Tracing Language Movement in Africa [online ed.] 019065757X, 9780190657574

Table of contents :
1. Toward an Interdisciplinary Perspective on Language Movement and Change / Ericka A. Albaugh and Kathryn M. de Luna

Part I: Describing and Classifying Language Movement and Change
2. Language Change and Movement as Seen by Historical Linguistics / Derek Nurse
3. The Ethnologue and L2 Mapping / Kenneth S. Olson and M. Paul Lewis
4. Understanding Distributions of Chadic Languages: Archaeological Perspectives / Scott MacEachern
5. 800 Languages and Counting: Lessons from Survey Research across a Linguistically Diverse Continent / Carolyn Logan

Part II: Forces of Fixity and Consolidation
6. Conquest and Contact in North African Languages / Moha Ennaji
7. Ajami Literacies of West Africa / Fallou Ngom
8. Vernacular Language and Political Imagination / Derek R. Peterson
9. Language Movement and Civil War in West Africa / Ericka A. Albaugh
10. How a Lingua Franca Spreads / Fiona McLaughlin

Part III: Influences on Fragmentation, Transformation, and Recombination
11. Scales and Units: Language Movement and Change in Central Africa / Kathryn M. de Luna
12. Localizing the Global: The Wanderwörter of Nineteenth-Century South Central Africa / David M. Gordon
13. The Invisible Niche of AUYL / Phillip W. Rudd
14. Language Movement and Pragmatic Change in a Conflict Area: The Border Triangle of Uganda, Rwanda, and DR Congo / Nico Nassenstein

Part IV: Traveling Remnants: African Languages and the Diaspora
15. The African Diaspora and Language: Movement, Borrowing, and Return / Maureen Warner-Lewis
16. Metaphors to Live By in the Diaspora: Conceptual Tropes and Ontological Wordplay among Central Africans in the Middle Passage and Beyond / Robert W. Slenes
17. Caribbean French-African Creole and African Metaphysics / Hanétha Vété-Congolo
18. Population Movements, Language Contact, Linguistic Diversity, Etc.: A Postscript/ Salikoko S. Mufwene

Citation preview

Tracing Language Movement in Africa Ericka A. Albaugh (ed.), Kathryn M. de Luna (ed.) https://doi.org/10.1093/oso/9780190657543.001.0001 Published: 2018

Online ISBN: 9780190657574

Print ISBN: 9780190657543

FRONT MATTER

Copyright Page  https://doi.org/10.1093/oso/9780190657543.002.0003 Published: February 2018

Page iv

Subject: Sociolinguistics Collection: Oxford Scholarship Online

p. iv

Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and certain other countries. Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America. © Oxford University Press 2018 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by license, or under terms agreed with the appropriate reproduction rights organization. Inquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above. You must not circulate this work in any other form and you must impose this same condition on any acquirer. Library of Congress Cataloging-in-Publication Data

Names: Albaugh, Ericka A., editor. | de Luna, Kathryn M., editor. Title: Tracing language movement in Africa / edited by Ericka A. Albaugh and Kathryn M. de Luna. Description: New York, NY : Oxford University Press, [2018] | Includes bibliographical references and index. Identi ers: LCCN 2017025625 (print) | LCCN 2017044859 (ebook) | ISBN 9780190657550 (updf) | ISBN 9780190657567 (epub) | ISBN 9780190641924 (ebook) | ISBN 9780190657574 (online course) | ISBN 9780190657543 (hardcover : acid-free paper) Subjects: LCSH: Languages in contact—Africa—History. | Linguistic change—Africa—History. | African languages—History. Classi cation: LCC f------ (ebook) | LCC P130.52.A35 T73 2017 (print) | DDC 409.6—dc23 LC record available at https://lccn.loc.gov/2017025625 9 8 7 6 5 4 3 2 1 Printed by Sheridan Books, Inc., United States of America

Tracing Language Movement in Africa Ericka A. Albaugh (ed.), Kathryn M. de Luna (ed.) https://doi.org/10.1093/oso/9780190657543.001.0001 Published: 2018

Online ISBN: 9780190657574

Print ISBN: 9780190657543

FRONT MATTER

Acknowledgments  Published: February 2018

Subject: Sociolinguistics Collection: Oxford Scholarship Online

This volume developed out of a symposium organized by Ericka Albaugh at Bowdoin College in September 2015. Many of the contributors participated in that symposium, including Kate de Luna, David Gordon, Carolyn Logan, Scott MacEachern, Fiona Mc Laughlin, Fallou Ngom, Ken Olson, Derek Peterson, and Hanétha Vété-Congolo. We were joined by a number of additional colleagues, including Femi Vaughan, Eileen Johnson, and Tom Spear, all of whom presented, commented, and otherwise added to the spirited conversation. These participants were tasked with teaching our small community how their discipline “worked” and, in particular, the various ways their own disciplines studied the relationship between language change and language movement. Perhaps not surprisingly for an interdisciplinary conference, we very quickly realized that we had to accept some di ering assumptions and develop some common de nitions about languages, change, causation, and movement to even begin to converse across the disciplines. This book, then, represents the teaching, learning, and indeterminate conversations that unfolded at the Bowdoin symposium. We are grateful that Moha Ennaji, Nico Nassenstein, Derek Nurse, Philip W. Rudd, Robert Slenes, and Maureen Warner-Lewis accepted our invitation to join the volume, expanding the disciplines and regions it represented. We are indebted to Salikoko Mufwene, who took on the monumental task of reading and commenting on the entire volume at several stages with a generous spirit. The conversations we have had in person and in correspondence with all of the contributors and participants have been energizing and fruitful. We thank Bowdoin College’s Dean for Academic A airs, Jen Scanlon, who saw the importance of the gathering, and for Bowdoin’s generous support through the Faculty Symposia Program. Emily Hricko facilitated the event’s organization. Bowdoin’s Department of p. viii

Government and Legal Studies provided student assistants to help with formatting the manuscript for peer review and submission. Femi Vaughan read and o ered helpful suggestions for the introduction; it is much improved as a result of his insight. Hallie Stebbins at Oxford has been enthusiastic from the start and has accommodated any number of requests—including an ever-increasing word count—with an unfailing patience. With Hannah Doyle and the rest of the team at Oxford, we have been in very good hands. We appreciate Trishula Patel’s excellent work on the index and support from Georgetown and Bowdoin for its timely completion.

Tracing Language Movement in Africa Ericka A. Albaugh (ed.), Kathryn M. de Luna (ed.) https://doi.org/10.1093/oso/9780190657543.001.0001 Published: 2018

Online ISBN: 9780190657574

Print ISBN: 9780190657543

FRONT MATTER

Contributors  Published: February 2018

Subject: Sociolinguistics Collection: Oxford Scholarship Online

Ericka A. Albaugh is Associate Professor of Government and Legal Studies at Bowdoin College and teaches on Africa, ethnic con ict, development, state-building and language politics. She has conducted eld research in Cameroon, Senegal, Ghana, Côte d’Ivoire, and Burkina Faso and has written articles on language politics, education, and elections in Africa that have appeared in International Studies Quarterly, Journal of Modern African Studies, and Democratization. Her most recent book is State-Building and Multilingual Education in Africa (Cambridge University Press, 2014). Currently, she is researching the spread of lingua francas within and across African state boundaries. Kathryn M. de Luna is Associate Professor of History at Georgetown University. She writes histories of food, emotions, and technology and publishes in the elds of history, linguistics, and archaeology. Her rst book, Collecting Food, Cultivating People: Subsistence and Society in Central Africa, was published in the Agrarian Studies Series at Yale University Press in 2016. She is currently working on two projects: a study of the changing a ective and sensory world of mineral use and mining in central Africa and a study of the relationship between mobility, climate, and environment in the middle Kafue region. Moha Ennaji is Professor of Linguistics and Cultural and Gender Studies at Sidi Mohamed Ben Abdellah University and President of the International Institute for Languages and Cultures in Fez, Morocco. He has researched and written on languages, gender, and culture in Morocco and in wider North Africa. He is director of the journal Languages and Linguistics, President of the South North Center for Intercultural Dialogue, and the author or editor of numerous books. Most recently, he edited Multiculturalism and Democracy in North Africa: Aftermath of the Arab Spring (Routledge, 2014). David Gordon is Professor of History at Bowdoin College. He received his PhD from Princeton University p. x

and writes on a range of subjects relating to

southern and central African history, including Atlantic

and Indian Ocean trading networks, British and Belgian colonialism, environmental cultures, and contested secular and spiritual sovereignties. His recent books include a history of how spiritual beliefs have in uenced human agency, entitled Invisible Agents: Spirits in a Central African History (Ohio University Press, 2012), and a collection of primary documents, Apartheid in South Africa: A Brief History with Documents (Bedford/St. Martins, 2017). Currently he is writing about the history of the Congo region over the long nineteenth century. M. Paul Lewis served as general editor of Ethnologue: Languages of the World from 2005 to 2016 and is a Senior Consultant in sociolinguistics with SIL International. His primary research and publication

interests are in language maintenance, shift, and death; language policy and planning; and language documentation. He holds an MA in Linguistics from the University of Texas at Arlington and a PhD in Sociolinguistics from Georgetown University. He did eldwork in Guatemala and has taught or consulted in many other parts of the world. Carolyn Logan is Associate Professor of Political Science at Michigan State University and Deputy Director of the Afrobarometer, a collaborative survey research project that conducts public opinion research on the quality of democracy and governance in 36 African countries. Logan received her PhD in International Relations from the Fletcher School at Tufts University in 2002. Carolyn’s research interests are in democratization and political development in Africa, especially the role of “traditional” leaders and institutions in democratization, and “citizen versus subject” attitudes among African publics. Scott MacEachern is Professor of Anthropology at Bowdoin College, in Brunswick, Maine. He holds MA and PhD degrees in archaeology from the University of Calgary. He has been involved in archaeological research projects in Cameroon, Chad, Nigeria, Kenya, Ghana, the United States, and Canada. His current research examines the evolution of political relationships around the Mandara Mountains of northern Cameroon and Nigeria, from the Iron Age to the colonial period. His main research interests are in state formation processes in Africa, the archaeological study of ethnicity and social boundaries, African cultural heritage management issues, and African and global historical genetics. Fiona Mc Laughlin is Professor of Linguistics and African Languages at the University of Florida. She has worked extensively on the phonology and morphology of Pulaar, Wolof, and Seereer, and her current p. xi

research

focuses on language contact and multilingualism in urban Africa, with a focus on Dakar. Her

translation of Boubacar Boris Diop’s novel Murambi, le livre des ossements was published by Indiana University Press in 2006, and her edited volume, The Languages of Urban Africa, appeared in 2008 with Continuum Press. Fiona’s work has been supported by fellowships from the National Endowment for the Humanities, the American Council of Learned Societies, the Camargo Foundation, and Fulbright. She has taught at the Université Abdou Moumouni in Niamey, Niger, and the Université Gaston Berger in SaintLouis, Senegal, and is a former director of the West Africa Research Center in Dakar. Salikoko S. Mufwene, the Frank J. McLoraine Distinguished Professor of Linguistics and the College at the University of Chicago, is a leading name in the elds of language ecology and evolutionary linguistics. He is author of The Ecology of Language Evolution (Cambridge University Press, 2001), Créole, écologie sociale, evolution linguistique (L’Harmattan, 2005), and Language Evolution: Contact, Competition and Change (Continuum Press, 2008). He is the founding editor of the series “Cambridge Approaches to Language Contact.” He also holds professorial appointments to the Committee on Evolutionary Biology and to the Committee on the Conceptual and Historical Studies of Science and is an a

liate of the

Department of Comparative Human Development and the Center for the Study of Race, Politics and Culture, all at the University of Chicago. Nico Nassenstein is Assistant Professor at the Department of Anthropology and African Studies at the Johannes Gutenberg-Universität Mainz. He has recently completed a grammar of Rufumbira (Uganda) with a focus on the pragmatics of con ict, language change, and border thinking. His research interests include variation and change in Congo Swahili (Western Swahili) varieties, youth language practices in Africa, and sociolinguistic variation in general. Apart from the study of Swahili, Kinyarwanda, and Lingala, he has recently developed an interest in the sociolinguistic aspects of language and (sex) tourism along the Kenyan coast. Recent publications include an edited volume Youth Language Practices in Africa and Beyond (coedited with A. Hollington, Mouton de Gruyter, 2015), and a monograph entitled Kisangani Swahili: Choices and Variation in a Multilingual Urban Space (LINCOM, 2015). Fallou Ngom is Professor of Anthropology and Director of the African Studies Center at Boston

University. His PhD is in French linguistics, and his research interests include the interactions between p. xii

African languages and non-African languages, the adaptations of Islam in sub-Saharan

Africa, and

Ajami literatures—records of African languages written in Arabic script. He seeks to understand the knowledge buried in African Ajami literatures and the historical, social, cultural, and religious heritage that has found expression in this manner. He is the founder of the online African Ajami Library (ALL) at Boston University. Another area of his work is LADO (Language Analysis for the Determination of National Origin), a sub eld of forensic linguistics. His work has appeared in several scholarly journals, including African Studies Review, Journal of Arabic and Islamic Studies, Language Variation and Change, and International Journal of the Sociology of Language. He is the author of Muslims beyond the Arab World: The Odyssey of ‘Ajamī and the Murīdyya (Oxford University Press, 2016). He has held Fulbright, ACLS/SSRC/NEH, and Guggenheim fellowships. Derek Nurse is Professor Emeritus of Linguistics at Memorial University of Newfoundland; he now lives in British Columbia. He has worked in a number of linguistic elds: historical linguistics, comparative linguistics, ethnolinguistics, language contact, language change, East African languages, Bantu languages, tense and aspect systems, and LADO. He has written, cowritten, or edited a dozen books, including The Swahili (1985), Swahili and Sabaki (1993), African Languages (2000), Bantu Languages (2003, and forthcoming), Tense and Aspect in Bantu (2008) with attention also to Niger Congo, and A Linguistic Geography of Africa (2007), among many others. He has also written over 80 chapters, articles, and web documents. Kenneth S. Olson is Senior Linguistics Consultant for SIL. His specialties are phonological theory, acoustic and articulatory phonetics, historical linguistics, and language documentation. He holds a PhD in linguistics from the University of Chicago. He has conducted eld research in the Democratic Republic of the Congo, Central African Republic, Cameroon, the Philippines, and France. Currently, he is researching rare speech sounds, particularly bilabial trills. Derek R. Peterson is Professor of History and African Studies at the University of Michigan. He is the editor of several books, including The Politics of Heritage in Africa: Economies, Histories, and Infrastructures (2015, with K. Gavua and C. Rassool) and African Print Cultures: Newspapers and their Publics in the Twentieth Century (2016, with S. Newell and E. Hunter). His most recent monograph, Ethnic Patriotism and the East African Revival (2012), won the Herskovits Award and the Martin Klein Prize. He is presently writing about Idi Amin’s Uganda. Philip W. Rudd is Associate Professor in the Department of English and Modern Languages at Pittsburg p. xiii

State University, Kansas. He has

a long-standing interest in stigmatized dialect and a research focus

on Nairobi’s urban vernacular. His publications include “A case study of the stigmatized code Sheng: The AUYL Syndrome” in Ufahamu, “The Ultimate Matthew E ect for Sheng” in Youth Linguistic Practices in Africa—Linguistic and Sociolinguistic Perspectives on Codes, Secret Languages and Speci c Writings (Springer, 2017), and (with Anne Schröder) “Language Mixing and Ecology in Africa: Focus on Camfranglais and Sheng” in Linguistic ecology and language contact (CUP, 2017). Robert Slenes is Professor of History at the São Paulo State University of Campinas (Unicamp) in Brazil. He received his doctorate from Stanford University with a thesis on “The Demography and Economics of Brazilian Slavery, 1850–1888” (1976). His subsequent research has concentrated on the social history of slavery, with an emphasis, since the late 1980s, on the experience of Central Africans in the diaspora to Brazil. His publications include several articles on this subject and a book about the slave family, focused on the provinces of São Paulo and Rio de Janeiro: Na senzala, uma or—Esperanças e recordações na formação da família escrava: Brasil Sudeste, século XIX (1999; 2nd ed., 2011). He is currently preparing a book on Central African culture and slave identity on the sugar and co ee plantations of the same region.

Hanétha Vété-Congolo is Professor of Romance Languages and Literatures at Bowdoin College. Her scholarship focuses principally on Caribbean and African ideas, philosophy, literature, and oraliture. Interdisciplinary and comparative, her work also pays particular attention to discourses by women and about women of the Caribbean and West and Central Africa. Her most recent books are L’interoralité caribéenne: Le mot conté de l’identité (Vers un traité d’esthétique caribéenne) and The Caribbean Oral Tradition: Literature, Performance, and Practice. Her poetry collections, Avoir et Être: Ce que j’Ai, ce que je Suis and Mon parler de Guinée were respectively published in 2009 and 2015. Maureen Warner-Lewis is Professor Emerita of African-Caribbean Languages and Orature at the University of the West Indies, Mona, Jamaica. She has written several books on the connections between African and Caribbean languages and cultural practices, including Yoruba Songs of Trinidad (1994); From Mother Tongue to Memory (1996, 1997); Guinea’s Other Suns: The African Dynamic in Trinidad Culture (1991, 2015); and Central Africa in the Caribbean: Transcending Space, Transforming Cultures (2003). She served on the Board of the African Caribbean Institute of Jamaica for many years. p. xiv

Tracing Language Movement in Africa Ericka A. Albaugh (ed.), Kathryn M. de Luna (ed.) https://doi.org/10.1093/oso/9780190657543.001.0001 Published: 2018

Online ISBN: 9780190657574

Print ISBN: 9780190657543

CHAPTER

1 Toward an Interdisciplinary Perspective on Language Movement and Change  https://doi.org/10.1093/oso/9780190657543.003.0001 Published: February 2018

Pages 1–22

Abstract This chapter begins the multidisciplinary conversation that will continue through the volume, beginning with the perspectives of political science and history on language movement and change. Political scientists view language alternatively through the lens of policy, as a variable a ecting other outcomes, as a product of history and individual choice, or as a normative right. Historians view language as an entity and an identity, as well as a repertoire of speech and belonging. These varying but intersecting approaches demonstrate that adopting an interdisciplinary perspective forces scholars to hold multiple views at the same time: language as an object and a subject for research, speakers as victims and agents, and language as xed and fragmented. The volume is organized around this latter tension. Common themes that run through the volume are the counting of data, the construction of boundaries, the pace of change, and the impact of power.

Keywords: Africa, language, linguistics, multidisciplinarity, interdisciplinarity, political science, history Subject: Sociolinguistics Collection: Oxford Scholarship Online

Introduction Just as languages are commonly studied as discrete units, disciplines often remain islands of research, isolated by their traditional methods, questions, and discourse. This book creates bridges between several disciplines as we look for common approaches to language change and language movement in Africa. The project began as a symposium convened by a political scientist (Albaugh) who had studied language on the continent for many years and had noted within her discipline the prevailing perception of Africa as deeply fragmented by language. Language maps perpetuate the image by representing separated groups. And yet people communicate within and across borders, because most Africans are multilingual, with facilities in many more languages than their mother tongue. They may not speak “pure” versions of any single language, but use portions of many. Moreover, individuals’ capacities change over time, responding to di erent pressures and opportunities. These realities call for a much wider concept of language than political science typically considers. Knowing that other disciplines had studied language on the continent, Albaugh gathered scholars from several elds—history, archaeology, anthropology, linguistics, and philosophy, as well as political science—to discuss how languages form and transform in Africa as speakers and languages move. p. 2

This volume is the outcome of that symposium. The collection here is empirical—aiming to represent language more accurately in Africa—as well as theoretical. The contributors identify in plain terms the theories that their disciplines use to make sense of language movement in Africa and provide empirical case studies to explore the themes that cut across all of our disciplines: how we use data, how we understand boundaries, how we trace change, and how we represent power. The volume opens up a conversation about interdisciplinary approaches to language movement and language change in Africa. This introduction, short introductions to the four parts of the volume, and the postscript by Salikoko Mufwene draw out the shared questions and approaches, but also show how the assumptions and goals of one discipline impact those of another. This is how a multidisciplinary method can lead to interdisciplinary insights. We, as editors, begin this multidisciplinary conversation from the perspectives of our own disciplines, political science and history, two elds whose engagement constitutes a bridge between the social sciences and humanities. By laying out the concerns of political science and history, we raise issues shared by the other disciplines represented in this volume, even as our methods and theoretical frameworks di er. To be sure, the scholars in this volume speak to each other as Africanists and share a common literature, even if we have lacked a multidisciplinary venue to discuss our disciplines’ mutual interest in language movement and change on the continent. Thus, we use this initial consideration of our own disciplines’ approaches as a way to identify those common threads and unifying themes that resonate through the following chapters. As we imagine our words’ reception by practitioners in other disciplines, we are compelled to reexamine our assertions and engage with novel rejoinders. Such re nement must by de nition create something new, a point to which we will return at the end of the introduction.

Political Approaches to Language in Africa Political scientists study the authoritative allocation of power and resources. They look at actions taken by the state, as well as choices by individuals, mediated through social norms and formal institutions. Groups are a frequent focus of study, since group identi cation often determines the power and resources to which people have access. Political scientists typically subsume language under the heading of ethnicity, as both are commonly assumed to de ne discrete groups. Much recent scholarly work insists, however, that groups p. 3

are not in fact distinct, but multilayered and

situational. In articulating this tension, there are four major

ways in which language is implicated in political science: language is seen as a policy; language is treated as a variable a ecting other outcomes; language identity is acknowledged as a product of history and choice; language is viewed as a right.

Language as Policy In its original form, political science as a discipline simply looked at state organs and the allocative policies they produced. Discussion of language usually involved governments’ attempt to achieve orderly administration, or “rationalization” in Max Weber’s (1968, 809) terms. Charles Tilly (1990, Ch. 2) explains rationalization as an outcome of the transition from indirect rule through intermediaries to direct contact between the state and its citizens. Household taxation, mass conscription, censuses, police systems, and, in the European context, national education facilitated this process. For example, in Eugen Weber’s study of the transformation of peasants into Frenchmen through militarization, urbanization, and education, public schools integrate citizens around national symbols, history, and a standard language (1976, 337). Europe’s national consolidation coincided with heightened colonial in uence in Africa. Achieving power over populations similarly included the mapping and demarcation of groups, an activity greatly facilitated by the missionaries that often preceded formal colonization. French and British colonial methods di ered in their treatment of African languages, particularly in education, with French policy excluding them completely and British using them in early years, re ecting rivaling ideologies of integration and separate development. Phillipson (1992) wrestles with the apparent liberal attitude of the British, compared to the French. Both achieved a “linguistic imperialism” in which African languages were devalued as inferior, an e ect that Kenyan writer Ngugi wa Thiong’o (1986) poignantly critiques as a “colonization of the mind.” Because of the strong in uence of the European experience on our understanding of state formation, there has been an expectation that African states will be similarly “rationalized,” primarily through education and language standardization. Modernist scholars such as Karl Deutsch (1953/1966) explored the bene cial nationalism that would derive from increased social communication through higher levels of education. Daniel Lerner (1958, 60) predicted that the combined forces of urbanization, literacy, and media growth p. 4

would propel the transition to an integrated,

participative society. Africanists such as James Coleman

also contemplated the socializing and integrative roles of education. Education systems could be “powerful instruments in forging national unity, in developing a common language of political communication, and in providing exposure to, if not inculcating a positive a ect for, national symbols and goals” (Coleman 1965, 227). And yet, education systems in Africa have not performed the socializing, homogenizing function that was expected. O

cial languages have not spread widely (Albaugh 2014, 61). Language policy after independence

was not a revolutionary break from colonial rule, but a continuation of its in uence. Overall, the view of language as policy attempts to capture the decision-making rationale behind government interest in language and the passage of policies prescribing or prohibiting its use. Some scholars use game-theoretic modeling to examine the strength of preferences and costs of adopting certain

policies (Pool 1991; Laitin 1988; 1992; de Swaan 2001). Others rely on large-N quantitative studies to pinpoint institutional and demographic variation that lead to certain policies toward languages (Liu 2014). Still others examine the weight of historical institutions that persist in current language regimes (Albaugh 2014; Cardinal and Sonntag 2015).

Language as a Variable Another way political scientists approach language is to treat it as causal. Ethnic diversity, commonly equated with linguistic diversity, has generally been seen as problematic for many valued outcomes, such as democratic stability, lack of violence, and economic growth. Theorists since John Stuart Mill have warned of trying to forge democracy among di erent nationalities: “Among a people without fellow-feeling, especially if they read and speak di erent languages, the united public opinion necessary to the working of representative government cannot exist” (1882 [1861], 310). Political scientists Rabushka and Shepsle (2009 [1972], 92) argued in a similar vein: “The plural society . . . does not provide fertile soil for democratic values or stability.” Africa’s tremendous ethnolinguistic fragmentation has been blamed for its poor growth, because multiple groups are assumed to make agreement over public goods more di

cult (Easterly and

Levine 1997; Rodrik 1999; Keefer and Knack 2002). While lay observation equated higher diversity with violence, careful qualitative work identi ed certain con gurations as more prone to con ict (Horowitz 2000 [1985], 37–38), and quantitative studies (e.g. p. 5

Fearon and Laitin 2003)

found the correlation between diversity and con ict di

cult to establish. This

partly came down to a problem of group measurement, and much e ort has been expended trying to be more precise. Virtually all quantitative work that examines diversity in Africa has relied to some extent on a measure created by Soviet researchers in the 1960s (Atlas narodov mira/Atlas of Peoples of the World), which estimated the relative size of ethnic groups within countries worldwide. This was used to create an “index of ethnolinguistic fractionalization” or ELF (Taylor and Hudson 1972), a numerical index that ranged from 0 to 1 and allowed quantitative comparison and analysis. Some argued that divisions should be disaggregated into di erent types of groups, most notably ethnic, religious, and linguistic (Alesina et al. 2003). It was an Africanist scholar, Daniel Posner, however, who challenged fractionalization indices most directly, noting that in many African countries with a high number of languages or ethnic groups, only a few are politically salient at any given time. He proposed instead a measurement of “politically relevant” ethnic groups, which, importantly, could change over time (Posner 2004). Scholars beyond Africa have taken up the challenge to more accurately represent the politically active groups, adding measures of group access to state power with the Ethnic Power Relations dataset and subsequent geospatial variants (Wimmer et al. 2009; Wucherpfennig et al. 2011). While precision and multidimensionality has improved scholars’ ability to study ethnicity or language as factors in uencing democracy, violence, and growth, it is also the case that language is itself an outcome.

Language as a Product The third way, then, that language enters political science scholarship is as a product. Recent literature has grappled with the constructed nature of groups and the institutional arrangements that may alter their composition and actions. Scholars working in this vein are unrepentant disciplinary appropriators, freely borrowing from anthropology, sociology, and psychology. This has produced a much more uid picture of language and ethnic identities, which can be di

cult to reconcile with other approaches.

Anthropologist Fredrik Barth (1969) set the foundation for much of this work, with his insight that ethnic groups were not static clusters of cultural content, but instead emerged and were reinforced when groups in contact classi ed themselves and others. Building on this, sociologist Rogers Brubaker (2004) challenged a

p. 6

xation on “groups,” wanting instead to account for when “groupness” crystallizes into intense activity, and

when it does not. The state plays a central role in classifying citizens, and individual cognitive

processes reinforce social categories. Political scientist Donald Horowitz (2000 [1985]) similarly identi ed general psychological tendencies to cleave and to compare, the stereotypes that are reinforced by authoritative a

rmation, and the depth of feeling that accompanies the striving for recognition and

esteem. This nuanced understanding of ethnicity’s uidity yet potential intensity infuses much work on Africa (Young 1976; Laitin 1986; Newbury 1988; Mamdani 1996; 2002; Posner 2003). Variation in intensity seemed to involve codi cation of language. Crawford Young asserted that “mobilized identities alter less readily than do unmobilized ones,” suggesting that a written language made much of the di erence: “the move from the oral repository of the traditional elders to the written page multiplies the potential mobilization of identity” (1976, 45, 121). Posner (2003) identi ed colonial transcription as a central variable in the transformation of Zambia’s linguistic cleavages. More instrumentally, Posner explored the power of political institutions to induce the formation of di erent social divisions. Individuals’ reactions to institutions and their environment—adaptation, choice, and demand—complement the exclusive focus on state actions that may shift language identities. Generally, individuals learn languages with higher value, in economic terms and prestige. While certainly material advancement is crucial, the power and status associated with a language also play a central role (Bourdieu 1982; Edwards 1985). The value of language capacity and the relative prestige such capacity a ords is determined by what is socially valued. The colonial imprint in Africa continues to reverberate when individuals prefer to learn European languages. This provides a bridge to the nal area in which political science intersects with language: rights.

Language as a Right Contemporary to the modernist scholars of the 1960s focusing on national integration, another group of sociolinguists and political scientists (e.g. Fishman, Ferguson, and Das Gupta 1968) aimed instead to preserve or resurrect threatened languages. Though these scholars conceded that much language change is not planned, they were con dent that deliberate attempts to in uence the social use and status of language would be successful (Ferguson 1977, 9). As they readily admitted, the new states of Africa and Asia became a laboratory for language planning. Concerns with language revitalization have become more prominent for Africa over the last two decades. p. 7

Part of this may be due to timing. Since the democratic wave arrived in Africa at the same time as several con icts escalated, many experiments in democracy were forged in post-con ict situations. Popular accounts attributed con icts to deep ethnic divisions, and institutional solutions seemed to follow this narrative. Achieving peace seemed to require guarantees for excluded groups. Lijphart’s (1977) ideal type of power sharing included cultural autonomy and group recognition, recommendations that dovetailed nicely with the growing concern of international organizations and activists in the 1990s with cultural and linguistic rights. Many of these scholars question the presumption that homogeneous nations are a natural endpoint toward which states should aspire. A prominent scholar of language rights, Will Kymlicka (2001, 2), argued that “we cannot simply take for granted that it is legitimate for a liberal-democratic state to pressure minorities into operating in the majority language.” Stephen May (2012, 187) equated the “ ‘philosophical matrix’ of a common language and culture that underlies nation-state formation” with a “longstanding ‘ideology of contempt’ toward minority languages.” Drawing from theorists of nationalism such as Anthony D. Smith (1986), May strongly criticized the dominant view that nation-states must necessarily forge a national language and culture through mass education. Language rights scholars typically champion linguistic rights

uncompromisingly (Skutnabb-Kangas and Phillipson 1994), but also do so more cautiously (Laponce 1987; Van Parijs 2011). Many of these writers are Europeans or Canadians focusing on the “glottophagie” (Calvet 1979) or “linguicide” (Phillipson 2008) brought about by the global spread of English. This concern seems most relevant to long-established western democracies, and the debate may not resonate similarly in the relatively young states of Africa. In situations of tremendous multilingualism, the alternatives to the homogeneous nation state might be more nuanced. To arrest current language movement seems very much to privilege the present and ignore long historical precedent. Through all, the tension between uidity and xity remains. As policy and product, language is treated in political science as changeable—ripe with possibilities for engineering or the contingent outcome of many historical forces. As variable and right, language is treated as much more xed—an in uence on other outcomes or a timeless treasure to be protected. Political science demonstrates within itself the tension this volume aims to illuminate. While political scientists have borrowed freely from other disciplines to support their research, a more intentional dialogue with other disciplines may help to break through some of the eld’s impasse.

p. 8

Historical Approaches to Language in Africa It bears stating explicitly that, like political scientists concerned with policy, rights, variables, and identity politics, historians most often study language change and movement as a lens through which to understand other historical processes. Historians’ approach to language has done much to illuminate the invention of new ethnicities, the development of colonial sciences, or the ways in which literate Africans’ participation in the e orts of the colonial state shaped the reach, tenor, and e

cacy of state power. Even when language

change and movement are the subjects of study among historians employing the methods of comparative historical linguistics to produce language classi cations, studies of sound change, and the reconstructed vocabulary of protolanguages, such products of the comparative method are valuable because they illuminate some other historical problem, such as changes in the subsistence practices or social organization of early Bantu-speaking communities. In this section, we explore a few of the many ways historians have approached language change and language movement before considering how parallels and di erences between the disciplines of political science and history open up common concerns across the disciplines studying the movement of African languages, approaches that might bene t from more robust interdisciplinary awareness.

Language as an Entity and an Identity Historians sometimes study language change and movement directly and sometimes through heuristics like “discourse” or “protolanguages” and their “reconstructed words.” But, like political scientists, linguists, anthropologists, and others, historians must grapple with the legacy of a European conceptualization of languages as autonomous, distinct, systematic, rule-bound, enumerable entities that was rst developed during the Enlightenment. With the rise of European nation-states, this conceptualization of languagesas-entities shifted slightly and languages came to be understood as entities related to other seemingly natural phenomena like race, ethnicity, nationality, and even subsistence (Anderson 1983; Bauman and Briggs 2003; de Luna 2016; Fabian 1991; Foucault 1970; Joseph 2004; Makoni and Pennycook 2005). Not surprisingly, it is this European conceptualization of language as related to identity that saturates the documents that came to constitute important archives for writing African history. As scholars working in a p. 9

eld that developed out of

European empiricism, Africanist historians are the inheritors of this

intellectual legacy but also study its e ects on interactions between Europeans and Africans, who, of course, had their own ideas about the politics of distinct ways of speaking.

Language has been a particularly important lens through which historians have studied and contributed to the politics of identity in Africa. To give one impactful example, Nurse and Spear (1985; see also Spear 2000) applied the method of comparative historical linguistics to languages of the Sabaki group to trace language change and movement on the Swahili coast of eastern Africa. Paired with other forms of data, including oral traditions and archaeology, Nurse and Spear used the study of language to make the case that Swahili—the people, their language, and their coastal culture—were products of the continent, rather than created wholesale by Persian or Arab immigrants to the East African coast. Contact, movement, and borrowing are integral to the history of the Swahili coast, as Nurse and Spear themselves suggest. Nurse and Spear’s achievement was nothing short of turning on its head the presumption that loanwords, scripts, and other attributes of language from the Indian Ocean world were the remnants of a civilization forged by outsiders. Signi cantly, they could make this argument in part by mobilizing the idea that language indexes identity; if Swahili was an African language, it was one form of evidence (among many put forward by Nurse and Spear) that Swahili culture should be attributed to Africans. More often, however, historians have tracked language change and language spread in order to understand how Europeans and Africans alike mobilized the European idea that language was conjoined with innate identities, often simultaneously but with very di erent goals. Europeans marshaled this way of thinking with language as a “tag” for identities like ethnicity, race, and nationality in e orts to categorize and govern African societies. Work in the 1980s and 1990s on the “invention of tradition” suggested that this process unfolded across colonial Africa, sometimes with roots in “scienti c” studies undertaken in the last decades of the precolonial period. The documentation, translation, and even creation of African languages were key parts of the process of inventing African tribes in the colonial era (Ranger 2015 [1983]; see also Spear 2003; compare discussions of ChiShona: Chimhundu 1992; Jeater 2002; 2006; Makoni 1998). Perhaps the most notorious example of the con ation of language and naturalized identities relates to the study of Bantu languages and the demographic history of Bantu populations in by South African “scientists” (Dubow 1995). Whites working under the guise of scienti c research sought a biological p. 10

basis for racism;

language was a key component in identifying and classifying populations onto which identities like “Bantu” were forced. In a critique of the invention of tradition, historians have recovered rich stories of Africans’ participation in this process. Like Europeans, Africans in the early 20th century also crafted new traditions out of longerstanding ideals, forms of speech, political associations, and regional cultures; a discursive approach was vital to this historical insight (Feierman 1990; Spear 2003; Vail 1989). Sometimes this “neo-traditionalism” was a strategy for making claims on state and church institutions. Other times, as was the case with the mines of South Africa, Africans mobilized preexisting ethnolinguistic categories as they created cultures of belonging and work in a context of extreme mobility and linguistic diversity (Harries 1987, 1994). But, as James Pritchett has beautifully illustrated through narratives of migrant labor among a cohort of Lunda men, these identities were situational, temporary, and pluralistic, bound to places and people connected by the travels of the individual becoming Shangaan in South African mines, “deep” Lunda in Angola, Luba in the Lubumbashi mines, and Bemba on the Copperbelt (Pritchett 2007, 238). Sometimes, as was the case with the creation of Kriolu, such languages, cultures, ties of birth, and networks of mobility and trade were crafted with Europeans or used by them (Green 2012 summarizes recent scholarship on Kriolu). This strategy of emphasizing the relationship between language and identity was also turned around by Africans and applied to European languages. Africans and Europeans alike promoted the relationship between nationalism and French in parts of French West Africa (Chafer 2001; Wilder 2005). Importantly, the most recent historical scholarship has demonstrated that the relationship between ways of speaking and the changing logics of belonging did not always follow the crude logics of instrumental motives (Glassman 2000).

Long before Europeans brought their understanding of language to their dealings with the continent’s residents, Africans had similarly mobilized languages (as entities) in the service of identify politics. These politics of speech were not erased by European ideas, as was demonstrated by the “neo-traditionalist” historiography discussed above. Sometimes, in parallel to European ideas, language marked identities that Africans imagined as ascribed. New work on race has helped us understand the role of language change, language spread, and their mechanisms (including literacy) in identity politics (Hall 2011) in earlier periods. But attention to the politics of language in earlier periods or in institutions with earlier roots also illuminates ways in which Africans politicized speech without necessarily thinking about languages as entities.

p. 11

Repertoires of Speech and Belonging Approaches that foreground formal languages or manipulated discourses miss other ways of thinking about how speech was used by Africans to mediate boundaries. For example, protected, restricted vocabularies and modes of speaking are an important component of the specialized knowledge of both secret societies and castes (Storch 2011) or occupational professions (consider loanwords in Tamari 1991). Anthropologists, linguists, and art historians have taught us much about the role of secret and sacred languages in West African secret societies, such as Poro (Bellman 1984). Similarly, Patrick McNaughton (1982) has described both the way secrecy is spoken about in Mande societies and the role of particular formulas of secret speech (kilisi) as one among many ingredients in “recipes” or “packets” of goal-oriented knowledge (daliluw) used to create the power necessary to carry out acts, cures, diagnoses, and so forth (1982, 488). As cosmological devices to solve problems, daliluw are common, but they are divided into di erent varieties. Some daliluw are “virtual blueprints for the creation of things” and are frequently employed by particular occupations groups, including blacksmiths (McNaughton 1982, 488). As Didier Goyvaerts (1996; see also Rudd, this volume) demonstrates with the “anti-language” Kibalele in Bukavu, some forms of speech combine or confuse other local languages in order to both indicate membership and hide illicit activities. Some historians have similarly focused directly on fragments of languages, particularly as they connect to specialists’ activities, trade, and mobility (de Luna and Gordon, this volume) or the networks of specialized knowledge connecting disparate ports in the Atlantic (Slenes and Warner-Lewis, this volume; among others: Sweet 2003; 2011). Forms of communication among professionals or members of secret societies or with spirits don’t meet the typical de nition of a full and distinct language. Here, we can recognize how European thinking about language as nite, bounded, encompassing grammar and vocabulary, and as associated with an entire “people” shaped the scales and units of research on the politics of language in the African context.

Misfits: A Return to Language and Identity Perhaps not surprisingly, Europeans often “read” a common ethnic identity onto specialist groups who marked themselves (in part) through language. Perhaps all the more telling is the confusion that ensued p. 12

when groups and ways of speaking did not overlap as expected. For example, travelers,

colonial o

cials,

and scholars long wondered why some hunter-gatherers in central Africa spoke Bantu languages. Bantu languages, it was believed, were to be associated with farmers who had moved into the central African forests several millennia ago, encountering autochthonous hunter-gatherer groups who, it was presumed, must have spoken non-Bantu languages. Some historians and anthropologists suggested that contemporary ties of clientage linking farmer-patrons to hunter-gatherer-clients obtained in the deep past as well, leading to the loss of hunter-gatherers’ original, non-Bantu languages (e.g. Vansina 1990). Kairn Klieman (2003) argues instead that those hunter-gatherer communities speaking Bantu languages today are not the descendants of the original, autochthonous hunter-gatherers encountered by the rst Bantuspeaking immigrants. Rather, they are the descendants of Bantu-speakers who elected to adopt a foraging lifestyle around the tenth century to meet demands for forest products generated by the extension of longdistance trade networks across the continent.

Language beyond the Politics of Identity This discussion has foregrounded the relationship between language and the politics of identity because this was an important paradigm for historical scholarship on language change and movement. But a range of motives inspired choices shaping language change and movement. The rich literature on the transformation and spread of Arabic provides a wealth of examples that connect the historian’s interest in language to a host of additional topics beyond but often related to identity politics, including: 1) changing mechanisms of communication, from the spread of literacy through education, to works of translation and documentation, to the interface of various media; 2) religious conversion and the embodied experience of religious practice; and 3) long-distance trade, to name a few (within a vast literature, recent works include Glassman 2011; Hall 2011; Lydon 2009; de Moraes Farias 2003; Ware 2014). Many of these topics obtain in other parts of the continent as well.

Multiple Disciplines on Language in Africa The approaches of political scientists and historians o er several points of departure to initiate this volume-long multidisciplinary conversation on the study of language change and movement in Africa within the context of our common concerns as Africanists. Political scientists and historians are interested p. 13

in language change and movement as consequences of social

acts. Both disciplines often accept the

ction of language as a bounded, single entity in order to use it as an explanation for outcomes of interest (such as economic growth, democracy, cultural boundaries, or nationhood) or an explanation for historical processes (such as the invention of new social groups or strategies of political resistance). This is a simple point, but an important one. In fact, we share with other humanists and social scientists the tendency to study language in order to understand other kinds of interactions, but di er in the degree to which we foreground language itself as the subject of inquiry. We begin our interdisciplinary conversation on language change and language movement, then, with a common standing as students of language politics. Part of the reason that we focus on language as a way to see something else has to do with the legacy of an Enlightenment conceptualization of language. Our interdisciplinary endeavor may simply make us more honest in admitting when we are perpetuating the ction that languages are discrete, bounded entities. There may be times when the ction is justi ed for illuminating something more clearly, and others when the misrepresentation elides more than it reveals. Therefore, it is important for us to adapt our theories,

models, and questions to new ways of thinking about the de nition of “language.” Putting disciplines in conversation brings scholars to the point of productive tension—where multiple ideas are kept in play. In the eld of linguistics, for example, the integrationist school argues that linguists should reject the idea that signs are arbitrary and linear, that words have meanings, that grammar has rules, and, signi cantly, that “languages” exist (Harris 1990, 45). In contrast, the status of language as an “entity” is crucial to political scientists concerned with language rights, who often seek to preserve and protect minority languages against absorption by European languages. Historians’ interest in how pieces of language survive and recombine (an interest shared by linguists and other humanists) brings to the conversation hybrids, creoles, and new forms of language that people invent, as many chapters in this volume suggest. What do these units of analysis o er those who, like political scientists, labor under the myth of languages as entities? Which questions will be answered and which lost by subjecting hybrids or creoles to the same analysis applied to languages—perhaps classifying them in order to count speakers and observe their political behavior? Interdisciplinary conversations about language change and language movement in Africa bring to the fore the question of what “counts” as a language and the stakes of thinking through languages as the primary unit of analysis. Problematizing the analytical units we use to study the politics of di erent ways of speaking necessarily p. 14

opens new ways of thinking about

agency. Historians’ work on the “invention of tradition” informed

political scientists’ recognition of the role of colonial administrators and missions in xing identities. The reaction to the arti cial notion that placed all the power (and blame) in the hands of outsiders was to highlight the agency of Africans themselves in using ethnicity and language as powerful strategies to counter the hegemonic aims of the colonizers. Later, the study of Africans’ agency complicated the binary framework of insider/outsider by considering the hierarchies of authority shaping the practice of power within African communities. What is called a minority language today might have dominated others historically and survives to claim victimhood at the hands of a new invader. Some political scientists similarly recognize elite instrumentalism and individual choice in claiming identities, often drawing from historians’ accounts. But fragmenting the “language” and tracking the change and movement of di erent units of speech may well foreground di erent actors and other scales of agency beyond those who claim the authority or responsibility to recognize, police, and resist language boundaries, orthographies, policies, and so forth. The simple juxtaposition of approaches common to the study of African language change and movement in the two disciplines political science and history (and the many other disciplines on which they draw) reveals that interdisciplinary approaches to language force us to hold multiple views at the same time: language as an object and a subject for research; speakers as victims and agents; language as xed and fragmented. It is just this multiplicity, this web of intersections and gaps we hope to capture in the organization of the volume and the shared themes we have asked authors to consider.

Themes and Organization of the Volume The tension between languages as xed or fragmented is a compelling problem to explore in the context of the great linguistic diversity for which the African continent is famous. As contributors to this volume nd themselves closer to one pole or the other, four key themes unite their approaches: Data—All of the contributions touch on how we use data: the di

culty of collecting it, its patchy nature, and

what we consider evidence, from oral surveys to pottery fragments and photos. Many contributors observed the tension that arises between striving for accuracy and accepting simpli cation. p. 15

Boundaries—Our historians noted that the colonial association of language with a territory continues to reverberate in the way we draw maps and craft survey questions. A uid notion is more precise, with

languages connected to people, rather than to space. Complicating matters even further is the question of which linguistic units, capacities, and practices should be studied and mapped. Change—Many of the contributions wrestle with the shifting nature of boundaries in maps, and how quickly the lines become obsolete. Normative advocacy to preserve languages in their present form seems to contradict the historical record of constant change, because speakers are always inventing new ways of communicating. Writing and transcription x languages and thereby preserve them, but also force speakers of other varieties to adapt. Power—Many chapters note the power of an administrator, a missionary, or a researcher to de ne the reality s/he is observing. Di erentiation is necessary to classify, but also has the potential to distort and privilege. The state has awesome power to alter language capacities through schooling and conscription, but it does not always choose to do so. Individuals’ capacity to learn languages and employ them for personal advancement, for protection, and for rebellion must be considered in any exploration of power. Although these four core themes unite the chapters that follow, we have not organized the book to address them directly. Nor have we organized the volume by region or time period, as is common in edited volumes on Africa. We found organization by discipline to hinder conversation, though we asked all authors to provide a short explanation of their discipline’s theoretical and methodological approaches to language movement and change to ensure that the chapters are accessible to nonspecialists. Instead, we have arranged the chapters to re ect di ering conceptions of language that became evident in our interdisciplinary discussions: tendencies to study changes that consolidate language or those that splinter it, viewing languages as whole or in part. Each author conceptualized language movement somewhere along a continuum of xation or fragmentation. We begin the volume with a collection of chapters that take for granted languages as units of study, providing a linguistic view of the continent from perhaps more familiar disciplinary vantage points. Then we present collections of chapters that take an approach to language that is increasingly more fragmented, concluding with a collection of chapters that track bits and pieces of p. 16

discourse, grammar, and lexis throughout Africa and across the Atlantic

into the Diaspora. Brief

introductions to each grouping of chapters pull out the contributions of di erent authors to our ongoing multidisciplinary conversation. The postscript by Salikoko Mufwene re ects this continuum, as it highlights the multiple ways that populations, and therefore languages, move, generating constant linguistic change. At the same time, he notes the remarkable resilience of Africa’s indigenous vernaculars relative to European colonial languages, a xity where the expectation was fragmentation. In grouping chapters into units that are closer to one pole or the other of the continuum between language xedness and fragmentation, we hope the experience of reading the volume sparks insights that are only available by jumping between case studies that elucidate the same core themes and similar approaches to language change and movement in di erent time periods, regions, and disciplines.

The Flux of Language: Concluding Thoughts on New Openings Beyond linguistics, traditional approaches in the other disciplines represented in this volume have tended to use language as an index or evidence for identity, power relations, trade, contact, or meaning. In contrast, linguistics, as a discipline, could and did focus on the study of language itself. But, as interdisciplinarity increased over recent decades, insights from beyond linguistics inspired new approaches like sociolinguistics and cognitive linguistics in that eld. And, as the chapters in this volume illustrate, other discipline’s practitioners began to consider the motion and divisibility of language—the particular characteristics that make language di erent as an index, as evidence, or as constitutive of other dimensions of life, that might be compared to, for example, material culture. From language’s parasitic qualities (Mufwene 2001; and already implicit in the comparative method as described by Nurse, this volume) to the variety of media that can record it, the qualities of language matter for how we and the communities we study use it. Sounds carry the capacity to be like something else (ideophones and onomatopoeia), but they can also sound alike, allowing for puns and mistakes. Register, dialect, accent, and other di erences both make meanings and ensure that thoughts get lost in translation, opening space for contestation and change. Morphology and grammar inspire the grafting and compounding of old parts into surprising new wholes, the novelty of which is sometimes masked for political reasons. Above all, language is de ned by its plasticity. p. 17

These particular qualities of language give power to the speakers we study, facilitating but not dictating change, struggle, and even continuity. The malleability of languages shapes many of the approaches to language movement and change pursued by nonlinguists and linguists alike in the chapters that follow. Like an image captured by a camera, language can be seen as an entity, enabling comparison across time and space, but of course the assembled characters change the instant after the click. This motion is unwieldy to control, study, and understand, but it provides the linkages between the still images. Tracing fragmentation requires attention to xity. After all, fragments must be of something, and they are powerful because they can be part of the becomings of something more. Authors might emphasize one pole or the other in the continuum between xity and fragmentation, but they nd greatest insight when they keep both in sight at the same time. The chapters that follow address a range of disciplinary problems, but, taken as a whole, they raise further questions. If the qualities of language were rst studied by linguists and subsequently borrowed by other disciplines’ practitioners, what contributions do those borrowings and borrowers make in return to the discipline of linguistics? Perhaps philosophers are best positioned to insist on the moral potency of human behavior. Political scientists’ endeavors to compare, explain, and predict might blunt categories we know as complex even as they illuminate possibilities for change. Historians and anthropologists might best argue for the signi cance of the contingencies of context over the ordering power of rules and patterns. How do such conversations across disciplines become “interdisciplinary”? Language is supple, pliable, powerful in the mouths of individuals but ultimately uncontrolled by a single speaker. It is a whole greater than the sum of its parts. Might interdisciplinarity become, like the creative utterances studied here, something more than the sum of its parts? What new questions can only be asked when grounded in the vigorous demands of disciplines and explored through the promise of interdisciplinary research? We explore these questions and o er new ones in the brief introductions to each part of the volume. Many are revisited in Salikoko Mufwene’s postscript. The collection of chapters that follow, their common themes, and their relationships to one another reimagine the African linguistic landscape from fragmented and rigidly demarcated to linked, overlapping, and continually in motion. This reimagining has broad implications for the global study of language. Reactions against a Eurocentric modernization paradigm, with the monolingual nation-state as the

p. 18

ultimate goal, tend to

produce categorical alternatives that are equally Western-derived. Taking Africa’s

experience seriously demonstrates not the victimhood of languages and the people who speak them, but adaptation and resilience, though perhaps under di erent guises and in novel combinations.

References Albaugh, Ericka. 2014. State-Building and Multilingual Education in Africa. New York: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Alesina, Alberto, et al. 2003. “Fractionalization.” Journal of Economic Growth 8: 155–194. Google Scholar WorldCat Anderson, Benedict. 1983. Imagined Communities: Reflections on the Origin and Spread of Nationalism. London: Verso. Google Scholar Google Preview WorldCat COPAC Barth, Fredrik. 1969. Ethnic Groups and Boundaries: The Social Organization of Cultural Di erence. Boston: Little, Brown. Google Scholar Google Preview WorldCat COPAC Bauman, R., and C. Briggs. 2003. Voices of Modernity: Language Ideologies and the Politics of Inequality. Cambridge: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Bellman, Beryl. 1984. The Language of Secrecy: Symbols and Metaphors in Poro Ritual. New Brunswick: Rutgers University Press. Google Scholar Google Preview WorldCat COPAC Bourdieu, Pierre. 1982. Ce que parler veut dire: Lʼéconomie des échanges linguistiques. Paris: Fayard. Google Scholar Google Preview WorldCat COPAC Brubaker, Rogers. 2004. Ethnicity without Groups. Cambridge: MA: Harvard University Press. Google Scholar Google Preview WorldCat COPAC Calvet, Louis-Jean. 1979. Linguistique et colonialisme: Petit traité de glottophagie. Paris: Payot. Google Scholar Google Preview WorldCat COPAC Cardinal, Linda, and Selma K. Sonntag. 2015. State Traditions and Language Regimes. Montreal: McGill-Queenʼs University Press. Google Scholar Google Preview WorldCat COPAC Chafer, Tony. 2001. “French African Policy in Historical Context.” Journal of Contemporary African Studies 19 (2): 165–182. Google Scholar WorldCat Chimhundu, H. 1992. “Early Missionaries and the Ethnolinguistic Factor during the Invention of Tribalism in Zimbabwe.” Journal of African History 33: 87–109. Google Scholar WorldCat Coleman, James, ed. 1965. Education and Political Development. Princeton, NJ: Princeton University Press. Google Scholar Google Preview WorldCat COPAC de Luna, Kathryn. 2016. Collecting Food, Cultivating People: Subsistence and Society in Central Africa. New Haven: Yale University Press. Google Scholar Google Preview WorldCat COPAC de Moraes Farias, P. F. 2003. Arabic Medieval Inscriptions from the Republic of Mali: Epigraphy, Chronicles and Songhay-Tuareg History. Oxford: Oxford University Press. Google Scholar Google Preview WorldCat COPAC de Swaan, Abram. 2001. Words of the World. Malden: Polity/Blackwell. Google Scholar Google Preview WorldCat COPAC

Deutsch, Karl. 1953. Nationalism and Social Communication. Cambridge, MA: Technology Press. Google Scholar Google Preview WorldCat COPAC Dubow, Saul. 1995. Scientific Racism in Modern South Africa. Cambridge: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Easterly, William, and Ross Levine. 1997. “Africaʼs Growth Tragedy: Policies and Ethnic Divisions.” Quarterly Journal of Economics 112: 1203–1250. Google Scholar WorldCat Edwards, John. 1985. Language, Society and Identity. New York: Basil Blackwell. Google Scholar Google Preview WorldCat COPAC Fabian, Johannes. 1991. Language and Colonial Power: The Appropriation of Swahili in the Former Belgian Congo, 1880–1938. Berkeley: University of California Press. Google Scholar Google Preview WorldCat COPAC p. 19

Fearon, James D., and David D. Laitin. 2003. “Ethnicity, Insurgency and Civil War.” American Political Science Review 97 (1): 75–90. Google Scholar WorldCat Feierman, Steven. 1990. Peasant Intellectuals: Anthropology and History in Tanzania. Madison: University of Wisconsin Press. Google Scholar Google Preview WorldCat COPAC Ferguson, Charles. 1977. “Sociolinguistic Settings of Language Planning.” In Language Planning Processes, edited by Joan Rubin et al., 9–29. New York: Mouton. Google Scholar Google Preview WorldCat COPAC Fishman, Joshua, Charles Ferguson, and Jyotirindra Das Gupta, eds. 1968. Language Problems of Developing Nations. New York: John Wiley & Sons. Google Scholar Google Preview WorldCat COPAC Foucault, M. 1970. The Order of Things: An Archaeology of the Human Sciences. New York: Vintage. Google Scholar Google Preview WorldCat COPAC Glassman, Jonathon. 2000. “Sorting Out the Tribes: The Creation of Racial Identities in Colonial Zanzibarʼs Newspaper Wars.” Journal of African History 41: 395–428. Google Scholar WorldCat Glassman, Jonathon. 2011. War of Words, War of Stones: Racial Thought and Violence in Colonial Zanzibar. Bloomington: Indiana University Press. Google Scholar Google Preview WorldCat COPAC Goyvaerts, Didier. 1996. “ʻKibaleleʼ: Form and Function of a Secret Language in Bukavu (Zaire).” Journal of Pragmatics 25: 123– 143. Google Scholar WorldCat Green, Toby. 2012. The Rise of the Atlantic Slave Trade in Western Africa, 1300–1589. Cambridge: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Hall, Bruce. 2011. A History of Race in Muslim West Africa, 1600–1960. Cambridge: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Harries, Patrick. 1987. “The Roots of Ethnicity: Discourse and the Politics of Language Construction in South Africa.” African A airs 86: 25–52. Google Scholar WorldCat

Harries, Patrick. 1994. Work, Culture, and Identity: Migrant Laborers in Mozambique and South Africa, c. 1860–1910. Portsmouth: Heinemann. Google Scholar Google Preview WorldCat COPAC Harris, R. 1998. Introduction to Integrational Linguistics. London: Pergamon. Google Scholar Google Preview WorldCat COPAC Horowitz, Donald. 2000 [1985]. Ethnic Groups in Conflict. Berkeley: University of California Press. Google Scholar Google Preview WorldCat COPAC Jeater, Diana. 2002. “Speaking like a Native.” Journal of African History 43: 449–468. Google Scholar WorldCat Jeater, Diana. 2006. Law, Language, and Science: The Invention of the Native Mind in Southern Rhodesia, 1890–1930. Portsmouth: Heinemann. Google Scholar Google Preview WorldCat COPAC Joseph, John. 2004. Language and Identity: National, Ethnic, Religious. New York: Palgrave Macmillan. Google Scholar Google Preview WorldCat COPAC Keefer, Philip, and Stephen Knack. 2002. “Polarization, Politics and Property Rights: Links Between Inequality and Growth.” Public Choice 111: 127–154. Google Scholar WorldCat Klieman, Kairn. 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. Google Scholar Google Preview WorldCat COPAC Kymlicka, Will. 2001. Politics in the Vernacular: Nationalism, Multiculturalism, and Citizenship. Oxford: Oxford University Press. Google Scholar Google Preview WorldCat COPAC Laitin, David D. 1986. Hegemony and Culture: Politics and Religious Change Among the Yoruba. Chicago: University of Chicago Press. Google Scholar Google Preview WorldCat COPAC Laitin, David D. 1988. “Language Games.” Comparative Politics 20 (3): 289–302. Google Scholar WorldCat Laitin, David D. 1992. Language Repertoires and State Construction in Africa. New York: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Laponce. Jean. 1987. Languages and Their Territories. Toronto: University of Toronto Press. Google Scholar Google Preview WorldCat COPAC Lerner, Daniel. 1958. The Passing of Traditional Society. Glencoe: Free Press. Google Scholar Google Preview WorldCat COPAC Lijphart, Arend. 1977. Democracy in Plural Societies: A Comparative Exploration. New Haven: Yale University Press. Google Scholar Google Preview WorldCat COPAC Liu, Amy. 2014. Standardizing Diversity: The Political Economy of Language Regimes. Pittsburgh: University of Pennsylvania Press. Google Scholar Google Preview WorldCat COPAC p. 20

Lydon, Ghislaine. 2009. On Trans-Saharan Trails: Islamic Law, Trade Networks, and Cross-Cultural Exchange in Nineteenth Century

Western Africa. Cambridge: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Makoni, Sinfree. 1998. “In the Beginning Was the Missionariesʼ Word: The European Invention of an African Language: The Case of Shona in Zimbabwe.” In Between Distinction and Extinction: The Harmonisation and Standardisation of African Languages, edited by K. K. Prah, 157–164. Johannesburg: University of Witwatersrand Press. Google Scholar Google Preview WorldCat COPAC Makoni, Sinfree, and Alastair Pennycook. 2005. “Disinventing and (Re)Constituting Languages.” Critical Inquiry in Language Studies 2 (3): 137–156. Google Scholar WorldCat Mamdani, Mahmood. 1996. Citizen and Subject: Contemporary Africa and the Legacy of Late Colonialism. Princeton, NJ: Princeton University Press. Google Scholar Google Preview WorldCat COPAC Mamdani, Mahmood. 2002. When Victims Become Killers: Colonialism, Nativism, and the Genocide in Rwanda. Princeton, NJ: Princeton University Press. Google Scholar Google Preview WorldCat COPAC May, Stephen. 2012 [2001]. Language and Minority Rights: Ethnicity, Nationalism and the Politics of Language. New York: Routledge. Google Scholar Google Preview WorldCat COPAC McNaughton, Patrick. 1982. “Language, Art, Secrecy and Power: The Semantics of Dalilu.” Anthropological Linguistics 24 (4): 487–505. Google Scholar WorldCat Mill, John Stuart. 1882 [1861]. “Of Nationality as Connected with Representative Government.” In Considerations on Representative Government, 308–319. New York: H. Holt. Google Scholar Google Preview WorldCat COPAC Mufwene, Salikoko S. 2001. The Ecology of Language Evolution. Cambridge: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Newbury, Catharine. 1988. The Cohesion of Oppression: Clientship and Ethnicity in Rwanda 1860–1960. New York: Columbia University Press. Google Scholar Google Preview WorldCat COPAC Ngugi wa Thiongʼo. 1986. Decolonizing the Mind. Portsmouth: Heinemann. Google Scholar Google Preview WorldCat COPAC Nurse, Derek, and Thomas Spear. 1985. The Swahili: Reconstructing the History and Language of an African Society, 800–1500. Philadelphia: University of Pennsylvania Press. Google Scholar Google Preview WorldCat COPAC Phillipson, Robert. 1992. “Colonial Linguistic Inheritance.” In Linguistic Imperialism, 109–135. New York: Oxford University Press. Google Scholar Google Preview WorldCat COPAC Phillipson, Robert. 2008. “Lingua Franca or Lingua Frankensteinia? English in European Integration and Globalization.” World Englishes 27 (2): 250–267. Google Scholar WorldCat Pool, Jonathan. 1991. “The O icial Language Problem.” The American Political Science Review 85 (2): 495–514. Google Scholar WorldCat

Posner, Daniel. 2003. “The Colonial Origins of Ethnic Cleavages: The Case of Linguistic Divisions in Zambia.” Comparative Politics 35 (2): 127–146. Google Scholar WorldCat Posner, Daniel. 2004. “Measuring Ethnic Fractionalization in Africa.” American Journal of Political Science 48 (4): 849–863. Google Scholar WorldCat Pritchett, James. 2007. Friends for Life, Friends for Death: Cohort and Consciousness among the Lunda-Ndembu. Charlottesville: University of Virginia Press. Google Scholar Google Preview WorldCat COPAC Rabushka, Alvin, and Kenneth A. Shepsle. 2009 [1972]. Politics in Plural Societies: A Theory of Democratic Instability. New York: Pearson Longman. Google Scholar Google Preview WorldCat COPAC Ranger, Terence. 2015 [1983]. “The Invention of Tradition in Colonial Africa.” In The Invention of Tradition, edited by Eric Hobsbawm and Terence Ranger, 211–262. New York: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Rodrik, Dani. 1999. “Where Did All the Growth Go? External Shocks, Social Conflict and Growth Collapses.” Journal of Economic Growth 4: 385–412. Google Scholar WorldCat Skutnabb-Kangas, Tove, and Robert Phillipson. 1994. Linguistic Human Rights: Overcoming Linguistic Discrimination. Berlin: Mouton de Gruyter. Google Scholar Google Preview WorldCat COPAC p. 21

Smith, Anthony D. 1986. The Ethnic Origin of Nations. Cambridge, MA: Wiley/Blackwell. Google Scholar Google Preview WorldCat COPAC Spear, Thomas. 2000. “Early Swahili History Reconsidered.” International Journal of African Historical Studies 33 (2): 257–290. Google Scholar WorldCat Spear, Thomas. 2003. “Neo-traditionalism and the Limits of Invention in British Colonial Africa.” Journal of African History 44: 3– 27. Google Scholar WorldCat Storch, Anne. 2011. Secret Manipulations: Language and Context in Africa. Oxford: Oxford University Press. Google Scholar Google Preview WorldCat COPAC Sweet, James. 2003. Recreating Africa: Culture, Kinship, and Religion in the African-Portuguese World, 1441–1770. Chapel Hill: University of North Carolina Press. Google Scholar Google Preview WorldCat COPAC Sweet, James. 2011. Domingos Álvares, African Healing, and the Intellectual History of the Atlantic World. Chapel Hill: University of North Carolina Press. Google Scholar Google Preview WorldCat COPAC Tamari, Tal. 1991. “The Development of Caste Systems in West Africa.” Journal of African History 32 (2): 221–250. Google Scholar WorldCat Taylor, Charles, and Michael Hudson, eds. 1972. World Handbook of Political and Social Indicators, 2nd ed. New Haven: Yale University Press. Google Scholar Google Preview WorldCat COPAC

Tilly, Charles. 1990. Coercion, Capital, and European States: AD 990–1990. Cambridge: Blackwell. Google Scholar Google Preview WorldCat COPAC Vail, Leroy, ed. 1989. The Creation of Tribalism in Southern Africa. Berkeley: University of California Press. Google Scholar Google Preview WorldCat COPAC Van Parijs, Philippe. 2011. Linguistic Justice for Europe and the World. Oxford: Oxford University Press. Google Scholar Google Preview WorldCat COPAC Vansina, Jan. 1990. Paths in the Rainforests: Toward a History of Political Tradition in Equatorial Africa. Madison: University of Wisconsin Press. Google Scholar Google Preview WorldCat COPAC Ware, Rudolph. 2014. The Walking Qurʼan: Islamic Education, Embodied Knowledge, and History in West Africa. Chapel Hill: University of North Carolina Press. Google Scholar Google Preview WorldCat COPAC Weber, Eugen. 1976. Peasants into Frenchmen: The Modernization of Rural France 1870–1914. Stanford, CA: Stanford University Press. Google Scholar Google Preview WorldCat COPAC Weber, Max. 1968 [1922]. Economy and Society, Vol. 2. Berkeley: University of California Press, 809–838. Google Scholar Google Preview WorldCat COPAC Wilder, Gary. 2005. The French Imperial Nation-State: Negritude and Colonial Humanism Between the Two World Wars. Chicago: University of Chicago Press. Google Scholar Google Preview WorldCat COPAC Wimmer, Andreas. 2009. “Ethnic Politics and Armed Conflict: A Configurational Analysis of a New Global Dataset.” American Sociological Review 74 (2): 316–337. Google Scholar WorldCat Wucherpfennig, J., N. B. Weidmann, L. Girardin, L.-E. Cederman, and A. Wimmer. 2011. “Politically Relevant Ethnic Groups across Space and Time: Introducing the GeoEPR dataset.” Conflict Management and Peace 28 (5): 423–437. Google Scholar WorldCat p. 22

Young, Crawford. 1976. The Politics of Cultural Pluralism. Madison: University of Wisconsin Press. Google Scholar Google Preview WorldCat COPAC

Tracing Language Movement in Africa Ericka A. Albaugh (ed.), Kathryn M. de Luna (ed.) https://doi.org/10.1093/oso/9780190657543.001.0001 Published: 2018

Online ISBN: 9780190657574

Print ISBN: 9780190657543

PART FRONT MATTER

Published: February 2018

Subject: Sociolinguistics Collection: Oxford Scholarship Online

We begin with a group of chapters that lay a broad but deep foundation for how various disciplines think about language and its movement across space and time. Historical linguist Nurse opens a window onto the origins of the eld that originally identi ed the relationship between language movement and change, outlining its methodology and the factors that inspired a renewed interest in exploring processes of language change. Linguists Olson and Lewis introduce the widely used Ethnologue database, describing the di

culties involved with identifying and classifying languages, explaining the rationale and methods for

reporting language status within countries, and articulating the ways of incorporating information about second-language capabilities. Archaeologist MacEachern discusses various, often contradictory, evidence used in the quest to determine the demographic history implicated in the present geographic distribution and density of languages and material culture in the greater Chad region of Central Africa. Finally, political scientist Logan, deputy director of the Afrobarometer, demonstrates the complexity of organizing language information gleaned from recent public opinion surveys across Africa to begin to answer questions about individuals’ political attitudes. The four authors employ di erent forms of data. In addition to the words, letters, sounds, and grammatical practices that one might expect when studying language, they show that evidence can come from pottery shards, traded beads, genetic traces, and geological clues, as well as surveys and recorded interviews. The availability and quality of these data vary widely, and the very paucity of written records for many African p. 24

languages,

along with the ambition of historical retrieval, forces special creativity and necessary

interdisciplinarity on those looking deep in time. All of the authors employ and interrogate the outlines and intensity of boundaries. They use classifying schemes to identify language families, macrolanguages, distinct languages, and dialects. Olson and Lewis acknowledge that dividing lines between distinct languages are imperfect, as does Nurse for the distinction between a Creole and a “regular” language. Similarly, though they recognize as futile the attempt to precisely enumerate second-language speakers, Olson and Lewis introduce a scale to measure the degree of second-language use among groups. Finally, MacEachern and Nurse take the continent, regions, and groups as units of analysis, while Olson and Lewis and Logan begin at the country level, with Logan extending even to micro-communities, all of which has consequences for their questions and arrangement of data. Obviously, the four chapters contain vastly di erent time scales of change. MacEachern and Nurse focus on changes over centuries and millennia, Olson and Lewis on observable changes since about the 1960s, and Logan on a single moment in time, which allows comparison with past and future evidence. Nurse is cognizant of shifts in scholars’ understandings of how change comes about, from assumptions of major population movements that brought languages to new places to recognition that change could be induced by

the constant “jostling” of languages in proximate contact. Similarly attempting to explain outcomes without major population movements, MacEachern observes geographic patterns and climate in uences on language, as well as shorter population movements produced by warfare and slaving. Change can “simply” be described, or it can be judged. Olson and Lewis assume that second language use is “usually (if not always) the major factor that contributes to the disuse” of a rst language, and Ethnologue’s creation of a scale of “disruption of language vitality” reveals its concern with threatened languages. Power is evident implicitly or explicitly. Even before the modern state, violence and trade moved languages (MacEachern). When the state is the unit of analysis (Olson and Lewis; Logan), the status and spread of a language have much to do with how it is treated by a government. Power relations between populations have determined which languages have been “donors” (Nurse) and which the receivers, which the colonizers (MacEachern) and which the dominated. Finally, all of these chapters demonstrate the power of scholars themselves to name languages and dialects, to o er them as categorical options, to render them historically visible or invisible. p. 25

This section opens up questions and dialog that connect the project as a whole. To di ering extents, they use language as a tool to answer other questions. Accepting the “ ction” of language boundedness may serve purposes of preservation or comparison. The tension between language and identity comes through in the criteria used to distinguish languages in the Ethnologue: languages otherwise di erent but spoken by groups who claim similar culture and heritage are classi ed as related, while languages otherwise similar but spoken by groups who consider themselves di erent are classi ed as distinct. This role of agency is highlighted as well in MacEachern’s observation that attempts to classify using material culture are confounded by human manipulation of that evidence in order to maintain or reduce divisions between them and others. MacEachern cautions methodological modesty; Nurse chides fellow linguists for disciplinary

p. 26

narrowness. All point to the fruitfulness of this enterprise.

Tracing Language Movement in Africa Ericka A. Albaugh (ed.), Kathryn M. de Luna (ed.) https://doi.org/10.1093/oso/9780190657543.001.0001 Published: 2018

Online ISBN: 9780190657574

Print ISBN: 9780190657543

CHAPTER

2 Language Change and Movement as Seen by Historical Linguistics  Derek Nurse https://doi.org/10.1093/oso/9780190657543.003.0002 Published: February 2018

Pages 27–44

Abstract The focus of this chapter is on how languages move and change over time and space. The perceptions of historical linguists have been shaped by what they were observing. During the owering of comparative linguistics, from the late 19th into the 20th century, the dominant view was that in earlier times when people moved, their languages moved with them, often over long distances, sometimes fast, and that language change was largely internal. That changed in the second half of the 20th century. We now recognize that in recent centuries and millennia, most movements of communities and individuals have been local and shorter. Constant contact between communities resulted in features owing across language boundaries, especially in crowded and long-settled locations such as most of Central and West Africa. Although communities did mix and people did cross borders, it became clear that language and linguistic features could also move without communities moving.

Keywords: historical linguistics, language movement, language change, overview, Africa Subject: Sociolinguistics Collection: Oxford Scholarship Online

Background and Definitions Early European and Indian linguistic scholars separated the synchronic study of language from the study of language origins. Synchronic study of a language focuses on one point in time, the present (e.g. a grammar of modern Hausa) or the past (e.g. a grammar of Classical Latin). The oldest known synchronic tradition is in India: although the best known name is Panini, who in the 4th century BC . wrote a sophisticated set of grammatical statements, he was only one of several scholars of his era, and the grammatical tradition preceded and followed him. Greek and Roman traditions started a few centuries later and developed into a codi ed view of grammar which continued through the Middle Ages and has left clear traces in contemporary thinking and terminology.

On the other hand, most thinking all over the world about the origins of language had little or nothing to do with analysis of language. The origin of language(s) was frequently attributed to divine or nonhuman action. A well-known account is that of the Tower of Babel, in which a single population speaking a single language came to the land of Shinar, where they built a city and a tower, but God intervened, confounding their speech and scattering them across the world. Many such mythical accounts exist in communities around the world. p. 28

The focus of Western thinking about language has swung between synchronic and diachronic (the study of language over time). From the Roman period to the late 18th century, the emphasis was on synchronic linguistics. Then, although the synchronic tradition continued, the pendulum swang dramatically to diachronic investigation, only to steadily swing back again in the mid-20th century. The birth of modern historical linguistics is conventionally attributed to Sir William Jones, a polyglot Anglo-Welsh judge working in Calcutta. In 1786 he gave a speech in which he suggested that Sanskrit, Greek, Latin, Persian, Gothic, and Celtic were related and had a common ancestor. This was taken up by several European linguists, and from then till well into the 20th century, historical linguistics created various language 1

families, starting with Indo-European and eventually covering the world. Serious scienti c application of historical linguistics to African languages and the establishment of language families started in the early 19th century and continues today (Dimmendaal 2011, 3–4, 81–92). Although many disagreements still exist on the number and composition of some language families, this classi cation of the world’s languages into related groups was and continues to be a major achievement. It provides a framework against which scholars and scientists from other disciplines—historians, archaeologists, geneticists, and biologists—can measure their ndings. The term historical linguistics is used in two ways. First, it is used to contrast with comparative linguistics. At the center of comparative linguistics is the comparative method, which works backwards in time, or 2

upstream, from today’s languages to establish genetic relationships (families, phyla ) among them and to reconstruct their ancestors using exclusively linguistic data. By contrast, historical linguistics charts the forward, downstream, or vertical development of languages from an earlier to a later point, and may draw on nonlinguistic data. But the term historical linguistics is also used in a second, more general way, to cover both historical—as just outlined—and comparative linguistics. It can be seen that what William Jones and his successors really did was comparative linguistics, even though it is generally called historical linguistics.

p. 29

Comparative and Historical Linguistics: Focus, Methodology, Implications It is worthwhile sketching brie y how comparative linguistics works, because that is central to how historical linguists came to view change and movement. Although Jones pointed to the possibilities of interconnectedness between languages, the methodology itself was steadily developed during the 19th 3

century by scholars such as Bopp, Rask, Verner, Grimm, and others. A set of steps is followed in applying the comparative method. First, assemble lists of words from the languages suspected of being related. Here is an abbreviated example (English serves the double function of gloss and control language):

A

dada

dadu

dega

mudi

duma

B

haha

hahu

hega

muhi

huma

C

rara

raru

rega

muri

ruma

D

tata

tatu

tega

muti

tuma

E(nglish)

father

three

trap

tree

send

Second, inspect the sets to determine regular sound correspondences. Here we see an obvious set of such correspondences: where language A regularly has [d], B has [h], C has [r], D has [t]. Language E has no regular correspondences (we ignore English dad). Repeat this for as many sets of correspondences as can be found. The more such phonetic correspondences, instances of correspondences, and words involving those correspondences that can be found, the greater the certainty that the correspondences are not accidental but indicate relatedness between and a common origin for the languages involved (English clearly does not t here, as none of the English words is cognate). Third, triangulate backwards from (two or more) contemporary languages to a protolanguage (Nurse 1997, 361 ; Dimmendaal 2011, 9 ). For each set of sound correspondences, posit an original single sound in the protolanguage from which today’s sounds derive. Most linguists would posit *t as the original reconstruction for the d-h-r-t correspondence above, even though that would not be obvious to non-linguists. Fourth, organize the reconstructed consonants and vowels into a sound system for the protolanguage. Fifth, having thus reconstructed a sound system for the protolanguage, the linguist can proceed to reconstruct its lexicon and morphological system. Important in p. 30

this process is identifying

innovations, because they indicate shared developments in the related

languages after the protolanguage disintegrated. In the set above, most linguists would not simply say that reconstructed *t changed separately to d in Language A, to r in Language C, and to h in Language B, but that these changes are linked and could be arranged as *t > d > r > h, so that A, B, and C all shared the change of *t > d, that B and C shared the change of d > r, and that B alone underwent r > h. This has obvious historical implications. Likewise, if other related languages in the area were found to have d, r, or h in these and other words, that has historical implications, since it is methodologically easier to assume that shared changes occurred at one time and place, rather than severally in di erent places and times. So languages sharing these changes are probably related, implying that they moved to their current locations from a single historical point. In this way, comparative linguistics, going backwards from the present, paves the way for historical linguists to examine inter alia the changes that took place over time. All this has implications for how historical linguists came to see change and movement. Over recent centuries, we have moved away from the ideal of the Renaissance scholar, who aimed at being acquainted with many elds of human enquiry, to become more specialized, knowing more and more about ever smaller elds, even within one discipline. Because of the central role of the comparative method, historical linguists came to know more and more about sound change. In the 19th century, they discovered much about sound change by examining in detail what had happened historically within Indo-European languages, changes referred to as “laws,” such as Grimm’s Law, Verner’s Law, Grassmann’s Law, etc. (or in African languages, not to be outdone, Meinhof’s Law, Meeussen’s Rule, Dahl’s Law, etc.). At rst they thought that sound changes operated without exception, but they came to realize that there were apparent exceptions which often had to do with consonant or vowels changing di erentially, depending on surrounding sounds. They also came to realize that the same kinds of changes operated in many languages across the world. In the 20th century, the emphasis changed. It became harder to mine interesting new historical material from the older languages, while at the same time linguists had access to more and more contemporary

languages. The second half of the 20th century saw the dramatic rise of phonology, concerned with the systematic organization of sounds in languages, how sounds behave and interact with each other. It has become increasingly theoretical and removed from phonetics, the straight study of sounds, and is a discipline unto itself. William Jones, his peers, and the historical linguists of the 19th century would have p. 31

trouble following it. It also means that nonlinguists—historians, archaeologists, political scientists, and others—have trouble understanding what phonologically informed historical linguists are talking about (“stops became spirants before superclose vowels”) and have to hope they are being told the truth. The emphasis on phonology is part of a wider phenomenon—the tendency among linguists in general, historical linguists, and phonologists to focus on language alone and to cut themselves o

from wider

discourse. They view language as a system or a set of subsystems, they know that certain kinds of internal changes occur often across natural languages. They are interested in how the systems function and change, and in expanding their knowledge of the possible systems and changes. They view change as primarily language-internal. They are happy in their world. Historians, archaeologists, and anthropologists are quite interested in hearing linguists’ opinions on topics such as relating ceramic styles, pastoralism, or graves to language groups, or how one group relates to another, but phonologists are rarely interested in input from nonlinguists about tone raising or vowel shifts, nor are they much interested in how they might relate to events outside language. When many nonlinguists think of how one language di ers from another, or of language change, they think rst of vocabulary, and when they think of causation they tend to think of language contact, but for most linguists vocabulary is rather super cial and less interesting. So linguists and nonlinguists focus on di erent phenomena. Another result of being involved in classi cation, in language families, and in how to subdivide the families is that historical linguists are better at thinking about certain time periods than others. We might divide human and language history in Africa into four periods: 1) the rst migration out of Africa, starting some 4

60,000 or more years ago; 2) the emergence of recognizable language families in Africa; 3) the steady disintegration and spread of those families; and 4) the modern period. Ignoring the details of their composition and some language isolates, African was until recently thought to be home to four language 5

phyla: Khoisan (35 languages), Afro-Asiatic (370), Niger-Congo (>1,430), and Nilo-Saharan (200). People speaking Khoisan languages probably roamed Africa as long ago as 60,000 BP (years before present). ProtoAfro-Asiatic is variously dated at between 9,000 and 18,000 BP. Proto-Niger-Congo and Proto-NiloSaharan are a little younger, at around 12,000 BP. The older a language grouping, the less certain its status, p. 32

because any grouping depends on the availability of

material to compare. Since languages steadily

replace materials over time, there comes a point where there is very little left to compare, which is the basic reason for the uncertainty of some phyla. Linguists are not agreed on how far back in time reconstruction can be done. Some would say to 8,000 BP, others would say to 15,000 BP or a little longer. That means that many language phyla can be reconstructed, but not beyond 15,000 BP or so, as most historical linguists are reluctant to join them and to think beyond that period. Thus most historical linguists have little to say about period 1) above, when humans left Africa. Comparativists have a good deal to say about periods 2) and 3), when protolanguages emerged and constantly split up, respectively. Period 3) especially covers many millennia, into the AD period, and has attracted much attention from historical linguists. In Africa, where few written records of any length exist, it covers down into recent centuries, so period 4), that of modern languages, is short, and historical linguists in the narrow sense have less to contribute. In China, India, Egypt, and Europe, written records are more plentiful, so comparative linguists have less to say and historical linguists play a larger role. All this in turn a ects how historical linguistics in the wider sense came to view language movement. A language family or phylum was viewed as deriving from a protolanguage, spoken by a community located at one point in time and space, constantly splintering, moving from one place to another. That tends to

encourage a view of languages moving with their community as it moved along—what has come to be called demic movement—and as a language divided, the community divided with it. It also encouraged a view of language communities moving long distances, and often fast. Early IndoEuropean horsemen raced west across the Russian steppe. Other Indo-Europeans poured down into northern India. Genghis Khan was born in Mongolia, but he and his descendants rapidly set up the Mongolian Empire, controlling much of Eurasia from China to Eastern Europe. A century after Islam was founded in Saudi Arabia, Arabic speakers had galloped west across Egypt and the Maghreb to Iberia, and down to the Indus Valley in India. In early days, Khoisan speakers covered the eastern side of Africa from Ethiopia to southern Africa; however that is interpreted, it must have involved lengthy journeys. Bantuspeaking communities expanded east and south out of the Nigeria/Cameroon borderland to inhabit most of central, eastern, and southern Africa in a millennium or so. Afro-Asiatic communities stretch from Mauritania to Iraq, so whatever homeland is posited, there must have been lengthy movement. Finally, the Fula (Fulbe, speaking Fulfulde) have expanded to cover a 5,000-kilometer (>3,000-mile) area from p. 33

Mauritania in the west to Sudan in the east. All carried their languages with them. Long distances were associated with ways of life such as pastoralism or a hunter-gatherer existence. Those who travelled on horse or camel moved fast, while subsistence farmers just trudged along. To sum up, in the 19th century, use of comparative linguistics and particularly of the comparative method led to the perceptions that languages moved with their communities, often over long distances, and change was vertical and largely language-internal.

Changing Perceptions in the 20th Century Many advances were made in historical work on Africa languages during the 20th century, but I concentrate here on those that led to modi cation of views of language movement and language change. The three major factors, which spawned others, were our expanded knowledge of African languages, the e ect of language contact on language change, and the rise—outside Africa at rst—of sociolinguistics. Prior to the 19th century, although Europeans had certainly started to explore Africa, very few synchronic accounts had been written of African languages (Dimmendaal 2011, 4), and it was only in the 19th century that colonial expansion led to a real increase in interest in, and knowledge of, the languages of Africa. This increase continued into the 20th century. Few linguists in the 19th century would have had any idea of the total number of languages in Africa or in any African country, but by the second half of the 20th century it was realized that Africa is home to over 2,000 languages, slightly fewer than in Asia but signi cantly more than in other areas of the world, and that it is a continent of staggering linguistic diversity. Africa was a linguistically crowded continent. This growing realization, plus a lessening of interest in large, higher-level groupings, led a shift in emphasis from comparative to historical linguistics. Granted that the splintering of early language families might have resulted in longer and possibly faster movements, there arose a di erent question: after the initial large movements, what then? Movements in more recent times, especially the last two millennia, must have been shorter. And indeed, had all of those initial large fast movements always been large and fast? While Bantu-speaking communities undoubtedly traveled a long way from their homeland in the Nigeria-Cameroon borderland to southern Africa and probably moved faster across areas with fairly harsh farming conditions and limited access to water, in other areas movement would have been much slower. Farmers perhaps only moved as land became exhausted, or families were large so some sons p. 34

had to move on, or better land was found over the horizon. Early

long distances and rapid movement

apparently needed to be supplemented later with shorter, gradual moves. And how, on the other hand, to explain that in many places in Nigeria it is possible to walk along a path only a few miles from one village to another and nd a di erent but related language, a situation that can be replicated in many areas in Africa?

One of the obvious outcomes of the realization of the wealth of languages in Africa is that many of them are constantly jostling their neighbors, and have been for centuries or longer. Although the account in the previous section emphasized language change as an internal process, linguists in the 19th century were certainly aware of the possibility of change occurring as the result of contact with other languages, of mixed languages, but these were not a major topic. The most dramatic examples of contact-induced change came in what are called Sprachbunds, areas of linguistic convergence and di usion, which involve sets of languages that have come to share features as the result of geographical proximity and language contact. The best known is the Balkan Sprachbund, rst mentioned in the 19th century but only fully explored in the 20th century. This is an area in which Slavic-, Albanian-, Greek-, Romance-, and Romani (Gypsy)-speaking populations had mingled for nearly two millennia. It involved the mixing of morphological, syntactic, and phonological features—vocabulary was less involved—due to communities having lived side by side for many centuries, to bilingualism, and to some communities giving up their languages but carrying features from their old into their new languages. The local versions of the languages involved di er considerably from the versions spoken outside the area. Since then, Sprachbunds have been described in several places: northern India, where Indo-Aryan and Dravidian came together; the Ethiopian highlands (Cushitic and Semitic, 1st millennium AD ); western Tanzania (Cushitic, Bantu, Nilotic); the West African Sahel; the Kalahari Basin; and others. Areas involving fewer, often just two, communities are more common, and those involving languages with unusual features are easiest to identify—click sounds in Bantu are an obvious example. The third factor also came from outside historical and comparative linguistics: the appearance of sociolinguistics. Social aspects of language had been acknowledged and studied since the late 19th century, but it was only in the 1960s that what we now call sociolinguistics really took o : the study of the e ects of aspects of society, including cultural norms, expectations, and context, on the way language is used, and the e ects of language use on society. Sociolinguists realized for the rst time that language change could be observed and studied as it was occurring, whereas historical linguists had only recognized the starting and p. 35

end points of change. Labov

(1972, 164–176) initially studied language use in the 1960s in Martha’s

Vineyard, originally a 19th century whaling port which had fallen onto hard times with the collapse of whaling. By the 1960s, some in Martha’s Vineyard couldn’t wait to get out, while others liked its old ways. Labov discovered that the two populations behaved di erently phonetically—beyond their control or ability to describe—so that those who identi ed with the place had changed their pronunciation of some vowels to older norms, while those who wanted to leave rejected the older norms. Labov went on to examine linguistic features in African American English and then current vowel changes in various places across North America.

Outcomes of the Changed Perceptions These three factors together—expanded knowledge of African languages, the e ect of language contact on language change, and the rise of sociolinguistics—led to an explosion of models of change starting in the 1960s. Historical linguists saw they had to take these into account, because they recognized the identity of language processes across time and space: the kinds of changes that can be shown to be happening or to have recently happened can be assumed to have happened in the past. Many new instruments have appeared in the historical linguistic tool kit over the last fty years. Given space constraints, only the main ones are treated here. It had long been recognized that when languages came in contact, various results were possible, of which the most recognized was “borrowing,” de ned as the “incorporation of foreign features into a group’s native language: the native language is maintained but changed by the addition of the incorporated

6

features” (Thomason and Kaufman 1988, 37). In the popular mind and in the past, words were easily borrowed, but it has become accepted that almost anything may be borrowed depending on the circumstances. It is also accepted that there is a general—not universally agreed on—order of borrowing: rst, lexicon, especially content words (e.g. clothing, food, animals, religion); second, as preceding, plus function words (adverbs, prepositions) and some nonlexical items (foreign sounds via loanwords, minor syntactic orderings); third, as preceding, plus more abstract lexicon (a p. 36

xes on borrowed words, pronouns,

numerals), phonological processes, additional syntactic processes); fourth, major structural

changes in

phonology, morphology, and syntax, but not leading to major typological change; fth, as preceding, plus major structural features that do lead to signi cant typological change. All these correspond to increasing levels of contact, from light, casual contact at the rst stage up to very strong cultural pressure at stage ve. If the pressure continues, the speakers of the language could give up and just move to speaking the donor/dominant language. The dominant earlier view was that most languages were the result of organic development from their earlier stages and that if borrowing occurred, it would mainly be of vocabulary. That has now been replaced by this more nuanced view of borrowing. It is accepted that many languages, large and small, inside and outside Africa, have undergone extensive borrowing: Hausa, Zulu, Lozi, Swahili, English, and Spanish. A second process resulting from contact and not much recognized before the 20th century is “language shift” (also “language replacement/assimilation”), in which a speech community shifts to speaking another language, a shift which often takes several generations to complete. In some cases, the features they bring with them become part of the target language. A major di erence between this and the foregoing is that while early and/or lexical transfer characterizes borrowing, in language shift it is rather nonlexical features which are transferred. Vocabulary is not usually transferred, because the shifting population has access to the vocabulary of the target language. Vocabulary may be transferred if the shifting community has a specialized vocabulary for some area not part of the target community’s culture. For instance, the Bajuni community of southern Somalia has shifted almost completely in two decades to a more or less standard variety of Swahili, but they keep their traditional vocabulary for food, dances, and kinship terms. When early Indo-Aryan was carried into northwestern India some 3,800 BP, it came to be spoken by many people indigenous to northern India. The major phonological and syntactic changes brought about by this became part of the language, despite the e orts of the guardians of the language. In Africa, language shifts have occurred widely over millennia, because in a linguistically crowded continent communities are always in contact. They continue today, as the larger or more prestigious swallow up the smaller. The larger are, for example, Akan, Amharic, Ewe, Fulfulde, Manding, Swahili, and Yoruba, and as they are increasingly used by smaller communities, they spawn local varieties characterized by the absorption of features from the smaller language communities. A third, widely recognized phenomenon is pidginization, often leading to pidgin and creole languages. It was rst acknowledged in the Caribbean: slaves from various communities in West Africa brought to work p. 37

by European colonizers often did not share a single language and had

to use forms of the colonial

languages, to which, however, they had limited access. The result was a series of modi ed versions of the European languages—pidgins—which initially were no one’s rst language. Pidgins worldwide show common features, involving simpli cation: reduced phonology (e.g. basic vowels, loss of tones, simpli ed syllable structure), little morphology, grammatical categories such as tense in verbs or number in nouns being expressed by separate words or reduplication, uncomplicated clause structure. Often much of the vocabulary of a pidgin comes from the dominant language, while structure comes from a combination of the substrates and universal linguistic processes. If the pidgin survives and is used by the next generation, it becomes a creole—in a commonly cited phrase, a creole is a pidgin that has acquired native speakers. Creoles move on to become “regular” languages, although the distinction between creole and regular is thin. After some centuries and in the absence of written records, it is not always easy to identify former creoles.

7

Pidgins do not always survive, because the circumstances that spawned them cease to exist. Often mentioned is Fanagalo, a Zulu-based pidgin used in mines in South Africa and neighboring countries but slowly declining. In Kenya and Zambia, a pidgin called Settla (“settler”) was used in the 20th century but is almost gone because the circumstances changed. Pidgins come in many varieties, and because they arise in di erent circumstances, there is not total agreement on the details of their genesis or development. Africa is home to so many pidgins and creoles, a few based on European but most on African languages, that it is only possible to give an overview. In West Africa, West African Pidgin English was the lingua franca spoken along the West African coast during the Atlantic slave trade era. The various pidgin and creole languages still spoken in West Africa today—Aku in The Gambia, Sierra Leone Krio, Nigerian Pidgin English, Ghanaian Pidgin English, Cameroonian Pidgin English, Fernando Poo Creole English, and others— all derive from it. There are other pidgins in West Africa, not all necessarily called pidgins or creoles, arisen from African languages such as Wolof, Ewondo, Hausa, and Fulfulde. In central Africa are others, spoken by millions of people: Kituba, a creole based on Kongo; Lingala, based originally on several local languages in p. 38

the Democratic Republic of Congo (DRC)

and now used more widely; Sango in the Central African

Republic. These were all initially lingua francas or trade languages which took on wider functions. It is the tradition when discussing Arabic to talk of dialects or varieties rather than pidgins or creoles, but pidgin varieties of Arabic do exist. In the Sudan we nd Juba Arabic, spoken in South Sudan, and Nubi, spoken originally by the descendants of Sudanese soldiers from di erent ethnic groups, who used it for inter-group communication. The best example in East Africa is Swahili, originally spoken mainly in small settlements on the coast. During the 19th century, it was carried west as far as the DRC, where locals had limited access to coastal Swahili. Simpli ed varieties developed, with much input from local languages. One of these is now one of the national languages of the country. During colonial times and later, simpli ed local varieties developed in Kenya and Tanzania. Pidgins and creoles are so widespread in Africa that we must assume they existed there long before outsiders arrived. What precedes raises an interesting issue: when did simpli cation occur in some of Africa’s most widespread languages? If compared to some of their less widespread relatives, Hausa, Kituba, Lingala, and Swahili show signs of simpli cation, but when did this occur? The path sketched above—pidgin, creole, then “regular” language—suggests that simpli cation and reduction would have been present at the start of the path and continued until later linguistic elaboration obscured it, but how long would that be? Alternatively, if a “regular” language is long and widely used as a lingua franca, that could also lead to simpli cation and reduction. English has been so used for centuries and is reduced compared to many other Indo-European languages (e.g. it no longer has gender or case). Looking at a language in retrospect, it is not always easy to distinguish these two possibilities. Associated with this is another question. Of the languages spoken by millions of people across Africa, some have not moved very far from their original homeland where they arose millennia or many centuries ago and do not seem to have had much success in persuading outsiders to speak their tongue, others—some mentioned above—have been quite successful in spreading their language. What determines this? Social issues clearly play a role in the kind of ongoing change investigated by Labov (ibid.) and in the processes described by him, but historical sociolinguistics is notoriously tricky, as I know from personal experience. After Thomason and Kaufman’s book (1988) rst appeared, I chose a set of East African languages with the aim of identifying the outcome of the three processes above, and thinking it would be then easy to identify the social situations that had produced them, but it was not. p. 39

The intense interest in language contact has led to a new eld, areal linguistics, spearheaded in Africa initially by Heine (e.g. Heine and Leyew 2008). A feature rst perceived as shared by two languages could in fact be shared by several, even many, and often formed part of a bundle. Some of the bundles are fairly local, others stretch across half the continent. The most extensive and striking occur when two conditions are met: they started early (more than two millennia ago, roughly) and they involve languages from two or

more of the four phyla. Since Niger-Congo covers much of the continent and is in contact at some points with each of the others, it is most often involved. The largest is the Macro-Sudan belt, associated with Gueldemann (e.g. 2008), stretching from Gambia in the west to South Sudan in the east, south of the Sahara, and north of the rain forest. Medium-sized areas are Chad-Ethiopia, the Kalahari Basin, the Sahara Spread Zone, and part of the Bantu area (especially west of the Great Rift, in Gabon, Congo, and the DRC). Smaller ones are too numerous to list. This interest in areal features—and the genetic input mentioned below in the fth section—has had the unintended result that the continent’s four language phyla are no longer as certain as they appeared only a few decades ago. Work by Greenberg, starting in the 1950s (see Greenberg 1963), set out four phyla, listed in 8

the second section above. These were then widely accepted, but in the last two decades some have cracked: Khoisan is no longer seen as a single unit, doubts have been cast on some members of Niger-Congo, Songhay and Kuman/Gumuz are questionable members of Nilo-Saharan, and there is disagreement about the internal composition of the African parts of Afro-Asiatic (formerly Hamito-Semitic). A fourth innovation, having to do with internal development rather than external contact, is grammaticalization. The concept goes back two centuries, the term to the early 20th century, but in the 1970s interest took o

and was quickly extended to African languages. It is hardly a surprise to most

speakers of English that “will,” as in “we will go,” derives from an older English verb willan “want,” because it is a short, fairly transparent step from “we want to go” to “we will go.” Similarly, only a few centuries ago “we are going” could only be used if the speaker was actually on the way to somewhere or something, but today it is as frequent as “we will go” to express general futurity (“we are going to Timbuktu next year”), where neither the time nor the place is close. Similarly, it is of no great surprise to English, French, German, or Spanish speakers that the idea of perfect action is expressed by using the p. 40

present form of “have” and a past participle

(“we have walked here”). It is a process whereby full words

– verbs, nouns, and other categories – evolve to become grammatical markers. Independent words move along a scale to become a

xes, they become shortened, and concrete reference becomes abstract

grammatical meaning. Changes occur most often in certain predictable directions and paths and not others, and certain processes occur repeatedly across languages. Sometimes the end results are transparent, as in the examples above, but other times they are not. For example, some Swahili speakers might be able to analyze the ta in tu-ta-kwenda “we will go” as being connected to the verb “want” (taka), but no speakers would think of the perfect marker me, as in wa-me-kwenda  “they have gone,” as deriving from an older verb “ nish” (mala). Within African language studies, Heine is most often associated with grammaticalization (for African examples, see Heine et al. 2003). This concept has entered into the consciousness of how all historical linguists view internal change. In summary, to the perceptions inherited from the 19th century that languages moved with their communities, often over long distances, and change was vertical and largely language-internal, the 20th century added that languages can spread without their communities, that communities can move short distances and in fact do not need to move at all for change to occur, and that change often occurs horizontally as the result of contact.

Historical Linguistics, Language Change, Language Movement, Other Disciplines Most historical linguists have always been aware of possible interfaces with other disciplines. Historical archaeology started somewhat earlier in Egypt and has been more comprehensive there than in subSaharan Africa, where most work has been done in the second half of the 20th century. Since then, historical linguists have become used to being presented with scenarios by archaeologists and historians and asked for their reaction. Linguists cannot respond to questions about the very far past, for example associated with the work rst done by Leakey and then others in Kenya, Ethiopia, and South Africa. They can respond best to matter dating from the start of the phyla mentioned in the second section above, say 15,000 to 10,000 BP. They get better nearer to the present but cease to be of much use after 1,000 BP. They look at the linguistic changes they have before them, and ask what is regular organic internal change and what might be due to contact between the groups involved in archaeological scenarios. This is most easily and visibly done in vocabulary: contact and changes between communities with di erent economies (hunterp. 41

gatherers, pastoralists, farmers) rub o

and leave lexical traces. The name most closely associated with

this kind of work is Ehret (e.g. 2010). In the last decade or so, the work of evolutionary biologists and geneticists has introduced a new dimension and new data into thinking about prehistory in sub-Saharan Africa. So far, with using genetic phylogenies and usually working in teams, they have mainly explored southern (e.g. Pickrell et al. 2012) and eastern (e.g. 9

De Filippo et al. 2012) Africa. Another team has used a slightly di erent approach, climatic factors, to examine Bantu expansion across the Rainforest (Bostoen et al. 2015). It is to be expected they will expand their geographical coverage in the future. In many cases, their results and genetic phylogenies t well with previous linguistic phylogenies while injecting more detail into previous knowledge. In other cases, where there were two competing linguistic scenarios they come down in favor of one. For example, there were two existing explanations for Bantu expansions cross the Rainforest—one that involved ancestral Bantu splitting earlier, with a western stream going down into southwestern Africa and an eastern stream going east across the Rainforest into East Africa and then turning south, the other involving the ancestral community rst moving south and then a stream moving northeast into East Africa and eventually south— but recent genetic evidence supports the latter. Also, where some linguists saw one Khoisan phylum and others saw several families, the genetic evidence supports the latter, with three families in southern African 10

and one in Tanzania.

While reading about these new approaches and the results they bring is exciting, our purpose here is to deal with what they tell us about language change and language/community movement that we didn’t already know. In the case of how language change occurs, the answer is: hardly anything. As Bantu communities moved south and east, language speciation occurred, and the languages changed again and again, vertically, even as communities kept moving. That was the predominant model. Sometimes, as they came in contact with non-Bantu communities, in the Rainforest, in the Savannah, and especially in southern Africa, that contact modi ed the languages. In some cases, we already knew the details of those contacts; in other cases, the linguists involved need to con rm more of these contact and absorption claims by identifying borrowed material. For language movement, the answer is similar. Predominantly, languages moved along with their p. 42

communities. As they moved along, some

Bantu communities stayed, while others kept moving. Where

they stayed adjacent to existing substrate non-Bantu communities, those communities started to abandon their languages in favor of the new Bantu language. Thus some Bantu languages spread horizontally, without any movement of the community, as happened notably with Pygmies in the Rainforest and Khoisan in the south. No Pygmies speak their original language any more, and many Khoisan have given or are giving up their language in favor of a nearby and dominant Bantu language. For many readers whose rst

language is West African, this talk of perpetual movement must seem a rather biased view, because Bantu 11

expansion is recent and fast, whereas West Africa is a crowded linguistic area

(Nigeria is home to some

500 languages) where many language communities sit side by side, and have done so for millennia. There has been relatively little recent movement of communities but much horizontal mixing, and therefore much contact-induced change. How language changes depends on the circumstances.

How Languages Move and Change over Time and Space: Summary The focus of this chapter is on how languages move and change over time and space. The perceptions of historical linguists have been shaped by what they were observing. During the owering of comparative linguistics, from the late 19th into the 20th century, the dominant view was that people moved, their languages moved with them, often over long distances, sometimes fast, and that language change was largely internal. That changed in the second half of the 20th century. We recognized that in recent centuries and millennia most movements of communities and individuals have been local and shorter. Constant contact between communities resulted in features owing across language boundaries, especially in p. 43

crowded and long-settled locations such as most of Central and West Africa. Although

communities did

mix and people did cross borders, it became clear that language and linguistic features could move without communities moving.

References Bostoen, Koen, Bernard Clist, C. Doumenge, R. Grollemund, J.-M. Hombert, J. K. Muluwa, and J Maley. 2015. “Middle to Late Holocene Paleoclimatic Change and the Early Bantu Expansion in the Rain Forests of Western Central Africa.” Current Anthropology 56: 354–384. Google Scholar WorldCat De Filippo, Cesare, Koen Bostoen, M. Stoneking, and B. Pakendorf. 2012. “Bringing together Linguistic and Genetic Evidence to Test the Bantu Expansion.” Proceedings of the Royal Society B 279 (1741): 3256–3263. Google Scholar WorldCat Dimmendaal, Gerrit J. 2011. Historical Linguistics and the Comparative Study of African Languages. Amsterdam: John Benjamins. Google Scholar Google Preview WorldCat COPAC Ehret, Chris. 2010. History and the Testimony of Language. Berkeley: University of California Press. Google Scholar Google Preview WorldCat COPAC Greenberg, Joseph H. 1963. The Languages of Africa. Bloomington: Indiana University Press. Google Scholar Google Preview WorldCat COPAC Gueldemann, Tom. 2008. “The Macro-Sudan Belt: Towards Identifying a Linguistic Area in Northern Sub-Saharan Africa.” In African Languages, edited by Bernd Heine and Derek Nurse, 151–185. Cambridge: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Heine, Bernd, Tom Gueldemann, C. Kilian-Hatz, D. A. Lessau, H. Roberg, M. Schladt, and T. Stolz. 2003. “Conceptual Shi : A Lexicon of Grammaticalization Processes in African Languages.” Afrikanistische Arbeitspapiere 34/35: 1–322. Google Scholar WorldCat Heine, Bernd, and Zelealem Leyew. 2008. “Is Africa a Linguistic Area?” In African Languages, edited by Bernd Heine and Derek Nurse, 15–35. Cambridge: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Heine, Bernd, and Derek Nurse, eds. 2000. African Languages. Cambridge: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Heine, Bernd, and Derek Nurse, eds. 2008. A Linguistic Geography of Africa. Cambridge: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Labov, William. 1972. “Phonological Correlates of Social Stratification.” In Directions in Sociolinguistics, edited by J. J. Gumperz and D. Hymes, 164–176. New York: Holt, Rinehart, Winston. Google Scholar Google Preview WorldCat COPAC Lewis, M. Paul, ed. 2015. Ethnologue. Dallas: SIL International. Google Scholar Google Preview WorldCat COPAC Nurse, Derek. 1997. “The Contributions of Linguistics to the Study of History in Africa.” Journal of African History 38: 259–391. Google Scholar WorldCat Pickrell, Joseph K., Nick Patterson, Chiara Barbieri, Falko Berthold, Linda Gerlach, Tom Güldemann, Blesswell Kure, et al. 2012. “The Genetic Prehistory of Southern Africa.” Nature Communications 3: 1143. Google Scholar WorldCat

p. 44

Thomason, Sarah, and T. Kaufman. 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press.

Google Scholar

Google Preview

WorldCat

COPAC

Notes 1. 2. 3. 4. 5. 6. 7.

8. 9. 10. 11.

The languages mentioned (Sanskrit, Greek, Latin, Persian, Gothic, and Celtic) and many others (a total of some 445) are all members of the Indo-European family. A phylum is a group of related families, so a phylum is larger than a family and its status less certain. Bantu languages would be a family; the larger grouping to which Bantu belongs, Niger-Congo, is a phylum. For earlier forays into historical comparisons and reconstructions, see https://en.wikipedia.org/wiki/Comparative_method_(linguistics). It is impossible to list the over 2,000 African languages, or even their subfamilies, for reasons of space. For the languages readers should consult Lewis et al. (2015) and for the families Heine and Nurse (2000). All dates and total language numbers given here are approximate. There are many good books on these topics. Thomason and Kaufman is among the best. Over the past decade, I have listened to many recording of Bajunis from southern Somalia who are refugees in Europe. Some have lived for some time in camps where they speak a form of Swahili with other refugees from di erent parts of East Africa. Some speak a Swahili which has features not directly attributable to Bajuni or Standard Swahili. I would guess these features have to do with the variety of Swahili used for inter-group communication in the camps. It will not last, for obvious reasons. With a few isolates. See MacEachernʼs chapter 4 in this volume for a discussion of climate factors in the Chad Basin. Dahalo in NE Kenya is omitted. Not only linguistically crowded. An issue not raised here is the possible connection between the distribution of language communities and that of flora and fauna. A glance at maps of the world makes it clear that there is a great abundance of species and numbers of flora and fauna in the tropics, which biologists interpret as the result of ecological factors. The same distribution is also true of language communities, of which there are many more and with greater typological diversity within the tropics. This appears to be more than coincidence. Until recently, humans were hunters/gatherers or farmers, and they clustered where the resources were. Today, when humans donʼt depend so much on flora and fauna, that clustering is less important, and far greater numbers of people now live outside the tropics.

Tracing Language Movement in Africa Ericka A. Albaugh (ed.), Kathryn M. de Luna (ed.) https://doi.org/10.1093/oso/9780190657543.001.0001 Published: 2018

Online ISBN: 9780190657574

Print ISBN: 9780190657543

CHAPTER

3 The Ethnologue and L2 Mapping  Kenneth S. Olson, M. Paul Lewis https://doi.org/10.1093/oso/9780190657543.003.0003 Published: February 2018

Pages 45–66

Abstract The early focus of the Ethnologue was on L1 use and is re ected in the maps that are included with each new edition. Typically, the maps show locations and boundaries corresponding to the distribution of L1 speakers. The location of widespread, second, or additional languages (such as national languages, lingua francas, and languages of wider communication) is only occasionally represented by maps, using a variety of methods. Major factors a ecting this e ort are related to language identi cation (ISO 639-3), categorization (status: sociohistorical, o

cial recognition, vitality), and analytical and

research methods (lexical similarity, intelligibility, bilingual pro ciency). This chapter examines the Ethnologue’s approach in all of these areas. Currently, signi cant e ort is being made to structure the Ethnologue database to provide expanded data on the ecological setting of each language. This should signi cantly increase capacity for mapping the use of widespread L2s. A sample map showing the use of Lingala/Bangala is provided.

Keywords: bilingualism, Ethnologue, intelligibility, ISO, L2, language ecology, language mapping, language status, lexical similarity, lingua franca Subject: Sociolinguistics Collection: Oxford Scholarship Online

Introduction The Ethnologue (Simons and Fennig 2017) was founded by Richard S. Pittman in 1951 ( rst edition) with the primary goal of providing information on Bible translation needs around the world. This initial motivation, combined with the perspective of many Bible translators that “the mother tongue speaks to the heart,” in uenced the development of the resource over the years leading to its longstanding focus on rstlanguage (L1, native language, mother tongue) use. The initial focus also was predominantly on smaller, less-well-known languages rather than major national and international languages for which Bible translations already existed. In 1971, the scope of the research was expanded to include “all known languages of the world,” and that expanded data began to appear in the eighth edition (1974). Later, with the advent of ISO 639-3 (see below), that scope was constrained somewhat to include primarily the living 1

languages of the world. p. 46

2

This early focus on L1 use is evident in the maps included in the Ethnologue. Typically, the maps show locations and boundaries corresponding to the distribution of L1 use. The locations are indicated by language names, or, if the names do not t on the map, reference numbers. Polygons are used to approximate language boundaries when the boundaries are known. The location of widespread, second, or additional languages (L2, such as national languages, lingua francas, and languages of wider communication) is only occasionally represented by maps, and this in a variety of ways—for example, by the inclusion of separate insets or by overlays (grid patterns) within a map. The representation of L2 use is currently not exhaustive. In the 15th edition (2005), the Ethnologue’s system of three-letter codes was adapted to match the 3

International Organization for Standardization (ISO) 639 draft standard for language identi cation. The existing codes in the ISO 639 standard had been used primarily by librarians as a system for cataloging materials by their language of publication and provided a set of codes for about 300 languages. Computer software developers as well as linguists found those to be inadequate for other purposes. ISO 639 part 3 (ISO 639-3) was developed to enable any user to identify all of the languages of the world, including ancient, 4

classical, extinct, and living languages, and received nal approval in 2006.

Given the Ethnologue’s status as a primary encyclopedic resource on language identi cation, and especially given its close historic ties to the ISO 639-3 language codes, a closer scrutiny of the constraints and possibilities that contribute to its currently limited mapping of L2s is merited. The Ethnologue contains a vast amount of information, yet it is properly viewed as a work in progress, with a new edition currently released annually. The ongoing research program of the Ethnologue seeks to p. 47

5

continuously improve the accuracy of the data included in the resource. It has

been recognized that

providing more exhaustive coverage of L2 use in the Ethnologue, including improved mapping of L2s, would 6

better re ect the actual linguistic situation on the ground (the language ecologies), which in turn would make the resource more valuable to linguists, educators, economists, government workers, language policy experts, and others who rely on it. In this chapter, we discuss aspects of the Ethnologue and ISO 639-3 that inform the mapping of L2s. First, we provide an overview of the principles employed by ISO 639-3 for language identi cation and re ected in the Ethnologue and how the Ethnologue categorizes language status and vitality. Then we examine recent modi cations to the Ethnologue database in order to better describe L2 use. Next we discuss how this enhanced database structure might be exploited in creating L2 maps. Finally, we end with some concluding remarks.

Language Identification The Ethnologue follows ISO 639-3 in reporting the inventory of languages identi ed by that standard. ISO 639-3 includes three primary criteria for language identi cation. The rst (and basic) criterion used by ISO 639-3 is mutual intelligibility: Two related varieties are normally considered varieties of the same language if speakers of each variety have inherent understanding of the other variety at a functional level (that is, can 7

understand based on knowledge of their own variety without needing to learn the other variety).

This statement invites three comments. First, the notion that a community shares a “speech variety” is an approximation. Each individual speaker has his/her own idiolect in which he/she exhibits distinct and unique language use. So in order to speak of a community speech variety we need to establish that the p. 48

community shares the vast majority of their speech habits, approximations can be useful but are also fraught with di

with only minor di erences. Such culties in some cases.

The Ethnologue, in its introductory material, describes three di erent perspectives on language identi cation: (1) language as particle (discrete, countable units), (2) language as wave (waves of change through time and through geographic and social space), and (3) language as eld (languages in ecological systems and hierarchies) (Pike 1959; Lewis 1999). The dominant focus of the Ethnologue has been the rst of these, identifying, categorizing, and counting discrete “languages.” It has struggled with and been criticized for failing to deal adequately with the second perspective, as it is the wavelike nature of linguistic variation that makes the language-as-particle perspective problematic. Increasingly, linguistics itself has focused more on the variation than on the similarities in the linguistic behavior of individuals and communities, and so the identi cation, categorization, and counting of “languages,” while still serving as a starting point for most sociolinguistic descriptive work, is being given less importance in the overall program of linguistics in general. More recently the research focus of the Ethnologue has shifted to pay greater attention to the third perspective, languages and their roles in ecological contexts. The data with which the Ethnologue originally worked in making language identi cation decisions (a function now passed to ISO 639-3, see below) consists of bundles of shared linguistic features. Based on experience, research, and the triangulation of di erent kinds of measures of “sharedness,” more or less arbitrary thresholds have been identi ed which serve as the structural boundaries between languages. Much of the discussion about whether a particular language identi cation is correct or not centers around di erences in the criteria that are to be used and the levels of similarity at which the thresholds should be set. There is a general consensus on these points, perhaps because of the Ethnologue’s early dominance in the language identi cation arena. The second observation on the ISO 639-3 statement is that the level of intelligibility between two speech varieties is a gradient measure rather than a binary opposition, so that we need to speak of degrees of intelligibility. In other words, it’s not always a matter of whether or not two speech communities can understand each other, but rather how much they can and are willing to understand each other. As a result, deciding whether two speech varieties constitute two separate languages or two dialects of a single language requires de ning the degree of intelligibility where the distinction should be made. Thirdly, the identi cation of one linguistic variety as a separate language from another linguistic variety is p. 49

not entirely a matter of linguistic

similarity or di erence. The perceptions and attitudes of the speakers

of the varieties in question play an important role in determining whether the two will be united and assigned a single ISO 639-3 code or separated and assigned separate codes (see below).

These two categories of criteria—(1) the structural similarity or di erence as measured by linguistic (usually lexical) comparisons and inherent intelligibility, and (2) the functional language-in-use perspective—were identi ed early on by Haugen (1966). Sometimes they align well with each other, making language identi cation fairly clear and obvious. In other cases, however, the structural description of the varieties in question lead to one conclusion, but the perceptions and actual behavior of the users of the varieties indicate that a very di erent decision should be made. A particular challenge for de ning an individual language is the notion of dialect continuum (Bloom eld 8

1933, 51). Consider three speech varieties: A, B, and C. In a dialect continuum, varieties A and B are mutually intelligible and varieties B and C are mutually intelligible, but varieties A and C are not mutually intelligible. According to the structural criterion of mutual intelligibility, varieties A and C should be considered separate languages. But the question that arises is this: Should variety B be considered a part of language A or language C? The ISO standard establishes the principle that in such cases, varieties A and C should be considered dialects of B, which is viewed as the “center” of the linguistic space denoted by the assigned ISO code. Sociolinguistic factors may also come into play in making such decisions. Well-known examples of dialect continua in Africa include the Fula speech varieties that are scattered from Senegal in the west (Pulaar) to Sudan in the east (Adamawa Fulfulde) (Breedveld 2009) and the Colloquial Arabic varieties across North Africa (Watson 2002). An additional observation regarding language identi cation is a more practical one. Measuring intelligibility can be a challenge. It is impossible to measure directly what any individual understands. That can only be assessed indirectly. The established method involves a somewhat more direct measure, such as the Recorded Text Test (RTT), which was pioneered by John Crawford and elucidated by Casad (1974). This method is an adaptation of Voegelin and Harris’s (1951) earlier work, taking into account Wol ’s (1959) critique. The RTT uses recorded texts with a set of questions. The ability of a speaker of a variety other than that used p. 50

in the recorded text to answer the

questions is taken as a measure of the intelligibility of the recorded

variety. There are strict controls on subject selection, the nature of the texts that are recorded, and the nature and number of the questions (which are translated into the variety of the subject). Even so, the RTT methodology has been critiqued (cf. O’Leary 1994) for the number of other intervening variables which could a ect the outcome and the kinds of conclusions that have been drawn based on this one measure. While generally e ective as a measure of intelligibility, implementation of the Recorded Text Test can be logistically challenging. As a result, other methods have at times been used, which assume a correlation between intelligibility and some other measure. The most common proxy is the use of lexical similarity as an estimate for intelligibility. Swadesh (1952) proposed that two speech varieties are distinct languages if lexical similarity is less than 81%. However, the received view is that use of lexical similarity is not an ideal measure of intelligibility, as many additional factors besides the lexicon, such as phonological, grammatical, and semantic di erences, contribute to the increase or decrease of intelligibility. At the same time, it can be a useful initial tool for making some broad decisions. For example, if the lexical similarity between speech varieties is less than 60%, mutual intelligibility between them is highly unlikely (Grimes 1992, 32). If lexical similarity is more than 60%, then more direct intelligibility testing is warranted. Even where the structural features may seem clear, a second, functional factor that informs the identi cation of a variety as a separate language from another is the following:

Where there is enough intelligibility between varieties to enable communication, the existence of well-established distinct ethnolinguistic identities can be a strong indicator that they should nevertheless be considered to be di erent languages. Kinyarwanda and Kirundi are mutually intelligible (Kimenyi 2009). However, each of the two speech varieties relates to a strong national identity (Rwanda and Burundi, respectively), since each is a statutory national language (Simons and Fennig 2017). In addition, the two have extensive separate literatures. As a result, the two varieties are considered to be separate languages within the ISO 639-3 standard. And a third factor, also functional in nature, which informs the identi cation of a language is the following: Where spoken intelligibility between varieties is marginal, the existence of a common literature or p. 51

of a common ethnolinguistic identity with a central

variety that both understand can be a strong

indicator that they should nevertheless be considered varieties of the same language. In the ISO 639-2 language code standard, several “languages” were included which actually consisted of multiple mutually unintelligible speech varieties. When the ISO 639-3 standard was formulated, the notion of “macrolanguage” was created as a pragmatic tool to map the multiple codes in the ISO 639-3 standard to the corresponding single code in the ISO 639-2 standard. One example of a macrolanguage is the “Arabic language” (ISO 639-2 code: [ara]). The Arabic world has a very strong common ethnolinguistic identity, but the term “Arabic language” actually comprises several speech varieties: (1) Classical Arabic, which is employed in the Quran and treated in the Ethnologue as a dialect of (2) Modern Standard Arabic (ISO 639-3 code: [arb]), a modern variety that derives from Classical Arabic, which is learned in formal education and is the most common written form today, and (3) Colloquial Arabic, which refers to the regional varieties spoken across the Middle East from Morocco to Oman. The many varieties of Colloquial Arabic form a dialect continuum. Watson (2002) writes, “Dialects of Arabic form a roughly continuous spectrum of variation, with the dialects spoken in the eastern and western extremes of the Arab-speaking world being mutually unintelligible.” The common situation in places where Arabic is in use is one of diglossia, where individuals use several speech varieties in di erent sociolinguistic domains—they speak a colloquial variety in day-to-day communications (such as Moroccan Spoken Arabic, Sudanese Spoken Arabic, etc.); read Modern Standard Arabic in books, newspapers, and other literature; and recite Classical Arabic in the mosque. There are anecdotes related by foreign diplomats who have studied Modern Standard Arabic of reading the same newspaper as their chau eur but being unable to understand each other’s speech. Given the challenges of de ning and measuring intelligibility, as well as issues of incorporating sociolinguistic factors into the assessment, one could ask, “Why pursue language identi cation?” Simons (2009) points out the value of a discipline holding to standards, particularly when it is an approximation of the reality. Simons presents as an analogy the standardization of time. Without a division of time into time zones, local time would be a continuous function, shifting as one travels from east to west (four minutes per degree of longitude). There would e ectively be an in nite number of noons, and the local time would depend on one’s location on the east-west continuum. This perspective re ects physical reality, but it su ers from p. 52

impracticality.

You would have to reset your watch every time you take a step to the east or west.

On the other hand, the division of time into time zones results in local time being a step function. Each of 9

the 24 time zones comprises 15 degrees of longitude, so that there are exactly 24 noons. Changing time zones simply involves adjusting a timepiece forward or backward in time by the appropriate number of hours. While this less closely re ects the physical reality, it greatly improves interoperability. Everyone

within the same time zone shares the same time, simplifying its measure for uses such as commerce, utilities, government services, and social interaction. The identi cation of languages as discrete units based on both structural and functional criteria as described above provides a reasonable starting point for the description of language ecologies. The generalization, as with standardized time zones, loses the ner details of the variation inherent in the reality, but it provides a way to make global comparisons and categorizations possible.

Language Status The Ethnologue currently reports on three aspects of a language’s status: sociohistorical status, vitality status, and o

cial recognition status. One major innovation, introduced in the 19th edition, is to categorize

each language-in-country entry in terms of its sociohistorical status in the country. Some languages are “indigenous,” having been present in the geographic territory now identi ed by a modern nation-state’s borders for a long enough time depth so as to be understood, for all practical purposes, to have “always been there.” Native American and Australian Aboriginal languages provide clear examples of the kinds of languages that fall into this category. At the other extreme are recent arrivals with no longstanding multigenerational community of users present within a country. These are categorized in the Ethnologue as “immigrant” languages. Diaspora communities eeing war or searching for economic opportunity are those most often included in this category. Because of their transitory nature, the Ethnologue has very little data on immigrant language communities and provides only a listing of those that are known, with an estimate of their population when known. p. 53

Of particular interest for this discussion, however, are those languages now identi ed as “non-indigenous.” These at one time in history would have been categorized as immigrant languages but through either (1) longstanding presence and continued intergenerational transmission or (2) widespread acquisition as a second language are now considered to have become fully and rmly established in the country. South African English is an example of the rst type (Bekker 2012), while English, French, and Portuguese in Africa are examples of the second type. Widespread second language education has made them established languages of wider communication used extensively in some domains (e.g., education, government) and with multiple functions assigned to them. Until this distinction between indigenous and non-indigenous languages was made, it was di

cult for the

Ethnologue to report on the presence of major languages (and others) used primarily as L2s and as Languages of Wider Communication in many parts of the world. The number of languages categorized as indigenous and non-indigenous is reported in the Language Counts section for each country. The Language Status eld in each Ethnologue language entry provides a second set of categorizations. First of all, this eld identi es the language’s position on the Expanded Graded Intergenerational Disruption Scale (EGIDS; Lewis and Simons 2010), which is a revision of Fishman’s (1991) GIDS scale. Fishman’s scale is a measure of the disruption of intergenerational transmission of a language. The EGIDS builds on that, using other factors beyond intergenerational transmission, to provide an estimate of the overall development versus endangerment of a language (see Table 3.1).

1

1

Table 3.1 Expanded Graded Intergenerational Disruption Scale

1

Level

Label

Description

0

International

The language is widely used between nations in trade, knowledge exchange, and international policy.

1

National

The language is used in education, work, mass media, and government at the national level.

2

Provincial

The language is used in education, work, mass media, and government within major administrative subdivisions of a nation.

3

Wider Communication

The language is used in work and mass media without o icial status to transcend language di erences across a region.

4

Educational

The language is in vigorous use, with standardization and literature being sustained through a widespread system of institutionally supported education.

5

Developing

The language is in vigorous use, with literature in a standardized form being used by some though this is not yet widespread or sustainable.

6a

Vigorous

The language is used for face-to-face communication by all generations and the situation is sustainable.

6b

Threatened

The language is used for face-to-face communication within all generations, but it is losing users.

7

Shi ing

The child-bearing generation can use the language among themselves, but it is not being transmitted to children.

8a

Moribund

The only remaining active users of the language are members of the grandparent generation and older.

8b

Nearly Extinct

The only remaining users of the language are members of the grandparent generation or older who have little opportunity to use the language.

9

Dormant

The language serves as a reminder of heritage identity for an ethnic community, but no one has more than symbolic proficiency.

10

Extinct

The language is no longer used and no one retains a sense of ethnic identity associated with the language.

Source: Simons and Fennig 2017.

The numbers on the EGIDS range from 0 to 10. The smaller the number on the scale, the more robust is the use and vitality of the language. The larger the number, the more endangered the language is. Level 10—the largest number, indicating the greatest level of disruption in language vitality—corresponds to an extinct 10

language.

In terms of endangerment, the division between levels 6a and 6b is important. Level 6a refers to vigorous oral language use by all generations of speakers (both old and young), so that long-term use of the language is not threatened. This has generally been considered to be the normative case, but that assumption has been increasingly called into question. The most recent statistics available to us indicate that 2,468 of the p. 54

7,097

known living languages are at this level. Level 6b is the tipping point; it also refers to vigorous use

by all generations of speakers, but the language is starting to lose users due to various sociolinguistic factors. Just over 1,000 of the world’s languages are at this level of vitality. Though L2 use can be present at any level on the EGIDS, for level 6b and below, crucially, there is an encroaching language that needs to be identi ed. That is, second-language use (and related sociolinguistic

p. 55

factors)

is usually (if not always) the major factor that contributes to the disuse of an L1. Bilingualism is a

necessary but not a su

cient condition for language shift.

For some languages, the available data do not allow an accurate assessment of the language’s position on the EGIDS. In these cases, the Ethnologue initially assigned the default status of EGIDS 6a, pending further research. In a few cases, informed investigators proposed that EGIDS 6b should be set as the default EGIDS level assignment in speci c countries. With one exception, the scale assumes that each stronger level of vitality entails the characteristics of the levels below it. Thus, for example, if a language is characterized as EGIDS 5 (Developing), this assumes that it is also characterized as EGIDS 6a (Vigorous). The one exception is that EGIDS 3 (Wider Communication) entails neither the existence of written literature (EGIDS 5) nor its use in formal education (EGIDS 4). This exceptional case in the EGIDS merits further comment for this discussion. In evaluating the factors that a ect the vitality of a language, the general chain of logic builds upon the notion that a language associated with an identity (EGIDS 9), even with no fully pro cient speakers, is stronger than a language with neither speakers nor an associated identity (EGIDS 10). EGIDS levels 7 through 8b use the state of intergenerational transmission as the primary distinguishing factor, with the loss of each succeeding older generation marking the decline of the language on the scale. As described earlier, EGIDS 6a and 6b represent a kind of fulcrum or tipping point between full vigorous use with diglossia (at EGIDS 6a) and without diglossia (at EGIDS 6b). Succeeding stronger levels on the scale then represent stages of language development rather than decline. The presence and strength of literacy in the language is the most signi cant factor, with incipient literacy (EGIDS 5) gaining acceptance and becoming well-established, with institutionally supported use in education (EGIDS 4). At EGIDS 3, the focus of the scale shifts to “vehicularity,” the use of the language to transcend ethnolinguistic boundaries. Consequently, the languages classi ed as EGIDS 3 (Wider Communication) are, by de nition, used as L2s in addition to their use as L1s. Widespread second-language use may also be a prevalent feature of languages at EGIDS 0 through EGIDS 2. However, at the EGIDS levels that are stronger than EGIDS 3, it is the level of o

cial recognition of the language that is most in focus.

A Language of Wider Communication (EGIDS 3) does not necessarily have to be a written language. Many lingua francas and LWCs are not written in any standardized way, but serve as oral means of intercommunication across ethnic and linguistic boundaries. The EGIDS takes this vehicularity to be an p. 56

indicator of stronger vitality than the use of the language for

reading and writing. The scale does assume,

however, that a language that is at EGIDS 3 also meets the requirements to be at EGIDS 6a. That is to say, a language that is not fully in use by its L1 community does not meet the criteria to be classi ed as EGIDS 3 even if it does have some L2 users. The Ethnologue classi es only a few, mostly liturgical, languages as Second Language Only when it is the case that there are no L1 users anywhere, but the language is used more or less widely as an L2. Well-known LWCs in Africa would include all of the former colonial languages and widespread languages such as Lingala and Sango. The evaluation of all of the languages of a country, region, or continent using the EGIDS has enabled us to develop vitality pro les of those units of analysis. Figure 3.1 shows the vitality pro le of the languages of Africa.

Figure 3.1

EGIDS Profile of Language Vitality in Africa Source: Simons and Fennig 2017 In contrasting the pro les of Europe, the Americas, and Africa, Simons and Lewis (2013) proposed that these pro les might be correlated with the di ering patterns of colonization as suggested by Mufwene (2002). The pro le for Africa shows relatively few languages at the higher (EGIDS 0—4) and lower (EGIDS 8a—9) ends of the scale, with the preponderance of the languages being maintained orally or in the initial stages of literacy. Secondly, the Language Status eld also presents information concerning o

cial recognition of the

language. The Ethnologue’s categorization is based on Cooper (1989, 99–103), but adds further distinctions. O

cial recognition can be described in terms of three parameters: the scope of the recognition (national,

provincial), the nature of the recognition (statutory, de facto) and the function of the recognized language p. 57

(working, identity, both). The combination of these three parameters and their values of categories for describing the level and nature of o

provides a rich set

cial recognition.

These three major categories of language status description—sociohistorical status, vitality status, and o

cial recognition status—then work together to provide a fuller picture of the language ecology. In many

cases, particularly when a language is functioning as a working language within a country, it is a second language for many, if not all, of the citizens of the country. So, for example, English or French as an established, EGIDS 1, working language of governance may not be the L1 of any of those who use the language regularly in the operation of the government. Though there may be high levels of pro ciency among its users, it is often a language that they learned in school and not in their home.

Language Use: Bilingualism The categorization of a language as a speaker’s “ rst” or “second” language is problematic. The predominant language ecologies, however, at the time when the Ethnologue project was begun were, at least for the languages of primary interest to the project, made up of monolingual and relatively isolated (or segregated) groups where the complexities of multilingualism were not much in evidence. This meant that the Ethnologue research could proceed and develop without having to wrestle with the de nitional complexities that the much more prevalent multilingual language ecologies now present. Most minority language communities around the world today are multilingual, and most descriptions of language use in these communities make use of the concept of “domains of use.” Technically, a domain of

use is de ned as the constellation of participants, location, and topic that is strongly associated with a particular language. In practice, domains are most often identi ed by location. So a language will be said to be used in the Home domain, or in the Market, or in domains that are Public or Formal (Ervin-Tripp 1964; Ferguson 1959; Fishman 1965). In many cases, the domains of use of the various languages are in complementary distribution (e.g. diglossia), so that there is no pressure toward loss of L1. This behavior corresponds to EGIDS level 6a and above. In other cases, two or more languages are in competition for the same domains of use. Such a case can contribute to younger speakers beginning to use a second language instead of the language of their parents. This behavior corresponds to EGIDS level 6b and below. p. 58

The Language Use category in Ethnologue entries pulls together several di erent types of information, including viability remarks, domains of use, user age groups, language attitudes, bilingualism remarks, and use as a second language. Historically, the content of this eld has been populated on an ad hoc basis with only loosely de ned terminological constraints. Of particular interest to us here are the bilingualism remarks and use as a second language. In 2014, the separate database elds that comprise the Language Use category in the Ethnologue were restructured in order to allow for better tracking of L2 use. The use of a second language by the speech community is now tracked more precisely in the Ethnologue database, and the use of a second language is indicated in published language entries by the phrase “Also use . . .” followed by the name and ISO 639-3 code for the additional language. This data structure was further consolidated and expanded in the 19th edition (2016) with the more frequent inclusion of quanti ers (All, Most, Many, Some, Few) to indicate the scope of the bilingualism where it is known (see below). This database structure allows the database to be queried to identify all languages that are identi ed as being “also used.” That use of a language as a second language by other speech communities is then indicated by the phrase, “Used as L2 by . . .” followed by the name and ISO code of each language. Currently, the Ethnologue is in the process of populating both the Language Use elds that describe L1 use and the Language Contact data elds that focus on L2 use with much more structured information on the overall language ecology as data become available. Given the enormity of the task, this may take some time, though signi cant progress is being made. The most signi cant complicating factor when dealing with the description of bilingualism is a de nitional one. What level of pro ciency in an L2 must an individual reach in order to be considered a “speaker” of that language? And what percentage of the total population of a community must be at that level of pro ciency in order to make the claim that the language in question is being used widely enough to be included in a map showing L2 use? The existence of a few individuals who have, perhaps, studied a second language in school is a threshold set too low. But requiring that a majority of the population have high levels of pro ciency in an L2 very likely sets the bar too high. As with language identi cation, there are assessment tools that can be used to gauge an individual’s pro ciency in an L2, though the more thorough and accurate the tool, the greater the practical and logistical

p. 59

issues that accompany its implementation. Even so, the characterization

of societal L2 use requires an

adequate sampling of individuals in a community (region, country) and an analysis of the levels of pro ciency represented within that sample. When L2 use information is reported to the Ethnologue, it is often in the form of somewhat ambiguous statements such as “High level of bilingualism in Language X.” It is almost impossible to determine what such statements mean. Are there many (most? all?) who use the L2? Or, do those (few) who use the L2 have high levels of pro ciency in it? Some sociolinguistic surveys have gone so far as to engage in relatively thorough investigations of L2 pro ciency and use, but there are not nearly enough of those to be able to use the data for global or even regional comparisons.

It is often the case that L2 use is closely associated with particular domains. An LWC may be most often used in the Market domain, for example, so pro ciency will be high when evaluated in terms of an individual’s ability to navigate buying and selling. Pro ciency in that same language may be extremely low when the individual is confronted with di erent topics in a di erent location. And such use may be strati ed within the community—women who go to the market, but not men, older people but not the young (or vice versa). Given the global scope of the research project, the Ethnologue makes no attempt to describe levels of L2 pro ciency. Even when those are reported, it is di

cult to determine what criteria have been used to

evaluate the pro ciency of speakers. With di erent criteria being used by di erent sources in di erent parts of the world, the resulting data reported would lack any basis for reliable global comparisons.

Toward Creating L2 maps As described above, the new structure for representing second-language use in the Ethnologue consists of data elds that identify the L2s used by any given language community, a set of quanti ers that describe the scope of that use (All, Most, Many, Some, Few), and a brief clarifying comment. Combined with a more structured approach to identifying the location (region in country) of each identi ed language that is still only partially implemented, the capacity to produce L2 maps (in contrast to the existing L1 maps) is greatly enhanced. Though desirable, it is almost a futile task to try to get precise counts of L2-speaker populations, though those are occasionally reported to the Ethnologue. Few, if any, national censuses investigate secondp. 60

language use in a rigorous way, and few local investigators manage to get even accurate

L1-speaker

populations. In addition, the de nitional issues of what constitutes a “speaker” (level of pro ciency, frequency of use, scope of domains of use, etc.), as described above, render unfeasible a detailed investigation on a global level. The quanti er categories are a reasonable approach to getting estimates in a way that can be applied consistently across all of the languages of the world. Each quanti er level is de ned in terms of a percentage range rather than in terms of speci c numbers (see Table 3.2). Table 3.2 Quantifiers Describing the Degree of L2 use in a Speech Community Quantifier

Criteria

Few

2% to 10% of the ethnic population use the Contact Language as an L2

Some

Greater than 10% but less than or equal to 40% of the ethnic population use the Contact Language as an L2

Many

Greater than 40% but less than or equal to 60% of the ethnic population use the Contact Language as an L2

Most

Greater than 60% but less than 95% of the ethnic population use the Contact Language as an L2

All

95% or more of the ethnic population use the Contact Language as an L2

For the purposes of display on a map, the polygons for the L2 locations could be displayed in a di erent color or with a di erent pattern according to the value of the quanti er. As with all generalizations, it is likely that there will be exceptions, outliers, or nuances that the Ethnologue will continue to report, albeit tersely, in the comment eld.

In mapping L2s, the same issues apply that must be dealt with when mapping L1s. First, where multiple languages of either sort are used in the same geographical region, polygons will overlap. Frequently this situation is mapped in the Ethnologue using distinctive crosshatch or grid patterns in the polygons. Similarly, widespread languages, such as English in the United States, generally have a polygon that is contiguous with the borders of the country that is in focus. Sometimes those polygons are simply not displayed in Ethnologue maps, though there may be a note or comment identifying the widespread languages. Second, some languages are scattered, used by nomadic groups (e.g. Tuareg, Fulani, Maasai) or more frequently by people in isolated small pockets within a country. The latter is often the case, for example, p. 61

with Deaf sign languages, which tend to concentrate around the locations of

Deaf schools or community

clubs. Mapping those dispersed communities often becomes an issue of scale, with small-format maps not adequately able to identify the location with more than a dot on the map. On large-format maps, the opposite problem occurs where there isn’t enough data available to accurately draw a polygon that represents the boundaries of the language community or where the users of the language are so dispersed in geographic space that no polygon can adequately cover them all and any attempt to do so would imply that the speaker community occupies a much larger space than they actually do. A third consideration is that frequently lingua francas and Languages of Wider Communication cross national borders. The Ethnologue is organized around the notion of “language-in-country,” and maps are generally created country by country. Languages that spread beyond and across national borders may not always be treated equally in the maps of di erent countries. For example, a widespread language may be the national language in one country but only the language of a signi cant minority population in another (e.g. Swahili). Increasingly, diaspora language communities that are widely scattered globally, but often concentrated locally in multiple locations, also present challenges for cartographers. The representation of not only the presence but the patterns of use in urban settings is of particular interest, but still lacking in su

cient data to be usefully presented. While regional, multinational, and other specialized “theme” maps

can be created, the current priority of the Ethnologue cartography team is to maintain, update, and gradually expand the inventory of country maps that can be viewed. In general, however, L2s, by the nature of their use as Languages of Wider Communication (EGIDS 3) or as Provincial (EGIDS 2) or National (EGIDS 1) languages, more often share the features of widespread languages and the geographic scope of their use is more likely to parallel provincial, regional, or national boundaries (or beyond). The distribution of national and provincial languages, in particular, very likely correlates to the existence and accessibility of education where those languages are taught and learned. Figure 3.2 provides an example of the type of L2 map that can be generated from the Ethnologue database. The map shows the geographic distribution of Lingala in the northwestern part of the Democratic Republic of the Congo (DRC) and the northern part of the Republic of the Congo (RC), indicated by gray shading. Lingala serves the function of lingua franca throughout this region. Also included on the map is the distribution of Bangala, a speech variety closely related to Lingala, which serves the same function in p. 62

northeastern DRC. Bangala is indicated by a diagonal line pattern. Overlap

p. 63

of the two speech varieties is indicated by both gray shading and the diagonal line pattern.

Figure 3.2

Geographic distribution of Lingala and Bangala In and around Kinshasa, Lingala has emerged as the L1 of a majority of speakers. This is indicated by dark gray shading. Note that this does not correspond to the place of origin of the language. Rather, Lingala arose as a lingua franca in the late 19th century in the mid-river region, about halfway between Kinshasa and Kisangani (Meeuwis 2010). Lingala use crosses the national boundary between DRC and RC, and it has similar—but not identical— o

cial recognition in the two countries. In DRC, it is a statutory provincial language in the northwestern

regions (2006, Constitution, Article 1(8)), whereas in RC, it is a statutory language of national identity (2002, Constitution, Article 6). Due to lack of data, we have not attempted to account for the degree of Lingala use in the various languages in the region.

Conclusion The production of accurate maps showing the distribution of second languages in Africa or anywhere else in the world has primarily been hindered by a lack of data. The Ethnologue has traditionally focused primarily on L1 use and has mapped as many of the rst languages identi ed by the ISO 639-3 standard as possible. As language contact increases as a result of increased mobility and access to means of communication, the use of second languages has become an area of increased interest. The Ethnologue has made the acquisition of better data on second-language use a priority in its research program and has restructured its database to accommodate more ne-grained data about the use of languages as L2. As more and more data are collected, this more granular data structure should increase our capacity to produce maps that show the distribution and levels of use of widespread second languages. The ability to produce maps of both L1 and L2 distribution, to identify communities where multiple languages are in use, and to identify languages used by multiple speech communities can provide useful tools for those engaged in language development work, language policy formulation, and in particular for educators and formulators of education policy and practice. In addition, the ability to correlate other kinds

of data with the L1 and L2 distribution data opens up the potential for new insights in the areas of political science, economics, sociology, and other areas of scienti c inquiry.

p. 64

References Bekker, Ian. 2012. “The Story of South African English: A Brief Linguistic Overview.” International Journal of Language, Translation and Intercultural Communication 1: 139–150. doi:10.12681/ijltic.16. Google Scholar WorldCat Bloomfield, Leonard. 1984 [1933]. Language. Chicago: University of Chicago Press. Google Scholar Google Preview WorldCat COPAC Breedveld, J. O. 2009. “Fulfulde.” In Concise Encyclopedia of Languages of the World, edited by Keith Brown and Sarah Ogilvie, 430–433. Oxford: Elsevier. Google Scholar Google Preview WorldCat COPAC Casad, Eugene H. 1974. Dialect Intelligibility Testing. Summer Institute of Linguistics Publications in Linguistics and Related Fields 38. Norman, OK: Summer Institute of Linguistics of the University of Oklahoma. Google Scholar Google Preview WorldCat COPAC Cooper, R. L. 1989. Language Planning and Social Change. Cambridge: Cambridge University Press. doi:10.1017/CBO9780511620812. Google Scholar Google Preview WorldCat COPAC Ervin-Tripp, Susan. 1964. “An Analysis of the Interaction of Language, Topic, and Listener.” American Anthropologist 66 (6) (Part 2): 86–102. doi:10.1525/aa.1964.66.suppl_3.02a00050. Google Scholar WorldCat Ferguson, Charles A. 1959. “Diglossia.” Word 15: 325–340. Google Scholar WorldCat Fishman, Joshua A. 1965. “Who Speaks What Language to Whom and When?” La Linguistique 1 (2): 67–88. http://www.jstor.org/stable/30248773. Google Scholar WorldCat Fishman, Joshua A. 1991. Reversing Language Shi . Clevedon, UK: Multilingual Matters. Google Scholar Google Preview WorldCat COPAC Grimes, Joseph E. 1992. “Correlations Between Vocabulary Similarity and Intelligibility.” In Windows on Bilingualism, edited by Eugene H. Casad, 17–32. Summer Institute of Linguistics and the University of Texas at Arlington Publications in Linguistics 110. Dallas: Summer Institute of Linguistics and the University of Texas at Arlington. http://www.sil.org/resources/archives/8408. Google Scholar Google Preview WorldCat COPAC Haugen, Einar. 1966. “Dialect, Language, Nation.” American Anthropologist 68 (4): 922–935. doi:10.1525/aa.1966.68.4.02a00040. Google Scholar WorldCat Haugen, Einar. 1972. The Ecology of Language: Essays by Einar Haugen. Stanford, CA: Stanford University Press. Google Scholar Google Preview WorldCat COPAC Kimenyi, Alexandre. 2009. “Kinyarwanda.” In Concise Encyclopedia of Languages of the World, edited by Keith Brown and Sarah Ogilvie, 604–610. Oxford: Elsevier. Google Scholar Google Preview WorldCat COPAC Lewis, M. Paul. 1999. “Language (Development) as Particle, Wave, and Field.” Notes on Sociolinguistics 4 (1): 3–19. Google Scholar WorldCat Lewis, M. Paul, and Gary F. Simons. 2010. “Assessing Endangerment: Expanding Fishmanʼs GIDS.” Revue Roumaine de

Linguistique 55 (2): 103–120. http://www.lingv.ro/RRL-2010.html. Google Scholar WorldCat Meeuwis, Michael. 2010. A Grammatical Overview of Lingála. LINCOM Studies in African Linguistics 81. Munich: Lincom Europa. Google Scholar Google Preview WorldCat COPAC Mufwene, Salikoko S. 2002. “Colonisation, Globalisation, and the Future of Languages in the Twenty-First Century.” International Journal on Multicultural Societies 4 (2): 162–193. http://www.unesco.org/shs/ijms/vol4/issue2/art2. Google Scholar WorldCat OʼLeary, Clare F. 1994. “The Role of Recorded Text Tests in Intelligibility Assessment and Language Program Decisions.” Notes on Literature in Use and Language Programs, Special Issue 3: 48–72. Google Scholar WorldCat Pike, Kenneth L. 1959. “Language as Particle, Wave, and Field.” Texas Quarterly 2 (2): 37–54. Google Scholar WorldCat

p. 65

Simons, Gary F. 2009. “Linguistics as a Community Activity: The Paradox of Freedom through Standards.” In Time and Again: Theoretical Perspectives on Formal Linguistics; In Honor of D. Terence Langendoen, edited by William D. Lewis, Simin Karimi, Heidi Harley, and Scott O. Farrar, 235–250. Linguistik Aktuell/Linguistics Today 135. Amsterdam: John Benjamins. doi:10.1075/la.135.18sim. Simons, Gary F., and Charles D. Fennig, eds. 2017. Ethnologue: Languages of the World, 20th ed. Dallas: SIL International. http://www.ethnologue.com. Google Scholar Google Preview WorldCat COPAC Simons, Gary F., and M. Paul Lewis. 2013. “The Worldʼs Languages in Crisis: A 20-Year Update.” In Responses to Language Endangerment: In Honor of Mickey Noonan, edited by Elena Mihas, Bernard Perley, Gabriel Rei-Doval, and Kathleen Wheatley, 3– 19. Studies in Language Companion Series 142. Amsterdam: John Benjamins. doi:10.1075/slcs.142. Google Scholar Google Preview WorldCat COPAC Swadesh, Morris. 1952. “Lexico-Statistic Dating of Prehistoric Ethnic Contacts: With Special Reference to North American Indians and Eskimos. Proceedings of the American Philosophical Society 96 (4): 452–463. http://www.jstor.org/stable/3143802. Google Scholar WorldCat Voegelin, C. F., and Zellig S. Harris. 1951. “Methods of Determining Intelligibility Among Dialects of Natural Languages.” Proceedings of the American Philosophical Society 95 (3): 322–329. http://www.jstor.org/stable/3143070. Google Scholar WorldCat Watson, J. C. E. 2002. The Phonology and Morphology of Arabic. Oxford: Oxford University Press. Google Scholar Google Preview WorldCat COPAC

p. 66

Wol , Hans. 1959. “Intelligibility and Inter-Ethnic Attitudes.” Anthropological Linguistics 1 (3): 34–41. http://www.jstor.org/stable/30022192. Google Scholar WorldCat

Notes 1. 2. 3.

Languages considered to be extinct before 1950 were removed from the Ethnologue, while those which have more recently gone out of use are included. The World Language Mapping System is the geographic dataset used to produce the maps in the Ethnologue: http://www.worldgeodatasets.com/language. ISO invited SIL International in 2002 to participate in the development of ISO 639-3. About 600 Ethnologue codes were

modified to align with ISO 639-2 codes. ISO 639-3 also incorporated codes for ancient languages from the Linguist List code set. The ISO 639-3 standard was approved in 2006 and published in 2007: http://www.iso.org/iso/home/standards/language_codes.htm. 4. SIL International, but not the Ethnologue o ice, serves as the Registration Authority for ISO 639-3: http://www01.sil.org/iso639-3, providing this service without compensation from ISO. 5. The Ethnologue is informed, in part, by the extensive language survey work of SIL. Many survey results are reported in the SIL Electronic Survey Reports series: http://www.sil.org/silesr. 6. An early, widely known application of the notion of ecology to language was by Haugen (1972), defined as the interactions between any given language and its environment. He identified the ecology of language as encompassing 10 areas of inquiry: classification, users, domains of use, concurrent languages, internal variation, written traditions, standardization, institutional support, attitudes, and status. The Ethnologue deals with each of these in at least summary form. 7. http://www.ethnologue.com/about/problem-language-identification 8. Bloomfieldʼs term was “dialect area.” 9. This ignores time zones that donʼt follow the standard, the implementation of daylight saving time, the variations in boundaries based on political or social factors, etc. 10. The use of larger numbers to indicate greater disruption of language transmission can be confusing. With 0 as the strongest level on the scale, we generally refer to the smaller numbers as being “higher” and larger numbers as being “lower” on the scale. The seeds of this chapter germinated during the first authorʼs participation in the workshop “Economy and Language: An InterDisciplinary Workshop” at the University of Chicago Center in Paris, June 19–20, 2014, organized by Salikoko Mufwene and Cécile B. Vigouroux. They then sprouted during the “Mapping Language Movement in Africa” symposium at Bowdoin College, September 25, 2015, organized by Ericka Albaugh. The authors wish to thank the participants at both events, as well as Gary F. Simons, for helpful comments, and Marcus Love, for the map.

Tracing Language Movement in Africa Ericka A. Albaugh (ed.), Kathryn M. de Luna (ed.) https://doi.org/10.1093/oso/9780190657543.001.0001 Published: 2018

Online ISBN: 9780190657574

Print ISBN: 9780190657543

CHAPTER

4 Understanding Distributions of Chadic Languages: Archaeological Perspectives  Scott MacEachern https://doi.org/10.1093/oso/9780190657543.003.0004 Published: February 2018

Pages 67–90

Abstract The distribution of Chadic languages in Africa is extremely diverse, including the widely dispersed Hausa language, the more restricted Central Chadic languages in the southern Lake Chad Basin, and the poorly understood Eastern Chadic languages in Chad. These distributions are disjunct in complex ways, and the relationships between Chadic and neighboring language families is extremely complicated. The genesis of these distributions lies in the mid-Holocene, with the occupation of the Lake Chad Basin by populations faced by the desiccation of the Sahara and the opening of arable lands further south. Further di erentiation of Chadic languages appears to be associated with sociopolitical developments in the region, especially over the last 1,000 years. This chapter will consider the methodological challenges associated with studying the history of these populations using archaeological, linguistic, and genetic data, as well as providing an initial framework for understanding the social dynamics within which these linguistic distributions emerged.

Keywords: archaeology, Chadic languages, Lake Chad Basin, Hausa, prehistory Subject: Sociolinguistics Collection: Oxford Scholarship Online

Introduction Any archaeological contribution to understanding the distribution of Chadic languages in Central Africa will probably not involve the identi cation of direct material correlates for Chadic speech communities, as such correlates rarely exist and are very di

cult to prove. Rather, the contribution of archaeology will lie in

elucidating the chronological, environmental, and socioeconomic frameworks within which language development and the spread of languages took place. As we will see, understanding the distribution of Chadic will involve a more general analysis of the process by which the Lake Chad Basin was settled by humans after the Last Glacial Maximum; there is no way to understand the genesis of these languages without also paying attention to the initial movement into the region of peoples from di erent parts of the subcontinent, almost certainly in a situation of environmental crisis and signi cant economic stress.

Any such analysis of archaeological data will also require that equal attention be paid to genetic and linguistic data that may help us understand these processes; this might seem obvious, but it is almost a p. 68

truism that past attempts to compare archaeological and linguistic distributions in Africa

have involved a

good deal of talking past one another by researchers—and today, this is frequently true for genetic data as well. The Lake Chad Basin is one of the very few areas in Africa where signi cant amounts of data from all of these disciplines are available to researchers, and we must learn to make better use of them. Finally, we need to note the patchy nature of archaeological evidence in the region. We have comparatively good data on archaeological site distributions in the areas between the Mandara Mountains and Lake Chad, in Nigeria and Cameroon, and in the areas around the Logone and near the historic lake-edge in Cameroon and Chad. However, we have almost no data—and essentially none relevant to the periods in question—from northeast and northwest of Lake Chad and almost none from areas south of the Mandara Mountains. Given the dynamic environmental regimes in the Lake Chad Basin during the Holocene (the geological period extending from the end of the last Ice Age about 11,000 years ago until the present), which would probably lead to them having been centers of cultural development at di erent times, this is a major problem.

Archaeology and Population Movements The archaeological study of human population movements, including those that might be associated with language spreads, has been a central element of the discipline since its inception. This has involved, for example, 19th century archaeological research that sought to identify the material traces of the Biblical wanderings of the Israelites, as well as classically known migrations from di erent areas of Europe and the Mediterranean Basin (Celts, the barbarian invasions of the Roman Empire, and so on). Perhaps the most striking example from this period was the search for archaeological correlates for the spread of the IndoEuropean language family—one strand of which culminated to disastrous e ect in the fantasy of an “Aryan” migratory master-race (see for example Lapouge 1899, 105–292) at the end of the 19th century. These analyses involved a set of assumptions that dominated the discipline over much of the last 150 years: (1) cultural change takes places because of the migrations of culture-bearing humans; (2) these migrations are relatively discrete and clear-cut processes; (3) there is a general correlation between material, ethnic, and linguistic identities; and thus (4) human cultures in the past can be identi ed by the existence of particular “type-fossils,” artefacts or assemblages of di erent kinds of artefacts that were unique to and characteristic of those ancient cultures. p. 69

The e orts that archaeologists made through the 20th century to identify particular assemblages of ancient material culture with speci c ancient societies (essentially, the “tribes” of cultural anthropology, itself a concept subject to very serious critique (Abu-Lughod 1991; Amselle 1985)) became steadily more sophisticated, and at the same time, paradoxically, less convincing. Ethnoarchaeological research made obvious the quite elementary conclusion that humans actively manipulate material culture in social strategies, for boundary maintenance but also for boundary e acement and the management and deployment of di erent aspects of identity (see for example Dietler and Herbich 1989; Galaty 1993). Under these circumstances, assumptions that artefacts and identity will co-vary in any essentialist way seem extremely naive. At the same time, we have abundant evidence that people do migrate, and also that—at particular times and places—certain constellations of material culture will act as foci for self-de nition and identity-formation within human groups. For archaeologists, this must lead to a methodological modesty, a questioning of easy assumptions about materials and identity. Concerning migrations, it leads to a focus upon the contexts—as noted above, technological, environmental, economic, political, and ideological— within which di erent kinds of sociocultural processes, including identity formation and migration, take place.

The Linguistic Situation The southern Lake Chad Basin contains signi cant linguistic diversity today, with Nilo-Saharan (Saharan, Central Sudanic), Afro-Asiatic (Chadic, Shuwa Arabic), and Niger-Congo (Adamawa/Savanna, Fulfulde) languages found across the region and often in close proximity (Barreteau and Dieu 2000; Blench 2010; Lewis 2009). The existence of linguistic isolates like Laal and Jalaa in Chad and Nigeria respectively (Blench 2010, 161–162), the uncertainty around the status of the Adamawa-Ubangian language family and especially of the Ubangian languages (Dimmendaal 2008), and the presence of Jarawan Bantoid languages in northcentral Nigeria and Cameroon hint at the possibility of even greater levels of linguistic diversity in the wider region in the Holocene. The complex imbrication of languages from all of these di erent families in the southern Lake Chad Basin testi es to complex processes of language contact and population movement that have been going on for millennia in this area. There is some disagreement about the position of Chadic within the broader Afro-Asiatic language family, p. 70

with some researchers claiming that

Chadic’s closest linguistic relations are the Berber languages of the

Sahara (Ehret 2006a; Fleming 1983), while others claim a closer relationship with Cushitic languages in northeastern Africa (Blench 1999), a distinction that has archaeological implications (see below). This paper will broadly concentrate on the Central Chadic languages, since we know virtually nothing about prehistory in the areas now occupied by West and East Chadic speakers. The internal relations of the various Chadic language groups are reasonably well accepted (Barreteau 1987; Newman 2013; Schuh 2003), with the status of the Masa language as a coordinate branch within Chadic along with West, Central, and East Chadic the main point of contention, and there seems to be a broad correlation between linguistic a

nities of the

Chadic languages and their spatial patterning. Most Chadic languages are found in restricted territories and with relatively small numbers of speakers, with the expansion of Hausa through the last millennium perhaps the most striking exception to this general pattern. Most of the populations living in and around the Mandara Mountains and on the plains between the mountains and Lake Chad speak Central Chadic languages today, with Kanuri-, Shuwa-, and Fulfulde-speakers occupying most areas of the plains where Chadic languages are not found (Barreteau and Dieu 2000). The latter two languages are relatively recent additions to the area, but Kanuri has much deeper roots in the Lake Chad Basin. There is a remarkable diversity of Central Chadic languages spoken in the Mandara massif itself over very short distances, with extensive multilingualism facilitating interaction between di erent groups. This linguistic di erentiation is central to the ethnic diversity of the area (Barreteau 1987; Barreteau et al. 1984; MacEachern 2001). These Chadic montagnard languages can again be divided into a number of subgroups (Gbwata-Margyi, Wandala, Mafa, Mbuko-Pelasla, Daba, and Kada groups especially), with geographical positioning generally paralleling linguistic relationships. Chadic languages spoken in the northeastern and northwestern extremities of the Mandara Mountains are found in particularly small areas and are spoken by particularly small numbers of people.

The Archaeological Record of Chadic Languages Early Holocene Beginnings Saharan expansion during the Last Glacial Maximum seems to have resulted in the more or less complete depopulation of the Lake Chad Basin: there is, at any rate, no persuasive archaeological evidence for p. 71

occupation during the period 20–10,000 BP, and the presence of dune systems through the

southern

Lake Chad Basin probably dating from that period indicates that the Sahara extended perhaps 400 km further south at that time. This actually makes populational reconstructions more straightforward, as it implies that a “clean slate” reoccupation of the basin took place as humans moved in to an empty Lake Chad Basin. This colonization probably started at some point after approximately 12,000 BP, on 1

palaeoenvironmental grounds. It is likely that we will never locate evidence of the initial phases of reoccupation near modern Lake Chad, because the rapid expansion to a Mega-Chad size of approximately 2

360,000 km , the size of the Caspian Sea (Bouchette et al. 2010), would have buried such sites under lacustrine sediments. The earliest evidence for human occupation of the southern Lake Chad Basin comes from Dufuna in northeastern Nigeria, now more than 300 km from Lake Chad but only about 50 km from the lake at that time; this evidence is an 8,000-year-old canoe, found 5 m underground by a Hausa farmer digging a well (Breunig 1995). The location where the Dufuna canoe was found would have been close to one of the watercourses that would have fed Lake Mega-Chad at that time.

Mid-Holocene Occupations There is thus little evidence for occupation of the southern Lake Chad Basin before about 6000 BP (MacEachern 2012a; 2012b; Marliac et al. 2000). This is probably because (a) in the very wet early Holocene, large parts of the area were permanently or seasonally ooded and subject to cattle diseases like trypanosomiasis, and thus of interest primarily to hunter- sher populations, while (b) the southern Sahara and northern Lake Chad Basin would have been comparably far more attractive environments for agropastoralists during this “Green Sahara” period. However, as noted above, those latter regions are almost unknown archaeologically. In the mid-Holocene, however, this situation would begin to be reversed, with the deserti cation of the former Green Sahara (Kropelin et al. 2008), the shrinking of Lake Mega-Chad toward its historic extent, and the corresponding shifting of fertile range- and croplands south of the lake. This very dynamic environmental situation—and the larger reorganization of environments through Central Africa as a whole over the period—needs to be taken into account during any consideration of the p. 72

evolution of Chadic and other Lake Chad Tishko

Basin language groups through time (Blench 2010; Ehret 2006b;

et al. 2009). We cannot assume that modern linguistic distributions of these families re ect the

geography of their initial encounters. All of the archaeological evidence available indicates that the mid-Holocene occupation of the southern Lake Chad Basin involved populations with Saharan material culture a

nities—that is, populations moving

into the region from the north (Breunig et al. 1996; Wendt 2007). Archaeological evidence includes the very small collection of pottery resembling Saharan ceramics from the Konduga site on the Bama Ridge dating to about 6300 BP, and the much more extensive and better-known Gajiganna Culture sites dating to after 4000 BP. The location of the Konduga site in an area that must have been very wet through much of the year (Ballouche and Neumann 1996; Wang et al. 2008), as well as charcoal evidence for woodland or forest environments at the time (Breunig et al. 1996), indicate that the area would not have been much use to cattle pastoralists—although that would have been the economy of contemporary Saharan populations using similar pottery. This suggests that Konduga may have been a temporary camp, perhaps associated with shing and not herding.

On linguistic grounds, large areas of the Green Sahara were probably occupied by populations speaking ancestral Nilo-Saharan languages (Drake et al. 2011; Ehret 2006b), and very likely their descendants were numbered among the environmental refugees during the period of desert expansion in the mid-Holocene. The situation with Chadic is less clear, because of the disagreements about the position of Chadic within Afro-Asiatic noted above: if Chadic is more closely related to Berber, we might expect to nd ancestral Chadic-speaking populations occupying parts of the Green Sahara, while closer relation to Cushitic might imply that ancestral Chadic speakers migrated into the Lake Chad Basin from East Africa. As an archaeologist, I will not try to resolve that debate. It should be noted, however, that mid-Holocene pottery from the Wadi Howar, an ancient riverbed which stretches from the Nile northeast toward Lake Chad and would have been a natural route of westward migration for East African populations, seems less similar to Gajiganna ceramics than does material from the Sahara (Wendt 2007, 84). Ehret (2006b; see also Tishko

et al. 2009, 1041) suggests on linguistic grounds that the overall distribution

of Chadic languages is in part the result of the widespread adoption by Nilo-Saharan speakers, previously occupying much of this area, of the languages of immigrant Chadic-speaking groups. These reconstructions parallel some recent genetic reconstructions in the region, which show genetic similarities between Chadic p. 73

and Nilo-Saharan speakers (Tishko

et al. 2009). Such widespread change in languages would

imply that

cultural interactions in the Lake Chad Basin between these populations were extremely intense. Reconstructed elements of vocabulary suggest that both of these groups were herders and farmers, possibly to di ering degrees. Primarily on archaeological data, Ehret hypothesized that these encounters between Chadic and Nilo-Saharan speakers took place south of Lake Chad, at around 8000–6000 BP. This date broadly ts with most glottochronological estimates (Barreteau and Dieu 2000; Starostin, in Blažek 2012; Tishko

et al. 2009: Supplementary Information, Figure S30j), but the location would seem to be unlikely,

especially for agropastoralists, given that the area south of the lake would probably have included extensive swamps. It is more likely that these initial encounters took place north of Lake Chad, with subsequent migration to the south (Figures 4.1 and 4.2).

Figure 4.1

Lake Chad Region 8000–6000BP

Figure 4.2

Lake Chad Region 4000–3000BP

Population Expansions in the Later Holocene Extensive archaeological evidence for human occupation and food production in the southern Lake Chad Basin dates only to after 4,000 years ago, with the Gajiganna sites southwest of Lake Chad (Breunig et al. 2001; Wendt 2007). We would expect on linguistic grounds that by this time both Chadic and Nilo-Saharan languages would have been long established in the region, but in fact the Gajiganna sites seem to represent the rst extensive colonization of the plains opened up by the shrinking Lake Mega-Chad (Figure 4.2). This situation—where archaeological and linguistic evidence for the tempo and mode of human occupation of a particular region disagree—is not uncommon. Gajiganna pottery looks most similar to ceramics from a wide area of the southern Sahara and neighboring regions, perhaps closest to material from northwestern Niger and northwestern Sudan (Wendt 2007, 86– 87). The colonization of heavy rki clay soils southeast of the lake began approximately 1,000 years later, around 3000 BP, again by populations whose ceramics show Saharan a

nities (Gronenborn et al. 1996)—

although rather di erent than those in the Gajiganna area, and somewhat more similar to earlier sites in northern Chad. The complex patterns of ceramic similarity and di erence over this large part of the continent in the mid-Holocene implies signi cant long-distance contact, interchange and social maneuvering. However, this very complexity in social interactions makes it more or less impossible to identify these initial colonizing populations with particular sociolinguistic groups for the reasons noted above: in a complex social space, people use material culture in social strategies, rather than simply letting p. 75

it represent them.

p. 74 p. 76

What of the economies of these groups moving into the southern Lake Chad Basin? The linguistic evidence has been taken to suggest that they were herders and farmers of domesticated cereals (Ehret 2006b), but here we have a problem: the Gajiganna sites have been intensively studied by archaeobotanists (more so than most archaeological traditions in West Africa), and there is no evidence of the use of domesticated cereals (Pennisetum millet, in this case) before about 3,200 years ago (Ballouche and Neumann 1996; Klee et al. 2004; Neumann 2003). Before that, these communities seem to have pursued what in more recent times would be an unusual economy, combining herding with the gathering of the wild grasses that are extraordinarily abundant in the area. It is not impossible that a more conventional foraging/farming adaptation existed earlier north of Lake Mega-Chad, that Gajiganna ceased using domesticated cereals

when they migrated into the wetter lands south of the lake, and then adopted them again after 3200 BP as the area dried out. However, at some point Occam’s Razor needs to be invoked: there does not appear to be a dramatic change in environments in the 600-year period between the initial appearance of Gajiganna sites and the adoption of domesticated millet by Gajiganna people. It is hard to avoid the conclusion that archaeological and linguistic reconstructions for mid-Holocene culture history in the southern Lake Chad Basin lead to di erent results. If there were intense interactions between Chadic- and Nilo-Saharan-speaking communities in the region during the mid-Holocene, they probably took place north of Lake Chad, in areas that would have been initially more attractive for agropastoralists retreating to the south. Further, available evidence suggests that farming began in this area well after initial colonization, so that the inferred evidence for earlier agriculture might actually have involved the intensive collection of wild grains (cf. Ehret 2006a). In general, these data suggest that the colonization of the southern Lake Chad Basin after 4000 BP involved Chadic-speaking populations at least, some of whom were probably descendants of groups speaking Nilo-Saharan languages north of Lake Chad some thousands of years before that. It is quite possible that this diverse occupation of the southern Lake Chad Basin also involved some Saharan-speaking groups, and possibly peoples speaking other languages, but there is no positive evidence of this.

Transformations during the Iron Age Some of the challenges of reconciling archaeological and linguistic data are exempli ed by the events of the p. 77

period between 2,500 and 1,500 years

ago in the southern Lake Chad Basin. Archaeological evidence

indicates a real transformation in regional lifeways over this period. This transformation included a variety of di erent elements: the development of more intensive agricultural systems, and probably of specialized pastoralism; the accompanying spread of more productive crops like sorghum and cowpeas, which could support larger populations; and increased levels of sedentism and settlement size (Figure 4.3). At 2500– 2000 BP, communities like Zilum and Malankari were between 12 and 30 hectares in size (Breunig et al. 2006; Magnavita et al. 2006; Magnavita et al. 2010), and so comparable in extent to the famous large oppida sites being constructed at the same time in Western Europe. Large tell mounds, the remains of long-term occupation in culturally important locations using mud-brick architecture, begin to appear in the archaeological record at the same time period. In technological and economic terms, this period is marked by the appearance of iron technologies (MacEachern 1996), rouletted pottery (Livingstone Smith 2007), and horses (MacEachern et al. 2001)—all of which would involve fundamental changes in the lifeways of people living in the area.

Figure 4.3

Lake Chad Region 2500–1500BP Of these technologies, horses would have originally come from North Africa and/or the Nile valley, but we have as yet no evidence for their route of introduction into the southern Lake Chad Basin. The question of origins for African iron technology is a perpetually vexing issue, but it increasingly seems likely that at least the slag-pit smelting process in use across the continent is distinctively African (Craddock 2009; Killick 2004). The late Holocene spread of rouletting on pottery across Africa appears to originate in the TichittMéma-Niger Bend area and spread into the Lake Chad Basin from the west, as part of a much larger process through which this form of ceramic decoration would eventually be found between the Atlantic and East Africa (Livingstone Smith 2007). The initial appearance of roulettes in the Lake Chad Basin about 2,500 years ago appears through such diverse archaeological contexts, from the shores of the lake to the edges of the Mandara Mountains, that it probably involves the cross-cultural transmission of a new ceramic 2

technology, possibly associated with iron-working. The Iron Age populations of the southern Lake Chad Basin seem to have increased substantially through the early and mid- rst millennium AD , with increasing p. 78

evidence for large settlements, many of them certainly walled.

p. 79

These are major changes in lifeways for Lake Chad Basin populations, but their origins remain di

cult to

pin down. Evidence for external contacts is ambiguous: roulettes and horses entered the area at di erent times, and there is certainly evidence for the movement of raw materials within the region itself, but there is little evidence for sustained long-distance trade outside of the Lake Chad Basin before ca. AD 800 (MacEachern in press). More strikingly, all of these regional transformations take place without any evidence for major population movements or replacements from outside the southern Lake Chad Basin through this period—although there probably would have been a signi cant degree of migration within the region, as is attested historically. On present evidence, most communities in the region at this time would have spoken Chadic languages, as they did in earlier times and would continue to do through the late Holocene, probably at least until 800–500 years ago, when, as we will see, Kanuri, Shuwa Arab, and Fulbe communities began to become in uential in the region. Perhaps the only material di erentiation that we might correlate with a linguistic process is the boundary between archaeological occurrences of two di erent forms of pottery roulettes, made of ber and of carved wood respectively, south of Lake Chad between about 2,500 and 1,500 years ago. This boundary, with a subsequent northward movement of the carved wooden roulettes into the Lake Chad Basin from the Adamawa Plateau area (Livingstone Smith 2007, 205), may be correlated with contacts between Chadicand Adamawa-speaking populations in that area. This is probably part of a larger story concerning culture contacts across the grassland-forest boundary in Central Africa during the period 3000–2000 BP, one

manifested in the appearance of iron and elaborate terracottas in the Nok region of Nigeria and in the Lake Chad Basin, and ultimately of millet and sheep/goats in savanna-forest mosaic zones between southern Cameroon and the DRC (Kahlheber et al. 2014)—a story, however, where the prehistory of intervening areas of the lands further to the south of Lake Chad is almost completely unknown.

Developments through the Last Millennium: The Prehistory of Central Chadic The geographical and cultural dichotomy between plains and highland populations in the southern Lake Chad Basin is one of the central human facts of the region. West Chadic–speaking communities are predominant among the peoples of the Jos Plateau in Central Nigeria, while East Chadic groups make up a portion of the “Hadjerai” peoples of the Guéra highlands in Chad. Both of these populations historically p. 80

occupied somewhat similar situations,

as constellations of relatively small-scale and egalitarian

communities surrounded by much larger states. To some degree, this situation probably re ects the status of Chadic languages as having great time-depth in the area, along with the subsequent in uence of predatory slave-raiding states through the last millennium at least. The only one such case where we have any substantial archaeological (or genetic) data, however, involves settlement in and around the northern Mandara Mountains, on the Cameroon-Nigeria border, by Central Chadic speakers. It is to that case that we will now turn our attention. The linguistic relationships between Central Chadic languages indicate a period of initial di erentiation by the end of the rst millennium BC or early rst millennium AD (Barreteau and Dieu 2000; Tishko

et al.

3

2009: Supplementary Information, Figure S30j). Initial Bayesian analyses of pottery from around the Mandara Mountains from this period indicates structured di erences in decorative techniques within a wider Iron Age ceramic production system that might be related to this initial di erentiation of ethnolinguistic groups (O’Brien et al. 2016). This di erentiation was probably associated with the intensi cation of plains settlement at the foot of the Mandara Mountains from the late rst millennium BC through the rst millennium AD , which as noted above is very much visible in the archaeological record (Gronenborn 1998; MacEachern 2012c; Marliac et al. 2000). Today, however, many of these Central Chadic languages are spoken within the Mandara Mountains themselves (Barreteau and Dieu 2000), and so we must consider the circumstances in which the center of gravity of that group of languages turned to this region. Earlier archaeological reconstructions saw the northern Mandara Mountains as essentially uninhabited until about 500 years ago. Given the greater time-depth indicated for Chadic language di erentiation in the mountains, this made reconciliation of archaeological and linguistic data di

cult. However, it now seems

that at least some portions of the northwestern massif were occupied not much after AD 1000 (MacEachern 2012c; MacEachern et al. 2012), with the possibility of settlement a century or two before the turn of the millennia (David 1998), although not at modern population densities. Glottochronological estimations by Barreteau (2000, Fig. 4), more detailed than those in Tishko

et al. (2009: Supplementary Information,

Figure S30j), indicate a phase of di erentiation of modern Mandara Chadic languages at AD 900–1300, which if it involved progressive occupation of rst foothills and then the massif would t these p. 81

archaeological data quite well (Figure 4.4).

Figure 4.4

Lake Chad Region 1200BP onward p. 82

This period also coincides with two other sets of processes that may have impelled Central Chadic–speaking populations to begin movement into the mountains: (a) a progressive drying trend in the southern Lake Chad Basin in the late rst millennium AD , which may have contributed to the diminution of forest cover in the Mandara massif (Boutrais 1984, 26–28; Brunk and Gronenborn 2004; Street-Perrott et al. 2000); and (b) the appearance of long-distance trade goods, especially carnelian and amazonite beads originating in the Sahara or North Africa, in burials in the region after AD 800 (Connah 1981; MacEachern in press). In the historic period, at least, exotic trade goods were often paid for in this area by the sale of slaves, and by this time the ruler of the Kanem state northeast of Lake Chad was already known to Arabic chroniclers as a seller of slaves (Levtzion and Hopkins 1981). Although the expansion of enslavement in the region is usually associated with the displacement of Kanuri rule from Kanem to Bornu in the 14th/15th centuries AD , it is likely that the trade was signi cant some centuries before that, possibly with the populations of the small Chadic-speaking walled towns of the area acting as middlemen. In that context, movement into a highland zone that was becoming less heavily forested may have been an attractive alternative. The end point of this process would be the abandonment of many of the Iron Age plains sites by the midsecond millennium AD (Bourges et al. 1999; Connah 1981; Gronenborn et al. 1996; Lebeuf 1969), probably at least in part because of increasing con ict associated with warfare and slave raiding. We thus move from a situation at AD 500 where the plains appear to be quite heavily populated, presumably by people speaking ancestral Central Chadic languages, but the mountains are not inhabited, to almost the opposite situation a thousand years later—involving people speaking descendant languages of those spoken by their plains ancestors. There are some parallels between linguistic and ceramic distributions in this area, but this is again an area where material culture, very much including ceramics, was deployed in individual and group social strategies (MacEachern 2003 [1991]), thus blurring any such correspondences. The modern distribution of ber roulette subtypes in the area does not map on to language families very well at all (Langlois 2005), and the prehistoric distribution is correspondingly complex. The Central Chadic–speaking populations for which genetic data are available (Zulgo, Mada, Podokwo, Ouldemé, and Mandara/Wandala) form a relatively tight genetic cluster (Tishko

et al. 2009), although with

considerable variability in particular genetic systems (see for example Cerny et al. 2004). Given the abundant ethnohistorical evidence for a great diversity of origins within even small Mandara montagnard p. 83

populations, the

processes of selection of individuals for genetic analysis is almost certainly going to

introduce some degree of bias into any subsequent analysis. Thus, for example, the Ouldemé group, about 12,000 strong, includes communities claiming at least eight di erent points of origins within and beyond

the Mandara Mountains over the last two to four centuries (MacEachern 2003 [1991], 394). The di

culties

in understanding Mandara demography from a genetic point of view are illustrated by recent estimates (Cerezo et al. 2011) of ancient e ective population sizes for Chadic-speaking Mandara populations in the massif that are between six and a hundred times larger than modern total populations—a nonsensical result, one that can only be explained indirectly as a consequence of large-scale and continuous gene ow among populations all over the Lake Chad Basin. If genetic and linguistic relations among Mandara montagnard communities seem complex, the situation on the plains to the north and east of the Mandara massif is even more complicated. In that area, speakers of a number of Afro-Asiatic (both Chadic and Semitic [Shuwa/Baggara Arab]), Nilo-Saharan (Saharan [Kanuri/Kanembu] and Central Sudanic [Bulala]), and Niger-Congo (Adamawa) languages share a general genetic similarity, relatively close to that of Central Chadic–speaking Mandara groups (Tishko

et al.

2009). There are some striking cases of genetic resemblance associated with linguistic di erence. Ethnolinguistic groups like the Tupuri, for example, seem to be essentially Chadic populations that have adopted an Adamawa language, an observation in line with linguistic data (Seignobos and Tourneux 2001). Such genetic similarities among very disparate language groups are quite striking; indeed, there seem to be more such cases in the Lake Chad Basin than anywhere else in Africa according to the Tishko

et al. (2009)

data, underlining the status of the region as a crossroads in this part of Central Africa. It thus appears from both archaeological and linguistic data that a signi cant part of the cultural patterning of the southern Lake Chad Basin results from sociopolitical processes through the last thousand years, especially those involving state formation, the extension of trade routes, and slave raiding. We might also note that the upheavals and violence associated with Boko Haram in this region today can in part be traced to the same processes (MacEachern 2015).

Conclusions This survey takes essentially no account of the distribution of West and East Chadic languages, in large part p. 84

because of the lack of archaeological

data for regions where those languages are found. This is the case

even for Hausa, the Chadic language with the largest geographical extent and sociopolitical impact through the last 800 years at least. Blench (2010) associates the spread of Hausa with the development of sociopolitical centralization through northern Nigeria. Such processes are natural subjects for archaeological research, but the archaeology of Hausaland is really only beginning; the concentration of research e ort on large sites associated with historically known capitals does not allow archaeologists to understand the dynamics of historical process, and research is particularly bedeviled by a lack of absolute chronologies, sites frequently being dug with no or minimal radiocarbon dating. Certainly most informative on political dynamics has been analysis of the contents of the extraordinary elite burials at Durbi Takusheyi, between Daura and Katsina (Gronenborn et al. 2012), which vividly demonstrates the sophistication and intercontinental reach of Hausa elites during the 14th and 15th centuries AD However, this again does not inform us on the initial spread of the Hausa language. In the southern Lake Chad Basin, on the other hand, there has been a very signi cant advance in our understanding of the dynamics of human occupations and migrations over the last 30 years (although, as always, there are still plenty of holes to be lled in—after being excavated, of course). We now have far more knowledge of the environmental dynamics of the region through time, and the disjuncts between archaeological and linguistic data concerning the initial settlement of the Lake Chad Basin should not obscure the fact that those two disciplines, along with genetics, provide historical reconstructions that are generally compatible with one another. The main challenge in the region is our lack of understanding of the prehistory of the northern Lake Chad Basin—and unfortunately the political situation does not signal that

that will change at any point soon. When I wrote my PhD dissertation 25 years ago, it seemed that there were major disagreements between archaeological and linguistic evidence for the settling of the Mandara Mountains. Further research in both disciplines, again with the addition of genetic data, today seems to be making these historical reconstructions generally compatible, which is a heartening development. Priorities in this region will include extending research on the earliest occupations of the Mandara massif, in particular through geoarchaeological research. In the longer term, and again with improvements in security, research in Nigeria and Chad may allow us to gain a greater understanding of the dynamics of history for West and East Chadic populations.

p. 85

References Abu-Lughod, Lila. 1991. “Writing against Culture.” In Recapturing Anthropology: Working in the Present, edited by R. Fox, 137– 162. Santa Fe: School of American Research Press. Google Scholar Google Preview WorldCat COPAC Amselle, Jean-Loup. 1985. “Ethnies et espaces: Pour une anthropologique topologique.” In Au coeur de lʼethnie: ethnies, tribalisme et état en Afrique, edited by J.-L. Amselle and E. Mʼbokolo, 11–48. Paris: Editions la Decouverte. Google Scholar Google Preview WorldCat COPAC Ballouche, Aziz, and Katharina Neumann. 1996. “A New Contribution to the Holocene Vegetation History of the West African Sahel: Pollen from Oursi, Burkina Faso, and Charcoal from Three Sites in Northeast Nigeria.” Vegetation History and Archaeobotany 4 (1): 31–39. Google Scholar WorldCat Barreteau, Daniel. 1987. “Un essai de classification lexico-statistique des langues de la famille tchadique parlées au Cameroun.” In Langues et cultures dans le bassin du lac Tchad, edited by D. Barreteau, 43–78. Paris: Editions de lʼORSTOM. Google Scholar Google Preview WorldCat COPAC Barreteau, Daniel, and Michel Dieu. 2000. “Linguistique.” In Atlas de la province Extrême-Nord, Cameroun, edited by C. Seignobos and O. Iyebi-Mandjek, 64–70. Paris: Editions de lʼIRD/MINREST. Google Scholar Google Preview WorldCat COPAC Barreteau, Daniel, P. Gubry, and Alain Beauvilain. 1984. “Les langues.” In Le Nord du Cameroun: Des hommes, une région, edited by J. Boutrais et al, 159–180. Paris: Editions de lʼORSTOM. Google Scholar Google Preview WorldCat COPAC Blažek, Václav. 2012. “Afroasiatic Migrations: Linguistic Evidence.” Paper presented at the conference Rethinking Africaʼs Transcontinental Continuities in Pre- and Protohistory: An International Conference to Mark the Retirement of Wim van Binsbergen. African Studies Centre, Leiden. Blench, Roger. 1999. “The Westward Wanderings of Cushitic Pastoralists.” In Lʼhomme et lʼanimale dans le bassin du lac Tchad, edited by C. Baroin and J. Boutrais, 39–80. Paris: IRD. Google Scholar Google Preview WorldCat COPAC Blench, Roger. 2010. “The Linguistic Geography of Nigeria and its Implications for Prehistory.” In West African Archaeology: New Developments, New Perspectives, edited by P. Allsworth-Jones, 161–170. Oxford: Archaeopress. Google Scholar Google Preview WorldCat COPAC Bouchette, Frédéric, Mathieu Schuster, Jean-Francois Ghienne, Claude Roquin, Abderamane Moussa, Patrick Marsaleix, and Philippe Duringer. 2010. “Hydrodynamics in Holocene Lake Mega-Chad.” Quaternary Research 73 (2): 226–236. Google Scholar WorldCat Bourges, Claire, Scott MacEachern, and Maureen Reeves. 1999. “Excavations at Aissa Hardé, 1995 and 1996.” Nyame Akuma 51: 6–13. Google Scholar WorldCat Boutrais, Jean. 1984. “Les unités naturelles.” In Le Nord du Cameroun: Des hommes, une région, edited by J. Boutrais et al, 23–62. Paris: Editions de l/ORSTOM. Google Scholar Google Preview WorldCat COPAC Breunig, Peter. 1995. Der Einbaum von Dufuna (Nigeria).” Berichte des Sonderforschungsbereichs 268 (5): 17–26. Google Scholar WorldCat

Breunig, Peter, Barbara Eichhorn, Stefanie Kahlheber, Veerle Linseele, Carlos Magnavita, Katharina Neumann, Martin Posselt, et al. 2006. “G(l)anz ohne Eisen: Große Siedlungen aus der Mitte des ersten Jahrtausends BC im Tschadbecken von NordostNigeria.” In Grundlegungen: Beiträge zur europäischen und afrikanischen Archäologie für Manfred K. H. Eggert, edited by H.-P. Wotzka, 255–270. Tübingen: Francke. Google Scholar Google Preview WorldCat COPAC Breunig, Peter, Abubakar Garba, and Musa Hambolu. 2001. “From Ceramics to Culture: Studies in the Final Stone Age Gajiganna Complex of NE Nigeria.” Berichte des Sonderforschungsbereichs 268 (14): 45–53. Google Scholar WorldCat p. 86

Breunig, Peter, Katharina Neumann, and Wim van Neer. 1996. “New Research on the Holocene Settlement and Environment of the Chad Basin of Nigeria.” African Archaeological Review 13 (2):111–143. Google Scholar WorldCat Brunk, Karsten, and Detlef Gronenborn. 2004. “Floods, Droughts, and Migrations: The E ects of Late Holocene Lake Level Fluctuations and Climate Fluctuations on the Settlement and Political History in the Chad Basin.” In Living with the Lake: Perspectives on History, Culture and Economy of Lake Chad, edited by M. Krings and E. Platte, 101–132. Cologne: Rüdiger Köppe. Google Scholar Google Preview WorldCat COPAC Cerezo, María, Viktor Černý, Ángel Carracedo, and Antonio Salas. 2011. “New Insights into the Lake Chad Basin Population Structure Revealed by High-Throughput Genotyping of Mitochondrial DNA Coding SNPs.” PLoS–One 6 (4):e18682. Google Scholar WorldCat Černý, V., M. Hájek, R. Čmejla, J. Brůžek, and R. Brdička. 2004. “mtDNA Sequences of Chadic-Speaking Populations from Northern Cameroon Suggest Their A inities with Eastern Africa.” Annals of Human Biology 31 (5): 554–569. Google Scholar WorldCat Connah, Graham. 1981. Three Thousand Years in Africa: Man and His Environment in the Lake Chad Region of Nigeria. Cambridge: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Craddock, P. T. 2009. “New Paradigms for Old Iron: Thoughts on É. Zangato & A. F. C. Hollʼs ʻOn the Iron Front.ʼ” Journal of African Archaeology 8 (1): 29–36. Google Scholar WorldCat David, Nicholas. 1998. “The Ethnoarchaeology and Field Archaeology of Grinding at Sukur, Adamawa State, Nigeria.” African Archaeological Review 15 (1): 13–63. Google Scholar WorldCat Dietler, Michael, and Ingrid Herbich. 1989. “Tich Matek: The Technology of Luo Pottery Production and the Definition of Ceramic Style.” World Archaeology 21 (1): 148–164. Google Scholar WorldCat Dimmendaal, G. J. 2008. “Language Ecology and Linguistic Diversity on the African Continent.” Language and Linguistics Compass 2 (5): 840–858. Google Scholar WorldCat Drake, Nick, Roger M. Blench, Simon J. Armitage, Charlie S. Bristow, and Kevin H. White. 2011. “Ancient Watercourses and Biogeography of the Sahara Explain the Peopling of the Desert.” Proceedings of the National Academy of Sciences 108 (2): 458– 462. Google Scholar WorldCat Ehret, Christopher. 2006a. “Linguistic Stratigraphies and Holocene History in Northeastern Africa.” In Archaeology of Early Northeastern Africa, edited by K. Kroeper, M. Chlodnicki, and M. Kobusiewicz, 1019–1055. Poznan: Poznan Archaeological Museum.

Google Scholar

Google Preview

WorldCat

COPAC

Ehret, Christopher. 2006b. “The Nilo-Saharan Background of Chadic.” In West African Linguistics: Studies in Honor of Russell G. Schuh, edited by P. Newman and L. M. Hyman, 56–66. Columbus: Ohio University Press. Google Scholar Google Preview WorldCat COPAC Fleming, Harold. 1983. “Chadic External Relations.” In Studies in Chadic and Afroasiatic Linguistics, edited by E. Wol and H. Meyer-Bahlburg, 17–31. Hamburg: Buske. Google Scholar Google Preview WorldCat COPAC Galaty, John. 1993. “ʻThe Eye that Wants a Person, What Can It Not See?ʼ: Inclusion, Exclusion and Boundary Shi ers in Massai Identity.” In Being Maasai, edited by T. Spear and R. Waller, 174–194. London: James Currey. Google Scholar Google Preview WorldCat COPAC Gronenborn, Detlef. 1998. “Archaeological and Ethnohistorical Investigations along the Southern Fringes of Lake Chad, 1993– 1996.” African Archaeological Review 15 (4): 225–259. Google Scholar WorldCat Gronenborn, Detlef, Paul Adderley, James Ameje, Arun Banerjee, Thomas Fenn, Gerhard Liesegang, Claus-Peter Haase, et al. 2012. “Durbi Takusheyi: A High-Status Burial Site in the Western Central Bilad al-Sudan.” Azania: Archaeological Research in Africa 47 (3): 256–271. Google Scholar WorldCat Gronenborn, Detlef, B. Wiesmüller, T. Skorupinski, and Barbara Zach. 1996. “Settlement History in the Kala-Balge Region of Borno State, Nigeria.” Berichte des Sonderforschungsbereichs 268 (8): 201–213. Google Scholar WorldCat p. 87

Kahlheber, Stefanie, Manfred K. H. Eggert, Dirk Seidensticker, and Hans-Peter Wotzka. 2014. “Pearl Millet and Other Plant Remains from the Early Iron Age Site of Boso-Njafo (Inner Congo Basin, Democratic Republic of the Congo).” African Archaeological Review 31 (3): 479–512. Google Scholar WorldCat Killick, David. 2004. “Review Essay: What Do We Know about African Iron Working?” Journal of African Archaeology 2 (1): 97–112. Google Scholar WorldCat Klee, Marlies, Barbara Zach, and Hans-Peter Stika. 2004. “Four Thousand Years of Plant Exploitation in the Lake Chad Basin (Nigeria), Part III: Plant Impressions in Potsherds from the Final Stone Age Gajiganna Culture.” Vegetation History and Archaeobotany 13: 131–142. Google Scholar WorldCat Kropelin, S., D. Verschuren, A. M. Lézine, H. Eggermont, C. Cocquyt, P. Francus, J. P. Cazet, et al. 2008. “Climate-Driven Ecosystem Succession in the Sahara: The Past 6000 Years.” Science 320 (5877): 765–768. Google Scholar WorldCat Langlois, Olivier. 2005. “Distributions ancienne et actuelle des décors imprimés au Diamaré (Nord-Cameroun) et à ses marges.” Préhistoire Anthropologie Méditerranéennes 13: 109–126. Google Scholar WorldCat Lapouge, Georges Vacher de. 1899. LʼAryen: Son rôle social; Cours libre de science politique professé à lʼUniversité de Montpellier (1889–1890). Paris: A. Fontemoing. Google Scholar Google Preview WorldCat COPAC Lebeuf, Jean-Paul. 1969. Carte archéologique des abords du lac Tchad (Cameroun, Nigeria, Tchad). Paris: Editions du CNRS. Google Scholar Google Preview WorldCat COPAC

Levtzion, Nehemiah, and J. H. Hopkins. 1981. Corpus of Early Arabic Sources for West African History. Cambridge: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC Lewis, M. P. 2009. Ethnologue: Languages of the World, 16th ed. Dallas, Texas: SIL International. Google Scholar Google Preview WorldCat COPAC Livingstone Smith, Alexandre. 2007. “Histoire du décor à la roulette en Afrique subsaharienne.” Journal of African Archaeology 5 (2): 28. Google Scholar WorldCat MacEachern, Scott. 1996. “Iron Age Beginnings North of the Mandara Mountains, Cameroon and Nigeria.” In Aspects of African Archaeology: Proceedings of the Tenth Pan-African Congress, edited by G. Pwiti and R. Soper, 489–495. Harare: University of Zimbabwe Press. Google Scholar Google Preview WorldCat COPAC MacEachern, Scott. 2001. “Setting the Boundaries: Linguistics, Ethnicity, Colonialism, and Archaeology South of Lake Chad.” In Archaeology, Language, and History: Essays on Culture and Ethnicity, edited by J. Terrell, 79–101. Westport: Bergin & Garvey. Google Scholar Google Preview WorldCat COPAC MacEachern, Scott. 2003 [1991]. Du Kunde: Processes of Montagnard Ethnogenesis in the Northern Mandara Mountains of Cameroon. London: Mandaras. Available online at http://www.mandaras.info/electronicisbnpublication/maceachern_dukunde_phd.pdf. Google Scholar Google Preview WorldCat COPAC MacEachern, Scott. 2012a. “The Holocene History of the Southern Lake Chad Basin: Archaeological, Linguistic and Genetic Evidence.” African Archaeological Review 29: 1–19. Google Scholar WorldCat MacEachern, Scott. 2012b. “The Prehistory and Early History of the Northern Mandara Mountains and Surrounding Plains.” In Metals in Mandara Mountains Society and Culture, edited by N. David, 29–67. Trenton, NJ: Africa World Press. Google Scholar Google Preview WorldCat COPAC MacEachern, Scott. 2012c. “Rethinking the Mandara Political Landscape: Enslavement, Climate and an Entry into History in the Second Millennium a.d.” In Power and Landscape in Atlantic West Africa: Archaeological Perspectives, edited by A. Ogundiran and J. C. Monroe, 309–338. Cambridge: Cambridge University Press. Google Scholar Google Preview WorldCat COPAC p. 88

MacEachern, Scott. 2015. “Boko Haram, coupeurs de route and Slave-Raiding: Identities and Violence in a Central African Borderland.” Paper presented at the annual meetings of the Society for American Archaeology, San Francisco CA, April 2015. MacEachern, Scott. in press. “Settlement, Burial Practises and External Relations in the Lake Chad Basin, 1500 BC–AD 1500.” In Trans-Sahara: State Formation, Migration and Trade in the Central Sahara (1000 BC–AD 1500), edited by D. Mattingly et al. London: Society for Libyan Studies. Google Scholar Google Preview WorldCat COPAC MacEachern, Scott, Claire Bourges, and Maureen Reeves. 2001. “Early Horse Remains from Northern Cameroon.” Antiquity 75 (287): 62–67. Google Scholar WorldCat MacEachern, Scott, Jean Marie Datouang Djoussou, and Rébecca Janson. 2012. “Research at DGB-1 and DGB-2, Northern Cameroon, 2010–2011.” Nyame Akuma 78: 6–16. Google Scholar WorldCat Magnavita, Carlos, Peter Breunig, James Ameje, and Martin Posselt. 2006. “Zilum: A Mid-First Millennium BC Fortified Settlement

near Lake Chad.” Journal of African Archaeology 4 (1): 153–169. Google Scholar WorldCat Magnavita, Carlos, Peter Breunig, Daniel Ishaya, and Olusegun Adebayo. 2010. “Iron Age Beginnings at the Southwestern Margins of Lake Chad.” In Crossroads/Carrefour Sahel: Cultural Developments and Technological Innovations in First Millennium BC/AD West Africa, edited by S. Magnavita, L. Koté, P. Breunig and O. Idé, 27–58. Frankfurt: Journal of African Archaeology Monograph Series. Google Scholar Google Preview WorldCat COPAC Marliac, Alain, Olivier Langlois, and Michèle Delneuf. 2000. “Archéologie de la région Mandara-Diamaré.” In Atlas de la province Extrême-Nord, Cameroun, edited by C. Seignobos and O. Iyebi-Mandjek, 71–76. Paris: Editions de lʼIRD. Google Scholar Google Preview WorldCat COPAC Neumann, Katharina. 2003. “The Late Emergence of Agriculture in Sub-Saharan Africa: Archaeobotanical Evidence and Ecological Considerations.” In Food, Fuel and Fields: Progress in African Archaeobotany, edited by K. Neumann, A. Butler and S. Kahlheber, 71–92. Cologne: Heinrick-Barth-Institut. Google Scholar Google Preview WorldCat COPAC Newman, Paul. 2013. “The Chadic Language Family: Classification and Name Index.” Mega-Chad Research Network/Réseau Méga-Tchad. Available online at http://www.megatchad.net/publications/Newman-2013-Chadic-Classification-and-Index.pdf. WorldCat OʼBrien, John D., Kathryn Lin, and Scott MacEachern. 2016. “Mixture Model of Pottery Decorations from Lake Chad Basin Archaeological Sites Reveals Ancient Segregation Patterns.” Proceedings of the Royal Society of London B: Biological Sciences 283 (1827). doi: 10.1098/rspb.2015.2824. Google Scholar WorldCat Schuh, Russell. 2003. “Chadic Overview.” In Selected Comparative-Historical Afrasian Linguistic Studies in Memory of Igor M. Diakono , edited by L. Bender, G. Takacs and D. Appleyard, 55–60. Munich: Lincom Europa. Google Scholar Google Preview WorldCat COPAC Seignobos, Christian, and Henry Tourneux. 2001. “Contribution à lʼhistoire des Toupouri et de leur langue.” In Leçons dʼAfrique: Filiations, ruptures et reconstitution de langues, edited by R. Nicolai, 255–284. Louvain: Peeters. Google Scholar Google Preview WorldCat COPAC Street-Perrott, F. A., J. A. Holmes, M. P. Waller, M. J. Allen, N. G. H. Barber, P. A. Fothergill, D. D. Harkness, et al. 2000. “Drought and Dust Deposition in the West African Sahel: A 5500-Year Record from Kajemarum Oasis, Northeastern Nigeria.” The Holocene 10 (3): 293–302. Google Scholar WorldCat Tishko , S. A., F. A. Reed, F. R. Friedlaender, C. Ehret, A. Ranciaro, A. Froment, J. B. Hirbo, et al. 2009. “The Genetic Structure and History of Africans and African Americans.” Science 324: 1035–1044. Google Scholar WorldCat p. 89

p. 90

Wang, H., J. A. Holmes, F. A. Street-Perrott, M. P. Waller, and R. A. Perrott. 2008. “Holocene Environmental Change in the West African Sahel: Sedimentological and Mineral-Magnetic Analyses of Lake Sediments from Jikariya Lake, Northeastern Nigeria.” Journal of Quaternary Science 23 (5): 449–460. Google Scholar WorldCat Wendt, Karl Peter. 2007. Gajiganna: Analysis of Stratigraphies and Pottery of a Final Stone Age Culture of Northeast Nigeria. Journal of African Archaeology Monograph Series 1. Frankfurt: Africa Magna. Google Scholar Google Preview WorldCat COPAC

Notes

1. 2.

3.

This also makes unlikely on archaeological grounds the model of gene flow posited for the L3e5 mtDNA variant (Podgorná et al. 2013). My suspicion is that this involves a shi from pottery to iron-working as a central “technology to think with,” since roulettes involve a simplification and mechanization of pottery decoration, and one probably with implications for gender roles. Glottochronological estimations of this di erentiation at ca. 3700–3500 BP by Starostin, in Blažek (2012), are far too early to reconcile with any culture-historical reconstructions for the region.

Tracing Language Movement in Africa Ericka A. Albaugh (ed.), Kathryn M. de Luna (ed.) https://doi.org/10.1093/oso/9780190657543.001.0001 Published: 2018

Online ISBN: 9780190657574

Print ISBN: 9780190657543

CHAPTER

5 800 Languages and Counting: Lessons from Survey Research across a Linguistically Diverse Continent  Carolyn Logan https://doi.org/10.1093/oso/9780190657543.003.0005 Published: February 2018

Pages 91–116

Abstract Using data on more than 800 home languages identi ed during Afrobarometer Round 5 surveys in 35 countries, as well as information on multilingualism gathered in 20 countries in Round 4, this chapter explores linguistic diversity and multilingualism at the individual level, within communities, and across countries. Afrobarometer data o er a unique perspective on the distribution of languages and language capabilities from the viewpoint of the users of language rather than those who study it. The chapter also identi es some of the challenges encountered in collecting public opinion data in linguistically diverse environments. The ndings reveal that even in many rural zones many Africans are living within ethnically and linguistically diverse communities, and preliminary analysis suggests this may have important implications for social and political attitudes. The data have untapped potential for understanding language evolution and for studying language both as a product and as a variable driving attitudes and outcomes.

Keywords: Afrobarometer, public opinion, languages spoken, home language, linguistic diversity, multilingualism Subject: Sociolinguistics Collection: Oxford Scholarship Online

Introduction In its rst 16 years, Afrobarometer completed six rounds of surveys across 36 African countries. Reaching this point—conducting more than 225,000 individual interviews and 145 national surveys—presented countless political, logistical, nancial, and conceptual challenges. One of the most important challenges the network confronts on its way to producing high-quality data is linguistic. It is not news that Africa is a linguistically diverse continent, but Afrobarometer ndings certainly recon rm this perception: during Round 5 (2011/2013), 53,973 respondents across 35 countries identi ed more than 800 African languages as their “home language.” These include some languages that

were identi ed by more than 1,000 respondents (e.g., Akan, Chichewa, Kirundi, Sesotho, and SiSwati), but also many others identi ed by only a single respondent in our sample. Under these circumstances, ful lling Afrobarometer’s twin linguistic goals of conducting interviews “in the language of the respondent’s choice” while maintaining stringent quality standards has required both adaptability and rigor in the development and implementation of Afrobarometer survey protocols. p. 92

This analysis will have three main parts. First, it will review what Afrobarometer data reveal about language and multilingualism on the continent, at the individual level as well as within communities and across countries. It is well known that most African countries are linguistically heterogeneous, but how linguistically diverse are the local communities where people live, and how do individuals and communities adapt to this diversity? Second, we will discuss how Afrobarometer “manages language,” considering challenges ranging from translating complex concepts into dozens of languages to the logistical di

culties

of linguistically matching respondents and interviewers. Finally, we will brie y consider some of the potential social and political implications of community-level linguistic homogeneity versus linguistic diversity, suggesting areas for further research and analysis.

Language and Survey Research I should begin by admitting that neither I nor my Afrobarometer colleagues are linguists or language specialists. We are primarily political scientists and public opinion specialists who have long treated language as more of a means than an end. Afrobarometer’s primary goal is to gather public opinion data on our signature topics of democracy and governance. And you cannot capture good-quality survey data on much of anything in Africa without a wholesale grappling with language, a topic that I will return to later in the chapter. But while we gathered a wealth of data on language in Africa over our rst 16 years, we have done little to examine language more closely either in its own right or as a means for explaining other phenomena that we study. The editors of this volume convinced me to take a closer look at our language data and what they can tell us about the many phenomena we study as well as about how we study them. This chapter re ects an initial e ort to troll through our language data. As such, it just begins to scratch the surface of what this rich trove of survey ndings has to o er. I will limit myself here primarily to an initial descriptive look through these data. But I will close with a discussion of several possible directions for further analysis. Although much of the focus of this book is on language movement, change, and transition, Afrobarometer data o er more of a snapshot of language capacity and distribution at a given point in time. But even in this snapshot, we may observe change, for example by considering di erences in how youth identify their p. 93

languages and language capabilities compared

to their elders, or studying the e ects of education and

urbanization. And, uniquely, Afrobarometer data o er a perspective on the distribution of languages and language capabilities from the viewpoint of the users of language rather than those who study it. Afrobarometer respondents self-identify their “home languages” according to how they themselves understand and name the languages they speak. But even in this there may be much for a professional linguist to glean. Do individuals in a particular locale identify themselves, for example, with local dialects or larger language groups? Does this identi cation vary with social status or age? Do the names and language patterns revealed by respondents match the patterns that experts report? The same applies to our data on multilingualism. Respondents also self-identify their capacity to speak other languages “well.” This is clearly an imperfect measure of language capacity—not least because each respondent sets his or her own standard for what “well” means. Olson and Lewis (this volume), in contrast, describe some of the challenges inherent in more ne-tuned measurement of language pro ciency for the

purposes of the Ethnologue. But there is rich potential for deepening our understanding of multilingualism and second-language capacity and change, especially when multiple data sources like Afrobarometer and Ethnologue can be analyzed alongside one another. Once we have explored and elaborated these patterns, we can move on to the next layers of analysis, exploring what might explain the patterns we see—including many of the questions of power, policy, trade, migration, and related issues that are discussed in this volume—as well as how these patterns can help us understand phenomena such as trust and social cohesion, national identity, and commitment to the state and its institutions. Many of these questions are beyond the scope of the present analysis, but by providing an initial description of these rich data, I hope to open the door for further exploration and analysis of the role language plays in shaping society, politics, and governance across the African continent.

An Overview of the Data Afrobarometer is a pan-African, nonpartisan research network that conducts public attitude surveys on democracy, governance, economic conditions, and related issues across more than 30 countries in Africa. Six rounds of surveys were completed between 1999 and 2015, and a seventh was launched in late 2016. p. 94

Afrobarometer conducts face-to-face interviews

in the language of the respondent’s choice with

nationally representative samples of between 1,200 and 2,400 respondents. Samples of this size yield results with a margin of error of +/−2% (for a sample of 2,400) or +/−3% (for a sample of 1,200) at a 95% con dence level. The ndings reported here are drawn primarily from two Afrobarometer data sets. Speci cally: • In every round, Afrobarometer has asked respondents to identify their “home language, . . . that is, the language of your group of origin.” The ndings presented here regarding home language draw from Afrobarometer Round 5 surveys, conducted with 53,973 respondents in 35 countries between 2011 and 2013. • In addition, in Round 4 (2008/2009) only, Afrobarometer asked individuals, “What languages do you speak well?” This question was asked of 27,713 respondents across the 20 countries covered in Round 4. • Every interview also captures the “language of interview.” We use ndings on “language of interview” from Round 5 to test success in matching interviewers and respondents and from Round 4 to check and correct each respondent’s list of languages spoken.

Measuring Linguistic Diversity The question about the respondent’s “home language” is one of the rst asked during an Afrobarometer 1

2

interview. The question is open-ended, and interviewers capture the individual’s exact response. In p. 95

almost 54,000 interviews across 35 countries in 2011/2013, more than 800 languages

were identi ed in

3

response to this question. With the addition of ve North African countries (Algeria, Egypt, Morocco, Sudan, and Tunisia) to Afrobarometer in Round 5, Arabic became the most widely spoken home language (5,748 respondents, slightly more than 10% of all interviews), followed by Chichewa/Chinyanja (2,055 respondents), Akan (1,279), SiSwati (1,233) and Cape Verde’s Crioulo (1,206). However, at the other end of the spectrum, some 1,583 respondents (3% of the sample) named 532 languages that were identi ed by 10 respondents or fewer as their home language. In Cameroon alone, respondents identi ed 167 home languages.

National-Level Diversity Several African countries are widely recognized as being linguistically relatively homogenous, especially in North Africa, where Arabic is nearly universal in Egypt, Sudan, and Tunisia (while Algeria and Morocco are home to sizeable minorities of Amazigh speakers). This is also the case in a handful of small countries such 4

as Cape Verde (where 100% identi ed Crioulo as their home language), Burundi (100% Kirundi), Lesotho (98% Sesotho), Mauritius (96% Creole), and Swaziland (98% SiSwati) (Figure 5.1). Across the rest of the continent, however, linguistic diversity reigns, with respondents reporting anywhere from a mere seven home languages in Niger and 12 in South Africa to 100 or more in Tanzania (100 languages), Nigeria (134), and Cameroon (167). The lone large-country exception is Madagascar, where 100% report speaking either “o

cial Malagasy” or a “Malagasy dialect” as their home language.

Figure 5.1

Number of “Home Languages” Identified (35 countries in 2011/2013 [Round 5]) Respondents were asked: “Which [Kenyan] language is your home language?” [Interviewers would prompt, only if necessary, “That is, the language of your group of origin.”] Despite the increasing foothold of various lingua francas across the continent, including European o

cial

languages as well as African languages of business and commerce (especially Kiswahili, various forms of Pidgin, and Hausa), the resilience of African languages is evident in these ndings. Although English, French, and Portuguese are increasingly the main languages spoken in the home in some urban areas on the continent, especially where parents are from di erent ethno-linguistic backgrounds, just 2% of respondents identi ed one of these three languages as their home language.

Some of the implications of Africa’s linguistic diversity are also revealed when we consider the proportion of p. 96

respondents in each country who

named the most frequently identi ed home language for the country as

their own home language. The gure ranges from a low of 14% who speak Kisukuma in Tanzania (where just 9% identi ed Kiswahili as their home language) to 98% or more in the seven most linguistically 5

homogenous countries (Egypt, Tunisia, Cape Verde, Burundi, Sudan, Swaziland, and Lesotho) (Figure 5.2). p. 97

Keep in mind, however, that this does not mean

that Tanzanians, Ivorians, or Cameroonians cannot

communicate with one another, but only that most of the time they must do so in second or third languages rather than their home language. The incentives to learn a national language or a local lingua franca are consequently higher. This helps explain why the vast majority of Round 5 surveys in Côte d’Ivoire, for example, were done in French, which is widely spoken there, in contrast to Niger, another francophone country, where less than 10% of interviews were done in French, while nearly two-thirds were done in the p. 98

widely shared Hausa language.

Figure 5.2

Proportion Naming most Frequently Identified Home Language (35 countries in 2011/2013 [Round 5]) Respondents were asked: “Which [Kenyan] language is your home language?” (% naming the countryʼs most frequently identified home language)

Individual-Level Linguistic Diversity We now turn to the data collected during Afrobarometer Round 4 regarding the languages each respondent reported speaking well. This question, too, was open-ended, and respondents were given the opportunity to list as many languages as they wanted, all of which were recorded by interviewers. As noted, the results are thus based entirely on respondents’ self-assessments of their language abilities, which, aside from the 6

language in which the interview was conducted, were not tested by the interviewer.

The results con rm the widely held notion that Africans are quite multilingual. More than two-thirds (70%) of respondents across these 20 countries reported speaking at least two languages, and more than onequarter (29%) reported speaking three or more languages (Figure 5.3).

Figure 5.3

Average Number of Languages Spoken by Respondents (20 countries in 2008/2009 [Round 4]) Respondents were asked: “What languages do you speak well?” Note: The question was open-ended. All languages identified were recorded, and a count of the number of languages was post-coded for each respondent. The mean number of languages spoken across all respondents was 2.1. Zambians (2.8 languages), Kenyans (2.7), and Namibians (2.6) tend to be far more multilingual than Basotho (1.6) and Malagasy (1.4) (Figure 5.4).

Figure 5.4

Average Number of African and European Languages Spoken per Individual (20 countries in 2008/2009 [Round 4]) On average, men speak more languages than women: 74% of men reported two or more languages spoken, compared to 65% of women. There is a similar gap between urban residents (77% speak more than one language) and rural dwellers (65%). The young are also much more multilingual than their elders. Threequarters (76%) of those age 25 and under speak at least two languages, compared to just half (52%) of those over age 65. Similarly large di erences are linked to education. Half (50%) of those with no formal schooling reported speaking at least a second language, compared to 83% among those who have attended or completed secondary school and 95% among those with any post-secondary education. Preliminary analysis suggests another important factor contributing to multilingualism: the size of each individual’s ethnic group. When we scale all sample sizes to 1,200 and then compare the number of languages each respondent reported speaking to the number of respondents in the sample who reported p. 99

speaking the same home language as

p. 100 7

the respondent, it is evident that, as Albaugh (2016) posits, individuals from smaller ethno-linguistic groups are considerably more likely to speak multiple languages (Figure 5.5).

Figure 5.5

Average Number of Languages Spoken by Size of Home Language Group (20 countries in 2008/2009 [Round 4]) It is worth noting just how much of this “extra” linguistic capacity—i.e. the number of languages each individual speaks beyond his or her mother tongue or primary language—corresponds to non-African (European) o

cial languages. English, French, and Portuguese are often the language of educational 8

instruction, especially in secondary schools. This “school French” or “school English” appears to result in a quite high proportion (48%) of respondents who claimed to have achieved competence in at least one European language. This includes fully 89% of Liberians who either claimed to speak English well or were coded as speaking it based on the language of their interview, along with 76% of Namibians; 83% of p. 101

9

Mozambicans claimed to speak Portuguese well. Respondents in most

francophone countries appear to

be somewhat more cautious in claiming pro ciency in French; the greatest proportion was in Benin, where 10

49% said they speak French well (Figure 5.6).

Figure 5.6

Proportion Speaking Main European Languages (English, French, Portuguese) (20 countries in 2008/2009 [Round 4])

Tanzania reveals a unique linguistic pro le. The rst independence president, Julius Nyerere, made Kiswahili, rather than English, the national language to be used in educational instruction and government. Mandating Kiswahili was a key part of his Ujamaa platform aimed at promoting Tanzanian national identity over tribal a

liations. Pro ciency in Kiswahili is essentially universal: 99% reported speaking it well.

Kiswahili’s dominance in Tanzania is especially evident in the nding that only 10% of respondents—the lowest level across 20 countries—reported competence in a non-African language. No other country has achieved this level of success in instilling widespread pro ciency in a national language other than the mother tongue, although Kenya comes close: Just 1% identi ed Kiswahili as a home language, while 92% claimed to be pro cient. But in sharp contrast to Tanzania, 60% also claimed pro ciency in English, the p. 102

main language of education in Kenya. Despite the widespread use of

Kiswahili in education, commerce,

and government in Tanzania, just 9% identi ed it as their home language; most identi ed one of 99 other languages as their mother tongue. If we exclude European “national languages,” then the mean number of (African) languages spoken per respondent is 1.6. Zambia still tops the list, with 2.4 African languages spoken per person, and Tanzania and Kenya—where many speak a home language as well as Kiswahili—reported 2.1 and 2.0 languages per person, respectively.

Community-Level Linguistic Diversity Africa is understood to be linguistically diverse in most cases at the country level, and it is commonly assumed that African cities are melting pots of languages and cultures. But assumptions about what nonurban communities look like tend to be much di erent. Rural Africa is typically seen as more homogenous terrain, with villages inhabited by people with a shared ethnic or tribal identity and language. These individuals may be exposed to people from other ethno-linguistic communities when they travel to shared markets or urban centers, but we often assume that close to home they live largely among people who, at least in a linguistic and cultural/ethnic/tribal sense, are much more like themselves. Returning to the Round 5 data (2011/2013 in 35 countries) on respondents’ home languages discussed above (see Figures 5.1 and 5.2), we can test these assumptions and see just how homogenous—or not—Africa’s rural communities are. As noted, Afrobarometer sample sizes range from 1,200 to 2,400 respondents per country. Samples are distributed across a country’s administrative units (provinces, districts, regions, etc.) and urban-rural areas in proportion to the share of each in the total population of the country based on the most recent census gures. Thus, if according to the census 12% of a country’s population lives in the urban portion of District A, then 12% of Afrobarometer’s sample in the country will be allocated within that same stratum. All individual respondents in a sample are randomly selected from “primary sampling units” (PSUs), usually census enumeration areas (EAs), that are randomly selected from lists of all PSUs in each stratum. EAs are small units, typically assumed to encompass roughly 100 households, although in reality they can vary considerably both within and across countries, from as few as 20 to as many as several hundred households. But the key points are that they are one of the smallest standard units of measure of population and area in a country, and they represent a very small slice of a country’s population. p. 103

An ideal sample would be completely random. That is, Afrobarometer would randomly select 1,200 or 2,400 EAs (depending on sample size) from around the entire country and then randomly select one individual from within each EA. However, from the perspective of logistics, time, management, and cost, it is not feasible to send interviewers to 1,200 (or 2,400) EAs. Afrobarometer therefore clusters samples, conducting eight interviews in each of 150 EAs (or 300, for a sample size of 2,400). Eight respondents are randomly selected from the entire adult population of the EA using randomly selected start points, walk patterns, and

11

individual selection protocols within selected households.

Although clustering makes for a less-perfect

sample in terms of national representativeness, it o ers other advantages, because it allows us, to a limited extent, to look at local community pro les. While a sample of eight individuals from a given EA is not representative, in a statistical sense, of everything about the EA (the margin of error on such a tiny sample is extremely high), we can nonetheless develop some broad conclusions about the characteristics of an EA from the collective pro le of these eight respondents. The ndings regarding the linguistic diversity among these eight respondents in each EA across much of the continent—including in rural areas—are revealing. We counted the number of home languages spoken by 12

the eight respondents within each EA across 22 of the 35 Round 5 countries.

We expect to nd linguistic

diversity even at the EA level in urban areas. But if rural inhabitants largely live in ethnically or tribally based villages and communities, then at the ne-grained EA level we would expect that all respondents from within an EA might be members of the same tribe or ethnic group and share a common home language. But this is frequently not the case. Judging by respondents’ self-reported home languages, there is, in fact, a remarkable degree of heterogeneity even in the rural zones of many countries. In Senegal, Ghana, and even relatively homogenous Botswana, only about half of rural respondents live in what appear to be ethnically p. 104

homogenous local communities where

all eight respondents from the EA reported speaking the same

home language, and in Sierra Leone, Togo, and Zambia, only about one in four do (Figure 5.7). In Cameroon, there appear to be virtually no ethnically homogenous communities. In contrast, rural communities in Niger, Namibia, Guinea, Kenya, and Benin tend to be more segregated linguistically—just 24% of rural Nigeriens are from linguistically mixed communities, along with 29% of Namibians and a surprisingly low p. 105

39% even in extremely diverse Kenya.

Figure 5.7

Linguistic Diversity in Rural Communities (22 countries in 2011/2013 [Round 5]) Respondents were asked: “Which [Kenyan] language is your home language?” [Interviewers would prompt, only if necessary, “That is, the language of your group of origin.”]. Note: We tallied the number of languages identified by the eight respondents in each rural EA. The figure shows the proportion of rural EAs in which the eight respondents combined reported one, two, or three or more home languages. But overall, diversity is the norm. In 11 of 22 countries, at least one-fourth of rural EAs recorded three or more home languages among the eight respondents, and in 13 of the 22, fewer than half of rural EAs were monolingual. Unsurprisingly, urban areas have a still more diverse pro le. In Kenya, for example, 57% of urban EAs recorded three languages or more (Figure 5.8). The mean number of home languages identi ed per urban EA was 3.0, compared to 1.5 in rural areas. But it is worth noting that in some countries, urban areas are also relatively segregated, most notably in Malawi, where more than half (56%) of urban EAs reported only one home language, and, more surprisingly, in Nigeria, where one in three urban EAs (36%) appear to be relatively homogenous. These data o er insight into urban settlement patterns at the micro-community p. 106

(EA) level and the tendency of

some urban communities to self-segregate, patterns that may have

important implications for social cohesion and national identity.

Figure 5.8

Linguistic Diversity in Urban Communities (22 countries in 2011/2013 [Round 5])

Managing Surveys amid Linguistic Diversity This incredible linguistic diversity has some obvious implications for managing a survey research project. Afrobarometer implements a relatively long questionnaire that includes some conceptually complex questions. Maintaining comprehensibility and comparability even as questions are translated into dozens of languages is essential to producing high-quality results. But producing high-quality translations is, as it turns out, only half the battle. Being able to put them to good use—i.e. getting the right translation in front of the right respondent in the hands of an interviewer who is quali ed to use it—proves to be equally challenging. In this section, I will brie y review some of Afrobarometer’s core protocols aimed at achieving these goals and discuss challenges posed by the degree of linguistic diversity reported here.

Selecting Interview Languages Producing reliable survey data requires that survey samples be representative and that the surveys accurately capture the views of respondents. Two key linguistic conditions required to meet these goals are, rst, that we minimize the number of people who are excluded from the sample because they do not speak a survey language, and second, that we ensure that respondents are interviewed in languages in which they are pro cient enough to understand the meaning of the questions and core concepts. Shortcuts, such as limiting survey languages to a common lingua franca or a couple of the largest languages spoken, will often leave us short of these goals. For example, implementing a survey in Kenya using only a Kiswahili translation would be problematic, because the roughly 8% of respondents who do not identify Kiswahili as a language they speak well (based on Round 4 data) would be excluded, potentially introducing bias into the ndings. And in a majority of countries, there is no single lingua franca that is spoken even this widely.

It has also been a core Afrobarometer principle that respondents be interviewed, to the extent possible, in “the language of the respondent’s choice.” Yet it is clearly not feasible to interview all respondents in their p. 107

mother tongue. Afrobarometer’s protocol therefore relies on a “5% rule,” i.e. in addition to any

key

lingua franca, “every language group that is likely to constitute at least 5% of the sample should have a translated questionnaire” (Afrobarometer 2014, 19). Using this rule, it is often possible to capture the home languages of a signi cant proportion of the population with four or ve translations. For example, in Benin, ve translations into Adja, Bariba, Dendi, Fon, and Yoruba covered the home languages of 87% of respondents in Round 5. An additional 6% of respondents spoke Otamari, so a translation into this language would have been justi ed. But no other group constituted more than 3% of the sample. Realistically, the only ways that smaller linguistic groups can be captured is by using national-language questionnaires (Arabic, French, English, Portuguese, or Kiswahili), using a lingua franca such as Pidgin in Cameroon, and taking advantage of the fact that individuals from smaller language groups are likely to speak one of the major local languages. However, in some countries, the 5% rule generates very low levels of home-language coverage. In Cameroon, for example, o

cial translations into Fulfulde and Ewondo cover only 18% of the population.

French (16%) and English (1%) questionnaires cover the home languages of another 17%. But for the remaining 65%, there is no language group that constitutes more than 3% of the sample, so we must rely on the multilingualism of members of the many smaller language groups. Kenya, South Africa, and Uganda present a contrasting challenge. In these countries, following the 5% rule covers a large share of the population but results in as many as nine translations. Working with too many translations can be both costly and unwieldy; Afrobarometer prefers a maximum of six. We therefore sometimes drop some of the smaller languages, even if they meet the 5% threshold. In short, Afrobarometer’s 5% rule—with occasional adjustments—is imperfect, but may remain the best option for balancing competing demands. Ultimately, for Round 5, Afrobarometer produced more than 80 translations of the questionnaire across 35 countries. By at least one key metric, the process has worked reasonably well: we have relatively few failed contacts with respondents based on available interview languages. Across all 35 countries included in Round 5, 53,973 face-to-face interviews were completed (“successful calls”) and 16,866 “unsuccessful calls” were reported where a selected household or respondent could not complete an interview. Of these, 581—less than 1% of all calls made—failed due to the selected respondent either being deaf or not speaking a survey 13

language.

p. 108

Producing Translations To ensure comparability of responses within and across countries, it is essential that, to the extent possible, all respondents be asked the same question in the same way, even when the language changes. One of the greatest challenges for Afrobarometer is making sure that all language, especially core concepts and words such as “democracy,” “accountability,” “freeness and fairness,” or even “trust,” is translated correctly, consistently, and e ectively. Ensuring the conceptual equivalence of local-language translations is one of the most critical steps in the entire survey process—a potential “weakest link” if it is not done well. In the end, Afrobarometer data are only as good—and as comparable both within and across countries—as the quality and consistency of the local-language translations used to collect them. Afrobarometer has developed rigorous translation protocols to ensure the success of this process. They include not just forward and back blind translations done by translators in isolation, but also synchronization meetings. These meetings o er an opportunity for translators and the national investigator to discuss each question to come to a common understanding of the intent, so that there is as

14

much consistency as possible in how key concepts are translated across languages.

During this process,

Afrobarometer prefers to rely as much as possible on experienced translators who are familiar with daily usage rather than linguists or academics who may produce more formal, but less accessible, translations that may be di

cult for day-to-day users of the language to understand.

One of the most common questions asked of Afrobarometer is how we handle the “d-word,” i.e. “democracy.” In each round, several questions on the Afrobarometer questionnaire use the word “democracy.” But, as many a challenger has pointed out, even in a single language the word “democracy” can mean di erent things to di erent people. So how do we know what our respondents are thinking of when we ask them whether they support democracy or think they are getting it in their country? The problem is compounded by the even more diverse meanings associated with the word “democracy” as it is translated into dozens of local languages. 15

Afrobarometer deals with this challenge in several ways.

First, we triangulate. While a number of

questions ask speci cally about “democracy,” many more ask about various processes and institutions of p. 109

democracy without using the d-word itself. These questions o er insight into

how individuals

understand democracy without using the word. In some rounds, we also ask respondents directly, “What, if anything, does the word ‘democracy’ mean to you?” “Democracy” is also the one word in the questionnaire that we do not translate into a local language unless it is essential. The survey manual describes the protocol for asking d-word questions by saying, “Always read the question in the language of the interview, but always read ‘democracy’ in the national language (i.e. English, French, Portuguese, or Kiswahili). Translate ‘democracy’ into the local language only if the respondent does not understand the national-language term.” This is, again, an imperfect solution to a challenging problem, but by using this approach, and recording whether the term was used in the national language or the local language, we aim to reduce the impact of its variable de nitions, while also collecting enough data to analyze those di erences when necessary. Finally, in Round 5 surveys, Afrobarometer also experimented with “anchoring vignettes” as a systematic solution. To elucidate individual understandings of democracy, we asked respondents to rate the level of democracy in several hypothetical political systems that were described to them. The results can be used to rescale responses from questions that speci cally mention “democracy” (Bratton, 2010, 109–112).

The Questioners: Selecting and Deploying Multilingual Fieldworkers The nal challenge arises when it comes time to put these carefully crafted translations to use. There are many important and complex questions related to how best to match interviewers and respondents in terms of ethnicity, race, gender, religion, and other factors. As recent work by Adida et al. (2016) has shown, racial and ethnic matching, or mismatching, between interviewers and respondents can a ect how respondents answer some types of sensitive questions. But Afrobarometer’s primary goal is to select and deploy eldworkers so as to best manage “language of interview,” ideally by matching interviewers and respondents as closely as possible in terms of language capacity. It is an unspoken ideal that wherever possible, respondents should be interviewed in their home language by another native speaker of that same language, although it is clear that in most countries this ideal cannot always be met. The evidence described above indicating that the linguistic heterogeneity we see at the national level often extends even into rural communities suggests that the challenge is even more complex than previously recognized. p. 110

Afrobarometer protocol calls for teams of four interviewers and a supervisor to work and travel together throughout eldwork. Teams are constructed and assigned routes and EAs with the goal of placing eldworkers so that their linguistic skills best match the pro le of the region they will cover. Each eld

team must realistically cover a signi cant number of EAs, as elding too many interviewers and teams can pose training and supervision challenges and prove detrimental to the overall quality of data. But it quickly becomes clear just how imperfect this approach may be in terms of our ability to linguistically match eldworkers and respondents. Fieldwork teams and their linguistic skills are essentially static, but the linguistic pro les of selected communities and respondents can vary widely. However, it is not feasible to ne-tune the composition and skills of interview teams to closely match the pro le of each separate community they will visit. In short, a eldwork team deployed to a region where one or two languages are dominant will be selected to ensure that the team members are native speakers of, or at least highly competent in, those languages. But once in the eld, they may frequently encounter communities, and individual respondents, with di erent linguistic pro les. One way to assess how well Afrobarometer is doing in terms of linguistically matching respondents and interviewers is to look at the language of interview and the actual utilization of the local-language translations that are produced. We can do this by looking at the proportion of individuals who speak a survey language as their home language who are actually interviewed in that language. In the Burkina Faso Round 5 survey, for example, 619 respondents (52% of the sample) identi ed Moore as their home language. Of those, 444 were actually interviewed in Moore, a 72% match rate. Others Moore speakers were interviewed in French (132 respondents) or Dioula (44 respondents). In total, 69% of Burkinabe identi ed a home language that matched an interview language (Moore, Dioula, Fulfulde or French), and 48% used these same languages in their interview, a 70% match rate for the country as a whole (Figure 5.9).

Figure 5.9

Match Rate for Home-language Interviews (34 countries in 2011/2013 [Round 5]) Note: Total % shown is the % speaking an interview language as their home language. The dark portion of the bar represents the share that were actually interviewed in their home language; the light-gray portion of the bar represents the share who speak an interview language as their home language but were interviewed in a di erent language; and % not shown represents those who did not have the option of being interviewed in their home language. Not surprisingly, we obtained very high match rates in the most linguistically homogenous countries, such as Sudan, Egypt, Cape Verde, Burundi, Lesotho, and Mauritius. Swaziland is a partial exception, because a considerable number of native SiSwati speakers chose to be interviewed in English. Among more linguistically diverse countries, however, the results were much more mixed. On the one hand we have Malawi, where home-language questionnaires were available for 80% of respondents and used by 78%, a match rate of 98%. At the other end of the spectrum we nd Kenya, where 67% could have done an interview in their home language, but only 16% actually did, a match rate of just 24%. Instead, 59% of p. 111

interviews in

Kenya were conducted in the main lingua franca, Kiswahili, and another 24% were

conducted in English. These apparent mismatches could arise from several factors, including: • Interviewer-respondent mismatch—A respondent’s home-language questionnaire may not be usable, p. 112

either due to poor positioning of

eldworkers (i.e. not sending teams with the right mix of language

skills) or because of challenges imposed by community-level linguistic diversity (i.e. respondents from many language-groups in a small area, some of which may not be recognized as common in the region). • Inter-ethnic relationships—These may a ect the choice of interview language when interviewer and respondent are from di erent ethno-linguistic groups. • Respondent choice—Afrobarometer protocol requires interviewing in “the language of the respondent’s choice,” and some respondents may prefer to be interviewed in a language other than their mother tongue for status, ease of communication, or other reasons. This might, for example, explain the signi cant number of native SiSwati speakers in Swaziland who chose to be interviewed in English rather than SiSwati. • Interviewers may prefer working in some languages rather than others and may therefore, consciously or unconsciously, steer respondents toward those languages. In short, matching failures could re ect weaknesses in survey management in terms of selection and deployment of eldworkers with appropriate language skills. But it is possible that even with careful attention to eldworker recruitment and deployment, the linguistic pro les of some countries may make achieving higher levels of home-language matching prohibitively di

cult and expensive. This could

explain at least part of the low match rate in Kenya, for example, where considerable internal migration has occurred, especially into the most agriculturally productive regions, resulting in diverse local-level language pro les that are di

cult to match.

This suggests that each country where match rates are low needs to be subjected to an in-depth analysis to determine the underlying reasons. In some cases, adjustments to management practices may be called for, while in others, the outcome may be to drop underutilized translations in future rounds.

Social and Political Impacts of Community-Level Linguistic Diversity My focus here has been primarily descriptive—revealing individual, community, and national linguistic pro les—and practical, in terms of understanding the implications of this diversity for doing the kinds of survey research that allow us to collect ne-grained individual-level information. But several possibilities p. 113

exist for further analysis, both to more fully

understand what explains the patterns we see here and to

explore the implications of language patterns, language distribution, and language capacity as variables that may help explain other social and political phenomena captured by Afrobarometer. While there are rich possibilities for further analysis with respect to all levels (individual, communal, and national) of language capacity and diversity, I want to focus in particular on the potential implications of the community-level diversity revealed in this analysis. Aside from its implications for survey management, does community-level linguistic diversity matter? Does it, for example, shape individuals’ attitudes toward their own identity, their neighbors, or their country? In particular, does living in a more diverse community draw people outward toward trust in, tolerance for, and acceptance of others, and toward a multi-ethnic national identity? Or does it accentuate di erences, pushing people apart and back toward the comfort of their own people, their own linguistic or ethnic group? A number of analysts have begun to look at these questions in the African context. For example, Albaugh (2016) analyzes the relationships between learning European languages and size of language group on the one hand, and preferences for parochial ethnic versus national identities on the other. The ndings presented here on community-level linguistic diversity suggest that this variable could be added into that analysis to deepen our understanding of intercommunal trust. Along these lines, Kasara (2013) explores the

impact of ethnic composition at the location level in Kenya (approximately 13,000 people) on inter-ethnic trust, and Robinson (2016) explores similar issues looking at district-level ethnic composition, across countries. A brief exploration of the data suggests that disaggregating even further, to investigate the impacts of ethnic composition at the EA level in respondents’ local communities, could be a fruitful extension of this research. We can illustrate this potential with a few initial tests. For example, focusing again on the 22 more linguistically diverse countries identi ed earlier, we nd that the number of home languages spoken in an EA is negatively correlated with trust in neighbors (Pearson’s r = −.154, p