other world regions. Unexpectedly, this led to a new, supplemental approach to using linguistic and material evidence to know the past. By way of opening an ending to our conversation, this conclusion summarizes some of the key lessons we learned in the process of writing this book. Here we address what we think are common challenges in crafting interdisciplinary narratives: training, dating, preservation, and scale. We then return to the supplemental approach that developed out of the revision process, identifying common elements in our supplemental readings. We include a set of steps researchers might consider, though we do not presume this is the only way to work supplementally with this particular combination of records. And we reflect on the opportunities we think the supplemental approach opens for our two disciplines in our common effort to know the lived history of early African communities.

© The Author(s), under exclusive licence to Springer International Publishing AG, part of Springer Nature 2019 K. M. de Luna, J. B. Fleisher, Speaking with Substance, SpringerBriefs in Archaeology, https://doi.org/10.1007/978-3-319-91036-9_6

[email protected]

127

128

6 Conclusion

Learning the Methods and Reading the Evidence We learned many lessons from the exercise of writing this book. Perhaps the most important was predictable: we, as scholars, suffered from similar kinds of challenges in using each other’s data. We certainly “caught” each other misusing data or drawing conclusions that would not be accepted as valid in the discipline from which the data was taken. We each wanted to use the other discipline’s data with nuance that would allow for a rich historical narrative of the sort produced by specialists. But we lacked that ineffable instinct of knowing how to handle a type of data that comes with close training, experience, and practice. What, then, would make us better consumers of each other’s data? How could we do interdisciplinary work better? Training is an important answer to these questions. We need to teach not only the results of interdisciplinary research, but also the process. We are sure that many of the issues we raised in our commentaries and dialogues are, in fact, common pitfalls and problems. Training can come in many forms and this volume is our modest contribution, as one among many other creative possibilities. Professional organizations have a particularly important place, we think, in providing workshops for interdisciplinary research for both graduate students and faculty. Although training is important, collaboration is even better. Historians tend to work in isolation, but a team-based approach to historical research with specialists in all data classes is the best way to do interdisciplinary research that combines data from highly specialized methods. Archaeologists work in teams, but most members of those teams are archaeological specialists (ceramicists, zooarchaeologists, geoarchaeologists) and not colleagues from other disciplines, like history or linguistics. This is certainly a direction that hard sciences funding models are moving in North America, the UK, and Europe. The KongoKing project directed by Koen Bostoen is a wonderful model that makes full use of the funding, research, and publication possibilities of the hard sciences model in sub-Saharan Africa. The project assembled a team from disciplines including archaeology, history, and linguistics, but also anthropology, art history, GIS analysis, and paleoenvironmental studies in order to understand the origins and early history of the Kongo Kingdom (www.kongoking.org). Our commentaries and dialogues also reveal that our very distinct data suffers from similar kinds of challenges, often in ways we did not expect. We share common problems with the limitations and, at times, uncertainty of our dating methods and resulting chronologies. We were both struck by the difficulties of different technical methods of determining dates. But even fairly simple matters like what reported dates actually mean were not as straightforward as we had expected. In both dialogues, we noted problems with the way that date ranges were reported and used, and how other scholars often collapse or narrow these ranges to suit particular research questions and interpretations. As outsiders dealing with an unfamiliar set of data and interpretations, it proved difficult to assess the quality of the dating, and adequately see possible issues and concerns.

[email protected]

Toward a Supplemental Approach to Material and Speech

129

Perhaps surprising to us both, we faced similar challenges with respect to preservation. The particular impact of preservation on the data is something that is rarely considered by non-specialists who use the evidence produced in other disciplines. In focusing on what data had been produced in each other’s fields, we lacked the ability to see what was missing and how those absences shape the data at hand. But, as we learned through our dialogues, it is one of the most important issues that a non-specialist must consider, as it has the potential to render data sets problematic for the type of social reconstructions we seek. Finally, the issue of scale came up frequently. Scale matters because it impacts how our data sets can be linked together. Other researchers combining archaeology and linguistics have grappled with the issue of scale, often arguing that there must be a general correspondence between the scale of language reconstructions and material culture patterns. We are now less concerned with having fully “matched” scales of data. However, we realize that there is a significant difference to the directionality of scaling. In terms of geographical scale, archaeology tends to work up from artifacts to regions (with contexts and sites in between) while historical linguistics works down from larger language areas to slightly more restricted regions of protolanguages. In terms of the social scales to which our evidence speaks, archaeologists seek out sample sizes that allow them to identify broader patterns of social practice, while the data of language already holds such broader patterns. Historical linguists struggle to scale down to the experiences of smaller social groups like households, individuals and the narrower events and episodes contributing to the broader processes reflected in vocabulary.

Toward a Supplemental Approach to Material and Speech Through a close reading of the documentary and material records of Banda, Ann Stahl (2001) demonstrates sensitivity to the silences and mentions resulting from the creation, curation, and reading of the historical records she deploys and the forms of historicity they illuminate. At the heart of her supplemental approach is the ability to keep historical records and the methods and disciplines that sustain them in productive tension to recover lived history in Banda. Our efforts build off these insights, but the supplemental approach to material and language evidence explored here necessarily engages with a different configuration of historicities and a set of tensions around scale, intentionality, preservation, and curation that ensnare the interdisciplinary scholar seeking to fully resolve, rather than exploit, those tensions. When applied to historical evidence embedded in material and speech, the supplemental approach advanced here exploits connections between two key discipline-­ specific methodological concepts that enable particular modes of contextualized thinking: the assemblage and the lexicon. Assemblages and lexicons help specialists recognize the social, material, and conceptual contexts of what people did in the past and, thereby, begin to apprehend what such actions meant to those who undertook them. In what follows, we discuss how engagement with assemblages and

[email protected]

130

6 Conclusion

lexicons opened up new interpretive possibilities for our research, focusing on assemblages in the case of south central Africa and lexicon for the Swahili case study, since the addition of these were crucial to our supplemental readings and reinterpretations. In archaeology, assemblages are related groups of artifacts associated with particular features and contexts. The analysis of these materials and recognition of their association in particular assemblages allows archaeologists to interpret the process and practices that led to their creation. Because they often involve collections of materials in context, such interpretations can move beyond the more functional interpretations that might be made from individual artifacts. In the south central African example, de Luna worked with assemblages in a couple of different ways. First, she worked with site assemblages of faunal materials from across the Batoka Plateau to understand changes in hunting strategies, butchery practices, and shifts from wild to domestic meat consumption. In a sense, she had thus constructed an assemblage of archaeological materials that could be associated with the regional scale of her word reconstructions. What was lost in this act, however, was the more fine-grained contextual data in which these artifacts were found (rather than just looking at which site and when, this requires asking where were they found within the site). Her use of smelting assemblages, however, captured the contextual data in this way—by charting the changing geography of smelting (from village to bush), she drew on not only the collection of materials that allowed for the interpretation of smelting, but also its physical context to draw out the meaning of these changing practices. In de Luna’s supplemental reading on south central Africa, focusing on the relationship between lexicon and assemblage forced further consideration of specific species buried beneath floors and their role in the process of constituting the households for which the houses were constructed. In this reinterpretation, we can see de Luna putting faunal data in particular contexts—in the floor deposits of houses— and thereby gaining greater interpretive potential from these well-defined assemblages. Although de Luna already had developed reconstructions for just these species based on their unique gendered form in the linguistic record, their location in the context of the pit beneath the house constituted an example of the process by which hunted animals were put into use and provided insight as to why hunting was meaningful to the admirers who spoke of hunters’ fame: some hunted animals had the power to make households and extend lineages. Here, the power of the language evidence to predict the species that would be most relevant to a space gendered feminine is a startling affirmation of the possibility of reconstructing the meaning of protoforms to the speakers who used them. Without the association between antelope species whose names were gendered feminine in the linguistic record and the pit beneath a house floor, de Luna could not make the leap between the reproductive and social practices of these antelope species and the changing practices of marriage and kinship undertaken by families in the Kafue period. In this way, the very specificity of the assemblage offered an iteration of a broader, more general process of demographic change described in the linguistic record and new insights into the meaning of hunters’ pursuits.

[email protected]

Toward a Supplemental Approach to Material and Speech

131

When thinking through lexicons, scholars assemble collections of words or lexemes that are related through meaning (e.g. synonyms [shared meaning], antonyms [opposite meanings], homonyms [near meanings], etc.) or through an association with the same practice, though probing the distinction between meaning and practice is part of a lexicon’s power. In this way, words associated with each other link the ideas and materials they name, extending our understanding of the broader social, material, and conceptual contexts in which communities of the past acted on the world. In the Swahili example, the lexicon of feasting—words for the actions, sites, accouterments and foods, and the participants and their relative responsibilities—was absent from the published linguistic record. Instead, Fleisher mined the available linguistic archive, intuitively extracting for analysis a lexicon that seemed to describe the sorts of communication and verbal jousting he could imagine of feasters (Chap. 3). In her commentary, de Luna proposed preliminary analysis of some of those terms’ derivations and morphology, incorporating into the lexicon of verbal jousting the older meanings on which it was built. This work is the domain of specialists, of course, and only accessible through collaboration, but it highlights the fact that lexicons emerge and change over time, a point also seen in the lexicon of hunting explored in Chap. 2. Even when the full suite of associations visible to an historical linguist are not available, it is powerful for a non-specialist to think about the development of a lexicon—how particular kinds of action and thought are brought into association and how those associations change over time (when such data is available, as it is on the Swahili coast). In the supplemental reading, the proto-Swahili lexicon of verbal competition is contextualized in the broader history of the lexicon of skilled speech, with feasting proposed as an eventual site for performances and negotiations of the value of good speaking. Fully aware of the limited sample size of the archive at hand, it is possible that earlier strategies to acquire skill in powerful speech in the NEC period were few or were replaced by innovations in the proto-Swahili period. The multiple words invented at this time for skilled and unskilled speech constitutes the innovation of a lexicon that developed in the context of emerging intercontinental trade in the proto-­ Swahili period. The archaeological record helps us understand why speaking was an increasingly important skill worthy of a new lexicon because it illuminates the sorts of new contexts in which speech was used and judged: concentrated settlement patterns (perhaps rife with eavesdropping and neighborly competition but enriched with larger audiences) and cross-cultural contacts (to hinterland and transoceanic communities—the latter perhaps seeking to make themselves understood with new, peculiar forms of speech learned by skilled coastal speakers). The fate of this lexicon was a story of durability from the proto-Swahili period forward: these words were remembered and their meanings were fairly stable, according to the glosses reported by Nurse and Hinnebusch (1993). It is meaningful that the lexicon persisted and was not later replaced. It was not a foregone conclusion that ambitious coastal traders would crystalize the proto-Swahili lexicon of verbal jousting; it took strategies like feasting, which concentrated control over acquiring, staging, practicing, and performing skilled speech, to curtail change in the ideas and practices of which the lexicon spoke and the aspirations and competitions it enabled. Eventually, even feasting

[email protected]

132

6 Conclusion

became banal, reflective of but no longer constitutive of such aspirations and competitions. From a vibrant flurry of innovation, proto-Swahili lexicons of verbal jousting stilled and settled back into the micro-politics of neighborly life in well-established Swahili towns, now governed by an established elite. Linking our best efforts to know how people worked on the world through speech and materials by bringing together assemblages and lexicons should allow certain incommensurabilities between our data sets (such as scale) to remain unresolved, thereby yielding insights impossible without holding these bodies of evidence in supplemental tension. We see examples of this in the supplemental readings we craft at the end of our dialogs. In the case of the Swahili coast, Fleisher was initially confused by the lexicon of verbal competitions because they seemed to be associated with an earlier period. But this lexicon forced Fleisher to look anew at the pre­AD 1000 material, and allowed him to imagine the micro-politics of language in settlements of this period. Because archaeologists work with material remains that are the result of practices, it is remarkable to have access to word innovations that refer to the creation of new ideas and practices. This lexicon made clear the silences that existed within the archaeological record: the material record of a site like Tumbe includes a rich assemblage of ceramics and other finds, but contrasts sharply with the monumental, permanent forms of construction (stone houses, mosques, tombs) found in later towns. And yet the lexicon of verbal competition from the earlier period alerts us to the types of debates and conversations that were happening in earlier towns like Tumbe. Such specific, and enduring verbal practices are not discoverable in the archaeological record, which thus effectively silences them, but they are revealed through the linguistic archive. For archaeologists, this offers the remarkable potential of exploring historical continuities between periods that are often regarded as either discontinuous or ‘transformational.’ In the case of south central Africa, focusing on the relationship between lexicon and assemblage allowed de Luna to make use of “oddities” of the archaeological archive (Bernault 2015): the only two known Iron Age floors built over the top of intentionally-crafted pits filled with animal bones in the region between the middle Kafue and middle Zambezi (or, at least, the only two so far recovered). These two assemblages are silenced by traditional archaeological analysis as being too small a sample to constitute evidence of broader social practice. After all, only two houses and quite likely a few generations of a single family deployed this strategy. But these oddities can be contextualized in the regional histories of changing hunting, marriage, and kinship practices illuminated by linguistic evidence. And the insights of the linguistic record take on new meaning when, for example, the reproductive habits of specific species gendered feminine are recovered in association with the material context in and with which households were created. These spatially-­specific, temporary, even familial strategies are creative iterations of broader patterns of practice that are simply inaccessible through the linguistic record but revealed through the material archive. Such iterations illuminate the diverse strategies by which the men and women we study sought to make the world or make their way in the world by mobilizing in unique ways a regional repertoire

[email protected]

Speaking with Substance

133

of ideas, associations, and materials named by the words we recover. In this way, the supplemental approach offers historians working with linguistic evidence a granularity lacking in the linguistic record alone: it both situates individuals, events, and experimentation in longue durée narratives of regional practices and concepts and demonstrates that such sweeping stories are always the aggregate of individuals’ experiences and efforts. The supplemental approach does not replace the important foundations crafted through correlations and direct associations, but rather builds on them. After all, we must begin with an idea of which bodies of evidence can be brought together and held in supplemental tension. For us both, correlations and direct associations are crucial steps in the process of working between our sources, and are thus necessary to the supplemental approach. Of course, there are other ways of pulling these archives together. For example, scholars have used conceptual metaphors as a method to associate processes of meaning-making preserved independently in the linguistic and archaeological records (Ortman 2012; Schoenbrun 2016). Such efforts to connect and, thereby, confirm or reconcile the evidence of distinct archives are an important step in the interdisciplinary enterprise. The supplemental approach contributes to such efforts the flexibility to exploit silences as much as mentions, oddities as much as common patterns.

Speaking with Substance The approaches advanced in this volume develop from the particularities of working on early African history. On the one hand, illuminating early African pasts is challenging work: there are far fewer scholars covering a much larger area than in many other regions of the world where this kind of interdisciplinary work is done. Much of that work remains descriptive—with good reason. The richest linguistic and archaeological records seldom overlap in time and space. And, the kinds of additional archives mined in other world regions as a way to augment linguistic or archaeological studies—written materials, monumental architecture, durable plastic arts, and so on—are rare in the African context. But it is unpalatable to suggest that the early histories of the continent primarily focus on processes over people, models over aspirations and accomplishment. In this way, the politics of knowledge in African Studies inspires creative approaches to interdisciplinary historical scholarship in the absence of the archives traditionally used to augment the histories generated from material and language. The supplemental approach presented in this volume offers another way to access rich, human-centered interpretations through linguistic and archaeological records. It could be observed that this is an approach born out of necessity. Yet, shifting focus from the problem of absence to archives’ presence immediately reverses the proposition: the usual additional archives mined to augment linguistic and archaeological records are decidedly rare in the scope of human history, while linguistic and

[email protected]

134

6 Conclusion

archaeological archives are common, if not ubiquitous (geographically if not chronologically).1 We need better methods for accessing the lived experience of all human societies, which requires experimentation with our most common archives, for they are, in fact, our richest archives. If we can develop ways to access the experiences of people underrepresented in the historical record, we will, indeed, be speaking with substance.

 There are debates about how far back in time historical linguistic evidence might reach.

1

[email protected]

Appendix A: A Glossary for Non-specialists

Artifact Assemblage Association Matrix Calibration

Classification Cognate rate Context

An object or part of an object made or altered by humans. A group of different artifacts found in association with one another, that is, in the same context. The relationship between objects or contexts. The soil or other material archaeologists excavate that contains artifacts and other objects. A process that transforms the derived ‘radiocarbon years’ of radiocarbon dating to estimates of calendrical years. Calibration is necessary to account for changes in the global radiocarbon concentration over time. A representation of the genetic relationships within a language family. The number of cognates shared by a pair of languages, usually measured from a standard list of core or basic vocabulary. A part of archaeological deposits that are the remains of an individual stratigraphic event. These may be deposited by humans (e.g., fill of a pit) but may also form in situ (such as the natural development of soils).

© The Author(s), under exclusive licence to Springer International Publishing AG, part of Springer Nature 2019 K. M. de Luna, J. B. Fleisher, Speaking with Substance, SpringerBriefs in Archaeology, https://doi.org/10.1007/978-3-319-91036-9

[email protected]

135

136

Core vocabulary

Correlation

de facto refuse Dendrochronology

Direct association

Appendix A: A Glossary for Non-specialists

A standard collection of either 100 or 200 words identified as being culturally neutral across languages and, therefore, more resistant to historical change. Also known as the Swadesh list or basic vocabulary, cognates within core vocabulary can be used as an initial measure of relatedness within a group of languages. However, this is not a reliable method of classifying languages and should not replace the comparative method. Rates of cognition within core vocabulary are also used by some scholars in glottochronological computations of the range of time during which a protolanguage began to diverge. A posited link between the linguistic and archaeological records developed by identifying overlaps in the proposed location and periodization of protolanguages and archaeological traditions. In some instances, researchers correlate protolanguages and archaeological traditions not on the basis of the overlap in time and space, but on the basis of the presumed commensurable effect on those two records of an underlying common historical process, such as arguments about the spread of maize in the Andes. Correlations serve as a first step in other methods used to more closely connect the linguistic and archaeological records. The refuse of abandonment. An absolute dating method that reads the pattern (number and condition) of annual rings formed in the trunks of trees. When compared to a master sequence, a calendrical date can be derived. A posited link between the linguistic and archaeological records developed by identifying the first instance of an object, practice, organism, etc. in the archaeological record of an area associated with the protolanguage exhibiting the first

[email protected]

Appendix A: A Glossary for Non-specialists

Feature

Formation processes

Geophysical survey

Glottochronology

Internal innovation

137

instance of a name for the recovered archaeological material. The more ­specific the item is to a particular region, the firmer the direct association. The greater the density of direct associations, the more reliable the posited link between the material culture of a period / place and a protolanguage. One or more contexts that represent some human non-portable activity. The determination of a feature is an act of interpretation and examples include pits, walls, and ditches. The processes of natural activities or human behavior that affect or transform artifacts and contexts after their initial period of use in a given activity. Ground-based physical sensing techniques used for archaeological imaging or mapping (e.g., ground penetrating radar, magnetometry). A contested method of dating the beginning of the process of linguistic divergence calculated by applying a formula estimating changes in core vocabulary over long periods of time. Africanists who use this method typically calculate the central ranges (identified by plotting bell curves) of cognate rates within a subgroup to produce a date range for the beginning of the divergence of a protolanguage. An innovation in a domain of language (phonology, morphology, lexis, etc.) produced by the speakers of a protolanguage (rather than borrowed by them). Internal innovations are identified by their distribution and form (especially the degree to which they follow the phonological rules definitive of a language group) in the extant languages developed from the protolanguage to which the innovation is attributed.

[email protected]

138

Lexicostatistics

Lexis

Lithics Loanword

Morphology

Phonology

Phytoliths

Primary refuse

Principle of least moves

Appendix A: A Glossary for Non-specialists

A quick but imprecise method of measuring relatedness within a group of languages by analyzing cognate rates within core vocabulary. In this method of ­classification, cognate rates are tabulated for every language against every other language within a group of languages hypothesized to be related to one another. Individual words but also assemblages of words, whether lexicons relating to an activity like iron-smelting or collections of synonyms, antonyms, or homophones. Stone tools and other chipped stone artifacts. A word borrowed from one language or protolanguage into another. Loanwords (and their language sources) can be identified through their phonology. A system of the units that comprise a word, often characteristic of a language or language family. Noun classes are an example of a morphological unit in Bantu languages. A system of sounds characteristic of a language. The study of such sound systems, particularly as they change over time and, relatedly, within a language family, facilitates the identification of inherited and borrowed words. Durable, microscopic silica bodies that form in living plants. When processed, identified and counted, these allow for the identification of plant remains in archaeological deposits. Refuse that is disposed of in the context of its production and thus has relatively direct spatial association with activities that produced it. To locate the speech communities of particular protolanguages in space, linguistic migration theory assumes a homeland in the area with the greatest linguistic diversity and for which the minimal number of moves would bring speakers

[email protected]

Appendix A: A Glossary for Non-specialists

139

of the languages that diverged from the protolanguage in question back to a common nucleus, or a homeland. This method has been critiqued for ignoring relationships between human mobility, environment, disease, and geography; however, it is merely a method of approximation that can be refined through direct associations. Principle of greatest linguistic diversity To locate the homeland of a protolanguage, linguists identify the area of greatest linguistic diversity within the broader geography of a particular language family. Probabilistic sampling An archaeological sampling method that uses statistical criteria to select the units in which archaeologists carry out their sampling procedure (e.g, pedestrian survey, excavation test pits, surface collection areas). This sample is used to draw reliable, representative general conclusions about a site or region. Protoform A proposed reconstruction of the phonology, morphology, and semantic domain of a word, including the earliest protolanguage to which it can be traced. Also called a starred form. Protolanguage An ancestral language with its associated linguistic attributes (e.g. phonology, lexis, etc.). Protolanguages do not represent actual historical languages. Rather, they are a heuristic device that serves as a label or container for a body of linguistic attributes proposed to have been inherited by a group of related languages and, therefore, to have characterized the common ancestral language from which extant languages developed. Protolexicon The vocabulary proposed for a protolanguage or a subset thereof. Radiocarbon dating An absolute dating method for determining the age of an object containing organic material by using the properties of radiocarbon(c14), a radioactive isotope of carbon.

[email protected]

140

Appendix A: A Glossary for Non-specialists

Secondary refuse

Seriation Speech community Starred form

Subgroup

Systematic pedestrian survey

Words and things

Refuse that is moved from the location of its production to locations dedicated to refuse accumulation (trash pits, middens). A relative dating method in which assemblages or artifacts from numerous sites are placed in chronological order. A social group defined by a common language or, in the case of early history, a proposed protolanguage. A proposed reconstruction of the phonology, morphology, and semantic domain of a word, including the earliest protolanguage to which it can be traced. Also called a protoform. A branch or intermediate group within a language classification. In the comparative method, subgroups are identified by innovations attested in the languages that constitute the subgroup. In lexicostatistics, subgroups within the broader group of languages should share higher cognation rates with each other than with all other languages; skews in these patterns usually indicate processes such as interaction across language boundaries resulting from proximity. Archaeological survey of a defined region applying a clear probabilistic sampling scheme, carried out by walking over and inspecting the survey area. A method of historical linguistic analysis focused on lexis. The approach assumes that a word reconstructed to a protolanguage constitutes indirect evidence of the existence of the idea, thing, or practice it names. This method is frequently used by historians of early African communities and can be tested through comparison to findings in the archaeological record.

[email protected]

Bibliography

Methods: Suggested Readings Archaeology Carver, M. O. (2010). Archaeological investigation. London: Routledge. Hodder, I., & Hutson, S. (2009). Reading the past: Current approaches to interpretation in archaeology. Cambridge: Cambridge University Press. Johnson, M. (2010). Archaeological theory: An introduction. Malden: Wiley-Blackwell. Livingstone-Smith, A., Cornelissen, E., Gosselain, O. P., & MacEachern, S. (Eds.). (2017). Field manual for African archaeology. Tervuren: Royal Museum for Central Africa. http://www.africamuseum.be/research/publications/rmca/online/fmaa Meskell, L., & Preucel, R. W. (2008). A companion to social archaeology. Malden: Blackwell. Mitchell, P.,& Lane, P. (2013). The Oxford handbook of African archaeology. New York: Oxford University Press. Renfrew, C., & Bahn, P.  G. (2016). Archaeology: Theories, methods, and practice. London: Thames and Hudson. Stahl, A. B. (2005). African archaeology: A critical introduction. Malden: Blackwell.

Historical Linguistics & African Languages Bostoen, K. (2017). Historical linguistics. In A. Livingstone Smith, E. Cornelissen, O. P. Gosselain, & S.  MacEachern (Eds.), Field manual for African archaeology (pp.  257–260). Tervuren: Royal Museum for Central Africa. Crowly, T., & Bowern, C. (2010). An introduction to historical linguistics (4th ed.). New York: Oxford University Press. Dimmendaal, G. J. (2011). Historical linguistics and the comparative study of African languages. Amsterdam/Philadelphia: John Benjamins. Ehret, C. (2010). History and the testimony of language. Berkeley: University of California Press.

© The Author(s), under exclusive licence to Springer International Publishing AG, part of Springer Nature 2019 K. M. de Luna, J. B. Fleisher, Speaking with Substance, SpringerBriefs in Archaeology, https://doi.org/10.1007/978-3-319-91036-9

[email protected]

141

Bibliography

142

Nurse, D. (1997). The contributions of linguistics to the study of history in Africa. Journal of African History, 38(3), 359–391. Nurse, D., & Philippson, G. (2003). The bantu languages. New York: Routledge. Schoenbrun, D. L. (forthcoming). Words, things, and meaning: Linguistics as a tool for historical reconstruction. In R. Vossen & G. Dimmendaal (Eds.), The Oxford handbook of African languages. New York: Oxford University Press.

General Bibliography Allen, J. V. (1993). Swahili origins. London: James Currey. Ansell, W. F. H. (1955). The declining red lechwe. Oryx: The International Journal of Conservation, 3(1), 15–18. Anthony, D.  W. (2007). The horse, the wheel, and language: How bronze-age riders from the Eurasian steppes shaped the modern world. Princeton: Princeton University Press. Antonites, A. R., Scott, K., & Grody, E. (2016a). New directions in South African archaeozoology of the last 2,000 years. African Archaeological Review, 33(4), 345–351. Antonites, A. R., Bradfield, J., & Forssman, T. (2016b). Technological, functional and contextual aspects of the K2 and Mapungubwe worked bone industries. African Archaeological Review, 3(4), 437–463. Ashmore, W., & Sabloff, J. (2002). Spatial orders in Maya civic plans. Latin American Antiquity, 13(2), 201–215. Baldi, P. (1988). Comment on ‘archaeology and language: The puzzle of Indo-European origins’. Current Anthropology, 29(3), 445–447. Banning, E. B. (2002). Archaeological survey. New York: Kluwer Academic/Plenum Publishers. Bastin, Y., Coupez, A., & Mann, M. (1999). Continuity and divergence in the bantu languages: Perspectives from a lexicostatistic study. Tervuren: Musée royal de l'Afrique centrale. Bastin, Y., Coupez, A., Mumba, E., & Schadeberg, T. C. (Eds.) (2002). Bantu lexical reconstructions 3. Retrieved from http://www.metafro.be/blr. Accessed Oct 2008 through May 2016. Bayliss, A. (2015). Quality in Bayesian chronological models in archaeology. World Archaeology, 47(4), 677–700. Bell, C. M. (1992). Ritual theory, ritual practice. New York: Oxford University Press. Bellwood, P. S. (2005). First farmers: The origins of agricultural societies. Malden: Blackwell. Bellwood, P. & Colin, R. (2002). Examining the farming/language dispersal hypothesis. Cambridge: McDonald Institute for Archaeological Research. Beresford-Jones, D., & Heggarty, P. (2011). What role for language prehistory in redefining archaeological ‘culture’? A case study on new horizons in the Andes. In B.  W. Roberts & M. Vander Linden (Eds.), Investigating archaeological cultures material culture, variability, and transmission (pp. 355–386). New York: Springer. Beresford-Jones, D., & Heggarty, P. (2012a). Introduction  – Archaeology and language in the Andes: A much-needed conversation. In P. Heggarty & D. Beresford-Jones (Eds.), Archaeology and language in the Andes: A cross-disciplinary exploration of prehistory (pp. 1–42). Oxford: Published for the British Academy by Oxford University Press. Beresford-Jones, D., & Heggarty, P. (2012b). Broadening our horizons: Towards an interdisciplinary prehistory of the Andes. In P.  Heggarty & D.  Beresford-Jones (Eds.), Archaeology and language in the Andes: A cross-disciplinary exploration of prehistory (pp.  57–84). Oxford: Published for the British Academy by Oxford University Press. Bernault, F. (2015). Suitcases and the poetics of oddities: Writing history from disorderly archives. History in Africa, 42(1), 269–277. Bisson, M.  S. (1976). The prehistoric coppermines of Zambia. (Unpublished PhD thesis), University of California, Santa Barbara.

[email protected]

Bibliography

143

Blench, R., & MacDonald, K. (Eds.). (1999). The origins and development of African livestock: Archaeology, genetics, linguistics and ethnography. London: Routledge. Boivin, N., Crowther, A., Prendergast, M., & Fuller, D. (2014). Indian Ocean food globalisation and Africa. African Archaeological Review, 31(4), 547–581. Bourdieu, P. (1977). Outline of a Theory of Practice. Bostoen, K., Clist, B., Doumenge, C., Grollemund, R., Muluwa, J. K., Maley, J., & Hombert, J.-M. (2015). Middle to late holocene paleoclimatic change and the early Bantu expansion in the rain forests of Western Central Africa. Current Anthropology, 56(3), 354–384. Boyle, C. L. (1958). The secretary’s African tour: An account by the secretary of his visit with his wife to east and Central Africa between June and October 1957. Oryx: The International Journal of Conservation, 4(4), 216–269. Bronk Ramsey, C. (2009). Bayesian analysis of radiocarbon dates. Radiocarbon, 51(1), 337–360. Bronk Ramsey, C. (2017). OxCal 4.3 manual. http://c14.arch.ox.ac.uk/oxcalhelp/hlp_contents. html Bronk Ramsey, C., Buck, C., Manning, S., Reimer, P., & van der Plicht, H. (2006). Developments in radiocarbon calibration for archaeology. Antiquity, 80, 783–798. Brück, J. (2005). Experiencing the past? The development of a phenomenological archaeology in British prehistory. Archaeological Dialogues, 12(1), 45–72. Casteel, R. W. (1978). Faunal assemblages and the “Wiegemethode” or weight method. Journal of Field Archaeology, 5(1), 71–77. Chami, F. (1994). The Tanzanian coast in the early first millennium AD: An archaeology of the iron-working, farming communities (Vol. 7). Uppsala: Societas Archaeologica Uppsaliensis. Chami, F. (1998). A review of Swahili archaeology. African Archaeological Review, 15(3), 199–218. Childe, V. G. (1936). Man makes himself. London: Watts. Chittick, H. N. (1974). Kilwa: An Islamic trading city on the East African coast (Vol. 5). Nairobi/ London: British Institute in Eastern Africa. Chittick, H. N. (1984). Manda: Excavations at an island port on the Kenya coast (Vol. 9). Nairobi/ London: British Institute in Eastern Africa. Colson, E. (1958). Marriage and the family among the plateau Tonga of Northern Rhodesia. Manchester: Manchester University Press for the Rhodes-Livingstone Institute. Colson, E. (1959). The plateau Tonga of Northern Rhodesia. In E. C. a. M. Gluckman (Ed.), Seven tribes of British Central Africa (pp. 94–162). Manchester: Manchester University Press for the Rhodes-Livingstone Institute. Colson, E. (1960). The social organization of the Gwembe Tonga. Manchester: Manchester University Press for the Rhodes-Livingstone Institute. Colson, E. (2006). Tonga religious life in the twentieth century. Lusaka: Bookworld Publishers. de Filippo, C., Bostoen, K., Stoneking, M., & Pakendorf, B. (2012). Bringing together linguistic and genetic evidence to test the Bantu expansion. Proceedings: Biological Sciences, 279(1741), 3256–3263. de Luna, K. (2012a). Surveying the boundaries of historical linguistics and archaeology: Early settlement in South Central Africa. African Archaeological Review, 29, 209–251. de Luna, K. (2012b). Hunting reputations: Talent, individualism, and community-building in precolonial South Central Africa. Journal of African History, 53(3), 279–299. de Luna, K. (2013). Affect and society in precolonial Africa. International Journal of African Historical Studies, 46(1), 123–150. de Luna, K. (2015). Marksmen and the bush: The affective micro-politics of landscape, sex, and technology in precolonial South Central Africa. Kronos: Southern African Histories, 41, 37–60. de Luna, K. (2016). Collecting food, cultivating people: Subsistence and Society in Central Africa. New Haven: Yale University Press. de Luna, K.  M., Fleisher, J., & McIntosh, S.  K. (2012). Thinking across the African past: Interdisciplinarity and early history. African Archaeological Review, 29(2–3), 75–94. Deetz, J.  (1996). In small things forgotten: An archaeology of early American life. New  York: Doubleday.

[email protected]

Bibliography

144

Delius, P., & Schoeman, M. (2010). Special issue: History and archaeology in conversation  – South Africa meets East Africa workshop. African Studies, 69(2). https://doi.org/10.1080/00 020184.2010.499194. Denbow, J. (1986). A new look at the later prehistory of the Kalahari. Journal of African History, 27, 3–29. Denbow, J. (1990). Congo to Kalahari: Data and hypotheses about the political economy of the western stream of the early Iron Age. African Archaeological Review, 8, 139–176. Denbow, J. (1999). Material culture and the dialectics of identity in the Kalahari: AD700-1700. In S. K. McIntosh (Ed.), Beyond chiefdoms: Pathways to complexity in Africa (pp. 110–123). Cambridge: Cambridge University Press. Denbow, J. (2011). Excavations at Divuyu, Tsodilo Hills. Botswana Notes & Records, 43, 76–94. Denbow, J. R. (2014). The archaeology and ethnography of Central Africa. Cambridge: Cambridge University Press. Denbow, J. R., Mosothwane, M., & Ndobochani, N. M. (2008). Finding Bosutswe: Archeological encounters with the past. History in Africa, 35, 145–190. Derricourt, R. M. (1985). Man on the Kafue: The archaeology and history of the Itezhitezhi area of Zambia. New York: L. Barber Press. Diamond, J. (1997). Guns, germs, and steel: The fates of human societies. New York: W.W. Norton. Dietler, M. (2001). Theorizing the feast: Rituals of consumption, commensal politics, and power in African contexts. In M.  Dietler & B.  Hayden (Eds.), Feasts: Archaeological and ethnographic perspectives on food, politics, and power (pp. 65–114). Washington, DC: Smithsonian Institution Press. Dimmendaal, G. J. (2011). Historical linguistics and the comparative study of African languages. Amsterdam/Philadelphia: John Benjamins Pub. Co.. Dirks, N. B. (1996). Is vice versa? Historical anthropologies and anthropological histories. In T. J. McDonald (Ed.), The historic turn in the human sciences (pp. 17–51). Ann Arbor: University of Michigan Press. Dobres, M.-A., & Robb, J. E. (2000). Agency in archaeology. London/New York: Routledge. Earle, T., & Preucel, R. (1987). Processual archaeology and the radical critique. Current Anthropology, 28(4), 501–538. Eggert, M. K. H. (2005). The Bantu problem and African archaeology. In A. Stahl (Ed.), African archaeology: A critical introduction (pp. 301–326). London: Blackwell. Ehret, C. (1971). Southern Nilotic history; linguistic approaches to the study of the past. Evanston: Northwestern University Press. Ehret, C. (1998). An African classical age. Charlottesville: University Press of Virginia. Ehret, C. (2000). Testing the expectations of glottochronology against the correlations of language and archaeology in Africa. In C. Renfrew, A. McMahon, & L. Trask (Eds.), Time depth in historical linguistics (pp.  373–399). Cambridge: MacDonald Institute for Archaeological Research. Ehret, C. (2001). Bantu expansions: Re-envisioning a central problem of early African history. International Journal of African Historical Studies, 34(1), 5–41. Ehret, C. (2011). History and the testimony of language. Berkeley: University of California Press. Ehret, C., & Posnansky, M. (Eds.). (1982). The archaeological and linguistic reconstruction of African history. Berkeley: University of California Press. Fagan, B. M. (1967). Iron age cultures in Zambia (Vol. 1). London: Chatto and Windus. Fagan, B. M., & Phillipson, D. W. (1965). Sebanzi: The Iron Age sequence at Lochinvar, and the Tonga. The Journal of the Anthropological Institute of Great Britain and Ireland, 95, 253–294. Fagan, B. M., Phillipson, D. W., & Daniels, S. G. (1969). Iron age cultures in Zambia. London: Chatto & Windus. Fielder, R. (1979). Economic spheres in pre-colonial lla society. African Social Research, 28, 617–641. Finlayson, B., & Warren, G. (2010). Changing natures: Hunter-gatherers, first farmers and the modern world. London: Duckworth.

[email protected]

Bibliography

145

Fleisch, A., & Stephens, R. (Eds.). (2016). Doing conceptual history in Africa. New  York: Berghahn. Fleisher, J. (2018). Town and village. In , W.-J., Stephanie and L. Adria (Eds.), The swahili world (pp. 194–204). London: Routledge. Fleisher, J. B. (2003). Viewing stonetowns from the countryside: An archaeological approach to Swahili regional systems, AD800–1500 (Unpublished PhD thesis), University of Virginia. Fleisher, J. B. (2010). Rituals of consumption and the politics of feasting on the eastern African coast, AD 700–1500. Journal of World Prehistory, 23(4), 195–217. Fleisher, J. B., & LaViolette, A. (2013). The early Swahili trade village of Tumbe, Pemba Island, Tanzania, AD 600-950. Antiquity, 87(338), 1151–1168. Fleisher, J.  B., & Wynne-Jones, S. (2011). Ceramics and the early Swahili: Deconstructing the early Tana tradition. African Archaeological Review, 28(4), 245–278. Fleisher, J. B., & Wynne-Jones, S. (2012). Finding meaning in ancient Swahili spatial practices. African Archaeological Review, 29(2/3), 171–207. Fleisher, J.  B., de Luna, K., & McIntosh, S.  K. (2012). Thinking across the African past: Archaeological, linguistic and genetic research on the precolonial African past. African Archaeological Review, 29(2/3), 75–317. Fleisher, J., Lane, P., LaViolette, A.  J., Horton, M., Pollard, E., Morales, E.  Q., Vernet, T., Christie, A., & Wynne-Jones, S. (2014). When did the Swahili become maritime? American Anthropologist, 117(1), 100–115. Flexner, J. L., Fleisher, J. B., & LaViolette, A. (2008). Bead grinders and early Swahili household economy: Analysis of an assemblage from Tumbe, Pemba Island, Tanzania, 7th – 10th centuries AD. Journal of African Archaeology, 6(2), 161–181. Fowler, D. (2000). A dictionary of Ila usage, 1860–1960. Hamburg: Lit. Freeman-Grenville, G. S. P. (1962). The East African coast. Select documents from the first to the earlier nineteenth centuries. London: Clarendon Press. Fuller, D., Boivin, N., Hoogervorst, T., & Allaby, R. (2011). Across the Indian Ocean: The prehistoric movement of plants and animals. Antiquity, 85, 544–558. Garrow, D. (2012). Odd deposits and average practice : A critical history of the concept of structured deposition. Archaeological Dialogues, 19(2), 85–115. Gifford-Gonzalez, D. (2014). Constructing community through refuse disposal. African Archaeological Review, 31(2), 339–382. Glassman, J. (1995). Feasts and riot: Revelry, rebellion, and popular consciousness on the Swahili Coast 1856–1888. London: James Currey. Gonzales, R. M. (2009). Societies, religion, and history: Central-east Tanzanians and the world they created, c. 200 BCE to 1800 CE. New York: Columbia University Press. Gramly, R. M. (1978). Expansion of Bantu-speakers versus development of Bantu language and African culture in situ: An archaeologist’s perspective. South African Archaeological Bulletin, 33, 107–112. Grollemund, R. (2012). Nouvelles approches en classification : Application aux langues bantu du Nord-Ouest. Ph.D Dissertation, Université Lumière Lyon 2. Grollemund, R., Branford, S., Bostoen, K., Meade, A., Venditti, C., & Pagel, M. (2015). Bantu expansion shows that habitat alters the route and pace of human dispersals. Proceedings of the National Academy of Sciences, 112(43), 13296–13301. Guthrie, M. (1967–1971). Comparative Bantu, 4 vols. Farnborough: Gregg International. Haller, T. (2007). The contested floodplain: Institutional change of common pool resource management and conflicts among the Ila, Tonga, and Batwa, Kafue Flats (Southern Province), Zambia. (Habilitation) University of Zurich. Hanks, W. (2010). Converting words: Maya in the age of the cross. Berkeley: University of California Press. Heggarty, P. (2007). Linguistics for archaeologists: Principles, methods and the Case of the Incas. Cambridge Archaeological Journal, 17(03), 311–340. Heggarty, P. (2015). Prehistory through language and archaeology. In C.  Bowern & B.  Evans (Eds.), The Routledge handbook of historical linguistics (pp. 598–626). London: Routledge.

[email protected]

Bibliography

146

Heggarty, P., & Beresford-Jones, D. (2010). Agriculture and language dispersals: Limitations, refinements, and an Andean exception? Current Anthropology, 51(2), 163–191. Heggarty, P., & Beresford-Jones, D. (2012). Archaeology and language in the Andes: A cross-­ disciplinary exploration of prehistory. Oxford: Published for the British Academy by Oxford University Press. Heggarty, P., & Renfrew, C. (2014). Introduction: Languages. In C. Renfrew & P. Bahn (Eds.), The Cambridge world prehistory (Vol. 1, pp. 19–44). Cambridge: Cambridge University Press. Helm, R. (2000). Conflicting histories: The archaeology of Iron-working, farming communities in the central and southern coast region of Kenya (PhD thesis), University of Bristol. Herbert, E.  W. (1993). Iron, gender and power: Rituals of transformation in African societies. Bloomington: Indiana University Press. Herbert, R. K., & Huffman, T. N. (1993). A new perspective on Bantu expansion and classification: Linguistic and archaeological evidence fifty years after Doke. African Studies, 52(2), 53–76. Hodder, I. (1982). Symbolic and structural archaeology. Cambridge: Cambridge University Press. Hodder, I. (2006). Catalhoyuk, the leopard’s tale: Revealing the mysteries of Turkey’s ancient ‘town’. London: Thames & Hudson. Hodder, I., & Berggren, A. (2003). At the trowel’s edge: An introduction to reflexive field practice in archaeology. Boulder: Westview. Holden, C., & Gray, R. (2006). Rapid radiation, borrowing and dialect continua in the Bantu languages. In P. Forster & C. Renfrew (Eds.), Phylogenetic methods and the prehistory of languages (pp. 19–31). Cambridge: MacDonald Institute for Archaeological Research. Hollis, A.  C. (1900). Notes on the history of the Vumba, East Africa. The Journal of the Anthropological Institute of Great Britain and Ireland, 30, 275–297. Horton, M. C. (1986). Asiatic colonization of the East African coast: The Manda evidence. Journal of the Royal Asiatic Society, 2, 201–213. Horton, M. C. (1987). Early Muslim trading settlements on the East African coast: New evidence from Shanga. Antiquaries Journal, 67, 290–322. Horton, M. C. (1991). Primitive Islam and architecture in East Africa. Muqarnas, 8, 103–116. Horton, M.C. and Mudida N. (1993). Exploitation of marine resources: Evidence for the origin of the swahili communities of East Africa. In T. Shaw, P. Sinclair, B.W. Andah and A. Okpoko (Eds.), In the archaeology of Africa: Food, metals and towns (pp. 673–693). Routledge, London. Horton, M. C. (1996). Shanga: The archaeology of a Muslim trading community on the coast of East Africa. Nairobi: British Institute in Eastern Africa. Horton, M. C., & Middleton, J. (2000). The Swahili: The social landscape of a mercantile society. Oxford: Blackwell. Huffman, T.  N. (1989). Ceramics, settlements and late Iron Age migrations. The African Archaeological Review, 7, 155–182. Huffman, T. N., & Herbert, R. K. (1994/1995). New perspectives on Eastern Bantu. Azania, 29/30, 27–36. Inskeep, R. (1962). Some Iron Age sites from Northern Rhodesia. South African Archaeological Bulletin, 14, 91–96. Inskeep, R. (1978). An iron-smelting furnace in southern Zambia. South African Archaeological Bulletin, 33, 113–117. Joyce, R. (2008). Practice in and as deposition. In B. Mills & W. Walker (Eds.), Memory work: The materiality of depositional practice (pp. 25–40). Santa Fe: School of American Research Press. Joyce, R., & Pollard, J.  (2010). Archaeological assemblages and practices of deposition. In D.  Hicks & M.  Beaudry (Eds.), The Oxford handbook of material culture studies. Oxford: Oxford University Press. Juma, A. (2004). Unguja Ukuu on Zanzibar: An archaeological study of early urbanism (Vol. 3). Uppsala: Societas Archaeologica Upsaliensis. Katanekwa, N. M. (2016). The prehistory of the 73+ Bantu languages and Bantu language groups of Zambia 3000 BC to 1600 AD. Livingstone: Lendekwa Heritage Consultancy & Services.

[email protected]

Bibliography

147

Kelly, R. L., & Thomas, D. H. (2011). Archaeology: Down to earth. Belmont: Wadsworth, Cengage Learning. Killick, D. (1990). Technology in its social setting: Bloomery iron-smelting at Kasungu, Malawi, 1860–1940 (PhD), Yale University. Kirch, P. V., & Green, R. C. (2001). Hawaiki, ancestral Polynesia: An essay in historical anthropology. Cambridge/New York: Cambridge University Press. Klieman, K. A. (2003). The pygmies were our compass: Bantu and Batwa in the history of west central Africa, early times to c. 1900 C.E. Portsmouth: Heinemann. Kuper, A., & van Leynseele, P. (1978). Social anthropology and the “Bantu expansion.”. Africa, 48, 335–353. Kusimba, C. M. (1999). The rise and fall of Swahili states. Walnut Creek: AltaMira Press. Kusimba, C. M., & Kusimba, S. B. (2005). Mosaics and interactions: East Africa, 2000 b.p. to the present. In A. B. Stahl (Ed.), African archaeology: A critical introduction (pp. 392–419). Malden: Blackwell. Landau, P. (2010). Popular politics in the history of South Africa, 1400–1948. New  York: Cambridge. Lau, G. (2012). The 1st millennium ad in north central Peru: Critical perspectives on a linguistic prehistory. In P. Heggarty & D. Beresford-Jones (Eds.), Archaeology and language in the Andes: A cross-disciplinary exploration of prehistory (pp. 163–196). Oxford: Published for the British Academy by Oxford University Press. LaViolette, A. (2008). Swahili cosmopolitanism in Africa and the Indian Ocean world, A.D. 600-­ 1500. Archaeologies: Journal of the World Archaeological Congress, 1, 24–49. LaViolette, A., & Fleisher, J. B. (2005). The archaeology of sub-Saharan urbanism: Cities and their countrysides. In A. B. Stahl (Ed.), African archaeology: A critical introduction (pp. 327–352). Oxford: Blackwell. LaViolette, A., & Fleisher, J. B. (2009). The urban history of a rural place: Swahili archaeology on Pemba Island, Tanzania, 700–1500 AD. International Journal of African Historical Studies, 42(3), 433–455. Livingstone, D. (1865). Narrative of an expedition to the Zambesi and its tributaries; and of the discovery of the Lakes Shirwa and Nyassa, 1858–1864. London: John Murray. Livingstone-Smith, A., Cornelissen, E., Gosselain, O., & MacEachern, S. (Eds.). (2017). Field manual for African archaeology/Manuel de terrain en Archéologie africaine. Tervuren: Royal Museum for Central Africa. Loftus, E., Sealy, J., & Lee-Thorp, J.  (2016). New radiocarbon dates and Bayesian models for Nelson Bay Cave and Byneskranskop 1: Implications for the South African later stone age sequence. Radiocarbon, 58(2), 365–381. Logan, A., & Stahl, A. (2017). Genealogies of practice in and of the environment in Banda, Ghana. Journal of Archaeological Method and Theory, 24, 1356–1399. de Luna, K. M. (2010). Classifying Botatwe: M.60 and K.40 languages and the settlement chronology of South Central Africa. Africana Linguistica, 16, 65–96. Lwanga-Lunyiigo, S. (1976). The Bantu problem reconsidered. Current Anthropology, 17, 282–286. Lyman, L. (1994). Vertebrate Taphonomy. Cambridge: Cambridge University Press. MacDonald, K. (1995). Analysis of the mammalian, avian, and reptilian remains. In S. McIntosh (Ed.), Excavations at Jenne-Jeno, Hambarketolo, and Kaniana (Inland Niger Delta, Mali), the 1981 season (pp. 291–318). Berkleley: University of California Press. MacDonald, K., & MacDonald, R. (1995). Mammalian, avian, and reptilian remains. In R. McIntosh, S. McIntosh, & H. Bocoum (Eds.), The search for Takrur: Archaeological excavations and reconnaissance along the Middle Senegal Valley (pp. 311–334). New Haven: Yale University Press. Makoni, S., & Pennycook, A. (2005). Disinventing and (re)constituting languages. Critical Inquiry in Language Studies, 2(3), 137–156.

[email protected]

Bibliography

148

Marten, L. (2006). Bantu classification, Bantu trees and phylogenetic methods. In P.  Forster & C.  Renfrew (Eds.), Phylogenetic methods and the prehistory of languages. Cambridge: McDonald Institute for Archaeological Research, University of Cambridge. McIntosh, S. K. (1995). Excavations at Jenne-Jeno, Hambarketolo, and Kaniana (Inland Niger Delta, Mali), the 1981 season. Berkleley: University of California Press. McIntosh, R. J. (1998). The peoples of the Middle Niger: The island of gold. Malden: Blackwell Publishers. McIntosh, S. K. (2005). Archaeology and the reconstruction of the African past. In J. E. Philips (Ed.), Writing African history (pp. 51–85). Rochester: University of Rochester Press. Middleton, J. (1992). The world of the Swahili: An African mercantile civilization. New Haven: Yale University Press. Mills, B., & Walker, W. (Eds.). (2008a). Memory work: Archaeologies of material practices. Santa Fe: School for Advanced Research Press. Mills, B., & Walker, W. (2008b). Introduction: Memory, materiality and depositional practice. In B.  Mills & W.  Walker (Eds.), Memory work: The materiality of depositional practice (pp. 3–23). Santa Fe: School of American Research Press. Möhlig, W. (1989). Sprachgeschichte, Kulturgeschichte und Archäologie: die Kongruenz der Forschungsergebnisse als methodologisches Problem. Paideuma, 35, 189–196. Moreau, R.  E. (1941). The joking (utani) relationship in Tanganyika. Tanganyika Notes and Records, 12, 1–10. Moreau, R. E. (1944). Joking relationships in Tanganyika. Africa, 14, 386–400. Mudida, N., & Horton, M. C. (1996). Subsistence at Shanga. In M. C. Horton (Ed.), Shanga: A Muslim trading community on the East African coast (Vol. 14, pp. 373–393). Nairobi: British Institute in Eastern Africa. Mukanzubo Kalinda Institute. (2011). Chitonga-English dictionary. Monze: Mukanzubo Kalinda Institute. Mulaudzi, M., Schoeman, M.  H., & Chirikure, S. (2010). 500 year initiative. South African Historical Journal, 62(2), 219–286. Nurse, D. (1997). The contribution of linguistics to the study of history in Africa. Journal of African History, 38, 359–391. Nurse, D., & Hinnebusch, T.  J. (1993). Swahili and Sabaki: A linguistic history. Berkeley: University of California Press. Nurse, D., & Spear, T. (1985). The Swahili: Reconstructing the history and language of an African society AD800–1500. Philadelphia: University of Pennsylvania Press. Oliver, R. (1966). The problem of the Bantu expansion. Journal of African History, 7, 361–376. Ortman, S. G. (2012). Winds from the north Tewa origins and historical anthropology. Salt Lake City: University of Utah Press. Pakendorf, B., Bostoen, K., & de Filippo, C. (2011). Molecular perspectives on the Bantu expansion: A synthesis. Language Dynamics and Change, 1(1), 50–88. Pawley, A., & Green, K. (1971). Lexical evidence for the Proto-Polynesian homeland. Te Reo, 14, 1–35. Pawley, A., & Ross, M. D. (1993). Austronesian historical linguistics and culture history. Annual Review of Anthropology, 22, 425–459. Pearce, F. B. (1920). Zanzibar, the Island metropolis of Eastern Africa. Fisher Unwin: London. Peterson, D. R. (2004). Creative writing: Translation, bookkeeping, and the work of imagination in colonial Kenya. Portsmouth: Heinemann. Phillipson, D. (1968). The early Iron Age in Zambia-regional variants and some tentative conclusions. Journal of African History, 9(2), 191–211. Phillipson, D. W. (1976). The prehistory of Eastern Zambia. Nairobi: British Institute in Eastern Africa. Phillipson, D. W. (1977). The later prehistory of Eastern and Southern Africa. London: Heinemann. Plog, S., Fred, P., Walter W. (1978). Decision making in modern surveys. Advances in Archaeological Method and Theory, 1, 383–421.

[email protected]

Bibliography

149

Pollard, J. (2008). Deposition and material agency in the early Neolithic of Southern Britain. In B. J. Mills & W. H. Walker (Eds.), Memory work: Archaeologies of material practices. Santa Fe: School for Advanced Research Press. Quintana Morales, E. (2013). Reconstructing Swahili foodways: The archaeology of fishing and fish consumption in coastal East Africa, AD 500–1500 (PhD), University of Bristol. Quintana Morales, E. M., & Horton, M. (2014). Fishing and fish consumption in the Swahili communities of East Africa, 700-1400 CE. Internet Archaeology, 37, 329–345. Ramos, G. (2011). Language and society in early colonial Peru. In P. Heggarty & A. J. Pearce (Eds.), History and language in the Andes. New York: Palgrave Macmillan. Ray, D. E. (2014). Disentangling ethnicity in East Africa, ca. 1 – 2010 CE: Past communities in present practices (PhD), University of Virginia. Reitz, E., & Wing, E. (2008). Zooarchaeology (2nd ed.). Cambridge: Cambridge University. Reitz, E., Quitmyer, I., Hale, H. S., Scudder, S., & Wing, E. (1987). Application of Allometry to Zooarchaeology. American Antiquity, 52(2), 304–317. Renfrew, C. (1987). Archaeology and language: The puzzle of Indo-European origins. New York: Cambridge University Press. Renfrew, C. (1988). Archaeology and language: The puzzle of Indo-European origins. Current Anthropology, 29(3), 437–468. Rennie, J. K. (1982). Traditional society and modern developments in Namwala district. In G. W. Howard & G.  J. Williams (Eds.), Proceedings of the national seminar on environment and change: The consequences of hydroelectric power development on the utilization of the Kafue Flats, Lusaka, April 1978 (pp.  35–46). Lusaka: Kafue Basin Research Committee of the University of Zambia. Reynolds, B. (1968). The material culture of the peoples of the Gwembe Valley. Manchester: Manchester University Press. Richards, C., & Thomas, J.  (1984). Ritual activity and structured deposition in later Neolithic Wessex. In R.  Bradley & J.  Gardiner (Eds.), Neolithic studies: A review of some current research (pp. 189–218). Oxford: British Archaeological Reports British Series 133. Robertshaw, P. (2000). Sibling rivalry? The intersection of archaeology and history. History in Africa, 27, 261–286. Robertson, J. (2001). South African early Iron Age in Zambia. In P. Peregrine & M. Ember (Eds.), Encyclopedia of prehistory vol. 1 Africa (pp. 260–271). New York: Springer. Robertson, J.  H., & Bradley, R. (2000). A new paradigm: The African early Iron Age without Bantu migrations. History in Africa, 27, 287–323. Robin, C. (2013). Everyday life matters: Maya farmers at Chan. Gainesville: University Press of Florida. Russell, N. (2012). Social zooarchaeology: Humans and animals in prehistory. Cambridge/New York: Cambridge University Press. Saidi, C. (2010). Women’s authority and society in early east-central Africa. Rochester: University of Rochester Press. Sapir, E. (1916). Time perspective in aboriginal American culture, a study in method, Geological survey memoir 90: no. 13 anthropological series. Ottawa: Government Printing Bureau. Schadeberg, T. (2003). Derivation. In D.  Nurse & G.  Philippson (Eds.), The Bantu languages (pp. 71–89). New York: Routledge. Schiffer, M. B. (1976). Behavioral archeology. New York: Academic Press. Schmidt, P. R. (1975). A new look at interpretations of the early Iron Age in East Africa. History in Africa, 2, 127–136. Schmidt, P.  R. (1978). Historical archaeology: A structural approach in an African culture. Westport: Greenwood Press. Schoenbrun, D. L. (1997). The historical reconstruction of Great Lakes Bantu cultural vocabulary: Etymologies and distributions. Köln: Rüdiger Köppe Verlag. Schoenbrun, D. L. (1998). A green place, a good place: Agrarian change, gender, and social identity in the Great Lakes region to the 15th century. Portsmouth: Heinemann.

[email protected]

Bibliography

150

Schoenbrun, D. L. (1999). The (in)visible roots of Bunyoro-Kitara and the Buganda in the Lakes region: AD 800–1300. In S. K. McIntosh (Ed.), Beyond chiefdoms: pathways to complexity in Africa. Cambridge: Cambridge University Press. Schoenbrun, D. (2012). Mixing, moving, making, meaning: Possible futures for the distant past. African Archaeological Review 29(2–3), 293–317. Schoenbrun, D. (2016). Pythons worked: constellating communities of practice with conceptual metaphor in northern Victoria Nyanza, ca. 800 to 1200 C.E. In A. B. Stahl & A. P. Roddick (Eds.), Knowledge in motion constellations of learning across time and place (pp. 216–246). Tucson: The University of Arizona Press. Scudder, T. (1962). The ecology of the Gwembe Tonga. Manchester: Manchester University Press. Siamwiza, B. (2007). Famine and hunger in the history of the Gwembe Valley, Zambia, c. 1850-­ 1958. In C.  Lancaster & K.  P. Vickery (Eds.), The Tonga-speaking peoples of Zambia and Zimbabwe (pp. 237–261). Lanham: University Press of America. Skoglund, P., Thompson, J. C., Prendergast, M. E., Mittnik, A., Sirak, K., Hajdinjak, M., Salie, T., et al. (2017). Reconstructing prehistoric African population structure. Cell, 171(1), 59–71.e21. Smith, M.  L. (2015). Feasts and their failures. Journal of Archaeological Method and Theory, 22(4), 1215–1237. Smith, E.  W., & Dale, A.  M. (1920). The Ila-speaking peoples of Northern Rhodesia (Vol. 2). London: Macmillan. Spear, T. T. (2000). Early Swahili history reconsidered. International Journal of African Historical Studies, 33(2), 257–290. Stahl, A. B. (1993). Concepts of time and Approches to analogical reasoning in historical perspective. American Antiquity, 58(2), 235–260. Stahl, A.  B. (1994). Change and continuity in the Banda area, Ghana: The direct historical approach. Journal of Field Archaeology, 21, 181–203. Stahl, A. B. (2001). Making history in Banda anthropological visions of Africa’s past. Cambridge: Cambridge University Press. Stahl, A. B. (2008). Dogs, pythons, pots, and beads. The dynamics of shrines and sacrificial practices in Banda, Ghana, 1400-1900CE.  In B.  Mills & W.  Walker (Eds.), Memory work: The materiality of depositional practice (pp.  159–186). Santa Fe: School of American Research Press. Stahl, A. B. (2009). The archaeology of African history. International Journal of African Historical Studies, 42, 241–255. Stahl, A. B. (2010). ‘Route work’ through alternative archives: Reflections on cross-disciplinary practice. South African Historical Journal, 62(2), 252–267. Stahl, A. B., & LaViolette, A. (2009). Introduction: Current trends in the archaeology of African history. International Journal of African Historical Studies, 42(3), 347–350. Stahl, A. B., Mann, R., & Loren, D. D. (2004). Writing for many: Interdisciplinary communication, constructionism, and the practices of writing. Historical Archaeology, 38(2), 83–102. Stephens, R. (2013). A history of African motherhood: The case of Uganda, 700–1900. Cambridge: Cambridge University Press. Stephens, R. (2016). ‘Wealth’, ‘poverty’ and the question of conceptual history in oral contexts. In A. Fleisch & R. Stephens (Eds.), Doing conceptual history in Africa (pp. 21–48). New York: Berghahn. Stephens, R. & Fleisch, A. (2016). Theories and methods of African conceptual history. In A. Fleisch and R. Stephens (Eds.), Doing conceptual history in Africa (pp. 1–20). New York: Berghahn. Steward, J. (1942). The direct historical approach to archaeology. American Antiquity, 7, 337–443. Sutton, D. (1996). Historical linguistics and Pacific archaeology: The unreality of ancestral Polynesian society. In R. C. Green & J. M. Davidson (Eds.), Oceanic culture history: Essays in honour of Roger Green (pp. 377–385). Dunedin: New Zealand Journal of Archaeology. Swanepoel, N., Esterhuysen, A., & Bonner, P. L. (Eds.). (2008). Five hundred years rediscovered: Southern African precedents and prospects. Johannesburg: Wits University Press.

[email protected]

Bibliography

151

Swenson, E. (2014). The archaeology of ritual. Annual Review of Anthropology, 44(1):329–345. Tishkoff, S. A., Reed, F. A., Friedlaender, F. R., Ehret, C., Ranciaro, A., Froment, A., Hirbo, J. B., Awomoyi, A. A., Bodo, J. M., Doumbo, O., Ibrahim, M., Juma, A. T., Kotze, M. J., Lema, G., Moore, J. H., Mortensen, H., Nyambo, T. B., Omar, S. A., Powell, K., Pretorius, G. S., Smith, M. W., Thera, M. A., Wambebe, C., Weber, J. L., & Williams, S. M. (2009). The genetic structure and history of Africans and African Americans. Science, 324(5930), 1035–1044. Trouillot, M.-R. (1995). The silences of the past the power and the production of history. Boston: Beacon Press. Tuden, A. (1962). The political system of the Ba-Ila: An analysis of change and stability in the patterns of authority (PhD), Northwestern University. Vansina, J. (1978). The children of Woot: A history of the Kuba peoples. Madison: University of Wisconsin Press. Vansina, J. (1986). Do pygmies have a history? Sprache und geschtichte in Afrika 7, 1, 5–34. Vansina, J. (1990). Paths in the rainforests: Toward a history of political tradition in equatorial Africa. Madison: The University of Wisconsin Press. Vansina, J. (1995). Historians, are archaeologists your siblings? History in Africa, 22, 369–408. Vansina, J.  (2004). How societies are born: Governance in west Central Africa before 1600. Charlottesville: University of Virginia Press. Vogel, J. O. (1971a). Kamangoza: An introduction to the Iron Age cultures of the Victoria Falls region. Nairobi/London: Oxford University Press for the National Museums of Zambia. Vogel, J.  O. (1971b). Kumadzulo: An early Iron Age village site in southern Zambia. Lusaka/ London: Oxford University Press for the National Museums of Zambia. Vogel, J. O. (1975). Simbusenga: The archaeology of the intermediate period in southern Zambia. London: Oxford University Press for the National Museums of Zambia. Walshaw, S. C. (2010). Converting to rice: Urbanization, Islamization and crops on Pemba Island, Tanzania, ad 700-1500. World Archaeology, 42(1), 137–154. Walshaw, S. C. (2015). Swahili urbanization, trade, and food production: Botanical perspectives from Pemba Island, Tanzania AD 600–1500. Oxford: Archeopress. Walz, J. R. (2010). Route to a regional past: An archaeology of the Lower Pangani (Ruvu) Basin, Tanzania, 500–1900 C.E (PhD dissertation), University of Florida. Watt, E. (1980). Excavations at matobo iron age site, Kalomo, Zambia. Zambia Museum Journal, 5, 79–85. Wilmsen, E. (2011). Nqoma. Botswana Notes & Records, 43, 95–114. Wilmsen, E., Killick, D., Rosenstein, D.  D., Thebe, P., & Denbow, J.  (2009). The social geography of pottery in Botswana as reconstructed by optical petrography. Journal of African Archaeology, 7, 3–39. Wright, H. T. (1984). Early seafarers of the Comoro Islands: The Dembeni phase of the IXth-Xth centuries AD. Azania, 19, 13–59. Wright, H. T. (1993). Trade and politics on the eastern littoral of Africa, AD800-1300. In T. Shaw, P. J. J. Sinclair, B. Andah, & A. Okpoko (Eds.), The archaeology of Africa: Food, metals and towns (pp. 658–670). London: Routledge. Wylie, A. (1989). Archaeological cables and tacking: The implications of practice for Berstein’s ‘options beyond objectivism and relativism. Philosophy of the Social Sciences, 19, 1–18. Wynne-Jones, S. (2010). Remembering and reworking the Swahili Diwanate: The role of objects and places at Vumba Kuu. International Journal of African Historical Studies, 43(3), 407–427. Wynne-Jones, S. (2015). Biographies of practice and the negotiation of Swahili at nineteenth-­ century Vumba. In F. G. Richard (Ed.), Materializing colonial encounters: Archaeologies of African experience (pp. 155–175). New York: Springer. Wynne-Jones, S. (2016). A material culture: Consumption and trade in precolonial East Africa. Oxford: Oxford University Press. Wynne-Jones, S., & Fleisher, J.  B. (2013). Ceramics and society: Early Tana Tradition and the Swahili coast (data paper). Internet Archaeology, 35. https://doi.org/10.11141/ia.35.7.

[email protected]

Index

A Africa eastern coast, 52, 61, 75–125 precolonial, 1 south central, 29–73, 98, 130, 132 sub-Saharan, 59, 66, 113, 128 west central, 107 African archaeologists, 1, 66 African archaeology, 10, 54 African coastal sites, 61 African history, 95 African Studies, 133 Agriculture agrarian economy, 33 AMS dates, 64 Analogic reasoning, 11 Analogy analogic reasoning, 11 Anatolian, 22 Ancestor, 12, 26, 43, 44, 57, 62, 65, 70, 71, 73, 109 Ancestral Polynesian society, 26 Animal domestic, 51, 56–58, 70, 76 wild, 51, 56–58, 60, 65, 70, 71 (see also Antelope) Anthony, D., 25, 108 Antelope, 36, 40, 44–46, 69–72, 97, 130 Anthropology anthropologist, 1, 64 Antonym, 13, 18, 102, 131, 138 Arabic, 103 Archaeological data, 6, 56 Archaeological evidence, 108

Archaeological and historical linguistic records, 20 Archaeological interpretations, 11 Archaeological and linguistic distributions, 109 Archaeological and linguistic evidence, 2, 25 correlation, 21 Archaeological materials assemblage of faunal, 55 excavations, 55 faunal assemblages, 57 household level data, 56 preservation issues, 59 smelting activities, 55 taphonomy, 59 Archaeological publications, 58 Archaeological record, 8, 67 Archaeological regions, 114 Archaeological reporting, 48 Archaeological survey, 114 Archaeological temporality phonology, 13 radiocarbon calibration, 10 radiocarbon dating, 10 source- and subject-side concerns, 11 Archaeologist’s interpretations, 11 Archaeology, 66, 127–130 archaeological data, 2, 3, 6, 11, 12, 47, 48, 54–56, 58, 66, 82, 89, 94, 112, 118 deposits, 8, 9, 135, 138 record, 8, 10, 11, 14, 16, 19, 20, 24, 27, 29, 31–33, 35, 36, 40, 41, 47, 49, 56, 57, 59, 67–69, 71, 73, 97, 101, 107, 112, 119, 125, 131–133, 136, 140

© The Author(s), under exclusive licence to Springer International Publishing AG, part of Springer Nature 2019 K. M. de Luna, J. B. Fleisher, Speaking with Substance, SpringerBriefs in Archaeology, https://doi.org/10.1007/978-3-319-91036-9

[email protected]

153

Index

154 Archaeology (cont.) archaeologist, 1–3, 5–12, 15, 18–21, 23, 28, 29, 36, 47–49, 51–54, 56–61, 63–69, 78, 80, 87, 93, 95, 99, 101, 102, 107, 109, 110, 112–118, 121, 123, 128–130, 132, 135, 139 and historical linguistics artifacts, 6 assemblage, 8 association, 7 context, 7 excavations and surveys, 7 features, 6 lithics, 6, 9 materials and samples, 9 methodologies, 5 primary/secondary refuse, 8 sampling strategy, 9 space and time, 7 supplemental approach, 27–29 survey procedures, 7 survey units, 7 of region, 66 Archaeology and Language in 1987, 22 Architecture, 120, 124, 125, 133 Archive, 3, 11, 16–18, 20, 27–30, 67, 73, 116, 117, 131–134 Artifact, 6, 7, 9, 11, 22, 48, 54, 55, 60, 61, 67, 68, 81, 129, 130, 135, 137, 138, 140 Aspirational and empty speech, 98 Assemblages, 48, 113 ceramics, 48, 52, 82–85, 88, 90, 91, 93, 95, 100, 113, 123, 124, 132 faunal, 9, 36, 48–51, 55–60, 62 and lexicons, 129 B Bantu, 34–36, 41, 43 communities, 115 expansion, 21 language family, 31 languages, 14, 36, 79 lexical reconstructions, 15, 35, 103, 105 missionary, 17 origin, 21, 22 -speaking, 17, 26, 30, 41, 109 toolkit, 21 Batoka Plateau, 40, 43, 44, 48, 62, 68, 72, 130 Bayesian modelling, 10, 64 Bead, 84, 93, 101, 111, 121, 124, 125 Beresford-Jones, D., 23–25, 108, 109 Bone, 6, 9, 10, 19, 40, 49–51, 56, 57, 60–62, 65, 69, 70, 72, 73, 132

Borrowing, 12, 13, 15, 43, 96, 101, 108, 110 Botatwe communities, 41 family, 31 languages, 32, 35, 41 speech communities, 31 Boundary language, 12, 18, 124, 140 linguistic (see Language) (see also Frontier) Bowl feasting, 95, 99 Bride, 71, 72 See also Marriage Bush -craft, 42, 44, 46, 63 Butchery, 40, 41, 43, 44, 51, 55, 60, 65, 130 C Calibrate, 10, 39, 53, 54 Carbon-dated archaeological evidence, 25 Cattle, 19, 25, 57, 62, 69–71, 108 Causation, 23, 25, 64 Celebrity, 33 See also Fame Central Eastern Botatwe and Kafue periods, 35 Central Eastern Botatwe villagers, 34 Ceramic, 82, 84, 89, 100 assemblage, 48, 52, 82–85, 88, 90, 91, 93, 95, 100, 113, 123, 124, 132 forms and decorations, 113 record, 114 vessel types, 91 (see also Pottery) Cereal, 24, 31, 33, 81 Chavín Early Horizon, 23 Chibuene, 90, 91, 100, 101 Chief, 46, 106, 120 Chila, 43, 44 Childe, V.G., 21, 22 Christian missionary, 17 Choreography, 99 Chronology, 5, 11, 14, 23, 24, 53, 54, 58, 64, 75, 79, 80, 89, 98, 101, 106–108, 116, 118, 125, 128, 140 Classification, 13–15, 31, 32, 42, 52, 71, 96, 106, 107, 109, 111, 112, 135, 138, 140 Climate, 33 Coast coastal resident, 98, 121 coastal settlement, 94, 121

[email protected]

Index

155

Coastal Authority, 119 Coastal Ceramics Project, 59, 90, 99, 124 Coastal languages, 108 Cognate rate, 13, 14, 32, 96, 99, 101, 108, 110, 135, 137, 138 Collaboration, 1, 2, 23, 24, 27, 29, 30, 34, 44, 71, 78, 114, 128, 131 See also Interdisciplinary Colonial colonization, 21 Communal hunts, 43 Community speech, 12, 14, 15, 19, 24, 25, 31, 35, 40, 43, 44, 96, 98, 99, 112, 131, 138, 140 Comorian languages, 101, 106 Comoros, 91, 100, 101, 124 Comparative comparison, 13, 19, 52 method, 13 Competition, 75, 76, 82, 88, 95–98, 118, 131, 132 See also Politics, competitive Conservative, 13, 110 Consumption practice, 41, 62, 87 Construction reconstruction, 2, 4, 15–17, 20, 23, 25–29, 35, 53, 54, 61, 75, 76, 78, 80, 81, 88, 93, 102, 103, 105–107, 115, 119–121, 123, 129, 130, 139, 140 Contact, 12–15, 17, 19, 96, 98–101, 108, 110, 112, 122, 124, 131 Context, 5–9, 11–26, 28, 31, 43, 49–52, 54–60, 62, 63, 65–68, 72, 73, 80–82, 85, 89, 92, 93, 97–99, 101–106, 116, 118, 120–124, 129–133, 135, 137, 138 Core, 13, 14, 17, 105, 135–138 Correlating time and place, 23 Correlation, 20, 21, 23–26, 54, 64, 80, 93–99, 109–113, 123, 124, 133, 136 Bantu expansion, 21 interdisciplinary research, 21

-set, 8, 20, 23, 25, 27–30, 47, 48, 51, 52, 55, 56, 63, 64, 66, 73, 93, 107, 113, 117, 125 Dating methods, 107, 108, 113, 128, 136, 139, 140 protolanguages, 95 radiocarbon, 10, 21, 25, 52–54, 64, 135, 139 de facto refuse, 8 Demographic shift, 96 Dendrochronology, 9, 136 Deposit intentional, 65, 70 -ional, 8, 57, 65 Derivation, 15, 18, 98, 102–105, 131 Derricourt, R., 35–37, 40, 44, 48, 50, 52–54, 57, 58 excavations, 50 periodization, 54 Descendant, linguistic, 33, 34, 46 Dialect cluster, 79, 99–101, 106, 112, 120, 123, 124 (see also Language) Diaspora, 110, 111 Difference, social, 32–42 See also Distinction, social Diffusion, 21, 23, 25–27, 29, 106 See also Dispersal; Movement Direct association, 14, 20, 25, 26, 29, 31, 47, 80, 97, 101, 108, 111, 118, 125, 133, 136, 137, 139 Dispersal, 22–25 See also Diffusion; Movement Distinction, social, 76, 92, 94, 121 See also Difference, social Distribution, 5, 15, 17, 18, 23, 35, 44, 48, 52, 54, 57, 58, 61–63, 89, 91, 99, 100, 109, 113, 123, 124, 137 Dirks, N.B., 3 Durable durability, 8, 19, 131 Documentary and material records, 129

D Data archaeological, 2, 3, 6, 11, 12, 47, 48, 54–56, 58, 66, 82, 89, 94, 112, 118 linguistic, 2, 14, 20, 22, 23, 25, 47, 48, 54, 56, 72, 76, 78, 80, 88, 89, 93–95, 99, 106, 115, 118, 120, 124

E Early Tana tradition (ETT), 84, 89, 100 Eastern African Coast dating and correlations, 95 feasting bowls, 99 languages, 101 lexical innovations, 96 linguistic study, 100

[email protected]

Index

156 Eastern African Coast (cont.) Proto-Swahili innovations, 96 social competition and speech, 98 vocabulary and material cultural, 97 Easy-to-use programs, 53 Ecology, 6, 11, 24 Ehret, C., 14, 21, 22, 26, 96, 107, 108, 110, 111 Elite, 9, 22, 62, 78, 81, 91, 99, 119, 132 Empiricism, 28 empirical, 28 Empowering feasts, 82, 89, 92 Environment, 6, 14, 28, 35, 45, 139 Equifinality, 61 Ethnic ethnography, 11, 17–19, 26, 41, 45, 76, 80, 87, 88, 97, 98, 120 -ity, 63 Ethnohistorical data, 18 European colonization, 21 Evidence historical, 6, 13, 15, 26, 30, 105, 109, 110, 113, 129 linguistic, 2, 4, 5, 12, 13, 16, 18–29, 31, 36, 68, 76, 95–97, 99–101, 107, 111, 116, 117, 127, 132–134 material, 6, 127 Excavation, 6–9, 11, 31, 40, 48, 50, 51, 53–55, 57–59, 66, 83, 84, 114, 139 F Fagan, B.M., 36–39, 41, 44, 48, 50–53, 56–58, 64, 69 Fame, 44, 46, 71, 130 See also Celebrity Fauna faunal assemblages, 9, 36, 48–51, 55–60, 62 material, 63 record, 36, 40, 44, 70 Feast AD 700, 83 after AD 1000, 84, 92 before AD 1000, 90–93 18th- and 19th-century feasts, 92 19th-century Swahili, 87 ceramics assemblages and political forms, 82 competitive, 75, 90–93, 120 empowering feasts, 89 and patron-client feasts, 92 and patron-role, 82

ETT materials, 91 -ing, 30, 61, 75–95, 97–99, 104, 118–124, 131 joking and mocking relationships, 88 Pemba Island, 82 Persian Gulf and celedon bowls, 84 practices, 87 PSW, 88 rethinking, 86–89 sgraffiato bowls, 84 white-glaze bowls, 85 Fishing, 31, 33, 35, 36, 41, 42, 62, 88, 121 Food collection archaeological evidence, 31 Botatwe language, 35 butchery and consumption practices, 41 butchery patterns, 40 chila, 43, 45 fishing, 41 hunting, 41, 42 Ila language, 36 Kafue hunters, 44 Kafue people, 45 linguistic evidence, 36 linguistic geography, 31 Lukanga swamps, 42 middle Kafue and Batoka sites, 40 Proto-Central Eastern Botatwe, 33, 34 skill, 35 spearcraft, 35 spearmen, 41 and politics, Ancient Eastern Africa, 81–82 production, 30, 31, 33 Formation processes, 8 Friendship, 33, 46 Frontier, 14 See also Boundary G Game, 35, 40, 43, 45, 51, 52, 69 See also Animal; Antelope; Meat Gathering, 31, 121 Geeraerts’ prototype theory, 17 Genetic geneticist, 1 history, 1, 15 Geographical scale, 129 Geography, 6, 15, 21, 23, 24, 31, 33, 34, 41, 68, 71, 95, 97, 99–100, 110, 111, 113, 116, 130, 139 Geology geoarchaeological, 7

[email protected]

Index

157

Geophysical surveys, 7 GIS analysis, 128 Glassman, J., 76, 87, 88, 92, 97 Glottochronology, 14, 24, 32, 89, 95, 96, 107–109, 111, 137 Government governance, 119, 120, 123 Graphite, 85, 87, 90, 91, 93, 95, 96, 99, 100, 106, 121–124 Great Lakes region, 98 H Heggary, P., 20, 23, 109 Hinnebusch, T.J.., 64, 75, 78–80, 88, 89, 92, 96, 98, 100–112, 119, 120, 122, 123, 131 Hierarchy hierarchical, 76, 119, 120 Hinterland, 68, 89, 100, 105, 131 Historical linguistic data, 115, 118 Historical linguistic evidence assumption, 12 classification, 13 cognate rates, 13 core vocabulary, 13 direct associations, 14 ethnographic evidence, 19 formal/relational analogies, 19 glottochronology, 14 interdisciplinarity and comparative historical linguistics, 19 lexicostatistics, 13 morphology, 13 phonology, 13 protoforms, 16 protolanguage, 12 social context & ethnographic record, 18 and meaning, 16 speech communities, 12 subgroups, 13 synonyms and antonyms, 18 words’ construction, 18 words’ protoforms, 19 words and things approach, 15 word use, 18 Historical linguistics, 66, 67, 127, 129 after AD 1000, 75, 76 archaeological materials, 79 and archaeology, 78–81, 86–89 13th and 14th centuries AD, 78 eastern African coast, 75 evidence, 76

Islam, 76 reconstructions and archaeology, 119 social/community, 80 Historical linguists, 13, 17, 99, 107 Historical sensibility, 117 Historicity, 3, 6 History genetic, 1, 13, 15 historian, 1–3, 17–21, 26, 27, 29, 47, 58, 64, 78, 80, 87, 89, 97, 107, 111, 112, 115–117, 128, 133, 140 historicity, 3, 4, 6, 12, 129 historical data, 18, 20, 28, 95 linguistics, 1–6, 12, 13, 16, 19, 20, 26–28, 47, 48, 54, 66–68, 72, 75–95, 108, 113, 115, 118–120, 125, 127, 129, 140 source, 3, 4, 6, 116 text, 2 Horton, M., 11, 61, 62, 76, 78, 86, 91, 119, 120, 122 Household, 71, 72 Batoka integration, 71 Isamu Pati, 72 Kafue-era speech, 72 Kafue-speaking brides, 72 linguistic record, 71 regional linguistic record, 71 Hunting, 42 communal, 36, 40, 41, 43, 45, 46, 56, 97 and fishing, 42 hunter, 11, 34–36, 40–45, 61, 62, 69, 73, 130 I Identity, 11, 17, 63, 80, 82, 97 Independent historical records, 120 Indian Ocean, 76, 78, 81, 83, 91, 100, 121 Indo-European, 16, 21, 22, 24–26, 108 Andean cases, 26 languages and cultures, 22 literature, 26 Inheritance, 13, 15, 101 Innovations, 115 economic, 31 and inheritance, 101, 102 internal, 12, 13, 15, 137 political, 33 social, 12, 31, 33 word, 75, 76, 88, 92, 98, 106, 115, 118, 120, 124, 132

[email protected]

Index

158 Interdisciplinarity, 1, 3–5 anthropology, 3 archaeology, 5 historical linguistics, 5 historical research, 4 language history, 4 methodology, 3 productive tension, 3 sources, 5 Stahl’s approach, 3 Interdisciplinary argument, 4 approach, 3–5 (see also Collaboration) collaboration, 29 epistemology, 3 learning, 1, 29 project, 3 reconstruction, 4 research, 1, 4 scholarship, 1, 20 synthesis, 21 Invention of words, 4, 98 Iron early iron age, 109 middle iron age, 40, 54 (see also Smelting; Smith) Isamu Pati, 39, 40, 50–52, 55–58, 68–73 Islam, 76 Islamic practice, 121 Iteration and process, 72, 73 J The Journal of African History, 21 Jousting, verbal, 76, 92, 95, 97, 118–120, 131, 132 K Kafue hunters, 45 languages, 43, 71 language group, 23, 96, 109, 114, 137 period, 40, 72, 130 proto-, 33, 34, 40, 43, 45, 69, 71 region, 32, 33, 35, 36, 52, 71 River, 45 speakers, 42 Kalala Island, 40, 50, 53, 57, 58 Kalundu, 39, 40, 48, 50, 51 data, 51 Kin -ship, 42, 43, 46, 56, 71–73, 130, 132 (see also Marriage; Matrilineal)

Kingdom, 46, 128 See also Royal Knowledge politics of, 133 technological, 36 Kongo Kingdom, 128 KongoKing project, 128 Kuoba banyama, 43 Kusokoma, 42 Kuweza-weza, 36 Kuwezya, 36 Kwale/early iron ware, 110 See also Iron L Landscape, 6, 24, 33, 35, 41, 42, 44, 71 Language evidence, 62, 130 family, 14, 15, 23, 24, 27, 31, 33, 66, 96, 108, 135, 137–139 multilingualism, 14 protolanguage, 12–15, 21, 22, 25, 32–34, 79, 80, 95–99, 104, 106–112, 115, 129, 136–140 spread of, 23, 24 subgroup, 13, 15, 31, 32, 45, 111, 140 (see also Dialect) Large-scale hunts, 71 Learning, 128–129 Lexicon’s power, 131 Lexicostatistics, 13, 109 Lexis lexicon proto-, 15, 22, 139 lexicostatistics, 13, 89, 107–109, 138, 140 Linguistic and archaeological records, 31, 73 chronology, 89 data, 2, 14, 20, 22, 23, 25, 47, 48, 54, 56, 72, 76, 78, 80, 88, 89, 93–95, 99, 105, 115, 118, 120, 124 descendants, 34 evidence, 2, 4, 5, 12, 13, 16, 18–21, 25, 27–29, 31, 36, 68, 76, 95, 97, 99–101, 107, 111, 116, 117, 127, 132, 133 linguistics (as discipline) historical, 4 and material culture change, 124–125 paleontology, 22 source, 3–5, 17 Lithic phytolith, 6 Lived history, 2, 4, 5, 27, 28, 30 Loanword, 12, 13, 101, 105, 124, 138

[email protected]

Index

159

M Manda, 78, 90, 91, 100, 101, 121, 124 Material culture, 6, 21–26, 42, 47, 55, 78, 99, 110, 114, 124, 129, 137 object, 2, 4, 12, 28, 47, 80, 81, 91, 123, 124 wealth, 36 Matrix, 7, 135 Marriage, 16, 17, 27, 42, 43, 46, 71–73, 130, 132 See also Bride; Kin; Reproductive practice Matrilineal matrilocal, 42, 72, 73 (see also Kin) Maya cosmological and political order, 12 Mcintosh, R., 6, 7, 11, 115 Meat, 40, 41, 43, 44, 48, 52, 60, 62, 63, 130 See also Animal; Antelope; Game Memory practice, 57 Method methodology, 1, 3, 9, 90 Metal Metallurgy, 21, 33, 41 (see also Iron) Middle Iron Age, 40 Middleton, J., 76, 88, 119, 120, 122 Migration immigrant, 43, 44 (see also Newcomer) (see also Movement) Minimum number of individuals (MNI), 49, 60, 61 Morphology, 13, 15, 18, 23, 35, 79, 102, 103, 131, 137–140 Mosque, 76, 82, 83, 93, 119, 120, 122, 125, 132 See also Islam Movement mobility, 14, 112 (see also Diffusion; Dispersal; Migration) Multi-ethnic society, 62 Mutual-intelligibility, 112 Mwiini, 102, 105 N Narrative, 1, 3, 4, 6, 12, 14, 21, 29, 30, 32, 47, 57, 58, 67–69, 73, 81, 89, 95, 115–117, 127, 128, 133 Newcomer, 44 Northeast Coastal (NEC), 79, 88, 108–112, 131 Northeast Coast Bantu group, 79 Northern Tanzanian coast, 92 Number of identified specimens (NISP), 48, 49, 51, 60, 61 Nurse, D., 64, 75, 78–80, 88, 89, 92, 96, 98, 100–112, 119, 120, 122, 123, 131 Nurse and Hinnebusch’s study, 79, 89, 110

O Oliver, R., 21 P Pastoral -ism, 62 Patron -client, 93 Paths in the Rainforests (1990), 107 Pattern, 5, 9, 11, 13, 16–19, 23, 24, 40, 42, 47, 49, 51, 54–57, 59–63, 67, 87, 89, 91, 98, 99, 113, 114, 116, 117, 124, 129, 131–133 Pemba, vii, viii, 81–84, 86, 87, 90, 91, 96, 100, 111, 119, 121, 124 Performance, 19, 82, 88, 98, 104, 120–122, 131 Periphery, 2, 14 Persian Gulf, 76, 84, 124 Phonology, 13, 17, 23, 45, 79, 103, 104, 112 Phytoliths, 6 Politics competitive, 75, 93 (see also Competition) political practice, 75 Pompeii Premise, 8 Portuguese, 112, 125 Post-AD 1000 ceramic assemblages, 88 Post-AD 1000 period, 118 Postprocessual archaeology, 65 Pottery, 8, 16, 40, 54, 68, 84, 85, 89–91, 96, 97, 99, 100, 112, 114, 121, 125 See also Ceramic Power, 4, 11, 20, 26, 30, 35, 41, 42, 44, 46, 58, 69–72, 75, 76, 78, 80, 82, 87, 88, 92, 94, 95, 98, 100, 103–105, 119, 120, 122, 123, 130, 131 Practice biography of, 76, 93, 94 consumption, 41, 62, 87, 89, 130 memory, 57 political, 75 reproductive, 72, 130 subsistence, 32, 33 Prehistory, 15, 21, 24 Preservation, 5, 28, 49–52, 59, 93, 127, 129 Principle of greatest linguistic diversity, 14 of least moves, 14 Process formation, 8–10, 49, 59 Proto-Central Eastern Botatwe, 33, 40 Proto-Kafue speakers, 43 Protolanguages, 12, 96, 106, 107 Proto-Northeast Coast Bantu (PNEC), 79

[email protected]

Index

160 Proto-Sabaki (PSA), 79, 88, 96 Proto-Swahili (PSW), 76, 88, 97, 98 innovations, 96 lexicons, 132 period, 131 speakers, 98 R Radiocarbon calibration, 10 Radiocarbon (14C) dates, 9, 10, 64 calibrated, 53 date range, 53 one- and two-sigma levels, 54 periodization, 53 Regional processes, 22 Renfrew, C. anatolian hypothesis, 22 approach, 25 Refuse de facto, 8, 136 primary, 8, 138 secondary, 8, 140 Reproductive practice, 72, 130 Residue analysis, 57 Ritual, 26, 33, 41, 43, 44, 56, 57, 65, 66, 80–86, 122 Royal, 46 See also Kingdom S Sabaki languages, 79, 110 proto-, 79, 88, 96, 99, 110 Sample size, 47, 49–52, 68, 80, 113, 129, 131 Sampling strategy, 7, 9, 68 Sasanian Islamic jars, 84 Scale, 4, 5, 7, 9, 21, 23–28, 42, 44, 46, 48, 51, 54, 55, 57, 63, 64, 66–73, 78, 93, 108, 113, 114, 116, 119, 123, 127, 129, 130, 132 Schadeberg, T., 103, 105 Separate communities of economic groups, 62 Seriation, 9, 140 Settlement coastal, 94, 121 Shanga, 61, 62, 78, 91, 100, 121 Sherd, 84, 86, 87, 90, 113 Shungwaya migration narrative, 89 Silence, 2, 3, 17, 28, 30, 105, 122, 129, 132, 133 Siraf storage jars, 84

Site, 4, 6–9, 11, 35–40, 48–62, 64, 66–68, 70, 73, 77, 81, 83–87, 89–91, 93, 97, 99, 100, 105, 109, 111–114, 120, 121, 124, 129–132, 139, 140 Smelting, 13, 35, 41, 55, 67, 68, 130, 138 See also Iron Smith, 34, 43, 45, 82 See also Iron Sociality, 46, 116 Social zooarchaeology, 61 Source historical, 3–6, 17, 116 linguistic, 3–5, 17 Space spatial, 7, 40, 58 Speaking, 133–134 Spear -craft, 33–36, 40–44, 46, 56 Spear, T., 64, 78, 80, 89, 96, 107, 109–111 Speech, 12, 14, 15, 19, 20, 24, 25, 29, 31, 35, 40, 43–45, 72, 88, 89, 96, 98–100, 102, 104, 105, 110–112, 118, 122, 123, 129, 131, 132, 138, 140 Stahl, A.B., 1–5, 11, 16, 27, 57, 58, 60, 61, 65, 66, 93, 129 approach, 3, 5 distinction, 4 source, 4 supplemental approach, 2, 4 Starred form protoform, 15, 140 Stone, 6, 9, 34, 76, 92, 103, 119, 120, 122, 124, 132, 138 Stratigraphy, 9, 40, 47, 52–54, 117 Structured deposits, 65, 70 Subsistence practices, 32, 33 system, 32, 46, 115 technologies, 32 Supplemental approach, 27–29 empiricism, 28 interdisciplinary collaboration, 29 linguistic and archaeological records, 27 material objects and word, 28 verifiability, 29 words and things method, 27 Survey procedure, 7, 139 unit, 7 universe, 7, 66 The Swahili (1985), 109 Swahili case study, 108, 120, 130

[email protected]

Index

161

coast, 29, 30, 76, 78, 79, 81–86, 94, 103, 106, 110, 114, 122, 131, 132 group, 101 historical linguistic material, 79 and the Indian Ocean world, 81 kiSwahili, 75, 89, 103 language, 79 proto-, 75, 76, 79, 88, 89, 96–101, 103–106, 110–112, 121, 124, 131, 132 town, 78, 89, 92, 119, 123, 132 Symbolic and structural archaeologies, 65 Synonym, 13, 18, 102, 131, 138 Systematic pedestrian surveys, 7 T Tana, 111 Tana tradition, early, 52, 84, 85, 89, 100, 121 Tanzania, 1, 66, 76, 80, 83, 84, 86, 91, 120 Taphonomy, 59 Technique, 7, 9, 10, 16, 31, 43, 49, 58, 60, 97, 99–101, 106, 107, 115, 137 Technology, 11, 16, 20, 22, 32–34, 36, 41, 43, 46, 47, 68, 108 Tool, 6, 11, 21, 25, 31, 34, 41–44, 46, 60, 71, 97, 98, 108, 116, 138, 142 Tomb, 119, 125, 132 Town, 30, 76, 78, 79, 85, 89, 92–94, 96, 98, 101, 103, 118–120, 122, 123, 132 See also Urban Trade trading, 76 trader, 91, 131 Training, 127, 128 Triangular Incised Ware (TIW), 84, 111 Triangulation method, 26 Tri-partite system, 53 Trouillot, M.-R., 3, 6 Tumbe, 83, 84, 90, 91, 93, 100, 101, 121, 122, 132

U Unguja Ukuu, 91, 93, 100, 101, 121 Urban center, 76 living, 98 (see also Town) Upstreaming, 5 V Vansina, J., 1, 5, 14, 25, 27, 43, 80, 107, 108 Vessel, 7, 8, 57, 84–86, 90, 91, 118 Village, 9, 34, 40, 41, 44, 46, 51, 54–57, 68–72, 76, 83, 91, 93, 100, 118, 119, 121–123, 130 Verbal extensions, 105 Vocabulary, 65 core, 13, 14, 135–138 W Wari Middle Horizon, 23, 24 Wealth, 7, 35, 36, 61, 73, 78, 92, 99 Weapon, 35 White-glaze bowls, 85, 90 Wild vs. domestic fauna, 51 Witchcraft, 105 Word history, 117, 123 innovation, 4, 15, 43, 45, 63, 75, 76, 88, 92, 95, 96, 98, 101, 103, 106, 108, 115, 118, 120, 124, 125, 131, 132 ‘words and things’, 15, 18, 20, 26, 27 World War II, 21 Wylie, A., 11, 88, 93 Wynne-Jones, S., 52, 59, 75, 76, 78, 84, 90, 91, 93, 94, 113, 123 Z Zambezi, 132 Zambezi Valley, 35, 42, 71 Zanzibar, 66, 79, 100, 124 Zooarchaeological, 50, 61 Zooarchaeology, 61

[email protected]