The Leopard's Spots

The Leopard’s Spots Brill’s Studies in Language, Cognition and Culture

Series Editors

Alexandra Y. Aikhenvald (Cairns Institute, James Cook University) R.M.W. Dixon (Cairns Institute, James Cook University) N.J. Enfield (Department of Linguistics, University of Sydney)

VOLUME 11

The titles published in this series are listed at brill.com/bslc The Leopard’s Spots

Essays on Language, Cognition and Culture

Gerrit J. Dimmendaal

LEIDEN | BOSTON Cover illustration: Gertrud Schneider-Blum

Dimmendaal, Gerrit Jan, author. The Leopard’s spots : essays on language, cognition, and culture / By Gerrit J. Dimmendaal. pages cm. — (Brill’s Studies in Language, Cognition and Culture; 11) Includes bibliographical references and index. ISBN 978-90-04-22244-1 (hardback : alk. paper) — ISBN 978-90-04-22414-8 (e-book) 1. Language and languages—Variation—Africa. 2. Cognitive grammar—Africa. 3. Language and culture—Africa. I. Title. P40.45.A35D54 2015 496--dc23 2014045280

This publication has been typeset in the multilingual “Brill” typeface. With over 5,100 characters covering Latin, ipa, Greek, and Cyrillic, this typeface is especially suitable for use in the humanities. For more information, please see www.brill.com/brill-typeface. issn 1879-5412 isbn 978-90-04-22244-1 (hardback) isbn 978-90-04-22414-8 (e-book)

Copyright 2015 by Koninklijke Brill nv, Leiden, The Netherlands. Koninklijke Brill nv incorporates the imprints Brill, Brill Nijhoff and Hotei Publishing. All rights reserved. No part of this publication may be reproduced, translated, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission from the publisher. Authorization to photocopy items for internal or personal use is granted by Koninklijke Brill nv provided that the appropriate fees are paid directly to The Copyright Clearance Center, 222 Rosewood Drive, Suite 910, Danvers, ma 01923, usa. Fees are subject to change.

This book is printed on acid-free paper. Contents

Preface ix Abbreviations xii List of Tables, Maps and Figures xiv

1 By Way of Introduction 1

Part 1 Language Ecology

2 Language Ecology and Linguistic Diversity on the African Continent 9 2.1 The Genetic Classification of African Languages: A Brief State of the Art 9 2.2 Accretion Zones, Spread Zones, and Their Ecological Bases 15 2.3 Esoterogeny, Metatypy and Other Miracles of Language Contact 21

3 Accretion Zones and the Absence of Language Union in the Nuba Mountains 25 3.1 Genetic Diversity 25 3.2 Typological Diversity 32 3.2.1 Phonology 32 3.2.2 Lexical Diffusion 40 3.2.3 Morphological Diversity 42 3.2.4 Clause Structure 45 3.3 A Natural Refugium Zone 54

4 Esoterogeny and Localist Strategies in a Nuba Mountain Community 64 4.1 Introduction 64 4.2 Internal Variation in Tima and the “Apparent-time” Approach 64 4.3 An Initial Comparison with Katla 68 4.4 Metatypy, Esoterogeny, or Neither? 74 4.5 Localist Strategies 79 vi contents

Part 2 Language and Co-Evolution

5 Some Observations on Evolutionary Concepts in Current Linguistics 85 5.1 Evolutionary Concepts and the Study of Language: Some Earlier Attempts 85 5.2 A Closer Look at the Brown and Witkowski Hypothesis 88 5.2.1 The Historical-Comparative Evidence and Counter-Evidence 90 5.2.2 The Empirical Basis: Synchronic Evidence and Counterevidence 93 5.3 An Alternative Account: Language and Cognition 97 5.4 Conclusions and Prospects 99

6 Studying Lexical-Semantic Fields in Languages: Nature Versus Nurture, or Where Does Culture Come into It These Days? 101 6.1 Investigating the Interaction between Language and Cognition: Research on Colour Terminology 101 6.2 Language Typology and the Study of Language Universals 103 6.3 The Berlin and Kay Framework 104 6.4 Some Problems with the Berlin and Kay Model 105 6.5 Extending the Greenbergian Framework to Other Lexical Domains 112 6.6 The Expression of Space and Direction in a Cross-Linguistic Perspective 115 6.7 The Problem of the “Radical Translator” 120 6.8 Implicational Scales and Historical Reconstruction 121 6.9 Some New Evidence for Linguistic Relativity? 122 6.10 Some Final Observations 123

7 Lexical-Semantic Fields in Tima 127 7.1 Introduction 127 7.2 Bio-nomenclature 128 7.3 Colour 130 7.4 Shape and Texture 135 7.5 Taste 138 7.6 Body Parts 139 Contents vii

Part 3 Conversational Styles

8 Colourful psi’s Sleep Furiously: Depicting Emotional States in Some African Languages 147 8.1 Introduction 147 8.2 Categories and Event Structures 149 8.3 A Closer Look at Two African Language Families: Nilotic and Bantu 152 8.4 Interpreting Colourful psi’s 165 8.5 On psi’s and fta’s 167

9 Perception of the Living Dead and the Invisible Hand in Teso-Turkana 170 9.1 Introduction 170 9.2 Invisible Forces in Teso-Turkana 171 9.3 Perception of the Invisible Hand 175 9.4 A Note on Perception Verbs in a Nilotic Context 179

10 Conversational Styles in Tima 181 10.1 Introduction 181 10.2 Ideophones 182 10.3 Emotional States 187 10.4 Exoteric and Esoteric Languages 193

References 197 Language Index 216 Subject Index 220

Preface

The main impetus for this collection of essays came from a language documentation project “A linguistic and anthropological documentation of Tima”, financed by the Volkswagen Foundation between 2006 and 2012. I would like to express my deeply felt gratitude to the Foundation for making this research possible. Special thanks go to Dr. Vera Szöllösi-Brenig, director of the programme Documentation of Endangered Languages of the Volkswagen Foundation, for her personal dedication and warm personality. I would also like to thank the wonderful research project team for this project, consisting of Suzan Alamin, Abeer Bashir, Meike Meerpohl, Abdelrahim Mugaddam, Gertrud Schneider- Blum as well as two dedicated student assistants, Meikal Mumin and Nico Nassenstein, for making this project such a success. The project allowed all researchers involved to acquaint themselves with a fascinating language spoken by a captivating group of people in the Nuba Mountains of Sudan, who play a central role in several chapters of this volume. I would also like to express my sincere thanks to the Tima community in the Nuba Mountains and the Tima diaspora in Khartoum for their kind hospitality and their interest in documenting their language and culture. Their “habitus” formed a major inspi- ration for me to reflect more deeply upon the link and interaction between language, culture and cognition. I am particularly grateful to the two main language helpers in the Tima documentation project, Nassardeen Abdallah and Hamid Kafi, whose pictures are shown on the front page of this monograph. The Tima documentation project would not have become such a success without the perseverance and positive thinking of Gertrud (Trudel) Schneider- Blum, whose tremendous contributions are gratefully acknowledged here. I am also deeply indebted to the editors of the series in which this monograph appears, Sasha Aikhenvald, Bob Dixon and Nick Enfield, as well as Stephanie Paalvast, Paige Sammartino and Irene van Rossum (Brill Publishers), for their untiring support and interest in the topic. Special thanks are also due to Prof. Al-Amin Abu-Manga, former Director of the Institute for African and Asian Studies, University of Khartoum, for his unflagging support and personal interest in languages and cultures of Sudan. This monograph could not have been finished without the indefatigable support, at different points in time, of four wonderful student-assistants, Silke Focke, Marvin Kumetat, Jan Peters, and Tobias Simon. Furthermore, the help of the cartographer Monika Feinen in preparing the maps is gratefully acknowledged here. Also, the positive and collegial atmosphere at the Institute for African Studies and Egyptology, University of Cologne, provides a perfect x preface intellectual basis for scientific research. I would like to express my deeply felt gratitude to all my colleagues for this. Special thanks are also due to the rector of the University of Cologne, Prof. Axel Freimuth, for bestowing the Leo-Spitzer-Prize for excellent research upon me in 2013. This prize allows me to spend more time on research for a period of three years. I would also like to express my sincere gratitude to Bompeti Ngila for his help in interpreting the Mongo data in Chapter 5, to Thérèse Harlemann-Ndobo for data on Baca, to Osman Adam Ismail for data on Daju of Lagowa, and to Birgit Hellwig for making her data on Katla available. The extensive comments and criticism at different points in time from Felix Ameka, Nick Enfield, Axel Fleisch, Eva Lindström, Steffen Lorenz, Nico Nassenstein, Anna Wierzbicka, and Andrea Wolvers on different chapters are also gratefully acknowledged here. Last but not least, I would like to thank the wonderful Dave Roberts for his corrections and stylistic recommendations. Some readers may be surprised—or even shocked—by the title of this monograph if they are familiar with a book with an identical title published more than a century ago by Thomas F. Dixon, Jr., which was the first work of a trilogy on the Ku Klux Clan in the United States. Human beings construct identities on “self” and “otherness” in sometimes abstruse and morally despi- cable ways; unfortunately, this is part of human history. The present monograph, however, grew out of admiration and respect for cultural diversity and the various ways the human mind has constructed cultural identities and created language diversity. The monograph is divided into three parts, each starting with previously published articles and followed by others elaborating on these. I would like to thank the following publishers for making this approach possible:

“Language ecology and linguistic diversity on the African continent”, with kind permission of Wiley Publishers and the editors of Language and Linguistics Compass.

“Esoterogeny and localist strategies in a Nuba Mountain community”, with kind permission of Rüdiger Köppe, and the editors of Sprache und Geschichte in Afrika.

“Some observations on evolutionary concepts in current linguistics”. In: Walburga von Raffler-Engel et al. (eds.), Essays in Language Origins ii, Preface xi

pp. 225–44. With kind permission of John Benjamins Publishing Com- pany, Amsterdam and Philadelphia.

“Colourful psi’s sleep furiously: Depicting emotional states in some African languages”, with kind permission of the editors of Pragmatics and Cognition 10: 55–80 (2002).

“Studying lexical-semantic fields in languages: Nature versus nurture, or Where does culture come in these days?”, with kind permission of the editor of Frankfurter Afrikanistische Blätter, Rainer Vossen.

Some of the ideas presented in Chapter 3 were part of a keynote address deliv- ered at the Nuba Mountain Languages Conference, Leiden University (the Netherlands), in August 2011. I would like to thank the organisers for their kind invitation, and members from the audience for their critical comments and suggestions. Abbreviations abl ablative abs absolutive ap antipassive as aspect marker acc accusative case agr agreement APPL applicative ben benefactive caus causative cl noun-class marker conv converb dat dative dem demonstrative dim diminutive dsc different subject converb erg ergative marker f feminine foc focus fut future fv final vowel gen genitive hab habitual id identifier ideo ideophone imp imperative inf infinitive ipfv imperfective instr instrumental loc locative log logophoric m masculine neg negative nom nominative ob object past past tense per perfect pfv perfective Abbreviations xiii pl plural poss possessive pp pronominal prefix pr pronominal pred predicative marker prep preposition prog progressive pro pronominal form prs present PRT particle psit perfective iterative qv quotative verb ref reference refl reflexive rel relative rtc root clause sg singular sm subject marker su subject tr transitive ven ventive List of Tables, Maps and Figures

Table caption

1 The genetic classification of languages in the Nuba Mountains 27 2 Kordofanian Nubian Languages 31 3 The vowel system of Tima 34 4 The vowel system of Koalib 36 5 The vowel system of Gaahgm 37 6 The vowel system of Shatt 38 7 The vowel system of Lagowa 38 8 Constituent order and head marking in languages of the Nuba Mountains 50 9 Lexical cognates between Tima and Katla 69 10 Pronominal subject markers in Katla and Julud 69 11 Pronominal subject markers in Tima 70 12 Pronominal subject marking in languages of the Nuba Mountains 71 13 Constituent order in languages of the Nuba Mountains 73 14 Pronominal subject marking in Otoro 73 15 Perception verbs in Tima 127

Map caption

1 Language families and linguistic isolates in Africa 14 2 The spreading of Eastern Sudanic languages 18 3 Verb-final Nilo-Saharan languages 19 4 Genetic diversity in the Nuba Mountains 28 5 Vowel systems in Nuba Mountain Languages 35 6 Constituent order in Nuba Mountain languages 51 7 Ergativity in languages of Northeastern Africa 72

Figure caption

1 Subclassification of Eastern Sudanic 30 2 Kordofanian (Schadeberg 1981a, 1981b) 65 3 Encoding sequence for botanical terms (Brown and Witkowski) 88 List Of Tables, Maps And Figures xv

4 Encoding sequence for zoological life classes (Brown and Witkowski) 89 5 Body part partonomy 98 6 Berlin and Kay’s Basic Colour Terms 104 7 Colour terms in Tima 131 8 Lexical Iconicity Hierarchy (Akita 2009) 185

Picture caption

1 Body part taxonomy in Tima 140 2 Grainstore in the Tima area 144

CHAPTER 1 By Way of Introduction

Evolutionary biology has served as a model science for linguists for more than a century now. This influence is reflected, not only in the still widely accepted representation of language relationships by means of family trees, but also in the conceptualisation of the nature of language change. “These involve the speed with which changes may occur, the degree of genetic and typological diversity in some areas as opposed to others, and so-called self-organising principles” (Dimmendaal 2011: 348). Biologists have claimed that biological evolution is concentrated in rapid events of speciation—the formation of new and distinct species—next to periods of relative equilibrium. Dixon (1997) argues for a similar model with respect to the divergence of genetically related languages. When looking at the linguistic map of Africa, it is also clear that certain areas manifest a much higher degree of genetic and typological diversity than others. Self-organisation, the process whereby organisms which are not closely related, or which are unrelated, independently acquire similar characteristics while evolving in varying ecosystems, has been argued to have parallels in language change as well. In other words, structural similarities between languages may result from inheritance, from areal contact, but also from self- organising principles. Dimmendaal (2011: 365–370) describes the emergence of verbal compounding from a combination of converb and main verb constructions in languages in Central Asia and Northeastern Africa as an example. Self- organisation also plays a role in Chapter 3 on typological diversity between languages spoken in the Nuba Mountains of Sudan. Evolutionary biologists like the late Stephen J. Gould have also invoked exaptation—the process by which features acquire functions for which they were not originally adapted or selected—as an important principle explaining speciation or biological diversity in the natural environment. For example, specific anatomical or behavioural traits may become dysfunctional or shift in their function. Bird feathers are a classic example. They probably evolved for temperature regulation originally, and were only adapted for flight at a later stage in their evolution. Indeed, the title of the present monograph is specifically inspired by the work of Stephen J. Gould, this 20th century genius and one of the most creative evolutionary biologists of our times, producing fascinating studies like The Panda’s Thumb (1980), Hen’s Teeth and Horse’s Toes (1983), and The Flamingo’s Smile (1985).

Linguists have also used exaptation as a concept in order to explain changes in languages. Lass (1990, 1997: 317), among others, argues that this process plays a role in languages historically. Specifically, Lass shows that late West- ern Indo-European aspectual grades were redeployed as number markers in Germanic. But human languages are also different to some extent from other products of evolution because of their highly adaptive and flexible nature. Humans can freely manipulate the structure of languages—if there is a need to do so. They may attach new meanings to words or constructions at any point, either con- sciously or subconsciously—through the “invisible hand”, as Keller (1994) calls this process of language use—again, if there is a need to do so. Also, whereas biologists usually have one dominant theory or a few competing explanatory theories in order to explain specific phenomena in the natural world surrounding us, one may find a huge variety of etiological myths in different speech communities around the world explaining one and the same phenomenon. Take the leopard’s spots as an example. Evolutionary biologists relate the rosette-like markings to selective advantages for nocturnal hunting. The leopard hunts mainly in forested habitats at night. The spots or rosettes improve camouflage, and this adaptation made it a more effective predator. Leopards have the largest distribution of any wild cat in the world, which shows that their adaptation has been quite effective. In contrast, lions, with their plain coats, are better adapted to open savannah country. Savannah animals like cheetahs, which also have spotted coats, need to rely less on camouflage because of their athleticism and speed. Because of their wide distribution throughout the world, leopards also turn up in numerous tales and myths. But whereas the scientific work of modern evolutionary biologists suggests one explanation (or at best a few concurring explanations) for the rosettes of leopards, the various etiological stories (e.g. in trickster tales) about the origin of the leopard’s spots or its social behaviour as found in different speech communities around the world show that the non- scientific mind may construct its own reasons why the spots exist. This is also reflected maybe in the English name for the animal, which derives from the Greek compound for ‘lion’ (λέων, leōn) and ‘male panther’ (πάρδος, pardos). In the trickster tale appropriated by Rudyard Kipling (1902) in his Just So Stories, for example, an Ethiopian hunter paints spots on Leopard with the tips of his fingers, having asked the latter whether he would prefer stripes like Zebra. Leopard chooses spots, but points out that they should not be as “vulgar-big” as those of Giraffe. Humans construct meaning through language, and they may do so in highly variable ways. This also applies to their description of the social behaviour of By Way of Introduction 3 leopards; see Paye and Lippert (1998) for examples from Liberia, and McKee (2010) for examples from the Democratic Republic of Congo. Leopards may symbolise courage, as in Chinese trickster tales, but also strength or virility. Rulers in the Kingdom of Dahomey, for example, were thought to originate from man and leopard, according to oral traditions. In Pharaonic Egypt, the god Osiris was depicted wearing a leopard skin, thereby providing him with a nimbus of virility. Among Shona speakers in Zimbabwe, there are stories about leopards visiting homesteads, where they may remain impervious to any attempts to dislodge them. These leopards may proclaim their identity to a medium, state the purpose of their visit, then vanish. It is this cultural construction of meaning through language which is central to the present volume. Such constructions are presumably constrained by our cognitive system, but leave lots of space for culture-specific interpretations, thereby reflecting the highly adaptive nature of human language. The present monograph, which seeks to investigate this interaction between language, cognition and culture, is divided into three parts. In Part 1, language ecology is the central theme. In Chapter 2, evidence is adduced for the role played by climate change and technological innovation in explaining linguistic diversity (or the lack thereof), and above all for the quintessential role played by social factors when it comes to explaining language diversity on the African continent. In a follow-up study (Chapter 3), an attempt is made to explain how and why the Nuba Mountains in Sudan developed into an accretion zone, i.e. an area characterised by tremendous genetic and typological variation. Chapter 4 focuses on the so-called “localist strategies” as found among speakers of Tima—a Niger-Congo language spoken in the Nuba Mountains in Sudan which plays a central role in the book—and other speech communities in the genetically and typologically heterogeneous Nuba Mountain area. Language and the structures which co-evolved with culture are central to Part 2. Starting with an investigation of differences between languages in their structuring of semantic fields, Chapter 5 presents a critique of evolutionary explanations in this respect. Chapter 6 evaluates different views on the relative role played by cognitive principles as against culture (“nature” versus “nurture”) within specific semantic fields. These two chapters provide the framework for an additional contribution (Chapter 7) involving the study of a range of lexical and grammatical phenomena in Tima. These domains relate to bio- nomenclature, body part nomenclature, shape and texture as well as odour and taste, taking into account current views from cognitive anthropology and linguistics on these topics. Part 3 takes differences in conversational styles as a central theme. Chapter 8 addresses the role played by so-called colour terms in metaphorical 4 CHAPTER 1 and metonymic expressions related to mental states, which commonly involve the liver, heart or stomach as the main “container” or seat for emotions. It is followed by a contribution (Chapter 9) addressing lexical-semantic changes and the role played by culture, more specifically the “invisible hand” in this respect. These two chapters set the framework for an investigation of speech styles in Tima in Chapter 10. Interestingly, the liver or other organs do not appear to play a role in constructions expressing mental states in this language. Instead, the bone functions as the container for emotions and related phenomena. The same chapter discusses other properties of conversational strategies in Tima. These include interactional domains such as the use of ideophones, expressing experiential knowledge, and the zest for metaphorical and metonymic extensions in this language. The chapter concludes with a brief discussion of so-called esoteric and exoteric languages. Within the “nature versus nurture” debate in the humanities and the social sciences, the present monograph places a strong focus on the important role played by nurture, in particular when it comes to understanding human language in all its diversity, without denying the potential role played by cognitive constraints or more universal principles. The intellectual framework adhered to here is probably formulated in its most succinct form in the seminal article by Evans and Levinson (2009), which adduces extensive evidence against the existence of Universal Grammar, as advocated in particular within the Generative Enterprise (Chomsky 2011). Moreover, Evans and Levinson (2009) identify a strong link between language and cognition, also emphasising the importance of co-evolution of language and culture. Other authors take similar positions, for example Enfield (2004) and the various authors in De Busser and LaPolla (to appear). Their intellectual stance takes the tremendous typological variation between languages seriously, thereby avoiding reductionist models. The latter often operate with (sometimes non-falsifiable) ad hoc stipulations and assumptions about structural properties, in particular syntactic ones. From an evolutionary perspective, it may be argued that if languages were as rigidly “universal” in their structure as some current theoreticians claim they are, one would expect them to be far more uniform syntactically and morphologically—which they obviously are not, in the same that phonological systems are not. Rigid and highly specialised structures are disadvantageous, as evolution tells us. Flexible systems on the other hand, whether physical or social in nature, provide selective advantages. And human languages in all their diversity, and in the different cultural contexts where they play a role, show that this evolutionary principle also extends to them. This does not necessarily lead By Way of Introduction 5 to “Bongo-Bongoism”, as Douglas (1970) calls the anthropological practice of countering every generalisation with an exception. It is not the zest for exoti- cism which lies at the heart of this intellectual stance, but rather respect and admiration for the ingenious linguistic structures and impressive structural variation between languages which co-evolved with culture. This is what the present monograph is about.

Part 1 Language Ecology

∵

CHAPTER 2 Language Ecology and Linguistic Diversity on the African Continent

Africanists have been criticised by comparative linguists working on language families in other parts of the world for being lumpers (Newman 1990). This chapter reviews current views among specialists on genetic diversity on the African continent. In addition, it investigates some of the causal mechanisms behind this language diversity. More specifically, it discusses the role played by innovations in subsistence economies and climate change, with a special emphasis on attitudes towards the role of language as a marker of social identity and their effect on language diversity.

2.1 The Genetic Classification of African Languages: A Brief State of the Art

How well established are the genetic units or phyla proposed over 50 years ago by Joseph Greenberg (1963) in his seminal contribution on the genetic classification of African languages? Among the four phyla he originally proposed, the one most securely established today is Afroasiatic. Here, Greenberg followed up on pioneering work by nineteenth century scholars such as Müller (1876– 1888) on what was then called Hamito-Semitic, and twentieth century scholars like Delafosse (1914), who seems to have coined the term “afroasiatique”. In his initial series of studies on the genetic classification of African languages, published between 1949 and 1954 and reprinted as a monograph in 1955, Greenberg accepted this family as a valid genetic grouping. But he also added a branch, Chadic, which included languages spreading out in various directions from Lake Chad. As pointed out in Greenberg (1955), the term “Hamito-Semitic” for this family should be avoided, because Semitic constitutes only one of the five branches; moreover, the concept “Hamitic” had developed racist connotations during the preceding decades, in particular after Meinhof’s Die Sprachen der Hamiten (1912), which constituted a mixture of genetic, typological and (physical) anthropological criteria; Hamitic languages, according to Meinhof, were originally spoken by livestock rearing peoples of Caucasian origin. In a sequel to Greenberg (1963), Fleming (1969) proposed to excise one group referred to as West Cushitic from Cushitic, and to accord it the status of a

© koninklijke brill nv, leiden, ��5 | doi ��.��63/9789004224148_003 10 CHAPTER 2 separate primary branch within Afroasiatic. For this new branch, Fleming proposed the name Omotic (after a major river in the area, the Omo). Although its Afroasiatic affiliation has been disputed, the allocation of Omotic within this family is now well established, based on the attestation of morphological properties which this family shares with other Afroasiatic branches; see Hayward (2000) for a succinct survey. At the same time, it has to be said that Greenberg’s intuitions on the genetic unity of Khoisan has not been confirmed by subsequent research. Today, the few scholars working on these languages treat the three units which Greenberg grouped together under one phylum, Northern Khoisan, Central Khoisan and Southern Khoisan, as independent language families which cannot or can no longer be shown to be genetically related. Here, Greenberg may indeed have been misled to some extent by the extensive borrowing which occurred between different languages in the Khoisan area. See, for example, the study by Traill and Nakagawa (2000) on borrowing between the Northern Khoisan language !Xóo and the Southern Khoisan language Gui as a case study. Interestingly, Greenberg (1963) also included two languages spoken in Tanzania, Sandawe and Hadza, in the same Khoisan family. As Güldemann and Elderkin (2010) show by way of the comparative method, Sandawe forms a genetic unit with Central Khoisan and Kwadi (in Angola),1 and Hadza is a linguistic isolate. Niger-Congo (called Niger-Kordofanian in Greenberg 1963) is sometimes presented as the best established phylum on the continent. All specialists, however, do not share this optimistic view. According to Greenberg, this family consists of Atlantic, Gur (Voltaic), Mande, Kwa, Benue-Congo, and Adamawa-Eastern. In addition, he included a group of languages spoken in the Nuba Mountains (Sudan), which have come to be known as Kordofanian. There is indeed a core of language groups, each of which is fairly well established as a genetic unit itself, which includes Greenberg’s Benue-Congo, Kwa, Gur and Adamawa; the latter two form a larger subgroup, according to Kleinewillinghöfer (1996) and their genetic relation is beyond reasonable doubt. The evidence is not only lexical in nature, it is based primarily on a range of cognate grammatical morphemes. The so-called “Eastern” branch of Greenberg’s Adamawa-Eastern, however, does not fit into Niger-Congo; “Eastern” probably constitutes an independent language family (or independent families) which cannot or can no longer be shown to be related to Niger-Congo (or any other family), and is now usually referred to as Ubangian.

1 Kwadi was thought to be extinct, but more recently the Khoisan specialist Anne-Marie Fehn (personal communication) found two elderly ladies in Angola still speaking it, and using it as a kind of secret language between the two of them. Language Ecology and Linguistic Diversity 11

The status of Greenberg’s Atlantic group within Niger-Congo is still unclari- fied. The internal diversification within this presumed primary branch is indeed so huge that some scholars would argue that “Atlantic” is primarily an areal grouping representing a number of independent, early descen- dants of Niger-Congo. A few have challenged this view and would go as far as saying that some of the languages originally included in this family may not even belong to Niger-Congo; for a more detailed account of classificatory issues as well as Greenberg’s method of mass comparison, see Childs (2003: 19–53), Dimmendaal (2011), and Dimmendaal and Storch (To appear) for a recent update. Actual comparative evidence for Niger-Congo as a family using classical Neogrammarian methods has emerged in particular through the scholarly work of the late John Stewart. See the obituary by Mous (2007) for a full list of Stewart’s publications. In his comparative endeavours, Stewart (e.g. 2002, 2007) focused on a systematic phonological comparison between some members of this language family, in particular between Kwa and Bantu, as a major subgroup within Benue-Congo. He further compared his Proto-Potou-Akanic-Bantu (Proto-pab) with languages from Greenberg’s Atlantic branch, and argued that “[. . .] Proto-pab has the potential to serve as a pilot Proto-Niger-Congo in essentially the same way as a “Proto-Germanic-Latin-Greek-Sanskrit” served the pioneers of linguistic reconstruction as a pilot Proto-Indo-European” (Stewart 2002: 197). It is no coincidence that Stewart did not include two other families assumed by Greenberg to constitute primary branches of Niger-Congo, Mande and Ubangian, in his comparative studies. The actual comparative evidence for them having Niger-Congo affiliation is indeed rather slim, and no convincing evidence has been added over the past decades. Consequently, Mande and Ubangian are best treated as independent language families. Greenberg’s inclusion of a group of languages spoken in the Nuba Mountains of central Sudan, which have come to be known as Kordofanian, into a larger family termed Niger-Kordofanian by Greenberg (1963) and subsequently renamed Niger-Congo by Williamson (1989), has received wide acceptance among scholars. Greenberg (1963) assumed that the Kordofanian branch consists of five subgroups, today usually referred to as Heiban, Talodi, Rashad, Katla and Kadu. But as Schadeberg (1981a) argues, the Kadu group (also referred to as the Kadugli group) should be excised from Kordofanian, or Niger-Congo, and it is questionable whether the Kadu(gli) group should be affiliated with Nilo- Saharan. More recent research on these languages has made clear that the four remaining subgroups manifest more internal diversity than originally thought. Consequently, what has been called Kordofanian, in actual fact consists of two 12 CHAPTER 2 groups which are closely related, Heiban and Talodi, plus two more distantly related Niger-Congo families, Rashad and Katla. Nevertheless, their common origin as Niger-Congo languages is beyond reasonable doubt. Not only do they have reflexes of widespread, cognate noun-class markers as found elsewhere in Niger-Congo, they also manifest an amazingly detailed formal identity of verbal derivational markers with, for example, Proto-Bantu. This is not necessarily due to the fact that Bantu or Benue-Congo and these languages in the Nuba Mountains are closely related, but rather because they, and other language groups in the peripheral zones of Niger-Congo such as Atlantic, are more conservative in their morphological structure, retaining cognate verbal derivational markers as well as noun-class markers.2 As Bender (2000: 43) states, “[. . . o]f the four ‘Greenbergian phyla’ [. . .] Nilo- Saharan is probably the least widely accepted.” Indeed, this view seems to be widespread, in particular among non-specialists. But it is not clear what this scepticism is based upon. To the outsider it may look as if Greenberg simply grouped together a range of languages formally treated as isolated units and situated mainly between Afroasiatic to the north and Niger-Congo to the south. However, contrary to a widely held belief that Greenberg was just lump- ing together left-over languages, in actual fact his classification was based on a judicious evaluation of the existing evidence. As was the case with other linguistic groupings, Greenberg elaborated upon research of investigators such as Margaret Bryan, Archibald Tucker and also Dietrich Westermann. Greenberg started with his own identification of a Macro-Sudanic family (1955), which was subsequently renamed Chari-Nile. Then, in his 1963 classification, language groups and languages formerly considered to be isolated units, such as the Songhai cluster, Saharan, the Maban group and Mimi, Fur, the Kunama cluster, and the Koman group and Gumuz were grouped together in a new family called Nilo-Saharan. More and more grammatical evidence has emerged over the past decades for the existence of this family, as a result of improved descriptions and historical-comparative studies on lower-level units. Still, there are two groups and one isolated language that do not fit in with the emerging historical reconstructions: Songhai, Koman and Gumuz. Nicolaï (2003) has reviewed the historical evidence for the affiliation of Songhai to Nilo-Saharan, and concludes that the actual evidence is very poor indeed. Similarly, very few of the more widespread nominal and verbal morphological markers of Nilo- Saharan are attested in the Koman languages or Gumuz, which are spoken

2 Extensive lexical data are available for only a few Niger-Congo languages in the Nuba Mountains. This complicates the search for lexical cognates between these languages and remaining Niger-Congo subgroups. Language Ecology and Linguistic Diversity 13 in the border area between Ethiopia and Sudan. Their genetic status remains debatable, mainly due to lack of more extensive data. Fortunately enough, a number of Koman languages and Gumuz are currently being studied in more detail. Paucity of new linguistic material on these and other extant languages has indeed formed a major handicap for our comparative endeavours. Obviously, this makes Greenberg’s deep insights on genetic relationships even more impressive, with the establishment of a Nilo-Saharan family being his most important contribution to the genetic classification of African languages. The only all-embracing comparative study of the Nilo-Saharan family today is Ehret (2001), who in fact includes Songhai and Koman as members of this family. Ehret presents some of the grammatical evidence for Nilo-Saharan now available, e.g. with respect to case marking, number marking, and other morphological properties of major categories like nouns and verbs. However, there are a number of methodological flaws in Ehret’s study (Blench, 2000). The actual number of convincing cognates is still too small to be able to reconstruct the original sound system, and much more comparative work needs to be done at lower level genetic units within Nilo-Saharan in order to understand the historical development of its phonological and grammatical systems. In summarising the current state of knowledge, we can state the following:

In addition to Afroasiatic, Niger-Congo, and Nilo-Saharan (the latter two in a modified or “reduced” form) the following language families or phyla can be identified: Northern Khoisan, Central Khoisan and Sandawe as well as Kwadi, Southern Khoisan, Mande, Songhai, Ubangian, Kadu, and Koman. Whether all languages that Greenberg (1963) originally included in his Atlantic branch of Niger-Congo in fact belong there will remain a bone of contention for years to come. Next to these eleven (plus) language families or phyla, there are several isolated languages. Apart from Gumuz and Hadza, there are a number of others, e.g. Baŋgi Me in Mali, Dompo and Mpra in Ghana (Blench 1999, To appear, Dimmendaal 2011), Jalaa in Nigeria (Kleinewillinghöfer 1996), and Laal in Chad. Boyeldieu (1977) states that the latter language shows grammatical and lexical similarities to Adamawa (Niger-Congo), Chadic (Afroasiatic) and an unknown source. The problem of historical “layering” of inherited and borrowed language material, as Aikhenvald (2007) has called this process, is indeed the main reason why it is no longer possible to arrive at more convincing hypotheses about genetic relationships in these cases. This problem also applies to two linguistic isolates in southern Ethiopia, Ongota (otherwise known as Biraile) and Shabo. This would bring the total number of languages families in Africa, including those with single members, 14 CHAPTER 2

map 1 Language families and linguistic isolates in Africa

to nineteen.3 Several of them may indeed be the last representatives of language families that are otherwise extinct. It is rather striking, when plotting these isolates—many of which are endangered—on a map (see Map 1), that most of them are spoken at the fringe of so-called spread zones (in the sense of Nichols 1992, 1997) like Nilotic, Bantu or Cushitic.4

3 The list of linguistic isolates is not necessarily exhaustive, as several areas are still poorly studied. 4 See Dimmendaal and Voeltz (2007) for a recent survey of endangered languages on the African continent. Language Ecology and Linguistic Diversity 15

The revised picture of the classification of African languages shows that genetic diversity is more extensive than assumed in Greenberg (1963), partly because some of the language isolates mentioned above were not known to him or to his contemporaries. It seems equally likely that in certain areas genetic diversity was much greater in the past, and that expanding language families absorbed these speech communities linguistically. It further makes clear that some areas are still much more genetically diversified than others. Genetic diversity and sheer language density are particularly prominent in the belt of land on either side of the River Niger, the Nuba Mountains (Sudan), the Ethiopian Highlands and southern Africa. Section 2.2 discusses some of the potential causes of this historical situation.

2.2 Accretion Zones, Spread Zones, and Their Ecological Bases

The “Khoisan” area, claimed by Greenberg (1963) to be a genetic grouping but now widely held to be an areal grouping instead, embraces three distinct language families spread across southern Africa: Northern Khoisan, Central Khoisan and Southern Khoisan (Vossen 2013). In spite of some common traits, the most prominent ones being the use of clicks and the widespread use of verb concatenation, the typological distinctions within the Khoisan area are huge (Güldemann 1998). This makes the Khoisan area an accretion zone according to Nichols’ (1997) terminology, i.e. an area with a high genetic and typological diversity. Here, climate conditions varied considerably in the past; during the dry period between 7500 bp and 4500 bp some areas would have been uninhabited. The most dramatic innovation in subsistence economy involved the introduction of agriculture and pastoralism some 2000 years ago, when speakers of Central Khoisan languages entered the area from the northeast, where they had likely acquired agriculture from the expanding Bantu at a time when the Kalahari was more amenable to agriculture (Güldemann 2006). The subsequent desiccation of the Kalahari apparently led some groups to re- adopt a hunter-gatherer economy. We may assume that here—as elsewhere on the continent—ethnic fission and fusion accompanied by language shift occurred, leading towards both linguistic convergence and divergence over the past millennia.5 The Khoisan accretion zone is interspersed with representatives from a Niger-Congo subgroup, namely Bantu, which cover major areas from Cameroon

5 As Watkins (2001) points out, there may be both divergence and convergence in situations of so-called equilibrium as well as punctuation in the sense of Dixon (1997). 16 CHAPTER 2 in the northwest to Kenya in the east and South Africa in the south. The Bantu subgroup is a proto-typical spread zone within the Benue-Congo branch of Niger-Congo. Bantu expansion is widely assumed to have involved a number of independent movements (Vansina 1995: 191). Consequently, spread zones and accretion (or residual) zones occur in neighbouring areas. This raises the more general question of what drives phylic dispersals. It is a widely held view by archaeologists that technological innovations played an important role in the expansion of specific groups and, consequently, of the languages they spoke.6 Blench (2006) argues that such a model is problematic when applied to Africa in general because no terms unambiguously related to agriculture have been successfully reconstructed in the protolan- guage of any African phyla. The evidence for agriculture in sub-Saharan Africa is late; there is no evidence for any cultivated plants before 4000 bp. The main expansions of African language families such as Niger-Congo and Nilo-Saharan therefore seem to have started when the speakers were hunter-gatherers. It has been pointed out in the archaeological literature (e.g. Muzzolini 1993) that West Africa was populated sparsely by modern humans until the end of the Pleistocene (around 12,000 bp). This hyper-arid palaeo-geological phase was followed by a much wetter Holocene period. It seems plausible that during this later period inquisitive minds exploited new territories suitable for human habitation because of the wetter climate, and gradually migrated into previously unoccupied areas. Such expansions do not necessarily require technological innovations, although the invention of the bow and arrow and the introduction of domesticated dogs probably improved the success of foraging economies (Blench 2006: 129). The expansion of the Niger-Congo family is presumably related to both these technological changes, as well as to climate change. Linguistic isolates like Bangi Me, Dompo, Jalaa, Laal, and Mpra, and also larger units like Songhay and Mande are remnants of an earlier diversity that must have characterised West Africa, as well as other parts of the continent. Climate change must have affected the linguistic picture dramatically in a number of regions, for example in what is now the eastern Sahel zone. One notable consequence of the wetter climate that set in around 12,000 bp was the emergence of a major riverine system, the Wadi Howar or Yellow Nile, which connected the Ennedi Mountains in Eastern Chad with the Nile between the third and the fourth cataract (Pachur and Kröpelin 1993: 20). These aquatic resources teemed with flora and fauna roughly between 10,500 bp and 3500 bp,

6 In the case of the Bantu expansion from the Nigeria-Cameroon borderland across major regions of central, eastern and southern Africa, specialists consider that knowledge of iron working played a major role among agriculturalists. Language Ecology and Linguistic Diversity 17 when marginalisation of the area set in again. Whereas the earliest humans inhabiting this ecological zone were hunter-gatherers, pastoralism was introduced into the area probably as early as 7000 bp. There is solid archaeological evidence that the Wadi Howar played an important role in the diffusion of material culture among cattle-keepers in the area, also revealing intensive contacts with the Nile valley. The Lower Wadi Howar was mainly occupied between 7000 and 5000 bp, after which it seems to have been abandoned, whereas the occupation of the Middle Wadi Howar continued to the fourth millennium bp. The archaeological findings provide a plausible explanation for a number of linguistic phenomena, which remain enigmatic otherwise (Dimmendaal 2007, to appear a). These include the following:

1. The current distribution in particular of Eastern Sudanic, the largest and best established subgroup within Nilo-Saharan. The Taman and Nubian languages, all members of the Northern subgroup within Eastern Sudanic, are spoken in the border area between Sudan and Chad and in the Nuba Mountains, where other Nubian languages as well as additional members of the Northern Subgroup, Nyimang and Dinik, are spoken. As for Nara, it is spoken in Eritrea. Rilly (2004) has provided convincing evidence that the extinct Meroitic language also belonged to Northern Eastern Sudanic. See Map 2 for further details on the distribution of these languages. The Central subgroup, consisting of the (Eastern) Jebel languages (number 6 on Map 2), is found east of the Nuba Mountains, whereas members of the Southern subgroup (containing Temeinian, the Daju group, Surmic, and Nilotic) are spoken in the Nuba Mountains as well as west and south of this area. The principle of least effort suggests a centre of gravity or original homeland for Eastern Sudanic north of the Nuba Mountains. For this reason, the diffusion has been referred to as the “Wadi Howar diaspora” (Dimmendaal, 2007).7 2. Heine (1976a) was the first to point out that Nilo-Saharan languages stretching from Ethiopia and Eritrea in the east across northern Sudan to Chad in the west manifest typological similarities to Afroasiatic languages in Ethiopia in terms of constituent order (involving a verb-final structure) and extensive case marking. Additional typological features characterising this convergence area, covering Nilo-Saharan subgroups like Nubian, Taman, Fur, Maban and Saharan, are described in Dimmendaal (2008a).8 They include Differential Object Marking as a case-marking strategy, the frequent use of light verb plus coverb constructions (‘do/say x’), and the use of converb constructions.

7 This hypothesis has received support from recent archaeological findings (Becker 2011). 8 For additional case studies on convergence zones in Africa see Heine and Nurse (2008). 18 CHAPTER 2

map 2 The spreading of Eastern Sudanic languages

Today, however, the Nilo-Saharan languages sharing these typological features with Afroasiatic languages in Ethiopia are found in a non-contiguous area of northern Sudan and Chad (see Map 3). The presence of a former contact zone in the Wadi Howar region, as already evidenced through similarities in material culture, provides a natural historical explanation for these typological features, which tend to spread through multilingualism (Dimmendaal, to appear a). Language Ecology and Linguistic Diversity 19

map 3 Verb-final Nilo-Saharan languages

3. The typological pattern of Central Eastern Sudanic and Southern Eastern Sudanic languages deviates radically from the pattern found in Northern Eastern Sudanic, although remnant features may still be found in the Southern subgroup (Dimmendaal, 2004). The latter languages are typologically similar to some of the Niger-Congo languages in the Nuba Mountains in terms of constituent order and the use of extensive verb morphology to express clausal relations. Also, Southern Eastern Sudanic groups, more specifically those belonging to Nilotic and Surmic, developed split ergativity, a phenomenon also found in Niger-Congo languages like Tima in the Nuba Mountains. The earliest speakers of Eastern Sudanic languages were probably pastoralists, since roots related to this subsistence economy can be reconstructed for its earliest stages (Dimmendaal, 2007). As pointed out above, there is evidence for structural borrowing between Niger-Congo languages in the Nuba Mountains and Eastern Sudanic (Nilo-Saharan) groups like Nilotic and Surmic. 20 CHAPTER 2

This is supported by evidence for lexical borrowing. For example, the widespread Nilo-Saharan root for a typical savannah dweller such as ‘elephant’ is also attested in at least one Niger-Congo (Kordofanian) branch in the southeastern Nuba Mountains, the Heiban group. Schadeberg (1981b: 159) reconstructs a root *-oŋor for Proto-Heiban. This is also the common Nilo-Saharan form for ‘elephant’. Compare, for example, ɔŋɔr in the Taman language Tama (spoken in Darfur), and ŋɔrɔ in the Surmic language Tirma (spoken in southern Ethiopia), or aŋar (plural form) in the Nilotic language Anywa (South Sudan); see Chapter 3 for further details. The lexical and grammatical convergence suggests that Eastern Sudanic (Nilo-Saharan) pastoralists of the savannah surrounding the Nuba Mountains and agriculturalists in the Nuba Mountains themselves speaking Niger-Congo languages were in contact with each other, exchanging material culture and thereby reducing ecological risks. The southward migration of some Eastern Sudanic groups, more specifically of the ancestors of Nilotic and Surmic groups, illustrates the important role of pastoralism for language expansion in the area. Other Eastern Sudanic groups ended up in the Nuba Mountains, where they now concentrate on agriculture, although most do have some livestock as well. Along the southern range, we find the Kadu languages, which probably constitute an independent family (see Map 4, p. 28). The Benue-Congo and Kwa groups appear to be the closest linguistic relatives of these Niger-Congo languages in the Nuba Mountains, and are spoken thousands of miles further west. So the genetic and typological diversity in this area is not necessarily the result of an ancient diffusion zone for human occupation as with the Khoisan area in southern Africa. Rather, we are dealing with a refugium, i.e. a region that has attracted migration because it has remained unaltered by climate change affecting surrounding regions. Another such linguistically diverse area functioning as an ancient refugium or retreat zone hardly affected by desertification is found in the Ethiopian Highlands, where Afroasiatic and Nilo-Saharan languages are spoken, as well as the linguistic isolates Ongota and Shabo. In the Bantu case, the adoption of new cultivation systems led to demo- graphic and linguistic expansion from the Cameroon-Nigeria area. This historical process thus supports the farming-and-language dispersal model (Bellwood and Renfrew, 2003). Bellwood (2003) asserts that, whether agriculture was being spread by converting hunter-gatherers or by range-expanding farmers, both would have become subject to population increase in favourable environments. This no doubt is true for Bantu, but less convincing for the Nuba Mountain area. Here, we are dealing with non-expansionist or non-spread Language Ecology and Linguistic Diversity 21

“agricultural languages”; see Campbell (2003) for a discussion of other such cases elsewhere in the world. In a brief synoptic survey like this, it is not possible to go into the details of all potentially relevant parameters. Nevertheless, we may already conclude that a combination of climate change and changes in subsistence economies led to the spreading, displacement and reduction of language families in Africa as it did in other parts of the world. But the social factors and cultural choices should not be neglected either, as argued next.

2.3 Esoterogeny, Metatypy and Other Miracles of Language Contact

One frequent effect accompanying shift from a source language into a target language involves interference, as Thomason and Kaufman (1988) point out in their classic contribution on language contact phenomena. Let us examine one such case of so-called shift-induced interference as well as two other types of contact-induced language change in an African context in order to show how and why such processes may influence linguistic diversity. The Nilotic (Nilo-Saharan) language Luo, is spoken today by well over three million people in western Kenya as well as neighbouring regions of Tanzania and Uganda. One important reason for this large population was a massive shift in language solidarity from neighbouring Bantu languages. The result was a dramatic restructuring of Luo. The task of distinguishing inherited from dif- fused material becomes difficult if there are no closely related languages, as in the case of Laal or Ongota discussed above. But in the case of Luo this is not problematic. Luo is part of a cluster of closely related languages, including Alur, Chopi, Kumam, Acholi and Lango. None of the other members of this cluster manifest a strong influence from Bantu, either lexically or grammatically. Luo developed three past tense markers on the verb from adverbs of time. The result was a tense-marking system which parallels that of neighbouring Bantu languages. Whereas Luo has retained two Nilotic number-marking suffixes in its nominal morphology, it has also developed nominal class prefixes, again parallel to neighbouring Bantu languages. The prefixes came about, not only through extensive borrowing from these, but also through language- internal developments whereby former compounds were reinterpreted as nominal prefixes plus roots, e.g. ja-luo/jo-luo ‘Luo person/people’. Whereas neighbouring Bantu languages tend to use a class prefix ki- in order to express instruments, Luo uses a prefix ra- (e.g. ra-ŋɪy ‘mirror’, from ŋɪyɔ ‘recognize’). The prefix ki- is also used in these Bantu languages to refer to individuals with 22 CHAPTER 2 specific handicaps, and exactly the same alternation occurs again in the corresponding Luo forms with ra- (e.g. ra-ŋɔl ‘handicapped person, from ŋɔl ‘be handicapped, limp’); see Dimmendaal (2001a) for further details. The second case study concerns Baale, a language spoken by a community forming an ethnic unit with the neighbouring Tirma and Chai, called Suri or Surma, all of whom live in the border area between southwestern Ethiopia and South Sudan. Baale is closely related to the Didinga-Murle languages, with which it forms the South-Western branch of Southern Surmic. The Tirma-Chai- Mursi dialect cluster on the other hand is part of the distantly related South- Eastern branch of Southern Surmic; Surmic is part of the Eastern Sudanic branch within Nilo-Saharan. The Baale number around 9,000 or less, whereas Tirma-Chai speakers number around 30,000. The latter see themselves as transhumant pastoralists, whereas the Baale concentrate on agriculture, because they live in an area infected by tsetse flies. Most Baale are bilingual in Tirma and/or Chai, but the latter rarely speak Baale. Whereas the Baale form one ethnic group with the Tirma and Chai, they have not given up their language, presumably because the former still form a tightknit community with a partly separate identity from the pastoral Tirma-Chai. Nevertheless, their use of the inter-community language Tirma-Chai has apparently been so extensive, that a dramatic phonological restructuring and massive morphosyntactic restructuring involving calquing, i.e. the copying of constructional meanings from Tirma-Chai, occurred in Baale. We thus have a classic case of what Ross (e.g. 2001) has called metatypy. Unlike the closely related Didinga-Murle languages, for example, Baale does not allow closed syllables in word-final position. Thus, the Didinga and Murle word for ‘axe’, mɛɛlɛk, corresponds to mɛɛlɛ in Baale. Similar calquing towards Tirma-Chai can be observed in Baale morphology and syntax. Unlike the closely related Didinga-Murle languages, which are verb-initial, Baale has a relatively free constituent order, allowing for svo, vso, ovs or sov depending on the pragmatic context. This system parallels that of Tirma-Chai. Similarly, one finds that lexical idioms, for example compounds, are formed after a pattern also attested in Tirma-Chai. Thus, Baale ɔt̪a ʊt̪ʊ ‘nipple, lit. breast-mouth’, corresponds to Tirma way tugo. For additional instances of metatypy in Baale see Dimmendaal (2001b) and Yigezu (2006). The result again is an increase in linguistic diversity, because of dramatic structural innovations in Baale. Unlike Luo, however, the restructuring process did not result from imposition of second-language learners in a language shift situation, but rather from calquing. The third and final case discussed here, again showing how people’s construction of their social environment may affect the linguistic picture, involves Language Ecology and Linguistic Diversity 23 a poorly understood aspect of historical language modification: deliberate language change. (See the succinct survey by Thomason, 2007). This is well attested in modern youth languages from a range of African metropoles (Kießling and Mous 2004; Nassenstein and Wolvers, to appear). Ross (2001), for example, has argued that such restructuring processes whereby speakers of a language add linguistic innovations that increase the language’s complexity in order to highlight their distinctiveness from neighbouring groups characterised language change among entire speech communities in the Pacific; Ross calls this phenomenon esoterogeny. As shown in Chapter 4, the Tima community in the Nuba Mountains in Sudan also have an oral tradition referring to such deliberate changes by their ancestors, so that neighbouring groups could no longer understand them. Yet, when comparing Tima with the closely related Katla language, one does not find an elaboration of the lexicon, or an obvious increase in the frequency of opaque idioms (these being the pro- claimed properties of esoterogeny). Both languages have rather complex verb morphology, with at least twelve slots for verbal inflectional and derivational markers, with a plethora of allomorphs conditioned by atr and fronting harmony, and other types of morphophonemic processes. What appears to make Tima more complex than Katla is the fact that its variable constituent order (svo, ovs, vso, sov) is strongly governed by pragmatic principles, whereas Katla has a fairly strict svo structure. In Tima, this affects the inflectional morphology of the verb, i.e. the presence or absence as well as the shape of person and tense-aspect markers amongst other things. There is a widespread view held by outsiders that Tima is hard to learn, and Tima people are aware of this. However, as argued in Dimmendaal (2009a; reproduced here as Chapter 4), the deviant structure probably resulted, not from language manipulation as oral traditions would have it, but rather from another type of contact-induced change: the imposition of structural properties from specific source languages into Tima in a language shift situation, similar to what must have happened in Luo (as discussed above). So the Tima oral tradition of deliberate language change by their ancestors must be based on a post hoc rationalisation, in order to explain why their language is so different from those of their neighbours. In the same way, their language also became an important emblematic sign of ethnic identity, a process reminiscent of Hill’s (2001) “localist” strategies of closed agricultural language communities in Central America which, like Tima, did not expand, and whose “insider/outsider” boundaries are marked by correspondingly abrupt linguistic discontinuity. Tima is one of over forty speech communities in the Nuba Mountains, most of whom are self-sufficient without any larger networks for the exchange of 24 CHAPTER 2 material or immaterial culture. Consequently, no lingua francas other than Arabic, whose spreading is a more recent development, or larger convergence zones developed in this area. Where individuals have large and dispersed social networks, one may expect linguistic uniformity. Conversely, where social networks are small and tightly self-contained, many distinct languages will ultimately evolve (Nettle 1999: 59). This is indeed the pattern one observes in the Nuba Mountains. Similar patterns seem to have emerged among predomi- nantly agricultural economies in central and northern Nigeria, where numerous Chadic (Afroasiatic), Benue-Congo, and Adamawa (i.e. Niger-Congo) languages are spoken. Apart from these language contact situations discussed above, there are of course others, for example pidginisation, creolisation or the formation of syn- cretic or intertwined languages. The best known examples of the latter outside Africa are Michif (Canada) or Mednyj Aleut (Russia), or the more recent case of Gurindji Kriol (Australia); see Dimmendaal (2011: 237–252) for further details. Similar restructuring may have taken place in northern Songhai lects such as Tadaksahak (Mali), and Songhai varieties in Niger and the Algeria-Morocco border zone, all of which are strongly influenced by Berber languages in the area. The Berber branch within Afroasiatic is spread over large parts of northern Africa, from Egypt in the east to Mauritania in the west, but many Berber varieties are now being replaced by Arabic dialects. Nettle (1999: 150) argues that the ethnolinguistic map is a product of people’s social behaviour, a view also defended here. But Nettle also claims that social behaviour is motivated by the economic necessities of subsistence, the latter in turn being linked to the ecological setting. It would seem, however, that reducing all social behaviour and people’s rational choices to changing economic and ecological settings would do no justice to reality. These factors may set the scene, but they cannot predict whether, for example, language shift will occur or not. The emerging picture of language ecology and corresponding genetic and typological diversity in Africa and elsewhere is more likely to be one resulting from linguistic change mediated by social factors, enacted by humans whose behaviour is only predictable to a limited extent.9

9 This chapter is a slightly modified and updated version of Dimmendaal (2008a). The final section of the original publication has been left out, as it overlaps with the contents of Chapters 3 and 4. CHAPTER 3 Accretion Zones and the Absence of Language Union in the Nuba Mountains

The Nuba Mountains in Sudan, an area roughly the size of Scotland, are characterised by a large number of languages belonging to three different phyla, Nilo-Saharan, Niger-Congo and Kadu (Dimmendaal 2009b: 90). The same area features a considerable degree of typological variation and consequently may be characterised as an accretion, or residual zone. It is this linguistic complexity that this chapter seeks to explain. I provide a description of the genetic and typological variation and its possible historical origin, then attempt to explain the low degree of convergence or language union in the area.

3.1 Genetic Diversity

The linguistic complexity of the Nuba Mountains in Sudan has impressed many scholars. In one of the earliest sources, MacDiarmid and MacDiarmid (1931), language groups are identified which are still considered valid today (in alphabetical order): Daju, Heiban, Kadu, Katla, Lafofa, Nubian, Nyimang, Rashad, Talodi, and Temein. Authors such as Nadel (1947) use phrases like ‘bewildering complexity’ and ‘cultural constellation’ in his detailed anthropological account of the Nuba Mountains.1 From the late 1930s onwards, Roland Stevenson also carried out research on the languages and cultures of the Nuba Mountain area. His pioneering work resulted in a doctoral dissertation presented at the School of Oriental and African Studies, University of London in 1951, sections of which were published in volumes 40–41 of Afrika und Übersee (Stevenson 1956, 1957). His findings were incorporated by Tucker and Bryan (1956, 1966) in their survey of the “Non-Bantu” languages of Northeastern Africa. In his ethnography of the Nuba Mountains (that also contains an extensive bibliography), Stevenson (1984: 7) points out like previous authors

1 Rüppell (1829) is probably the earliest source on languages of the Nuba Mountains. For further details concerning linguistic and anthropological sources, see Schadeberg and Blench (2013a).

© koninklijke brill nv, leiden, ��5 | doi ��.��63/9789004224148_004 26 CHAPTER 3 that “[. . . t]his diversity of the Nuba peoples clearly presents difficulties of classification [. . .]. A long recognised feature of the Nuba Mountain area is the multiplicity of its languages and dialects. Communities only a few miles apart, and even on the same hill-range (such as Eliri and parts of the Moro hills) may speak mutually unintelligible languages: In Eliri, in fact, there are three distinct languages. The actual number of language groups in the Nuba Mountains as a whole is fewer than at one time supposed, but nevertheless confusing when added to the other diversities of stock and culture.” In his genetic classification of African languages, Greenberg (1955) classifies the various language clusters identified by Stevenson and other scholars into three different families or genetic groupings. Table 1 lists these, as they are also found in the follow-up classification (Greenberg 1963). Apart from the descriptions of different languages in the Nuba Mountains as provided by Stevenson (1956), Schadeberg (1981b, 1981c) and Schadeberg and Blench (2013) are the only more extensive recent sources. Schadeberg (1981b) presents a comparative survey of the lexicon and noun-class system of the Heiban group within Greenberg’s Kordofanian, whereas Schadeberg (1981c) investigates the lexicon and nominal class system of the genetically related Talodi group from a historical point of view. Schadeberg (1981a) argues that Greenberg’s Tumtum group, renamed Kadugli by the former, is not part of Kordofanian or even Niger-Congo for that matter. Schadeberg (1981a: 304) further argues that “[. . . t]he lexical and grammatical similarities linking KADUGLI with Nilo-Saharan are in no way inferior to those that have been adduced for a number of other language groups. It is for these reasons that I recommend that KADUGLI may be included in the search for substantial Nilo-Saharan comparisons.” Bender (1996) also includes Kadu (as this family is referred to in his study) as a member of the Nilo-Saharan phylum, but Ehret (2001: 65) calls Kadu “[. . .] a group of doubtful connection to the Nilo-Saharan family at all [. . .]”. For the remaining two branches of Greenberg’s Kordofanian, Rashad and Katla (hereafter called Katloid rather than Katla), no comparative studies similar in size to Schadeberg (1981b, 1981c) have yet been published. Schadeberg (2013) presents comparative data on the Rashad group (without attempting to reconstruct earlier forms), whereas Dimmendaal (2014a) presents a historical- comparative study of the noun-class system of the Katloid group, together with a lexical and grammatical comparison with the Rashad group. Dimmendaal finds no evidence for close genetic affiliation between the four Niger-Congo groups spoken in the Nuba Mountains. Whereas Talodi and Heiban appear to form a genetic unit, there is little evidence for close genetic affinity of these groups with the Rashad and Katloid group. The latter two, however, probably do form a genetic unit. Blench (2013) reaches the same conclusion. Accretion Zones and the Absence of Language Union 27 table 1 The genetic classification of languages in the Nuba Mountains, Sudan

Stevenson (1984) Greenberg (1955) Greenberg (1963)

1. Koalib-Moro Kordofanian Kordofanian (a) Koalib (b) Heiban, Laro, Otoro (c) Shwai, Tira, Moro, Fungor (with Kau and Nyaro) 2. Talodi-Mesakin Kordofanian Kordofanian (a) Talodi (b) Eliri (c) Mesakin (d) Acheron, Tacho, Torona, Kuku-Lumun (a dialect cluster in Moro Hills) 3. Lafofa Kordofanian Kordofanian 4. Tegali-Tagoi Kordofanian Kordofanian (a) Tegali sub-group: Tegali, Rashad, Kajakja (b) Tagoi sub-group: Tagoi, Tumale, Morab 5. Kadugli-Korongo Kordofanian Kordofanian (a) Tulishi, Keiga, Kanga (b) Miri, Kadugli, Katcha, Tumma (c) Korongo, Tumtum 6. Temein Isolate Eastern Sudanic (a) Temein (Nilo-Saharan) (b) Keiga Jirru, Teisei-Umm-Danab 7. Katla Kordofanian Kordofanian (a) Katla, Julud (b) Tima 8. Nyimang Eastern Sudanic Eastern Sudanic (a) Nyimang (Nilo-Saharan) (b) Afitti (eastern part of Jebel Dair) 9. Hill Nubian Eastern Sudanic Eastern Sudanic (a) Dair, Kadaru, Ghulfan, and dialect of some (Nilo-Saharan) small western hills (Tabag, Abu Jinūk) (b) Dilling, western Kadaru, Karko, Wali 10. Daju Eastern Sudanic Eastern Sudanic (a) Daju of western Kordofan (near Lagowa) (Nilo-Saharan) (b) Liguri (c) Shatt 28 CHAPTER 3

Dimmendaal (To appear b) gives a lexical and grammatical reconstruction of the Katloid cluster and compares the results with data from the Rashad group. The same contribution adduces further evidence against Kordofanian as a genetic grouping. Instead, he argues that Katloid-Rashad is more closely related to Benue-Congo.

map 4 Genetic diversity in the Nuba Mountains Accretion Zones and the Absence of Language Union 29

The genetic position of another group of languages classified as Kordofanian by Greenberg (1963), the Lafofa group (consisting of Tekeim, and Al Amira), can only be clarified once more detailed grammatical data become available. It may either be most closely related to the Talodi group or form an independent branch in a genetic grouping with Talodi and Heiban. There is general consensus amongst scholars that language groups such as Nubian as well as Nyimang and Afitti belong to the Eastern Sudanic branch within Nilo-Saharan. In his detailed analysis of Meroitic, Rilly (2011) concludes that this extinct language of the ancient Meroitic kingdom belongs to a subgroup within Eastern Sudanic called Northern Eastern Sudanic, which also includes the Taman group (in Sudan and Chad), Nara (in Eritrea), the above- mentioned Nubian languages, as well as Nyimang and Afitti. The other Eastern Sudanic language groups represented in the Nuba Mountains, Temein, Keiga and Tese (Teisei-Um-Danab), referred to by the name Temeinian in the present study, are more closely related to Daju, Nilotic and Surmic, with whom they form the Southern branch of Eastern Sudanic (Dimmendaal 2007: 57). One piece of evidence comes from the irregular (suppletive) alternation for ‘cow’ attested in all four groups:

Singular Plural (1) Daju (Lagowa) teɲe tukke ‘cow’ Temein ntɛŋ kituk ‘cow’ Proto-Nilotic d̪ɛŋ d̪ʊk ‘cow’ Majang (Surmic) taŋ togi ‘cow’

This shared lexical innovation fits in with other typological innovations shared by these language groups, such as a Verb-Object constituent order, in contrast to Object-Verb order in Northern members of Eastern Sudanic. There is additional evidence that the Southern group within Eastern Sudanic forms a genetic unit with the Jebel languages, for example the shared innovation of the post- verbal case markers for A-roles (subjects) *-ɛ and *-i. These Ergative case markers are not only attested in Surmic and Nilotic, but also in one other Eastern Sudanic group, Jebel, more specifically in the only Jebel language still spoken today, Gaahmg (Dimmendaal 2014a, Stirtz 2014). Consequently, these case markers must go back to their latest common ancestor, called Southeastern Eastern Sudanic in Dimmendaal (2014a), who arrives at the following subclassification for Eastern Sudanic languages: 30 CHAPTER 3

Taman

Meroitic

Northern Nubian

Nara

Nyimang, A tti Eastern Sudanic

Jebel

Southeastern Daju

Temeinian

Southern Surmic

Nilotic figure 1 Subclassification of Eastern Sudanic

There is no evidence for these Ergative case markers in the Daju cluster or in Temeinian. But as these subject markers only occur in the Southeastern Eastern Sudanic languages that allow for postverbal subjects (Gaahmg and a range of Nilotic and Surmic languages), and as languages belonging to the Daju and Temeinian group do not allow for postverbal subjects, they consequently must have lost these markers. Preverbal subjects and corresponding absence of case marking is also found in some Nilotic languages, e.g. Eastern Nilotic Bari or Western Nilotic Luo, both of which have generalised svo constituent order. Whereas the majority of Daju languages are spoken in Chad and Western Sudan, there are three in the Nuba Mountains: Shatt and Liguri, which belong to Eastern Daju, and Lagowa, which belongs to the other primary branch, Western Daju, according to Stevenson (1956–57: 112). Thelwall (1981) argues that Daju of Lagowa is most closely related to Nyalgulgule in southwestern Sudan. Whereas the Temeinian and Daju languages listed above became part of an accretion zone in the Nuba Mountains, Nilotic and Surmic developed into expansion or spread zones, as Map 2 (p. 18) illustrates. Their rapid spreading (all the way into central Tanzania) is most likely due to the important role played Accretion Zones and the Absence of Language Union 31 by pastoralism and the search for “cool grounds” over the past millennia in these communities south and southeast of the Nuba Mountains. Nubian languages are spoken over a vast area, stretching in discontinu- ous locations from southern Egypt and northern Sudan (Nobiin, Kenuzi- Dongolawi, mainly along the Nile) into Darfur (Midob, Birgid) and the Nuba Mountains, where speakers of so-called Kordofanian Nubian languages commonly refer to themselves as Ajang. Rilly (2010: 164–165) gives the following list for varieties of Kordofanian Nubian:

table 2 Kordofanian Nubian Languages

Dair Dilling Tagle Kasha Kadoro Karko Koldegi Fanda Dabatna Kujuria Habila Wali Ghulfan Tabag Debri Abu-Jinuk Kudur El-Hugeirat

As this brief introductory survey should have made clear, there is considerable genetic variation in the Nuba Mountain area. Moreover, there is very little evidence for so-called “language union” in the sense of Trubetzkoy (1923: 116, quoted in Toman 1995: 204):

[. . . L]anguages in a region defined in terms of geography and cultural history acquire features of a particular congruence, irrespective of whether this congruence is determined by common origin or only by a prolonged proximity in time and parallel development. We propose the term language union (jazykovyj sojuz) for such groups which are not based on the genetic principle.

As shown next, the genetic diversity described for the Nuba Mountains above corresponds to an equally high degree of structural (typological) diversity and lack of language union in the sense of Trubetzkoy. 32 CHAPTER 3

3.2 Typological Diversity

3.2.1 Phonology Most languages spoken in the Nuba Mountains are tone languages. However, as this is the norm in sub-Saharan languages (except for certain geographically peripheral zones mainly along the West African and East African coast), this feature is not particularly significant from an areal point of view. See Wedekind (1985) and Clements and Rialland (2007: 70–74) for a discussion of this phonological feature of African languages. Interestingly, a number of Daju languages spoken in the Nuba Mountains appear to have tonal accent (as do Cushitic languages like Somali). Ismail (2007) argues that there is stress in Daju of Lagowa, a claim that does not tally with my own unpublished data. According to my data, there appears to be one prominent high peak (involving pitch rather than intensity) per word in most cases. But as the following examples show, which particular syllable in a word carries a high tone is unpredictable:

(2) ɛ́ɛ́gɛ̀ ‘giraffe’ wànɛ́ ‘body, skin’

There may be certain restrictions on tone melodies accompanying number inflection with nouns or verbal paradigms, but again the position for high tone assignment (essentially one per word) appears to be largely unpredictable in Daju of Lagowa:

(3) ààgɪ́ á-ɗòhó tɨ́k 1sg 1sg-go today ‘I am going today’

(4) ììgí ɗó↓hó tɨ́k 2sg 2:go today ‘you are going today’

As example (4) shows, a low tone may assimilate to the height of the following high tone (as in a classical two-tone system with automatic and non-automatic downstep). Moreover, more than one high tone may occur on a single word, under conditions not yet understood, as example (5) shows:

(5) àp ɗóhó tɨ́k 3sg go today ‘(s)he is going today’ Accretion Zones and the Absence of Language Union 33

Boyeldieu (2009: 13), in his description of the Daju language Shatt (spoken south of Kadugli in the Nuba Mountains) also observes: “High pitch does not normally appear more than once in a single word. Furthermore, its position is not invariant in all morphosyntactic contexts.” In this respect, Shatt appears to be similar to Daju of Lagowa. Alamin Mubarak (2006, 2013a) describes Daju of Laggori as a language with stress rather than pitch accent or tonal distinctions. Although this possibility cannot be excluded, a more detailed phonetic and phonological investigation of the prosodic system would seem to be in order. The actual tone systems in the Nuba Mountains vary even between neighbouring languages. For example, the Niger-Congo language Tima, which is spoken northeast of the Daju of Lagowa area, has a two-tone system with automatic and non-automatic downstep. Example (6) shows a tone minimal triplet in Tima:

(6) kʊ́dà ‘python’ kʊ̀dà ‘shoe’ kʊ̀dá ‘tree species’

Neighbouring Nubian (Nilo-Saharan) languages such as Tabaq, Wali or Uncunwee (Ghulfan) on the other hand distinguish between three contrastive level tones, and do not appear to have either automatic or non-automatic downstep. Example (7) from Uncunwee (Comfort 2013) illustrates this:

(7) ātɛ́ ‘large water pot’ ɔ́tɪ̀ ‘water’ kàrɛ́ ‘sorghum flour’

Similar variation between neighbouring languages can be observed for other parts of the Nuba Mountains. The degree of typological variation corresponds to what can be observed for sub-Saharan Africa as a whole, with the exception that so far no languages have been found in the Nuba Mountains with four or five contrastive level tones. A further interesting phenomenon, again attested in numerous African tone languages, is upstep or key raising, for example in the Talodi language Dagik (John Vanderelst, personal communication) or the Katloid language Tima (Gertrud Schneider-Blum, personal communication). The conditioning for this phenomenon awaits further investigation. 34 CHAPTER 3

Similar observations in terms of typological variation can be made with respect to segmental structures. In the Niger-Congo language Tima, for example, we find twelve vowels, which can be divided into [-atr] and [+atr] sets:2

table 3 The vowel system of Tima

[+atr] vowel set [-atr] vowel set i ɨ u ɪ ɘ ʊ e o ɛ ɔ ʌ a

Tima has a classical system of cross-height vowel harmony, involving the following alternations: ɪ ~ i, ɛ ~ e, a ~ ʌ, ɔ ~ o and ʊ ~ u. This pattern is widespread in a range of Niger-Congo, Nilo-Saharan and Afroasiatic languages from Senegal to Ethiopia. In addition, however, Tima has two high central vowels (ɘ ~ ɨ), which also participate in vowel harmony rules. atr harmony systems are also found in Nubian, Nyimang, Temein and Kadu. Given the widespread presence of atr harmony across sub-Saharan Africa (Dimmendaal 2001), its absence in several languages in the Nuba Mountains (e.g. in Daju languages) becomes even more significant; see Map 5 for the areal distribution of atr harmony. So far, Tima appears to be the only language in the area with an additional system of fronting harmony, whereby underspecified vowels in affixes copy their features front/central/back from the first vowel in the adjacent root, as example (8) illustrates (See Bashir 2010 for further details):

Singular Plural (8) kì-híì ì-híì ‘needle’ kɪ̀-dɪ́ɪ́ ɪ̀-dɪ́ɪ́ ‘leg’ kɨ́-bʌ́ʌ́ŋ í-bʌ́ʌ́ŋ ‘friend’ kɘ̀-ráŋ ɪ̀-ráŋ ‘kind of grass’ kʊ̀-bɔ́ŋ ɪ̀-bɔ́ŋ ‘bracelet’ kù-hú ì-hú ‘giraffe’

2 The atr symbols used in this monograph are those commonly used by Africanists in to represent these vowels, rather than the ipa symbols. Accretion Zones and the Absence of Language Union 35

map 5 Vowel systems in Nuba Mountain Languages

Non-low central vowels are more common in Niger-Congo than in the other phyla represented in the Nuba Mountains, as the Tima examples above illustrate. Stevenson (2009) gives examples from Tira and Otoro, which belong to the same group as Koalib (Heiban).3 Outside the Nuba Mountain area, such

3 Schadeberg (1981b: 132) reconstructs a seven-vowel system, with *i, *ɪ, *e, *a, *o, *ʊ and *u, and with contrastive vowel length for Proto-Heiban. In other words, he does not reconstruct high or mid-central vowels, neither does he reconstruct such vowels for Proto-Talodi. For the common ancestor of the latter group, Schadeberg (1981c: 94) reconstructs five short vowels 36 CHAPTER 3 central vowels are common in Adamawa languages in Chad and Nigeria, but also in Gur languages like Lama (mainly in Togo), for example. These parallel systems may therefore be reminiscent of former language contact in that area. Quint (2006: 31–40) shows that Koalib, a member of the Heiban group, has eight vowel phonemes.

table 4 The vowel system of Koalib

front central back closed i u mid e ɐ o open ɛ a ɔ

Quint (2006) shows that Koalib has fronting assimilation of affixes, resulting in alternations between /e/ and /o/ depending on whether a front vowel or back vowel occurs in lexical roots to which affixes are attached. This type of fronting assimilation is also found outside the Nuba Mountain area, for example in the Eastern Sudanic language Tama, spoken in the border area between Sudan and Chad (Dimmendaal 2009c: 310):

(9) ŋáy-ɪ́r ‘the back’ sʊ́ʊ́t-ʊ́r ‘the wind’

Similar phenomena in Heiban again may be reflexes of much earlier areal contacts which occurred outside the Nuba Mountains, more specifically in zones west of the Nuba Mountains. In his description of the phonology of Koalib, Quint (2006: 36ff) also demonstrates height assimilation in affixes, as illustrated by example (10):

(10) kwáarál-kè ‘with the antelope’ kwɐ̀ɐlúŋ-kì ‘with the liar’

(*i, *e, *a, *o, *u) and six long vowels (*ii, *ɪɪ, *ee, *aa, *oo, *uu). In other words, for neither of the two groups are high or mid-central vowels reconstructed. It should be kept in mind, however, that these reconstructions are based on limited datasets. Accretion Zones and the Absence of Language Union 37

Height assimilation again is also found in languages spoken outside the Nuba Mountains. Stirtz (2012: 56–59) describes such a system for the Eastern Sudanic language Gaahmg, spoken in an area east of the Nuba Mountains (see Map 2, p. 18). According to Stirtz (2012: 33), Gaahmg has six vowel phonemes. He further argues that the position of the tongue root is one of the distinctive features controlling the distribution of vowels within a word.

table 5 The vowel system of Gaahgm

[-round] [+round] [-back] [+back] [+atr] i ǝ u [-atr] ɛ a ɔ

“Only vowels with the same [atr] value occur together in the same root. Across morpheme boundaries in the same word, [+atr] quality spreads to all vowels unspecified for [atr], either from root to bound morpheme or from bound morpheme to root, whereas [-atr] quality never spreads” (Stirtz 2012: 33). Consequently, vowels in morphemes alternate between ɛ and ɪ, a and ǝ, or ɔ and u, as shown in example (11):

Singular Plural (11) ún-g ún-íígg ‘tear’ mɔ̀ggɔ̀r-d̪ mɔ̀ggɔ̀r-ɛ̄ɛ̄gg ‘cane’ cɛ́ld cɛ́ld-āgg ‘local broom’ mīīd̪ mīīd̪-ə́gg ‘stone’ kɔ̄rd̪ kɔrd̪-ɔgg ‘bird type’ cūld̪ cúld̪-ūgg ‘birth sack’

Whether manipulation of the tongue root does indeed play a role when producing these vowels in Gaahmg cannot be ascertained without instrumental phonetic research. Interestingly, Clements and Rialland (2008: 53) make similar observations for vowel systems in typologically similar languages elsewhere in Africa: “Whether the operative feature in such systems is [atr] or a feature of relative vowel height remains a matter of debate.” They also refer to Maddieson (2003), who provides phonetic evidence that both types of systems may be present among Bantu languages. 38 CHAPTER 3

Alamin Mubarak (2006) describes Laggori Daju as a language with six vowels /i, e, a, ǝ, o, u/, although in her examples an additional vowel, ɛ, occurs. Boyeldieu (2009) shows that Shatt, another Daju language in the Nuba Mountains, has the following vowel distinctions:

table 6 The vowel system of Shatt

[-round] [+round] [-back] [+back] [+high] i / ii ɨ u / uu [-high] e / ee a / aa o/ oo

Daju of Lagowa on the other hand has ten, or possibly eleven vowels with atr harmony according to my own (unpublished) data. It is not yet clear whether there is an atr distinction for the high central vowel.

table 7 The vowel system of Lagowa

[+atr] vowel set [-atr] vowel set i ɨ u ɪ (ɘ) ʊ e o ɛ ɔ a

Example (12) illustrates this:

(12) íyyè ‘meat’ ɲɪ́rt̪ɛ̀ ‘tooth’ ɛ́ɛ́gɛ̀ ‘giraffe’ èègínjè ‘giraffes’ àkkâŋ ‘rib’ cɨ́mɨ̀rɛ̀ ‘chest’ ɔ́ŋɔ́nɛ̀ ‘eye’ kód̪òs ‘three’ wʊ́rɛ̀ ‘woman’ m(w)únè ‘nose’ Accretion Zones and the Absence of Language Union 39

The vowel systems of the other two Daju languages spoken in the Nuba Mountains, Laggori and Shatt, thus differ from Daju of Lagowa, but are similar to other varieties of Daju spoken west of the Nuba Mountains which do not have atr harmony either. For a recent study of such a Daju language in Chad see Palayer (2011) on Dadjo D’Eref. Consonantal systems also vary considerably between languages in the Nuba Mountains, and again there is little evidence for areal convergence in this respect. Whereas it is common for them to distinguish between dental and alveolar (or postalveolar) stops, this distinction again is also attested in numerous languages east and southeast of the Nuba Mountains; it is not a feature unique to the area. There is one language, Katla, with labio-velar (or labial-velar) stops, /kp/ and /gb/. These of course are common in Niger-Congo and Nilo-Saharan languages (as well as a few Chadic languages) west and southwest of the Nuba Mountains. However, Katla’s close relative Tima does not have these stops. A comparison with cognates in Katla shows that this is due to an innovation in Tima (Dimmendaal, to appear b) whereby */kp/ and */gb/ shifted to and merged with /k/ or weakened to /y ~ w ~ h/, depending on its position in the word. The following cognate forms illustrate the shift */g/ > ø in Tima (tonal pattern for Katla examples not known):

Katla Tima (13) -ɪgɪt̪ -ɪ́ɪ̀(ð) ‘eye’ -ògíl -éèl ‘buy’ -aga -áá ‘porcupine’ -agam -áàm ‘hair’

In prefixes, */g/ and */gb/ shifted to and merged with */k/, whereas intervo- calically lenition occurred, resulting in an approximant /w ~ y/ or loss of this voiced stop depending on the environment in Tima:

Katla Tima (14) g-ògùnèn k-ʌ̀wúnèn ‘woman’ g-agam k-áàm ‘hair’ g-àtà k-átáh ‘leaf’ gb-ɔlɔ k-ʌ́(ʌ́)lù ‘mud’ gb-àjà k-ʊ̀dà ‘shoe’

The loss of the labial-velar stop in Tima may indeed be an instance of areal adaptation, since no other language in the Nuba Mountains apart from Katla has labial-velar stops, as far as we know. As argued in the next chapter, the 40 CHAPTER 3 range of clan names which Tima shares with neighbouring groups suggests that its speech community emerged through the fusion of various groups probably speaking different languages at an earlier point. This process in turn may have resulted in the loss of consonants that are marked from an areal point of view.

3.2.2 Lexical Diffusion The languages of the Nuba Mountains are among the least studied on the African continent. This paucity of information affects our knowledge of their lexical structure. Black and Black’s (1971) Moro dictionary remains one of the few lexical sources to date, and Schneider-Blum’s (2013a) multi-media Tima dictionary is the only modern dictionary of a Nuba Mountain language so far. In a fascinating contribution on the lexical structure of Tima, Schneider- Blum (2012) shows that metaphors and metonymic extensions constitute prominent lexical strategies. Whether this property extends to other languages in the Nuba Mountains can only be determined once similar dictionaries become available for them. The cultural background to these strategies in Tima is taken up again in Chapter 10, as part of a discussion of so-called esoteric and exoteric languages. Another fascinating property of the Tima lexicon concerns the structure of different semantic fields, as shown in Chapter 4. One other typologically interesting idiosyncrasy of the Tima lexicon, for example, is the possibility of grada- tion between the two antonyms ‘empty’ and ‘full’:

(15) -hʊ̀waŋ́ ‘empty (also: dry)’ -cìɽìŋkìŋ ‘less than half full’ -rɨ̀mʌ́n ‘half full’ -mʌ̀ʔʔɨ̀r ‘almost full’4 -bɪ̀lɪ̀bɪ̀lɪ̀k ‘full’

Hayward (1991) has shown that Semitic, Cushitic and Omotic languages in Ethiopia share patterns of polysemy, for example the extension of ‘narrow, constricted’ to ‘in difficulties’. These convergence phenomena at the lexical level correspond to a range of phonological, morphosyntactic and pragmatic features shared by these languages. Whether anything similar can be observed for the Nuba Mountain area can only be determined once more extensive lexical material becomes available. None of the phonological phenomena discussed above nor the morphological and syntactic phenomena discussed

4 The alternative lexeme for ‘almost full’ is -rɔ̂y, or the more archaic form -rɔ̂ð. Accretion Zones and the Absence of Language Union 41 below, however, are indicative of a Nuba Mountains convergence zone paralleling the Ethiopian area.5 Lexical borrowing between languages in the Nuba Mountains does not appear to be very prominent either. In Tima, for example, one can identify a number of lexical items shared with the neighbouring Kadu language Tulishi, but this appears to involve the usual borrowing of non-basic vocabulary, as with the following example:6

Tima Tulishi (16) pampaŋ pampa ‘drum’

There is no indication from oral histories or other sources that people in the Nuba Mountain area try to avoid lexical borrowing. Aikhenvald (1996) has described such attitudes for the Vaupes region in Brazil, where this tendency to keep languages apart is related to the practice of exogamy, whereby females come from another speech community. As endogamy is the more common pattern in the Nuba Mountains, resistance to lexical borrowing is not necessarily expected there. There is some evidence for ancient lexical borrowings between Niger-Congo languages spoken in the Nuba Mountains today and Nilo-Saharan languages outside the Nuba Mountain area, for example the widespread Northeastern Nilo-Saharan root for ‘elephant’, ɔŋ(ŋ)ɔr, as discussed in Chapter 2. Schadeberg and Blench (2013: 6) point towards additional widespread words in the Nuba Mountains such as the lexemes for ‘pig’ and ‘horse’. But it should be pointed out that the same “Wanderwörter” (wandering words) are attested in various languages outside the Nuba Mountains, as the following example for ‘pig’ shows (adapted from Blench 2006: 267):

(17) kudur Nyimang (Eastern Sudanic; Nuba Mountains) kudur Temein (Eastern Sudanic; Nuba Mountains) kudur Otoro (Niger-Congo; Nuba Mountains) kadruuk Sudan Arabic (Afroasiatic) kuturu Anej (Koman; Ethiopia) kutun Old Nobiin (Eastern Sudanic; Sudan) kutsū Nupe (Niger-Congo; Nigeria)

5 Of course, this convergence does not affect all languages spoken in Ethiopia. Languages spoken in the western and southwestern zones of the country have not been affected by these processes of areal diffusion. 6 Data from Schneider-Blum 2013 and http://www.rogerblench.info/Language/NiloSaharan/ Kadu/Kadu%20composite%20datasheets.pdf. 42 CHAPTER 3

Consequently, their existence in languages of the Nuba Mountains again is not indicative of diffusion just within this area. A detailed comparison of kinship terminology on the other hand may lead to interesting historical evidence for contact between specific groups in the Nuba Mountains. But again, here too lexical evidence points towards contact with languages outside the area. One such example is a widespread Nilo- Saharan root for ‘maternal uncle’, which is also found in at least one Niger- Congo language in the Nuba Mountains, Tima:

(18) mʌ́múŋ ‘maternal uncle’ (Tima, Niger-Congo) mamaɪ ‘maternal uncle’ (Teso, Nilotic, Nilo-Saharan) mamai ‘maternal uncle’ (Murle, Surmic, Nilo-Saharan) maama ‘maternal uncle’ (Fur, Nilo-Saharan)

There is a widespread tradition, in particular among pastoral groups in Eastern Africa speaking Eastern Sudanic languages, of extracting the lower front teeth as a sign of beauty. A similar tradition is attested among the Tima in the Nuba Mountains, who refer to this act as kɨ̀ɨ̀hɨ́l; the same word also means ‘peeling (of sugarcane), stripping off’. Again, this presumably reflects ancient cultural contacts between these Nilo-Saharan and Niger-Congo groups in Eastern Africa (and not necessarily in the Nuba Mountain area). As argued below, there is some morphological evidence for areal contact between speakers of these languages, but on the whole the evidence is rather limited.

3.2.3 Morphological Diversity The phonological diversity of the Nuba Mountain languages is paralleled by morphosyntactic diversity. Let us take nominal morphology as an example. Number marking systems for nouns differ widely, although there is some evidence for areal diffusion of specific features, as the following examples illustrate. Nubian languages in the Nuba Mountains have retained the tripartite number system that is characteristic of Northeastern Nilo-Saharan, i.e. the Nilo- Saharan group that forms a coordinate, primary branch with Central-Sudanic, where such number-marking systems are absent (Dimmendaal 2000, 2010). Data from the Kordofan Nubian language Uncunwee (Ghulfan) illustrate a system whereby nouns are morphologically unmarked either in the singular (with plural inflection), or in the plural or collective (with corresponding singulative marking), or both singular and plural are marked (Jakobi and Williams 2007): Accretion Zones and the Absence of Language Union 43

Singular Plural (19) kà-tʊ̀ kà-ní ‘field’ ɔ̀gʊ̀r ɔ̀gìr-í ‘shoulder’ kɔ̀ɲɔ̀l-tʊ̀ kɔ̀ɲɔ̀l ‘egg’

This three-way number marking system has also been retained in a number of Daju languages in the Nuba Mountains, for example in Shatt (Boyeldieu 2009):

(20) ɲèɲémès-ìc ɲèɲémès ‘star’ tùmòs tùmós-ìɲ ‘elephant’ kùmù-x kùmù-ɲ ‘antelope sp.’

The same system occurs in the Daju language Liguri (Alamin Mubarak 2006):

Singular Plural (21) ŋoŋŋi-s ŋoŋŋi ‘bone’ ba ba-tta ‘house’ kamlakk-a kamlakk-uda ‘camel’ píib-ì píib-u ‘child’

However, there is no trace of this tripartite system in Daju of Lagowa, which only has plural marking. The other member of Southern Eastern Sudanic, Temein, again turns out to be conservative in terms of number marking for nouns in that it has retained the three-way division for number marking. But apart from this suffixation system, it also uses a plural-marking prefix with nouns (Lorenz 2013):

Singular Plural (22) jɛ́ʈ-ɪ̀ʈ kɪ́-jɛ̀ʈ ‘hair’ ʈúúkúr-ìs ku-túúkúr-i ‘ankle’ wór k-wòr-áɁ ‘shield’

A similar strategy with a plural marking prefix ka- (rather than ku-) is found in Western Nilotic languages like Belanda Boor, spoken southeast of the Nuba Mountains (von Heyking 2013).

Singular Plural (23) rɪ́ŋɔ́ ká-rɪ́ŋɔ́ ‘meat’ fíɲ ká-fíɲ ‘place’ 44 CHAPTER 3

The origin of this typological similarity remains to be clarified, but the prefix- ation of a plural number suffix /kv-/ cannot yet be simply attributed to areal diffusion, as Temein and its closest relatives Keiga Jirru and Tese appear to be the only languages with this feature within the Nuba Mountains. See Blench (2013) for a grammatical overview of this cluster. Interestingly, number marking on nouns is virtually absent in one other Northern member of Eastern Sudanic spoken in the Nuba Mountains, Nyimang. In line with more universal tendencies, following Corbett’s (2000) animacy hierarchy, Nyimang distinguishes number in independent pronouns and nouns denoting persons, but not in inanimate nouns, for example (Tucker and Bryan 1966: 246):

Singular Plural (24) mɔr mɔr-gi ‘friend’ yɔŋ nyi ‘you’

Nyimang uses distinct singular and plural verb stems, i.e. pluractionality as well as dual marking (Tucker and Bryan 1966: 247), but this latter phenomenon is attested as an areal feature in different languages in Northeastern Africa (Storch and Dimmendaal 2014), and therefore again it is not indicative of convergence in the Nuba Mountains. This also applies to the morphological distinction between inclusive and exclusive pronouns, which is found in a range of languages in the Nuba Mountains as well as outside the area. Niger-Congo languages in the Nuba Mountains have all retained traces of the proto-typical noun-class system found across this phylum. However, in the Katloid group and in particular in Tima this system has been strongly reduced (Dimmendaal 2014a). Indeed, the only productive singular/plural (or singulative/collective) alternation found in Tima is between a kV- (with a vowel whose qualities are determined by the following root vowel) and ɪ-/i-. The same morphemes occur as agreement markers on dependent categories such as adjectives or numerals.

Singular Plural (25) kɨ́-bʌ́ʌ́ŋ í-bʌ́ʌ́ŋ ‘friend’ kɘ̀-rá ɪ̀-rá ‘kind of grass’ kɪ̀-dɛ́k ɪ̀-dɛ́k ‘neck’ kù-hú ì-hú ‘giraffe’ kʊ̀-bɔ́ŋ ɪ̀-bɔ́ŋ ‘bracelet’

Stevenson (1956–1957) observes that Niger-Congo (that he terms “Kordofanian”) languages belonging to the Rashad group not only have noun-class prefixes, but Accretion Zones and the Absence of Language Union 45 also plural suffixes. This system is particularly prominent in Tagoy. Schadeberg (2013: 330–331) presents additional examples with number suffixation in the Rashad group. The common ancestor of the Katloid group also has remnants of such a system of number suffixation. Compare the following cognates for ‘female breast’, where the reflex in Tima contains a final vowel going back to a number suffix (with the loss of the final /t/ in this language being regular):

Singular Plural (26) ti-min nyi-mind-it ‘female breast’ (Moreb) ti-miny i-miny-it ‘female breast’ (Tagoi) ki-mindi i-mindi ‘female breast’ (Tima)

Number marking by way of suffixes is characteristic of Eastern Sudanic (Nilo- Saharan) languages. Hence, this additional system in two Niger-Congo groups (Rashad and Katloid) may be reminiscent of areal contact at some point in the past. As illustrated below, case marking in the Niger-Congo group Heiban is indicative of contact with Eastern Sudanic (more specifically Nubian) languages. But with just these two instances of areal contact, one has to conclude that areal diffusion plays hardly any role in morphosyntax. Moreover, these Niger-Congo languages have not abandoned their system of noun-class prefix- ation. Also, as we shall see below, Accusative case marking in Heiban (Niger- Congo) and Kordofan Nubian developed in slightly different directions. In the third language family found in the Nuba Mountains, Kadu(gli), number marking is expressed by way of prefixes. Contrary to the system found in Niger-Congo languages in the area, however, agreement marking on nominal modifiers is based on the inherent gender of the head noun (Reh 1985: 249–261):

(27) bìitì ŋ-álímì ‘cold water’ water agr:m-ipfv:be.cold

(28) mòtò m-àdéelá ‘good labour’ labour agr:f-ipvf:be.good

3.2.4 Clause Structure A typological model proposed by Nichols (1986) classifies languages into four types based on the formal expression of syntactic (or semantic) relations at the clausal level. In a dependent-marking language, such relations are expressed on (dependent) arguments of a verb, e.g. by way of case markers. In a head- marking language, such relations are expressed morphologically on the verb. Alternatively, languages may combine head marking and dependent marking, or both strategies may be absent i.e. zero marking may occur. 46 CHAPTER 3

When comparing languages in the Nuba Mountains from this perspective, again one observes dramatic differences both between related und unrelated languages. Take the Nilo-Saharan languages in the area for example. Nubian languages manifest a high degree of dependent marking at the clausal level, more specifically between a morphologically unmarked Nominative, Accusative case and several peripheral case distinctions, e.g. for Location or Instrument. Jakobi (To appear) shows that Accusative case marking in Uncunwee follows more general principles of “Differential Object Marking”. Thus, object nouns are marked for Accusative case if they are definite, animate, or if they occupy a prominent status in discourse, e.g. when carrying focus. (See also Dimmendaal 2011 for a discussion of Differential Object Marking in Nilo-Saharan). Jakobi (To appear) also shows that there is an interesting exception in Uncunwee to a more general Animacy Hierarchy proposed, for example, by Aissen (2003), in that pronominal objects only receive Accusative case in this language when carrying focus.

(29) yě à ɖʊ́k-kɛ̀rɛ̀ 1sg 2sg:acc beat-fut.1sg ‘I will beat you’

(30) yě à-gɪ̀ ɖʊ́k-kɛ̀rɛ̀ 1sg 2sg-acc beat-fut.1sg ‘I will beat you (picked out of a group of people)’

According to Aissen (2003), object pronouns are at the top of the hierarchy, and always receive Accusative case in languages with Differential Object Marking. The data from Uncunwee contradict this claim. Nubian languages are strongly dependent marking at the clausal level. Interestingly, the Dative (covering semantic roles such as Benefactive or Malefactive) is marked on the verb by way of a morpheme derived from the verb ‘give’ rather than by way of a case marker (Comfort and Jakobi 2011). Ibrahim and Jakobi (2013) present parallel examples from Taglenna:

(31) ɪdʊ id=gi kamɛ = gɪ kiŋ-g-ʊmin woman man=ACC food = acc make-appl-PAST.3sg ‘The woman prepared food for the man.’

In related languages like Nyimang we find a certain degree of head marking on the verb, e.g. for Directional and Benefactive or Applicative roles. Example (32) illustrates this (Fiedler and Berlin 2013: 440): Accretion Zones and the Absence of Language Union 47

(32) ɛ̄ lé-ɔ̀ ànī lād̪ī, kwoní kàdōŋ-ōŋ àrān and when 3pl walk.ipfv animals man-gen near t̪à-ān go.pfv-ben ‘and while they were walking, a lot of animals came near’

Unlike Nubian languages and other members of Northern Eastern Sudanic, however, Nyimang does not mark pronominal subjects on the verb. Instead, it often uses “suppletive” verb stems for singular and plural, as with ‘put’ (singular bi, plural áŋ(g)-) (Tucker and Bryan 1966: 247). Moreover, Nyimang also distinguishes between singular, plural and dual objects by way of different suffixes on the verb, as with the verb ‘eat’ in example (33) (Tucker and Bryan 1966: 250).

(33) țal (singular subject, singular object or no object) țal-ɪn (dual subject, singular object or no object) țal-d̦i (plural subject, singular object or no object) țal-d̦-ɪn (dual subject, plural object)

Whereas Niger-Congo languages do not generally use case marking as a dependent-marking strategy at the clausal level, there is one interesting exception: case marking, specifically for objects, is a well-established property of the Heiban group, as example (34) illustrates (Ackerman and Moore 2013: 89):

(34) í-g-ʌləŋ-í kúku-ŋ 1sg.sm-clg-sing-caus.pfv clg. Kuku-acc ‘I made Kuku sing’

This system of Accusative case marking parallels the system found in Nubian languages in the Nuba Mountains. If case marking in the Heiban group is indeed the result of areal influence from Nubian (or some other Nilo-Saharan branch), this influence must be old, leaving enough time for the Heiban languages to develop in a different direction typologically. Contrary to Kordofan Nubian languages, Accusative case marking in Heiban appears to be obligatory, i.e. it is not subject to Differential Object Marking principles. This constellation again provides evidence of former language contact, as with number suffixation in these Niger-Congo languages (discussed above), but that this convergence was interrupted fairly rapidly. There is no evidence for deep language contact or areal diffusion (as an exponent of long-term multilingualism), for example when comparing constituent order. Whereas these Nubian languages 48 CHAPTER 3 are verb-final, the Niger-Congo languages belonging to the Heiban group have svo constituent order. Example (35) is from Moro (Rose 2013: 40):7

(35) um:iə g-a-land̪-ó ʌurí boy sm.cl-rtc-close-pfv door ‘the boy closed the door’

The system in the other Niger-Congo branches in the Nuba Mountains again differs from that found in Heiban. There is no evidence for case marking in the Talodi group and the Rashad group, which tend to be strongly head marking at the clausal level. Schadeberg (2013) gives examples from one Rashad language, Tagom (example (36)) suggesting that sov constituent order is common in this language.

(36) ŋindá ŋgər kɛ́-más they eye they-rub ‘they rubbed their eyes’

In the Katloid group, there is again an interesting case phenomenon whose origin nevertheless remains enigmatic. One member of this cluster, Tima, has a split ergative system (Dimmendaal 2010) whereby postverbal subjects of transitive predications take a proclitic (homorganic) N- marker in the case of nouns (or noun phrases). Example (37) is from Schneider-Blum (2013: 292):

(37) cɛ́-dàh-ɪ̀ ŋ̀-kànʊ́ŋ cɛ́-yɛ̀ táàn-ŋùŋ pɨ́nʌ̀ per-say-tr erg-Kanung per-rep beat-log3sg pron3sg ‘Kanungi said that (s)hej beat him/herk’

This element fuses with pronominal subjects and encliticises onto the verb (Schneider-Blum 2013):8

7 Contrary to proper names, common nouns functioning as objects do not require an Accusative case marker in Moro. 8 So far, Tima is one of only two languages in the Nuba Mountains which have been shown to have logophoric pronouns (Schneider-Blum 2013). Reh (1985: 356–361) describes this phenomenon for the Kadu language Krongo. Again, it should not come as a surprise if additional languages in the Nuba Mountains manifest logophoricity, as this feature is widespread in major parts of sub-Saharan Africa. Accretion Zones and the Absence of Language Union 49

(38) ŋààŋ-á húm-áá-yáŋ-nʌ́ cídʌ̀ 2sg-foc depend-instr-loc3sg/pl-1sg.erg body ‘I depend on you’

This system is similar typologically to case-marking systems found in Eastern Sudanic (Nilo-Saharan) groups east and south of the Nuba Mountains (Dimmendaal 2014a). In a range of Nilotic and Surmic languages and in the Jebel language Gaahmg, postverbal subjects of transitive predications (i.e. postverbal A-roles) are also marked for case. However, Eastern Sudanic languages in the Nuba Mountains like Temeinian and Daju, which belong to the same Southeastern subgroup of Eastern Sudanic, do not allow for postverbal subjects and hence do not have such case-marking strategies. The origin of the ergative system in the Niger-Congo language Tima consequently remains a puzzle. Kadu languages manifest double marking at the clausal level. Reh (1985) provides a detailed account of one such language, Krongo. In this language, extensive case marking (Dative, Locative, Ablative, Instrumental, Comitative, Genitive, Possessive) on noun phrases is combined with head marking on the verb. Apart from subject marking, we find Beneficiary, Directional, Itive, and Ventive marking on the verb. Example (39) is an example of this (Reh 1985: 222):

(39) òrùbò-òkò-ŋ còorì à-káaw y-íkkì inf:build-ben-tr house dat-person M-that ‘to build a/the house for that man there’

The Nuba Mountain area is thus a facsimile of the typological diversity found across Africa as a whole when it comes to head-marking and dependent-marking strategies at the clausal level and in terms of constituent order. The position of the verb varies between verb-initial (as in the Kadu languages), verb-second (as in the Katloid group), and verb-final (as in the Nubian languages). Similarities in constituent order may of course become significant from an areal point of view once they are part of a set of typological features shared between two or more unrelated languages. Constituent order in and of itself is hardly a predictor of other syntactic features in languages generally (Dimmendaal 2008b). Although there is a statistical tendency for verb-final languages to use postpositions rather than prepositions, and for verb-initial languages to use prepositions, there are of course numerous exceptions. Reh (1985: 290), for example, shows that verb-initial Kadu languages like Krongo use postpositions: 50 CHAPTER 3

(40) n-àbà-ŋ nì nkí-mìsì kúbú agr-appear-tr snake loc-stone under ‘The snake appeared from underneath the stone.’

Dimmendaal (1987) reanalyses these location marking elements as postnomi- nal modifiers, but the fact remains that they follow rather than precede the locative noun. An additional parameter of typological variation concerns the syntactic position for the marking of tense, aspect, and person, either on the verb or as a separate auxiliary constituent. Also, the question to what extent peripheral semantic roles such as Instrument, Location, or Direction are expressed on the verb (as a head-marking strategy) or on dependent categories (e.g. by way of case) is typologically interesting. Table 8 summarises variation with respect to these typological properties in the Nuba Mountain area; map 6 shows the areal distribution of constituent order in this area. table 8 Constituent order and head marking in languages of the Nuba Mountains

Genetic affiliation Case marking Constituent order Peripheral role marking on the verb

1. Koalib-Moro Yes svo Dative, Instrumental, Locative 2. Talodi-Mesakin No svo, vso Associative, Dative, Locative 3. Tegali-Tagoi No sov Directional 4. Lafofa No sov (No information) 5. Katloid Ergative marking svo in Katla and Dative, Directional, in Tima Julud; relatively Instrumental free constituent order in Tima 6. Nyimang and Extensive case marking s (aux) ov Dative, Directional Afitti 7. Hill Nubian Extensive case marking sov Dative, Directional 8. Temeinian Remnants of Locative svo Dative, Directional case marking only 9. Daju No case marking svo Directional 10. Kadu Case marking vso, svo Dative, Directional Accretion Zones and the Absence of Language Union 51

map 6 Constituent order in Nuba Mountain languages 52 CHAPTER 3

All Niger-Congo languages in the Nuba Mountains are strongly head-marking at the clausal level, as shown in Table 8, which constitutes a non-exhaustive listing of verbal derivation. The verb in the Talodi language Dagik, for example, potentially carries lexical-derivational markers expressing Passive, Reciprocal, Causative, and Applicative amongst others. But whereas Heiban languages tend to express pronominal subject and object as well as tam markers by way of one phonological and grammatical word, members of the Talodi cluster tend to use either a separate word (or auxiliary element) for the subject and tense-aspect marker (s aux vo), or a verbal enclitic in the perfective and with content questions. Example (41) illustrates the first of these strategies (Vanderelst 2013: 170).

(41) a-ŋ:ɪ b-ɔ nɛmː-ɔ ŋɛr kɜl:-ɛ ref-pr1:1sg cl1-pfv take-fv cl.11.water pour-pr6:1sg ‘I took the water [and after that] I poured [it].’

Example (42) illustates the alternative, with a verbal enclitic (Vanderelst 2013: 166):

(42) mara rɜg-ɔ-ŋɔ̃ Kura. ɪ ŋabɛ how eat-fv-pr2:3sg Kʊraðɪ cl11-fish ‘How did Kuradi eat the fish?’

This inversion as well as the position of auxiliaries in Dagik is reminiscent of “verb second” properties of West Germanic languages like Dutch, English, or German. Constituent order in Tima is strongly conditioned by pragmatic considerations. Whereas preverbal subjects are common, ovs occurs when the object carries focus, as examples (37) and (38) above illustrate. The other two members of the Katloid cluster, Katla and Julud, appear to have generalised svo order. Nyimang (Ama) is strictly verb-final, as example (43) shows (Fiedler 2013: 442):

(43) wìdɛ́ŋ á bóŋ sòbájè child prt water move.ipfv ‘The child is swimming.’

As for narrative discourse structure, the rather limited knowledge we have about Nuba Mountain languages nevertheless allows for some initial observations. Nilo-Saharan languages in the northern part, in particular Nubian Accretion Zones and the Absence of Language Union 53 languages, commonly use converbs plus a final main clause verb in order to enhance the storyline. This property is shared with other Eastern Sudanic and more distantly related Nilo-Saharan languages north, northeast and northwest of the Nuba Mountains. These in turn share this property with Afroasiatic languages in Ethiopia (Dimmendaal 2009c). One detailed account of this phenomenon can be found in Ibrahim Gulfan (2013) on the Kordofan Nubian language Taglennaa, where converbs can be categorised according to the number of subjects involved or the succession of events they denote. The following sentences contrast juxtaposition of two independent utterances with a complex sentence whereby the first verb is rendered as a converb in Taglenaa (Ibrahim Gulfan 2013: 376):

(44) Ahmed kɔyɛ-gɪ kel-un. Ali kal-jɪ kel-un Ahmed meat-acc eat-prs Ali porridge-acc eat-prs ‘Ahmed eats meat. Ali eats porridge.’

(45) Ahmed kɔyɛ-gɪ kel-ndɛ Ali kal-jɪ kel-un Ahmed meat-acc eat-dsc Ali porridge-acc eat-prs ‘Ahmed eats meat and Ali eats porridge’

Interestingly, Taglennaa also manifests verbal compounding, thereby con- firming the claim made in Amha and Dimmendaal (2006) that self-organising principles easily lead to parallel structures in genetically unrelated languages. They show that in the Omotic (Afroasiatic) language Wolaitta of Ethiopia, the juxtaposition of converb plus main verb has resulted in idiomatic verbal compounds. Masica (1976) observes exactly the same phenomenon in South Asia. Amha and Dimmendaal (2006) argue that these parallel structures are due to the facts that no other constituent intervenes between the converb plus main verb, and that the subject is identical. Ibrahim Gulfan (2013: 377) argues that in Taglennaa, a verbal compound consisting of a converb plus main verb “[. . .] describes a single event in which both verbs contribute to the resulting meaning [. . .]”.

(46) id ʃar-gɪ ʃirk-i et-un man things-acc squeeze-conv put.in-pres ‘The man squeezes the thing in.’

(47) ol-i ŋay-i come.out-conv walk-imp ‘Die!’ 54 CHAPTER 3

As a result, these Nubian languages show structural similarities to Omotic languages in Ethiopia, rather than to unrelated neighbouring languages in the Nuba Mountains. Although relatively little is known about narrative discourse structures in Kadu languages, Reh (1985: 193–195, 329–361, 375–399) gives us a typological introduction for one member of this language family, Krongo. In addition to asyndetic coordination, common strategies include the use of a sentential coordinator /ŋ-/, a strategy which is also attested in Niger-Congo languages in the area such as Tima. Again, the variation between different language families in the Nuba Mountains parallels that found in language families outside this area. Longacre (1990) argues that, to a considerable extent, these variant strategies in the rendering of narrative discourse correspond to variant constituent order types found in different parts of Africa. Hence, the Nuba Mountains again constitute a facsimile of different narrative discourse strategies found elsewhere on the continent. The genetic and typological diversity in the Nuba Mountain stands in rather sharp contrast with the surrounding area. To the north and south, mainly Eastern Sudanic languages are spoken. To the east and southeast, there is admittedly a degree of genetic diversity: distantly related Nilo-Saharan languages like Berta and Gaahmg, as well as Koman languages and Gumuz, which Dimmendaal (2011) considers to form an independent family and a linguistic isolate, respectively. But in spite of this, Berta, Gaahmg and Coman languages like Uduk allow for postverbal subjects, which are also marked for case. This property is shared with Nilotic and Surmic languages to the south, and thus constitutes an areal feature (Dimmendaal, 2014b). This again stands in sharp contrast with the dramatic typological diversity observed in the Nuba Mountains. How did this genetic diversity come about in the first place? And why are there so few clear cases of convergence in the area? These questions are central to the following section.

3.3 A Natural Refugium Zone

One of the earliest authors writing extensively about the Nuba Mountains, Nadel (1947), gave the following description of the natural habitat of its inhabitants:

The Nuba Mountains [. . .] form an irregular, broken pattern of long mountain ranges, squat massifs and rugged rocks, separated by broad val- leys and stretches of plain. They are bounded to the east, west and north Accretion Zones and the Absence of Language Union 55

by the semi-arid thin bush country typical of the Sudan in this latitude and rich in the south almost to the marches of the Nile valley. The hill country itself is well watered [. . .]. It also has a much richer vegetation is is more densely wooded than the surrounding plains, although intensive cultivation of the hillsides has resulted in considerable de-forestation, most marked in the neighbourhood of settlements. (Nadel 1947: 1)

As the survey above should have made clear, this attractive niche in the Sudanese landscape can be viewed as an accretion or residual zone linguistically, following the criteria developed by Nichols (1992: 21). Such regions are characterised by language families representing a considerable time depth. As is common with areas where an accretion of languages occurs, there is no traditional or pre-colonial lingua franca in the Nuba Mountains. Whereas Sudanese Arabic is the common language in the area today, this is a fairly recent development dating back only to the 19th century. Instead, local bilingualism was—and to a considerable extent still is—the main means of inter-ethnic communication. The few instances of convergence through areal contact (involving case and number marking with nouns, as illustrated above) suggest a historical scenario of “interrupted convergence”.9 It is possible that additional instances will be identified once more extensive data become available. Indeed, topics such as lexical-semantic fields and speech styles are included in the present monograph (Chapters 7 and 10) with the aim of providing a comparative basis for future research on these domains. In order to understand how this situation came about in the Nuba Mountains, it is necessary to bring in current knowledge about different factors that are known to affect linguistic maps: geography, technology, and social constellations. There is solid palaeogeographical evidence that the areas west, north and east of the Nuba Mountains have been affected by dramatic climatological changes over the past 10,000 years. At the end of the last ice age, global warm- ing resulted in much higher precipitation in much of Africa. One of the consequences was the emergence of a major riverine system connecting the Ennedi Mountains in Chad with the Nile in Sudan. This tributary, which is commonly known as the Wadi Howar or the Yellow Nile (Pachur and Kröpelin 1993), ran

9 As Storch (2006) argues in her study of contact phenomena in Western Nilotic, historically only those linguistic features are prone to diffusion in this family that do not violate emblematic patterns and that can be integrated into the system without altering its most basic structures. In other words, speakers can be selective. 56 CHAPTER 3 eastwards from the mountainous Ennedi region in Eastern Chad and entered the Nile between the third and fourth cataracts. There is archaeological evidence for a continuous occupation of the Wadi Howar area by human beings until it became extinct around 3000 years ago. Desertification forced people to move to wetter regions west, east and south of this former riverine system. The current distribution of Nilo-Saharan, in particular of the Eastern Sudanic sub- branch, reflects this situation (Dimmendaal 2007, to appear a; see Chapter 2). The Nubian group within Eastern Sudanic, for example, is spread over western Sudan, along the Nile into southern Egypt and into the northern part of the Nuba Mountains. Two of the closest relatives are found in eastern Chad and Western Sudan (the Taman group) and Eritrea (the Nara group). Nubian, Taman and Nara form a genetic subgroup together with the extinct Meroitic language, Nyimang and Afitti (Rilly 2004). Meroitic used to be spoken north of the Nuba Mountains in the vicinity of the Nile (see Map 2, p. 18). With respect to Nyimang, Stevenson (1984: 85) observes: “Their genealogies do not reach back to a time before their occupation of the present Nyimang hills, but they have a tradition asserting that in an unknown period before this they migrated from a place ‘in the west’ beyond Tima and Abu Jinūk which they call Kwuja or Kwija.” The Nuba Mountains probably provided an attractive natural refugium during periods of severe drought (of which there have been many) at earlier stages, i.e. before speakers of Niger-Congo or Nilo-Saharan languages moved in. There is a third phylum represented in this area, Kadu, today situated along the southern range of the Nuba Mountains (as shown on Map 3, p. 19). This current situation is due either to newcomers speaking Niger-Congo and Nilo- Saharan languages who forced these groups to move southward, or to a migration of Kadu speakers into the mountains from the south. The Kadu languages all appear to have a verb-initial structure, a typological property also found in different pockets south and east of the Nuba Mountains, occupied by speakers of Surmic, Nilotic and Rub (Kuliak) languages as well as Berta, all of which belong to Nilo-Saharan. Possibly, then, speakers of Kadu languages moved into the Nuba Mountains, where they were not dominated by neighbouring pastoralists who preferred savannah. Today, the Nuba Mountains still constitute a climatic zone with much more rainfall than surrounding areas and permanent water supplies. Some of the language families represented in the Nuba Mountains have relatives in geographically distant regions. The Niger-Congo languages are a case in point: their closest relatives are spoken much further west in Nigeria, Cameroon and Chad. With respect to the Kadu languages, it is not clear whether their Accretion Zones and the Absence of Language Union 57

presence along the southern range of the Nuba Mountains is due to migration from neighbouring savannah regions or not, as stated above. Their speakers may have lived in the Nuba Mountains and may have been pushed south as a result of groups entering from the west and north. Alternatively, speakers of Kadu languages may have migrated into the Nuba Mountains as a result of expanding Eastern Sudanic groups south and southeast of this area. The verb- initial structure of Kadu languages at least suggests that the ancestral speakers may have been in contact with speakers south and east of the Nuba Mountains, where verb-initial languages are common today. This latter typological feature in terms of constituent order contrasts with the structure of Eastern Sudanic languages further north in Sudan and Chad (Nubian and Taman). Nilotic and Surmic languages (not to mention the Jebel language Gaahmg) probably underwent a typological adaptation, presumably when its speakers entered the area south and southeast of the Nuba Mountains (Dimmendaal 2005). The presence of Daju languages in the Nuba Mountains appears to be the result of a fairly recent migration, probably not earlier than the sixteenth century. Stevenson (1984: 37) makes the following statement in this respect: “The Daju, as is well known, had been at a much earlier period rulers in Darfur until they were ousted by the Tunjūr, who in turn were succeeded by the Fur. One section of the Daju had moved westwards to Dar Sila (in Chad) and beyond, but others moved to the east, and their representatives today are the Daju of the hills near Lagowa, the people of Liguri and Saburi near Kadugli, and the Shatt, south and south-west of Kadugli.” Thelwall (1981: 172) observes with respect to these Daju languages, which are not closely related: “There is definite evidence for some movements east into west Kordofan, and that these movements happened separately over a period of time long enough to keep these groups separate, or perhaps to allow them to develop separate linguistic features.” These various external genetic links between languages in the Nuba Mountains strongly suggest that the area functioned as an important natural refugium for various groups. The most likely immediate cause was climate change over the past millennia. Slave raids probably played only a minor role in the dispersion of people into the Nuba Mountains. It only occurred on a large scale in the nineteenth century (see Stevenson 1984: 43–50). Moreover, the Nuba Mountains do not provide any natural defence against enemies. According to Tima oral tradition, for example, their ancestors moved into the higher zones (including caves) of the mountains during the 19th century in order to escape slave raids. In other words, their homes in the lower zones did not provide any natural protection. When the Mahdist state was dissolved in 58 CHAPTER 3

1898 and Sudan became a Condominium, the British apparently encouraged Tima and other groups to leave their temporary homes and to return to the lower regions. An additional factor suggesting that the complex linguistic map of the Nuba Mountains is the result of gradual migrations into this area during the past millennia rather than any other more recent cause is the absence of “emergency languages”. These typically come about when speakers of many different languages come into contact with each other and start searching for a common channel of communication. This can be observed, for example, with respect to the emergence of Nubi, the creolized variety of Sudanese Arabic, which developed within a relatively short period in the 19th century among soldiers. Other examples of “emergency languages” are found among Maroon communities in Suriname and other South-American countries. Their creolised varieties of Indo-European languages came into being when speakers with different linguistic backgrounds who had escaped the slave plantations created new (Maroon) communities. There is no evidence for any such process in the Nuba Mountains. Another fact which does not fit the scenario of communities taking refuge in the Nuba Mountains in recent times as a result of political upheavals or enslavement of communities outside these zones is the presence of extensive lists of common clan names shared by communities which speak genetically unrelated languages. An example of this is given in Chapter 4 with respect to Tima and Katla, which are Niger-Congo, and Temein and Wali, which are Nilo- Saharan (Meerpohl 2013). Exogamy is part of the cultural history of the area, and appears to have played a role mainly whenever new ethnic groupings come into existence. Indeed, common clan names usually come about as a result of exogamy or ethnic fission and fusion. Their presence matches Stevenson’s (1984: 122) observation: “How the Temein people came to their present habitat in the Temein hills (and part of Wali) is obscure, for like many Nuba peoples they preserve no clear accounts of their migration.” If the migrations had taken place relatively recently (e.g. in the 19th century), one would expect at least some sort of recollection of dramatic historical events like political domination and enslavement. Consequently, a far more plausible explanation for the complex linguistic picture of the Nuba Mountains is a gradual migration of speech communities into these higher elevations, as they reacted to dramatic climatic changes. Their presence in the Nuba Mountains for several millennia now has presumably led to further genetic diversification. It is quite likely that human beings have lived in the Nuba Mountains since time immemorial because of the ecological conditions, but virtually no Accretion Zones and the Absence of Language Union 59 archaeological research has been carried out in the area so far. Nadel (1947: 4) notes that “[. . . a] few rock-paintings in widely scattered places (in the Koalib country and one or two other hills), about which local people can give no information, and which they could never have produced, suggest an older population which preceded the present inhabitants.” Thelwall and Schadeberg (1983) present the following hypothesis regarding the relative chronology of linguistic settlement of the Nuba Mountains:

1. Kordofanian 2. Nyimang, Temein, Kadugli 3. Daju i: Shatt, Liguri 4. Hill Nubian 5. Daju ii: Lagawa

According to Thelwall and Schadeberg (1983: 226), “Kordofanian” is much more diversified than any of the other language families in the Nuba Mountains. Moreover, “[. . . t]he whole Kordofanian language family is located within the Nuba Mountains where it occupies the most central and most widespread geographical position [. . .]. There appears to be a continuous history of branch- ing, beginning with a presently assumed four-way split into Katloid, Heiban, Talodi and Rashad. This primary split must have preceded the subsequent split of Talodi into Tegem and Narrow Talodi [. . .]. On the basis of this evidence it is clearly indicated that the development of Kordofanian occurred in the Nuba Mountains, and that Kordofanian has the longest linguistic history in this area”. Indeed, genetic differences between these Niger-Congo languages appear to be huge. I consequently assume that Kordofanian is not a valid genetic grouping. If Katloid-Rashad is indeed more closely related to Benue-Congo than to Talodi-Heiban, as Dimmendaal (to appear b) claims, their ancestral communities may well have migrated into the Nuba Mountains independently of each other. Given that Talodi-Heiban speakers live in the central and eastern parts of the Nuba Mountains, and that Katloid-Rashad speakers occupy the western and central parts, the former may have entered the area from the west before the latter entered it from the west. But whether these migrations preceded those of the ancestral Kadu(gli) or the Temeinian communities cannot be determined. The remnants of number suffixation in Katloid-Rashad and of case marking in the Heiban group (as discussed above) suggests contact at an early stage with languages with these typological properties. Number suffixation and case marking of course are found in Hill (Kordofan) Nubian languages. But whether this Eastern Sudanic group indeed was the source, or another earlier 60 CHAPTER 3

Northeastern Nilo-Saharan group, remains pure speculation. Sagar (1922: 138– 139) mentions a legend that the Nyimang and Koalib (who speak a Heiban language) were once contiguous, occupying Dilling and other small, now uninhabited hills, and that the coming of the Hill Nubians from the north drove a wedge between them. It is very likely that the Koalib were once much further west than they are now. Nadel (1947: 358) records a tradition in Delami that the northern Koalib once lived in Kortala in the Kadaru hills, and even today they share Habila (between Kadaru and Dilling) with the Hill Nubians.10 Thelwall and Schadeberg (1983: 220) also assume that in more recent times people may have found it expedient to escape the influence of the Sahel empires and slave-raids, and to move into the Nuba Mountains in order to take refuge there. Possibly, the slave trade over the past few centuries—or even over the past millennium—had a catalysing effect. But it cannot explain the genetic and typological diversity found in the Nuba Mountains for the reasons already explained. If the complex linguistic picture found in the Nuba Mountains today is the result of numerous migrations by adaptive communities originating outside the area, why did they keep their languages and why is there so little evidence of convergence between them? Contrary to what the name suggests, the Nuba Mountain area is for the most part merely a plateau interspersed with hills. Admittedly, there is a major mountain range between the Tima (Niger-Congo) and Tulishi (Kadu) area, but people do cross it regularly, for example to visit the hospital. It is very unlikely, therefore, that this mountain range played a significant role as a natural barrier. The emergence of autarkic economies associated with a wide range of different speech communities migrating into the Nuba Mountains independently of each other helped to maintain this linguistic diversity. Once these different speech communities had established themselves, presumably over a period of several thousand years, ritual sites (e.g. burial sites of important ritual leaders) emerged, which further contributed to localist strategies (a phenomenon elaborated upon in the next chapter). As Nadel (1947: 14) states: “It seems significant that none of the Nuba groups has a word for tribe or ethnic group, although they all have terms for village, hill community and clan. These three concepts can be defined, concretely, by

10 Nubia was first mentioned about 4,300 years ago in accounts of trade missions during the Old Kingdom in pharaonic Egypt, when Aswan formed the southern limit of control. The ancient Egyptians imported gold, ebony, wood and ivory from the south. Nubia, which consisted of a number of kingdoms, probably served as a trade corridor between Egypt and zones further south. Accretion Zones and the Absence of Language Union 61 reference to some other entity—habitation, locality, or descent. Tribe could be defined only tautologically, as a group which is itself”. The Tima people, for example, associate particular mountain ranges with each group in the Nuba Mountains and also extend this concept to the homeland of foreigners who live on the plains. When enquiring about someone’s origin, they literally ask ‘from which mountain does this person come?’. By extension, the Tima word for ‘hill, mountain’, kwálʊ̀ŋ, may refer to ‘country’, ‘territory’ or ‘nationality’. As Meerpohl (2013) shows in her anthropological study of the Tima, people also hold strong emotional ties to the areas where they live, amongst other reasons because this is where their ancestors, in particular important ritual leaders of the community (so-called kujurs) are buried. Most groups in the Nuba Mountains are autarkic in terms of their subsistence economies. They are essentially sedentary agriculturalists who also own cattle, sheep and goats. The favourable ecological conditions presumably enabled them to develop self-sufficient economies in the course of history. Consequently, one would not expect them to develop any trade languages or larger networks. These latter conditions would have resulted in the spreading of linguistic features and the development of major contact zones as a result of multilingualism. The limited degree of multilingualism may also be the main reason why the historical scenario that Dixon (1997) proposes for languages in Australia does not seem to apply to the Nuba Mountain area. According to this model, extensive borrowing between related and unrelated languages is common once equilibrium has set in. This would have been the situation in the Nuba Mountains after a period of dramatic climate changes when new groups began to move into the Nuba Mountains. After the initial expansion (punctuation) of families such as Nubian or Talodi and Heiban, some contact occurred, as argued above, but this apparently never resulted in extensive lexical borrowing or replication of grammatical patterns. In the case of Australian aboriginal communities, patterns of exogamous marriage and the formation of bonds for common hunting presumably resulted in frequent lexical and grammatical borrowing. But hunting plays a very limited role in the Nuba Mountain communities. Moreover, the same area is not conducive to a nomadic, transhumant pastoral economy. Instead, agriculture and the sedentary lifestyle that accompanies it is the common pattern. Once speech communities settling in the Nuba Mountains had adapted to the area, taking advantage of the variety of ecological zones available, they apparently did not feel any need to build up larger networks for the exchange of commodities or, linked to this, to become multilingual or maintain exogamy. The Koalib, for example, are one of the larger speech communities in the Nuba 62 CHAPTER 3

Mountains with between 100,000 and 200,000 members (Quint 2006: 16). But many speakers of Koalib speak only Arabic as a second language, fluency in any of the surrounding languages being a relatively rare phenomenon (Nicolas Quint, personal communication). This appears to be typical in the area. Although I have met individual Tima speakers who have also mastered three or more neighbouring languages, this ability is rare. In other words, multilingualism is not very common in the area. The sociolinguistic surveys by Mugaddam (2006) and Mugaddam and Abdelhay (2013) confirm that Arabic tends to be the only second language. Because of its dominant role in the educational system and public administration, it is now tending to replace various local languages, but this is a recent phenomenon. Consequently, the absence of any significant instances of convergence in the Nuba Mountains is probably due to the fact that multilingualism was never widespread to begin with. Whereas common clan names attest to contacts in the past, these do not necessarily play a role today. Although the Tima feel a kindred spirit with neighbouring communities such as the Wali with whom they share clan names, this does not necessarily imply that there is much interaction between them today. One indication of this is that they do not necessarily speak each other’s languages. Whereas Eastern Sudanic groups speaking languages belonging to Daju, Temeinian, Nyimang and Afitti found their niches in the Nuba Mountains, the ancestors of pastoral Nilotic and Surmic groups (whose languages are closely related) migrated to areas southeast and south of the Nuba Mountains. Their speech areas constitute “open spread zones”, i.e. zones where languages can enter from more than one side (Nichols, to appear). These contrast with closed spread zones, which are characterised by barriers to entry on all sides. The historical picture emerging from our current knowledge concerning the Nuba Mountains is similar to that of the Upper Xingu region in the Brazilian Amazon. Apart from the high number of languages belonging to genetically unrelated language families, there are also few if any traces of convergence or diffusion in this area. Aikhenvald (2012: 89) points out that few people in the area know languages other than their own (though many can be considered passive bilinguals). She concludes that this is due to the shallow time-depth involved, various groups having moved into the area in more recent times. The few shared typological features between genetically unrelated languages are therefore seen as manifestations of an incipient convergence zone. In contrast, as argued above, the Nuba Mountains in all likelihood constitute an ancient refugium where convergence came to a halt at an early point. This follows naturally from a social context where the need for multilingualism is limited. The Accretion Zones and the Absence of Language Union 63 latter situation in turn may lead to a conscious valorisation of linguistic differences, or iconisation, as Irvine and Gal (2000) have called this phenomenon. Such attitudes do not preclude the possibility of sharing “common ground” when national or global politics require this. Schadeberg and Blench (2013: 16) point out that “[. . .] it is clear that Nuba peoples are strongly attached to their culture, making every effort to reinforce and transmit it during periods of exile. Cultural festivals, the wrestling competitions in Khartoum, the rapid rebuilding of “traditional” house forms after 2005 all testify to a newly strengthened group identity.”11 A number of authors, such as Riefenstahl (1973, 1975) also treated the “Nuba” as a cultural unit. Of course, people in the Nuba Mountains are aware of such publications, and of the fact that communities elsewhere in Sudan tend to view them as a homogenous group politically. Still, these are the result of more recent developments, in particular globalisation. “Cultural otherness” as constructed and maintained by way of language and other distinctive features was a more prominent feature of Nuba Mountain communities until fairly recently. One such case of “alterity” is discussed in the next chapter.

11 Southern Sudan became an autonomous region in 2005 and an independent state (South Sudan) in 2011. CHAPTER 4 Esoterogeny and Localist Strategies in a Nuba Mountain Community

4.1 Introduction

Esoterogeny, a phenomenon whereby speakers add linguistic innovations that increase the complexity of the language in order to highlight their distinctiveness from neighbouring groups, has played a role in the historical development of languages in the Pacific (Thurston 1989). As Ross (1996: 184) has argued, “esoterogeny arises through a group’s desire for exclusiveness”. He refers to specific languages in Papua New Guinea which developed into ‘emblematic’ or ‘in-group’ codes excluding outsiders. Such speech varieties favour innovations leading to increased complexity and to differences from neighbouring lects. Ross (1996: 183) also points to such processes as elision and assimilation which result in phonological compactness, allophony and allomorphy, the accumula- tion of irregularities, an elaboration of the lexicon with numerous near syn- onyms, much borrowing, and an increase in the frequency of opaque idioms. Members of the Tima community (Sudan) also claim that their ancestors deliberately changed their language so that the neighbouring Katla, who speak a genetically related language, could no longer understand them. The present contribution sets out, first, to discuss synchronic variation between Tima speakers (section 4.2), followed by a comparison of the structure of Tima with the related language Katla, in order to establish the degree of variation between these languages (section 4.3). Whereas lexically these languages are fairly close, the grammatical differences between them are considerable. However, section 4.4 will argue that neither esoterogeny nor an alternative process, metatypy, were probably involved in the historical restructuring of Tima (or Katla for that matter). More likely, the deviant structure resulted, not from language manipulation, but rather from another type of contact-induced change, the imposition of structural properties from a specific source into a specific target in a language shift situation.

4.2 Internal Variation in Tima and the “Apparent-time” Approach

Tima is spoken by approximately 6,000 people of whom about 5,000 live along the western fringe of the Nuba Mountains; the others are based in greater

Khartoum. The Tima refer to their own language as dù-mùrík, and to themselves as ì-múrìk (singular kù-murìk). According to Greenberg (1963), Katla and Tima together form one of five subgroups within the Kordofanian branch of Niger-Kordofanian, which Williamson (1989) renamed as Niger-Congo. Schadeberg (1981b, 1981c) classified the Kordofanian languages into four groups:

Heiban

Talodi

Kordofanian

Rashad

Katla figure 2 Kordofanian According to Schadeberg (1981a, 1981b)

Greenberg (1963) assumed that the fifth group, Kadugli(-Krongo), was also part of Kordofanian, and referred to it as Tumtum. But Schadeberg (1981c) removed it from Niger-Congo, recommending that it be included in the search for substantial Nilo-Saharan comparisons. Apparently, he did not question the status of the Katla group as a proper member of the Kordofanian branch of Niger-Congo even though the actual evidence for close genetic affiliation with Heiban, Talodi, and Rashad appears to be rather slim. Lexically, the Katla group (or Katloid, as it is called in the present study) differs considerably from the other three. Also, although the languages of the Katla group do have noun-class systems, several of the actual forms do not appear to be cognate with Schadeberg’s (1981b, 1981c) reconstruction of Heiban and Talodi. In fact, there appears to be more grammatical evidence for a closer genetic affiliation between the Katla group and Niger-Congo subgroups like Benue-Congo and Kwa (Dimmendaal, to appear b.). In the present study, the Katla group is therefore treated, not as a member of the Kordofanian branch, but as a separate Niger-Congo split-off. The question of the wider genetic affiliation of the Katla group, however, does not play a crucial role in 66 CHAPTER 4 the ethno-history which is central to the present contribution, and accordingly is not further discussed here.1 Linguistic change is drawn from a pool of synchronic variation, as John Ohala has argued in a number of publications (e.g. Ohala 1989). Every time speakers of a specific language produce a particular word there is minor variation. For example, speed of pronunciation affects the realisation of sounds to a considerable degree. This results in variation within the utterances of a single speaker, as well as between speakers. Ohala (1993: 239) points out the “[. . . t]remendous variability that exists in what we regard as the ‘same’ events in speech, whether this sameness be phones, syllables, or words [. . .]”. It is from this pool of variation that sound change is drawn or actuated. From an analytical point of view, there is a related issue (in addition to the actuation problem) involving the diffusion of sound change and other types of linguistic innovations across a speech community, i.e. the transition problem. According to Labov (2001: 33–34), the leaders of linguistic change are not inventors of a certain form, but rather those who, by reason of their social histories and patterns of behavior, will advance the ongoing change more vigor- ously. Based on his research mainly in American metropoles, Labov arrives at the conclusion that “[. . . t]he ‘leaders of change’ [. . .] are people at the centre of social networks within a community, often combined with a broad range of connections outside of their immediate locality (sometimes called ‘expanded centrality’).” This type of research concerning language change is virtually non-existent within African linguistics, partly because the “real-time” approach requires longitudinal research of speech behaviour within a community. I used the complementary method, the “apparent-time” approach, during a research trip to the Tima speech community in 2007. This involved interviews with about forty males and females between 20 and over 80 years of age, originating from the villages of Käyyä, Märyäng, Märyäm, Timmä, and Wahyah. From this preliminary survey, it became clear that the Tima language is quite homogeneous, with only minor phonological variation between speakers. The most distinctive property, occurring in the speech of the oldest speakers and separating them

1 Stevenson (1957: 51) points out, with respect to the Rashad group, that “[. . . t]here appear to be some links with the KATLA group in the western part of the Nuba Mountains which require further investigation.” There is indeed both lexical and grammatical evidence that the Katloid group is distantly related to the Rashad group, and that these form an early split- off from Niger-Congo rather than being part of Kordofanian. Roger Blench (personal communication) also takes this view. See Dimmendaal (2013). Esoterogeny and Localist Strategies 67 from younger speakers, is the presence of a strongly velarised [ð] occurring in word-final position:

(1) elderly middle aged young cɪ́ɪ̀ð cɪ́ɪ̀ cɪ́ɪ̀ ‘eye’ kɨ̀-ɗíð kɨ̀-ɗíí kɨ̀-ɗíí ‘leg’ or: kì-ɗíí kì-ɗíí

A second point of variation concerns the vowel harmony of noun-class prefixes. In addition to atr harmony, vowels in prefixes and suffixes correspond in terms of fronting and rounding. For all speakers, the nominal prefix /kV-/ (whereby V represents an underspecified vowel) contains a back rounded vowel if the root contains a back rounded vowel, and a central unrounded vowel if the root contains a central unrounded vowel:

(2) elderly middle aged young kʊ̀-lʊ́ŋ kʊ̀-lʊ́ŋ kʊ̀-lʊ́ŋ ‘thigh’ kɨ́-bʌ́ʌ́ŋ kɨ́-bʌ́ʌ́ŋ kɨ́-bʌ́ʌ́ŋ ‘friend’

Additional allomorphs, containing a front vowel in the noun-class prefixes whenever the root contains a front vowel, are used by certain speakers, as in the word for ‘leg’ above. Whereas none of the old people appears to apply such a fronting rule before front vowels—they use a centralised variant instead— speakers from both the middle and younger generations sometimes front the prefix vowel ([ɨ] -> [i] before [i] or [e], and [ɘ] -> [ɪ] after [ɪ] or [ɛ]) in such cases.

(3) elderly middle aged young kɘ̀-bɛ̀t̪ɛ́r kɘ̀-bɛ̀t̪ɛ́r kɘ̀-bɛ̀t̪ɛ́r ‘fiction’ or: kɪ̀-bɛ̀t̪ɛ́r kɪ̀-bɛ̀t̪ɛ́r

Whereas the use of a strongly velarised [ð] by the elderly is seen as a distinctive property of “archaic Tima”, the variation in the fronting harmony is something many speakers are not even aware of. So far, no clear-cut social stratification has been established between those middle-aged and young people who apply fronting and those who do not. It is certainly not a matter of dialectal variation, as speakers from the same village may pronounce either. What is more, a number of speakers who were aware of this variation indicated that they alternate between the prefix forms because “it’s all the same”. Consequently, it seems 68 CHAPTER 4 more likely that different social networks influence whether people apply the rule or not. It remains to be determined who the “leaders of change”, in this case the innovators of the fronting harmony rule, are. Similarly, it remains an intriguing question why some individuals copy such innovations and others do not. Tima villages are organised according to clan membership, with specific clans being concentrated in certain villages. But all Tima villages are “mixed” in the sense that representatives from different clans may be found in one and the same village, and one and the same clan may be represented in different villages. Consequently, the phonological variation described above may be linked to social networks involving clan membership and intensive contacts between members of the same clan.2

4.3 An Initial Comparison with Katla

Whereas Tima appears to be fairly homogeneous language-internally, it deviates from Katla in a number of respects.3 Phonologically, Katla and Tima vary only in detail. Katla has labial-velar stops (Tucker and Bryan 1966: 262), a feature found in a range of languages stretching from Senegal to South Sudan (Greenberg 1983). It would seem to be most likely, therefore, that the labial-velar stops in Katla are a retention. As Stevenson (1957: 191) notes, they correspond to k(w), h(w) and g in Tima, and “[. . . e]ven in KATLA they may vary with kw, gw and sometimes k, g: rakpas or rakwas ten, ɡbɔn or ɡɔn house”. However, only a detailed comparison of cognates between Katla and Tima on the one hand and Benue-Congo, Kwa and Kordofanian languages on the other can provide a conclusive answer to this question. Moreover, a study of borrowings in Katla and Tima from language families such as Adamawa and Chadic may provide additional insights into the history of their languages. Table 9 shows that lexically, Tima and Katla also appear to be quite similar. Note that the examples were selected for their presumed cognacy; other basic words are not necessarily cognate.

2 In addition to these subtle phonological variables, there appear to be some lexical differences between speakers. 3 The third member of the Katla group, Julud, which is closely related to Katla, does not seem to play a role in oral traditions about historical conflicts. This language as well as its speech community therefore is not further considered here. Esoterogeny and Localist Strategies 69 table 9 Lexical cognates between Tima and Katla

Tima Katla kɔ́-nɔ̀ ɡ-ʊ́nʊ̂ ear k-áàh ɡ-àc head kɪ́-ŋɛ̀ ŋɛ̂ŋ mouth k-áàm ɡ-áɡâm hair k-úù ɡ-û dog kɨ́-bʌ́ʌ́ŋ bɔɔŋ (no tones indicated) friend -ʌ́lùk -o-lak (no tones indicated) eat -bul- (no tones indicated) -bul- (no tones indicated) die

Both Tima and Katla have reduced noun-class systems, with a number of prefixes still being used productively, whereas others occur only in closed sets. The most common alternation in Tima involves a singular or singulative prefix kV- (as in the examples in Table 9), and a plural or collective prefix /ɪ- ~ i-/. The fact that agreement markers related to different noun classes have come to be used as concordance markers on modifying categories such as demon- stratives, adjectives, numerals further suggests that Tima had a more elaborate noun-class system at an earlier stage (Dimmendaal 2014b). It is mainly in the verb system and the organisation of information structure as reflected in constituent order that Tima and Katla differ most dramatically from each other. For Tima’s closest relatives, Katla and Julud, Hellwig (To appear) gives the following set of pronominal subject markers (orthographic symbols replaced by ipa symbols):4

table 10 Pronominal subject markers in Katla and Julud

Katla Julud

-atr +atr -atr +atr

1sg ɲà- ɲò- ɲà- ɲò- 2sg ŋà- ŋò- ŋà- ŋò- 3sg ø- ø- ø- ø- 1pl.excl nɪ̀- nì- nɛ̀- ~ ndɛ̀- nè- ~ ndè-

4 Table 8 replaces the one in Dimmendaal (2009a). 70 CHAPTER 4 table 10 Pronominmal subject markers in Katla and Julud (cont.)

Katla Julud

-atr +atr -atr +atr

1pl.incl ɪ̀- ì- ɪ̀- ì- 2pl nà- nò- nà- nò- 3pl nì- nì- ɪ̀- ì-

Tima attests a similar system of pronominal subject prefixes, but in addition, it has a system of post-verbal subject-marking clitics, as shown in Table 11. The choice between type B and C pronominal enclitics in this table is determined by the information structure as expressed in a verbal sentence. Type C pronouns occur with transitive expressions where the pre-verbal object carries focus (Schneider- Blum 2013). Post-verbal nominal subjects in such transitive constructions are preceded by a proclitic nasal element, whereas pronominal subjecs encliticise onto the verb and contain an initial nasal as well (Table 11). This system of post- verbal subject marking is only found with transitive verbs; consequently, Tima manifests split ergativity (Dimmendaal 2010; Schneider-Blum 2013).

table 11 Pronominal subject markers in Tima

Prefix marker A Enclitic marker B Enclitic marker C (Ergative)

1sg N- -da, -dɔ, -dʌ -na, -nɔ, -nʌ 2sg a- -naŋ -naŋ 3sg ø-, N- -ø -mɨnʌ 1pl excl ɪ-, i- -nɛy -nɛy 1pl incl ɪ-, i- -nin -nin 2pl na-, nʌ- -nan -nan 3pl ø- -ø -ɲihinʌ

So far, no evidence has been found for such an ergative system in Katla or Julud (Birgit Hellwig, personal communication). More generally, neither Katla nor Julud seem to allow post-verbal subjects, either as free nouns or noun phrases or as free or bound (clitical) pronominal subject markers. Some examples showing the use of pronominal subject (or A-role) marking in Tima: Esoterogeny and Localist Strategies 71

(4) kí-hìʌ́-↓ɗʌ́ tamáa dumurík=ʌ́ʌ̀ŋ neg-speak-1sg language Tima=neg ‘I don’t speak Tima.’

(5) ɪ́mmɔ̀ŋ-ɛ́ ŋ̀-kéél-nʌ́ fish-foc 1sg-buy-erg.1sg ‘I buy/bought some fish.’

(6) kɨ́-ŋ̀-kʌ́↓luk kɨ́ɗʌ̀ kaɓʊ̀h=ɔ́ɔ̀ŋ neg-1sg-eat 1sg meat=NEG ‘I don’t eat meat.’

Contrary to Katla, Tima has a highly variable clause level constituent order. Whereas svo is the dominant word order in Katla, Tima permits ovs, svo, vso, or sov, depending on pragmatic conditions. And, as examples (4–6) show, the occurrence of post-verbal as opposed to pre-verbal subjects (whether nominal or pronominal) is associated with different inflectional markers on the verb. Consequently, the Tima system also appears to be more complex in this respect. Outside the Nuba Mountains, additional languages with pronominal prefixes, enclitics or suffixes and variable constituent order (conditioned by clause level information packaging, as in Tima) are found southeast and east of the area where Tima is spoken (Map 7). These languages are all characterised by case marking of the post-verbal subject (A-role) in transitive constructions, resulting in a system of split ergativity. Table 12 (based on data in Tucker and Bryan 1966) summarises what is known about pronominal subject marking on verbs in languages in the Nuba Mountains. I was able to collect some additional material on Nyimang and Temein, for example, but did not find any evidence for subject-marking enclitics or suffixes.

table 12 Pronominal subject marking in languages of the Nuba Mountains

Nyimang, Afitti absent Ghulfan (Nubian, Eastern Sudanic) suffix Daju of Lagowa (Daju, Eastern Sudanic) prefix Temein (Temeinian, Eastern Sudanic) prefix Keiga Jirru (Temeinian, Eastern Sudanic) prefix Talodi-Masakin (Talodi, Niger-Congo) prefix, prefix + suffix Koalib-Moro (Heiban, Niger-Congo) prefix, prefix + suffix Lafofa (Lafofa, Niger-Congo) prefix Kadu (independent family) suffix 72 CHAPTER 4 Ergativity in languages of in languages Africa Northeastern Ergativity map 7 Esoterogeny and Localist Strategies 73

Table 13 summarises what is known about basic constituent order in languages of the Nuba Mountains.

table 13 Constituent order in languages of the Nuba Mountains

Nyimang sov Ghulfan sov Daju of Lagowa svo Temein svo Keiga Jirru svo Talodi-Masakin svo (vso) Koalib-Moro svo (vso) Lafofa sov (osv, svo) Kadu vso

There is some typological similarity between the system found in Tima (but not Katla or Julud) and a number of languages belonging to the Heiban and Talodi groups. In these languages, specific tense-aspect forms require subject prefixes for all persons, whereas other constructions use a combination of prefixes and suffixes. Compare Table 11 (Tima) with the data in Table 14 from Otoro, a member of the Heiban cluster (Tucker and Bryan 1966: 280).

table 14 Pronominal subject marking in Otoro

Prefix Suffix

1sg i- -i 2sg ŋa- -a 3sg ŋu- -u 1pl ana- -ana 2pl nya- -anya 3pl al- -alɔ 74 CHAPTER 4

The Katla group is only distantly related to the Heiban or Talodi group, as pointed out above. As these subject-marking affixes in Otoro show, there is no obvious formal similarity to those found in Tima (or Katla for that matter), except for the second person singular and plural. Consequently, there is no immediate reason to believe that the Tima system with its pronominal prefixes and enclitics was inherited from a common Niger-Congo ancestral language. Whether the post-verbal pronominal markers for subjects (or A-roles) in Otoro are indeed suffixes, or enclitics like in Tima, is not clear at this point. But their presence puts the Tima system in a different typological perspective; apparently, we are dealing with an areal feature shared between Otoro (and other Heiban languages) and Tima. Was it indeed language manipulation which resulted in the deviant grammatical structure of Tima (in particular its verb system and constituent order rules), as the oral tradition of the Tima suggests, or is there an alternative explanation, such as shift-induced interference from a language type as found in Otoro? Or is there maybe another explanation? These questions are addressed next.

4.4 Metatypy, Esoterogeny, or Neither?

I started research on Tima in 2004, when I was approached by a group of Tima speakers in Khartoum. According to these spokespeople their language was vanishing. Although the majority of Tima speakers based in Khartoum do indeed seem to have switched to Sudanese Arabic as a first language, the Tima language is still spoken in the Nuba Mountain area. Nevertheless, its role in daily interaction is diminishing even there. In daily conversations there is a lot of code-mixing between Tima and Sudanese Arabic. Also, Tima often use Arabic for daily interactions, in spite of everyone present being able to speak Tima. These are strong signs that Tima is becoming obsolescent. But the language shift appears to be relatively recent, caused by the dominant role of Sudanese Arabic in education and administration. The fact that many Tima people find it highly important that their language is going to be documented and developed for educational purposes shows that they consider it to be a crucial, emblematic feature for them as a social group. We have already seen that Tima differs from Katla in its verb structure and constituent order. According to one oral tradition among the Tima, their ancestors manipulated their language after a conflict with the Katla, in order to set themselves apart from them. But language manipulation or engineering by ethnic groups as a whole typically involves the introduction of phonetic Esoterogeny and Localist Strategies 75 features and lexical innovations as social registers (Thurston 1989, Ross 1996). Kießling and Mous (2004) observe this with different social groups within one speech community, in their survey of urban youth languages in Africa. Conscious language change is attested in a wide variety of languages across the world; compare Thomason (2007) and Storch (2011) for a succinct survey. Claiming that the Tima manipulated their grammatical, rather than their lexical, structure, and consequently claiming that the Tima case is an exception to this widespread and well-attested phenomenon without providing any additional, concrete evidence is not a very attractive hypothesis. So are there alternative explanations? There is another type of language contact process involving what Ross (2001) calls metatypy, i.e. a change in the grammatical organisation of a language as a result of its speakers’ bilingualism in another intercommunity language. The restructuring is driven by grammatical calquing from this intercommunity language. The modified language, i.e. the one undergoing metatypy, tends to be emblematic of the speaker’s social identity. Such a process appears to have taken place among the Baale in the border area between Sudan and Ethiopia (Dimmendaal 2000, Chapter 2). The Baale consider themselves to be part of a larger ethnic grouping which they refer to as “Suri”, and which is known to neighbouring groups in south-western Ethiopia as “Surma”. At the same time, the Baale form a tight-knit community which is aware of its separate identity, with the Baale language as a marker of this identity. The Baale people live in an area infested with tsetse flies, causing them to shun pastoralism in favour of agriculture, whereas the Tirma and Chai remained pastoralists. Baale is a member of the Western branch within the Surmic language family. Tirma(ga)-Chai is part of a dialect continuum which includes Mursi (spoken by a distinct ethnic group) and belongs to the Southeastern branch of Surmic. Southwestern and Southeastern Surmic are only distantly related with relatively few lexical cognates (Yigezu 2002). Tirma(ga)-Chai speakers number around 20,000, whereas the number of Baale is estimated at around 9,000. Tirma(ga)-Chai functions as the inter-community language between them, i.e. the Baale speak Tirma(ga)-Chai but very few Tirma or Chai speak Baale. Metatypy typically occurs when bilingual speakers use such an inter-community language so extensively that they may be even more at home in this language than in the (emblematic) language of their own community. Tirma(ga)-Chai clearly provided the metatypic model for grammatical calquing in Baale, which has also been heavily influenced phonologically and lexically by Tirma(ga)-Chai (Dimmendaal 2000, Yigezu 2006). The deviant structure of Tima, however, cannot be explained along these lines, since at least today there is no such inter-community language serving as 76 CHAPTER 4 a metatypic model. Neither can the deviant structure be explained as being the result of Sudanese Arabic influence. First of all, the influence is too recent; second, Sudanese Arabic lacks the verbal inflectional properties concerning pronominal subject marking one finds in Tima. The deviating structure of Tima (in comparison with Katla) consequently does not appear to be the result of metatypy either, as the sociolinguistic conditions (e.g. diglossia) necessary for this process are absent. A more likely explanation for the deep structural differences between Tima and Katla (and Julud) would be shift-induced interference and imposition from some other languages as a result of imperfect language shift (in the sense of Thomason and Kaufmann 1988). But which of these is the innovating language: Tima or Katla?

Scenario 1: Tima as the Innovating Language I originally hypothesised that Tima was the innovating language, as this would be in accordance with their own oral tradition. As the deviating features (when comparing the language with Katla) did not involve the features usually associated with esoterogeny, and as there was no evidence for metatypy either, I deduced that these features might have come about as a result of language shift from one or more Kordofanian languages (in the sense defined by Greenberg 1963), where similar features are attested, as we have already seen. Under such historical conditions, however, one should be able to find common clan names, and there is as yet no evidence for this. Although one cannot exclude the possibility that these may be found, the Tima deny having any blood ties with any of the other groups speaking Niger-Congo languages other than Katla or Julud, or with the Tulishi, who live south of the Tima area and who speak a Kadu language. What we do find is common clan names between the Tima and the neighbouring Katla, as might be expected. And, interestingly, one also finds common clan names between Nubian groups like the Ajang (Ghulfan) or Wali and the Tima. These blood ties are very much present in the minds of speakers. An anecdote may help to illustrate the importance of clan relations. During a fieldtrip to the Tima area in September 2007, the car we were driving in was stopped while crossing the Wali area north of the Tima zone. The Wali had erected a road block, to control traffic crossing their area, which was a transition zone between the region controlled by the Sudanese army and the region controlled by the Southern Peoples Liberation Movement. Whereas we met with some initial suspicion, the atmosphere completely changed when one of the Tima informants in the car introduced himself and told the Wali guards that he belonged to the kukiŋ clan. All of a sudden, the Wali guards became Esoterogeny and Localist Strategies 77 friendly and relaxed and allowed us to continue our journey, because apparently they have the same clan. It is these blood ties which results in a kindred spirit and a sense of common history. To conclude, the deviant structure of Tima (concerning the inflectional structure of the verb and constituent order) cannot easily be explained as being a result of shift-induced interference from one or more Niger-Congo languages. Possibly, we are dealing with an archaic areal property which probably goes back much further in history.

Scenario 2: Katla as an Innovating Language The alternative explanation for the structural differences between Tima and Katla is to assume that structural innovations occurred in Katla. Although this hypothesis contradicts the oral tradition, it turns out to be slightly more plausible for at least two reasons. As we might expect in a language with ergative properties (compare Dixon 1994), Tima has an antipassive marker -ak / -ʌk, which is used with inherently transitive verbs when the object is absent (deleted) or expressed as a peripheral argument.

(7) ɲ̀cʌ́-ŋ̀-kɔ́yɔ̀-ɔ̀k kɨ́ɗʌ̀ prog-1sg-prepare-ap 1sg ‘I am (in the process of) preparing something.’

Preliminary investigations (Birgit Hellwig, personal communication) show that Katla has a cognate marker -ak, in spite of the fact that it does not manifest any (split) ergativity synchronically. This “Activity” marker (as Birgit Hellwig calls it) is cognate with the antipassive marker in Tima. Its presence as a verbal valency-changing marker in Katla is best explained as a remnant feature of an earlier system with (split) ergativity. Ergativity was probably lost as a structural property of Katla and Julud as a result of the loss of post-verbal subject marking (for transitive constructions). This same process, which probably resulted in the loss of ovs or ova order, i.e. of the post-verbal marking of A-roles in transitive clauses in Katla, presumably occurred with pronominal subject marking on verbs by way of enclitical markers. The generalisation of svo or avo constituent order was thus accompanied by a simplification of the inflectional marking for A-roles on verbs. Was this development in Katla a simple historical process, or is there another explanation, e.g. shift-induced interference? If the latter, the structural changes in Katla and Julud would have been caused by speakers of an svo language without subject-marking verbal enclitics (or free post-verbal subject 78 CHAPTER 4 noun phrases) who imposed structural properties from their original language into the target language Katla, in a language shift situation. But are there languages of this type spoken in the same area as Katla? Indeed, there are. The neighbouring language Temein, which forms a separate branch of Eastern Sudanic (Nilo-Saharan) together with Tese has svo order (Tucker and Bryan 1966: 260, based on data collected by Roland Stevenson). My data confirms this hypothesis. Also, Temein uses pronominal subject marking prefixes; there is no evidence so far for subject marking verbal suffixes or enclitics. Is there any evidence that there are blood ties between the speakers of these languages and the Katla? Interestingly, there are common clan names in Katla and Temein, thus providing strong, initial evidence for the alternative scenario: that Katla is the innovating language and that shift-induced interference was the mechanism that caused this historical process. A preliminary investigation of Katla clan names (Birgit Hellwig, personal communication) with those in Temein (Stevenson 1984: 148), make this clear:

Katla Temein curki suroki kukunaŋ kokwunaŋ kukuny kokwiny

More extensive research in this area may reveal additional shared clan names between the two groups.5 Also, a detailed comparison between Katla and Temein may reveal additional evidence for lexical and structural borrowing between the two languages. Consequently, it seems slightly more plausible to conclude that Katla lost (split) ergativity as a morphosyntactic and pragmatic property due to the absorption of people who either spoke Temein or a closely related language.6 Conditions for such a process usually involve a rapid shift whereby the target language model (in this case from Katla) is not fully available to all its members, so that imperfect learning is a probability; moreover, the shifting group has to be so large numerically that the learners’ errors are likely to spread throughout the target speech community. The type of restructuring that apparently took place in Katla is not uncom- mon in languages which have been strongly influenced by other languages

5 The clan name “Curki” is also found in Tima. 6 There is one Daju language, Daju of Lagowa, spoken in the same area as Katla and Tima (see Map 4) which also has svo order and which also lacks enclitical verbal subject-markers. But neither the Tima nor the Katla or the Julud seem to claim any common historical ground with this group, which is assumed to have originated in Darfur. Esoterogeny and Localist Strategies 79 through shift-induced interference. English has undergone many similar changes distinguishing it significantly from most other Germanic languages, particularly from its closest relatives (e.g. unrounding of front rounded vowels, loss of verb-second and verb-final properties, loss of much inflectional morphology, the acquisition of many idioms etc.), not to mention “substrate” influences from Celtic and the Romance language French. One important dif- ference with Tima and Katla concerns the oral tradition that seeks to explain these language changes, a tradition which points towards cultural motives, and which consequently involve conscious efforts to exclude certain outsiders from a specific speech community.

4.5 Localist Strategies

In spite of the fact that there is no evidence for deliberate language change in Tima—or in Katla for that matter—it is still interesting to ask why such an oral tradition emerged in the Tima speech community. What was the cause of this post hoc rationalisation among Tima people for the observed structural differences between their own language and Katla? The Tima appear to be a close-knit society, whose members prefer to marry within their own group. My interviews with a range of ethnic Tima also made clear that they find it important to be open towards the outside world when it comes to the introduction of technical innovations or opportunities to improve their socio-economic conditions. They are very proud, for example, of the academics in their community. Like most other groups in the Nuba Mountains, the Tima have a mixed economy. Traditionally, they do not produce for an external market, nor does there seem to be a big need for trade or exchange of material goods with other communities in the area, as they are essentially self-sufficient, as are neighbouring groups. Claims about ethnic identity among the Tima and other groups in the area are associated with territorial claims. For them, their language clearly functions as an important emblematic feature of their ethnic identity, setting them apart from neighbouring groups like the Katla. In this respect, the Tima seem to follow, what one might call a “localist” strategy, following Hill (e.g. 2001). In her research among certain groups in Mexico, Hill concluded that communities may follow either “localist” or “distributed” strategies. In the localist strategy, speakers hold an opinion that they have a rightful and primary claim on valuable and dependable local resources adequate to sustain their well-being. In the distributed strategy, speakers may argue that they have no 80 CHAPTER 4 such rightful and primary claim on valuable and dependable local resources. This strategy also corresponds to interesting linguistic differences. According to Hill, in a “localist” strategy the speaker decides to select a particular kind of person as his or her model, and (s)he will try to sound as much like that particular kind of person as possible. In a “distributed” strategy, the speaker is not sure what kind of person (s)he want to sound like, and will try to sound like a variety of different kinds of people. The speech of any single person and the patterns of variation in any community will always be the product of a combination of these two strategies, according to Hill (2001: 261), who associates each of these strategies with a different set of ecological and socio-cultural constraints. Whereas Hill (2001) relates these alternative social strategies to differences at the dialect level within a speech community, there does not seem to be any principled reason why they could not apply to differences between different speech communities as well, in particular in Africa, where many people are polyglot. One striking feature common to many pygmy groups in Central Africa, for example, is that they tend to speak either the same language as, or one closely related to, a non-pygmy group with which they live in a kind of symbiotic relationship. The languages in question are Central Sudanic, Bantu, Adamawa or Ubangian. This strongly suggests that pygmies shifted their language in order to adapt to local circumstances, i.e. whenever a shift in language solidarity could be associated with positively regarded traits and social privileges with membership in a given social context. Whereas pygmy groups are traditionally small hunter-gatherer communities, non-pygmy groups tend to be numerically dominant, sedentary social groupings with an agricultural economy. In order to be able to add to limited resources and to sustain or improve the well-being of the community, these pygmy groups switched their language solidarity towards those spoken by socially and economically dominant groups, a behaviour that may be interpreted as a distributed strategy. The Tima, and presumably other groups in the Nuba Mountains, on the other hand, followed a localist strategy.7 Speakers of Tima seized on observed structural differences with the closely related Katla language (differences which themselves probably resulted to a large degree from shift-induced interference). The oral tradition among the Tima concerning language manipulation by their ancestors thus probably resulted from post hoc rationalisations among its speakers, who were trying to understand the causes for the observed, deviant structures of Tima and Katla. This rationale became important when social conflicts occurred between the

7 This section has been reduced in order to avoid overlap with the contents of Chapter 3. Esoterogeny and Localist Strategies 81 two groups. Consequently, the Tima language became an important emblematic feature of a distinct social and ethnic Tima identity, an identity which was also associated with territorial claims. Although it is often the case in Africa that language and ethnicity are not isomorphic, and although in quite a few cases additional identities (defined, for example, along clan affinity, as we have seen above) play an important role as well, there are instances where language apparently does constitute an important emblematic feature of social and ethnic identity, as with the Tima people. Thelwall and Schadeberg (1993) also assume that the desiccation of the Sahara time and again impelled people to migrate in search of more abundant water, either to remaining rivers and lakes, or just more generally southwards, but they also emphasise that people may have found it expedient to escape the influence of the Sahel empires and slave-raids, taking refuge in the Nuba Mountains. This chapter argues that there may have been one additional factor contributing to the maintenance, and in fact the creation of additional, genetic and typological diversity: Language shift between typologically and genetically distinct languages accompanied by shift-induced interference and imposition of linguistic features. Speech communities are often clearly aware of relationships between their languages and those spoken by other communities in the area.8 Consequently, they may also reflect upon the “how and why” of observed differences with closely related languages and, through post hoc rationalisations, arrive at an explanation for these differences, also in order to define their distinct, social identities. It remains to be determined to what extent this model—with its “localist strategies” as a central concept—helps to explain the linguistic and genetic diversity in other genetically and typologically diversified regions on the African continent, like central Nigeria.

8 At the same time, speech communities may also be aware of the fact that other languages are not related to their own. All Tima speakers interviewed on this question pointed out that the language of their southern neighbours, the Tulishi, is not related to theirs. This is also the view held by comparative linguists, who treat Tulishi as a member of the Kadu group, which is probably an independent language family.

part 2 Language and Co-Evolution

⸪

CHAPTER 5 Some Observations on Evolutionary Concepts in Current Linguistics

Generations of linguists and anthropologists have tried to key features of particular languages to the physical and social environment of their speakers. Most obviously, relations of this type are to be found in specialised semantic fields in the lexicon, where items reflect phenomena of the physical and social environment. Although trivial in one sense, such items may become of interest when compared cross-linguistically and cross-culturally. Brown (e.g. 1976, 1977a, 1977b), Brown and Witkowski (e.g. 1981), and Brown, Witkowski and Chase (1981) argue that certain social conditions need to be fulfilled before specific lexical distinctions occur in languages. They claim that there is a direct relation between societal complexity and the emergence of lexical differentiation for certain shades of colour and bio-nomenclature. This strongly evolutionary approach to the structure of the lexicon in human language is tested below for some African languages. In fact, the evidence they furnish does not support this evolutionary interpretation of language as a vehicle of culture. Moreover, this chapter will show that their model fails on empirical as well as on general methodological grounds, and, as an alternative to their behaviorist semantic position, will argue for a cognitive approach.

5.1 Evolutionary Concepts and the Study of Language: Some Earlier Attempts

Hypotheses such as those put forward by Brown, Witkowski, and Chase (1981) about the observability of evolutionary traits in modern human languages are far from new. Ever since the birth of comparative linguistics, an idea has been fixed in the mind of many comparativists that language structure is under the functional control of some determinate and recurrent stimulus from the environment in which that language is spoken. Early cross-linguistic studies of morphological systems, for example, resulted in a typology whereby agglutina- tive, inflectional and isolating languages represented evolutionary stages in the development of morphological systems; such an evolutionary interpretation of parametric variation was common in the work of some nineteenth-century morphologists. This conceptualisation of variation between languages is now

© koninklijke brill nv, leiden, ��5 | doi ��.��63/9789004224148_006 86 CHAPTER 5 known to be based on cultural prejudice rather than any solid linguistic principle. Languages may lose affixational morphology and switch to a more analytical system, but they can subsequently develop bound morphemes again. In other words, the variation is essentially due to a shift in morphological type which is of a cyclic, rather than an evolutionary, nature. The limited distribution of specific sounds such as glottalised consonants or clicks in certain areas of the world has led some scholars to contend that archaisms may be found in the phonological systems of modern languages. Hagège and Haudricourt (1978: 49–65) offer an interesting survey of such various historical ideas about somatic, racial or climatological traits conditioning the presence of sounds. According to van Ginneken (1938, 1939), for example, the high number of back (glottalised) consonants in Caucasian and Salish languages are archaisms which have a racial cause: brachycephalism. Due to contacts with races that had dolichocephal traits and narrow mouths and tongues, van Ginneken maintains, these consonants were gradually replaced by consonants reproduced with more front points of articulation. Similarly, van Ginneken asserts that clicks, with their limited geographical distribution (mainly in Southern Africa), are archaisms. Clicks are relatively marked compared to other sounds, and acquired at a relatively late stage by children. If ontogeny recapitulates phylogeny, then the phonological structure of more ancient languages must have been close to structures existing in child language of modern human beings (see also Jakobson, 1962: 531 for similar observations). Connelly’s (1984) and Demuth’s (1992) longitudinal study of Sotho and Mowrer and Burger’s (1991) study of Xhosa, another Southern Bantu click language, confirm this. Attempts have also been made to relate noun-class systems to specific concepts of the natural and supernatural world. Most Bantu languages of Central and Southern Africa have noun-class systems, which, according to some scholars (e.g., Merkies 1980) is a direct reflection of the classification of referents in the natural world. Such a claim lacks empirical support for at least three reasons: These languages do not belong to a single culture; the observed correlations are far too weak (with numerous nouns in particular noun classes that do not fit the semantic characterisation of that class); and the semantic fields assumed to be represented in the noun-class system are not culture- specific. Comparable studies on grammatical gender in languages and the natural world, for example with regard to the sex of the heavenly bodies, show that there is no necessary correspondence between linguistic oppositions and those which are expressed in religious beliefs, ritual, myths or folklore (Levi- Strauss 1970). With regard to constituent order, claims have also been made about the relative chronology of certain types. In his classic study on constituent order, Some Observations on Evolutionary Concepts 87

Greenberg (1966) argues that a useful typological differentiation can be set up on the basis of the position of the verb (v) relative to other syntactic constituents such as subject (s) and object (o). This resulted in a distinction between sov, svo and vso language types. According to Givón (1979), sov represents the earliest attested word order type, contemporary languages with this word order being instances or relics of an earlier evolutionary stage. Givón considers that we have been drifting away from the sov typology—i.e., from a pre- syntactic, pragmatic mode of discourse with topics or agents—to a mode with syntactic subjects and objects as the most recent phylogenetic stage in the evolution of language. Givón takes it for granted that the reconstruction of an original sov word order is well-established for most (if not all) language families. However, this is far from obvious. Proto-Indo-European probably allowed verb- initial order in specific contexts. Old Egyptian, Proto-Semitic, and probably Proto-Chadic being verb-initial languages, it is possible to justify a hypothesis positing an original verb-initial syntax for their common ancestor, Afroasiatic; similar observations can be made for other proto-languages like Proto-Nilotic and Proto-Austronesian. Givón’s hypothesis is therefore creative, but lacks sup- porting evidence. Rather than assuming that one type is older than the other, it seems more likely that the variation currently found is due to a cyclic rather than an evolutionary development, as with morphological variation.1 The various attempts to relate the appearance of phonological, morphological or syntactic traits to cultural identity have failed so far. This leaves the lexicon as the sole component in which language and culture may bear a direct relation to each other. Culture may indeed influence the structure of certain lexical fields (rather than the other way round). Links between the two domains are evident if one looks at kinship terminology or at vocabulary items referring to specific technical domains. But the relation becomes less obvious where colour terminology is concerned. Berlin and Kay (1969) created a small revolution by arguing that there are hierarchical relations between certain colour terms. Though never stated unambiguously, they seem to have thought that the variation in the number of colour terms actually distinguished in different languages could only be accounted for by studying the social conditions of particular groups and their technological development. According to Tornay (1978), Berlin thought that specific social conditions needed to be satisfied before more elaborate colour systems could appear. However, Tornay also notes that the encoded colour term hierarchy is to be interpreted in historical, not in evolutionary, terms.

1 Givón (1979) further sets out an evolutionary scenario spanning the gap between pongid and early hominid communication, starting from monopropositional communication. 88 CHAPTER 5

Brown and Witkowski (1981) have added further cases of possible hierarchical and implicational relations in the structure of specialised lexical domains, in addition to Berlin and Kay’s colour terminology scale, in particular for botanical and zoological life-form classes and body part terminology. Their evidence is based on the synchronic study of a wide range of languages as well as on historical-comparative research. The present chapter argues not against the observed hierarchical relations—although modifications are no doubt needed— but rather against the evolutionary interpretation of these hierarchies.

5.2 A Closer Look at the Brown and Witkowski Hypothesis

According to the two main protagonists, Brown and Witkowski (hereafter B&W), there is a strong relationship between the degree of societal complexity and nomenclatural equation in lexical domains. The idea of ‘societal complexity’ is based on an article by Schaefer (1969). In B&W’s conceptualisation, distinctions in several domains of the lexicon, e.g., colour and bio- nomenclature, are added in fixed orders or sequences, whereby the number of life forms increases with societal scale. B&W claim that languages having few lexical items in these fields are spoken by people living in small-scale societies with a low level of political integration, technological development and social integration. In other words, the observed hierarchies reflect evolutionary stages in the growth of vocabularies. These observations about hierarchical and implicational relations in certain domains are a direct spin-off from research that Greenberg (e.g., 1966) initiated on the hierarchical organization of human language. Although observations such as those by Greenberg do not explain anything as such, they provide important generalisations about implicational relations in languages (when correct in terms of observational adequacy), on the basis of which some explanation may subsequently be founded. Berlin and Kay (1969) were the first to argue for such hierarchies with respect to colours. B&W extended claims about hierarchies in semantic domains to lexical domains related to bio-nomenclature and body part terminology on the basis of data from a large corpus of languages. B&W establish the following encoding sequence for botanical terms:

(1) No life forms (i.e., no basic term); (2) ‘tree’; (3) ‘grerb’ (i.e. ‘grass’ or ‘herb’) or ‘grass’; (4–6) ‘vine/grass/bush’ if (3) was ‘grerb’, or ‘grerb/vine/bush’ if (3) was ‘grass’. figure 3 Encoding sequence for botanical terms (Brown and Witkowski) Some Observations on Evolutionary Concepts 89

Thus, if two forms are named in a particular language, ‘grerb’ will be one of them; if there are three terms, ‘bush’ or ‘vine’ will also be labelled, and so on. B&W go on to propose the following encoding sequence for zoological life- form classes:

(1) No life-forms; (2–4) ‘bird/fish/snake’ (from one to three of the triplet, with no internal ordering); (5–6) ‘mammal’ or ‘wug’ (i.e. ‘worm’ or ‘bug’). figure 4 Encoding sequence for zoological life classes (Brown and Witkowski)

In other words, if more than three classes are distinguished, either one or both of ‘wug’ and ‘mammal’ or a conjoined ‘wug-mammal’ class is distinguished. B&W claim that the hierarchies are essentially unidirectional, i.e., regression in the number of lexical terms distinguished does not take place; lexical encoding sequences for these domains have expanded over the past millennia, the lexical expansion itself being caused or stimulated by a growth in societal complexity. With regard to body part terminology, B&W arrive at similar hierarchical organisations of terms. Four stages are recognised in the growth of partonomic nomenclature for upper and lower extremities of the body. For example, Stage 1 and 2 languages do not have labels for ‘hand’ or ‘foot’, while Stage 3 and 4 languages do. Stage 2, with a label for /forearm and hand/ or /lower leg and foot/, apparently does not constitute a necessary interval in the development of all languages. Before arriving at Stage 4, all languages do, however, pass through Stage 3, in which /hand/ or /foot/ is labeled by the same term labeling the native ‘arm’ or ‘leg’ respectively, or, as the case may be, the native ‘forearm’ or ‘lower leg’ ” (Brown, 1976: 421). No correlation between the growth sequence and general cultural development is immediately suggested in the case of body part terminology, according to Brown (1976: 419), contrary to the domain of colour terminology and bio-nomenclature. This gives rise to at least the following questions:

1. What kind of historical evidence supports the presumed unidirectional development from a less elaborate to a more elaborate system of colours and categorisation of the biological environment? 2. Are the hypothesised implicational relations synchronically correct? 3. Do the implicational scales reflect evolutionary stages? 90 CHAPTER 5

To help answer these questions, section 5.2.1 investigates B&W’s historical evidence, as an important part of it is based on the reconstructability of particular items in specific language families, mainly Polynesian and Mayan. Subsequently, section 5.2.2 discusses the synchronic validity of the B&W hypothesis, then section 5.2.3 considers the evolutionary interpretation of the observed synchronic and diachronic differences between languages.

5.2.1 The Historical-Comparative Evidence and Counter-Evidence B&W derive their evidence for an evolutionary interpretation of observed hierarchies in specific semantic fields mainly from the historical-comparative study of two language families, Mayan and Polynesian. According to B&W, traditional lexical reconstructions for these two families as well as others tend to be biased toward over-reconstruction, at least in areas of the lexicon which are additive. For example with regard to colour terminology they note:

They generally are as large as colour term systems formed in contemporary daughter languages. Considering the additive nature of the colour encoding sequence, these reconstructions are unlikely. (B&W 1981: 13)

According to B&W, there has been a gradual increase in colour terminology and bio-nomenclature during the historical development of these and other language families. For example, they claim that the common ancestors of the Mayan languages encoded only one botanical life form, ‘tree’ (Brown, Witkowski and Chase 1981: 1); they suppose that other terms, e.g., for ‘grass’ or ‘branch’ are later developments. They claim that Proto-Mayan had a three- way distinction for colours—‘black’, ‘white’, and ‘red’—other colours being the result of subsequent additions in daughter languages. In order to be able to test the validity of these claims from a methodological point of view, a brief digression into the methods of historical comparative linguistics is required. When can a particular root be reconstructed and when not? The answer to this question is: if the form is attested in daughter languages belonging to different primary branches, and if the various reflexes fit in with the regular sound correspondences between those daughter languages where the form is attested, the root in question has to be set up for their common ancestor. If an item is attested in related languages without the necessary formal correspondences, it cannot be reconstructed for the common ancestor unless the irregular sound correspondences can otherwise be explained. B&W seem to accept the general principles of historical-comparative linguistics. How then do they cope with positive evidence for the occurrence of particular lexical roots in Proto-Mayan, or in Proto-Polynesian for that matter? Some Observations on Evolutionary Concepts 91

B&W try to explain away such instances by assuming that over-reconstruction took place. They claim, for example, that Proto-Mayan only had terms for ‘white’, ‘black’, and ‘red’ but that the widespread terms for ‘grue’ (i.e., green and blue) or ‘yellow’ in Mayan are due to later borrowing (diffusion). Because of their evolutionary bias, they come to the conclusion that the existence of roots other than the three basic terms in Mayan is due to the vast changes that have occurred in Mayan societal complexity over the past millenia. But whatever one’s theoretical stance, if certain terms correspond both in form and in meaning between related languages, i.e., they show regular sound-meaning correspondences, such terms have to be reconstructed for their common ancestor, whether in the area of colour terminology or bio-nomenclature and other lexical domains. That is all one can say from the point of view of historical- comparative linguistics. A more serious methodological pitfall concerns so-called negative evidence. If a particular term is lost in all daughter languages, there is no way to retrieve the original form (unless there are written sources). For instance, if pastoralists lose their cattle and subsequently turn to hunting and gathering, the pastoral terminology may disappear from their language and there is no way of retriev- ing it. If such people are still in contact with neighbouring pastoralists such terms may still be retained, as with speakers of Kuliak languages in Uganda. Various pastoral terms can be reconstructed for Proto-Kuliak (Heine 1976b). which are not borrowed from neighbouring languages. If the original term for ‘iron’, for example, is replaced again and again in daughter languages, it can no longer be reconstructed. There is no way we can overcome such limitations of the comparative method. But the fact that a particular term cannot be reconstructed—and this constitutes the second major flaw in B&W’s reasoning— does not mean that the term was absent in the proto-language. For example, with regard to the Nilotic family of East Africa one cannot reconstruct a term for ‘white, light, bright’. But one would not want to conclude from this that the proto-language did not have any colour term for ‘white, light, bright’, since it is one of the two basic shades of colour attested in all modern languages. The problem is that this term has been replaced a number of times in individual languages and branches, as a result of which the original form can no longer be reconstructed, at least on internal evidence.2 Claims about the absence of

2 The only alternative left in such cases is to look at external evidence. In the case of Nilotic, for example, one could search for cognates in the closely related Surmic languages. If a widespread term found there is cognate with a form attested in a number of Nilotic languages (and if borrowing in these Nilotic languages is not a plausible scenario), this external evidence may still allow for the reconstruction of this term for Nilotic. 92 CHAPTER 5 such terms in the proto-language are not justified. Replacement can be caused by several different processes, e.g., semantic extension or narrowing of another term or borrowing. In a number of Nilotic languages the meaning of Proto- Nilotic ‘shine, be bright’ has been extended to ‘(be) white’. Once such a colour term becomes applicable to a wide range of items, it may become basic, and thereby replace the original term for ‘(be) white’. Even though ‘white’ is a ref- erentially basic colour term, its form is still replaced, in the same way as other colour labels. When certain members of a language family have fewer colour terms than others, and the cognate terms are absent rather than having shifted towards some other shades of meaning, we will not be able to reconstruct a more elaborate earlier system, simply because these terms have vanished. This could give rise to the erroneous claim that the number of colour terms in these particular members of the family increased, or did not change, neither of which are necessarily true. Kristol (1980), for example, shows that there are Italian dialects which have fewer colour terms than classical Latin. Only those still adhering to Gibbon’s magnificent but misguided synthesis of Roman history (in six volumes, published between 1776 and 1789), The History of the Decline and Fall of the Roman Empire, would want to claim that negative external stimuli gave rise to this simplification. We therefore have at least one case with written sources for the proto-language where “regression” took place, a clear case for a non- evolutionary approach. If there had been evidence for a unidirectional development, it would mean that our cognitive abilities have improved over the past few millenia. With regard to sensory functions such claims are simply unjustified, whether in the area of bio-nomenclature, or colour terminology. Often enough terms are not lost, but their denotational and connotational meaning changes, which complicates the discovery of lexical cognates. The semantic structure of languages is subject to change in the same way as are other parts of grammar. Semantic widening or narrowing may occur without being socially adaptive in any sense, nor does an external stimulus seem to be a prerequisite. It is often possible to reconstruct a root form without having unambiguous evidence for its meaning. This of course complicates the semantic interpretation of reconstructed forms, including the lexical domains considered here. With these limits of the historical-comparative method in mind, one should be extremely careful in interpreting historical “evidence” for a particular development of lexical domains. Moreover, for African language families like Bantu one can in fact reconstruct various terms in the domain of bio-nomenclature; this domain does not seem to have been less rich or complex than that of contemporary Bantu languages. There are therefore proto-languages for which a considerable time-depth may be assumed, and which provide positive Some Observations on Evolutionary Concepts 93 evidence for such terms as ‘tree’ or ‘grass’, as well as for rather extensive colour terminologies. Blust (1985: 895) criticises the way B&W manipulate historical data on the Polynesian language family. The implicational universals, as stated above, appear to be statistically well-grounded, but the evolutionary argument is gratuitous, and in certain cases demonstrably false. The next section argues that B&W’s implicational universals themselves require further adjustment from a synchronic point of view, because not only do perceptual considerations play a role but also functional and pragmatic ones.

5.2.2 The Empirical Basis: Synchronic Evidence and Counterevidence According to B&W, the order for the proposed implicational relations in the area of, for example, bio-nomenclature is essentially unidirectional. In their view, languages generally do not fall back to less elaborate systems (which are also assumed to represent earlier evolutionary stages). Instead languages only add terms, and they do this in the encoding sequence order. B&W suggest that the lexical encoding sequence for folk botanical life-forms bears a direct relation to the societal complexity scale referred to above. Those rare cases where a decrease can in fact be observed are, according to B&W, due to ecological changes. When people move to an area without snakes, the term may disappear from the set of bionomenclatural terms. Otherwise, lexical domains only expand, and the expansion itself is associated with the relative importance of the respective domains in societies of varying complexity. In order to test the observational adequacy of the W&B model, we will investigate three African languages spoken by hunter-gatherer communities which are sufficiently distinct both geographically and socially to exclude mutual contact. As hunter-gatherer communities are generally believed to represent the most archaic stages in the development of human subsistence economy, they should be representative of societies with lower societal complexity according to the B&W model (although such statements about the measur- ability of societal complexity are questionable in themselves). B&W did not consider the languages of these communities in their surveys. The first of these three groups is the !Kung, who live in the border area between Botswana and Namibia, and whose language belongs to Northern Khoisan. The !Kung are traditionally hunter-gatherers, and a major part of their diet consists of wild fruits and plants. In this respect they differ from other hunter-gatherer communities such as the so-called Central-African pygmies. Nowadays many !Kung people have abandoned the traditional way of life, and try to get work on farms. !Kung and Khoisan languages in general are sparsely documented. However, thanks to meticulous studies by Snyman (e.g., 1975) 94 CHAPTER 5 and others, we are now beginning to build up a more accurate picture of these languages. Lack of adequate language data easily leads researchers to make unjustified claims. Thus, Tanaka (1980: 48), in his discussion of a another Khoisan group claims that “the San have, as is usual among hunter-gatherers, no general term denoting ‘animal’, ‘plant’, or the like”, a claim which is unjustified. Snyman’s dictionary of one variety of !Kung makes clear that the language does have various bio-nomenclatural as well as colour and body part terms. The colour terminology domain contains terms not only for ‘white’, ‘black’ and ‘red’ but also ‘yellow’ and ‘brown’. The dictionary also shows that the language makes a classificatory distinction between ‘edible’ and ‘inedible’ for biological terms, a distinction which should be included in the analysis of !Kung taxonomy. One could argue that the counter-evidence presented here is not well- grounded statistically because B&W derive their evidence from a large corpus of languages. Counter-evidence from a few languages or language families does not necessarily dismantle a particular theory. It should be kept in mind, however, that !Kung and numerous other languages spoken in the developing world are disadvantaged simply because relatively little is known about them. This has led some researchers to the assumption that if nothing is known from the literature on a particular subject, nothing relevant exists. The lack of observational and therefore descriptive and analytical adequacy is an important problem in B&W’s attempts to make universally valid claims about folk taxonomies, since they base most of their evidence on available literature rather than on data they themselves collected. Also, where solid counter-evidence to their implicational and evolutionary claims exists, it has been ignored. Randall and Hunn (1984) raise this point, noting that the descriptive data on which B&W base their so-called universals have been subject to systematic reporting bias. An additional problem concerns the interpretation of the literature itself. As Randall and Hunn note, glosses often reflect an informant’s satisfaction that an anthropologist has learned an approximate meaning, but this may still be different from a correct definition of what a particular term actually refers to in the language. The second group considered here, the Okiek, are a hunting and honey col- lecting people who occupy zones in Western Kenya. Contrary to the !Kung, they live in an environment allowing for stationary residence. The modern Okiek are scattered over a large area, and sometimes the different groups are not aware of the existence of other communities speaking virtually the same dialect or language. There are basically three languages spoken by these groups generally known as the Okiek: Akie, Kinare and Sogoo. The languages are part of the Southern Nilotic branch of the Nilotic language family. Contrary to what Some Observations on Evolutionary Concepts 95 the B&W hierarchy leads us to expect, these languages have various terms relating to bio-nomenclature, and these are not recent innovations but reconstruc- table for the common ancestor of the language family of which they are part. Rottland (1982), in his comparative study of Southern Nilotic, reconstructs at least the following terms with regular reflexes in the Okiek group:

(1) Proto-Southern Nilotic Okiek reflex *kɛɛt keet-it ‘tree, wood, medicine’ *kɪsɪk kɪsɪk-ɛɛt ‘trunk’ *kwɛɛn kwɛɛn-ɪk ‘firewood’ *pɛɛr(t) pɛɛr-tɛɛt ‘bark’ *suus suuswe ‘grass’

In the same way Rottland reconstructs colour terms for Proto-Southern Nilotic such as ‘green’, ‘red’, ‘brown/black’, ‘white’, ‘black’, ‘ocre’, ‘yellow/brown’, again with reflexes in the Okiek group. In other words, these languages and their common ancestor have elaborate classificatory systems in these domains and not the basic system that B&W’s scheme suggests. The third group, the Aka pygmies, live in the tropical forest area of Congo Brazzaville and the Central African Republic. Aka is a Bantu language closely related to Nango, Pande and Mbate. The Aka are a foraging society. Contrary to South-African Khoisan groups, their economy depends on gathering wild fruit. Thomas and Bahuchet (1981) describe the language and culture of the Aka pygmies in great detail in their Encyclopédie des Pygmées Aka. The two volumes that have appeared so far are a rich source on the Aka language, particularly with regard to terminology related to the natural environment and culture, among which there are several terms for colour and bio-nomenclature. These original sources provide positive evidence regarding extensive vocabularies. Nothing of the elementary type that B&W’s unidirectional (and evolutionary) sequencing predicts is found in Aka. In many African languages the word for ‘tree’ is polysemous; it can also mean ‘medicine’, and such a root meaning ‘tree, medicine’ can be reconstructed for various language families, many of which have a considerable time depth. The claim that ‘tree’ has increased greatly in salience in time seems counter- intuitive, because trees are extremely important in building and in the prepa- ration of medicine. Accordingly, both historical-comparative and synchronic evidence suggest that such elaborate systems are not a recent innovation of these languages. African hunter-gatherer communities and their languages present a further significant case in point because they also illustrate that there is no 96 CHAPTER 5

unidirectional development of cultural stages. It would be wrong to assume that all present-day hunter-gatherers represent the earliest stage in the development of mankind. We know of pastoralist or agricultural people who have given up their means of subsistence, often due to ecological pressures, and have switched to hunting and gathering. A further modification required concerns the hierarchy itself. For example, Heine and König (1987) show that among the Samburu (Kenya) there are terms for ‘tree’, ‘grass’ and ‘weed’, but not for an all-inclusive taxon ‘plant’. ‘Vine’, ‘herb’, ‘shrub’ or ‘bush’ do not play any role in taxonomic groupings. ‘Weed’ refers to whatever does not fit into the category ‘tree’ or ‘grass’. As a category it tends to be disregarded as a term by men but not by women, presumably since it is of no use as cattle feed. In other words, there are also sex-based differences which are related to perceptual salience for different speakers. With colour terms among the Samburu and other pastoralists one often observes the opposite, i.e., males tend to use more terms than females, presumably because naming is strongly associated with cattle, which above all belong to the male realm. As Turton (1980) observes in a beautiful article on colour terminology among the Mursi of south-western Ethiopia, there is not a single colour term that does not also refer to shades of colour for cattle. A particular environmental domain, that of cattle, is used as a model “to represent differences between categories of colour and pattern which are universally recognised” (Turton, 1980: 320). Cattle are extremely important to the Mursi. Their colours and shades are used as a model for encoding or as a system of denomi-nation. Turton (1980: 334) further suggests that technologically simple cultures are more likely to use a single, highly salient and unitary model for colour naming than technologically advanced cultures as further discussed in Chapter 6. To sum up, B&W’s evolutionary sequencing hypothesis breaks down for a number of reasons:

1. There is no unambiguous diachronic evidence for an increase in the number of terms distinguished in B&W’s encoding sequences. 2. The correlation between the so-called societal complexity and number of terms can be falsified synchronically. 3. Functional and pragmatic criteria (which are culture-specific), not just morphological criteria (i.e., shape), play a role in the way people denominate their environment, for example in bio-nomenclature. This is true for the distinction between ‘edible’ and ‘inedible’ as focal terms but also for other dichotomies. Many pastoralists in East Africa distinguish between ‘domesticated animal’ and ‘wild animal’ by way of monolexemic units. Such terms are as much part of the basic taxonomy as ‘snake’ or ‘fish’. Some Observations on Evolutionary Concepts 97

5.3 An Alternative Account: Language and Cognition

The history of ideas in anthropology shows ample cases of attempts to relate social structure, law and religion in all its diversity to evolutionary stages in the cultural development of mankind; a link to linguistic structure is of course but a small additional step. Nineteenth century social scientists like Morgan assumed that cultural development went from a stage of savagery (hunting and gathering) through barbarism (pastoralism and agriculture) to civilisation. Morgan also tried to relate the evolution of technical knowledge to other cultural spheres, particularly social and family structure. The latter was supposed to have developed from “unlimited promiscuity” via matriarchical and patriar- chical polygamy to monogamy. According to Frazer, totemism represented the proto-religion, which then developed via a stage of polytheism to monothe- ism. The underlying assumption in all instances was a direct functional link between technological development and social structure. A functionalist or structuralist approach has now replaced this strongly evolutionary perspective. The question remains to what extent social structure in all its diversity is adaptive, and whether there is a dependency relation with, for example, linguistic structure. B&W’s observations on the hierarchical organisation of lexical domains are important, but their presumed direct stimulus-response relation between the cultural environment and the lexicon is misguided. Furthermore, their proposed implicational relations between the presence or absence of specific terms may not have universal validity in the case of bio- nomenclature. Bio-nomenclature is constrained by the species pool of the environment in which speakers live; this might also be interpreted as an adaptation in an evolutionary sense. A further necessary modification of the B&W hypothesis is that languages not only add categories or terms in a specific order (otherwise no such synchronic regularities between languages would exist), but that terms can also be lost. Both synchronic and diachronic evidence therefore falsifies B&W’s unidirectional hypothesis and its correlation with so- called cultural stages. In the case of colour and body part terminology, culture independent perception seems to be involved. There is probably a neuropsychological basis for the observed hierarchy. Here again B&W’s hierarchies should be reinterpreted in a non-evolutionary, atemporal and anticausal way. If one calls B&W’s approach “behaviouristic”, the alternative interpretation may be referred to as “cognitive‟. Andersen’s (1978) cross-linguistic studies show that the domain of human body part terminology “tends to be organised into a hierarchical structure with five (or occasionally six) levels”. This partonomy has the following structure for the domain of ‘arm’ and its sub-parts: 98 CHAPTER 5

body

head trunk arm leg

forearm hand

palm finger

fingernail figure 5 Body part partonomy

This categorisation or naming of body parts reflects ways of ordering the visu- ally perceptible properties of shape (e.g., round, long) and of relative salience. These are directly related to the neuro-physiology of perception.3 Andersen also makes it clear that there are more terms for the upper front parts of the human body than for other parts, a fact related to the spatial orientation of the human perceptual apparatus. Lexical regression or expansion seems to follow the pattern of relative salience, as with colour terminology to some extent, otherwise no cross- linguistic regularities and implicational universals could be formulated. When ‘hand’ is not labeled as a separate body part, as in many African languages, the term ‘arm’ refers to the limb which starts near the shoulder and ends at the point where the digits meet the hand. If the separate term for ‘hand’ disap- pears, speakers refer to the particular body part as ‘(part of the) arm’, and so on. The presumed direct link with cultural stages is not supported by empirical evidence. However, even if there is no direct link with culture in the case of body part terminology or colour terms, one would still like to know why and when changes take place in the lexicon, other than as a consequence of the minimalistic property of language that the lexicon must fit the needs of its speakers and may be changed accordingly—a necessary but not a sufficient condition for change. Language is utilitarian but also full of non-functional aspects, as are other biological systems. It is well-known from evolutionary

3 See, for example, Marr’s (1982) fascinating modular theory for visual perception which claims that our visual system depends on a number of autonomous modules operating independently from knowledge about the world. Some Observations on Evolutionary Concepts 99 theory that there are mutations which occur without being causal, functional or adaptive. Although such a lack of correlations with other stimuli may seem less exciting intellectually, at least to some of us, it is more in accordance with empirical observations with regard to changes in lexical domains. The ultimate answer to the question why certain changes took place may have to be that we just don’t know. There may be numerous reasons for the lack of correlation between communicative needs and what one finds in real language. For example, the con- stant interaction between speakers of contiguous languages may encourage retention or innovation of terms, i.e., hierarchically organised domains such as colour may expand or decrease through diffusion. Such developments should be interpreted from a geographical perspective, albeit with a historical dimension. Convergence phenomena are known to occur at the phonetic, morphological and syntactical level but they also occur semantically. Thus, in many languages of East Africa which are not necessarily related historically, no basic distinction is made between terms for ‘blue’ and ‘green’. Another fundamental reason why we no longer find archaic stages may be the fact that groups interact and shift their language solidarity. The Okiek hunter-gatherers seem to have given up their earlier language when they came into contact with Southern Nilotic pastoralists. In adopting the new language, they also took over the system of bio-nomenclature, colour, and body part terminology. There is ample evidence that shift in language solidarity has occurred and is still occuring again and again in the cultural history of Africa and elsewhere. Although not in the area of body part terminology or colour, some differences can in fact be related to cultural and ecological factors in the visual environment. This cultural relativism finds support in the perception of geometric or optical illusions (Segall et al. 1966). Such illusion-producing habits have to be acquired via experience. Another way in which culture and linguistic signs (i.e., lexical items) may interact relates to the nature of semantic shifts, as the final section suggests.

5.4 Conclusions and Prospects

At least in the past, the general idea behind the synchronic study of folk taxonomies seems to have been motivated by the wishful thinking that more primitive stages of nomenclature could be discovered through such analyses. So far the results do not seem to justify an evolutionary interpretation of observed 100 CHAPTER 5 differences between languages from a simpler to a more complex state of development in lexical domains. In other words, the synchronic study of languages is not going to reveal earlier stages.4 Recent studies in figurative language and metaphor have shown that universal principles of sensory perception are reflected in figurative speech, which admittedly Brown and Witkowski (1981) do observe. There are cross-language uniformities with regard to figurative names for certain body parts, presumably based on common cognitive principles. Interestingly enough, there is also evidence for culture-specific semantic extensions and metaphors, which cannot be understood without knowledge of those speaking the languages. Variational semantics and its relation to our cognitive system on the one hand and culture-specific cosmologies on the other seem to be an area where interesting correlations between language and thought may be found. With regard to the synchronic comparative study of language, culture-specific (rather than universal) principles may be discovered for the type of semantic ranges that certain lexical items may have. In an article on the semantics of verbs in Kwaio, spoken on the Solomon Islands, Keesing (1979) shows how the tran- sitional use of a series of verbs rests on cultural particularities and symbolic structures shared by Kwaio speakers. These lexically labeled chunks reveal information about the structure of the perceptual and conceptual world of the Kwaio, i.e., about Kwaio cosmology. The symbolic associations are culture- specific because their readings depend on symbolic structures and the religious premises of Kwaio culture. For example, the Kwaio verb meaning ‘jump on’ has an additional meaning ‘taking possession of a person’s mind by an ancestor)’. Although Keesing does not raise this point, one could well imagine that semantic changes occur whereby the concrete meaning of the verb is lost; such lexical changes from the concrete to the abstract are extremely common in languages. If we are looking for features that differentiate languages and people rather than those that bind them (as in the Generative Enterprise inspired by Chomsky), variational semantics of the type mentioned above seems to be an area of potential interest. Most likely, the emerging results will not correspond to the Eurocentric and simplistic Western/non-Western technological dichotomy, nor to the trichotomy of foraging, pastoral and agricultural economies. Rather, more subtle, intricate and no doubt more interesting differences should emerge, as the following two chapters illustrate.

4 This does not imply that an investigation of the phylogeny of human language should be dis- missed altogether; there have been fascinating syntheses of current knowledge of the ontogeny and phylogeny of human language and that of other creatures (e.g., Lieberman, 1984). CHAPTER 6 Studying Lexical-Semantic Fields in Languages: Nature Versus Nurture, or Where Does Culture Come into It These Days?

The study of the interaction between language, culture and cognition has a long tradition. It is now generally accepted that different speech communities may have different styles of speaking, reflecting differences in cultural values. Reflexes of such cultural behaviour may be found in, for example, verbal taboos. This type of research, usually subsumed under the heading of “ethnography of speaking”, is now well established in particular through the work of scholars such as Bauman, Gumperz, Hymes, Irvine, and Sherzer. The kind of research that is the focus of this chapter is more controversial. It involves the investigation of lexical-semantic fields, in particular those related to colour, the expression of space and direction, kinship terminology and the like. It is also concerned with language and thought interactions as cognitive processes; but at times authors hold diametrically opposed views when it comes to the evaluation of the relative role played by culture in the structuring of such semantic fields. Below, I first sketch the historical background of this type of research, and show how it has received new impetus over recent decades from research on language typology and universals. Somewhat paradoxically I then defend the view that, as a result of this more recent research paradigm, we may have fallen again into a “Whorfian” trap in our attempts to explain cross-linguistic variation.

6.1 Investigating the Interaction between Language and Cognition: Research on Colour Terminology

There is by now a huge bulk of literature on the interaction of language and cognition deriving particularly from the scholarly work of American and Russian researchers; see, for example, Rice (1980), and the influential work of Vygotsky (1962). Vygotsky examined the crucial role of language in mediating the growth of cognitive abilities in young children; it was his view, for example, that external, social reality interacts with the elementary mental functions of

© koninklijke brill nv, leiden, ��5 | doi ��.��63/9789004224148_007 102 CHAPTER 6 young children through language. (See Luria 1981 as a further example of such research from the time of the former Soviet Union.) Modern research on the interaction between language and cognition, using experimental methods, is generally assumed to have started in the 1950s when Brown and Lenneberg (1954) and Lenneberg and Roberts (1956) published the results of their experimental research. These American psychologists were aim- ing at objectively characterising and comparing language categories in terms of their denotational referents. Brown and Lenneberg, for example, decided to investigate the interaction between language and cognition through research on the description of colour, since colour was assumed to be a universal phenomenon, having boundaries which could be plotted on known dimensions, and which were contiguous with one another, sharing their boundaries. Actual scientific interest in colour terminology dates back to at least the nineteenth century. The German researcher Geiger (e.g. 1871, 1872) proposed what was probably the first “cross-cultural” evolutionary theory of colour names. Geiger assumed an additive progression of at least six stages. First, ‘black’ and ‘red’ (as meta-categories) were named, suggesting a vague conception of something coloured. Second, ‘black’ and ‘red’ stood in contrast to one another. Third, ‘yellow’ was registered. Fourth, ‘white’, previously included in ‘red’, was distinguished. Fifth, ‘green’ developed from ‘yellow’. Sixth, and last, ‘blue’ developed. Geiger argued that so-called “primitive” people had fewer colour names because they were physiologically underdeveloped. For Geiger, having no word ‘blue’ meant that the person could not see blue. A fellow German physical anthropologist and contemporary, Virchow, con- tested this view. He examined speakers of varieties of Nubian (a group of languages now classified as Nilo-Saharan), living in Berlin during the same period, who “although displaying anomalies of colour vocabulary were quite able to discriminate colours in all parts of the spectrum, demonstrating this by sort- ing and matching coloured papers and wools.” (See Andree 1878 and Kirchhoff 1879 for a discussion.) In their experimental studies of colour terminology seventy-five years later, Brown and Lenneberg (1954) used the so-called “Munsell Chart”, a stimulus array consisting of 329 colour chips. They defined the category of colour as being composed of three psychophysical dimensions or perceptual attributes:

1. hue (intensity) 2. value (lightness, dominant wavelength) 3. chroma (saturation, purity) Studying Lexical-Semantic Fields in Languages 103

“Colour” itself was not an issue for Brown and Lenneberg, it was presup- posed. Only variations in the manner of dealing with what one conceives of as “colour” was at issue, using English categories as the metalanguage for describing so-called objective reality. Brown and Lenneberg interviewed speakers of American English and Zuni for their experiments. One of the conclusions of their research was that there was a relationship between codability and rec- ognition of colours among those interviewed. This type of research would set the trend for investigations in the domain of colour terminology for the next decades.

6.2 Language Typology and the Study of Language Universals

During the 1950s and 1960s, the American anthropologist and linguist Joseph H. Greenberg not only worked on his seminal classification of African languages (Greenberg 1955, 1963), he also developed an interest in psycholinguistic experiments. In fact, Greenberg himself was one of the first linguists to do experimental psycholinguistic research in cooperation with psychologists. (See, for example, Greenberg and Jenkins 1964.) But above all Greenberg’s name as a scholar is associated with quintessential observations on (implicational) universals of language. Apparently, the Cross-Cultural Survey initiated in 1937 by the Institute of Human Relations at Yale University provided the real impetus to Greenberg’s own work on linguistic universals and typology.1 The most powerful results of this method were to be published in what is now a classic article on implicational universals in the area of morphology and syntax (Greenberg 1966). Greenberg, as well as others, also contributed ideas on implicational universals in lexical-semantic fields, e.g. with regard to kinship terminology, the development of numeral systems, and body part terminology. (See, for example, the collection of studies in Greenberg 1978; cf. above e.g. (Greenberg 1966).) An additional lexical domain where researchers tried to apply Greenberg’s methods of typological comparison, by using data from a wide variety of related and unrelated languages, was again colour. Whereas Brown, Lenneberg and Roberts had provided the blueprint for the main features of this tradition, Greenberg provided the typological perspective.

1 See Newman’s (1991) interesting interview, in which Greenberg points out that his typological research was also inspired intellectually by linguists like Wilhelm Schmidt (e.g. his 1926 book Die Sprachfamilien und Sprachenkreise der Erde). 104 CHAPTER 6

6.3 The Berlin and Kay Framework

In their seminal work Basic Color Terms, Berlin and Kay (1969) initiated a second era of colour research by using a Greenbergian typological framework. Berlin and Kay question the fundamental assumption of most previous colour research, namely that colours form an even perceptual continuum. Instead, they claim that the human visual system sets constraints on the range of variation in semantic structure, i.e. on universal colour space. Their measuring device again was the Munsell Colour system. Berlin and Kay conclude that, in the area of colour terminology, language structure and cognitive structure are determined by underlying perceptual regularities, more specifically by focal colours. On the basis of their sample studies, they arrive at the following conclusions:

1. There are eleven universal colour categories which are the referents of the eleven or fewer Basic Colour Terms in any language. 2. Encoding the perceptual categories into Basic Colour Terms follows a fixed, partial order (with seven evolutionary stages) in the history of any language (whereby ‘grey’ may emerge as a ‘wild card’ at any stage). black green yellow purple redblue brown pink white yellow green orange grey figure 6 Berlin and Kay’s Basic Colour Terms

3. The temporal order is an evolutionary one: Few colour terms occur in association with simple cultures and technologies; colour lexica with many terms tend to occur with complex cultures and complex technologies.

“Basic” in Berlin and Kay’s terminology implied that a colour term is monolexemic, not included in any other colour term, and that it should refer to a wide range of objects. Moreover, it should also be psychologically highly salient, easily elicited and appearing at the beginning of elicitation lists. The Berlin and Kay framework inspired several authors to test the proposed hierarchy for colour terminology in African languages other than those already referred to in Berlin and Kay (1969), for example Nyangatom (Nilotic, Ethiopia, Tornay 1973); Gusii, (Bantu, Kenya, Whiteley 1973); Luganda (Bantu, Uganda, Pollnac 1975); Kapsiki (Chadic, Cameroon, van Beek, 1977) and Bini (Edoid, Nigeria, Wescott 1970). Studying Lexical-Semantic Fields in Languages 105

6.4 Some Problems with the Berlin and Kay Model

Over the years, a number of problems have emerged related to the methods and results of Brown, Lenneberg and Roberts’ experimental research, and Berlin and Kay’s further elaboration of it. Errors of representation naturally creep in whenever large amounts of data are to be processed, as anybody who has undertaken such research will confirm. Thus, Berlin and Kay (1969) place languages such as Maasai (p. 85) and Nandi (p. 98) in Ethiopia, rather than Kenya and Tanzania (for Maasai) and Kenya (for Nandi), and they place Somali (p. 66) in Chad, rather than in Somalia. Such problems seem hard to avoid. But there are more fundamental, methodological problems with the Berlin and Kay model, which start already with data gathering. Saunders (1992: 147) informs us that in her research on colour terms among the Kwakiutl (Kwa’kwala) in Canada, the collection of colour names for plants, animals, beads and sequins was unproblematic. But the introduction of the colour chart at the end of each interview caused discomfort, anxiety and agitation among her language helpers because of the decontextualisation of colour terminology. Apparently, all but one person was reluctant to continue, although there was never total refusal to cooperate. I myself have had similar experiences when eliciting colour names in the Nilotic language Turkana (Kenya), or the Surmic language Chai (in Ethiopia). Among the Turkana or Chai, as well as other neighbouring pastoral groups for that matter, names for colours do not really mean anything outside the realm of cattle terminology. As the anthropologist Lienhardt (1970: 13) observed for the pastoral Dinka:

If they were to have their cattle-color terminology taken away they would have scarcely any way of describing visual experience in terms of color, light and darkness.

Accordingly, the Munsell Colour chart, presented as the neutral, universal and value-free referent of colour is really a translation machine. It is a device which transmutes indigenous meanings onto a single standard which decrees that English abstract colour terms have absolute ontological independence (Saunders 1992: 188–189). In other words, the lexical meaning of colour terms in languages are reduced to denotational meanings, more specifically the denotational value of English lexical items, which are used as the metacatego- ries for comparative work. Turton (1980) extends Lienhardt’s (1970) observations about the Nilotic language Dinka to Mursi, a Surmic language of southwestern Ethiopia closely 106 CHAPTER 6 related to Chai (referred to above). As Turton points out, there is not a single colour term in Mursi that does not also refer to shades of colour for cattle. In other words, cattle serve as a model to represent differences between categories of colour and patterns which are universally recognised. Turton arrives at the following structure for Mursi colour terms:

(1) koroi ‘black (including some blues)’ holi ‘white’ golonyi ‘red (brown with preponderance of red; reddish-brown, when applied to cattle)’ biley ‘yellow (brown with preponderance of yellow; yellowish-brown tan, when applied to cattle)’ chagi ‘green-blue (slate grey; bluish-grey; ash coloured, when applied to cattle)’ rege ‘pink (flesh-coloured)’ lulumi ‘brown’ gidangi ‘grey (dirty white; grey, when applied to cattle)’ sirwai ‘blue-violet (a combination of reddish-brown and black, the two colours shading into one another to give a satiny lustrous appearance, when applied to cattle)’.

In addition, some speakers use hurai or lele for certain shades of colour. (See Turton 1980: 326 for further details.) As this list of Mursi terms shows, the contrast between the so-called primary hue categories ‘green’ and ‘blue’ is overridden in favour of the less salient ‘red/ brown’ and ‘red/pink’ hue contrast. Whereas Mursi has a single colour term corresponding to the English focal colours ‘pink’ or ‘grey’, colour distinctions “lower on the hierarchy”, such as ‘green’ and ‘blue’ are not distinguished in this language. The reason for this property (which is widespread in East African languages spoken by pastoralists) appears to be that terms for focal green and focal blue cannot be “anchored” in the cattle model.2 Consequently, there is a “hole in the hierarchy” proposed by Berlin and Kay (1969).

2 See also Tornay (1978b) for a description of similar phenomena in the neighbouring Nilotic language Nyangatom. Note that it is quite possible for a culture to extend the usage of such terms to domains that may not be obvious to speakers of Standard Average European languages. Bender (1983), for example, observes that the colour terms for ‘green’ and ‘blue’ in northern Sudanese Arabic may also be applied to the perception of human skin colours. Bender based himself on the work of a B.A. student in linguistics from the University of Studying Lexical-Semantic Fields in Languages 107

A comparable issue emerging from the proposed hierarchy concerns early studies of Amerindian languages. Boas (1891) researching the Northwest Coast languages observed a common pattern in which ‘light blue’, ‘light green’, and ‘yellow’ were interlaced in diverse ways (“yeen”). The yellow-green “confusion” has also been reported for various African languages, in particular those spoken in the Sahel, as further discussed in Chapter 7. The nineteenth century explorer Gustav Nachtigal already stressed the inclination of certain Central African groups to unite ‘yellow’ with ‘green’ and also with ‘red’. (See Kirchhoff 1879 for a summary of observations by early investigators.) A further problem concerns the criteria for “Basic Colour Terms”. Hickerson (1971) points out that a stringent application of Berlin and Kay’s criteria could reduce a language from Stage vii to Stage ii. Wescott (1970) also demonstrates the multitude of problems which can arise in applying the Berlin and Kay criteria for basicness in his analysis of Bini (Nigeria). Depending on one’s criteria, one could reduce a language seemingly rich in colour vocabulary such as Bini to the lowest classificatory level, Stage i. In his analysis of colour categories in Karam (Papua New Guinea), Bulmer (1968) concludes that there are 14 or 15 categories. “At the same time no Karam concept has been recorded which approximates to that of ‘colour’ in English. While some Karam terms exist which can be glossed reasonably accurately by reference to the dimensions of hue, saturation (or chroma) and value (or brightness) which adequately specify the domain of ‘colour’ in English, other Karam terms which fall in the same contrast sets involve use of additional dimensions including unripeness/ripeness, succulence/dessication, dirtyness and patterned contrast” (Bulmer 1968: 131).3 Instead of a hierarchical taxonomy for representing Karam colour categories, a representation in terms of “target areas” or “semantic foci” would be more appropriate, according to Bulmer, who further notes that “[. . . c]onsider- ations of polysemy, together with some statements by informants themselves about the nature of their categories, suggest that this model has some psychological validity” (Bulmer 1968: 131).

Khartoum, Ahmed el Tinay, himself a speaker of Sudanese Arabic. This phenomenon is, of course, not totally absent from languages such as English, where ‘red’ may be used as a descriptive term for the colour of hair. 3 The case of Karam is multiplied elsewhere, for example Doumes’ (1978) description of Jörai (Vietnam) in which there are 23 basic colour terms but no single word that translates as ‘colour’. 108 CHAPTER 6

The Berlin and Kay framework also largely ignores a language’s grammatical formations. Whereas it is not immediately clear from a general linguistic point of view that the categorial status of terms referring to shades of colour is relevant, there are a posteriori reasons for assuming so. The Turkana colour system is structurally similar to that of the Nyangatom and Mursi, who speak related languages and who live northeast of the Turkana. Turkana uses the following terms:

(2) -kwaŋ ‘white’ -iryono ‘black’ -rɛŋ ‘red’ -ɲaŋ ‘yellow’ -pʊs(ɪ) ‘blue’ -ɔŋɔr(ɪ) ‘brown’ -kɪpʊra(t) ‘pink’ -mug(i) ‘sable’

The English translations of these terms are taken from Barrett (1988, 1990). A closer look at the functioning of these terms shows that not all of them occupy the same structural position in Turkana. Three of them, ‘white’, ‘red’ and ‘black’, are clearly verbs in their morphologically simplex form; the remaining ones are adjectives. In other words, the former are more commonly used as predicates expressing referent modification, whereas the latter are more commonly used as attributives expressing reference modification in the sense of Bolinger (1967). Correspondingly, the primary, proto-typical meaning of a verb such as -kwaŋ is not a stative situation, that of ‘being white’, but a dynamic property, ‘to shine, glitter, flicker’; the dynamic Aktionsart of this verb is further shown by its capacity to take an aspectual suffix -it in Turkana:

(3) -kwaŋ ‘shine, glitter, be white’ e-kwaŋ ‘it shines, it is white’ e-kwaŋ-ɪt ‘it is shining, glittering’

The verb ‘be black’, -iryono (singular) / -iryoko (plural), contains a petrified causative prefix (-i-). This suggests that its original meaning was ‘caused to be black, blackened (as with charcoal)’, again a more dynamic meaning. The origin of the verb root for ‘red’, -rɛŋ, is not known. The three colour terms for ‘white’, ‘red’, and ‘black’, expressing degrees of brightness, are verbal (and thereby dynamic in nature), whereas the remaining colour terms are adjectival. The tripartite Studying Lexical-Semantic Fields in Languages 109 division itself between ‘white’, ‘red’ and ‘black’, is widespread across Africa (see for example Wescott 1970 on Bini), but it is also found elsewhere.4 In her reconstruction of Proto-Polynesian colour terminology, Branstetter (1977) notes that the terms for ‘white’, ‘red’ and ‘black’ show little variation across the family. The importance of this tripartite division is well known from anthropological studies (Turner 1966, Tornay 1978a). The remaining Turkana colour terms are adjectives, whose proto-typical meaning includes that of an inherent and permanent property or state. Special lexical-derivational modification is applied to express these concepts as predicates. (Other such adjectival concepts expressing irreversible states in Turkana are notions like ‘bald’, ‘orphaned’.)

Adjective Verb (4) -ɲaŋ ‘yellow’ -ɲaŋaana (sg) ‘be yellow’ -ɲaŋaaka (pl) -mug(i) ‘brown’ -mugyaana (sg) ‘be brown’ -mugyaaka (pl) -pʊs(ɪ) ‘grue’ -pusyaana (sg) ‘be grue’ (green+blue) -pusyaaka (pl)

Over the years, there have been a number of revisions of the original Berlin and Kay framework (e.g., Kay and McDaniel 1975, 1978; Kay, Berlin and Merrifield 1991). Following developments in semantic theory, the authors of the notion “Basic Colours”, for example, incorporated proto-type theory, and categorisation in terms of “fuzzy sets” (rather than foci or discretely contrasting semantic features) in their representation of semantic categories. They assume that the latter are formed by “fuzzy logical operations” of “fuzzy union” and “fuzzy intersection”, resulting in so-called “composite” categories. These are unions of two or three fuzzy sets, i.e. “[. . .] named color categories that comprise more than one of the colors that have been established by neurobiologists, independently of language, to correspond to fundamentally distinct neural response processes”. These fundamental neural response categories correspond to the denotations of the English words black, white, red, yellow, green and blue and their translations into many other languages (Kay and McDaniel 1978). The authors refer to De Valois, Abramov and Jacobs (1966) for a neurophysiological underpinning of their statements.

4 As MacLaury (1992: 159) points out, “[. . . b]rightness may be a more general and impression- istic aspect of color perception that sweeps across the domain at large [. . .] whereas unique hues are located within confined areas of the color space.” 110 CHAPTER 6

According to Kay, Berlin and Merrifield (1991), only nine (out of 63 logically possible) types of composite colour categories are attested, including ‘yellow/ green/blue’ and ‘white/yellow’. As pointed out above, composite categories or “macro-colours” consist of various combinations of the basic hues (‘red’, ‘yellow’, ‘green’, and ‘blue’) and ‘black’ and ‘white’; macro-colours can have variable foci, both intra-societally and cross-societally. But this more recent approach summarized in Kay, Berlin and Merrifield (1991) and based on the World Color Survey with data from 111 languages, cannot solve or explain the “gaps” observable in the colour terminology of languages such as Mursi, or languages of the northwest coast of America. The problem with ‘yellow’ is that it may be associated in a composite category with (1) ‘red’ or ‘white’ or both, or (2) ‘green’ and ‘blue’ and so on. The proposed neurophysiological wiring is embedded in the opposition of “warm” and “cool” colours: ‘yellow’ and ‘red’ are “warm”, ‘green’ and ‘blue’ are “cool” colours. Kay, Berlin and Merrifield (1991: 18) in fact admit that no explanation is available for these problematic cases. Some might argue that universalists are interested in more widespread patterns, leaving “exceptions” to be further investigated, and possibly explained, by butterfly collectors. But the issue is not simply one of diverging academic interests; it is of far more fundamental significance. The emphasis on common patterns has led to a negligence of the possible (cultural) significance of the differences as manifested in the referential meaning of colours. The referential meaning of words manifests itself as a culture’s categorisation of the world. This is also true for colour terminology. The Mursi system is based on the various colour patterns and textures of cow skins; the meaning of colour terms in the non-bovine world is a derived one (Turton 1980). And the two are not necessarily isomorphic when speakers of different languages are asked to apply the terminology, as the following example illustrates. In Chai, a Surmic language of southwestern Ethiopia which forms a dialect cluster with Mursi (as well as Tirma(ga), one finds a colour system virtually identical to that of Mursi. The ownership of cattle and a corresponding bovine idiom is also central to Chai culture. Note Last’s (1995: 113) interesting observation in his sketch of Chai:

The division between bovine and non-bovine with respect to color terms is indeed striking: the term /cá!gí/ refers to the color of grass and trees, and to the color of the sky. If applied to cows, however, it denotes a particular shade of what I would probably call brown or reddish. Nevertheless, an attempt to describe an identically colored piece of cardboard as /cá!gí/ met with laughter from the side of our informants, since cardboard Studying Lexical-Semantic Fields in Languages 111

is not bovine. When referring to an object in the non-bovine part of reality, this particular color would rather be described as /gìda!ŋí/ ‘reddish, brownish, grey, dirty, whitish (of cows only)’ or /gòlèɲí/ ‘orange, red, pink, violet, reddish-brown (of cows only)’. To be honest, I have proved myself rather incompetent in distinguishing between /cá!gí/ and /gòlèɲí/ cows, despite the fact that I find myself able to distinguish between green and red in the non-bovine universe.

The statement on “identity” of course follows from the cultural categorisation of colour that Last himself is most familiar with, i.e. that of Dutch and possibly English. There may be many more characteristics besides brightness or hue, for example, texture, saturation, or pronouncedness (van Brakel 1993: 113), a conclusion which seems hard to avoid when considering the use of colour terminology in Mursi or Chai, for example. Clearly, the way in which colour terms are used or not used says absolutely nothing about the general cognitive ability of people to generate linguistic expressions for colours. This may be further illustrated with material from the widely quoted case of the Dani language of Irian Jaya. In the early Berlin and Kay framework, this was one of the few languages assumed to be a representative of an (evolutionary) early system with only “two basic colours”, ‘white’ and ‘black’. Rosch (1972a, 1972b) has shown that the Dani terms mili and mola were not brightness terms with foci in ‘black’ and ‘white’, but categories best classified as ‘dark-cool’ and ‘light-warm’. Mili refers to ‘dark, dull, dark-cool’, and also ‘bad’; mola refers to ‘light, bright, light-warm’, and also ‘good’. These terms, however, comprise but a small subset of the Dani speech community’s resources when talking about colour. They may use expressions such as gut, ‘white heron’ (also used to describe ‘albinos’); oagik, ‘cockatoes’ (also used for ‘white clay’); bimi- tet ‘mountain dove’ (also used for the rusty colour of some pigs); or getega, ‘bluish adze stone’ (also used for something that is greenish) etc.5 Despite the large differences between the colour vocabularies of the Dani people and, say, speakers of American English, there are no “cognitive” differences between the two which could be ascribed to differences in their “colour languages”, despite their positions at opposite ends of Berlin and Kay’s evolutionary typology of colour vocabularies. Here, then, is another manifestation of how a strict validation of Saunders’ (1992) notion of “Basic Colour Terms”

5 Rosch (1972b) also concludes that memory for colour does not vary with changes in linguistic code. But Lucy and Shweder (1979) conclude that language does in fact play a role as a factor in colour memory and codability. (See also Davies et al. 1991.) 112 CHAPTER 6 fails.6 The mere fact that not all languages have a word for ‘colour’ should at least be a warning not to treat the connection between the lexical item and the object or referent as a socially decontextualised “natural categorisation of nature”. Evidently, areal and language-specific or culture-specific differences or values do not override “neurophysiological constraints”. But the cultural meaning of colour would seem to be at least as insightful as the so-called neurophysiological constraints are for the way in which we as humans perceive, and more importantly—at least from a linguistic point of view—use colour. It is therefore rather unfortunate, to say the least, that philosophers of science such as Hardin (1993) should dismiss the potential for ethnocentrism inherent in the methodological pitfall of letting European languages and cultures contaminate experiments (van Brakel 1993). Hardin assumes that the dominant Berlin and Kay paradigm regarding the study of colour terminology is that of a “not so naked emperor”, thereby showing little sensitivity to cultural relativism. As should be clear from the discussion above, there is not only an imminent danger of subjectivism in our scientific endeavours: ethnocentrism has already had its effects on the so-called objective investigation of colour. This danger, and its potential effects, is not just restricted to the investigation of colour terminology, as we shall see below.

6.5 Extending the Greenbergian Framework to Other Lexical Domains

As the principal instigator of new methods of dynamic comparison in linguistics, Greenberg himself extended the study of implicational scales in lexical- semantic fields to such domains as numeral systems (Greenberg 1978) and kinship terminology (Greenberg 1980). Of course, he was fully aware of the potential complexity of explanatory mechanisms behind the observed diversity:

[. . . kinship studies] provide the possibility, through the analysis of highly specific and formally manageable phenomena, of disentangling the contribution various causal factors (social, historical, psychological, linguistic and, I should add, evolutionary) make to an important cultural phenomenon. (Greenberg 1980: 31)

6 See also the comments by various authors in Current Anthropology 33(2) on MacLaury (1992) in this respect. Studying Lexical-Semantic Fields in Languages 113

Kinship terms are genealogically based, but not all terms are necessarily biologically defined. Some anthropologists therefore argue for a social category approach. This would bring the important role of culture or environment back into the picture. Greenberg (1978) has also arrived at a number of interesting generalisations regarding numeral systems in languages, for example that the existence of multiplication implies the existence of addition, as well as several other implicational universals. But here it would seem that areal diffusion, as an explanatory mechanism for the observed diversity or lack of it, comes into the picture as well. Whereas decimal or base-five systems are widespread across Africa, there are areas where alternative systems are found. Greenberg (1978) mentions the duodecimal systems of Plateau Benue-Congo languages in Nigeria, which are being replaced by a decimal system through the influence of the Chadic language Hausa (and the British education system one might want to add). Kutsch Lojenga (1994: 355–58) describes the traditional base-four system of Ngiti, a Central Sudanic language spoken in northeastern DR Congo, that apparently only a few old people remember. A full-scale inventory of these alternative systems in Africa and elsewhere could provide interesting information on the diffusion of numeral systems, now resulting in a decrease in diversity.7 In the area of body part nomenclature, Andersen (1978) shows that languages tend to have more terms for the upper parts of the body than for the lower parts, and more for the front than the back. She also observes that there are implicational scales; for example, if a language has a separate word for ‘toe’, it also has word for ‘finger’. But languages may use the same term for both body parts, or refer to ‘toe’ as the ‘finger of the foot’. Again, certain authors claim that there are evolutionary stages in these implicational scales. Brown (1976: 404, 420) claims that ‘body’ is labelled in all body part taxonomies. But it appears not to be labelled in some Papuan languages, where a single term exists for ‘skin’, ‘trunk’ and ‘person’ (Wilkins 1993: 9).

7 Interviews with a speaker of the Cameroonian Bantu language Baca for example, in a field methods class at Leiden University in 1992, revealed some interesting aspects of its numeral system requiring further investigation. Baca uses the widespread Bantu roots for ‘two’, ‘three’, ‘four’, and ‘five’. But ‘six’ and ‘seven’ are based on the root for ‘three’ (-tát), namely bì-cí-ndát, ‘six’, and bì-cí-ndát nàwùmé, ‘seven’; the latter is also based on the root for ‘three’ as contained in the numeral for ‘six’, plus the root for ‘one’. I would like to thank Marie-Thérèse Harleman- Ndobo for her willingness to share her knowledge of Baca with the students. 114 CHAPTER 6

Brown and Witkowski (1981) show that certain body parts tend to have simple unitary labels (e.g., ‘eye’, ‘ear’) whereas others are complex (e.g., ‘pupil of the eye’). They further show that cross-linguistic regularities occur in the use of figurative language in body part labelling. These are built on perceived similarities and associations between body parts and other things in the physical world. ‘Person/digit’ is a widespread equation for ‘thumb’, for example ‘chief of hand’ (Igbo), ‘parent finger’ (Gikuyu), and ‘chief of finger’ (Amharic). Similar metaphors for ‘thumb’ are found beyond Africa, e.g. ‘old man of hand’ (Wappo), and ‘mother finger’ (Mandarin Chinese); see Brown (1976) for further details. The use of body part nomenclature as a metaphor to specify the location in terms of a partonomic property of an object is also widespread cross- linguistically. Metaphorical extensions such as kwe ma mɛrɛ ‘lit. head of mountain’ for ‘mountain top’ in Bari, and [waŋ otto] ‘lit. eye of house’ for ‘window’ in Päri (both Nilotic languages) are probably widespread. But there are also what would seem to be less common extensions with a more regional distribution, for example in Päri, ko otto ‘lit. chest of the house’ for ‘wall’ (data from Simeoni 1978: 22–23). Witkowski and Brown (1981) extend claims about lexical-semantic hierarchies to the domain of bio-nomenclature, as shown in the preceding chapter. According to them, synchronic differences between languages further represent a unidirectional development or specialisation, related to the level of technology and general means of subsistence of the speech communities involved (see also Brown 1977b). But as Heine and König (1987) show, in Samburu, a Nilotic language of Kenya, there are terms for ‘tree’, ‘grass’ and ‘weed’ but not for an all-inclusive taxon ‘plant’. Neither do ‘vine’, ‘herb’, ‘shrub’ or ‘bush’ play any role in taxonomic groupings; see Chapter 5. ‘Weed’ refers to whatever does not fit into the category ‘tree’ or ‘grass’. Moreover, ‘weed’ as a category tends to be disregarded by men but not by women, presumably since herding is essentially being men’s business and weeds are of no use as cattle feed. In other words, the proposed hierarchies are clearly susceptible to further modification, as in the case of colour terminology. See also Berlin (1992) for a survey of ethnobiological classification, where it is claimed that cognitive as well as utilitarian criteria are part of folk classification.8 Another interesting example of such functional adaptation may be found in several Surmic languages, as well as neighbouring Western Nilotic languages in East Africa. These have extensive sets of lexical distinctions with names for

8 One such additional area is that of odour terminology. Where some languages have a rather limited set of terms (e.g. most European languages), others employ far more elaborate distinctions. Studying Lexical-Semantic Fields in Languages 115 larger mammals, specifically those living in groups or herds, and for males and females of different ages, for example in Koegu (derived from Hieda 1991):

(5) ímish ‘buffalo’ dímak ‘big male buffalo’ bush ‘old male buffalo’ gidé ‘young male buffalo’ c’encela ‘little male buffalo’ kobor ‘adult male buffalo’ góngura ‘old female buffalo’ kaura ‘young female buffalo’ mógosh ‘little female buffalo’ carkeilban ‘adult female buffalo’

Comparable distinctions or different lexemes are found for male and female hippopotami, warthogs and elephants of different ages:

(6) núar ‘elephant’ gádi ‘big male elephant’ gorung’ ‘big female elephant’

Rather than being part of a more universal pattern, the extensive sets of lexemes would seem to be specific to a group of genetically related languages spoken in the same area. Presumably, the lexical specialisation in their lexi- cons derives from an extension, through analogy formation, of the “cattle complex” lexicon to other domains in the natural environment. Such an extension may be considered functional. At the same time, however, it should be borne in mind that several pastoral cultures in East Africa have no such lexical specialisation. The rather restricted geographical spread of the phenomenon suggests that areal diffusion should also be taken into account in any historical explanation of these properties.

6.6 The Expression of Space and Direction in a Cross-Linguistic Perspective

The classic assumption of psychologists is that space orientation is egocentric and forward-looking. This means that for human beings, abstract space consists of three planes defined in relation to the human body: vertical up/ down, horizontal front/back, and horizontal left/right. By extension, it is then 116 CHAPTER 6 possible to transfer the centre of the co-ordinates onto an object, i.e. assigning an object a front, a back and sides, in order to use that object, rather than ego, as a relatum. But cross-linguistic research over the past decades has shown that there are different solutions to spatial conception in languages (see Levinson 2003 for a summary). How, then, should one set about investigating spatial conception in little-studied languages? To this end, the Cognitive Anthropology Research Group from the Max Planck Institute at Nijmegen University, the Netherlands, has developed a so-called Space Kit, by means of which notions of relativity in spatial conception and description can be investigated. Although there are problems with the Space Kit (not all speakers interviewed are accustomed to looking at two-dimensional pictures representing objects situated in a three-dimensional setting), some exciting results have already emerged. In the Australian (Pama-Nyungan) language Guugu Yimidhirr, instead of using body-relative locational descriptors such as ‘in front of, or ‘at the back of’, speakers use roots about position, direction, orientation, and motion; these roughly correspond to English ‘north’, ‘south’, ‘east’ and ‘west’. In other words, these relational terms are not “body-centric” or “ego-centric”, but “earth-centric” (Haviland 1992). Meso-American languages such as Tzeltal use absolute angles of orientation, with reference to a fixed notional ‘uphill/downhill’ plane (Haviland 1992). The system corresponds to the overall fall of the terrain along a south/north axis; it is not essentially egocentric either. There appear to be no general terms like ‘to the left of’ or ‘to the right of’. For several Oceanic languages on the other hand, the directional system consists of two axes, one based on the rising and setting of the sun, and the other on the geographical reference points of “seaward” and “inland”. It is thus necessary to know the environment in which the language is being spoken, or rather the use of the system depends on knowledge of that environment. Clearly, such cognitive distinctions also have consequences for grammatical features. Australian languages, apart from a locative case, do not encode locative preposition-like or postposition-like concepts, i.e. they have absolute systems (see, for example, Dixon 1980). Instead, they use a system of absolute orientation similar to cardinal points, i.e. one finds expressions such as ‘to the north of’ or ‘to the south of’ (from millimetres to kilometers, but not ‘in front of’; see Levinson (2003) for a synthesis of relative and absolutive systems of spatial orientation). Most African languages on the other hand do use either prepositions or postpositions (or sometimes both, as in linguistic areas where verb-initial and verb-final languages border on each other). Such adpositions are commonly Studying Lexical-Semantic Fields in Languages 117 derived from body part terms in languages such as Hausa (Chadic), Swahili (Bantu), and Maasai (Nilotic). The deictic system of the Chadic language Hausa uses body part nomenclature as an axis of orientation in the physical world.

(7) cikī ‘belly’ ciki(-n) ‘in(side)’ ciki-n gidā ‘inside the house’ bāyā ‘back’ bāya(-n) ‘behind’ bāya-n gidā ‘behind the house’

But non-derived cardinal terms tend to be used in order to describe directions.

(8) arḕwa ‘north’ kudù ‘south’ gabàs ‘east’ yâmma ‘west’

For example:

(9) sai kà bi tītìn nan zuwā̀ arḕwa should 2sg:m:subj follow road dem towards north har wata mararrabā, kà bi hanyàr dà till certain crossing 2sg:m:subj follow road with ta mīƙḕ gabàs 3sg:f lead east ‘follow this (main) road heading north till you get to a crossing, then take the road leading east’

According to Michael Broß (personal communication), urbanized Hausa speakers are far less likely to using these cardinal directions than those in rural areas. Giving directions sometimes requires a special kind of spatial knowledge.9 Wassmann (1993) describes some interesting aspects of Yupno, a Papuan language of Papua New Guinea, in which the visual representation appears to have no obvious link to the actual cardinal directions. I had a similar experience when somebody drew a map representing settlements and directions for getting to a mountainous area in south-west Ethiopia (a countryside similar to that of the Yupno). Whereas the person giving the directions, as a former teacher, was well aware of cardinal directions and their role in topography, these seemed to play no role in the visualisation of his directions.

9 See also Klein (1982) for an analysis of citizens giving directions to some well-known land- marks in inner-city Frankfurt. 118 CHAPTER 6

Whereas Nilotic languages use prepositional nouns in order to further specify the location, the system of cardinal direction marking is different from that found in most Australian languages or Hausa for that matter. In Nilotic languages there is a basic distinction between ‘east’ and ‘west’, but the other basic and widespread notions are ‘upward’ and ‘downward’. Moore (1986) describes an interesting account of local deixis in the Southern Nilotic language Marakwet. Anthropomorphic centring and projection onto objects, widespread on the African continent, is also attested in this language. The Marakwet build their villages on the slope of the Cherangani escarpment in Kenya, but farm in the valley below, so they move down and up on a daily basis. They use the antonyms nwun ‘downhill’ and doka ‘uphill’ to describe the relative positions. They also refer to fields furthest from the villages as bar kel ‘land of foot’, those nearest the villages as bar mät ‘land of head’, and the fields in between as bar quem ‘land of middle/waist’.10 But there is an additional, and rather intriguing aspect of deixis in Marakwet, involving tai and let, which are also terms of direction and orientation related to movement up and down the valley. Moore (1986: 15) shows that these antonyms can have several meanings:

(10) tai let ‘front’ ‘behind’ ‘right’ ‘left’ ‘past’ ‘future’ ‘coming’ ‘going’ ‘ancestors’ ‘unborn’ ‘south’ ‘north’ ‘upper part of valley’ ‘lower part of valley’

The Marakwet system of geographical reference points is one instance of a widespread Nilotic conceptual dichotomy involving ‘up’ and ‘down’ and the direction of water flow. The antonyms ‘south’ and ‘north’ in example (10) refer to the direction in which the Kerio River flows, away from and towards Lake Turkana, respectively. Browsing through dictionaries of Nilotic languages, one finds that various researchers have been confused by this concept, presumably because of interference from their (Indo-European) mother tongue, in which distinctions between four cardinal wind directions are common. Savage (1954),

10 According to Moore (1986: 52), the Marakwet also place a great deal of emphasis on the importance of space in ordering social perceptions and experience. Studying Lexical-Semantic Fields in Languages 119 for example, gives the following entries for the Western Nilotic Acholi, next to entries for ‘east’ and ‘west’:

(11) tuŋ malo ‘north’ tuŋ piny ‘south’11

But Crazzolara (1955), who had been doing research on related languages for two decades, arrived at the following translations for these entries:

(12) maalo ‘up, above, aloft’ pɪɪɲ ‘ground, down, below’

Crazzolara translated (tuŋ) arii as ‘the cross direction to East and West, i.e. North or South’. This association of ‘up’ and ‘north’ or ‘south’, or ‘down’ and ‘north’ or ‘south’ is probably not a coincidence. Whereas ‘east’ and ‘west’ (where the sun rises and sets) are standard, the two remaining terms tend, or rather tended, to refer to ‘up’ and ‘down’ (where water comes from and moves towards). Presumably, the semantic content of these terms is changing rapidly because of the effect of geography lessons in schools, and the need to find translation equivalents for English and Swahili terms. Such processes will also make it difficult (and sometimes impossible) to retrieve the original meanings or conceptual content of these and other terms.12 Barrett’s (1988, 1990) practical dictionaries for the Eastern Nilotic language Turkana is typical of several that have been complied for educational purposes. He gives the following entries for cardinal directions:

(13) kɪdɛ ‘east’ tɔɔ ‘west’ kuju ‘north’ kwap ‘south’

11 Note also that the Acholi word piny ‘south’, has a cognate in Dinka which means ‘land’, ‘country’. Dinka associates ‘back’ and ‘front’ with ‘east’ and ‘west’: piny cien ‘east’ cien ‘behind, at the end’ piny tueng ‘west’ tuen ‘first, in front’ 12 Mol (1978) confirms this in the Eastern Nilotic language Maasai, where moi-kuape can be translated as either ‘north’ or ‘south’. 120 CHAPTER 6

There is nothing wrong with such translations because the terms may currently be used with these meanings. The point is, however, that many Nilotic languages, including Turkana, have, on the one hand, a basic distinction between ‘east’ and ‘west’ (where the sun rises and sets) and, on the other hand, two other terms meaning ‘up, in/towards the sky or some elevation, where the water flows from’, and ‘down, on/towards the earth or ground, where the water flows to’. Compare the following related Turkana forms and corresponding meanings which reinforce this (partly historical) interpretation:

(14) kwap 1. ‘south’ 2. ‘under, down, below, underneath’ a-kwap ‘land, country’ kuju ‘north’ a-kuj(u) ‘God’

6.7 The Problem of the “Radical Translator”

In any fieldwork, there is a fundamental problem in getting at the meaning of utterances in a language the structure of which one is in the process of trying to understand. As the philosopher Quine observes in his Word and Object (1960: 26–90), “alien concepts” can be carried into English in an unlimited number of ways, each acceptable on its own, so there is nothing to choose between various renditions. Quine calls this the problem of the “Radical Translator”. Approaches to lexical semantics, for example in dictionary work, often rely on denotational overlap with English words (or any other receptor language) to establish the meanings of lexemes. Glosses often reflect an informant’s satisfaction that the investigator has learned an approximate meaning, but this may still be different from a correct definition of what a particular term actually refers to in the language under investigation. It is important to know about local structural facts in order to get at the real meaning of lexical items, as the discussion of terms for cardinal points should have made clear. (Of course the same holds with respect to the correct translation of the function and meaning of grammatical morphemes.) And describing the meaning of such items requires a metalanguage with semantic primitives that are not culturally biased. Wierzbicka’s work offers one such promising line of thought. (See, for example, Wierzbicka 1994, 1996.) Studying Lexical-Semantic Fields in Languages 121

6.8 Implicational Scales and Historical Reconstruction

As may be expected, claims about implicational relations in lexical-semantic fields have had their effect not only on synchronic-descriptive research, but also on historical-comparative research of language families. According to Witkowski and Brown (1981), Proto-Mayan had only ‘white’, ‘black’ and ‘red’ as basic colours. In spite of the widely spread terms for ‘grue’ (i.e. green and blue) or ‘yellow’ in contemporary daughter languages, these are not reconstructed as part of the colour term system of the common Mayan ancestor. Witkowski and Brown argue that, considering the additive nature of the colour encoding sequence, these reconstructions are unlikely. But such reasoning is seriously flawed, as argued in Chapter 5. If a particular form is attested in daughter languages belonging to main branches of the family, and if the various reflexes fit in with the regular sound correspondences between those daughter languages in which the forms are attested, the root in question has to be set up for their common ancestor. Branstetter (1977: 22), in her presentation of colour terms in Polynesian languages, concludes that “[. . . i]n the approximately 3500 year history of Polynesian languages [. . .] the color systems have changed from a Stage iv system [. . .] to the modern Polynesian language color systems of which from a sample of thirteen languages [. . .] one [. . .] is Stage ii, two [. . .] are Stage iiia, one [. . .] is Stage v, and one [. . .] is Stage vi.” But in a comment on Branstetter’s reconstruction of Proto-Polynesian colour terminology, Hamp (1980: 391) concludes that the following seven lexemes can be reconstructed: ‘white’, ‘black’, ‘red’, ‘roseate’, ‘yellow’, ‘orange’, and ‘grue’, pointing out that these “[. . .] agree neither with Branstetter’s nor with Berlin and Kay’s results”. An interesting account of colour terminology in a reconstructed proto- language, entirely motivated by internal evidence, is that of Kinkade (1988) on Salish languages. According to Kinkade, some 90 different roots for colours are found in the 23 Salish languages. Their common ancestor, Proto-Salish, “[. . .] had a very strange color system indeed: black, white and gray, the yellow- green-blue area split into three almost horizontal strata (from light to dark), and three kinds of red” (Kinkade 1988: 448). Whereas it is less common cross- linguistically to have a single term referring to a light colour in the blue to yellow-green part of the spectrum, and another for blue itself, this is in fact in line with systems currently found in many languages of the American northwest coast. 122 CHAPTER 6

6.9 Some New Evidence for Linguistic Relativity?

Whereas developments in the cognitive sciences and in language typology research in the last twenty five years or so has cast doubts on any rampant Whorfianism, instead emphasising universal constraints on language structure, more recent results of cross-linguistic studies do in fact suggest a place for language influencing thought processes. Lucy (1992a, 1992b) is some of the most convincing research in this respect. And as pointed out by the same author, “[i]f there is a linguistic relativity, then it may create real dilemmas for the conduct of research because researchers themselves are not exempt from these linguistic influences” (Lucy 1992a: 2). As Lucy (1985, 1992a, 1992b) and Lucy and Shweder (1979) point out, Whorf has often been radically misunderstood. His primary concern was with wide scale cultural habits of thought that shape the thinking of both children and adults. But Whorf clearly rejected simple-minded correlations between the formal features of language and general cultural characteristics. Neither was Whorf primarily concerned with perception or with cognitive processes, but with conceptual content (Lucy 1992a: 82). He followed a tradition of research initiated by Boas that language represents a classification of experience, or rather that linguistic classification reflects thought. Boas’ foremost student Sapir went on to claim that organised linguistic classifications channel thought, i.e. that linguistic classifications cohere into formally complete systems. According to these views, language contains an implicit classification of experience which may vary from language to language. Speakers accept their linguistic categories—as formally systematised abstractions from experience—as guides in the interpretation of experience or habitual thought, “that is, the habits of thought into which individual speakers routinely fall (what speakers actually think as opposed to what they are able to think)” (Lucy 1985: 74). Lucy points towards the importance of systematically attending to local structural facts in order to formulate an adequate formal-functional theory of lexical meaning. Lucy (1992b) also presents a model by means of which the linguistic relativity hypothesis may be re-examined. He identifies different patterns of thought through a systematic assessment of memory and through classification preferences among speakers of English and Yucatec Maya. Lucy investigates consequences of given differences in number marking between English and Yucatec Maya. In Yucatec, noun phrases are neutral in number. He claims that English speakers habitually pay attention to the number of various objects of reference more (and for a wider array) than Yucatec Studying Lexical-Semantic Fields in Languages 123 speakers do. And, as Lucy’s tests—simple cognitive tasks involving attention, memory, and classification—demonstrate, there are indeed differences in habitual thought. Yucatec, on the other hand, uses numeral classifiers, which take care of shape; in the lexical structure of Yucatec, there is a fundamental orientation towards substance. Interestingly, English speakers pay special attention to shape whereas Yucatec speakers pay attention to material (substance) in their cognitive tests (Lucy 1992a: 89, 1992b: 144–45).13 Numeral classifier systems are extremely rare among African languages, but there is at least one group which has them, namely the Kegboid group of Cross River languages in Nigeria (including Kana, see Ikoro 1994 for a detailed description). Given this areal restriction, these languages would form an almost ideal test case for similar psycholinguistic tests on picture description and recall tasks. But no such experimental research testing has been carried out to date investigating whether there are indeed incursions of linguistic categorisation into non-linguistic processes of thinking.

6.10 Some Final Observations

Cognitive approaches point towards limits on diversity between human languages, and as such they constitute an extremely important scientific exercise. Pinker (1994), for example, claims that language is an instinct, with very specific knowledge being “wired” or “programmed” in our brains. Pinker strongly subscribes to Chomsky’s view that “[. . .] from a Martian’s-eye-view all humans speak a single language” (Pinker 1994: 237).14 Exciting and attractive as it may sound from a philosophical point of view, the empiricist would also want to know what cross-linguistic investigation tells us. And this approach reveals

13 This hypothesis concerning “habitual thought” corresponds to what Kay and Kempton (1984) call “the name strategy” hypothesis. This predicts that the speaker who is con- fronted with a difficult task of classificatory judgment may use the lexical classification of the judged objects as if it were correlated with the required dimension of judgment even when it is not, so long as the structure of the task does not block this possibility” (Kay and Kempton 1984). 14 According to Pinker (1994: 32), “[. . . a]ll languages have words for ‘water’ and ‘foot’ because all people need to refer to water and feet [. . .]”. One wonders whether the author is aware of Japanese, Russian, and several African languages. Moreover, one wonders what caused him to come to this conclusion? Could it be perhaps that the lexical structure of his mother tongue, English, moulded his concept of reality? 124 CHAPTER 6 that although structures are indeed governed by common innate cognitive principles—of course they are—at the same time there is also tremendous variation.15 The interesting question of course is: Why is there so much variation cross- linguistically? One possible answer might be: Human language is so effective and successful from a phylogenetic point of view because our cognitive system, as reflected in language, apparently leaves considerable space for local adaptation, thereby allowing language to be placed more squarely in the social world in which it is used. Even if cultural selection occurs from a finite inventory of possible concepts, it is still extremely interesting from a cultural-historical point of view to see which selections speech communities make and why I have therefore argued here for a hermeneutic approach to the study of lexical-semantic domains. (See also Mathiot 1979 for such an approach.) The emphasis over the past decades on cognitive constraints in lexical- semantic domains has led to a movement away from a proper appreciation of the important role of culture and the natural environment in language structuring. Because of this cognition-oriented research, important observations about cultural and linguistic differences in the area of semantic fields in the lexicon may be overlooked. Keesing (1992) points the finger at Chomsky for his emphasis on formal- ist approaches to language structure. But it could equally well be argued that the study of universals within a Greenbergian framework easily leads one away from socially constituted regularities or cultural differences. Of course, Greenberg himself is not to blame for this. As one of the foremost linguists of the twentieth century, he made tremendous intellectual contributions to the study of linguistic diversity, as I have argued elsewhere (Dimmendaal 1993). Cognitive factors constrain the range and types of linguistic variation. But what is so interesting about that, given the tremendous cross-linguistic variation in systems? Moreover, our current understanding of how the human brain works is still rather limited. In a world where so many languages are disappearing, and where there is a unidirectional process of globalisation, it is of utmost importance to document differences between languages. This tendency towards unification under the influence of major languages such as English and Spanish, is manifested not only in lexical-semantic modification, but also in grammar. For example, several Amerindian languages now use conjunctions to link clauses, thereby copying a strategy which is common in

15 For a brilliant synthesis of the nature versus nurture debate in the social sciences see Kuper (1994). Studying Lexical-Semantic Fields in Languages 125

Spanish, whereas they would have formerly used separate verb forms for the same purpose. (See Appel and Muysken (1986) for observations on grammatical borrowing and linguistic change.) Colour hierarchies as proposed in the scientific literature also show, at least to some extent, the history of the progressive domination of western categories (Tornay 1978b). These various developments would seem to be part of a more general phenomenon of linguistic acculturation, influencing the meaning and functioning of words under the domination of the major languages of the world. Claiming that human senses change with the dominant mode of produc- tion, and that, consequently, one can observe evolutionary forces triggered by techno-sociological development, is unwarranted. Given the variation in the organisation of lexical-semantic fields, one could make at least as convincing a case for the association of their structure with either culture area or language family. What has changed, presumably, is the geographical distribution of diversity in systems. The change in lexical-semantic diversity seems to have a parallel (in terms of space and time) in grammatical systems. Nichols (1986, 1992) presents a fascinating investigation along these lines showing that languages tend to use either a dependent-marking strategy in their coding of grammatical systems (for example through case markers on subject and object noun phrases operating as arguments of verbs), or a head-marking strategy (for example by marking the presence of subjects and objects on the verb). Contacts between languages belonging to either type may lead to areal diffusion. One rather serious methodological problem, potentially affecting the results of our cross-linguistic comparisons, revolves around the problem of “the Radical Translator”. As I have argued above, we are only too conscious of those “natives” whose language filters the world and reports on linguistic and cultural diversity; I mean us linguists, psychologists and (cognitive) anthropologists. The nub of the problem is indeed semantic in nature. There is a fundamental empirical problem in our search for cognitive principles and their derivatives, implicational universals, having to do with the translation of lexical items. Obviously, we need to combat ethnocentric views in the formulation of universals. Brown and Lenneberg, and their intellectual successors, wanted to con- test the claim associated with the names of Sapir and Whorf that the search for semantic universals was fruitless in principle because each language was assumed to be semantically arbitrary relative to every other language. But either approach would seem to take issues to an extreme and unwarranted end. Language does not mould our environment, but it helps us to see our world in an orderly fashion. As an implicit analysis of experience, language structure 126 CHAPTER 6 may vary from one speech community to another. Languages clearly differ semantically, although not without constraint of course. Such variations are not necessarily relative even if they are non-universal. In this chapter, I have therefore proposed a pendulum swing from relativism through universalism back to a (moderate) type of relativism, i.e. to a more limited kind of Whorfianism, or less dramatic determinism “in which linguistic differences may be capable of inducing cognitive differences” (Kay and Kempton 1984: 75). Silverstein (1976) warns us to be careful of terms like “semantic”, “meaning”, “function” and other lexical items referring to entities of semiotic theory, and “to shake well before using”. But this would also seem to be true for lexemes whose usage should indeed be scrutinised for actual referential content. Before translating a term, one should first read the directions for use in the language under study. Maybe it says:

Vor Gebrauch schütteln or:

agiter avant de servir or:

shake well before using

And this list may of course be extended with translations from any other language. But sooner or later one comes to realise that they do not express quite the same (cultural) meaning. CHAPTER 7 Lexical-Semantic Fields in Tima

7.1 Introduction

This chapter focuses on specific lexical domains in Tima. As Chapter 3 has already shown, the Nuba Mountain area is characterised by tremendous genetic and typological diversity. Whether this also applies to the lexical domains discussed below—bio-nomenclature, colour, shape and texture, taste, and body parts—can only be clarified once more descriptions become available. This chapter therefore serves as a pilot study. As for other lexical domains, Meerpohl (2013) investigates kinship terminology, Alamin, Schneider-Blum and Dimmendaal (2012) treat spatial orientation in considerable detail, and Dimmendaal (2014b) discusses the implication of location and direction marking in Tima from the point of view of evidentiality marking. Before discussing these various lexical domains, some introductory remarks on sense perception as reflected in the Tima lexicon are in order. The Tima lexicon distinguishes between vision, audition, olfaction, gustation, and tactition, as Table 15 from Schneider-Blum and Dimmendaal (2013) shows.

table 15 Perception verbs in Tima

Sense modality Verbal noun Verbal root English equivalent

Vision kɘ̀-ŋàhɛ́ɛ́l -ŋah- seeing, watching, looking, tending, caring Audition k-ɨ̀mɨ̀ntéél -mɨnt- listening, hearing Olfaction kɨ̀-ŋʌ̀ʌ́l -ŋʌl- smelling, sniffing kù-dùùhéél -duh- Gustation kɪ̀-lɛ̀ɛ̀mɪ́l -lɛm- tasting Tactition kɘ̀-dàál -da- touching, feeling

The Tima verbal nouns in Table 15 are forms which themselves have a relatively high frequency in Tima discourse. For example, the verbal noun kɘ̀-ŋàhɛ́ɛ́l

© koninklijke brill nv, leiden, ��5 | doi ��.��63/9789004224148_008 128 CHAPTER 7 derives from a transitive verb root ŋah-. The corresponding nominal stem has a complex internal structure, based on the antipassive form ŋah-ak followed by the nominalising Applicative suffix -ɪl (ŋàhɛ́ɛ́l<-ŋah-ak-ɪl). The presence of this suffix reflects the presence (whether formerly expressed or implied) of some object affected by the verbal act. This suffix is obligatory with certain nominalised forms (‘seeing, watching’, ‘listening, hearing’, and ‘tasting’), and optional with other nominalised perception verbs (‘smelling’, ‘touching’). The actual meaning of ‘seeing’ in Tima also extends into a more abstract meaning of ‘tending, caring for something, looking after’. Schneider-Blum and Dimmendaal (2013) discuss these and other semantic extensions. The prefix kV- is a noun-class marker whose vowel harmonises with the first vowel of the following verb root. These verbs play a central role in the description of different perceptual experiences discussed below.

7.2 Bio-nomenclature

Chapter 6 referred to strong claims that have been made about the internal structure and historical development of bio-nomenclature as a lexical domain. The interest in folk taxonomies goes back to the 19th century when Western colonisation brought knowledge about alternative ways of conceptualising the natural environment. The standard terminology for folk-biological ranking probably goes back to Berlin (1992), who refers to English concepts such as ‘plant’ or ‘animal’ as folk kingdom ranks. These can be divided into ‘life-form’ ranks such as ‘tree’ and ‘grass’ or ‘fish’ and ‘bird’. Each of these contains generic ranks such as ‘oak’ and ‘clover’. According to Berlin, these form the core of ethnobiological classifications. These generic species contain ‘folk-specific’ ranks like ‘white oak’, which in turn may be sub-divided into ‘folk varietal’ ranks (like ‘swamp white oak’). So taxonomies like these are based on relationships of contrast and inclusion (“kind of” relations). For example, in English, the specific term ‘oak’ contrasts with ‘birch’ and both are included in the next hierarchical level ‘tree’ which in turn contrasts with ‘herb’ and ‘grass’. According to Brown (1984), systems are organised into taxonomic structures with no more than six mutually exclusive ranks expressing growth stages. With respect to botany, one may come across languages without a separate word for ‘tree’ as a life-form taxon. Brown (1984: 133–134) claims that this applies to Southern Paiute (an Uto-Aztecan language spoken in the USA). At stage 2, life- form terms like ‘tree’ occur, whereas at stage 3 we find terms for ‘tree’ and ‘grerb (i.e. grass plus herb)’. Stage 4 languages distinguish between ‘tree’, ‘grerb’ and Lexical-semantic Fields In Tima 129

‘vine’, according to this typology, and stage 5 languages have lexemes for ‘tree’, ‘grerb, ‘grass’ and ‘vine’, as in Daga (a Trans-New Guinea language of Papua New Guinea):

(2) oma ‘tree, wood’ rarema ‘grerb (herb)’ ut ‘grass (grass + weed)’ damik ‘vine’

Polysemy is common cross-linguistically, as the Daga word oma ‘tree, wood’ illustrates. In many African languages, the word for ‘tree’ may also be translated as ‘medicine’, as for example yaàt in Acholi (Nilotic, Uganda). Determining the primary and extended meaning is not always easy. The Tima distinguish between annual plants (k-ʌ́yí (sg), y-ʌ́yí (pl) and perennial plants (c-ɪ̀bɪ́) (sg), ɪ̀-bɪ́ (pl), the latter referring not only to vegetation that grows over several seasons or years, but also to trees. When a field investigator asks for the word for ‘tree’, then, speakers might say c-ɪ̀bɪ́, and for ‘grass’, the answer would be k-ʌ́yí. Without additional research, the investigator might conclude that these responses confirm Brown’s (1984) hierarchy, whereas in fact this misses an important feature of Tima taxonomy: k-ʌ́yí refers to ‘annual plants’, not just ‘grass’. Such examples show that it is not just physical similarity that counts, but also function or behaviour.1 Such folk taxonomies also show that life-form taxa may be biologically highly diverse, covering a wide range of generic taxa. The Tima classification is essence-based and reflects utilitarian factors (e.g. edible versus inedible). Such taxonomies support the conclusion that typicality is not based on similarity, but rather reflects the salience of or human interest in a species in a specific community. Hence, they are culturally mediated. Brown (1984) argues for similar “growth stages” with respect to zoological terms, ranging from a life-form taxon like ‘fish’ (stage 1), ‘bird + snake’ (stage 2), ‘bird + fish + snake’ (stage 3), ‘bird + fish + snake + wug (worm + bug)’ (stage 4), ‘bird + fish + snake + wug + mammal’ (stage 5). Tima has separate lexemes for the following generic terms (and consequently its system would belong to stage 3):

(3) cɪ̀-hɔ́ɔ́k ‘bird’ kú-mòŋ ‘fish, ant’ kɨ̀-mɨ́nʌ̀ ‘snake’

1 For further details see Schneider-Blum (To appear). 130 CHAPTER 7

In addition, there are two overarching taxonomic concepts in Tima, kàn bàyʊ́k (sg) / yàn bàyʊ́k (pl) ‘living being(s), animals (lit. thing(s) alive)’ and kàn káárɘ́n (sg) / yàn káárɘ́n (pl) ‘wild animals (lit. things wilderness/uncultivated land)’. Over the past decades, there have been two competing models of folk biological classification. On the one hand, the Taxonomic Hierarchy Model (e.g. Berlin 1992) parallels a Linnean taxonomy of plants, with a set of taxa specifiable by relations of affinity and contrast. On the other hand, the Natural Core Model (e.g. Hunn 1982) allows for gradable membership or categorial contiguity, and thereby for transitions. This model also accounts for the fact that phenomena in the natural environment are not necessarily categorised (or lexicalised) and may be left unclassified instead. Data from Tima clearly support the Natural Core Model. Rather than having mutually exclusive taxa that jointly exhaust the domain, the Natural Core Model assumes that members are assigned a place in the category according to their degree of membership or to the features they share with the most typical or central members, rather than through discreteness.

7.3 Colour

Several of the conceptual problems with the Berlin and Kay paradigm discussed in Chapter 6 also apply to Tima. The visual perception of what English refers to as ‘colour’ is expressed by way of adjectives in Tima. For example, the macro-term ‘RED’ in Tima, -rdí, is used as a nominal modifier in its basic categorical form:

(4) kɨ́dʌ́ŋ kɨ́-rdí brick agr-red ‘red brick’

Adjectival roots in Tima can be turned into abstract nouns, and this also applies to lexemes translatable as colour terms in the English metalanguage, for example bí-rdì ‘redness’ (the prefix bí- being used to form abstract nouns). I elicited the terms reported in figure 5 by presenting a set of 52 colour stimuli shown to different speakers.2 When applying the Berlin and Kay (1969) notions of “focal colour” and “best representative” for a particular lexical term, the following schema emerges:

2 I would like to express my gratitude to Serge Tornay for making this set available. He and other investigators used the same set in Tornay (1979), an early study criticising Berlin and Kay’s (1969) methodology. Lexical-semantic Fields In Tima 131

-tún -tɨ́k -rdí -hɛ́h -kùlùmó BLACK WHITE RED BLUE BROWN GREEN YELLOW figure 7 Colour terms in Tima

The root -tɨ́k is used, for example, to describe the colour of a kind of white mud traditionally used for body painting.

(5) kʌ́↓lú kɨ́-tɨ́k mud agr-light ‘white mud’

The achromatic colour terms WHITE and BLACK also function as antonyms in the description of the bark of specific acacia trees:

(6) kɨ̀ɽɨ̀r kù-tún Acacia agr-dark ‘Acacia Laeta’

kɨ̀ɽɨ̀r kɨ̀-tɨ́k Acacia agr-light ‘Acacia Mellifera’

The term RED is also used to describe sunsets:

(7) kìnéé à-rdí sun pred:sg-red ‘it is getting dark’

(8) kìnéé ŋ=kɨ̀-rdì sun instr=agr-red ‘dusk, time around sunset’

According to Berlin and Kay’s (1969) presumed hierarchy for the division of the colour spectrum, these terms are “basic” in that they are not derived from some other lexeme. But Tima contradicts this, since it does not distinguish between English ‘blue’, ‘green’ and ‘yellow’ (although admittedly it does have a lexical root for English ‘brown’). This system is common in particular in languages west of the Nuba Mountains in Sudan and Chad, and also elsewhere in the 132 CHAPTER 7 world. Foley (1997: 157) points out that such systems are found in languages in Asia, Australia, Oceania, North America and South America. There are a range of conceptual problems with the Berlin & Kay paradigm in this respect, as the preceding chapter argues, and as Foley points out:

[. . . m]ost problematic of all are systems which have as a basic color term a YELLOW/GREEN composite category, in which YELLOW again splits from WARM, but this time merges with GRUE. How can YELLOW be simultaneously LIGHT/WARM and DARK/COOL? [. . .] This grouping of YELLOW/GRUE poses formidable problems for Kay and McDaniel’s grounding of generalities of basic color terms in innate perceptual properties of the human color vision system, specifically, the subsystems based on opposing colors. Yellow and blue (member of the composite GRUE) are opposing poles of the same subsystem and if these subsystems based oppositions are the universal grounding for human color cat- egorizations, it is hard to see how yellow and blue could be conflated in a single named category.

A closer look at the grammatical system of Tima shows that the term -hɛ́h, which is best translated as ‘light, bright’, may be modified by ideophonic adverbs, as example (9) shows with the predicative marker a-. (The gemination of the glottal approximant is an expression of emphasis or intensity):

(9) àhhɛ́h wɛ̀lwɛ̀l ‘it is light blue’ àhhɛ́h wʌ̀lwʌ̀l ‘it is bright blue’ àhhɛ́h t̪àrt̪àr ‘it is light green’ àhhɛ́h t̪ùrt̪ùr ‘it is bright green’ àhhɛ́h kɨ̀-rɨ̀ndì ‘it is light yellow’ àhhɛ́h kʌ́yí ‘it is green as grass’

A number of words modifying these terms also occur as independent nouns; the root -rɨ̀ndì is related to the word for ‘millet’, ì-rɨ̀ndì, whereas kʌ́yí may be translated as ‘grass’. The modifier -wàlwàl refers to ‘bright, vibrant’, as in kɨ̀ɽɨ̀r kʊ́-wálwàl ‘Acacia senegalensis’. It may also be translated as ‘loud’ when used in combination with a word like ‘song’. Defenders of the Berlin and Kay paradigm might argue that the presumed “under-differentiation” in the domain between RED and BROWN is compen- sated for through the additional use of these modifiers. However, -hɛ́h wɛ̀lwɛ̀l and -hɛ́h kɨ̀rɨ̀ndì are a kind of -hɛ́h in the same way that ‘indigo blue’ and ‘light Lexical-semantic Fields In Tima 133 blue’ are a kind of ‘blue’. So they are a proper subset of the latter, following Berlin and Kay’s definition of “basic colour term”. Moreover, Tima speakers make similar distinctions between different types of BLACK or RED:

(10) àttún kùlùmkùlùm 3sg.dark ideo ‘it is beautifully dark’

(11) àttún hɘ̀rhɘ̀r 3sg.dark ideo ‘it is dark and filthy’

In order to express a strong degree of redness (i.e. saturation), the adjectival root may be strengthened through the addition of an ideophonic adverb.

(12) à-rdí t̪ɪ̀bɪ̀t̪ɪ̀bɪ̀k 3sg-red ideo ‘it is very red, it is blood red’

The term -hɛ́h consequently refers to anything which is not dark or white, and not red or brown, i.e. to anything which is perceived as light. More than forty years ago, Tornay (1972) pointed towards some fundamental epistemological problems concerning the investigation of so-called colour terms using the Berlin & Kay (1969) paradigm. These are as valid today as they were then; see also Chapter 6.3 The fundamental problem with the Berlin & Kay perspective, namely the assumption that the meaning and function of such terms relates to colour, can be illustrated with numerous other examples. In his classical contribution on Hanunoo, Conklin (1964: 191) points out that the four terms in this Austronesian language bi:ru DARK/ BLACK/BLUE, lagtiq LIGHT/WHITE, raraq RED and latuy YELLOW/GREEN) appear to have certain correlates beyond

3 Tornay (1972: 85) discussed the symbolic functions of so-called colour terms in particular with respect to living creatures among the Nyangatom in south-western Ethiopia. With wild animals, these terms help to denominate subspecies or closely related types, whereas with domesticated animals (e.g. cows or goats) they serve to distinguish individuals. However, with respect to human beings, they mark social positions. For example, somebody who has killed a lion, leopard or buffalo (i.e. an animal which may kill human beings) will change the blue colour of his clay cap to red-ochre. 134 CHAPTER 7 what is usually considered the range of chromatic differentiation, and which are associated with nonlinguistic phenomena in the external world. More specifically, these phenomena relate to dessication and succulence as well as faded or indelible substance in visible components of the natural environment. Conklin (1964: 192) concludes that the perceptual categorisation for the four-way distinction in actual fact refers to DRYNESS, WETNESS, DARKNESS, and LIGHTNESS. Neglecting such referential meanings of these terms leads to arbitrary translations in a Berlin & Kay conceptualisation of the world, as shown in their (1969: 66) glosses of the Hanunoo terms. Such cases may be multiplied with data from a range of languages across the world, as shown in the literature over the past forty years. One recent example comes from the Kwa language Siwu (in Ghana). Dingemanse (2011: 198) observes that the verb rɛtɛ primarily means ‘ripe’ rather than ‘red’, as it is also used for fruits that are not red when ripe, such as bananas, whereas yuɛ primarily means ‘unripe’ and, by extension, ‘green’. Still, as Dingemanse makes clear (2011: 198), “[t]he verb yuɛ cannot be used for the striking colour of rice fields in the rainy season (which to the English eye looks green); for this visual sensation kpìnàkpìnà ‘dark/black’ is used instead.” The author further points out:

The overall picture that emerges from these tasks is that there appears not to be a relevant and coherent domain of experience in Siwu that neatly maps onto the English category colour: subjects use a mix of stative verbs, source-based descriptors, environmental terms and ideophones to describe colour patches of the Munsell charts. (Dingemanse 2011: 199)

Apart from the morphologically simple, descriptive terms corresponding to colour terms in English, Tima has several reduplicated, iconic adjectival stems:

(13) -pʌ̀ʌ̀pʌ̀ʌ́k ‘patchy (as for cows)’4 -cɛ̀ɛ̀rcɛ̀ɛ́r ‘striped (with narrow stripes)’ -cɔ̀ɔ̀rcɔ̀ɔ́r ‘striped (with wide stripes)’ -ɲɪ̀ɽɛ́ɲɪ́ɽɛ̀ɛ̀k ‘ringed, hooped (like a zebra)’ -t̪ɪ̀dɛ̀ɛ̀t̪ɪ̀dɛ́ɛ́k ‘spotted, speckled’

4 The final consonant is not copied in the first syllable whenever it is an obstruent. Lexical-semantic Fields In Tima 135

7.4 Shape and Texture

Lexemes describing shape or textures in Tima can be divided on formal grounds into those with a simple (usually -cvc) root structure and those with reduplicated forms. Terms describing haptic perceptions such as surface texture, flexibility or compliance correspondingly reflect a more general pattern with basic (non-derived) rather than morphologically derived terms. The verb kɘ̀-dàál covers only the meaning of active ‘touching’ or ‘feeling’. In order to express haptic experiences, Tima tends to use adjectives expressing specific textures or shapes.5 These may be used in order to modify a noun, or as predicates, in which case they are preceded by a marker a- (sg) / ɪ-, i- (pl).

(14) í-wúúŋ ↓ɪ́-pɔ́ cl-ocra agr-soft ‘finely ground/soft ocra’

(15) í-wúúŋ ɪ̀-pɔ́ cl-ocra agr-soft ‘the ocra is finely ground/soft’

Such adjectival roots may also occur in lexicalised forms such as káh ↓pɔ́ ‘fontanel’, lit. ‘soft head’. The root for ‘soft’ matches other monosyllabic (simple or underived) adjectives like -tí ‘short, small’ or -líl ‘cold’. The root -midin ‘frayed’, has a formal shape which is also common to basic (non-derived) adjectives, e.g. -pʊlʊŋ ‘wide’ or -yɔkɔm ‘solid, strong’. It may be used to describe the physical state of, for example, rope, but also for congealed porridge. Gemination of medial consonants occurs in emphatic speech:

(16) kwʌ́ʌ́=nʌ́ à-mìddín rope=DEM pred:sg-frayed ‘the rope is very frayed’

There are also several reduplicated adjectival stems in Tima, although synchronically this morphological operation is no longer productive:

(17) -dʌ̀ʌ̀ŋdʌ̀ʌ̀ŋ ‘crooked, askew’ -kɔ̀dɔ̀kɔ̀dɔ́k ‘round’

5 For a semantic grouping of Tima adjectives, following a Dixonian grouping into shape, pro- pensity etc., see Alamin (2013b). 136 CHAPTER 7

Such adjectival forms can also be used in an adverbial sense as secondary predications:

(18) cɪ̀bɪ́ nǎ àŋ-kááràk dʌ̀ʌ̀ŋdʌ̀ʌ̀ŋ tree dem per-grow crooked ‘this tree grew crookedly’

The reduplication itself appears to be a reflex of a former word formation process in Tima whereby nominal stems were reduplicated, possibly with some additional segmental or suprasegmental morphology, in order to express descriptive adjectives. These in turn could also be used adverbially. This is at least suggested by certain lexical forms based on or related to the word kɨ̀-mɨ̀lʌ́y ‘earwax’. An alternative form, kɨ̀-mɨ̀lʌ́ð, with a final dental fricative, is used only by old speakers. A presumably related adjectival stem occurs in the expression t̪àmáá mɪ̀lɛ̀ʔɛ̀y ‘soft and gentle voice’. The former nominal stem (which originally may have referred to ‘wax’ in general) can be reduplicated in order to express an adjectival meaning, -mɨ̀lʌ̀ʌ̀mɨ̀lʌ̀y ‘slimy (e.g. of okra), smooth’. Alternatively, this stem is pronounced -mɘ̀lɛ̀ɛ̀mɘ̀lɛ̀y. In fast speech, the fully reduplicated root may be changed into -mɨ̀lʌ̀ʌ́lʌ̌y, with a corresponding abstract noun bɪ̀-mɘ̀lɛ́ɛ́lɛ̀y ‘smoothness’. See Schneider-Blum (2013) for additional, related lexemes in Tima. The stem mɨ̀lʌ́y itself probably contains a petrified prefix formerly used with nouns referring to liquids, masses or collectives. Synchronically, a number of such stems with incorporated noun-class prefixes all describing shapes or textures occur in Tima, although there is no longer a corresponding noun from which they are derived.

(20) -mɘ̀ɽàhmɘ̀ɽàh ‘sticky (like gum or dough), gluey’ -mɘ̀dɘ̀kɪ̀mɘ̀dɘ̀kɪ̀k ‘sticky (like sugar)’ -mɘ̀nt̪ɛ̀hmɘ̀nt̪ɛ̀h ‘cracked’

Example (21) suggests that several noun-class prefixes (mɘ̀-, t̪ɪ̀-, nʌ́-) were incorporated into the root historically:

(21) -nʌ́ŋɨ́rnʌ́ŋɨ́r ‘rough (e.g. of skin after scratching)’ -t̪ʌ́ŋɨ́rt̪ʌ́ŋɨ́r ‘rough (of skin, tree bark, mountain ranges)’

Synchronically, however, the incorporated noun-class prefixes are an inherent part of the stem: Lexical-semantic Fields In Tima 137

(22) cídʌ́ ↓lɛ́ɛ́nɪ́ à-t̪ʌ̀ŋɨ̀rt̪ʌ̀ŋɨ́r sg-body 1sg:poss stat-rough ‘my skin (lit. body) feels rough’

Dimmendaal (2014b) describes this incremental process as the “annual ring” model for noun class modification in Niger-Congo languages. The relevance of the Frequency Code, i.e. the symbolic use of voice quality (Ohala 1994), becomes evident from the following, alternative Tima forms describing textures:

(23) -lbɛ́ɛ́↓bɛ́k ‘slightly soft’ -lbúú↓búk ‘soft (like a balloon, cushion or baby’s head)’

Dingemanse (2011: 201) observes that Siwu has few nouns for abstract shapes, but that it does have many shape-related ideophones which encode highly specific distinctions, in particular in collocations with body part terms. This parallels Tima, and presumably numerous other African languages. In Tima, the reduplication marks these terms as ideophones.

(24) -kɔ̀dɔ̀kɔ̀dɔ̀k ‘round’ -dààldààl ‘shallow, flat’ -bùɽùbùɽùl ‘massive, thick, big’ -ɽɪ̀bɪ̀ɽɪ̀bɪ́h ‘wrinkled’

Finally, a number of adjectival stems expressing shape appear to be morphologically complex historically, but synchronically they are best treated as being morphologically simple, due to lack of alternation with other forms. For example, the root -tʌ̀pùʔùk may refer to the size of palm leaves, an elephant’s ears, or a girl’s bottom.

(25) kudù cí↓ŋí láŋɪ́ à-tʌ̀puʔuk vagina faeces 2sg:poss pred:sg-huge ‘your bottom is huge, you are a lardass’

The etymology of this adjectival root, or of other terms referring to shape or texture, such as -múrúhìíŋ ‘tough, resilient’, remains to be determined. 138 CHAPTER 7

7.5 Taste

Storch (2014) shows that Luwo, a Nilotic language spoken southeast of the Nuba Mountains, has a particularly rich array of lexical items expressing olfac- tive experiences. There is no indication so far that the Tima lexicon is particularly elaborate with respect to the description of odours. On the other hand, this language does show some interesting properties regarding the description of tastes. Cross-linguistically, it is common to distinguish between five basic tastes: sweet, sour, bitter, salty, and umami—a term recently borrowed from Japanese to describe the flavour of glutamates (Dingemanse 2011: 189, 194). Tima has words for four of these basic concepts:

(26) -hín ‘sweet’ -dɛ́ ‘sour’ -kɨ́k ‘bitter’ -lɛ̀lɛ̀lɛ̀ ‘calcium tasty’

The term -hín can also be used to describe tasty meat. This semantic extension appears to be common in neighbouring languages, so a more precise translation would be ‘tasty, palatable’. It also has a more abstract meaning when describing someone’s speech as being ‘pleasant, agreeable’. The abstract noun derived from this adjectival root combined with the word for ‘bone’ (k-ûh), bì-hín, expresses ‘joy, pleasure, happiness (lit. sweetness of bone)’. As for the term -kɨ́k, it is used to describe the quality of a bladder:

(27) kɪ́↓pɛ́n kɨ́-↓kɨ́k bladder agr-bitter ‘bile, gall’

Another term related to gustation describes the taste of calcium or salt:

(28) íídì ɪ̀-lɛ̀lɛ̀lɛ̀ water AGR-calcium.tasty ‘the water tastes like calcium’

The latter term is part of a set of reduplicating adjectives, although none of these appear to be derived synchronically: Lexical-semantic Fields In Tima 139

(29) -hòhòk ‘salty’ -kààkààk ‘bitter-sweet (e.g. tea with little sugar)’ -hɘ̀lɛ̀hɘ̀lɛ̀m ‘sweet’ (e.g. sugarcane)’ -hùlùhùlùm ‘sweet-salty (e.g. sour dough), sweet-sour (e.g. a mango)’ -dɛ̀kʊ̀dɛ̀kʊ̀k ‘sourish’

The Tima system thus goes beyond the widespread division into five terms for the description of tastes. None of these terms, however, describe a taste similar to ‘umami’.

7.6 Body Parts

Research within the Greenbergian tradition of language typology, using an inductive method in order to arrive at universal tendencies and implicational universals, resulted in Anderson’s (1978) first inventory of common strategies in the visual perception of body part nomenclature. Her pioneering study reveals two characteristics of visual properties that play an important role in the lexical realisation of terms. Firstly, those related to form or shape are reflected in the frequent use of specific metaphors in order to refer to certain body parts cross- linguistically, for example the extension of ‘cheeks’ to ‘buttocks’ (or vice versa). Secondly, those related to spatial proximity are reflected in widespread strategies of using metonyms for the same lexical domain; the importance of the latter can be observed in the widespread tendency to extend the meaning of ‘eye’ to ‘face’ (and vice versa). Metaphorical extensions manifest both widespread (“universal”) and area-specific properties. The anthropomorphic projection is reflected in the fact that cross-linguistically the hand or foot is often looked upon as a family, whereby the fingers or toes are the children (as in Yoruba or Mixe), and/or the thumb or big toe is the father or mother (as in Chinese, Gikuyu, Quechua or Lakhota); see Brown (1976). The metaphorical extension from ‘seed, kernel’ to ‘eye’ on the other hand is characteristic of languages in Northeastern Africa, for example in Baale, a Surmic (Nilo-Saharan) language spoken in South Sudan and Ethiopia, where kɛɛrɛ means both ‘seeds’ and ‘eyes’. Andersen (1978) observes that languages do not usually distinguish between more than five levels with respect to body part terminology. For example, the body (as a first level) may be divided into the head, trunk, legs, and arms. The hand in turn is part of the arm, with the finger and the fingernail constituting the fourth and fifth level, respectively. Andersen (1978) also notes that the tendency to use morphologically complex forms increases when moving down this hierarchy of partonomic relations. Examples (30) and (31) from Tima illustrate the relevance of these cross-linguistic observations. 140 CHAPTER 7

kah

kɪdɛk kɘrabʊ

pɨrʌmpɨrʌŋ

kalʊk

kudu kaɽɘm

kɨdʌwun

kɪrɛmʊŋ

kɨmʌnʌ kɨdʌwun kʊkwɔlɔŋ

kuruŋo kɪdɪɪ

kɨhɨɨr

kudu kɔdɔr kɨmʌnʌ kɪdɪɪ kʊkwɔlɔŋ picture 1 Body part taxonomy in Tima

Tima uses morphologically simplex forms for body parts such as ‘arm (including the hand)’, but refers to the digits by means of complex lexemes. The diminutive marker kɨ̀-mʌ̀nʌ́ (singular) / ì-mʌ̀nʌ́ (plural) is used in combination with the word for ‘arm’ to express ‘finger’:

(30) kɨ̀-mʌ̀nʌ́ kɨ̀-dʌ̀wún cl-small agr-arm ‘finger’ Lexical-semantic Fields In Tima 141

A similar morphologically complex strategy occurs with other words referring to body parts occurring in a partonomic relation to other parts, for example:6

(31) kɪ́ɲɛ̀ ɨ̀ kìmídì mouth gen breast ‘nipple’

Example (32) also illustrates the common strategy of using basic terms for prominent body parts in order to refer to cognitively less prominent parts:

(32) kɪ́ɲɛ̀ ɨ̀ kɨ̀mʌ́mìì mouth gen nose ‘nostril’

The genitive marker in examples (31–32) is optional, presumably because of a tendency towards segmental contraction, which itself is a reflex of lexicalisation. See Schneider-Blum (2012) on the transition between freely generated syntactic structures and phrasal compounds in Tima. The word for ‘head’ in Tima, k-âh (sg) / y-âh (pl) is used in combination with various other body part terms to create compound words for cognitively less significant body parts:

(33) káh ↓cɛ́ɛ́n head face ‘forehead’

káh ↓kɪ́ɲɛ̀ head mouth ‘lip’

káh ↓cɪ́ɪ́ head eye ‘rib’

6 The metaphorical extension of ‘mouth’ to ‘nipple’ is widespread across Africa and consequently may be an archaism. It also occurs in creolised varieties of English in the Carribean as a result of a Kwa substrate (Huttar, Essegbey and Ameka 2007). 142 CHAPTER 7

kâh kɪ̀mɛ́ head penis ‘glans penis’

Note that cɪ́ɪ́ ‘eye’ in the word for ‘rib’ above is also used metaphorically, since the Tima apparently perceive the rib cage as resembling a pair of eyes. Tima also uses metaphorical extensions based on basic body terms to denote nouns in other semantic domains. For example, the word kú-ɽúún ‘belly’, not only occurs in kuɽún tʊ̀ɽɪ̀ ‘stomach’ (lit. belly (of) food), but also in kú-ɽúún(ɨ́) kɨ̀-dʌ̀wún ‘palm of hand’ (lit. belly of hand). The noun kánt̪ɘ̀, which is probably best translated as ‘inside’, co-occurs with lexemes some of which are no longer used independently (i.e. “cranberry” morphemes), again in order to refer to cognitively less significant, or less visible, body parts, as well as to words in other domains (example (34)).

(34) kánt̪ɘ̀ dʊ̀wàdɪ́ɪ́ inside ? ‘crotch’

kánt̪ɘ̀ mʌ́mʌ̀ŋ inside ? ‘armpit’

kánt̪ɘ̀ kʌ́dɨ́h inside ? ‘space between the stones underneath a granary’

The word kùdú ‘vagina’ is also used metaphorically to refer to ‘maternal family root’. Its occurrence in compounds referring to body parts at the lower end of another body part may mean that its original meaning was ‘lower part’. So its use as the word for ‘vagina’ may be the circumvention of a taboo word. For several of the examples of nominal compounds in example (35), the meaning of the second element has also become obsolete, that is, it is no longer used in isolation.

(35) kùdú kàɽɘ́m ~ kɘ̀dɘ́ kàɽɘ́m ‘elbow’ kùdú kùlúmbʌ́lìŋ ‘earlobe’ kùdú cɪ̀ɽɔ́k ‘throat’ kùdú cíŋí ‘bottom (lit. vagina faeces)’ kùdú kɔ̀dɔ̂r ‘heel’ kùdú círɨ́h ‘neck’ Lexical-semantic Fields In Tima 143

Note that the head noun of the word for ‘elbow’ has two variants, with the centralised vowels of the second variant possibly indicating further lexicalisation. Such sporadic changes are usually the result of accelerated speech, resulting in assimilation or deletion of segments (Dimmendaal 2011: 54–55). Several Tima body part terms—and this is more common cross-linguistically—are used not only to create intrafield metaphors or metonymic constructions (in the sense of Wilkins 1996), but also as interfield metonymic extensions. Thus, kùdú ‘vagina’ is used to refer to indicate the foot of a mountain, e.g. kùdú Píndíŋ ‘the foot of Mount Pinding’. Similarly, kɪ́ɲɛ̀ or kɪ́ŋɛ̀ ‘mouth’—depending on the dialect—occurs in combination with an ideophonic word in the following construction:

(36) kɪ́ɲɛ̀ kʌ́ŋkʌ̀ŋ mouth IDEO ‘riverbank’

There are other body part terms which serve as the heads of compounds, e.g. kâh ‘head’ which is commonly used in spatial expressions conveying the meaning of ‘top, upper end’, e.g. káh↓ Kákʊ́láŋkʊ̀ ‘summit of Mount Kakulanku’. The word kɔ́nɔ̀ ~ kwɔ́nɔ̀ ‘ear’, has been extended to refer to ‘cooking spoon, stirring spoon’ because of its shape. This also applies to ɪ̀-lɔ́lɔ̀y ‘scrota’, which is also used to refer to the stones underneath a granary, ɪ̀-lɔ̀lɔ́y kùdùŋkùdúŋ, lit. ‘scrota (of) granary’ (Picture 2, p. 144). The frequent use of such metaphorical and metonymic strategies in Tima is a highly economic way of creating new meanings in the lexicon (Schneider- Blum 2012). Chapter 10 takes up this issue again in the context of esoteric versus exoteric languages. 144 CHAPTER 7

picture 2 Grainstore in the Tima area Part 3 Conversational Styles

∵

CHAPTER 8 Colourful psi’s Sleep Furiously: Depicting Emotional States in Some African Languages

This chapter sets out to investigate the “poetry of grammar”, more specifically the role of the body in figurative speech, in African languages mainly belonging to Nilotic and Bantu. Understanding the semantics and pragmatics of metaphorical and metonymic expressions in these languages presupposes an interaction between a number of cognitive processes. Interestingly, these languages seem to use these strategies involving figurative speech in tandem with alternative strategies involving on-record statements. This multivocality only makes sense if we place language and language structure in the social world in which it is used.

8.1 Introduction

When American president Bill Clinton finally admitted, during a meeting with a special committee, to his extra-marital affairs with the White House intern Monica Lewinsky, Dutch newspapers commented: “Clinton met Lewinsky zaak in zijn maag” (“Clinton with Lewinsky [case] in his stomach”, i.e., Clinton embarrassed about the Lewinsky case). Of course, Clinton had good reasons not to speak “à coeur ouvert” (i.e., openheartedly), as some French news agen- cies put it, nor would it have been wise for him “das Herz auf der Zunge zu tragen” (“to carry the heart on the tongue”, i.e., to express his feelings), as German newspapers phrased it. As discussions lingered on, some American news agen- cies also started wondering whether Lewinsky could “stomach” any more. Whereas some claim that metaphors are language- and/or culture-specific, the use of body part terminology for the expression of mental states, as in these examples from a number of Indo-European languages, in fact appears to be common in languages across the world. In one of his pioneering studies, Matisoff (1986: 4) coins the label psychocollocation for such polymorphemic expressions referring as a whole mental process, quality, or state, one of whose constituents is a psychonoun, i.e., a noun with explicit psychological reference. The rest of the ‘psi-collocation—that Matisoff calls a ‘psi’—contains morphemes (usually action verbs or adjectives) that complete the meaning. In other words, psi’s are semantically defined

© koninklijke brill nv, leiden, ��5 | doi ��.��63/9789004224148_009 148 CHAPTER 8 constructions with syntactic ramifications, a formal aspect that we will return to in section 8.2. Matisoff presents examples from Sino-Tibetan languages and English. But there are also numerous examples from languages in other parts of the world including Africa. In one sense, such psi’s would ‘simply’ seem to reflect a linguistic translation of common human experiences involving perception of physiological changes coming from the activity of the visceral organs such as the heart or stomach. These smooth-muscle organs are innervated by the autonomic nervous system. The nerves and the brain in turn have the ability to receive and react to stimuli, such as light, sound, impact, construction, etc. This conceptualization of emotional states and neurophysiological changes has come to be known as the James-Lange theory of emotions (see, for example, James 1884). Expressions such as ‘das Herz auf der Zunge tragen’ in German, or its English equivalent ‘wearing one’s heart on one’s sleeve’, involve figurative speech where the ‘heart’ acts as a container or seat where feelings or emotions are based. Figurative speech of this kind usually obeys the syntactic rules of the language in question, but often involves a flouting of semantic rules, or a violation of pragmatic constraints. As Chomsky, one of the most influential architects of modern formal linguistics, points out (e.g., 1965), one can usually detect a hierarchy of deviations from grammaticalness in a certain set of sentences. A sentence such as ‘John became Bill to leave’ involves a breaking of strict subcategorisation rules (for transitivity). But there is another type of ungrammatical sentence, the latter involving grammatically correct but semantically anomalous statements (Chomsky 1965: 149), such as this famous and frequently quoted example (Chomsky 1957: 15):

(1) Colorless green ideas sleep furiously

Constructions of this type typically involve a relaxation of ‘selection restrictions’ (in terms of early generative models). Such utterances are ‘meaningless’ in one sense, but interestingly can often be interpreted metaphorically. During my initial survey of linguistic material of body part terminology and the expression of emotional states in two African language families, it became clear that the visceral organs figure prominently in psi’s in these languages as well. Interestingly, terms playing a role in colour description also play a prominent role in such constructions, as in the following example from the Nilotic language Maasai (data and translation based on Mol 1996): Colourful psi’s Sleep Furiously 149

(2) a-ibor oshoke inf-be.white stomach:abs ‘to be kind hearted (lit. to be white as to the stomach)’

This type of ‘colourful psi’ involves ‘stomach’ as a metaphorical seat for feelings and the predication ‘being white’ as one property or characteristic feature of the stomach, i.e., as a metonymic extension, since ‘stomach’ and ‘being white’ are related by contiguity, rather than similarity. It is in this sense that the title of this chapter (‘colourful psi’s sleep furiously’) is to be understood. However, before moving into the ‘poetry of grammar’ (in the sense of Jakobson 1971: 47), we will first look into the ‘grammar of poetry’ in section 8.2. More specifically, we will briefly consider alternative ways in which event structures of the type illustrated above may be categorised or framed syntactically cross- linguistically. Next, section 8.3 provides examples from African languages mainly belonging to Nilotic (Nilo-Saharan) and Bantu (Niger-Congo phylum). These language families cover major parts of eastern, central and southern Africa, so they are potentially an interesting sample from a genetic and areal point of view. Section 8.4 elaborates on the role of colour terminology in these same languages. Finally, Section 8.5 deals with how the understanding of the semantics and pragmatics of such metaphorical and metonymic expressions presupposes an interaction between various cognitive processes.

8.2 Categories and Event Structures

When comparing closely related Germanic languages such as Modern Dutch and German, one observes divergent strategies in the way semantic roles are linked to, or mapped onto, syntactic functions such as subject or (indirect) object. This appears to be the case in particular with respect to the expression of mental states. A comparison of cognate verbs and their syntactic behaviour further suggests that modifications must have taken place in one of the two, or both, historically.

German: (3) mich reut meine Sünde ob regret my sin ‘I regret my sin’ 150 CHAPTER 8

Dutch: (4) ik heb berouw van mijn zonde su aux regret of my sin ‘I regret my sin’

It is clear from older, written sources that Modern Dutch, rather than German, has been the innovating language in this respect. In Middle Dutch texts, such as the late 15th century drama Elckerlijc, the same verb still takes a (Dative) object, as in German:

Middle Dutch: (5) mi rouwet ob regret ‘I regret’

Modern Dutch has retained the verb, but its argument structure has been modified. This type of historical restructuring, which has been described by a number of authors (e.g., van der Horst 1981), has affected all dialects of Dutch, but interesting intra-dialectal variation exists with regard to specific verbs. Compare the following variation between Modern Standard Dutch, which is based on western varieties, and the more conservative eastern dialects (data by me as a native speaker). The latter form a dialect continuum with German dialects across the border, and are typologically similar to these.

Modern Standard Dutch: (6) ik lust dat niet 1sg:su like that:ob neg ‘I don’t like that (of food only)’

Eastern Dutch: (7) mie lust dat neet 1sg:ob like that:sub neg ‘I don’t like that (of food only)’

When comparing idiomatic expressions involving psi’s in Modern Dutch with those in German, such alternative mapping rules also occur. Whereas in German the affected entity tends to be expressed by means of a (Dative) object (8), the latter typically occurs in subject position in modern Standard Dutch (9): Colourful psi’s Sleep Furiously 151

German: (8) ihm ist etwas über die Leber gelaufen 3sg:ob is something over the liver gone ‘something is biting him’

Dutch: (9) hij heeft wat op zijn lever 3sg:su has something on his liver ‘he has got something on his mind’

Talmy (1988) proposes the label ‘force dynamics’ (how entities interact with respect to force) for one of the very basic imaging systems that organise scenes for language. The imagery of force dynamics affects the way we talk about natural events, social pressures, and psychological events. As examples (3) and (4) show, even closely related languages may differ considerably in this respect. Although the meaning of the two constructions in (8)–(9) is slightly different, the important point is that in Dutch the affected entity in the event description tends to appear in subject position, even when no volitionality is involved on the part of the affected entity; German, on the other hand, tends to avoid this strategy, the preferred syntactic slot for the affected entity being the (Dative) object. When comparing Nilotic languages such as Maasai with other languages belonging to this family, a comparable divergence occurs in the framing of force dynamics. Again, these frames appear to represent alternative strategies in linking semantic roles to syntactic functions or positions, strategies which themselves may change over time, even within a few centuries. Whether such differences do indeed have cognitive implications is an important question. Since no research has been done in this area on these African languages, however, this remains an open question for the time being. Along similar lines, words expressing emotional states may involve verbs in one language and nouns (sometimes requiring an auxiliary when used in a predicative sense) in another. Compare the Nilotic language Turkana (example (10), data from Dimmendaal 1983) and the Chadic language Hausa (example (11), data from Ma Newman 1990):

(10) ɛ̀-mɔnà 3-be.stingy ‘(s)he is stingy, mean’1

1 Third person singular forms translated by way of a masculine pronoun ‘he/him’ in the sources are translated as ‘(s)he/him/her’ in those languages that do not formally mark gender distinctions. 152 CHAPTER 8

(11) shii mài roowàà nee 3sg:m poss stinginess id ‘he is/was stingy’

Nouns generally express more permanent states or entities, whereas verbs tend to reflect temporary states (with adjectives forming an intermediate category). Since Turkana uses a verb whereas Hausa uses a noun, one might argue that ‘anger’ is conceptualized as a temporary state in Turkana, and as a permanent disposition in Hausa. This conclusion, however, is wrong or at least premature for a number of reasons. First, languages usually have morphological strategies allowing for categorial changes (e.g., from verb to noun, or vice versa). Second, languages may use different strategies in order to express temporal contingency; where one language uses a verb in order to express a temporary state (e.g., ‘be stingy’), other languages may express the same concept through a combination of a noun (‘anger’) with an auxiliary verb. Moreover, as typological studies show, the inflectional behaviour of adjectival or attributive constructions (i.e., similar to either verbs or nouns) is also linked to constituent order (Dixon 1997: 125). Consequently, one can make no further claims on the basis of such typological differences without a proper understanding of category shift or tense-aspect marking in such languages.

8.3 A Closer Look at Two African Language Families: Nilotic and Bantu

Nilotic languages are spoken in an area ranging from central Tanzania north- ward across Kenya, Uganda, DR Congo and South Sudan. They belong to the Nilo-Saharan phylum, one of the major phyla on the African continent. Nilotic is usually divided into three primary branches, Western, Eastern, and Southern. Examples below are drawn from representatives from each of these branches. In one of the first contributions on figures of speech expressing mental states in Nilotic languages, Nida (1955) observes that in Western Nilotic Anywa (a language spoken in South Sudan and Ethiopia) cwiny ‘liver’ figures prominently. The liver may be ‘good’ (i.e., ‘generous’), ‘bad’ (i.e., ‘unsociable’), ‘heavy’ (i.e., ‘sad’), or ‘sweet’ (i.e., ‘happy’). In fact, both ‘liver’ and ‘stomach/belly’ are dominant in expressing mental states in Anywa. Reh et al. (1999: 86) provide an example with (the locative form of) the word for ‘stomach’.

(12) yɪ̀-ɛ̄ ràác stomach-3sg bad ‘(s)he feels ashamed’ Colourful psi’s Sleep Furiously 153

In Päri, a closely related language spoken in South Sudan, ‘belly/stomach’ and ‘heart’ may apparently be used interchangeably (Simeoni 1978: 91):

(13) yi-a mita belly-1sg desire ‘I want/like/prefer . . . (lit. my belly desires)’

(14) cwiny-a mita liver-1sg desire ‘I want/like/prefer . . . (lit. my liver desires)’

In Luo, a Western Nilotic language spoken in Kenya, Tanzania and Uganda, the liver is the most frequently used body part in expressions relating to emotion (Reh 1998).2 The word for ‘stomach’ nevertheless also appears in one or two constructions expressing emotion, e.g., when angry, ic wang ‘the stomach burns’ (Reh 1998: 15). Body parts such as ‘head’ do not seem to figure in such expressions, although ‘head’ may figure as a metaphorical container for social image (or ‘face’ in the sense of Brown and Levinson 1987; see also Section 8.5 below); for example (Stafford 1967: 70):

(15) wiy-e tɛk head-3sg hard ‘(s)he is obstinate’

(16) ne wa-kuodo wi-gi past 1pl-shame head-3pl ‘we shamed them’

Example (15) has an interesting parallel in Dutch, where ‘stubborn’ (koppig) literally means ‘head-y’ (using the pejorative word for ‘head’ in Dutch, kop), and in French (têtu). As for example (16), the head presumably plays a role in such constructions because feelings of shame may be accompanied by narrowing of blood veins, and a corresponding raising of blood pressure, which is felt most prominently in the head. Similar figures of speech occur in Luo with ‘belly/stomach’ (data from Stafford 1967: 70):

2 Reh’s study of the lexical-grammatical means available in Luo to denote emotion is based on a Luo novel. 154 CHAPTER 8

(17) iy-a ɔ-wang stomach-lsg 3sg-burn ‘I am angry’

(18) ɔ-wang’ɔ iy-a 3-burn:tr stomach-my ‘it annoys me’

Noonan (1992: 189–190) presents an interesting discussion of ‘body-part imagery’, showing that Lango (a Western Nilotic language of Uganda) attributes emotions and personal characteristics to the stomach/belly and also to the liver:

(19) yɪ̀-ɛ́ yòm belly-3sg 3sg:soft:hab ‘(s)he is happy’

(20) cwɪ́ɲ-é yòm liver-3sg 3sg:soft:hab ‘(s)he is happy’

Crazzolara (1955: 209) translates the cognate form for ‘liver’ in the closely related language Acholi as ‘mind, temper, heart, sentiments, mental disposition’, and further observes that this word [. . .] forms the indirect subject or object of numberless phrases.

(21) chuny-e pɛk liver-3sg heavy ‘(s)he is sad’

(22) chuny-e o-duogo liver-3sg 3-return ‘(s)he is encouraged’

Whereas the liver, and to a lesser extent the stomach, seem to be prominent as a metaphorical container for emotions in Western Nilotic, the heart appears to be more prominent in Eastern Nilotic. In Bari (South Sudan), for example, the word for ‘heart’ töwili, figures prominently as the metaphorical container or seat for emotions. Spagnolo (1960: 306–307) gives the following translations for töwili: 1. heart; 2. soul, mind, conscience; 3. courage; 4. the inner part of a thing, illustrating this with various adjectival and predicative constructions: Colourful psi’s Sleep Furiously 155

(23) töwili lɪlɪk heart calm ‘calm, pacific’

(24) töwili logo heart hard ‘brave, steady’

(25) töwili papɛ: heart hot ‘quick-tempered, furious, irritable’

(26) töwili a rusökö heart be contracted ‘be sad (e.g., about a relative’s death)’

(27) töwili adɪ́ sö′busö′bu heart say tremble ‘the heart is trembling with fear’

Example (27), involving a ‘say’ verb plus an ideophonic predicate is common as a syntactic formula expressing sensory experiences in African languages; further examples of this type of construction are presented below for Bantu languages such as Zulu (Section 8.3). Whereas Bari is an svo language, many other Eastern Nilotic languages have a rather strict verb-initial structure, with subjects, objects and oblique phrases following the verb. These languages also use different strategies for the syntactic expression of event structures; unlike Bari, they tend to avoid non-volitional entities in subject position, as in example (28) in Maasai (Kenya, Tanzania; Mol 1978, 1996):

(28) a-dum-u oltau inf-pick-ven heart:abs ‘to pick up courage’

(29) e-nyor oltau lenye 3-love heart:abs self ‘(s)he loves himself/herself’ 156 CHAPTER 8

(30) a-ita-do-yio oltau inf-caus-go-down heart:abs ‘to heave a sigh’

Little can be said about figures of speech in Southern Nilotic since there are no extensive dictionaries. From the scanty material available on the role of the body and emotions in these languages, it appears that ‘stomach’ is the most common source for metaphorical and metonymic expressions, as in Päkoot, spoken in Kenya (data from Baroja 1989):

(31) ma mut 3:die stomach:nom ‘give up’

(32) kɪmɛɛɣiá múutná past:3pl.die heart:abs ‘they lost hope’

(33) karam mu good stomach:abs ‘be generous’

For the closely related Kipsikiis language (also spoken in Kenya), similar expressions with the cognate word for ‘stomach’ occur (data from Toweett 1979: 411):

(34) lál-è mààèèt hurt-as stomach:nom ‘feel emotional or physical pain’

For a number of Nilotic languages there is evidence that specific terms are used which also play a role in colour description. Concepts corresponding to ‘black, dark’, ‘white, bright’ and ‘red’ figure prominently in figurative expressions in this respect (as they do in Bantu, which we will explore in more detail below). For example, in Päri (Western Nilotic, spoken in South Sudan; data from Simeoni 1978: 91–92):

(35) cwiny-a col liver-lsg black ‘I dislike, I am angry/displeased’ Colourful psi’s Sleep Furiously 157

This begs the question to what extent these terms correspond to colour terms in the English meta-language. As examples (36) and (37) from Eastern Nilotic Bari show, some expression may have a literal reading in addition to the idiomatic meaning; for example, (36) may also mean ‘your teeth are red’ and (37) is also used to express ‘your eyes are red’.

(36) kala kulök a lɔ́tɔr teeth 2pl be red ‘you are very angry; your teeth are red’

(37) kɔnyɛn kulök a lɔ́tɔr eyes 2pl be red ‘you are a ferocious fellow; your eyes are red’

In Eastern Nilotic Maasai, the polysemous verb root -rok ‘black, empty, stubborn’ plus the derivational suffix adding an inchoative or inceptive meaning, expresses a physical as well as a mental state of affairs. Given that cognate forms in other Eastern Nilotic languages have a meaning ‘(be) black’ (e.g. Turkana: -iro- ‘be black’), other shades of meaning must be the result of secondary development.

(38) a-rok-u inf-black-ven ‘1. to become black; 2. be very angry’

(39) enkurma na-rok flour rel-be.black ‘flour without anything in it (e.g., oil)’

(40) enkaji na-rok house rel-be.black ‘empty house’

(41) a-rok oipi inf-be.black shadow:abs ‘to be ungrateful’

(42) a-rok oshoke inf-be.black stomach:abs ‘to be stubborn’ 158 CHAPTER 8

(43) a-rok ongu inf-be.black eye:abs ‘to be envious’

‘Black’ is also a holy colour among the Maasai, and may be used in a descriptive invocation for ‘God’: ee Papa lai o-rok, which Mol (1978: 32) translates as ‘oh my black father’. Consequently, the Maasai root -rok is polysemous at least synchronically. Similarly, the Maasai root corresponding to ‘(be) white’ in English also has some additional shades of meaning:

(44) a-ibor oshoke inf-be.white stomach:abs ‘to be kind hearted’

(45) a-ibor ongu inf-be.white eye:abs ‘to be immoral, to be promiscuous’

(46) ena-ibor-onyek rel-be.white-eyes ‘prostitute’

It would be incorrect to claim that ‘be black, dark’ versus ‘be white, bright’ inherently correspond to negative versus positive feelings in these languages, as examples (44) versus (45) illustrate; the emergent meaning clearly depends on the collocation of the verb plus its complement. When, in Maasai, -ibor ‘(be) white’, is combined with ‘eye’, for example, the resulting meaning reflects a disfavoured social disposition. Similar collocations of ‘white’ and ‘eye’ are attested elsewhere in Nilotic, as in Lango (Western Nilotic):

(47) wàŋ-ɛ́ tàr eye-3sg 3sg:white:hab ‘(s)he is lewd’

This property, of meaning emerging through collocations with specific nouns, is not unique to these languages. In Ewe, a Kwa (Niger-Congo) language of Togo and Ghana, expressions with ‘stomach’ may also yield antonymic interpretations (Felix Ameka, personal communication): Colourful psi’s Sleep Furiously 159

(48) é-ɸé dɔ-me nyó 3sg-poss stomach become.good ‘(s)he is kind, generous’

(49) é-ɸé dɔ-me vé-á 3sg-poss stomach be.painful-hab ‘(s)he is unkind, stingy’

(50) é-vé dɔ-me na-e 3sg-be.painful stomach dat-3sg ‘it angered him/her’

The variation in meaning can only be explained if we take into account the inherent meaning of lexemes as well as the meaning emerging from collocation with other lexemes. The approximately 400 languages constituting the Bantu group within Niger-Congo (the largest phylum in Africa), are spread over an area stretching from Cameroon east to Kenya and south to South Africa. The following examples of figurative speech involving body parts are drawn from three genetically and geographically distant Bantu languages: Mongo (Democratic Republic of Congo), Swahili (East Africa, in particular Tanzania and Kenya) and Zulu (South Africa). Taylor and Mbense’s (1998) study of Zulu is one of the rare studies of a Bantu language concerning the expression of emotions. They base their findings on fieldwork, as well as on Doke and Vilakazi (1953), who present various examples of attributive expressions involving the root for ‘heart’ as the locus of anger.

(51) -nhliziyo-hlutu heart-snap ‘quick-tempered’

(52) -nhliziyo-mbi heart-bad ‘evil-hearted’

(53) -nhliziyo-mbili heart-two ‘unreliable, double-minded’ 160 CHAPTER 8

(54) -nhliziyo-mfushane heart-short ‘short-tempered, impatient’

(55) -nhliziyo-ncane heart-small ‘impatient, quick-tempered’

(56) -nhliziyo-nye heart-one ‘unchanging, good-hearted’

(57) -nhliziyo-nhle heart-good ‘good-hearted’

(58) -nhliziyo-nde heart-long ‘patient, long-suffering’

As in many other African languages, constructions involving ‘say’ plus an ideophonic predicate (expressing a sensory reaction) are also common in Zulu (data from Taylor and Mbense 1998: 199):

(59) inhliziyo i-thé canu heart su-say:pfv ideo ‘I felt sick/angry (lit. my heart said canu)’

In the distantly related Bantu language Mongo, the word for ‘heart’ also figures prominently in figurative speech. Hulstaert (1957: 335–336), describing the northwestern dialect, translates bo-téma as ‘1. heart, inner part; 2. belly’. Given that in Meeussen (1967: 101) reconstructs *-tíma as a Proto-Bantu form for ‘heart’, the other shades of meaning in Mongo are presumably an extension or secondary development historically. (Data based on Hulstaert 1957.)

(60) botɛ́ma nd-âlikó heart prep-sad ‘fearful, sad’ Colourful psi’s Sleep Furiously 161

(61) ale botɛ́ma fwɛɛ 3sg:be heart boiling ‘(s)he is hurt, angry’

(62) ale botɛ́ma bobé 3sg:be heart malice ‘(s)he is hurt, angry’

(63) ale la botɛ́ma bobé 3sg:be prep heart malice ‘(s)he is selfish’

(64) botɛ́ma bǒolátsa heart sadness ‘(s)he is sad’

(65) botɛ́ma wɪ̂sei heart gen.compassion ‘a compassionate heart’

(66) botɛ́ma bǒtsitsima heart appeased ‘calm/appeased’

(67) ale la botɛ́ma béfé 3sg prep heart insincere ‘(s)he is insincere’

(68) botɛ́ma ˈôfʾa nkɛlɛ heart without anger ‘easily agitated but quickly appeased as well’

There is an additional term in Mongo, bolóko, which Hulstaert (p. 215) translates as ‘heart, interior, marrow’, and which is also common in figures of speech:

(69) bolóko bokáki interior hanging ‘fearful’ 162 CHAPTER 8

The Proto-Bantu root for ‘heart’ *-timà, which has been retained as such in languages like Mongo (but not in Zulu), in fact means ‘liver’ in a number of other Bantu languages (Guthrie 1970: 108). This type of shift in meaning is of course common cross-linguistically (Wilkins 1996). It may also be one of the reasons why ‘heart’ and ‘liver’ are sometimes used interchangeably in figures of speech, as observed for Nilotic languages like Päri above. In Zulu, there are quite a few expressions involving the word for ‘liver’, isiɓindi, which Doke and Vilakazi (1953: 78) translate as 1. courage, boldness; 2. liver; 3. essential internal part of anything; 4. species of hard tree-growing fungus. The translation ‘courage, boldness’ presumably emerges from metaphorical expressions such as the following:

(70) unesiɓindi lomuntu liver this.person ‘this person has courage’

The distantly related Bantu language Swahili did not retain the Proto-Bantu root for ‘heart’; instead, it uses a form m-oyo, which is a reflex of the Common- Bantu form *-yòyò ‘life’ (Guthrie 1970: 201). So this word, which is commonly used in figurative speech, is the result of metaphorical replacement, as in the following expressions:

(71) -ji-pa moyo refl-give heart ‘take heart, pluck up courage’

(72) -piga moyo konde hit heart fist ‘cheer up, make a bold resolve’

(73) -tia moyo place heart ‘encourage, hearten’

(74) -shuka moyo lower heart ‘be depressed’

(75) -shupaza moyo harden heart ‘harden the heart’ Colourful psi’s Sleep Furiously 163

A derived, diminutive form of the word for ‘heart’, ch-oyo is used in the following expression:

(76) ku-wa na choyo inf-be prep heart:dim ‘to be covetous, grudge’

Whereas constructions involving ‘heart’ as a metaphor predominate in Swahili, the ‘liver’, ini, is also considered to be a seat of feelings:

(77) maneno yale ya-li-m-kata ini words those su-past-ob-cut liver ‘those words cut him to the heart’

Kilian-Hatz and Schladt (1997), in their discussion of body part idioms in the expression of emotion, cognition and volition in a number of African languages suggest that ‘heart’ and ‘stomach’ are always positively marked, while the majority of the other inner organs like ‘liver’, ‘kidney’ or ‘gall’ are negatively marked (similar to English ‘he is hearty’ as opposed to ‘he is liverish’). But the examples above do not support this claim as a general principle, since they show that it is the collocation which determines the emergent meaning. There is some evidence that ‘colour’ terms also play a role in figurative speech in Bantu languages like Zulu. It is important to keep in mind that lexemes translated as ‘white’, ‘red’, or ‘black’ in the interlinear glossing in the examples below are polysemous in Zulu. -mhlophe, for example, corresponds to English ‘white’, but it also expresses ‘pale-coloured, faded’, as well as ‘pure, faultless, innocent’ and ‘destitute, empty’. This raises a more general question, that of how to distinguish between primary (inherent) and secondary meaning. Whereas in the case of Maasai it could be argued, on the basis of the comparative evidence, that ‘black, dark’ is an old meaning, and ‘empty’ or ‘angry’ is a secondary development, there is no a priori reason to believe that meanings referring to hue or colour are always the more basic. This in turn raises a wider, empirical problem in the study of these phenomena, that of the “primary” translation provided in interlinear glossing. Since detailed, etymological dictionaries are lacking for most of these languages, it is not always possible to address these important issues in a satisfactory manner. The following examples from Zulu (Doke and Vilikaze 1953) illustrate the glossing and translation. 164 CHAPTER 8

(78) isitsha esimhlophe geqe dish white ideo ‘an absolutely empty dish’

(79) -nhliziyo-mhlophe heart-white ‘calm, peaceful, unruffled, pure-hearted’

Similarly, the adjectival root -mnyama in Zulu corresponds to English ‘black, dark-coloured’, but it may also mean ‘deep, profound, unfathomable’, ‘confused, hazy, dizzy’, ‘ill-omened, dreaded’, ‘lacking in appetite’, or ‘gloomy, angry’.

(80) -nhliziyo-mnyama heart-black ‘lacking in appetite, gloomy’

Whereas some of these meanings, for example the link between ‘deep, profound’ and ‘dark’, as in ‘deep waters’, are straightforward, others (‘lacking in appetite’) would seem to require chains of meaning extensions, a type of semantic change that is still poorly understood cross-linguistically. Taylor and Mbense (1998: 200) make reference to a folk belief amongst Zulu speakers, that nausea is caused by an excess of black bile (inyongo), which can percolate into the heart giving a person a ‘black heart’. It may indeed be such beliefs which help to understand semantic widening of the type illustrated here. Similarly, the root -ɓomvu in Zulu could be translated as ‘red’, but it also means ‘angry’, as in the following expression:

(81) -nhliziyo-ɓomvu heart-red ‘bad-tempered, angry’

The same root occurs in the word for ‘ red soil’, isi-ɓomvu. So it would have to be considered ‘derived’ rather than ‘basic’ in the sense of Berlin and Kay (1969). From this it would also appear that the reference to a particular hue is more basic than the meaning ‘bad-tempered, angry’. Zulu terms like -ɓomvu are polysemous, presumably involving metonymic transfer from a physical symptom such as hue to some psychic effect or state historically. Whether this physical symptom is in fact to be equated with colour terminology in the sense of Berlin and Kay (1969) is a question addressed in the following section. Colourful psi’s Sleep Furiously 165

8.4 Interpreting Colourful psi’s

The arbitrariness of form-meaning relationships in languages has been central to linguistics ever since de Saussure. However, evidence has been accumulat- ing suggesting that this relationship is sometimes motivated or even iconic, even beyond areas where this is immediately obvious such as onomatopoeia. This has become particularly clear in the field of sound-symbolic or ideophonic words. For example, it has been known for some time that there is a cross-linguistic tendency for concepts such as ‘large’ to be conveyed by use of low tone, and/or vowels with low F2 (for example a, ɔ, u), and/or consonants with low acoustic frequency. The antonym ‘small’, on the other hand, may be conveyed with high tone, vowels with high F2 (e.g., i, ɪ, e) and/or consonants with high acoustic frequency. Ohala (1994) and others refer to this phenomenon as the ‘Frequency Code’. At the semantic level, the frequent use of body part terminology in metaphorical expressions is another universal: “Among the most easily adaptable tropes are those from domestic relationships and activities and from parts of the body” (Kennedy 1998: 61). Within this, the description of mental states through expressions involving body part terminology is such a widespread tendency. It remains to be determined to what extent colour terminology is subject to universally-based principles as well. The discussion below is intended as a contribution to the attempt to find such universally-based conceptualizations in this domain. The philosopher Quine (1960: 26–90) observes that ‘alien concepts’ can be carried into English in innumerable ways, each acceptable on its own, so that there is nothing to choose between various renditions. (See also Chapter 6.) Approaches to the semantics of lexical entries, for example in the dictionary work referred to above, often rely on denotational overlap with English or French words (or any other language into which terms are translated) to establish the meanings of lexemes. This is particularly clear in the area of colour terminology (Chapter 6). The fact that many Nilotic languages, for example, do not have a word for ‘colour’ suggests that we should not treat the connection between the lexical item and the object or referent as a socially decontextualised ‘natural categorisation of nature’.3 Lexical items in Bantu and Nilotic

3 This kind of criticism of the Berlin and Kay paradigm in fact goes back to the 1970s. See, for example, Tornay (1978) and also more recent studies such as Lucy (1996), Wierzbicka (1996: 287–334), Goddard (1999) and references therein, for similar observations. The actual categorization of “colour” terms as verbs, adjectives or nouns, has consequences for their referential meaning, for example as dynamic events or states (Chapter 6). It is also a widely accepted 166 CHAPTER 8 languages involving reference to colour are typically polysemous. It would be erroneous to claim that the primary meanings of -mhlophe and -mnyama in Zulu are indeed the denotational value of the English abstract colour terms ‘white’ or ‘black’ as manifested in the Munsell Colour chart, thereby viewing all other meanings or denotations as extensions. Note, for example, that the Zulu terms are not reflexes of Proto-Bantu ‘white’ (*-yédù) and ‘black’ (*-yi̦dù́ ), so their current meaning must have emerged from some other lexical item. Their meaning, when they are used as colour terms, may also depend upon the material or texture with which they are associated. For example, the word for ‘green-blue/GRUE’ may receive a slightly different interpretation in Nilotic languages depending on whether it refers to the sky or skins. This phenomenon is not unique to these languages. In Sudanese Arabic, the word axḍar ‘green’ may refer to the colour of leaves. But with reference to human skins, it denotes shades that would be referred to as ‘pitch black’ in English; see Bender (1983) for a description of skin colour terms in this variety of Arabic.4 In Sudanese Arabic, and in many other African languages, colour terms such as ‘red’ are also used in order to describe skin colours that would be considered ‘black’ from the point of view of a white English speaker. It is important to note, therefore, that these colour terms and their corresponding meanings are not necessarily to be identified with the abstract colour terms emerging from the use of the Munsell Colour Chart. A straightforward explanation in terms of physical properties, such as hue, for the way in which these terms are used in figurative speech (paralleling Ohala’s ‘Frequency Code’) is not yet available. These complicating factors may also turn out to be important for an understanding of cross-linguistic variation in the denotational and connotational meaning of colour terms in figurative speech of the type discussed here. Interestingly, such expressions always seem to be used in languages in tandem with non- idiomatic speech. The question arises, then, why languages use such alternative strategies in the first place. This issue is central to the next and final section.

contention that lexical richness in particular domains, whether expressed through nouns, verbs or adjectives, is culturally revealing. The question whether such formal distinctions in syntactic categorisation also reflect differences in conceptualizations between speakers of different languages as Wierzbicka (1995) argues, is still controversial. 4 According to Bosworth et al. (1986: 698–707), colour terms are commonly used in tropes in different varieties of Arabic. In Syrian Arabic, a ‘white heart’ designates a person of good companionship or noble nature; in Levant Arabic, a ‘yellow smile’ is a smile full of envy and a ‘blue enemy’ is an inveterate and mortal enemy. Colourful psi’s Sleep Furiously 167

8.5 On psi’s and fta’s

It is a rather striking fact that languages usually have alternative strategies, not involving figurative speech, for the expression of mental states. Thus, in the Nilotic language Maasai there are various verbs expressing emotional conditions, as in the following examples of infinitives, some of which are marked with a middle voice suffix (-a/-o).

(82) a-isin-a ‘to be sad’ a-gor-o ‘to be angry’ a-ng’id-a ‘to be proud of, rejoice, be happy, to boast’ a-lom ‘to be jealous’

For the Bantu languages discussed above, too, there are alternative strategies not involving metaphorical or metonymic extensions. These include verb roots which in some cases are so widespread and stable, that they can be reconstructed for their common ancestor, Proto-Bantu. Thus, Guthrie (1970) identifies a number of “common” verb roots with regular reflexes (and similar or identical meanings) in a range of Bantu languages:

(83) *-yí�̦m Common-Bantu -im- Duala -im- Nyoro -im- Taabwa -(j)im Yao ‘become mean, refuse to give’

(84) *-cáŋg- Common-Bantu -ʃaŋg- Mbomotaba -sāŋg- Kongo -hlak’ɛl- Sotho ‘be pleased’

Interestingly, languages use these lexical strategies in tandem with alternative strategies involving metaphorical and metonymic expressions. Indeed, there are often pairs of expressions, one from the plain series and the other involving figurative language. In the Nilotic Luo, for example, positively evaluated emo- tionals such as love, joy, and happiness are more frequently rendered by means of the verb lexeme than negatively evaluated emotions, with the latter often involving borrowings from neighbouring (Bantu) languages (Reh 1998: 21). 168 CHAPTER 8

One possible pragmatic motivation for the use of multivocality may be found in the role of language in social interaction. In their influential study of speech acts and their relevance for politeness strategies, Brown and Levinson (1987) argue that giving hints or association clues, understating or overstat- ing, being ironic, as well as the use of metaphors represent convenient ways of making off-record or indirect statements. Such speech act strategies contrast with on-record strategies, which involve pragmatically transparent ways of performing a (speech) act. The latter are potentially ‘face threatening acts’ (fta’s), and thus may require redress strategies. The more imposing and face- threatening a given speech act, the more indirect the strategy chosen by the (rational) speaker may be, according to Brown and Levinson. Commenting on the social behaviour of others, for example, is a potentially face-threatening act, which may affect ‘Negative Face’ wants (expression of restraint), and Positive Face wants (expression of solidarity). As an off-record (indirect) strategy, the use of metaphors or figurative speech possibly provides a way of minimising the imposition. (Of course, such expressions may serve other functions as well, including esthetic ones.) How this works is not quite clear. One relevant factor may be that such constructions are often in paradigmatic contrast with a non-idiomatic form with a more literal meaning. The usage and interpretation of the former requires verbal skill (and, as pointed out by one referee, the dis- play of one’s assumption that the addressee shares these skills which creates the effect). Consequently, the conversational implicature may be that the speaker has made an attempt to be indirect particularly in areas or domains associated with negative feelings or taboo. In the Africanist literature, there are few in-depth studies available on discourse strategies in social interaction, a rare exception being Strecker (1988) on the Omotic (Afroasiatic) language Hamar. Strecker shows that hinting as well as the use of metaphors abound as off-record strategies in this language. For example, when inviting a person to join in on a meal, the host may ask the guest to ‘sit down to spit on to his heart’. As a figure of speech, such an expression vio- lates (Gricean) conversational maxims of Quality and Relevance (Grice 1975). Its meaning remains obscure without specific cultural knowledge. In fact, it is common among the Hamar for the host to spit milk on the guest’s chest as a customary blessing before starting to eat. Similarly, the Hamar expression ‘my belly is green’, used in order to express happiness or satisfaction, is presumably another instance of culturally defined imagery related to the dominant role played by cattle in their culture; a green stomach is a stomach filled with grass. And again, amongst these pastoralists in southern Ethiopia, cattle raids are commonly referred to as ‘talks’, with the raiders being ‘hunters’. Here, the Roman concept nomen est omen may be invoked; such euphemisms allow for Colourful psi’s Sleep Furiously 169 a ‘hiding’ strategy, in particular in areas of high risk and taboo. Understanding their meaning thus requires integration of cultural knowledge. If one views the language faculty as an autonomous system, i.e., as a disembodied mental construct, common linguistic phenomena such as the types of figurative speech illustrated above remain rather enigmatic. As Strecker (p. 154) argues, “[i]rony, rhetorical questions and the like are the tools of the powerful and being such they encode the authority of those who may legitimately use them” amongst the Hamar; of course, one learns about such cultural practices through structural coupling within social groups. Knowledge of the cultural context (through primary language socialisation), of the inherent meaning of lexical items, and of the meaning emerging from collocation with other syntactic entities, allow native speakers of Hamar to function in a world that is meaningful to them. One additional, important topic related to figures of speech, not discussed until now, involves the rhetoric of emotion and the cultural appreciation of emotional states. The classic way of initiating such research which has long been known to provide important analytical insights into culture-specific con- figurations of ideas about thoughts or feelings, is through the study of lexical distinctions in the target language. It is rather striking that in the Bantu language Swahili, for example, there are various lexical items relating to feelings of anger which have been borrowed from the Semitic language Arabic.5 This phenomenon, also observed for Luo (Reh 1998), requires further research from a psychological and ethnographical point of view.

Swahili Gloss of Swahili Gloss of Arabic equivalent (85) hasira ‘anger, wrath, passion’ ‘oppression of heart’ ghadhabu ‘rage, fury, passion, anger’ ‘be angry’ kasirani ‘anger, bitterness of heart, vexation’ ‘cast down one’s looks’ ghaidi ‘1. anger; 2. determination, resolution’ ‘wrath, anger exasperation’

Fillmore’s (1975: 114) observation, “when you pick up a word, you drag along with it a whole scene”, is highly appropriate with respect to these lexemes. The translations also make clear another important phenomenon, namely the alternative conceptualizations of emotional experience across cultures; in the case of Swahili historical contact with Arabic and Persian Islamic culture presumably played a role. Understanding their rhetorical effect in social sym- bolisation, i.e., their cultural meaning, requires native-speaker competence in such languages, and knowledge of a wide set of appropriate discourse scenar- ios. Such studies are not yet available in the field of African linguistics.

5 See Wehr (1979) for the corresponding forms in Arabic. CHAPTER 9 Perception of the Living Dead and the Invisible Hand in Teso-Turkana

9.1 Introduction

In the border area between Kenya, Uganda, South Sudan and Ethiopia a group of Nilotic languages is spoken which has come to be known as the Teso-Turkana cluster. It forms part of the Eastern branch of Nilotic and consists of two languages: Teso, and a dialect continuum spoken by different ethnic groups, more specifically the Jiye, the Karimojong, the Nyangatom, the Toposa, and the Turkana. It should be pointed out, however, that the dialect boundaries within the latter cluster do not coincide with the traditional territorial boundaries of these respective ethnic groups. For example, northern Turkana is distinct from southern Turkana, but essentially identical with the neighbouring variety of Toposa. Traditionally, these groups are transhumant pastoralists, although all of them also practice agriculture. In the course of the twentieth century, many Teso became sedentary and started concentrating more on agricultural activities (Lawrence 1957, Karp 1978). Also, some Turkana groups have become fish- ermen along the shores of Lake Turkana, often as a result of cattle loss during severe draughts. Several anthropologists have studied the culture of a number of these groups, and this chapter discusses their contributions. In addition, I myself have carried out linguistic fieldwork in the Turkana area, resulting, among other things, in a grammatical description of the language (Dimmendaal 1983), so part of the data and analyses presented here are based on my own field observations. This chapter takes a traditional anthropological-linguistic approach sometimes referred to as “ethno-science”. It involves the investigation of lexical terminology as the “linguistic manifestation” of material and spiritual culture, and the study of contexts in which these lexical concepts are used. More specifically, section 9.2 investigates widespread terms for the “living dead” as (partly) invisible forces. The meaning of at least one cognate lexeme varies across these groups. It is this second aspect—semantic change and the metaphorical concept of the “invisible hand” of speakers using these terms in different contexts,

© koninklijke brill nv, leiden, ��5 | doi ��.��63/9789004224148_010 Perception of the Living Dead and the Invisible Hand 171 thereby changing their meaning (Keller, 1994)—which is central to section 9.3. The chapter concludes with some observation on perception verbs in these languages (section 9.4).

9.2 Invisible Forces in Teso-Turkana

Over the past few decades, a number of studies have appeared which investigate religious concepts among groups speaking Teso-Turkana languages, for example Barrett (1987; 1998) on Turkana, and Novelli (1999) on Karimojong cosmology. Both Barrett and Novelli were Catholic priests who lived among these groups for several decades. Van der Jagt (1989) also based his anthropological and theological account of Turkana religion on fieldwork in the area. As these and other studies on Teso-Turkana point out, there is a general term, ( ɲ)a-kuj(u), which may be translated as ‘God’. The prefix a-, which interestingly is a feminine gender marker, is pronounced as ɲa- in Jiye, Toposa and northern Turkana, a generalised dialect variant that applies to the other examples below. The vowel in parentheses indicates that it is devoiced prepausally and realised as an extra short vowel otherwise. Etymologically, the nominal root is related to kuju ‘up, higher elevation’, a term which is also used these days to refer to the cardinal direction ‘north’. Its antonym kwap may be translated as ‘down, on the ground’ (compare a-kwap ‘land, country’) or ‘south’. These directional markers are associated with the direction in which water flows, from higher (mountainous) zones towards lower regions, or from the sky towards the earth. As is common cross-culturally, specific human beings are assumed to be capable of acting as malefactors in Teso-Turkana speech communities. Teso refers to this act of bewitching as -cʊd (Hilders and Lawrance 1958: 33). A cognate form usually combined with a habitual suffix, -cʊd-aan, occurs in the remaining lects of the Teso-Turkana cluster. For example, the corresponding root form, -cʊd, expresses ‘weed, pluck, pick, shear, nibble, mow, cut’, in Turkana, according to Barrett (1990: 9). Somebody capable of bewitching other members of the community, however, is usually referred to by way of a nomen agentis form, ɛ-kapɪlan(ɪ) for a male and a-kapɪlan(ɪ) for a female, based on a verb root -pɪl. The latter is apparently no longer used in its root form. Instead, one finds a partially reduplicated adjectival stem -pɪpɪl ‘painful, aching, susceptible’. In addition, there is an abstract noun, a-kapɪlaanut, expressing the state or structural property of ‘causing pain, looking with evil eyes, thinking evil’. 172 CHAPTER 9

One widespread lexeme in Teso-Turkana, ( ɲ)ɛ-kɪ-pɛ, itself probably of considerable historical time depth in Eastern Nilotic (see section 9.3), may be translated as ‘evil spirit’. It contrasts with a noun referring to ‘ancestral spirit’ (example 1), which appears to be mostly benevolent.1

Singular Plural (1) Turkana ɛ-paara-ɪt ŋɪ-paara Karimojong e-parait ŋipara Toposa (no singular) ŋɪ-paara Teso e-para i-para

The etymology of the nominal root -paara is unclear. It may be related to a widespread root meaning ‘shine’; compare Bari -para ‘to shine far in the distance’ (Spagnolo 1960: 234). Related (derived) nouns refer to ‘day(time)’, as in Turkana paaran (Barrett 1990: 64); compare also Bari paran ‘at noon, at mid- day’ (Spagnolo 1960: 234). For the Turkana, maring ‘epilepsy’ and other diseases are caused by human actors, for example, malefactors such as ŋɪ-kapɪlak ‘(male) witches, individuals catching an evil eye’, or ŋa-kapɪlak ‘female witches’. Alternatively, ancestral spirits and A-kuj ‘God’, “[. . .] indeed [. . .] any ‘extraordinary’ thing like a person, animal, bird, snake or insect [. . .] may cause misfortune” (Barrett 1998: 92). There is a range of terms referring to ancestral spirits or the “living dead” in Turkana and other members of the Teso-Turkana cluster, although, as Barrett (1998: 97) points out with respect to the former, “[. . .] the Turkana do not live in a demon-filled world”. Schröder (1993a, b & c) makes similar observations with respect to the neighbouring Toposa. The bad ancestral spirits (ŋɪ-paara) that cause possession are also known as ngi-kerep, ngi-tuwa, and ngi-kamitiiri in Toposa. Ngi-kerep are very aggressive and cause symptoms associated with schizophrenia, while ngi-tuwa cause seizures, and ngi-kamitiiri cause idiocy. There does not appear to be one common, overarching or general name for ‘spirits’ in Teso-Turkana. Instead, a common strategy is to call spirits by the same names as their corresponding manifestation in humans. For example, the term ŋi-kerep (masculine plural) refers not only to ‘madness, epilepsy’, but also to the spirits causing this pathology. Then, as we will see below, there are

1 The prefixes in the nouns with the root -pa(a)ra are masculine gender markers. Teso-Turkana languages distinguish between masculine, feminine, and neuter gender as obligatory inflectional features of nouns. Number is indicated by way of suffixes; see Dimmendaal (1983) for further details. Perception of the Living Dead and the Invisible Hand 173 also some common and widespread general terms, not coterminous with the name of their manifestations or with the name of a specific disease. Members of the community who have been notable in preventing and cur- ing diseases during their lives all become ŋɪ-karam after they pass away.

Singular Plural (2) Turkana ɛ-karam-ɪt ŋɪ-karam Toposa ɲɛ-karam-ɪt ŋɪ-karam

The ŋɪ-karam appear a lot in dreams, and are very much involved in daily life, especially where ritual is concerned. They may, for example, influence the run- ning of the home. When people and animals are sick, these benevolent dead are sometimes thought to be annoyed. The etymological source of the nominal stem -karam is not clear. Its most likely source is the verbal root -ram ‘beat’ (a root which is also found elsewhere in Eastern Nilotic, e.g. in Bari, -ram ‘knock (down)’), preceded by a nominal gender prefix ka- (itself derived from a locative preposition with the same form). ‘Beating’ in this sense is presumably to be understood as a penalty or corrective measure, inducing proper social behaviour. The following lexeme refers to a kind of malevolent spirit in Teso-Turkana speech communities:

Singular Plural (3) Turkana e-cen-it ŋi-cen Karimojong e-cen-it ŋi-cen

This time, the lexical source is clear. The nominal form is derived from a verb -cen ‘anathematise, damn, curse, blame’. The ŋi-cen are revengeful or vindictive spirits of the departed. They are not necessarily bad by nature, but they must take revenge for the evil which they suffered during their lives. Consequently, they may blame the living and bring trouble or annoy them, or they may inflict punishment on them and bring diseases. Although there is a productive system of agentive noun formation in Teso- Turkana, and although in at least one case (ŋi-cen) the verbal origin of the noun is clear, none of the lexemes referring to ancestral spirits is derived through the regular nomen agentis formation process in these languages, in spite of the fact that these ancestral spirits are thought to be capable of acting or instigat- ing some process. Regular agentive nouns in these languages are derived from verbs by means of a prefix ka- and a suffix -an(ɪ) ~ -ɔn(ɪ) ~ -on(i) in the singular, and -ak ~ -ɔk ~ -ok in the plural, with the variants depending on the vowel 174 CHAPTER 9 harmony rules of the language in question. Compare ɛ-ka-pɪl-an(ɪ) ‘witch’ above, which has the prototypical structure of an agentive noun. Another interesting linguistic property of these terms, again suggesting that their morphological properties are semantically motivated, is that Teso-Turkana languages, as often in Nilo-Saharan, distinguish between three types of nouns with respect to number. Depending on the language, nouns are either (i) inherently singular, taking a plural suffix to mark plurality; (ii) inherently plural or collective, taking a singulative suffix to mark singularity; or (iii) inflected both in the singular and the plural. Interestingly, ( ɲ)ɛ-kɪ-pɛ ‘evil spirit’, takes a plural suffix -an (ŋɪ-py-an), thereby showing that its basic form is the singular. However, with the remaining terms discussed above, ŋɪ-paara, ŋɪ-karam, ŋɪ-cen, the collective (or plural) constitutes the morphologically unmarked and thereby the semantically basic form, since corresponding sin- gulatives are formed with a suffix -ɪt or -it. For Turkana, there are a number of additional terms referring to ancestral spirits, e.g. ( ɲ)a-wiyenit (sg), ŋa-wiyenit-o (pl). These spirits cause people to do bad things like stealing animals. Another type of malevolent spirit is e-pin-it (sg), ŋi-pin (pl). Its lexical source is not known, and neither is it clear whether the term is also used to refer to a particular disease. As Barrett (1987: 53) admits, “[ . . . t]he Turkana refer to their ‘living-dead’ with many words, which I have found very difficult to translate.” One of the additional terms he mentions is a noun ( ɲ)e-torube (sg), ŋi-torube-i (pl). The same term is attested in other members of the Teso-Turkana cluster. Its primary meaning appears to be ‘shadow which a person or object casts’. Ancestral spirits or ghosts like the ŋɪ-paara may be encountered at night, near graves for example, where they appear as ‘shadows’, i.e. as ŋi-torube-i. The ŋi-torube-i of good ancestors are not dangerous, unless someone has angered them. This term does not seem to refer to a separate category as such, but instead to a manifestation of a specific spirit. So the term ‘shadow’ appears to be a phenotypical label, rather than being a genotype of some sort. Linking spirits to shadows is traditionally widespread in the area and occurs cross-culturally. In his description of the Nandi, a Southern Nilotic community, Hollis (1909: 41) states that “[ . . . t]he human soul is embodied in a person’s shadow, and it is firmly believed that after death the shadows of both good and bad people go underground and live there.” Schröder (1993c) points out that for the Toposa there are two levels of supernatural beings. The supreme deity is nya-kuj ‘God’. The lesser divinities are ngi-pean ‘nature spirits’, and ngi-para ‘non-nature spirits’. The latter term only occurs in the plural in Toposa, according to Schröder (1993c), who further notes that two different spirit concepts are covered by this collective term, namely spirits of remembered dead and impersonal spirits (possibly of generations Perception of the Living Dead and the Invisible Hand 175 no longer remembered) who died of old age, i.e. those with sufficient authority. A more recent development further subdivides those ngi-para that cause diseases into those that are exorcised by established diviners and those that are exorcised by a new category of female diviners as part of the lo-poryang ritual in Toposa society. They are always invisible, except when they appear in dreams in the shape of the deceased. They do not come until several years after the person has died. The ngi-pean are associated with natural phenomena and water and their movement can be detected, whereas the ngi-para are not, and their movement is unknown. Ngi-pean are active both day and night, whereas ngi-para, especially the ancestral type, are active when people are asleep, typically by night. Members of the Ngi-kateok clan among the Toposa are said to have particular mediating powers to appease the ngi-pean. As with the Turkana, the ngi-pean particularly manifest themselves in natural phenomena like thunder, lightning, storms, and whirlwinds. Disasters like floods or falling trees are expressions of their anger; they also produce rainbows. For the Toposa, ngi-pean are unpredictable, mischievous and capricious; ngi-para on the other hand, both the personal and impersonal types, are more predictable and even-tempered. If an ancestor was easily angered in life, or if things went wrong at the time of the burial, then the spirit is also angry and may need to be appeased. The Toposa consider that crises are caused by spiritual forces, either bad ones, such as the capricious free nature divinities (ngi-pean) or the easily offended spirits of the dead (ngi-para), or by God himself (nya- kuj) (Schröder 1993c).

9.3 Perception of the Invisible Hand

Whereas the etymological source of some of the terms used in reference to benevolent and malevolent ancestral spirits in Teso-Turkana is unclear, there is one term whose lexical origin can be traced back to the Proto-Teso-Turkana stage and beyond with considerable certainty, given its widespread distribution across Eastern Nilotic. It may be reconstructed as *-pɛ for Proto-Eastern Nilotic. Compare the following examples with translations based on the original sources (Data from Farina (1986) for Karimojong, Schröder (1993c) for Toposa, and Barrett (1990) for Turkana):2

2 In Teso, there is another term probably containing an ancient lexical root, a-jok-it (sg), i-jok-in (pl), that Hilders and Lawrance (1958) translate as ‘evil spirit’. The root-jɔ(ɔ)k is widespread synchronically in Western Nilotic languages. 176 CHAPTER 9

Singular Plural (4) Karimojong ekipye ŋipyan ‘lightning’ Toposa nyekupe ngipean ‘natural spirit’ Turkana ɛ-kɪ-pɛ ŋi-py-an ‘anything extraordinary’

There are at least two reasons for assuming that the root form *-pɛ represents an archaic term in the cosmology of Eastern Nilotic speech communities. As Dimmendaal (1988: 22) argues in a contribution on Proto-Nilotic reconstructions, historically the prefix kɪ- was added to a set of monomoraic nouns in Eastern Nilotic, in order to avoid monomoraic stems. The Proto-Eastern Nilotic forms below are based on Vossen (1982).

Proto-Nilotic Proto-Eastern Nilotic (5) *ma(c) *kɪ-ma- ‘fire’ *pi(-R) *-pi ‘water’ *riŋ *ki-riŋ ‘meat’

According to Vossen (1982), Eastern Nilotic is divided into a Bari group and a Non-Bari group; the latter in turn consists of Lotuxo-Maa and Teso-Turkana. Whereas Vossen (1983) reconstructs a root *-pi ‘water’ for Proto-Eastern Nilotic, he reconstructs a form with a corresponding prefix *ki- (*ki-pi) for Proto-Non-Bari. One common innovation of the Non-Bari group involves a historical change from covert to overt gender marking for nouns (Heine and Vossen 1983). As a result, the prefix *ki-, originally added in order to create bimoriac foot structures for words, became incorporated into these nominal stems. Compare the reflexes for ‘meat’ in Turkana and Lopid, a Lotuxo dialect within Lotuxo-Maa:

Singular Plural (6) Turkana (ɲ)a-ki-riŋ ŋa-riŋo ‘meat’ Lopid hi-riŋ-o riŋ-oi ‘meat’

The Lopid singular form probably contains an original plural (or collective) suffix. Although this word does not have a nominal gender in this dialect of Lotuxo, the corresponding Lokoya dialect does: a-xi-riŋ-o (Vossen 1982: 393). The fact that the word for one type of evil spirit, -pɛ, also has an incorporated prefix kɪ- in the singular (where a monomoraic root would otherwise occur) already suggests that the lexeme goes back to an early stage in Eastern Nilotic. A second piece of evidence is that a cognate root is attested in both the Perception of the Living Dead and the Invisible Hand 177

Bari and Lotuxo-Maa groups within the Non-Bari branch. In Bari, a verbal root pɛ ‘fulminate, strike (of lightning)’ occurs (Spagnolo 1960: 235). In Maasai, a member of the Lotuxo-Maa branch, a verbal root -pɛ ‘be wild, troublesome, jit- tery, skittish’ is found (Mol 1996: 327). The past tense root form in Maasai, a-te- pi-a, manifests the same type of glide formation as found in the plural nominal form in Turkana and Karimojong (ŋi-py-an). This morphophonological rule is no longer productive synchronically in Teso-Turkana. The lexical meaning of the cognate forms in Bari and Maasai also matches the meaning of the lexical root in Teso-Turkana. The term ɛ-kɪ-pɛ in fact is polysemous in several Teso-Turkana lects. In Karimojong it not only expresses ‘evil spirit’, but also ‘flash, stroke of lightning’. In Teso only the latter meaning is attested: ekipie ‘striking (lightning)’ (Hilders and Lawrance 1958: 43). Whereas the masculine noun ( ɲ)ɛ-kɪ-pɛ is the general, generic term for this malevolent spirit, associated with long, delayed lightning, a corresponding feminine form may be used in order to express a female spirit, associated with sudden, quick lightning. The term in Karimojong or Turkana denotes both the natural phenomenon, and the spirit which causes its devastating effects, i.e. both the phe- notype and the genotype. Marshall Thomas (1965: 180), in her description of the culture of the Dodoth, a section of the Karimojong, quotes a man called Lopore:

The ngipian are seen in lightning and appear like mist below the clouds, for the ngipian can float in the air. Sometimes they roam the earth, following the rivers, and appear above wet places early in the morning like wraiths [. . .] If you displease them, they can make your cows stick in the mud or catch you in a whirlwind which will drive you mad [. . .] It will occupy your body and eventually kill you unless an emuron [‘traditional healer’] paints you with clay stripes like the stripes of a rainbow, then persuades the ekipi to enter a goat, leads the goat around a tree so the ekipi will climb the tree, and kills the goat.

Lawrance (1957: 183), quoting Kitching (1912), points out that among the Teso “[. . . s]pecial superstitions centre around lightning. A person injured by a flash had to wear bells round the ankles for weeks afterwards.” Being struck by lightning constitutes one of those facts of life that are totally inexplicable for the Karimojong: “During a storm, while you are walking with a friend on the path to the village, you are struck by the lightning, while your friend remains unscathed” (Novelli 1999: lvii). Such events are not taken to be acts of God because they escape every logical connection with the Karimojong concept 178 CHAPTER 9 of God. A lightning strike can leave somebody stunned for some time.3 This experience is so frightening that, by extension, every other mysterious hap- pening that negatively affects humans and is not attributable to a more precise cause, may be called ɛ-kɪ-pɛ. This tendency towards metonymic extension parallels the use of the term ŋi-kerep to refer to the specific spirits and their phe- nomenology. Turkana uses ( ɲ)ɛ-kɪ-pɛ to refer to a person, a domesticated or wild animal or anything behaving in an unusual or extraordinary way or found in the wrong place, hence Barrett’s (1998: 92) translation, ‘anything extraordinary’. A person who possesses certain talents, such as training a bullock’s horns or showing expressive dancing talents to his favourite ox, may also be called a ( ɲ)ɛ-kɪ-pɛ. A young man in the Turkana settlement where I lived had picked up some Dutch, simply by listening to the conversation of two Dutch tourists who had visited the Turkana area for one or two days. When I told an adult Turkana fellow villager how impressed I was with the language talents of this young Turkana man, his reaction was:

(7) mm, ɛkɪpɛ ŋesi yes extraordinary 3sg ‘yes, he is an ɛkɪpɛ’

From the tone of his voice, and from his subsequent comments about this Turkana youngster’s cunning behaviour, it was clear that the interpersonal meaning of this sentence was something like “yeah, he is a smart alec (be careful of him)”. As a result of the new contexts in which the term ɛ-kɪpɛ (or its plural form, ŋi-pyan) has apparently come to be used, particularly in urban contexts, it has developed into a generic term for ‘extraordinary thing’ through metonymic extension. And this is where we observe “the invisible hand” as a historical notion.4 As Keller (1994) points out in his analysis of semantic shifts in the historical development of languages, unintentional causal consequences emerge from the use of terms, more specifically from individual intentional acts by speakers in different social contexts. These are the invisible hands manipulating meaning and creating additional and thereby new meaning, in much the same way as when people trace a path that the landscape gardener never planned, simply by all of them crossing the lawn in the same place regularly.

3 Traditionally, those who died a violent death, for example after having been struck by lightning, were left where they died. 4 Keller in turn based his concept on a notion that Adam Smith developed for economics in his influential 1776 study, The Wealth of Nations. Perception of the Living Dead and the Invisible Hand 179

9.4 A Note on Perception Verbs in a Nilotic Context

In a seminal contribution on the role of culture in our cognitive system, more specifically our perception, Evans (2003) argues that the extension of perception verbs to cognition readings in Indo-European (Sweetzer 1990) does not necessarily apply to other language families, and by implication not to other cultures. Proto-Indo-European *weid- ‘see’ developed into a verb meaning ‘know, understand’ in Germanic languages, e.g. German wiss-en, or Dutch wet- en. Sweetzer (1990) assumes that these verbs, involving intrafield (or trans- field) extensions from ‘see’ to ‘know’, are based on more universal conditions or cognitive states. However, as Evans (2003) argues on the basis of the historical- comparative study of Australian languages, it is clear that ‘know’ commonly develops from a verb ‘hear’, rather than ‘see’ in these languages. One widespread verbal root for ‘know’ can probably be reconstructed for Proto-Nilotic as *na (Dimmendaal 1988: 46), its reflex in Teso-Turkana being -na-ɪkɪn. In this cluster the root is only used in combination with a dative extension -ɪkɪn in order to express ‘get accustomed to, get to know’. When combined with the stative formative suffix -a (-na-ɪkɪn-a), its meaning is ‘be familiar with, be used to each other’. But there is no evidence that this root goes back to a root either meaning ‘see’ or ‘hear’ in Nilotic. The verbal stem for ‘hear’ -ɪrar in Teso-Turkana expresses a cognitive sense of ‘understanding’; its etymological origin is unknown. Neither the term -na-ɪkɪn nor -ɪrar play a special role in talking about the perception of spirits. In terms of visual experiences, spirits in Teso-Turkana are associated with colour rather than smells, for example when they manifest themselves as ‘shadows’ by night or in sandstorms by day. Whirlwinds, absorbing yellow sand in desert-like areas, are also seen as manifestations of ŋi-pyan. A Turkana traditional healer (e-muro-n(i) (sg), ŋi-muro-k (pl)) may take a sick person to a tree that has been buffeted by the wind, and make a sacrifice there to appease the spirits. He may kill a “yellow” goat (a-kine na-ɲaŋ), yellow being the colour of the whirlwind.5 This buffeted tree is also referred to as the ‘tree of the spirit’, ( ɲ)ɛkɪtɔɪ a ( ɲ)ɛkɪpɛ since it is also considered to be the locus of spirits. The sick person is made to ‘eat the benevolent ancestors’(akɪ-ɲam ŋɪkaram), the act of

5 The adjective -ɲaŋ corresponds to ‘light brown’ or ‘beige’ in English, but also to the colour of urine; see Tornay (1978) for a discussion of colour terminology in the Nyangatom lect of Teso-Turkana. The yellow-coloured dust-devils which constantly harass the inhabitants of the desert in the Turkana area stand “[. . .] between black (good, cool) and red (dangerous and hot)” (Barrett 1998: 92). 180 CHAPTER 9 eating referring to chewing and spitting out parts of the slaughtered goat in a ritual manner. What about other types of cognitive activities potentially activated in the perception of the “invisible” in Teso-Turkana or other Nilotic speech communities? Can one also feel, hear or touch them, maybe? Accounts in the literature—confirmed by my own observations during an approximately 18 month stay among the Turkana—tell us that the spirits of the dead are thought to visit people at night and whisper in their ears, i.e. they may be heard as well as seen in dreams. This perceptive activity or process is usually referred to by way of the verb stem –aɲ-un in Teso-Turkana. Barrett (1990) translates this term as ‘perceive, apprehend, find, discover, get, achieve, acquire’. The root -aɲ- is no longer used in isolation, and is found only in combination with the marker -un, expressing movement towards the deictic centre (or speaker). The etymological origin of the verbal root again is unknown. Interestingly, the terms for ‘perceive (plus related senses)’ vary widely across Nilotic, which would make an investigation of their etymological origin in individual languages a rewarding task. More extensive dictionaries are required, however, in order to make this kind of research possible. Contrary to some other cultures, one cannot smell spirits, as far as our present understanding of Teso-Turkana cosmology goes. The lexical root to express ‘smell, stink’ appears to be fairly stable in Nilotic—compare the widespread Nilotic root -ŋʊ /-ŋu; Karimojong -iŋu-ore, Proto-Kalenjin (Southern Nilotic) -ŋʊ- but it does not seem to play a role in describing the perception of spirits. Neither is there any evidence in Teso-Turkana of the kind of ritualised language for communicating with spirits found in some Nilotic speech communities (e.g. special registers among the Luwo in South Sudan; Storch, 2014). This may be due to lack of proper understanding of these cultures, but given that some of the accounts in the literature are based on extensive fieldwork covering several decades, these phenomena—if present—do not appear to be very prominent in the area. CHAPTER 10 Conversational Styles in Tima

10.1 Introduction

Ever since Hymes’ (1968) seminal contribution on the “ethnography of speaking”, the description of cross-linguistic differences in speech styles has continued to receive fresh impetus in the scientific literature. With respect to African speech communities, numerous studies have appeared using Hymes’ framework to describe greeting strategies and other types of ritualised speech behaviour; see Ag Youssouf, Grimshaw and Bird (1976), McIntyre (1980), and Nwoye (1993), to mention but a few titles. One of the methodological principles that Hymes (1968) advocates is the investigation of local conceptualisations through the words which name them, an “ethno-scientific” approach followed in the discussion of Tima lexical domains in Chapter 4. The same emic approach forms the basis for an investigation of emotional states and the way Tima categorises them in this chapter.1 The descriptive model that Hymes developed for the description of speech behaviour in different communities includes a concept of “speech norms”. Brown and Levinson (1987), in their influential model on face-saving and face-threatening speech acts, developed this latter concept into a general model for the comparison of “politeness strategies” between different speech communities. Strecker (1988) remains one of the rare monographs to use this model for an in-depth analysis of the role language plays in day-to-day interaction in an African speech community, the Hamar, speakers of an Omotic language in south-western Ethiopia. One interesting discourse feature in Hamar is the use of ideophones in social interactions, as an important Positive-Face strategy (showing closeness, as against Negative Face strategies, which show deference or social distance). Through the use of such expressive strategies speech participants claim common ground. The expressive use of ideophones in Tima is one of two aspects of conversational styles discussed below (section 10.2). The second topic involves emotional states as reflected in Tima (section 10.3).2 With

1 The differentiation between etic and emic approaches goes back to Pike (1954). 2 Additional properties of the “ethnography of speaking” in Tima are described in Alamin, Schneider-Blum and Dimmendaal (2012) on spatial orientation, Schneider-Blum and Dimmendaal (2013) on perception, and Dimmendaal (2013) on the grammar of knowledge.

© koninklijke brill nv, leiden, ��5 | doi ��.��63/9789004224148_011 182 CHAPTER 10 respect to the latter, Wierzbicka (1999: 240) observes: “[ . . . a]lthough human emotional endowment is no doubt largely innate and universal, people’s emotional lives are shaped, to a considerable extent, by their culture”, a position I also subscribe to.

10.2 Ideophones

Ideophones are words depicting sensory imagery, used by speakers to share in sensory experiences and to sort out matters of experiential knowledge. As such, they play an important role in Tima as part of the Positive Face strategies inventory (Brown and Levinson 1987). It is sometimes thought that ideophones are particularly prominent in narrative discourse (storytelling). But as Dingemanse (2011) points out, in what is probably the most detailed account of ideophones in a particular speech community to date, ideophones in Siwu (a Kwa language spoken in Ghana) are also found in greeting routines, funeral dirges, and day-to-day interaction. A similar picture emerges in Tima: ideophones are indeed common in storytelling, but they also occur in different conversational contexts, including mundane social interactions. Ideophones are typically rendered as adverbs in Tima, and tend to occur at utterance boundaries, usually at the end of a sentence. But they may also stand by themselves, as in example 1, where someone is explaining how ocra is prepared:

(1) ì-túyùk-àt̪áŋ wʊ̀dàná twàc twàc cɛ́-ŋʊ̀dàná twàc twàc 1pl-beat-compl cry ideo ideo 3:impV.cry ideo ideo ‘we beat them until twac twac they cry twac twac’

Example (1) is part of a conversation between four speakers discussing the various procedures for harvesting and preparing beans. A gesture depicting bean threshing reinforced the perlocutionary force of the ideophones,3 one of four common strategies listed in McNeill’s (1992) gesture taxonomy. The other three are deictic gestures (pointing gestures that indicate either concrete or imagi- nary objects or people), emblems (highly conventionalised gestures that can be said to have a lexicalised meaning) and beats (movements that do not represent a discernible meaning). All of these gesture types are extremely important features of Tima conversational strategies.

3 Dingemanse (2011: 348) confirms that gestures co-occurring with Siwu ideophones are usually depictive in nature. Conversational Styles in Tima 183

Example (1) also illustrates an additional conversational strategy: the repetition of an ideophone. Other participants may add this to the conversation, as an “echoing” strategy that expresses common ground and agreement. Reduplication is the formal strategy par excellence for the expression of iconic relationships, as Chapter 4 illustrates with Tima ideophonic adverbs describing shape, taste, textures, and colours. But reduplication is actually found in all major grammatical categories. Take nouns for example. Even though it is not usually possible to identify the etymology when reduplication occurs, they do tend to refer to entities or objects which occur in large numbers or quantities in the natural environment:

Singular Plural (2) mɘ̀nɛ̀mɘ̀nɛ̀k ɪ̀-mɘ̀nɛ̀mɘ̀nɛ̀k ‘locust species’ kɨ̀-mʌ̀nɨ́mʌ̀n ì-mʌ̀nɨ́mʌ̀n ‘star’ kɪ̀-ŋɛ̀rɪ̀ŋɛ̀rɪ́m ɪ̀-ŋɛ̀rɪ̀ŋɛ̀rɪ́m ‘pebble’

As for verbs, modern Tima uses a number of morphological strategies to express repetition of an action or process. One pattern—relatively unproductive— involves suffixation of a pluractional marker -t ̪-; another pattern—irregular though slightly more productive—involves root-internal alternation. But a third pattern involves reduplication of the verb root, and in these cases the root itself does not usually occur in isolation:

(3) -mìlímìl ‘rub’ -mùnúmùn ‘smile’ -mʌ̀ɽɨ́mʌ̀ɽ ‘decorate, brand (bowls)’ -mòɽúmòɽʌ̀k ‘pray, ask for favours’ -lɛ̀wʊ́lɛ̀wʊ̀ ‘blink’ -tɔ́htɔ̀h ‘stutter’

Iconicity is also present in interjections. Whereas Tima uses a short (disyllabic) word to chase away a dog, it uses quadruplication to attract its attention:

(4) hɪ́yà used to chase away a dog yɛ̂hyɛ̂hyɛ̂hyɛ̂h used to call a dog kúúrùkúrkúrkúr used to call chickens hàwʊ̀hàwʊ̀hàwʊ̀ used to make a donkey stand still hàyhàyhày used to call goats múrràmúrràmúrràmúrrà used to call a cat 184 CHAPTER 10

Unlike interjections, ideophones are not immediate responses to events, but rather depictions of them (Dingemanse 2011: 155). They frequently relate to the manner in which a certain action or process takes place or is carried out, and to a lesser extent to the description of states. The following examples illustrate ideophonic adverbs that describe states (albeit often resulting from some action):

(5) yɛ̂h ɪ̀-tàrɘ̀mt̪àrɘ́m sorghum pred-roughly.ground ‘the sorghum is roughly ground’

Examples (6)–(8) illustrate the use of ideophonic adverbs with dynamic verbs:

(6) kàykàtʊ̀kwɛ́ɛ̀l cɛ́-láàl dɔ̀ŋ̀kɔ̀ldɔŋkɔ̀l hunter prog-sneak ideo ‘the hunter sneaks up with a stoop’

(7) án-dɔ̀ɔ̀ cɔ̀rtɔ̀cɔ̀rtɔ̀k 3:pfv-rise ideo (s)he / they got up stiffly’

(8) rííh yáàm t̪ɪ̀mɪ̀t̪ɪ̀mɪ̀k plait hair ideo ‘plait the hair densely!’

Ideophonic adverbs may also stand in paradigmatic contrast to each other, i.e. they are not necessarily tied to individual verbs:

(9) cɛ́m-pánʊ̀ʊ̀k hɔ̀rhɔ̀r 3:prog-breathe ideo ‘(s)he is / they are breathing quickly’

(10) cɛ́m-pánʊ̀ʊ̀k ŋʌ̀hŋʌ̀h 3:prog-breathe ideo ‘(s)he is / they are panting’

(11) díìk hùrtùhùrtùk walk ideo ‘walk hesitantly!’ Conversational Styles in Tima 185

Some ideophonic adverbs, including the one in example (11), can also be changed into verbs (example (12)):

(12) mánà hùrtútúúk-áá-dʌ̀ why hesitate-instr-like ‘why are you (pl) hesitating?

Dingemanse (2011: 150) points out that in Japanese the more frequently ideophonic adverbs are used, the more easily they can be used as verbs. As for Tima, such derived verbs also acquire other properties inherent to verbs, including nominalisation, e.g. kù-hùrtùtúúl ‘hesitation’. For further details, see Dimmendaal and Schneider-Blum (To appear). The use of a light verb like ‘do, make’ and/or ‘say’ in so-called coverb constructions is common in a wide variety of African languages. The complements of these verbs may be nouns, adjectives or adverbs, and the latter tend to be ideophonic. In some languages, the ‘do, make’ versus ‘say’ contrast corresponds to an intransitive versus transitive distinction. Amha (2001) describes such a system for the Omotic language Wolaitta. Devos and Bostoen (2012) show that in the Bantu language Shangaci the generic action verb ‘do, make’ *-kɪ́t- developed into a quotative verb ‘say’ (-ira) used, amongst other contexts, in combination with ideophones (Example 13, adapted from Devos and Bostoen 2012: 106):

(13) nkhirá wáawe osálá wíiré khóng’óngoo mu-khira o-awe o-sal-a o-ir-e khong’ongoo pp-tail pp-poss pp-remain-pfv pp-qv-psit ideo ‘his tail remains like a stump’

Tima uses a verb -kɔ- (or kwɔ-), ‘do, make’ in combination with ideophonic adverbs to express positions of the body:

(14) kɔ́yɔ̀ twʌ̀ŋkìŋtwʌ̀ŋkìŋ ‘hop on one leg’ kɔ́yɔ̀ tùrùtùrùk ‘do a headstand’

Akita (2009) argues that there is a kind of “Lexical Iconicity Hierarchy” whereby ideophones depicting sounds are the most likely candidates for lexical specification: sound < movement < visual patterns < other sensory perceptions < inner feelings and cognitive states figure 8 Lexical Iconicity Hierarchy (Akita 2009) 186 CHAPTER 10

Akita (2009) also states that iconicity in the Lexical Iconicity Hierarchy increases the higher a semantic field appears in the listing. According to this theory, if a language has only one type of ideophone, they tend to be very iconic, such as those denoting sound. But a language can never have only less iconic ideophones, such as those representing inner feelings and cognitive states. Akita further claims that cross-linguistically, the more iconic an ideophone is, the more likely it resembles ideophones in other languages depicting the same meaning. However, the Tima data does not support this claim, as far as present knowledge goes. Speakers appear to be more occupied with visual patterns, physical states, and movement than with sounds. Whereas lexically, Tima, Katla and Julud are quite close (together forming the Katloid group), a comparison of ideophonic adverbs in these languages does not render any obvious cognates. As in Tima, reduplication is one of the formal properties of Katla ideophones, although this language also attests quite a few monosyllabic roots ending in a consonant:4

(15) kpàkpàt ‘the cutting of sorghum’ fâk ‘breaking through a hole, bursting out of a hole’

The frequent coinage of new ideophones and the obsolescence of those that have become “worn out”, not only in these languages but also cross- linguistically, suggest a high degree of communicative dynamism. The relative instability of such ideophones (compared to other lexemes) historically indeed does not appear to be unique to these languages. Dingemanse (2011: 327) shows how creativity and idiosyncrasy in the use of ideophones play an important role in Siwu. He identifies eleven cases of ideophone coinage in a total sample of one hundred and five types. As one Siwu speaker points out (Dingemanse 2011: 337), “[ . . . ] in the heat of the moment, when you look at some event, you can sculpt your speech to form a depiction of it; and your fellow interactants will understand you, not because you have chosen the right sequence of speech sounds—for that is the point: the one right depiction does not exist—but because your fellow interactants will have understood that you are launching into a depictive performance and will treat it as such.”

4 I would like to express my gratitude to Birgit Hellwig for making Katla ideophones available from her dataset. Conversational Styles in Tima 187

10.3 Emotional States

The role of figurative language in the conceptualisation of emotions has been a popular topic in cognitive linguistics (see Chapter 8). Some authors have propagated an approach that considers the embodied conceptualisations of anger as being both physiological and socio-cultural, for example, Ansah’s (2011) study of anger in Akan, a Kwa language of Ghana. As Wierzbicka (1999: xx) argues “[. . .] given the universal real physiology, members of different cultures cannot conceptualize their emotions in a way that contradicts universal physiology (or maybe even their conceptualization of universal physiology); but nevertheless they can choose to conceptualize their emotions in many different ways within the constraints imposed on them by universal physiology.” Parallel to the nature-nurture debate with respect to the perception of colour or spatial orientation (Chapters 6 and 7), the debate about the nature of human emotions is fraught with equally contradictory views. Interpretations oscillate between biological reductionism (“emotions arise from human biology”) and social constructionism (“emotions are products of culture”).5 But metaphorical language about the emotions and human physiology in emotion are both part of an integrated system (Kövecses 2000). The range of metaphorical and metonymic expressions for mental states already discussed for a variety of African languages (see Chapter 8) is found in many other languages and this also applies to expressions reflecting emotional states. English expressions such as “boiling with anger”, “be filled with fear, shame or sorrow”, “overflowing with joy or love” have parallels in other languages. Kövecses (2000: 156) mentions Hungarian, Japanese, Chinese, Wolof and Tahitian as additional languages where fluid in a pressured container is a preferred metaphor when expressing feelings of anger. Conceptual metaphors bring two distinct domains into correspondence with each other, one typically more physical or concrete than the other. This linking helps to understand the less concrete concept in terms of the more concrete one. Metonymy on the other hand involves a “stand-for” relationship (Kövecses 2000: 5). Kövecses (2000: 187) presents a balanced view by pointing out that “[. . . f] eeling states are also, in part, culturally determined. This is because events that evoke parallel emotions in different cultures are unlikely to induce them in precisely the same way.” Anybody who has lived in a different culture for some time presumably subscribes to this view. Cultural aspects of emotions

5 This “oscillation” mirrors differences of opinion in research on gender or class in the social sciences, or the nature and origin of criminal behaviour. 188 CHAPTER 10 already appear to be reflected in the metaphorical uses of notions such as “hot” or “cool”. For the Turkana in Kenya (who speak a Nilotic language), a person described as “hot” is socially unpleasant (e.g. mean, stingy), rather than sexu- ally attractive and desirable, as in Anglo-American culture. Example (16) shows that coolness is associated with positive experiences in Tima:6

(16) kʊ́mbɔ́yɔ́ŋ à-lìl-àt̪áŋ á-kɘ̀ɽɘ̀lɛ̀ kind.of.wasp pred-cool-compl prep-kind.of.wasp ‘the kumboyong wasp is less harmful than (lit.: is cool from) the kerele wasp’

A core notion in Tima social interaction is the verb -hɛbʊk which may be translated as ‘socialising, chatting, showing emotions’ (Schneider-Blum 2013):

(17) cɛ̀-hɛ́bʊ̀k 3:prog-socialise ‘(s)he is / they are extrovert / showing positive emotions’

The positive connotations of this verb contrast with -kwárt̪àk, a verb that contains the (archaic) pluractional marker -t ̪-, suggesting that this phenomenon is not interpreted as a punctual event, but as a repetitive behaviour or character trait.

(18) cɛ́ŋ↓-kwárt̪àk 3:prog-show.off ‘(s)he is / they are arrogant, showing off’

Social interaction involving speech may also be characterised as -ŋúl ‘stinking’, a qualification not only used for physical objects but also for slandering. Interestingly, emotions or emotional states in Tima culture are not associated with the stomach, liver, or heart, but with the bone, k-ûh (sg)/y-ûh (pl), as a metaphorical container.7 The same lexeme is also used in the sense of ‘inside, interior’ in combination with the locative marker l-, which replaces the nominal class prefix k- (sg) or y- (pl) in such cases:

6 This preferred mental state of being “cool” (i.e. admirably up to date) is also reflected in (African) American English. 7 Whether this metaphor or metonym is widespread in languages of the Nuba Mountains is not yet known. Conversational Styles in Tima 189

(19) l-ûh kùrtú loc-bone house ‘inside the house’

(20) án-cɔ̀ɔ̀ ɨ̀-l-ûh 3sg:ipfv:past-go prep-loc-bone ‘(s)he / they went inside’

(21) pɨ́↓nʌ́ yánt̪ɘ̀ yúh ↓ɪ́ɪ́hwáá 3sg inside bones gen.people ‘(s)he is / they are among the people’

The same locative form is used to refer to the heart of Tima country, the valley basin surrounded by mountains:

(22) l-uh kɔ̀ɔ́ loc-bone family ‘at the heart of Tima country’ (lit. ‘at the bone of the family’)

As k-ûh is also used in a metaphorical sense to express ‘soul, spirit’, it is best translated as ‘1. bone; 2. interior; 3. soul, spirit’ as a lexical entry in a dictionary (see Schneider-Blum 2013: 310–311). But as example 23 illustrates, its meaning may also extend to ‘creativity’:

(23) cɛ́-dàhɪ́ɪ́-dʌ̀ ŋ̀-kúh ↓lɛ́ɛ́nɪ́ prog-say-1sg instr-bone my ‘I have an idea’

When a close relative passes away, Tima speakers may comfort the next of kin with the following idiom:

(24) tók k-ûh squeeze sg-bone ‘be strong!’

‘Refusing the bone’ is the Tima idiom for the expression of ‘envy’ or ‘jealousy’:

(25) kʌ̀húnèn cèŋ-kʌ́wúl-áát̪áŋ k-úh mʌ̀mì woman 3:prog-refuse-instr.compl sg-bone co-wife ‘the woman is jealous of the co-wife’ 190 CHAPTER 10

The same lexeme, k-ûh, is the central concept in a range of other idioms, all expressing different permanent character features or temporal emotional conditions or states, as the following antonyms (involving the concepts of ‘light’ and ‘dark’, or ‘white’ and ‘black’ in English) illustrate:

(26) k-ûh à-hɛ́l sg-bone pred-light ‘be good-natured, not resentful’

(27) k-ûh à-tún sg-bone pred-dark ‘be envious, hold a grudge’

To describe someone as having a certain character trait or being in a certain emotional state, Tima uses a syntactic frame whereby k-ûh occurs in subject position. The person affected by this mental condition is denoted by means of a prepositional phrase in combination with the main predication in the case of nouns or third person pronouns, or by means of a marker y- (presumably cognate with and derived from the locative preposition ɨ) followed by a pronominal marker encliticised on the verb in the case of first and second person pronouns. Example (28) and (36), expressing happiness and sadness, respectively, illustrate these syntactic strategies:

(28) k-ûh à-hín-yɛ̀ɛ́n sg-bone pred-sweet-loc.1sg ‘I am happy’

Several other adjectives related to perceptual stimuli are used in combination with k-ûh to differentiate between mental states. The person affected by this state tends to be presented as a topic when nominal (with k-ûh occurring in subject position). The topicalised noun or noun phrase is co-referential with a pronominal element encliticised onto the predicate and expressing the affected object.

(29) Tàmbóshé k-úh à-dɪ̀ʔɪ́l-yáŋ ɪ̀- ɪ̀hwáá índíʔíl Tamboshe sg-bone PRED-clean-LOC.3SG/PL PREP-people all ‘Tamboshe is generous with everybody (lit. Tamboshe, she has a clean bone with all the people)’ Conversational Styles in Tima 191

(30) k-ûh à-yàkɔ́m-yàáŋ sg-bone pred-strong-LOC.2sg ‘you are patient, persevering’

Examples (31–34) are additional idioms involving adjectival predicates:

(31) k-ûh à-dʊ̀ʔál sg-bone pred-hot ‘be furious’

(32) k-ûh à-líl sg-bone pred-cold ‘be calm, deliberate, considerate, harmless’

(33) k-ûh à-kɨ́k sg-bone pred-bitter ‘be brave’

(34) k-ûh à-rɔ̀ʔɔ́r sg-bone pred-bad ‘unkind, angry, resentful’

In addition, k-ûh may be used in combination with verbs to express more dynamic, temporary situations, as in (36) and (37).8

(35) y-ûh ɪ̀-rɔ̀ɔ̀r-yáŋ ì-ihínʌ́ pl-bones pred-forbidden-LOC.3SG/PL prep-3pl ‘they are sad (lit. forbidden)’

(36) k-ûh àn-cá↓kálɘ́k á-yɛ̀ɛ́n sg-bone 3:pfv-destroy prep-loc.1sg ‘I am in a bad mood (lit. the bone has been destroyed from me)’

(37) k-ûh cɛ́-ŋʊ́dánà yɛ̀ɛ́n sg-bone prog-sing loc.1sg ‘I have hunger pangs (lit. the bone is singing as to me)’

8 Baldi (2011) addresses the role ideophones play in rendering emotions. There is no evidence for this in Tima. 192 CHAPTER 10

In many African languages and Indo-European creoles in the Caribbean, the verb ‘eat’ occurs with inanimate subjects to express physical or mental processes9 and this is also the case in Tima, where such inanimate subjects include k-ûh:

(38) cè-kʌ́lùk ŋ̀-k-ûh 3:prog-eat erg-sg-bone ‘(s)he / they have a stomach ache (lit. the bone is eating her/him/them)’

In such dynamic constructions, the bone is not only the container, it also oper- ates as a force. The use of ‘bone’ as a container for emotions appears to be rare universally, at least in the existing literature. But perhaps this is because investigators have been focusing too much upon ‘heart’, ‘liver’, ‘stomach’ and other soft organs when analysing the expression of mental states (see Chapter 8). A closer look at German, for example, shows that there are in fact quite a few expressions for emotional states involving ‘bone(s)’:

(39) das geht auf die Knochen it goes on the bones ‘it exhausts you’

(40) ihm sitzt die Angst in den Knochen him sits the fear in the bones ‘he is scared stiff’

(41) keinen Mumm in den Knochen haben no energy in the bones have ‘to have no guts’

(42) sich bis auf die Knochen blamieren oneself till on the bones make.a.fool.of ‘to make a proper fool of oneself’

(43) er ist ehrlich bis auf die Knochen he is honest till on the bones ‘he is deadly honest’

9 See, for example, Haitian French: dan ap manje m ‘I have a toothache (lit. the tooth is eating me)’. Conversational Styles in Tima 193

Wierzbicka (1999) is probably one of the first linguists to take an “ethno-scientific” approach, by looking at expressions and at the same time developing a sensitivity to different emotional norms across cultures, as in the following statement (Wierzbicka 1999: 274–275):

Just as the English concept of “blue” doesn’t match the Russian concept “goluboj” (“sky blue”), the Polish concept “niebieski” or the Japanese concept’ “aoi” [. . .], so the English concept of “anger” doesn’t match the Ifaluk concept “song” or the Italian concept “rabbia”. This doesn’t mean that there are no “universals of seeing”, or that there are no “universals of feeling”, but it does mean that in our search for these universals we should carefully listen to how people in different cultures talk about what they see and how they feel [. . .]; and that we should avoid analytical categories based on culture-specific aspects of our own languages.

These are extremely important considerations when trying to describe emotional states in Tima and their corresponding translations in English. We can only arrive at more accurate translations of the various Tima idioms that use ‘bone(s)’, however, by studying their use in day-to-day interactions. The purpose of this survey of Tima idioms centring around the conceptualisation of human physiology and emotions is to lay the foundation for future investigations.

10.4 Exoteric and Esoteric Languages

Exoterogeny occurs if a speech community has extensive ties with other communities and its language is also spoken as a contact language by members of those communities. Swahili, today an important lingua franca for major parts of Eastern Africa, is an exoteric language that originated as a local contact medium on the coast. Extensive lexical borrowing appears to be one diag- nostic property of such language types. Its alternative, esoterogeny, whereby speakers of a language add linguistic innovations that increase the complexity of the language in order to highlight their distinctiveness, is less well known and understood. Language manipulation is known to play an important role in contemporary urban youth language across Africa. But the strategies attested in these languages are probably nothing but a continuation of ancient traditions; see Storch (2011) for an extensive survey of this phenomenon in Africa and Europe. Esoterogeny and exoterogeny do not reflect two alternative strategies; they are rather stages on a continuum involving different types of language 194 CHAPTER 10 ideologies. There is no immediate evidence for conscious manipulation in Tima (see Chapter 4). Nevertheless, Tima people adhere to an esoteric language ideology (their ancestors having manipulated the language in order to make it incomprehensible to their neighbours, the Katla), and this may be true for other speech communities in the Nuba Mountains. This, at least, is suggested by the low degree of multilingualism and corresponding absence of phonological or grammatical convergence phenomena (see Chapter 3). The lexical structure of Tima fits in with this concept of exclusiveness. In her analysis of lexical strategies in Tima, Schneider-Blum (2012) points towards the extensive use of metaphors, synecdoche and metonymic extensions, thereby compensating for the relatively small number of lexical roots. These parallel strategies in the Athabaskan language Dene Sᶙɫiné of Canada: “Minimal borrowing would lead to an over-reliance on the indigenous lexicon. However, when that endolexicon is as small as it is in the typical Dene language, there could be no recourse except to re-deploy and over-extend the basic vocabulary, either through conversion or through new composite constructions [ . . . A]ny new lexicalization would tend to be idiomatic and with idiomaticity would come a high degree of opacity that would keep any new generation of outsiders from cracking the code. The esoteric nature of the language would thus persist, if not strengthen over the generations” (Rice 2012: 71). Such predispositions of speakers can be attributed to a habitus (in the sense of Bourdieu 1977, i.e. the dispositions or schemata acquired through the experiences of everyday life) guiding the structuring of social practices, more specifically of individual actions in social contexts “[. . .] without in any way being the product of obedience to rules” (Bourdieu, 1977: 72). Models of turn-taking organisation as developed in models of conversation analysis (Schegloff 2007 and others) provide ideas on how the notion of habitus could be “cashed out” or applied to future studies on social interactions in Tima or other languages of the Nuba Mountains. The description of various Tima lexical-semantic fields in Chapter 7 serves as a basis for future comparative research in other languages in the area in order to test to what extent these strategies are unique to Tima, and to what extent they are ancient areal phenomena or Africanisms. The extensive documentation now collected and archived for Tima indeed allows for a much more extensive analysis of conversational strategies and interactional routines.10 One intriguing property of turn-taking devices in Tima social interactions is that the principle of “current speaker selects next

10 The Tima narrative discourse corpus can be accessed in the DoBeS archive (http://www .mpi.nl/dobes/projects/tima). Conversational Styles in Tima 195 speaker” (as a turn-taking device in theories of conversation analysis) is sometimes absent. Instead, one observes “current speaker continues” and “next speaker self-selects”. The turn-taking machinery as a universal basis for conversations hence may manifest “local flavors” and “tweaked versions” (Levinson 2006: 56, 61). Here too, conversational styles in this fascinating Nuba Mountain community may provide us with additional insights into human interaction. However, most of the video footage still awaits in-depth analysis. The domains discussed in this chapter will hopefully serve as initial steps towards a more detailed account of speech norms in the Tima community, but this endeavour requires a separate monograph.

References

Ag Youssouf, Ibrahim, Allen D. Grimshaw and Charles Bird. 1976. Greetings in the desert. American Ethnologist 3: 797–824. Aikhenvald, Alexandra Y. 1996. Areal diffusion in northwest Amazonia. Anthropological Linguistics 38: 73–116. ———. 2007. Grammars in contact: A cross-linguistic perspective. In: Alexandra Y. Aikhenvald and R.M.W. Dixon (eds.), Grammars in Contact: A Cross-Linguistic Typology, pp. 1–66. Oxford: Oxford University Press. ———. 2012. The Languages of the Amazon. Oxford: Oxford University Press. Aikhenvald, Alexandra Y. and R.M.W. Dixon (eds.). 2001. Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics. Oxford: Oxford University Press. Aissen, Judith. 2003. Differential object marking: Iconicity vs. economy. Natural Language and Linguistic Theory 21: 435–483. Alamin Mubarak, Suzan. 2006. An initial description of Laggori noun morphology and noun phrase. In: Al-Amin Abu-Manga, Leoma Gilley, and Anne Storch (eds.), Insights into Nilo-Saharan Language, History and Culture, pp. 9–24. Cologne: Rüdiger Köppe. Alamin, Suzan. 2013a. Verbs in Laggori. In: Schadeberg and Blench (eds.), pp. 451–462. ———. 2013b. Tima adjectives. In Schadeberg and Blench (eds.), pp. 251–268. Andersen, Elaine S. 1978. Lexical universals of body-part terminology. In: Joseph H. Greenberg (ed.), Word Structure: Universals of Human Language, pp. 335–68. Stanford: Stanford University Press. Andree, Richard. 1878. Über den Farbensinn der Naturvölker. Zeitschrift für Ethnologie 10: 323–34. Ansah, Gladys Nyarko. 2011. Emotion language in Akan: The case of anger. In: Gian Claudio Batic (ed.), Encoding Emotions in African Languages, pp. 119–137. Munich: lincom Europa. Appel, Rene, and Pieter Muysken. 1987. Language Contact and Bilingualism. London: Edward Arnold. Baldi, Sergio. 2011. Emotions, colours and qualities: an overview of Hausa ideophones. In: Gian Claudio Batic (ed.), Encoding Emotions in African Languages, pp. 56–72. Munich: Lincom. Baroja, Tomás H. 1989. Analytical Grammar of the Pokot Language. Trieste: Università di Trieste. Barrett, Anthony J. 1987. Dying and Death among the Turkana. Eldoret: Gaba Publications. ———. 1988. English-Turkana Dictionary. Nairobi: Macmillan. ———. 1990. Turkana-English Dictionary. Nairobi: Macmillan. 198 References

———. 1998. Sacrifice and Prophecy in Turkana Cosmology. Nairobi: Paulines Publications Africa. Becker, Erik. 2011. The Prehistoric Inhabitants of the Wadi Howar: An Anthropological Study of Human Skeletal Remains from the Sudanese Parts of the Eastern Sahara. PhD dissertation, Johannes-Gutenberg Universität, Mainz. Bellwood, Peter. 2003. Farmers, foragers, languages, genes: The genesis of agricultural societies. In: Peter Bellwood and Colin Renfrew (eds.), Examining the Farming/ Language Dispersal Hypothesis, pp. 17–28. Cambridge: McDonald Institute for Archaelogical Research. Bellwood, Peter and Colin Renfrew (eds.). 2003. Examining the Farming/Language Dispersal Hypothesis. Cambridge: McDonald Institute for Archaeological Research. Bender, M. Lionel. 1983. Color term encoding in a special lexical domain: Sudanese and Arabic skin colors. Anthropological Linguistics 25: 19–27. ———. 2000. Nilo-Saharan. In: Bernd Heine and Derek Nurse (eds.), African Languages, An Introduction, pp. 43–73. Cambridge: Cambridge University Press. Berlin, Brent. 1992. Ethnobiological Classification: Principles of Categorization of Plants and Animals. Princeton: Princeton University Press. Berlin, Brent and Paul Kay. 1969. Basic Color Terms: Their Universality and Evolution. Berkeley: University of California Press. Black, Mr. and Mrs. K. 1971. The Moro Language Grammar and Dictionary. Khartoum: Sudan Research Unit, University of Khartoum. Blench, Roger. 1999. Recent fieldwork in Ghana: Report on Dompo and a note on Mpre. Ogmios 11: 14–15. ———. 2000. The classification of Nilo-Saharan. Afrika und Übersee 83: 293–307. ———. 2006. Archaeology, Language, and the African Past. Lanham: Alta Mira Press. ———. 2013. Introduction to the Temein languages. In: Schadeberg and Blench (eds.), pp. 485–500. ———. To appear. Baŋgi Me, a language of unknown affiliation in Northern Mali. Mother Tongue. Blust, Robert. 1985. Review of Language and Living Things (Brown). Language 61: 891–97. Boas, Franz. 1891. Some general report on the Indians of British Columbia. Report of the British Association for the Advancement of Science 1890: 562–715. Bolinger, Dwight L. 1967. Adjectives in English. Lingua 18: 1–34. Bosworth, C.E. et al. (eds.) 1986. The Encyclopedia of Islam, Vol. v. Leiden: Brill. Boyeldieu, Pascal. 1977. Eléments pour une phonologie de Laal de Gori. In: Jean-Pierre Caprile (ed.), Etudes phonologiques tchadiennes, pp. 186–198. Paris: selaf. ———. 2009. The modified form of Shatt Damam nuns and its Daju cognates. Afrika und Übersee 90: 9–84. References 199

Bourdieu, Pierre. 1977. Outline of a Theory of Practice. Cambridge: Cambridge University Press. Branstetter, K.B. 1977. A reconstruction of Proto-Polynesian color terminology. Anthropological Linguistics 19: 1–25. Brown, Cecil H. 1976. General principles of human anatomical partonomy and speculations on the growth of partonomic nomenclature. American Ethnologist 3: 400–24. ———. 1977a. Lexical universals and the human language faculty. In: Muriel Saville- Troike (ed.), Linguistics and Anthropology, pp. 75–91. Washington, d.c.: Georgetown University Press. ———. 1977b. Folk botanical life-forms: Their universality and growth. American Anthropologist 79: 317–342. Brown Cecil H. and Stanley R. Witkowski. 1981. Lexical encoding sequences and language change: colour terminology systems. American Anthropologist 831: 13–27. Brown, Cecil H., Witkowski, Stanley R. and Paul K. Chase. 1981. Where do tree terms come from? Man 1: 1–14. Brown, Penelope and Stephen C. Levinson. 1987. Politeness: Some Universals in Language Usage. Cambridge: Cambridge University Press. Brown, Roger W. and Eric H. Lenneberg. 1954. A study in language and cognition. The Journal of Abnormal and Social Psychology 49: 454–62. Bulmer, R.N.H. 1968. Karam colour categories. Kivung 1: 120–33. Campbell, Lyle. 2003. What drives linguistic diversity and language spread? In: Peter Bellwood and Colin Renfrew (eds.), Examining the Farming/Language Dispersal Hypothesis, pp. 49–63. Cambridge: McDonald Institute for Archaelogical Research. Childs, Tucker. 2003. An Introduction to African Languages. Amsterdam and Phila delphia: John Benjamins. Chomsky, Noam 1957. Syntactic Structures. The Hague and Paris: Mouton. ———. 1965. Aspects of the Theory of Syntax. Cambridge: The mit Press. ———. 2011. The Generative Enterprise Revisited. Berlin: Mouton de Gruyter. Clements, G.N. and Annie Rialland. 2008. Africa as a phonological area. In: Bernd Heine and Derek Nurse (eds.), A Linguistic Geography of Africa, pp. 36–85. Cambridge: Cambridge University Press. Comfort, Jade. 2009. The locative enclitic in Uncunwee. Paper presented at the 6th World Congress of African Linguistics, University of Cologne. Comfort, Jade and Angelika Jakobi. 2011. The verb ‘to give’ as a verbal extension in Uncunwee (Kordofan Nubian). In: Raija Kramer, Holger Tröbs, and Raimund Kastenholz (eds.), Afrikanische Sprachen im Fokus, pp. 27–35. Cologne: Rüdiger Köppe. Conklin, H. 1964. Hanunoo color categories. In: Dell Hymes (ed.), Language in Culture and Society: A Reader in Linguistics and Anthropology, pp. 189–192. New York, Evanston, and London: Harper and Row. 200 References

Connelly, M. 1984. Basotho Children’s Acquisition of Noun Morphology. PhD diss., Essex University, Cochester. Corbett, Greville. 2000. Number. Cambridge: Cambridge University Press. Crazzolara, J.P. 1955. A Study of the Acooli Language. 2nd revised edition. London: Oxford University Press for the International African Institute. Davies, I.R.L. et al. 1991. Linguistic basicness and colour information processing. International Journal of Psychology 26: 311–27. De Busser, Rik and Randy J. LaPolla. To appear. The Shaping of Language: The Relationship between the Structures of Languages and their Social, Cultural, and Natural Environments. Amsterdam and Philadelphia: John Benjamins. Delafosse, Maurice. 1914. Esquisse générale des langues de l’Afrique, et plus particulière- ment de l’Afrique française. Paris: Masson. Demuth, Katherine. 1992. Acquisition of Sesotho. In: Dan Slobin (ed.), The Cross- Linguistic Study of Language Acquisition Vol. 3, pp. 557–638. Hillsdale: Lawrence Erlbaum Associates. De Valois, R.L., I. Abramov and G.H. Jacobs. 1966. Analysis of response patterns of lgn cells. Journal of the Optical Society of America 56: 966–77. de Voogt, Alex. 2009. A sketch of Afitti phonology. Studies in African Linguistics 38.1: 35–51. Devos, Maud and Koen Bostoen. 2012. Bantu DO/SAY polysemy and the origins of a quotative in Shangaci. Africana Linguistica 18: 97–132. Dimmendaal, Gerrit J. 1983. The Turkana Language. Dordrecht: Foris Publications. ———. 1987. Krongo: between universal, areal and genetic norms. Journal of African Languages and Linguistics 9: 161–77. ———. 1988. Proto-Nilotic lexical reconstructions. Afrikanistische Arbeitspapiere 16: 5–67. ———. 1991. Some observations on evolutionary concepts in current linguistics. In: Walburga von Raffler-Engel et al., Studies in Language Origins, pp. 225–44. Amsterdam: John Benjamins. ———. 1993. Review of On language (Joseph H. Greenberg). Language 69: 796–807. ———. 1995. Studying lexical-semantic fields in languages: Nature versus nurture, or Where does culture come in these days? Frankfurter Afrikanistische Papiere 7: 1–29. ———. 2004. Head marking and dependent marking in the Nilotic area. In: F.K. Erhard Voeltz (ed.), Studies in African Linguistic Typology, pp. 71–92. Amsterdam and Philadelphia: John Benjamins. ———. 2007. Eastern Sudanic and the Wadi Howar and Wadi el Milk diaspora. Sprache und Geschichte in Afrika 18: 37–67. ———. 2008a. Language ecology and linguistic diversity on the African continent. Language and Linguistics Compass 2.5: 840–858. References 201

———. 2008b. Africa’s verb-final languages. In: Bernd Heine and Derek Nurse (eds.), A Linguistic Geography of Africa, pp. 272–308. Cambridge: Cambridge University Press. ———. 2009a. Esoterogeny and localist strategies in a Nuba Mountain community. In Wilhelm J.G. Möhlig et al. (eds.), Language Contact, Language Change and History Based on Language Sources in Africa, pp. 75–95. Cologne: Rüdiger Köppe. ———. 2009b. Tima. In: Gerrit J. Dimmendaal (ed.), Coding Participant Marking: Construction Types in Twelve African Languages, pp. 338–355. Amsterdam and Philadelphia: John Benjamins. ———. 2009c. Tama. In: Gerrit J. Dimmendaal (ed.), Coding Participant Marking: Construction Types in Twelve African Languages, pp. 307–331. Amsterdam and Philadelphia: John Benjamins. ———. 2010. On the origin of ergativity in Tima. In: Frank Floricic (ed.), Essais de typologie et de linguistique générale, Mélanges offerts à Dennis Creissels, pp. 233–239. Paris: Presses Universitaires de l’École Normale Superieure. ———. 2011. Historical Linguistics and the Comparative Study of African Languages. Amsterdam and Philadelphia: John Benjamins. ———. 2013. Where have all the noun classes gone in Tima? In: Carole de Feral, Maarten Kossmann and Mauro Tosco (eds.), In and out of Africa Languages in Question In Honour of Robert NicolaI, Vol 2 Language Contact and Language Change in Africa, pp. 103–125. Louvain: Peeters. ———. 2000. Number marking and noun categorization in Nilo-Saharan languages. Anthropological Linguistics 42:214–261. ———. 2001. Language diffusion and genetic relationship: An African perspective. In: R.M.W. Dixon and Alexandra Aikhenvald (eds.), Language Diffusion and Genetic Linguistics, pp. 358–392. Oxford: Oxford University Press. ———. 2014a. Marked nominative systems in Eastern Sudanic and their historical origin. Afrikanistik Online Vol. 11. (urn:nbn:de:0009-10-38596). ———. 2014b. The grammar of knowledge in Tima. In Alexandra Y. Aikhenvald and R.M.W. Dixon (ed.), The Grammar of Knowledge, pp. 245–259. Leiden: Brill. ———. To appear a. Language diversification and contact in Africa. In: Mily Crevelt, Jean-Marie Hombert and Pieter Muysken (eds.), Language Dispersal, Diversification and Contact. Oxford: Oxford University Press. ———. To appear b. Katloid and Rashad: A historical comparison. Dimmendaal, Gerrit J. and Anne Storch. To appear. Niger-Congo: A state of the art. In: Sarah Thomason (ed.), Oxford Handbooks Online. New York: Oxford University Press. Dimmendaal, Gerrit J. and Gertrud Schneider-Blum. To appear. A two-stage model of word formation in Tima. Paper presented at the 2nd Nuba Mountain Languages Conference, Paris. 202 References

Dimmendaal, Gerrit J. and F.K. Erhard Voeltz. 2007. Endangered languages of Africa. In: Christopher Moseley (ed.), Encyclopaedia of the World’s Endangered Languages, pp. 579–634. London: Routledge. Dingemanse, Mark. 2011. The Meaning and Use of Ideophones in Siwu. PhD diss., Radboud University, Nijmegen. Dixon, R.M.W. 1980. The Languages of Australia. Cambridge: Cambridge University Press. ———. 1994. Ergativity. Cambridge: Cambridge University Press. ———. 1997. The Rise and Fall of Languages. Cambridge: Cambridge University Press. Doke, C.M. and B.W. Vilakazi. 1953. Zulu-English Dictionary. Johannesburg: Wit watersrand University Press. Douglas, Mary. 1970. Natural Symbols: Explorations in Cosmology. London and New York: Routledge. Dournes, Jacques. 1978. Les races de couleurs: une optique jörai (Viet-Nam). In: Serge Tornay (ed.), Voir et nommer les couleurs, pp. 369–399. Nanterre: Laboratoire d’Ethnologie et de Sociologie Comparative. Ehret, Cristopher. 2001. A Historical-Comparative Reconstruction of Nilo-Saharan. Cologne: Rüdiger Köppe. Enfield, Nick J. (ed.). 2002. Ethnosyntax: Explorations in Grammar and Culture. Oxford: Oxford University Press. Evans, Nicholas and Stephen C. Levinson. 2009. The myth of language universals: Language diversity and its importance for cognitive science. Behavioral and Brain Sciences 32: 429–492. Fillmore, Charles J. 1975. Topics in lexical semantics. In: Roger Cole (ed.), Current Issues in Linguistics, pp. 76–138. Bloomington: Indiana University Press. Fleming, C. 1969. The classification of West Cushitic within Hamito Semitic. In: Daniel McCall, Norman Bennett, and Jeffrey Butler (eds.), Eastern African History, pp. 3–27. Boston: Boston University Studies in African History 3. Foley, William A. 1997. Anthropological Linguistics: An Introduction. Oxford: Wiley- Blackwell. Geiger, Lazarus. 1871. Zur Entwicklungsgeschichte der Menschheit. Stuttgart: Cotta. ———. 1872. Ursprung und Entwicklung der menschlichen Sprache und Vernunft. 2 Volumes. Stuttgart: Cotta. Gibbon, Edward. 1776–1789. The History of the Decline and Fall of the Roman Empire. 6 Volumes. [Edited by David Womersley in 1994 and published by Penguin, New York.] Givón, Talmy. 1979. On Understanding Grammar. New York: Academic Press. Goddard, C. 1999. Review article: Color categories in thought and language. Linguistic Typology 3: 259–269. Gould, Stephen J. 1980. The Panda’s Thumb. New York: W.W. Norton. ———. 1983. Hen’s Teeth and Horse’s Toes. New York: W.W. Norton. ———. 1985. The Flamingo’s Smile. New York: W.W. Norton. References 203

Greenberg, Joseph H. 1955. Studies in African Linguistic Classification. New Haven: Compass Publishing. ———. 1963. The Languages of Africa. International Journal of American Linguistics Vol. 29.1, Part ii. ———. 1966. Some universals of grammar with particular reference to the order of meaningful units. In: Joseph H. Greenberg (ed.), Universals of Language, pp. 73–113. Cambridge: The mit Press. ———. (ed.). 1978. Universals of Human Language Vol. 3: Word Structure. Stanford: Stanford University Press. ———. 1978. Generalizations about numeral systems. In: Joseph H. Greenberg (ed.), Word Structure: Universals of Human Language, pp. 249–95. Stanford: Sanford University Press. ———. 1980. Universals of kinship terminology: Their nature and the problem of their explanation. In: Jacques Maquet (ed.), On Linguistic Anthropology: Essays in Honor of Harry Hoijer, pp. 9–32. Malibu: Undena Publications. ———. 1983. Some areal characteristics of African languages. In: Ivan R. Dihoff (ed.), Current Approaches to African Linguistics 1, pp. 3–22. Dordrecht: Foris Publications. Greenberg, Joseph H. and James J. Jenkins. 1964. Studies in the psychological correlates of the sound system of American English. Word 20: 157–77, 22: 207–42. Grice, Paul (1975). Logic and conversation. In: P. Cole and J. Morgan (eds.), Syntax and Semantics Vol 3: Speech acts, pp. 41–58. New York: Academic Press. Güldemann, Tom. 1998. The Kalahari basin as an object of areal typology—a first approach. In: Mathias Schladt (ed.), Language, Identity, and Conceptualization among the Khoisan, pp. 137–69. Cologne: Rüdiger Köppe. ———. 2006. Changing profile when encroaching on hunter-gatherer territory? Towards a history of the Khoe-Kwadi family in southern Africa. Paper presented at the conference on Historical linguistics and hunter-gatherer populations in global perspective at the Max Planck Institute for Evolutionary Anthropology, Leipzig, August 10–12. Güldemann, Tom and Edward D. Elderkin. 2010. On external genealogical relationships of the Khoe family. In: Matthias Brenzinger und Christa König (eds.), Khoisan Languages and Linguistics: The Riezlern Symposium 2003, pp. 15–52. Cologne: Rüdiger Köppe. Guthrie, Malcolm. 1970. Comparative Bantu. Farnborough: Gregg International Publishers. Hagège, Claude and André Haudricourt. 1978. La phonologie panchronique. Comment les sons changent dans les langues. Paris: Presses Universitaires de France. Hamp, Eric P. 1980. Notes on Proto-Polynesian colours. Anthropological Linguistics 22: 390–91. 204 References

Hardin, Clyde L. 1993. Van Brakel and the not-so-naked emperor. British Journal for the Philosophy of Science 44: 137–50. Haviland, John B. 1992. Anchoring, iconicity, and orientation in Guugu Yimidhirr pointing gestures. Cognitive Anthropology Research Group Working Paper 8, Max Planck Institute for Psycholinguistics, Nijmegen, the Netherlands. Hayward, Richard J. 1991. À propos patterns of lexicalization in the Ethiopian language area. In: Ulrike Claudi and Daniela Mendel (eds.), Ägypten im afro-orientalischen Kontext, pp. 139–156. [Special Issue of Afrikanistische Arbeitspapiere]. ———. 2000. Afroasiatic. In: Bernd Heine and Derek Nurse (eds.), African Languages, An Introduction, pp. 74–98. Cambridge: Cambridge University Press. Heckenberger, Michael J. 2005 The Ecology of Power: Culture, Place and Parenthood in the Southern Amazon, a.d. 1000–2000. New York and London: Routledge. Heine, Bernd. 1976a. A Typology of African Languages Based on the Order of Meaningful Units. Berlin: Dietrich Reimer. ———. 1976b. The Kuliak Languages of Northeastern Uganda. Nairobi: East African Publishing House. Heine, Bernd and Christa König. 1987. On the taxonomic status of folk botanic categories among the Samburu. Afrikanische Arbeitspapiere 10: 31–52. Heine, Bernd and Derek Nurse (eds.). 2008. A Linguistic Geography of Africa. Cam bridge: Cambridge University Press. Heine, Bernd and Rainer Vossen. 1982. On the origin of gender in Eastern Nilotic. In: Rainer Vossen and Marianne Bechhaus-Gerst (eds.), Nilotic Studies Vol. 2, pp. 247– 268. Berlin: Dietrich Reimer. Hellwig, Birgit. To appear. A Grammar of Katla. Hickerson, Nancy. 1971. Review of Berlin and Kay (1969). International Journal of American Linguistics 37: 257–70. Hieda, Osamu. 1991. Koegu vocabulary with a reference to Kara. African Study Monographs, Supplement 14: 1–70. Hilders, J.H. and J.C.D. Lawrence. 1958. An English-Ateso and Ateso-English Vocabulary. Nairobi, Kampala and Dar-es-Salaam: The Eagle Press and East African Literature Bureau. Hill, Jane. 2001. Language on the land: towards an anthropological dialectology. In: John E. Terrell (ed.), Archaeology, Language, and History: Essays on Culture and Ethnicity, pp. 257–82. Westport: Bergin and Garvey, Greenwood Publishing. Hollis, A.C. 1909. The Nandi, Their Language and Folk-Lore. Westport, conn: Negro Universities Press. Hulstaert, G. 1957. Dictionnaire lomongo-français. Tervuren: Musee Royal du Congo Belge. Hunn, E. 1982. The utilitarian factor in folk biological classification. American Anthropologist 84: 830–847. References 205

Huttar, George L, James Essegbey and Felix K. Ameka. 2007. Gbe and other West African sources of Suriname creole semantic structures: Implications for creole genesis. Journal of Pidgin and Creole Languages 22.1: 57–72. Ibrahim Ghulfan, Gumma. 2013. Converbs in Taglennaa. In: Schadeberg and Blench (eds.), pp. 371–379. Ibrahim, Ali and Angelika Jakobi. 2013. The relative clause in Taglennaa (Kordofan Nubian). In: Schadeberg and Blench (eds.), pp. 355–370. Ibrahim, Gumma and Piet Huttenga. 2007. The phoneme system of Tagle, a Kordofanian Nubian language. In: Doris Payne and Mechthild Reh (eds.), Advances in Nilo- Saharan Linguistics, pp. 99–113. Cologne: Rüdiger Köppe. Irvine, J.T. and S. Gal. 2000. Language ideology and linguistic differentiation. In: P.V. Kroskrity (ed.), Regimes of Language: Ideologies, Polities, and Identities, pp. 35–84. Santa Fe, nm: School of American Research Press. Ismail, Muhammad Abbaker. 2007. Noun classes in Daju Lagawa and the role of stress. In: Payne and Reh (eds.), pp. 115–122. Jakobi, Angelika. To appear. Differential Object Marking in Uncunwee (Kordofan Nubian). In: Angelika Mietzner and Anne Storch (eds.), Proceedings of the 11th Nilo- Saharan Linguistics Colloquium. Jakobi, Angelika and Robert Williams. 2007. Number marking in Ghulfan. Paper presented at the 10th Nilo-Saharan Linguistics Colloquium, Paris. Jakobson, Roman. 1960. Closing statements: Linguistics and poetics. In: Thomas A. Sebeok (ed.), Style in Language, pp. 350–377. Cambridge, ma: mit Press. ———. 1962. Selected Writings i: Phonological Studies. The Hague: Mouton. James, William. 1884. What is an emotion? Mind 9: 188–205. Karp, I. 1978. Fields of Change among the Iteso of Kenya. London: Routledge and Kegan Paul. Kay, Paul and Chad K. McDaniel. 1975. Color categories as fuzzy sets. Language Behavior Research Laboratory, University of California at Berkeley, Working Paper 44. ———. 1978. The linguistic significance of the meanings of basic color terms. Language 54: 610–46. Kay, Paul and Willett Kempton. 1984. What is the Sapir-Whorf hypothesis? American Anthropologist 86: 65–79. Kay, Paul, Brent Berlin and William Merrifield. 1991. Biocultural implications of color naming. Journal of Linguistic Anthropology 1: 12–25. Keesing, Roger M. 1979. Linguistic knowledge and cultural knowledge: some doubts and speculations. American Anthropologist 81: 14–36. ———. 1992. Anthropology and linguistics. In: Martin Pütz (ed.), Thirty Years of Linguistics Evolution, pp. 593–623. Amsterdam and Philadelphia: John Benjamins. Keller, Rudi. 1994. Sprachwandel: Von der unsichtbaren Hand in der Sprache. 2nd edition. Tübingen and Basel: utb-Verlag. 206 References

Kennedy, G.A. 1998. Comparative Rhetoric: An Historical and Cross-Cultural Introduction. New York and Oxford: Oxford University Press. Kießling, Roland and Maarten Mous. 2004. Urban youth languages in Africa. Anthropological Linguistics 46: 303–341. Kilian-Hatz, C. and M. Schladt. 1997. From body to soul: On the structure of body parts idioms. Afrikanistische Arbeitspapiere 49: 61–79. Kinkade, M. Dale. 1988. Proto-Salishancolors. In: William Shipley (ed.), In Honor of Mary Haas: From the Haas Festival Conference on Native American Linguistics, pp. 443–66. Berlin: Mouton de Gruyter. Kipling, Rudyard. 1902. Just So Stories for Little Children. [Republished 1993 by Wordsworth Editions Limited, Hertfordshire.] Kirchhoff, Alfred. 1879. Über Farbensinn und Farbenbezeichnung der Nuberi. Zeitschschrift für Ethnologie 11: 397–402. Kitching, A.L. 1912. On the Backwaters of the Nile. London: Fisher Unwin. Klein, W. 1982. Local deixis in route directions. In: R.J. Jarvella and W. Klein (eds.), Speech, Place and Action, pp. 161–75. Chichester: John Wiley and Sons. Kleinewillinghöfer, Ulrich. 1996. Relationship between Adamawa and Gur: The case of Waja. Gur Papers / Cahiers Voltaiques 1: 25–46. Kövecses, Zoltán. 2000. Metaphor and Emotion: Language, Culture, and Body in Human Feeling. Cambridge and Paris: Cambridge University Press and Editions de la Maison des Sciences de l’Homme. Kristol, Andres M. 1980. Color systems in southern Italy. Language 56: 137–145. Kuper, Adam. 1994. The Chosen Primate: Human Nature and Cultural Diversity. Cambridge, mass.: Harvard University Press. Kutsch Lojenga, Constance L. 1994. Ngiti: A Central-Sudanic Language of Zaire. Cologne: Rüdiger Köppe. Labov, William. 2001. Principles of Linguistic Change: Social Factors. Oxford: Blackwell. Lass, Roger. 1990. How to do things with junk: Exaptation in language evolution. Journal of Linguistics 26: 79–102. ———. 1997. Historical Linguistics and Language Change. Cambridge: Cambridge University Press. Last, Marco. 1995. A Sketch of Chai. M.A. thesis, Leiden University. Lawrance, J.C.D. 1957. The Iteso: Fifty Years of Change in a Nilo-Hamitic Tribe of Uganda. London: Oxford University Press. Lenneberg, Eric H. and J. Roberts. 1956. The Language of Experience; A Study in Methodology. Baltimore: Waverly Press. Levinson, Stephen C. 1991. Relativity in spatial conception and description. Cognitive Anthropology Research Group Working Paper i., Max Planck Institute for Psycho linguistics, Nijmegen, the Netherlands. References 207

———. 2003. Space in Language and Cognition: Explorations in Cognitive Diversity. Cambridge: Cambridge University Press. ———. 2006. On the human “interaction engine”. In: N.J. Enfield and Stephen C. Levinson (eds.), Roots of Human Sociality: Culture, Cognition and Interaction, pp. 39–69. Oxford and New York: Berg. Lévi-Strauss, C. 1970. The sex of the heavenly bodies. In: Michael Lane (ed.), Introduction to Structuralism, pp. 330–39. New York: Basic Books. Lieberman, P. 1984. The Biology and Evolution of Language. Cambridge: Harvard University Press. Lienhardt, G. 1970. Divinity and Experience: The Religion of the Dinka. Oxford: Clarendon Press. Longacre, Robert 1990. Storyline Concerns and Word Order Typology in East and West Africa. Studies in African Linguistics Supplement 10. Lorenz, Steffen. 2013. Number Marking as an Areal Feature of Languages in Northeastern Africa. ma thesis, University of Cologne. Lucy, John A. 1996. The linguistics of ‘colour. In: Clyde L. Hardin and L. Maffi (eds.), Colour Categories in Thought and Language, pp. 320–46. Cambridge: Cambridge University Press. ———. 1985. Whorf’s view of the linguistic mediation of thought. In: Elizabeth Mertz and Richard J. Parmentier (eds.), Semiotic Mediation: Sociocultural and Psychological Perspectives, pp. 73–97. Orlando: Academic Press. ———. 1992a. Language Diversity and Thought: A Reformulation of the Linguistic Relativity Hypothesis. Cambridge: Cambridge University Press. ———. 1992b. Grammatical Categories and Cognition: A Case Study of the Linguistic Relativity Hypothesis. Cambridge: Cambridge University Press. Lucy, John A. and Richard A. Shweder. 1979. Whorf and his critics: Linguistic and nonlinguistic influences on color memory. American Anthropologist 81: 581–615. Luria, Alexander R. 1981. Language and Cognition. New York: John Wiley and Sons. [Translated from Russian] Ma Newman, R. 1990. An English-Hausa Dictionary. New Haven and London: Yale University Press. MacDiarmid, P.A. and D.N. MacDiarmid. 1931. The languages of the Nuba Mountains. Sudan Notes and Records 14: 149–162. MacLaury, Robert E. 1992. From brightness to hue: An explanatory model of color- category evolution. Current Anthropology 33: 137–163. Maddieson, Ian. 2003. The sounds of the Bantu languages. In: Derek Nurse and Gerard Philippson (eds.), The Bantu Languages, pp. 15–41. London: Routledge. Marr, David. 1982. Vision. A Computational Investigation into the Human Representation of and Processing of Visual Information. San Francisco: Freeman and Company. 208 References

Masica, C.P. 1976. Defining a Linguistic Area: South Asia. Chicago and London: Univer sity of Chicago Press. Mathiot, Madeleine (eds.). 1979. Ethnolinguistics: Boas, Sapir and Whorf Revisited. The Hague: Mouton. ———. 1979. Folk-definitions as a tool for the analysis of lexical meaning. In: Madeleine Mathiot (ed.) Ethnolinguistics: Boas, Sapir and Whorf Revisited, pp. 121–260. The Hague: Mouton. Matisoff, J.A. 1986. Hearts and minds in South- East Asian languages and English: An essay in the comparative lexical semantics of psycho-collocations. Cahiers de Linguistique Asie Orientale 15.1: 5–57. McIntyre, J.A. 1980. The language of Hausa greetings: The social construction of hierarchy. Afrika und Übersee 63: 39–67. McKee, Rob. 2010. Mangbetu Tales of Leopard and Azapane: Trickster as Resistance Hero. http://www.gial.edu/gialens/issues.htm. Meeussen, A.E. 1967. Bantu grammatical reconstructions. Africana Linguistica 3: 79–121. Meinhof, Carl. 1912. Die Sprachen der Hamiten. Hamburg: Friederichsen. Merkies, H.C.G. 1980. Ganda Classification: An Ethno-Semantic Survey. PhD diss., University of Nijmegen, The Netherlands. Mol, Frans. 1978. Maa. A Dictionary of the Maasai Language and Folklore English- Maasai. Nairobi: Marketing and Publishing. ———. 1996. Maasai Language and Culture Dictionary. Limuru: Kolbe Press. Moore, Henrietta L. 1986. Space, Text and Gender: An Anthropological Study of the Marakwet of Kenya. Cambridge: Cambridge University Press. Mous, Maarten. 2007. John M. Stewart, 1926–2006. Journal of African Languages and Linguistics 28: 71–74. Mowrer, D.E. and S. Burger. 1991. A comparative analysis of the phonological acquisition of consonants in the speech of 21/2-6-year-old Xhosa- and English-speaking children. Clinical Linguistics and Phonetics 3: 139–164. Mugaddam, Abdelrahim Hamid. 2006. Language Status and Use in Dilling city, the Nuba Mountains. Journal of Multilingual and Multicultural Development 27: 290–304. Mugaddam, Abdelrahim Hamid and Ashraf Abdelhay. 2013. The socio-linguistic profile of Tima Language. In: Thilo C. Schadeberg and Roger Blench (eds.), Nuba Mountains Languages Studies, pp. 297–324. Cologne: Rüdiger Köppe. Müller, Friedrich. 1876–1888. Grundriss der Sprachwissenschaft. Hildesheim, Zürich and New York: Olms. Muzzolini, A. 1993. The emergence of a food-producing economy in the Sahara. In: Sinclair T. Shaw, B.P. Andah, and A. Akpoko (eds.), The Archaeology of Africa: Food, Metals and Towns, pp. 227–39. London: Routledge. Nadel, S.F. 1947. The Nuba. London: Oxford University Press. References 209

Nassenstein, Nico and Andrea Wolvers (eds.). To appear. Youth Languages in Africa and Beyond. Berlin & New York: Mouton de Gruyter. Nettle, Daniel. 1999. Linguistic Diversity. Oxford: Oxford University Press. Newman, Paul. 1990. On Being Right: Greenberg’s African Linguistic Classification and the Methodological Principles Which Underlie It. Bloomington: Institute for the Study of Nigerian Languages and Cultures, African Studies Program, Indiana University. ———. 1991. An interview with Joseph H. Greenberg. Current Anthropology 32: 453–67. Nichols, Johanna. 1986. Head-marking and dependent-marking grammar. Language 62: 56–119. ———. 1992. Linguistic Diversity in Space and Time. Chicago: Chicago University Press. ———. 1997. Modeling ancient population structures and movement in linguistics. Annual Review of Anthropology 26: 359–84. ———. To appear. Types of spread zones: Open and closed, horizontal and vertical. In: Rik De Busser and Randy J. LaPolla (eds.), The Making of Language: The Relationship between the Structures of Languages and their Social, Cultural and Natural Environments. Amsterdam and Philadelphia: John Benjamins. Nicolaï, Robert. 2003. La force des choses ou l’épreuve nilo-saharienne: Questions sur les reconstructions archéologiques et l’évolution des langues. Cologne: Rüdiger Köppe. Nida, E.A. 1955. Problems in translating the Scriptures into Shilluk, Anuak and Nuer. The Bible Translator 6: 55–63. Noonan, Michael. 1992. A Grammar of Lango. Berlin: Mouton de Gruyter. Norton, Russell. 2013. The Ama dual suffix: An internal reconstruction. Paper presented at the Nilo-Saharan Linguistics Colloquium, University of Cologne. Novelli, Bruno. 1999. Karimojong Traditional Religion: A Contribution. Kampala: Comboni Missionaries. Nwoye, G. Onuigbo. 1993. An ethnographic analysis of Igbo greetings. African Languages and Cultures 6: 37–48. Ohala, John J. 1989. Sound change is drawn from a pool of synchronic variation. In: E.L. Breivik and E.H. Jahr (eds.), Language Change: Contributions to the Study of its Causes, pp. 173–98. Berlin: Mouton de Gruyter. ———. 1993. The phonetics of sound change. In: Charles Jones (ed.), Historical Linguistics: Problems and Perspectives, pp. 237–78. London: Longman. ———. 1994. The frequency code underlies the sound-symbolic use of voice pitch. In: Leanne Hinton and John J. Ohala (eds.), Sound Symbolism, pp. 325–47. Cambridge: Cambridge University Press. Pachur, Hans-Joachim and Stefan Kröpelin. 1987. Wadi Howar: Paleoclimatic evidence from an extinct river system in the southeastern Sahara. Science 237: 298–300. Palayer, Pierre. 2011. Dadjo D’Eref. Louvain: Peeters. 210 References

Paye, Won-Ldy and Margaret H. Lippert. 1998. Why Leopard Has Spots: Dan Stories from Liberia. Golden, co: Fulcrum Books. Pike, Kenneth 1954. Language in Relation to a Unified Theory of the Structure of Human Behavior. Dallas, tx: Summer Institute of Linguistics. Pinker, Steven. 1994. The Language Instinct: The New Science of Language and Mind. London: Penguin Books. Pollnac, R.B. 1975. Intracultural variability in the structure of the subjective color lexicon of Buganda. American Ethnologist 2: 89–109. Quine, W.V. 1960. Word and Object. Cambridge: The mit Press. Quint, Nicolas. 2006. The Phonologie of Koalib, a Kordofanian Language of the Nuba Mountains (Sudan). Cologne: Rüdiger Köppe. Randall, Robert A. and Eugene S. Hunn. 1984. Do life-forms evolve, or do uses for life? Some doubts about Brown’s universals hypotheses. American Ethnologist 11: 329–349. Reh, M. 1998. The language of emotion: An analysis of Dholuo on the basis of Grace Ogot’s novel ‘Miaha’. In: A. Athanasiadou and E. Tabakowska (eds.), Speaking of Emotions: Conceptualisation and Expression, pp. 375–408. Berlin: Mouton de Gruyter. Reh, M. with the assistance of A.S. Akwey and C.U. Uriat. 1999. Anywa-English and English-Anywa Dictionary. Cologne: Rüdiger Köppe. Rice, Mabel. 1980. Cognition to Language: Categories, Word Meanings and Training. Baltimore: University Park Press. Rice, Sally. (2012). Our language is very literal: Figurative expressions in Dene Sᶙɫiné [Athabaskan]. In: Anna Idström & Elisabeth Piirainen in cooperation with Tiber F.M. Falzett (eds.), Endangered Metaphors, pp. 21–76. Amsterdam: John Benjamins. Riefenstahl, Lenie. 1973. Die Nuba—Menschen wie von einem anderen Stern. Munich: List. ———. 1975. Die Nuba von Kau. Munich: List. Rilly, Claude. 2004. The linguistic position of Meroitic. ARKAMANI Sudan Electronic Journal of Archaeology and Anthropology. http://www.ddl.ish-lyon.cnrs.fr/projets/ clhass/PageWeb/ressources/Isolats/Meroitic%20Rilly%202004.pdf. Rosch, Eleanor. 1972a. Probabilities, sampling, and ethnographic method: The case of Dani colour n ames. Man 7: 448–66. ———. 1972b. Universals in color naming and memory. Journal of Experimental Psychology 93: 10–20. Rose, Sharon. 2013. The morphological structure of the Moro verb. In: Schadeberg and Blench (eds.), pp. 25–55. Ross, Malcolm D. 1996. Contact-induced change and the comparative method: Cases from Papua New Guinea. In: Mark Durie and Malcolm D. Ross (eds.), The References 211

Comparative Method Reviewed: Regularity and Irregularity in Language Change, pp. 180–217. New York: Oxford University Press. ———. 2001. Contact-induced change in Oceanic languages in North-West Melanesia. In: Alexandra Y. Aikhenvald and R.M.W. Dixon (eds.), pp. 134–166. Rottland, Franz. 1982. Die südnilotischen Sprachen: Beschreibung, Vergleichung und Rekonstruktion. Berlin: Dietrich Reimer. Rüppell, Eduard. 1829. Reisen in Nubien, Kordofan und dem peträischen Arabien vorzüglich in geographisch-statistischer Hinsicht. Frankfurt am Main: Friedrich Wilmans. Sagar, J.W. 1922. Notes on the history, religion, and customs of the Nuba. Sudan Notes and Records 3: 137–156. Saunders, B.A.C. 1992. The Invention of Basic Color Terms. PhD Diss., University of Utrecht, the Netherlands. Savage, G.A.R. 1954. A Short Acoli-English and English-Acoli Vocabulary. Nairobi: Boyd and Co. Schadeberg, Thilo C. 1981a. The classification of the Kadugli language group. In: Thilo C. Schadeberg and M. Lionel Bender (eds.), Nilo-Saharan, pp. 291–305. Dordrecht: Foris Publications. ———. 1981b. A Survey of Kordofanian Vol. 1: The Heiban Group. Hamburg: Helmut Buske. ———. 1981c. A Survey of Kordofanian Vol. 2: The Talodi Group. Hamburg: Helmut Buske. ———. 1994. Comparative Kadu wordlists. Afrikanistische Arbeitspapiere 40: 11–48. ———. 2013. Rashad survey data. In: Schadeberg and Blench (eds.), pp. 325–345. Schadeberg, Thilo C., and Roger M. Blench. 2013a. The Nuba Mountain languages: An introduction. In Schadeberg and Blench (eds.), pp. 1–24. ———. 2013b. Nuba Mountain Language Studies. Cologne: Rüdiger Köppe. Schaefer, James M. 1969. A comparison of three measures of social complexity. American Anthropologist 71: 706–708. Schegloff, Emanuel A. 2007. Sequence Organization in Interaction: A Primer in Conversation Analysis Vol. 1. Cambridge: Cambridge University Press. Schmidt, Wilhelm. 1926. Die Sprachfamilien und Sprachkreise der Erde. Heidelberg: Winter. Schneider-Blum, Gertrud. 2012. Don’t waste words: Perspectives on the Tima lexicon. In: Matthias Brenzinger and Anne-Marie Fehn (eds.), Proceedings of the 6th wocal Conference, pp. 515–523. Cologne: Rüdiger Köppe. ———. 2013a. A Tima-English Dictionary. Cologne: Rüdiger Köppe. ———. 2013b. Personal pronouns in Tima. In: Schadeberg and Blench (eds.), pp. 281–296. ———. To appear. Classification of plants among the Ùmúrik (Tima). Ms, University of Cologne. 212 References

Schneider-Blum, Gertrud, and Gerrit J. Dimmendaal. 2013. Excite your senses: Glances into the field of perception and cognition in Tima. In: Alexandra Y. Aikhenvald and Anne Storch (eds.), Perception and Cognition, pp. 217–249. Leiden: Brill. ———. To appear. Lexicography of endangered languages in Africa. In: Patrick Hanks and Gilles-Maurice de Schryer (eds.), International Handbook of Modern Lexis and Lexicography. New York: Springer. Segall, M.H., D.T. Campbell, and M.J. Herskovits. 1966. The Influence of Culture on Visual Perception. New York: Bobbs-Merrill. Silverstein, Michael. 1976. Shifters, linguistic categories, and cultural description. In: Keith H. Basso and Henry A. Selby (eds.), Meaning in Anthropology, pp. 11–55. Albuquerque: University of New Mexico Press. Simeoni, Antonio. 1978. Päri: A Luo Language of Southern Sudan. Small Grammar and Vocabulary. Bologna: Editrice Missionaria Italiana. Snyman, J.W. 1975. Zu Hoasi: Fonologie en Woordeboek. Cape Town: A.A. Balkema. Spagnolo, Lorenzo M. 1960. Bari English Italian Dictionary. Verona: Missioni Africane. Stafford, R.L. 1967. An Elementary Luo Grammar with Vocabularies. Nairobi: Oxford University Press. Stevenson, Roland C. 1956–1957. A survey of the phonetics and grammatical structure of the Nuba Mountain languages, with particular reference to Otoro Katcha and Nyimang. Afrika und Übersee 40: 27–65, 41: 117–152, 171–196. ———. 1962. Linguistic research in the Nuba Mountains. Sudan Notes and Records 43: 118–130. ———. 1964. Linguistic research in the Nuba Mountains. Sudan Notes and Records 45: 79–102. ———. 1984. The Nuba of Southern Kordofan: An Ethnographic Survey. Khartoum: Khartoum University Press. ———. 2009. Tiro and Otoro: Two Kordofanian Grammars [edited by Thilo C. Schadeberg]. Cologne: Rüdiger Köppe. Stewart, John M. 2002. The potential of Proto-Potou-Akanic-Bantu as a pilot Proto- Niger-Congo, and the reconstructions updated. Journal of African Languages and Linguistics 23: 197–22. ———. 2007. Consonant mutation in Proto-Potou-Akanic-Bantu and the Fula-type languages of Senegal and Guinea. In: William A.A. Wilson and Anne Storch (eds.), Guinea languages of the Atlantic group: Description and internal classification, pp. 171–95. Frankfurt: Peter Lang. Stirtz, Timothy M. 2012. A Grammar of Gaahmg, A Nilo-Saharan Language of Sudan. Utrecht: Netherlands Gradual School of Linguistics. Storch, Anne. 2006. How long do linguistic areas last? Western Nilotic grammars in contact. In: Alexandra Y. Aikhenvald and R.M.W. Dixon (eds.), Grammars in Contact, pp. 67–93. Oxford: Oxford University Press. References 213

———. 2011. Secret Manipulations: Language and Context in Africa. New York: Oxford University Press. ———. 2014. A Grammar of Luwo: An Anthropological Approach. Amsterdam and Philadelphia: John Benjamins. Strecker, Ivo. 1988. The Social Practice of Symbolization: An Anthropological Analysis. London: The Athlone Press. Talmy, L. 1988. The relation of grammar to cognition. In: B. Rudzka-Ostyn (ed.), Topics in Cognitive Linguistics, pp. 165–205. Amsterdam: John Benjamins. Tanaka, Jiro. 1980. The San: Hunter-gatherers of the Kalahari. A Study in Ecological Anthropology. Tokyo: University of Tokyo Press. Taylor, John R. and Thandi G. Mbnese. 1998. Red dogs and rotten mealies: How Zulus talk about anger. In: Angliki Athanasiadou and Elżbieta Tabakowska (eds.), Speaking of Emotions: Conceptualisation and Expression, pp. 191–226. Berlin: Mouton de Gruyter. Thelwall, Robin E.W. 1981. The Daju Language Group: Systematic Phonetics, Lexicostatistics and Lexical Reconstruction. PhD Diss., School of Humanities of the New University of Ulster. Thelwall, Robin and Thilo C. Schadeberg. 1983. The linguistic settlement of the Nuba Mountains. Sprache und Geschichte in Afrika 5: 219–232. Thomas, Jacqueline M.C. and Serge Bahuchet (eds.). 1981. Encyclopédie des pygmées aka: Techniques, langage et societé des chasseurs-cueilleurs de la forêt centrafricaine (Sud-Centrafrique et Nord-Congo). Vol. 2: Dictionnaire aka-français (langue bantou (C.10) Paris: selaf. Thomason, Sarah G. 2007. Language contact and deliberate change. Journal of Language Contact 1: 41–62. Thomason, Sarah G. and Thomas Kaufman. 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press. Thurston, William R. 1989. How exoteric languages build a lexicon: Esoterogeny in West New Britain. In: Ray Harlow and Robin Cooper (eds.), vical 1: Papers from the Fifth International Conference on Austronesian Linguistics, pp. 555–79. Auckland: Linguistic Society of New Zealand. Tornay, Serge. 1973. Langage et perception: La dénomination des couleurs chez les Nyangatom du sud-ouest éthiopien. L’Homme 13.4: 66–94. ———. 1978. Perception des couleurs et pensee symbolique. In: Serge Tomay (ed.), Voir et nommer les couleurs, pp. 609–37. Nanterre: Laboratoire d’Ethnologie et de Sociologie Comparative. Toweett, T. 1979. A Study of Kalenjin Linguistics. Nairobi: Kenya Literature Bureau. Traill, A. and H. Nakagawa. 2000. A historical !Xóo-ǀGui contact zone: Linguistic and other relations. In: Herman M. Batibo and Joseph Tsonope (eds.), The State of Khoesan languages in Botswana, pp. 1–17. Gaborone: Tasalls Publishing. 214 References

Trubetzkoy, Nikolai Sergeevič. 1923. Vavilonskaja bašnja i smešenie jazykov [The tower of Babel and the Confusion of languages]. Evrazijskij Vrememennik 3: 107–124. Tucker, A.N. and M.A. Bryan. 1956. The Non-Bantu Languages of Northeastern Africa. London: Oxford University Press for the International African Institute. ———. 1966. Linguistic Analyses: The Non-Bantu Languages of Northeastern Africa. London: Oxford University Press for the International African Institute. Turner, Victor V. 1966. Color classification in Ndembu ritual. In: Michael Banton (eds.), Anthropological Approaches to the Study of Religion, 47–84. London: Tavistock. Turton, David. 1980. There’s no such beast: cattle and colour naming among the Mursi. Man 152: 320–38. van Beek, Wouter E.A. 1977. Color terms in Kapsiki. In: Paul Newman and Roxana Ma Newman (eds.), Papers in Chadic Linguistics, pp. 13–20. Leiden: Afrika- Studiecentrum. van Brakel, J. 1993. The plasticity of categories: The case of colour. British Journal for the Philosophy of Science 44: 103–35. Vanderelst, John. 2013. Personal pronouns in Dagik. In: Schadeberg and Blench (eds.), pp. 157–175. ———. To appear. A Grammar of Dagik. PhD diss., University of Cologne. van der Horst, J.W. 1981. Kleine Middelnederlandse Syntaxis [Small Middle-Dutch Syntax]. Amsterdam: Huis aan de Drie Grachten. van der Jagt, Krijn. 1989. Symbolic Structures in Turkana Religion. Assen and Maastricht: Van Gorcum. van Ginneken, J. 1938. Les clics, les consonnes et les voyelles dans l’histoire de l’humanité. Amsterdam: k.n.a.w. ———. 1939. La reconstruction typologique des langues archaïques d l’humanité. Amsterdam: k.n.a.w. Vansina, Jan. 1995. New linguistic evidence in the Bantu Expansion. Journal of African History 36: 173–195. von Heyking, Beatrix. 2014. A Grammar of Belanda Boor. Cologne: Rüdiger Köppe. Vossen, Rainer. 1982. The Eastern Nilotes: Linguistic and Historical Reconstructions. Berlin: Diedrich Reimer. ———. (ed.). 2013. The Khoesan Languages. London and New York: Routledge. Vygotsky, L.S. 1962. Thought and Language. Cambridge: The mit Press. Wassmann, Jürg. 1993. Finding the right path: The route knowledge of the Yupno of Papua New Guinea. Cognitive Anthropology Research Group Working Paper 19. Nijmegen: Max Planck Institute for Psycholinguistics. Wedekind, Klaus 1985. Thoughts when drawing a tone map of Africa. Afrikanistische Arbeitspapiere 1: 105–124. Wehr, H. 1979. A Dictionnary of Modern Written Arabic. Wiesbaden: Otto Harrassowitz. Wescott, R.W. 1970. Bini color terms. Anthropological Linguistics 12: 349–60. References 215

Whiteley, W.H. 1973. Colour-words and colour-values: The evidence from Gusii. In: R. Horton and R. Finnegan (eds.), Modes of Thought: Essays on Thinking in Western and Non-Western Societies, pp. 145–61. London: Faber and Faber. Wierzbicka, A. 1995. Adjectives vs. verbs: The iconicity of part-of-speech. In: M.E. Landsberg (eds.), Syntactic Iconicity and Linguistic Freezes: The Human Dimension, pp. 223–45. Berlin: Mouton de Gruyter. ———. 1996. Semantics, Primes and Universals. Oxford: Oxford University Press. Wierzbicka, Anna. 1999. Emotions Across Languages and Cultures: Diversity and Universals. Cambridge: Cambridge University Press. Wilkins, David P. 1996. Natural tendencies of semantic change and the search for cognates. In: Mark Durie and Malcolm Ross The Comparative Method Reviewed: Regularity and Irregularity in Language Change, edited by, 264–304 Oxford: Oxford University Press. Williamson, Kay. 1989. Niger-Congo overview. In: John Theodor Bendor (eds.), The Niger-Congo Languages: A Classification and Description of Africa’s Largest Language Family, pp. 1–45. Lanham: University Press of America. Yigezu, Moges. 2002. A Comparative Study of the Phonetics and Phonology of Surmic Languages. PhD diss., Université de Bruxelles. ———. 2006. Convergence of Baale, a Southwest Surmic language to the Southeast Surmic group: Lexical evidence. Annual Publications in African Linguistics 3: 49–66. Language Index

Abu-Jinuk (Abu Jinūk) 27, 31, 57 Dabatna 31 Acheron 27 Dadjo D’Eref 39 Acholi 21, 119, 129, 154 Daga 129 Adamawa 13, 24, 36, 69, 81 Dagik 33, 52, 53 Adamawa-Eastern 10 Dair 27, 31 Afitti 27, 29, 52, 56, 63, 72 Daju 17, 25, 27, 29, 30, 32–34, 38, 39, 43, 49, Afroasiatic 9, 10, 12, 13, 17, 18, 20, 24, 34, 41, 52, 58, 59, 63, 72, 79 53, 54, 87, 168 Daju of Lagowa 10, 27, 29–30, 32, 33, 38, 39, Ajang 31, 77 43, 59, 72, 74, 79 Aka 95 Dani 111 Akan 187 Debri 31 Akie 94 Dene Sᶙɫiné 194 Al Amira 29 Didinga 22 Alur 21 Dilling 27, 31, 60 Amharic 114 Dinik 17 Anej 41 Dompo 13, 16 Anywa 20, 152 Dutch 53, 11, 147, 149, 150, 151, 153, 178, 179 Arabic 24, 62, 75, 166, 169 Atlantic 10, 11, 12, 13 Eastern Sudanic 17–20, 22, 27, 29, 30, 36, 37, 41–45, 47, 49, 51, 53, 55–57, 60, 63, 72, Baale 22, 76, 139 79 Baca 113 El-Hugeirat 31 Baŋgi Me 13 Eliri 26, 27 Bantu 11, 12, 14, 15, 16, 20, 21, 25 (Non-Bantu), English 2, 53, 80, 103, 105, 106, 107, 111, 122, 37, 81, 86, 92, 95, 104, 113, 117, 147, 149, 152, 123, 124, 128, 134, 141, 148, 163, 187, 188, 193 155, 156, 159, 160, 162, 163, 165, 167, 169, Ewe 158 185 Bari 30, 114, 154, 155, 157, 172, 173, 176, 177 Fanda 31 Benue-Congo 10, 11, 12, 16, 20, 24, 28, 60, 66, French 80, 147, 153, 192 (Haitian French) 69, 113 Fungor 27 Berber 24 Fur 12, 17, 42, 58 Berta 55, 57 Bini 104, 107, 109 Gaahmg 29, 30, 37, 49, 55, 57 Biraile 13 German 53, 102, 147, 148, 149, 150, 151, 179, Birgid 31 192 Belanda Boor 43 Germanic 2, 11, 53, 80, 149, 179 Ghulfan 27, 31, 33, 42, 72, 74, 77 Central Khoisan 10, 13, 15 Gikuyu 114, 139 Chadic 9, 13, 24, 39, 69, 104, 113, 117, 151 Greek 2, 11 Chai 22, 76, 105, 106, 110, 111 Gumuz 12, 13, 55 Chari-Nile 12 Gur 10, 36 Chinese 3, 114 (Mandarin Chinese), 139, 187 Gurindji Kriol 24 Cross River 123 Gusii 104 Cushitic 9, 14, 32, 40 Guugu Yimidhirr 116 Language Index 217

Habila 31, 60 Koman 12, 13, 41, 55 Hadza 10, 13 Kordofanian 10, 11, 20, 26–29, 31, 44, 59, 60, Hamar 168, 169, 181 66–67, 69, 77 Hamito-Semitic 9 Krongo (Korongo) 27, 48–49, 54, 66 Hanunoo 133, 134 Kudur 31 Hausa 113, 117, 118, 151, 152 Kujuria 31 Heiban 11, 12, 20, 25–27, 29, 35–36, 45, 47, 48, Kuku 27 52, 60, 62, 66, 72, 74, 75 Kuliak 57, 91 Kumam 21 Igbo 114 Kunama 12 Indo-European 2, 58, 118, 147, 179, 192 Kwa 10, 11, 20, 66, 69, 134, 141, 158, 182, 187 Kwadi 10, 13 Jalaa 13, 16 Kwaio 100 Japanese 122, 138, 185, 187, 193 Kwakiutl 105 Jebel 17, 29, 30, 49, 57 Jiye 170, 171 Laal 13, 16, 21 Jörai 107 Lafofa 25, 27, 29, 52, 72, 74 Julud 27, 52, 53, 69, 70, 71, 74, 77, 78, 79, 186 Laggori (Liguri) 27, 30, 33, 43, 38, 39, 58, 59 Lagowa (Lagawa) 27, 29–30, 32, 33, 38, 39, Kadaru (Kadaro, Kadoro) 27, 31 43, 59, 72, 74, 79 Kadu 11, 13, 20, 25, 26, 34, 41, 45, 48, 49, 52, Lakhota 139 54, 57, 60, 61, 72, 74, 77, 82 Lango 21, 154, 158 Kadugli 11, 26, 27, 45, 59, 60, 66 Laro 27 Kajakja 27 Latin 11, 92 Kalenjin 180 Liguri 27, 30, 43, 58, 59 Kana 123 Lotuxo 176, 177 Kanga 27 Luganda 104 Kapsiki 104 Lumun 27 Karam 107 Luo 21–22, 23, 30, 153, 167, 169 Karimojong 170–173, 175–177, 180 Luwo 138, 180 Karko 27, 31 Kasha 31 Maa 176, 177 Katcha 27 Maasai 105, 117, 119, 148, 151, 155, 157, 158, 163, Katla 11, 12, 23, 25, 26–27, 39, 52–53, 59, 167, 177 65–67, 69–72, 74, 75, 77–80, 81, 186, 194 Maban 12, 17 Katloid 26, 28, 33, 44–45, 48, 49, 52–53, 60, Macro-Sudanic 12 66, 186 Majang 29 Kau 27 Mandarin Chinese 114 Kegboid 123 Mande 10, 11, 13, 16 Keiga 27, 29 Marakwet 118 Keiga Jirru 27, 44, 72, 74 Masakin (Mesakin) 27, 52, 72, 74 Kenuzi-Dongolawi 31 Mayan 90, 91, 121 Khoisan 10, 15, 20, 93, 94, 95 Mednyj Aleut 24 Kinare 94 Meroitic 17, 29, 30, 56 Kipsikiis 156 Michif 24 Koalib 27, 35–36, 52, 59, 60, 62, 72, 74 Midob 31 Koegu 115 Mimi 12 Koldegi 31 Miri 27 218 Language Index

Mixe 139 Proto-Indo-European 11, 87, 179 Mongo 159, 160, 161, 162 Proto-Mayan 90, 91, 121 Morab (Moreb) 27, 45 Proto-Nilotic 29, 87, 92, 176, 179 Moro 27, 40, 48, 52, 72, 74 Proto-Polynesian 109, 121 Mpra 13, 16 Proto-Potou-Akanic-Bantu 11 Murle 22, 42 Proto-Salish 121 Mursi 22, 76, 96, 105, 106, 108, 110, 111 Proto-Semitic 87

Nandi 105, 174 Quechua 139 Nara 17, 29, 30, 56 Ngiti 113 Rashad 11–12, 25–28, 44–45, 48, 60, 66, 67 Niger-Congo 3, 10, 10–13, 15–16, 19, 20, 24–26, Rub 57 33–34, 35, 39, 41–42, 44, 45, 47, 48, 49, 52, Russian 101, 122, 193 54, 57, 59–61, 66, 67, 72, 75, 77, 78, 137, 149, 158, 159 Saharan 12, 17 Nilo-Saharan 11–13, 16–22, 25–27, 29, 33, Salish 86, 121 34, 39–42, 45, 46, 47, 49, 53, 55–57, 59, 60, Samburu 96, 114 66, 79, 102, 139, 149, 152, 174, 197, 198, 201, Sandawe 10, 13 204–205, 209, 210, 212 Semitic 9, 40, 169 Nilotic 14, 17–21, 29–30, 42–43, 49, 55, 56–57, Shabo 13, 20 63, 91–92, 94–95, 99, 104–106, 114, 117–120, Shangaci 185 129, 138, 147, 148–149, 151–158, 162, 165–167, Shatt 27, 30, 33, 38–39, 43, 58, 59 170–176, 179, 180, 188 Shwai 27 Nobiin 31, 41 Siwu 134, 137, 182, 186 Northern Khoisan 10, 13, 15, 93 Sogoo 94 Nubian 17, 25, 27, 29, 30, 31, 33, 34, 52, Somali 32, 105 45–47, 49, 52–54, 56–57, 59–60, 62, 72, Songhai 12–13, 24 77, 102 Sotho 86, 167 Nupe 41 Southern Khoisan 10, 13, 15 Nyalgulgule 30 Southern Paiute 128 Nyangatom 104, 106, 108, 133, 170, 179 Spanish 124, 125 Nyaro 27 Sudanese Arabic 41, 55, 58, 75, 77, 106, 107, Nyimang 17, 25, 27, 29, 30, 34, 41, 44, 46, 47, 166 52–53, 56–57, 59–60, 63, 72, 74 Suri 22, 76 Surma 22, 76 Oceanic 116 Surmic 17, 19–20, 22, 29–30, 42, 49, 55, 57, 63, Okiek 94, 95, 99 76, 91, 105, 110, 114, 139 Old Egyptian 87 Swahili 117, 119, 159, 162–163, 169, 193 Omotic 10, 40, 54, 168, 181, 185, Ongota 13, 20, 21 Tabag (Tabaq) 27, 31, 33 Otoro 27, 35, 41, 74, 75 Tacho 27 Tagle (Taglennaa) 31, 46, 53–54 Päkoot 156 Tagoi 27, 45, 52 Päri 114, 153, 156, 162 Tagom 48 Polish 193 Talodi 11, 12 25, 26, 27, 29, 33, 35, 48, 52, 60, Polynesian 90, 93, 121 62, 66, 72, 74, 75 Proto-Bantu 12, 160, 162, 166, 167 Tama 20. 36 Proto-Chadic 87 Taman 17, 20, 29, 30, 56, 57 Language Index 219

Tegali 27, 52 Turkana 105, 108–109, 119, 120, 151, 152, 157, Teisei-Umm-Danab 27, 29 170–180, 188 Tekeim 29 Tzeltal 116 Temein 25, 27, 29, 34, 41, 43–44, 59, 72, 74, 79 Ubangian 10–11, 13, 81 Temeinian 17, 29–30, 49, 52, 60, 63, 72 Uncunwee 33, 42, 46 Tese 29, 44, 79 Teso 42, 170, 171, 172, 175, 177 Wali 27, 31, 33, 59, 63, 77 Teso-Turkana 42, 171–177, 179, 180 Wappo 114 Tima 3–4, 19, 23–24, 27, 33, 34, 35, 39–42, Wolaitta 54, 185 44, 45, 48–49, 52–54, 57–59, 61–63, 65–72, 74–82, 127–144, 181–183, 185–186, 188–193, Xhosa 86 194–195 Tira 27, 35 Yoruba 139 Tirma 20, 22, 76, 110 Yucatec Maya 123 Toposa 170–176 Yupno 117 Torona 27 Tulishi 27, 41, 61, 77, 82 Zulu 155, 159–160, 162–164, 166 Tumale 27 Zuni 103 Tumma 27 Tumtum 26–27, 66 !Kung 93–94 Subject Index

accretion zone 3, 15–16, 25, 30, 55 on-record 147, 168 actuation problem 66 Positive Face 168, 181, 182 apparent-time approach 64, 66 Frequency Code 137, 165, 166 ATR harmony; see vowel harmony fronting harmony; see vowel harmony basic colour terms 92, 104, 107, 111, 133 head marking 45–46, 48–50, 52, 125 bio-nomenclature 85, 88–93, 95–97, 99, 114, 127, 128 iconicity 183, 186 body part terminology 88–89, 97–99, 103, iconisation 63 139, 147, 148, 16 ideophones 4, 132–134, 137, 143, 155, 160, 165, borrowing 10, 19–21, 41, 61, 64, 68, 78, 91, 92, 181–186, 191 92, 125, 167, 193, 194 invisible hand 2, 4, 170, 175, 178 case 13, 17, 29, 30, 45–50, 54, 55, 59, 71, 116, language shift 15, 22–24, 64, 74, 76, 78, 81 125 Lexical Iconicity Hierarchy 185, 186 co-evolution 3, 83, 144 lexical-semantic fields; see semantic fields colour; see basic colour terms light verb 17, 185 constituent order 17, 19, 22–23, 29, 30, 47, 48, localist strategy 3, 23, 60, 64, 79–81 49–52, 54, 57, 60, 71, 73–74, 77, 87, 152 converb 1, 17, 53 metaphor 3, 4, 40, 100, 114, 139, 141–143, convergence 15, 17, 20, 24–25, 39, 40, 41, 44, 147–149, 153–154, 156, 162, 163, 165, 167, 168, 47, 54, 55, 60, 62, 99, 194 170, 187–189, 194 conversational strategies 4, 182, 183, 194 metatypy 21–22, 64, 74–76 conversational styles 3, 145,196 metonymy 4, 40, 139, 143, 147, 149, 156, 164, conversational maxim 168 167, 178, 187, 188, 194 coverb 17, 185 multilingualism 18, 47, 61–62, 194 Munsell colour chart 102, 104, 105, 134, 166 dependent marking 45–47, 49, 50, 125 Differential Object Marking 17, 46, 47 Natural Core Model 130 Negative Face; see Face emotion 4, 61, 147, 148, 151–154, 156, 159, 163, noun class 12, 26, 44, 45, 65, 67, 69, ,86, 128, 167, 169, 181, 182, 187–193 136, 137 equilibrium 1, 15, 61 number 13, 21, 32, 42–45, 47, 55, 59, 122, 172, ergative 29–30, 48–50, 70, 77 174 esoteric 4, 40, 143, 193, 194 nurture 3, 4, 101, 124, 187 esoterogeny 21, 23, 64, 74, 76, 193 ethno-science 170 off-record strategy; see Face exaptation 1, 2 on-record strategy; see Face exoteric 4, 40, 143, 193 expansion (zone) 16, 20, 30, 61, 89, 93, 98 perception 97–100, 106, 109, 118, 122, 127, 128, 130, 135, 139, 148, 170, 175–181, 185, 187 Face 153, 168, 181–182 Positive Face; see Face Negative Face 168, 181 PSI (psychocollocation) 147–150, 165, 167 off-record strategy 168 punctuation 15, 61 Subject Index 221

Radical Translator 120, 125 texture 3, 110, 11, 127, 135–137, 166, 183 real-time approach 66 tone 32–33, 69, 165 relativism 99, 112, 126 transition problem 66 residual zone 16, 25, 55 transition zone 76 rounding harmony; see vowel harmony turn taking 194–195 self-organising principles 1, 53 universal 4, 44, 93, 94, 97, 98, 100, 101, semantic fields 86, 90, 101, 103, 112, 121, 124, 102–106, 110, 113, 115, 122, 124–126, 132, 139, 125, 127, 186, 194 165, 179, 182, 187, 192, 193, 195 shape 3, 96, 98, 123, 127, 135, 137, 139, 143, 183 space 101, 115, 116, 118 vowel harmony 23, 34, 37–39, 67, 68, 70, Space Kit 116 97 ATR harmony 23, 34, 37–39, 67, 68, 70, 97 taste 3, 138–139, 183 fronting harmony 23, 34, 67, 68 Taxonomic Hierarchy Model 130 rounding harmony 67, 79