Digitized by tine InternetArcliive ' in 2007 with funding from IVIiq^osoft Corporation

Iittp://www.arcliive.org/details/comparativestudy01jolinuoft

A COMPARATIVE STUDY

OF THE

BANTU AND SEMI-BANTU LANGUAGES

OXFORD UNIVERSITY PRESS

LONDON EDINBURGH GLASGOW NEW YORK

TORONTO MELBOURNE CAPE TOWN BOMBAY

HUMPHREY MILFORD

PUBLISHER TO THE UNIVERSITY

A COMPARATIVE STUDY OF THE

BANTU AND SEMI-BANTU

LANGUAGES

BY

SIR HARRY H. JOHNSTON

G.C.M.G., K.C.IS., P.Sc. (Canibs.),

OXFORD AT THE CLARENDON PRESS 1919

PREFACE

This work on the Bantu and Semi-Bantu languages, of which the first volume is now published, was commenced about seven years ago, in pursuance of a plan determined — as related in the text — very much earlier in my life. It was already beginning to be printed in 19 14, when the War broke out. The effect of the War on the Clarendon Press, as on most other printing establishments, was greatly to delay the setting up in type ; at the same time, the indirect results of the War increased the material at my dis- posal and constrained me to write a longer and more comprehensive work than that which had been originally contemplated : for the campaigns in Africa took place for the most part in countries containing Bantu and Semi-Bantu languages, with the result that many vocabularies were supplied, and thus some languages were brought to light that were previously unknown.

It has been thought best to publish the first portion of this work on its completion. The second part, an analysis and comparison of the phonology and word-roots, and a comparative examination of the syntax of the Bantu and Semi-Bantu languages, together with the conclusions to be derived from this evidence, is also finished ; but its printing and publication must await some relaxation in the present stress.

The work has grown to such dimensions that I have to economize space and paper as much as possible. Therefore I must ask to be excused from tendering in detail my thanks and obligations to my numerous helpers, both 'living and dead. In a way, this acknowledgement is achieved by the very full Bibliography at the end of this volume, and I trust in that not to have omitted the name of any contributor of information. Of the personages in the past whose work has prompted this comparative study of two related language families of Western and Central Africa, there are five whom I should like specially to mention in a dedicatory sense : the Rev. Sigismund Koelle, the Agent of the Church of England Missionary Society, who composed the famous Polyglotta Africana and other studies of African languages at Sierra Leone between 1848 and 1854 ; Wilhclm Heinrich Immanuel Bleek, the Librarian of the Grey Library at Capetown, the real originator of Bantu studies ; Edward Stecrc, Bishop of Zanzibar, and author of many works on the East African Bantu ; the Rev. W. H. Stapleton, of the Baptist Missionary Society on the Congo ; and A. C. Madan, M.A., Student of Christ Church, Oxford, who has done so much to increase our knowledge of the East and South-central Bantu languages and of the Bantu syntax in general. It was mainly due to Mr. Madan's

vi PREFACE

intercession that the present work finally took shape and achieved publication ; and I have deeply regretted the fact that he has died (August 1917) before he could see it in a completed form.

I also desire to tender my thanks to the British South Africa Company, Chartered and Limited, for the efforts made by its London Direction and its employes in South- central Africa to record hitherto quite unknown languages for the purpose of this book : and to the F'rench Government and His Majesty's former Ambassador in Paris, Viscount Bertie of Thame, for the facilities accorded tome in 1 915 to visit the camps and hospitals of the ' Senegalese ' soldiers in France. This asscmbl^e of negroes from all parts of French West and West-central Africa was a singularly fortunate circumstance, inasmuch as it enabled me not only to get an increased acquaintance with the Semi-Bantu languages, but at the same time to check the accuracy of Sigismund Kocllc's invaluable vocabularies which were transcribed sixty to seventy years ago at Sierra Leone, from freed slaves arriving there from all parts of West and Central Africa.

I hope circumstances may permit of this study of the Bantu and Semi-Bantu languages being published in its entirety while I am still able to correct the proofs, and while most of my numerous helpers are alive to appreciate the results of their collaboration.

H. H. JOHNSTON.

Poling,

April 1919.

LIST OF CONTENTS

CHAPTER I

PAGE

A HISTORY OF RESEARCH INTO THE BANTU LANGUAGES .... i

CHAPTER n

THE DISTRIBUTION AND CHARACTERISTICS OF THE BANTU LANGUAGES . 15

CHAPTER HI

ILLUSTRATIVE VOCABULARIES OF 366 BANTU AND 87 SEMI-BANTU LAN- GUAGES AND DIALECTS 39

The Bantu Languages :

Group A. The Nyanza Languages 45

B. The Wunyamvvezi Languages 86

C. The British East Africa Languages 98

D, E. The Kilimanjaro) and Usambara Languages 113

F. The Zangian Languages 129

G. The Usagara-Ugcogo) Languages 141

H, I. The Upper Rufiji and Lower Rufiji-Ruvuma Languages 154

I (cont.), J, K. The Lower Rufiji-Ruvuma. North Ruvuma-North-east Nyasaland. and

Ukinga Languages 169

^ L. The Tanganyika-Bangweulu Languages 184

M. The North-west Nyasa Languages 208

N, O. The Vac)-Ngindco and Mogambique Languages 221

P. The South Nyasaland Languages 236

P (cont.), O. The South Nyasaland and Southern Rhodesia-Pungwe-Sabi Languages . 252

Q (cont.), R. The Southern Rhodesia-Pungwe-Sabi and ^engwe-Ronga Languages . 268

R (cont,), S. The ^engwe-Ronga and Becuana-Transvaal Languages .... 281

S (cont.), T. The Becuana-Transvaal and Zulu-Kafir Languages 298

U, V. The West Central Zambezia and Western Zambezia Languages . .... 318

V (cont.), W. The Western Zambezia and North-west Zambezia Languages . . . 333

X. The South-west Africa Languages . 350

Y. The Angola Languages 366

Z. The Kong(u or Westernmost Congoland Languages 378

AA. The South Congoland or Luba-Lunda Languages 394

AA (cont.), BB. The South Congoland and Upper Kwangu Languages . . . .411

BB (cont.), CC. The Upper Kwangco and Kwangco-Kasai Languages .... 427

CC (cont.), DD. The Kwangco-Kasai and Central Congoland (Luafige-L<omami) Languages 439

DD (cont.), The Central Congoland (Luange-Lcomami) Languages 451

DD (c^w/.), Sub-group DD I. The Central Congoland and Manyema Languages . . 462

viii LIST OF CONTENTS

PAGE

DD I {coni.), EE, FF. The Manyema, Middle Ltomami, and Elila-Lmwa-Lualaba (Bulega)

Languages 473

FF(«:<w«/.), GG, HH. The Elila-Lcowa-Lualaba, Ruwenzori-Semliki, and Upper Ituri

Languages 4*^4

HH (cont.), II. The Upper Ituri and Wele-Aruwimi (Ababua) Languages . • 496

JJ. The Aruwimi-Lcomami Languages 509

JJ {cont.), KK. The Aruwimi-Liomami and North Central Congoland Languages (Sub- groups KK 1, KK 2) 520

KK {cont.). North Central Congoland Languages (Sub-group KK 3) 533

¥iK(cont.). North Central Congoland Languages (Sub-group KK 4) 546

LL. The Kwa-Kasai-Upper COgtowe (Teke) Languages 553

LL {cont.), MM. The Kwa-Kasai-Upper OOgcowe (Teke) and Central a)g(o\ve Languages 565 MM (<r^«/.), NN. The Central OOgcowe and COgtowe-Gaboon Languages . . -578 NN (cont.), 00. The OOgcowe-Gaboon (Mpongwe) and Spanish Guinea-West Cameroons

Languages 5^9

00 (cont.). The Spanish Guinea- West Cameroons Languages (Sub-groups OO I, 00 2,

003,004) 604

00 {cont.). The Spanish Guinea- West Cameroons Languages (Sub-group 00 4 (cont.),

00 5) 618

00 (cont.), PP, OQ. The Spanish Guinea- West Cameroons (Sub-group OO 6), Manen-

guba, and Middle Sanagi Languages ■ 630

RR, SS. The Paiiwe or ' Faiig ' and Kadei-Saiiga-LtDbai Languages .... 642 SS (coni.), TT. The Kadei-Lcobai (Upper Saiiga) Languages and the Fernandian

Language and Dialects 655

The Semi-Bantu Languages :

A. The Cameroons-Cross River Languages (Sub-group A i) 672

A (cont.). The Cameroons-Cross River Languages (Sub-group A I (cont.), Sub-group A 2) 698

B-G The Northern Cross River Basin, Benue, Bauci Languages 716

H, I, J. The Central Nigerian, South-west Tcogcoland, and South Sierra Leone Languages 732

K, L, M, N. The South Guinea, Nalu, North Guinea, and Upper Gambia Languages . 750

CHAPTER IV

SUPPLEMENTARY ADDITIONS AND CORRECTIONS TO THE FOREGOING

VOCABULARIES 773

CHAPTER V BIBLIOGRAPHY OF THE BANTU AND SEMI-BANTU LANGUAGES . . .785

MAPS

SKETCH MAP OF BANTU AND SEMI-BANTU -AREAS 818

KEY-MAP GIVING NAMES S19

ERRATA ET ADDENDA

The printing of the first portions of this work began in 191 4, and the correction of the proofs was somewhat interfered with by the author's absence in America, and later on in France. A few errors, therefore, escaped his notice in the earlier chapters till after the printing-off had taken place ; or some- times words or phrases became erroneous on the receipt of further information from Africa.

Also, during the printing of this book, much new material has been received, dealing more especially with the Konjco (No. i) and other Nyanza tongues ; with Taita (14), Nika (16), Caga Dialects (17-17 e), pambala (19), Zigula (20), Swahili (21-21 g), Pcogorco (28), Kidonde or Kimawanda (33 a), North Ngindto (55n), Makua (56), North Kuanyama or Humbe (91a), Kuvale (92 bj, OOlundombe (93), Sumbi (94b), Songto (97), Minungoj (112a), Hcolu (113), Mpama (167), Batende (i68a), and Kiwumbu (177).

It has been thought more convenient to give the mass of the addenda and corrections at the end of the Vocabularies, the greater part of which were in print before this additional information was received ; and only to insert in the preliminary pages a correction of the errata in the earlier work. But readers specially interested in the languages above cited are advised in studying them not only to take note of the correction of errata here given, but of the additional information at the end of the book ; and students who wish to have before their eyes, without turning to the beginning or end of the volume, as complete a conspectus of the Bantu languages as can be produced at the present time, might insert with the pen the additions to and corrections of the Vocabularies in the appropriate places.

On pages 3, 5, 6, 13, and 14, and wherever else the word occurs, for 'pojsa' read 'posa'.

On page 7, and wherever else it occurs, for ' Herero' read ' Hererto '.

At the bottom of page 14 should be added this sentence : ' and of the Semi-Bantu of Senegambia and Nigeria'.

In the foot-note to page 16 a hyphen is missing in ' Tatoga-und-Irakuleute' .

On page 17, in line 5, insert the word ' one ' after ' differing '.

On page 19, line 20, 'preprefix ' must be substituted for 'prefix'. In line 34, instead of ' Class 6, its plural ', substitute : ' Class 6, apart from being the plural of Class 5 '. In line 37, instead of ' Classes 9 and 10 were', read 'Classes 9 and 10 are'. In line 39, instead of 'Classes 12 and 13 were", read ' Classes 12 and 13 are ', and instead of ' Class 13 (Ka-) could ', read ' Class 13 (Ka-) can '. In line 40, after ' Class 14 (Bu-) ', change 'was' to 'is'; and in line 41, 'Class 15 was ' should read 'Class 15 (Ku-) is '.

On page 20, in line i , the words ' but was ' should read ' but is '. In Hne 2, ' Class 16 represented ' should be ' Class l6 represents ' ; in line 3, change ' had ' and ' meant * to ' has ' and ' means ' ; in line 4, change ' existed, were ' into ' exist, are '. In line 5, for the word ' certainly ' read ' first quinary and then '. In line 1 7, after ' No changes ', insert the words ' in sense '. In line 19, after ' prepositions, and ', insert the word ' perhaps '.

On page 22, line 5, instead of the words ' The Fula ', read ' The Semi-Bantu, the Fula *, and in line 6, after ' Wolof ', strike out the following words, ' the Temne '. In the footnote, Hne 7, strike out ' Temne ', and after ' Bantu ' insert ' Semi-Bantu '. At the end of line 9, instead of ' Sudanese ' read ' Sudanic '.

On page 25, in line 3, after ' fusion ' insert the words ' some of.

On page 26, line 4, instead of ' akin ' read ' related '.

X ■ ERRATA ET ADDENDA

On page 27, in line 4, for ' dates ' read ' date '. At the end of line 14, instead of ' And here ' read ' And hence '. In line 25, instead of Africa has been a long process in unwritten history and cannot be compressed, &c. ', read ' Africa has not been a long process in unwritten history and can be compressed, &c. '.

On page 29, at the end of line 5, the letter a should be deleted from ' defecations *, which word should read ' defections'. In line 33, instead of ' Bantu languages which ' read ' Bantu languages that '.

On page 30, in line 8, the first two equivalerts of ' hair ' should be rendered -wele, -wili, and not •bele, -bUi. In line 17, among the word-roots for 'blood ' should be inserted -kila (after -rupa).

On page 31 a table is given of the original 19 prefixes of the Bantu languages. It should be premised that this is the full list of prefixes with concord particles or answering pronouns. But in addition there are now known to be five or more other prefixes — locatives, or indicative of sex or honorability. These, however, are without concords.

On page 32, line 4, ' m- before a nasal' should read ' m- before a labial '.

On page 36, in line 9, the simple vowel system of Old Bantu should read : ' — 5, a, e, i, 5, and perhaps a* — '.

On page 39, in the Title of Chapter III, the number of Bantu languages and dialects should be enlarged from ' 276 ' to ' 366 ' ; and of Semi-Bantu languages and dialects from ' 24 ' to ' 87 '. Similarly, on page 42, in the last paragraph but one and in the footnote, the figures 376 (total number of Bantu and Semi-n.intu languages and dialects) should be changed to ' 366, of which about 360 are illustrated '.

On page 43, n the list of Bantu language Groups, P should stand for 'THE SOUTH NYASA- LAND LANGUAGES ',Q for 'THE SOUTHERN RHODESIA-PUNGWE-SABI LANGUAGES', and R for 'THE ^ENGWE-RONGA LANGUAGES'.

On page 44, and paragraph, the Groups of the Semi-Bantu Languages should be rendered thus : Group A. (S-B.). The Cameroons-Cross River Languages.

B. (S-B.). The Northern Cross River Basin Languages.

C. (S-B.). The South-west Benue Languages.

D. (S-B.). The Southern Benue Languages.

E. (S-B.). The Central Benue Languages.

F. (S-B.). The South-west Bauci Languages.

G. (S-B.). The Central Bauci Languages.

H. (S-B.). The Kaduna Basin (Central Nigeria) Languages. I. (S-B.). The Tcogioland Languages.

J. (S-B.). The Southern Sierra Leone Languages.

K. (S-B.). The South Guinea Languages.

L. (S-B.). The Nalu Language.

M. (S-B.). The North Guinea Languages.

N. (S-B.). The Upper Gambia Language. Page 85. In the prefixes, &c., of Lu-Masaba and I.u-konde, among the forms of the sth prefix Ki- is erroneously given. It should read Ri-.

Page ICO, col. 16, the root for ' Ear ' should read Sikirco.

Page too, cols. 14. 15, 16, opp. ' Fire ', render the words Mwdcn, Moi-tu, Mw-hu as M>odo>, M-5tw,

. ' Frog ', for C-ua read Cua. . ' Foot ', for Su-aiyo read Lu-aiyo. ' Thorn ', for D-zala-g-umba read Dzala-gumba. . ' Wood ', the root should be spelt .kuni, -khuni, not •kumi, -khumi. ' Inside ', for -a n-denji read -a n-denyL Page 113. The alternative name of the principal Caga dialect should be Mcofsi, not Musi. Page 114, col. 20, opp. 'Bull', for N-jeka read N-jeku.

(Page 140. The concords of the 4th Class of prefixes in Swahili and Swahili dialects should be stated as mi", -ni, -i-), not (m-, mu-, i-).

Page 153, last paragraph, for ' Lake Manyasa ' read ' Lake Manyara'.

Page 198, col. 17. The word Ily-enza should be opposite ' Egg' and not ' Ear '.

Page 228, col. 54, opp. ' Stone '. for Li-gonga read Li-ganga.

and M^ohea.
Page 1 01, col.	12, opp.
Page 1 01, col.	14, opp.
Page 105, col.	16, opp.
Page 106, col.	J 3. opp.
Page no, col.	14, opp.

Page 5, thirteenth line from bottom of last paragrapn, tor ' l.ewis (irants ' read

Page 6, tenth line from bottom of last paragraph, for ' Ba-yeiyi ' read ' Ba-yeiye '.

Page 7, fifteenth line from top, for •// read -//'.

Page 10, fifteenth line from top, for 'Sudanese' read ' Sudanic ', and tenth line from bottom, for ' C. Hermann ' read ' C. Herrmann '.

Page II, fourteenth line from top, after ' Lutheran ' insert ' pastor and '.

Page 12, twenty-first line from top, for ' Dr. Eduard Sachau ' read ' P. Hendle ' ; and on p. 791, par. 28, instead of ' Prof. ' insert ' P. Hendle, edited by '.

Page 29, one line from bottom, for " two hundred and seventy-six ' read ' three hundred and sixty-six '.

Page 33, three lines from top, for ' consonantal ' read ' substantival '.

Page 37, eight lines from top, after 'twenty-six' insert 'main '.

Page 378, and again on page 803. In the summary of titles at the commencement of the Vocabu- laries on p 37S, and again in the bibliography on p. 803, Ki-yombe or Ci-luangoj should be numbered 102, Ki-vili or Ki-vumbu or Lu-wumbu (N.Luangco) 103, and Ki-lumbco of Mayiivtha. 103 a, conform- ably with the Vocabularies.

Page 792, par. 33, delete 'preliminary Corrigenda ct'.

Page 798. In the paragraph numbered 76, the name ' Elmsley ' should be spelt ' Elmslie '.

Page 81 5, twelfth line from bottom, for ' Sengalese ' read ' Senegalese '.

1734 Johnston^ Uaittu.

Sept. 1919, face p. \.

ERRATA ET ADDENDA xi

Page 231, col. 54, opp. ' This, these ', the demonstrative of the i6th Class should read Apa, not (ji)pa Page 237, col. 61 a, opp. ' Cloth ', for N-garu read N-saru. Page 315, col. 76, opp. ' Sit, remain, &c. ', instead of -jslaza read -slara.

Page 350. The beginning of 4th footnote should run : ' This language in different dialects is, &'c.\ Page 353, col. 92, opp. ' Fear ', for OOn-uma read COu-uma. Page 355. In the English column, ' Man, vir.' is missing opp. to Ci)mu-rumendu. Page 359. The word for ' Nine' in col. 94 is perhaps better spelt £-cia, E-cieha ; and for ' Hundred' OO'Cita. In col. 92 the roots for ' Hundred ' may be -tyita and Cita, instead of -ta.

Page 365. The last paragraph should read : ' 94. Umbundu is spoken in the Benguela and Bai- lundio districts of southern Aiigola, north of 14° 30' South latitude, south of the Kuvuj river, and of 11° 30' South latitude ; and west of the Upper Kwanza river.'

Page 377. The definition of the area of the Kisama language (95) should be limited on the south by the Kuvo) river.

Page 410. Last paragraph, the close of the definition of the area of 108-108 a should read, ' West of the Upper Lcomami, south of the Lubefu, and east of the Safikuru '.

Page 553. In list of titles for ' 177. Ki-wumbco' read ' 177. Ki-wumbu ', and Ki-wumbu likewise on page 564.

Page 606, col. 204, opp. ' Fire ', the figure ' which refers to the footnote should be removed from li-didi and placed after 'aji're', which concludes the sentence in brackets. And in the answering foot- note ' Mote-mote ' the two vowels o should be stressed o.

Page 675, opp. ' Child', cols. 230-231, after Ma-nku insert Mu-pie (231).

Page 690, col, 230, 231, opp. 'Nine', the whole passage of equivalents for 'Nine' should be rendered thus :

Li-vojoj.

Bojoj (230 b, 230 g). Sipo (230 i). Vua.

plbco (230 c)- M-bu, Pfcdoi (230 j). Bb, Beo> {230 d). I-fi(23of). Ve, Be, Bemi. Tsiaiiuco (231). Page 696, seventh line from bottom, at the end of line an 'a' should follow ' indicate '. Page 698, in the list of languages, the line beginning ' (238. Olulciimu. See Appendix, &c.' should read ' (238. Olulcomo). 238 a. Ikom. See Appendix) '.

Page 705, col. 244, opp. ' Thigh ', U-nau should becorrected to U-na&.

Page 715, third line from bottom, ' Kwoi-Ibco' river should be spelt 'Kwo-Ibai'.

CHAPTER I

A HISTORY OF RESEARCH INTO THE BANTU LANGUAGES

Outside the range of the Semitic and Hamitic families of languages, as represented by Arabic and Amharic, the first types of African speech to attract the attention of Europeans and to be written down by them were members of the widespread BANTU group ; an association of tongues as closely interrelated as clear-cut and unmistakable in their peculiar characteristics as the Aryan, the Semitic, or the Malay families of speech. The Bantu languages of the central prolongation of Africa were of course totally unknown to the Graeco-Roman civilization of the Mediterranean, the cultured minds of which may just have been able to perceive (if they collected and collated the evidence of travellers) that beside the recognized types of Hamitic and Semitic languages of north and north-east Africa there were the tongues of the Nilotic negroes, of the Nubians, of the Tibu (Garamantes), and of the Fula and Wolof negroids on the Atlantic coast of the Sahara and along the banks of the Senegal river. And although we may assume on fairly sufficient evidence that the Arabian trading-cities of the Yaman and Hadramaut coasts had founded depots for commerce on the Equatorial East African littoral as early as the commencement of the Christian Era (if not before), and had likewise got into touch with the north end of Madagascar and the Komorco Islands, it is more probable that in those days— eighteen hundred to two thousand years ago — there were no Bantu-speaking negroes on the east coast of Africa. Consequently, though the merchants of south- west Arabia, who acted as important intermediaries in the Indian trade with the Roman Empire in Egypt, may have conveyed slaves from the Zangian coasts and islands to Egyptian slave-markets, it is doubtful whether these brought with them any Bantu syllables into the medley of tongues talked in the Mediterranean basin.

But not long after Arab and Bantu first met on the east and south-east littoral of Africa there seems to have been a greater fusion of interests between them than between Arab and Kushite (Gala and Somali). The Arabs took up the stock language of the Zangian group somewhere on the coast between the Pafigani or Rufu river and the Ruvuma and turned it into the Swahili tongue of commerce that we know so well at the present day— the easiest and most widely spread of the Bantu languages.

When the Portuguese rounded Cape Verde and reached the mouth of the Congo during the second half of the fifteenth century, they nowhere found a native language (excepting, of course, North African Arabic) sufficiently easy of pronunciation or simple in structure to be used as a medium of instruction and intercommunication, until they reached Bantu Africa. Their missionaries soon began to learn Kifi-kofigto and a little later the Mbundu speech of Aiigola. Kifi- kofigo) words are given in Pigafetta's description of the Portuguese Congo explorations published in 1591. Kifi-kofigd) was committed to writing and to print (in the form of a treatise on Christian doctrine) as early as 1624 by a Jesuit (Portuguese) missionary ; and a Grammar (see Bibliography in Chapter IV) by an Italian missionary was published at Rome in 1659. Kimbundu of Aiig5la was

1784 B

2 A HISTORY OF RESEARCH INTO

illustrated by Italian missionaries in 1642 and 1661 ; and a Grammar of Kimbundu (by Diaz) was printed at Lisbon in 1697. A study of Luafigu) and Kakoiigco was carried out in the eighteenth century by French missionaries, such as' the Abbe Proyart.

As early as 1505, but chiefly between the middle of the sixteenth century and 1760, Portuguese soldier-explorers and missionaries put into their writings numerous phrases of two East African Bantu tongues: Karana (' Mocaranga ') and Swahili. The amount of the first-named language which can be gleaned from these early records of Portuguese expeditions in Zambezia and on the Sufala coast is considerable, and has been conveniently collected for me by Mr. R. W. Hall, the Warden of Zimbabwe. Though carelessly transcribed by these sixteenth-century Portuguese, with little regard for grammar or plural prefixes, it is nevertheless easily recognized as Karana and as I easily translated. It shows such comparatively little difference from the same tongue spoken to-day after an interval of between four hundred and one hundred and fifty years that we should be puzzled to account for the rapid formation and divergence of the two hundred and twenty-six distinct Bantu languages of present times ; did we not remember that the Romance languages were not in existence in the fourth century of the Christian Era, yet were evolved with all their characteristic features by the twelfth century ; and that the French and Italian of three hundred and fifty years ago are not more dissimilar from the speech of to-day than is the tongue of the Rhodesian Mafuna from the 'Mocaranga' of Monomotapa recorded by the Portuguese sixteenth-century pioneers.

A short list of words of the Komoro Islands Swahili— the dialect of Moihila— was written down as early as 1626 by Thomas Herbert, an Enghsh traveller who voyaged in that year round the Cape of Good Hope to Aden and thence to India, whence he returned overland through Persia, Caucasia, and Russia.' In the eighteenth century the Portuguese traders and officials on the Lower Zambezi took up the language of Sena and made it a kind of lingua franca in eastern Zambezia under the name of ' Chicunda '. [This Cikunda speech was the first Bantu language learnt by the English and Scottish missionaries who followed Livingstone, and a short manual of it was printed by the Rev. Dr. Robert Laws in 1879 or 1880.] As early as the seventeenth century, the Portuguese, with Kimbundu and Kipi-kongo) on their Atlantic coast possessions, Cikunda on the Lower Zambezi, the speech of Tete (Cinyungwe) on the Zambezi of the Cataracts, and the Citoiiga learnt by the Jesuit missionaries on the Middle Zambezi, had realized the truth about the wide range of the Bantu languages. In the eighteenth century they repeatedly referred in their writings to the similarity of speech between Aiigola and Mozambique.

The Dutch explorers, following in the footsteps of the Portuguese along the West African coast, also began to interest themselves in these harmonious, easily pronounced Bantu languages. Barbot, ' who wrote his Description of Guinea about 1682 (English translation first published in 1732), gave short vocabularies of Kip-kongoj and of a dialect of north-west Cameroons, probably Barundco.' Barbot was a Frenchman in the service of the French Royal Company of Africa, but he seems to have derived much of his information east of Senegal from the Dutch. In the latter half of the eighteenth century — the period of Cook, of La Perouse, of James Bruce — greater curiosity than ever before was evinced as to the languages of non-European peoples. The Dutch occupation of Cape Colony had attracted to South Africa a number of explorers— French, German, and Scottish. The strange Hottentot speech had already been transcribed to some extent by Wilson in 1691, and between 1705 and 1713 by Pieter Kolben, the industrious and intelligent Prussian explorer of South

' These words of the ' Mohella ' language recorded by Thomas Herbert do not seem to be met with in the first (1634) edition of his travels. They were presumably collected in 1626, but were not inserted into his book until the 1677 edition was produced. He does not give a ' vocabulary ', as stated by R. N. Gust (Modern Languages of Africa), but a few words, such as Coguo (fowl), Gumbey (ox), Mage (water), Sinzano (needle), and Buse (goat).

' As regards the words of Barundco said to have been collected by Barbot, I quote from Astley's Voyages and Travels, '^\:i\i\\%\\iiA.'va. 1746. They are the first five numerals: Mo; Ba ; Malela; Melei; Malau ; and are attributed to the ' land of Ambozes ' (Ambas Bay).

THE BANTU LANGUAGES 3

Africa. But the Dutch had been slow in coming into contact with the Bantu peoples of South Africa. To the tribes of stalwart, naked Bantu negroes dwelling on the south and south-east coasts they had given the name of'Caffer', copying in this the Portuguese who had picked up ' Cafre ' from the Arabic Kafir (a heathen). Beyond Hottentot- and Bushman-land, in the centre of southernmost Africa, the Dutch began (about 1779) to be aware that there existed a ' nation ' of tall black negroes whom they called 'Beetjuaan ' by some corruption of an unidentified native word which has since been perpetuated as ' Bechuana '. It was made evident by about 1806 that neither ' Beetjuaan ' nor ' Caffer ' spoke anything like Hottentot or Bushman ; and equally clear that many of the Dutch slaves from Mo9ambique, Scofala, or Angola could make themselves understood by these tall negroes beyond the Great Fish river or north of the Orange. In 1772 an unnamed French missionary compiled a Kakofiga)- French dictionary, and in 1776 the first vocabulary of p<osa (Kafir) was printed, in the work of Andrew Sparrman, a Swede. In 1804 the Portuguese Cannecattim published his Angola Dictionary. In 1809 Henry Salt, a pioneer diplomatist of the British Government in East Africa, wrote down vocabularies of Makua, of some Yaco dialect, and of Swahili.'

In 1808 the notable German traveller, Heinrich Lichtenstein, had grasped and expressed the idea (in a Berlin scientific periodical) that all the southern prolongation of Africa, from the Congo coast on the north-west, to the Equator on the north-east, was probably the domain of but one language family— excepting of course the narrow band of Hottentot-Bushman speech from Cape St. Francis to Cape Frio. [Lichtenstein's Travels in Southern Africa in 180J-6, were published in their English translation in 1812-15.] The pioneer missionaries of South Africa soon confirmed this theory by their practical studies, and there were always at hand in those days slaves from the Portuguese East and West African possessions and from Zanzibar to confirm the idea by contributing specimens of their more northern forms of Bantu. A vocabulary of the Makua language of Mozam- bique was, for purposes of comparison, written down from the dictation of a negro slave in Sumatra by William Marsden, the Orientalist, somewhere about 1778. He himself was an authority on the Malay language, but he became interested in African philology, and handed his Makua vocabulary to the scientific members of Captain Tuckey's staflT who accompanied that officer in 1816 in the attempt to explore the Congo from its mouth up stream. These men went there with the Bantu theory in their minds, though the comprehensive word ' Bantu ' was not then suggested. They refer to the existence of this one great family across the southern third of the continent in their contributions to the Narrative of an Expedition to explore the River Zaire (Report on the Tuckey Expedition) published in i8i8.

Adrien Balbi, in his Atlas ethnographique du Globe, printed at Paris in 1826, precognized the existence of one great language family over all Africa south of the Equator (exception being taken for Hottentots and Bushmen). He based this theory on the observations of Moritz Rugendas, a Bavarian artist who had travelled through eastern Brazil in the early nineteenth century and had taken down descriptions of West Africa and scraps of speech from negro slaves.'' Douville,

' Henry Salt was sent to convey presents to the Kings of Shoa and Abyssinia in 1809, to detach Ethiopia as much as possible from French intrigues against Egypt. On his way to Somaliland round the Cape of Good Hope, he stopped at Mozambique and there wrote down short vocabularies of ' Makooa ' (easily recognizable as Imakua) and ' Monjow ' (a jargon seeming like a mixture of Yaco, Gindto, and Northern Makua (Medio)). His words of ' Monjow' (re-spelt phonetically) are as follows : Mcalungon (God); Dyuva (sun); Muezi (moon); Tuundnwa. (stars); Muze (earth); Maiumbi (mountain); Mere (tree) ; Aumba . . . ? Numba (house) ; Etala (road) ; Meze (water) ; Sieinba (fish) ; Niyati (ox, evidently buffalo) ; Liguluwe (pig) ; Umpua (dog) ; Itepu (elephant or ivory) ; Nyama (flesh) ; Nune (bird) ; Eyunda (pigeon); Mulutpuiana (man); Ateate (father); Alumbu (brother); Bo) (sister); Manace (boy) ; Mutwe (head) ; Humpu (hair) ; Makutwe (ears) ; Unwa (mouth) ; Atigonda (war) ; (jiknazi (a bow) ; Impamba (arrows) ; LipongU) (spear) ; Uti (gun) ; Jete (salt) ; Jipanje (white) ; Ya-kuswers (red); -atigu (my). For a description of Salt's Swahili ('Sowauli', 'Sowaiel'), see the Bibliography of Chapter IV.

' The results of his observations are said to have been published in or concurrently with his Voyage pittoresqtte et descriptif au Brhil, ? 1824, but I cannot trace them. Rugendas dealt with the languages of Angola and Mozambique.

B 2

4 . A HISTORY OF RESEARCH INTO

a French traveller in Angola (probably in the interests of the slave trade) who much later became a secretary of the Paris Geographical Society, published a book on the Congo and Angola in 1832; and although his records of Bantu languages are very jumbled and full of inaccuracies, still they are not as apocryphal as they were afterwards esteemed. Douville not only mentions some of the principal language groups in Angola and the western Congo, but realizes through Portuguese information the importance of the Luba family beyond.

In the 'forties of the nineteenth century great progress was made in Bantu studies. The (English) Baptist Mission had been established in Fernando Po, and John Clarke, one of its missionaries, had already put down on paper information regarding the Fernandian or Bube language. He also published specimens of 209 African dialects, taken down from the slaves who were landed at the Admiralty depot at Fernando Po. Although there were very few words of each language (some are more illustrated than others), yet his hearing was fairly accurate and his writing phonetically correct. His work has a certain usefulness even at the present day. At this time also the great Sigismund Koelle, like Krapf and Rebmann, a native of the kingdom of WOrttemberg, in southern Germany, and a clergyman of the Church of England, was employed by the Church Missionary Society to watch the interests of the freed slaves landed at Sierra Leone. He took advantage of this opportunity to collect the materials for his remarkable Polyglotta Africana, published by the Church Missionary Society in 1854. This work deserves the highest praise, even in comparison with philological work at the present day. It is the more remarkable, since it was compiled over sixty years ago. So far as the Bantu languages are concerned, Koelle proves to be more interesting perhaps than he ever anticipated. He placed on record not only many facts of Inner African geographj' which were to remain unrealized for several decades longer, but he wrote down in 1851-2 the speech of Bantu peoples who were not to be actually visited by the white man till — in some cases— the beginning of the twentieth century. Excepting here and there a few lapsus calami and errors due to mishearing or misunderstanding, his work is remarkably correct and also shows that many of the languages transcribed by him have changed but little in the course of sixty years. Nearly all his Bantu vocabularies are included in the scope of the present work. Koelle appreciated to the full the fact of the existence of this great family of closely-related African languages, but preferred to call it the ' South African ' family. He was not sufficiently acquainted with the structure of Bantu speech to include all his Bantu vocabularies under the same heading ; several of them appear as ' unclassified ' languages. A few are incorrectly named, but are easily identified by their affinities and the geographical information he gives as to their position. Considering the absolute blank of the map of Africa in those days, behind the coastline, his geographical delimitations of the spheres occupied by the enormous number of languages he tran- scribed (due to the collation of the information he took down from the freed slaves) is remarkably accurate, and has been a great help to myself and other recent writers on the Bantu and Semi-Bantu languages in identifying and locating the forms of speech which he recorded.

Whilst Koelle was working at Sierra Leone and the Baptist missionaries at Fernando Po and the Cameroons,' the American Board of Foreign Missions was despatching evangelists to settle on

' In referring to the work of the Baptist missionaries in early days in the Cameroons, I must specially signalize for notice and gratilude on our part the remarkable work achieved by Joseph Merrick, who- under circumstances of great difficulty printed a dictionary of the Isubu tongue in 1842 and a Grammatical Note in 1854 ; and of Alfred Saker, who issijed a grammar and vocabulary from his printing press on the Cameroons estuary, in 1855 •'■^'^ 1862. Merrick's work was the more remarkable in that he was a negro or mulatto who had received his education chiefly, if not entirely, in the island of Jamaica. He was one of the early recruits of the Baptist Missionary Society in its desire to establish civilized settlements for the relief of freed slaves on the coast of Equatorial West Africa. It is difficult, as one reads Merrick's dictionary of Isubu, to realize that this work was compiled so anciently as the 'forties of the last century and by one who had such slight advantages in the way of education, though he had evidently made the utmost use of the instruction given him.

THE BANTU LANGUAGES 5

the coast of the Gaboon. Here they not only made the discovery of the Gorilla, but through the linguistic zeal of the Rev. J. L. Wilson, of Messrs. Preston and Best, and in later days of the Revs. R. H. Nassau, H. M. Adams, J. Bushnell, and of Mr. and Mrs. Arthur Marling, they recorded the Kele or Kalai, the Mpongwe, the Benga, and the Paiiwe (Fang) languages of French Congo and Spanish Guinea.

At the same period also— the 'forties and early 'fifties— the German naturalist. Dr. Wilhelm Peters, of Berlin,' was exploring Portuguese East Africa and writing down words of the Bantu languages spoken there. And those noteworthy pioneers of East Africa, the missionaries Johann Ludwig Krapf and Johann Rebmann (both of them natives of Wurttemberg, and, like Koelle, in the service of the English Church Missionary Society), were revealing the Bantu languages of Equa- torial East Africa from the Tana river to south-west Nyasaland, either through their travels or by the interrogation of slaves. Krapf realized the existence of the Bantu family, but called it the ' Nilotic ', and later the ' Orphno-Cushitic ' ; very inappropriate names, considering that the Nile basin and the Kushite (Gala-Bisharin) race barely come within the Bantu sphere, and that there exist the well-defined and quite independent families of Nilotic negro and Kushite languages."

Missionaries of the United Presbyterian Church of Scotland simultaneously were settled at Old Calabar on the eastern confines of Southern Nigeria, and were exploring the hinterland up the Cross river. Though they collected little or nothing in the way of Bantu languages, they were enabling philologists to define the limits of the Bantu family on the north-west.

In South Africa a great impetus was given in the first half of the nineteenth century to the study of languages by missionary work and by the explorations of Robert Moffat and David Livingstone. The first adequate grammar of pwsa (Kafir)' had been published by W. B. Boyce in 1834, and was succeeded in 1846 by a vocabulary or a short dictionary written by John Ayliffe, and Apple- yard's Grammar in 1850. Then came Lewis Grant's Zulu Grammar in 1859, and a Zulu Grammar and Dictionary by the eminent Bishop Colenso. But the great event in the history of Bantu studies was the appointment in i860 of Wilhelm Heinrich Immanuel Bleek to be Librarian of the collection of books presented by Sir George Grey as a public library to Capetown. Bleek was born in 1827 at Berlin, and was a student at Bonn University, where his attention was directed to the arrange- ment of South African languages and the affinities of Hottentot speech. He published a disserta- tion in Latin on this subject in 1851, and this attracting the attention of those who were advising the British Government in regard to the Niger expeditions, he was offered a post on the Niger in 1854 (in conjunction with Dr. W. A. Baikie) ; but his health prevented his taking up this appoint- ment. Instead he accompanied Bishop Colenso to Natal, where he began to study the Zulu language. Moving to Capetown in 1856 his value was at once appreciated by one of the few great Governors who ever went out to South Africa— Sir George Grey (a man whose services to the British Empire and the world at large have never yet been sufficiently appreciated). Sir George

' Dr. Peters's travels took place between 1842 and 1848.

" The Rev. Dr. J. L. Krapf issued from the Tubingen University Press (Wiirttemberg) Vocabularies 0/ Six East African Languages in 1850, a Vocabulary of ^ Wakuafi' (Masai) in 1854, and compiled in the course of many years his Swahili- English Dictionary, which was published in London in 1882. Although this work is of great interest to students of the origin and affinities of the Swahili dialects, its value is marred by the author's eccentric orthography and his preference for the Mombasa dialect over that of Zanzibar. He is also too ready to include as ' Swahili', words really belonging to quite different Bantu languages spoken in his time by the slaves of the Arab settlers. Krapf also collected materials for a Nika Dictionary which was afterwards completed under the editorship of the Rev. T. H. Sparshott (1887).

Johann Rebman or Rebmann^both spellings are used by him — one of Krapfs colleagues, had compiled an interesting study of ' Ki-niassa ' in 1853-5, which was not published till 1877. This was more or less the Ci-maravi dialect of the widespread Nyanja tongue. Rebmann also contributed to the above-mentioned Nika Dictionary.

' As already mentioned, the Kafir language was first transcribed in print by Andrew Sparrman in 1776. Sparrman was the naturalist to Captain James Cook's second expedition to the Pacific.

6 A HISTORY OF RESEARCH INTO

Grey gave him a post as Government Interpreter at Capetown, which he enlarged into that of Librarian in i860. As early as 1856, however, Bleek, through studying the vocabularies written down by Dr. Wilhelm Peters, had not only grasped the unity of the great language family of Central and South Africa, but had proposed as an appropriate name for it the term BANTU, a term which he derived from the wellnigh universal word for ' men '.' Bleek considered that, despite variations in pronunciation, this term was so universal and so characteristic of the languages to which it belonged in exhibiting the detachable prefix and unchangeable root, that it should preferably be selected for their universal designation, for the additional reason also of its shortness and easy pronunciation.

Bleek really laid the foundations of Bantu philology. In 1862 he published at Capetown the first volume of his Comparative Grammar of the South African Languages, and in 1869 the second volume (in London). He had previously brought out a handbook of Australian, African, and Poly- nesian philology between 1858 and 1863, and although his conclusions are too far-fetched, especially in detail, and are not borne out by a more scientific study of these African and Oceanic forms of speech, still he detected many an obscure truth or half-truth. In his scheme for the Comparative Grammar of the South African Languages he hampered himself by attempting to deal with Hottentot (and no doubt later on, Bushman) as well as with Bantu. As there is no affinity between the two groups other than that they are both forms of speech uttered by Africans, this was a mistake. His first volume deals only with phonology, and the second gets no farther than the nouns and their prefixes. But both volumes opened up as no other work has done, the fascinating interest of the Bantu languages to philologists, and even to the outside world. It is as easy to trace the affinities between one Bantu language and another as it is between those of the Aryan groups— perhaps easier. As a rule they are melodious languages with the phonology of Italian, and as easily pro- nounced. They are expressive, and compared to most other African speech families, comparatively simple and logical in their syntax. Bleek adopted an excellent system of orthography, one towards which we have somewhat reverted of late.

He made the best of the scanty material at his disposal, which was indeed small as compared with what we have before us at the present day. It consisted of Krapf's, Rebmann's, and Steere's East African vocabularies of Pojkcomo) (which he used very little), Kamba, Nika, Swahili, and ^ambala; Yao) (collected by Peters and Steere), Makua (Peters), Sena and Tete (Ci-nyungwi) ; of Tekeza (under which term he meant the Thonga or konga language of Delagoa Bay and Amatoiigaland), Sesutco, Sexlapi or Secuana; Zulu and Kafir (pcusa) ; of Yeye (Ba-yeiyi), with references to Rotse or Lui, Ta)ka or Toiiga, Nyefigto, Subia, Lujazi and Ma-ponda, all derived from Livingstone's MS. vocabularies ; * of Hercro, Ndofiga (Ovambco), Nan<o or Benguela, Angola or Bunda (properly, Ki-mbundu) ; of Kongto (Kipi-koiigio) ; Mpongwe, Kele (Dikele), Benga, Duala, Isubu, and Fernandian. Bleek's work, as already stated, did not get bej'ond the noun, its prefixes, and their concords. But it was of such an illuminating nature that its interest has prompted as many philologists to work for its completion as writers of fiction have been impelled to finish Edwin Drood. At any rate, the reading of his two slim volumes in 1883 inspired me with the determination to write this work now published, and to lose no opportunities for its preparation when my African travels or researches should enable me to collect material.

' In its fullest form with the preprefix it is Abaniu or Ba,ba.ntu. The singular is Unmntti (archai- cally, Gumuntu). The root is -ntu, apparently meaning no more than 'object' in its essence, and deriving its precise and differing interpretations from its prefixes : thus (discounting preprefixes) Mu-ntu = ' a human being ' ; Ba.-n/u = ' men, people ' ; Ki-ntu or Ka-ntii is ' a thing ', and Bi-ntii or Tu-ntu ' things ' ; Pd-ntu = ' a place ' ; Lu-tttit = ' a quality ' (of some kind) ; and Bu-ntu is ' humanity '.

• Which, scarcely used since, now contribute to the information of the present volume, through the kindness of lileek's ultimate successor, Dr. L. P^ringuey.

THE BANTU LANGUAGES 7

Bleek divined or actually exposed many of the facts connected with the origin and structure of the Bantu languages. But from the very insufficient material at his disposal he was led into several wrong deductions. One of these— evidently a pet theory of his — was in regard to the origin of a Zulu term for God — Cdmit-kulutikulu, which may (or may not) have been at one time the Zulu term for this concept. He realized that this word must mean either ' the great, great one ', or ' the old, old one ' (for -kulu can bear both meanings, even in the same language) ; and seeing the tendencies of Zulus to ancestor worship and the deification of dead-and-gone notabilities, he thought he found in this the etymology of the much more widely spread Bantu term for God —Mulungu, — which he considered to be a contraction oi Cdmii-kulunkulu. As a matter of fact, the two words have no connexion. Cdmu-ktdunkulu had the interpretation and origin given to it by Bleek, but Mulungu is really the rain or the sky god, and is derived from a root, -lungn, which is sometimes applied to the firmament and sometimes (though rarely) to rain (see numerous instances in the East and Central African vocabularies). Bleek's other error was in regard to the preprefixes or articles and the concord particles. He noticed that the vowels preposed to the Zulu-Kafir prefixes («mu-, aba-, «li-, &c.) were detachable from the second syllable, the prefix proper {niu-, ba-, or H-, &c.) : that the preprefixes had vanished from a good many of the Bantu tongues, but seemed to recur in the con- cord particles ; yet that these particles, though in general offering considerable resemblance to the prefix, in some cases exhibited a strange dissimilarity. Thus, for example, the concord particle of the Ma.- prefix might be sometimes nia- and yet at others ga-, ya-, or a-. The concord particle of Mil- might be gu-, yu-, or u-. Inasmuch, therefore, as we cannot postulate for the conjoint pre- fix and preprefix, a mere duplication of similar syllables (we can do so in regard to the majority in such forms as baba-, biibu-,kiki-, lulu-, Sic), and cannot assume a hypothetical tnumti- for the first and third, and mama- for the sixth, how are we to explain the dissimilarity between the concord particle and preprefix, and the prefix itself (often that between mh- and gu-, mi- and gi-, ma- and ga-) ? Bleek's attempts to do so in paragraphs 407-10, 461-5, and "536 of his second volume do not hit upon the right explanation. A step nearer in this direction was made by the Rev. F. W. Kolbe in his Language Study based on Bantu (London : 1888), and perhaps in earlier works dealing with the Herero language ; but it was reserved for the present writer to find the most reasonable solu- tion of this enigma, when he studied in 1901 the archaic Bantu languages of the West Elgon district (see his Uganda Protectorate, vol. ii, pp. 891-2).'

Bishop Edward Steere, who went out to East Africa as a member of the Universities' Mission in 1865, was a noteworthy pioneer in Bantu philology. He was the first to deal in a practical and intelligent manner with the Swahili language. His Handbook and Exercises were for twenty or more years after their first publication in 1875 the generally accepted means of acquiring that tongue, both for practical use and philological inquiry.^ Steere also was the first to set forth clearly the features of the Yaco language, of Ngindu), Dzalamoj, the Nyanyembe dialect of Nyamwezi, of ^ambala, and Makonde. In dealing with Ngindo) and Dzalamo) in 1869 he, at the same time, added a few words to our very slender knowledge of Ki-afigazija, the language of Great Komoro). This far-outlying ancient Zafigian ('Swahili') speech, together with the closely allied dialect of Anjuan, had, as already related, not only been one of the first written of the Bantu languages (by Thomas Herbert in 1626, in its Mwhila dialect), but had been studied by a Madagascar missionary named Elliott, and by Dr. Wilhelm Peters in the 'forties,^ and by Hildebrandt in the early 'seventies of the nineteenth century.

Livingstone's interest in African languages waned somewhat through the pressure of other

' See also p. 361 of the article on the Bantu Languages in vol. iii of the eleventh edition of the Ency- clopaedia Britannica.

' Later editions have been issued by Mr. A. C. Madan, M.A. ' His work is included in Bleek's Languages of Mozambique.

8 A HISTORY OF RESEARCH INTO

cares after his first great journey across the continent. His Zambezi expedition of 1858-63 resulted — strange to say— in absolutely no additions to our knowledge of the south-east African Bantu languages ; though it was eventually, through the planting of missionaries in Nyasaland, the cause of a great advance of Bantu study. Livingstone's last journeys (in which he discovered the great Lualaba-Congo, crossed and recrossed Taiiganyika, reached Lakes Mweru and Baiigweulu, and penetrated the Manyuema country into the heart of the Equatorial forest) might have helped philology but f6r the fact that the vocabularies and language notes which he made in the course of five years never saw the light. In common with all the rest of his scientific material they were put on one side by his literary executors and apparently have now been lost. Dr. W. A. Baikie, of Niger fame (another pioneer of British Empire whose scientific work met with insufficient appreciation from pietistic minds), collected much linguistic information dealing with the Semi-Bantu languages to the south of the River Benue, but very little of this material has been saved for our information.

In the middle of the 'seventies the Trans-African journey of Commander V. L. Cameron gave us (in all probability) the first definite illustration of the Lua (Luba) tongue, and enabled us better to understand old Portuguese records which suggested the continuity of Bantu speech across the continent. In 1877 the publication of H. M. Stanley's Through the Dark Continent conferred really g^eat benefits on students of the Bantu languages ; though I doubt if these were appreciated (except by Robert Needham Cust), for many years afterwards. Stanley indicated the Bantu character of the Victoria Nyanza and Tanganyika languages, and of those along the course of the Upper and Western Congo.

Meantime French and French Swiss Protestant missionaries and those of the London Missionary Society ' had been adding to our knowledge of the Becuana and Rofiga groups of tongues in South Africa, and the great German explorations of south-central Congoland had begun, together with the missionarj' settlements in Nyasaland. These last were to produce the brilliant language studies of Dr. Robert Laws, of David Rufl'ele-Scott, and Alexander Hetherwick, besides the work of Alexander Riddel, Dr. W. Elmslie, and J. A. Bain, and, farther to the east, the studies of the northern Makua dialects by Bishop (as he afterwards became) Chauncj' Maples.

Stanley's levin-stroke through the pitchy darkness of unknown Africa had started the opening up of the Congo from its estuary eastwards. George Grenfell had commenced his career as missionary-explorer in the Cameroons, and with his noteworthy colleagues, Thomas Comber and William Holman Bentley, was exploring western Congoland and commencing to note its languages. Kongo studies had been resumed, indeed, after a blank of nearly seventy years ^ by Dr. Adolf Bastian, who in 1875 published his work on the Luaiigco coast, which contains a good deal of interesting information regarding the western Congo and Gaboon languages.

In 1880 the Egyptologist C. R. Lepsius— the inventor likewise of the Standard Alphabet— ^rc- ceded his Nubische Crammatik by a famous Preface on the Languages of Africa, in which occurs his definition of Bantu characteristics. But this preface, though remarkable at the period at which it was written for its extended knowledge, is based on some erroneous conceptions due to the lack of material for comparison ; and Lepsius's definition of Bantu peculiarities no longer holds good, since some of these are found to be shared by other African language families.

When the last quarter of the nineteenth century opened, and with it began the tremendous move- ment of Europe on Africa, the time had arrived, evidently, in which some definite acknowledgement should be made of the interest offered by Negro languages io the world at large as a field of study.

' The Rev. John Brown's Secwana Dictionary (now in its enlarged 1895 edition) was first published about 1877.

' The last work on the speech of the estuarine Congo, published under the old Portuguese Roman Catholic missionary regime, was that of Father Cannecattim in 1809.

THE BANTU LANGUAGES 9

The task of tabulating all the extant information on this subject was undertaken by Robert Needham Cust, a retired Indian Civil Servant, who in 1883 published his Modern Languages of Africa. It does not pretend to be more than a history of linguistic research in Africa and a geographical enumer- ation and allocation of all the then known forms of speech in Africa from north to south, and east to west. But it is combined with the most extensive bibliography. Little attempt is made to describe the structure of these different types of speech, but evidence as to their relationships is sometimes adduced. Yet the work has been of the greatest help to students in many fields of African research, since its publication, and is still of value for the extent and correctness of its bibliography. It shows, in fact, a surprising amount of research, and personally I have seldom found any of its references wrong or misleading.

In 1882 an Italian professor of Palermo, Giacomo de Gregorio, issued a work at Turin entitled Cenni di Glottologia Bantu, which, however, was little more than an extended review of Bleek's published writings. In 1884 the present writer had in his book on the River Congo given a little new information regarding the Bantu languages of western Congoland. In 1885, however, in an appendix to his account of the Kilimanjaro) Expedition, he advanced slightly the development of Bantu philologj' by the information there given, and by one or two novel deductions drawn from the work of earlier writers. During the 'eighties and 'nineties some progress was made in the enumeration and study of the tongues of south-west Congoland and of eastern and southern Afigola, through the work of German explorers— A. E. Lux, Otto H. Schutt, Dr. Max Buchner— and of Portuguese expeditions under Serpa Pinto (1878-80) and Capello and Ivens (1878-85).

Between 1885 and 1888 the present writer investigated and wrote down ten or more of the Cameroons languages and several of the Semi-Bantu forms of speech on the border-line between the Cameroons and the Cross river region ; but little of this work has been published before its appear- ance in this book. During the same period Mr. J. T. Last, once of the Church Missionary Society, and later, and for long, a Government servant in Zanzibar, earned our gratitude by the printing of his Polyglotta Africana Orientalis, and by studies of the Kamba and Sagara languages. Mr. Last collected his information either direct from the natives of the different countries or by interrogating far-fetched slaves and porters in Arab caravans. The bulk of his work is surprisingly accurate, considering the difficulties of its transcription. He gave much new information for fixing the East African boundaries of the Bantu sphere.

Amongst the first Bantu scholars of German South-west Africa were the missionaries or the sons of missionaries, the Rev. Dr. C. H. Hahn, Theophilus Hahn, the Rev. J. Rath, G. Viehe, Rev. F. W. Kolbe, and the Rev. H. Brincker; whose studies were published between 1857 and 1891. The extremely interesting Herero language had been written down as early as 1857 by Dr. C. H. Hahn. Its characteristics were fully revealed by Rath, Kolbe, Brincker, and others before the German political occupation of that region.

In the 'seventies and 'eighties Primitive Methodist missionaries (coloured as well as white), such as William Luddington and Theophilus Parr, and Spanish Dominicans such as Juanola, did a great deal for the elucidation of the Bantu dialects of Fernando Po. Much of the work of the Primitive Methodists, however, has not yet been printed, but has been lent to the present writer for his assistance.

The 'nineties of the last century witnessed a great advance in Bantu studies. The languages of the Western and Upper Congo were taken in hand scientifically by the British Baptist missionaries —notably Dr. W. Holman Bentley, Dr. A. B. Sims,' the Revs. W. H. Stapleton- and George

' Dr. Sims did not belong to the British Baptist Mission, though in close relations of comradeship with its members.

' Stapleton's research work in regard to the tongues of the whole Congo basin might justly be described

10 A HISTORY OF RESEARCH INTO

Grenfell (though the work of this last was not revealed till after his death in 1906). Missionaries of the Congo-Balolo Mission— chiefly Messrs. Eddy, I. and F. T. McKittrick, did good work in con- nexion with the Lunkundu and Mongco dialects of the Lwlco language. To these workers in a field of exceptional interest must be added in the first decade of the twentieth century the Revs. John Whitehead, William Forfeitt, Lawson Forfeitt (at a later date), Robert Glennie, and John Weeks. Simultaneously, or earlier, the American missionaries were at work in southern Angola, where the great H^li Chatelain (of Swiss origin) dealt in a masterly way with the Mbundu language and kindred dialects, and where the Rev. Wesley M. Stover illustrated the Nana) or Umbundu of Bih6. In central Congoland, the Rev. W. M. Morrison, of the American Presbyterian Mission, gave us the first satisfactory exposition of the Luba or Lua language. George Grenfell had led the way (with Stapleton) in defining the northern limit of the Bantu family in the region of the Mubangi basin. Franz Thonner, a Viennese explorer, followed this up, and in 1899, as well as at later dates, revealed the existence of parcels of non-Bantu speech as far south as the northern banks of the Upper Congo. Stapleton, H. M.Stanley, William Forfeitt (especially), and Vice-Consul G. Babing- ton Michell, carried on this delimitation eastwards and showed us the existence of Sudanese tongues, like the so-called Bamanga, imbedded in areas of Bantu languages (Archdeacon Farler, of the Universities' Mission, and A. Downes Shaw, C.M.S., had earlier helped in the laying down of the northern Bantu limit in East Africa by recording the curious Mbugu of Usambara).

The agents of the London Missionary Society (notably the Revs. J. Griffiths (Ki-lega), David Picton Jones (Guha and Mambwe),and W. G. Robertson (Bemba)) had opened up to our knowledge the western and southern Tanganyika tongues, in which direction some additional or parallel information was obtained by the present writer in 1889. French, Belgian, Dutch, and German Roman Catholic missionaries, preceded by the explorers A. Marche and the Marquis de Compiegne in the COgwwe basin (early 'seventies), Monseigneur Le Berre, Fathers Delorme, Dahin, and Reeb, Lejeune, Alex. Visseq, Butaye, 6m. Cambier, and, above all, De Clercq; ' Gustave de Beerst, P. H. Trilles, Eugene Hurel, P. Schumacher, I.M.M. van der Burgt, and A. Capus and J. Calloc'h— were putting down on paper at the same time and thenceforward into the twentieth century the languages of the Gaboon, of north-west and south-central Congoland, and the region between Tafiganyika and the Victoria Nyanza. Paul Kollmann, a German official in East Africa, in 1899 published useful vocabularies of some of the south Victoria Nyanza languages, which were treated more scientifi- cally at an earlier or a later date by Peres Eugfene Hurel, Loupias, Dufays, Schumacher, Alex. Arnoux, Menard, and A. Capus, and by Hermann Rehse and Hauptmann C. Hermann. I should also mention with due acknowledgements the services of Messrs. J. F. Cunningham, J. A. Meldon, S. Ormsby, and G. A- S. Northcote, of the Uganda Administration, who, between 1899 and 1909, worked at the recording of Victoria Nyanza and Albert Nyanza dialects as specified in the Bibliography following my vocabularies. Research of value amongst the (still little-known) tongues of south-east Africa was being done in the 'nineties and in the early twentieth century by the Swiss or French Protestant missionaries, Paul Berthoud (who dealt with the ^i-gwamba of the north-east Transvaal), E. Jacottet (who also gave us our first scientific account of Subia and Luyi, two languages of the Upper Zambezi), Henri Junod (^i-rofiga and its dialects and folk-lore) ; and by Bishop Smyth and John Matthews (pilefige or Ciopi), the Revs. Theodor and Paul Schwellnuss (Venda), A. H.

as enormous ; only about a quarter has seen publication. A go,od deal of his work is incorporated in the present book. The remainder, unfortunately, is scarcely utilizable owing to accidents which attended it after his death, as we lack the English elucidation of the vocabularies, portions of which also are still missing.

' Bantu philology owes much to the labours of P^re Auguste De Clercq in south-central Congoland (the Luba and Kanyoika languages especially), and to P6re J. Calloc'h for his work on the Ifumu or eastern Teke language and the Sudanese tongues of the western and north-western basin of the Mubangi.

THE BANTU LANGUAGES n

Hartmann (Mapuna, Karafia), and W. A. Elliott, L.M.S. (Mapuna and pindebele). In northern and eastern Zambezia' good work was being done ere the close of the nineteenth century by Father J. Torrend, S.J., in regard to the Ci-tofiga of the Central Zambezi and Cuabco of the Zambezi delta. In the opening years of the twentieth century Edwin Smith gave a distinct uplift to Bantu studies by the publication of his Handbook of the lla (^ukulunibwe) Speech. The hnguistic work of the Universities' Mission between 1890 and 1906 must not pass unacknowledged. In addition to the remarkable contributions to Bantu hterature made by Mr. A. C. Madan, at one time a lay member of the Mission (contributions elsewhere enumerated), Archdeacon H. W. Woodward, and the Rev. W. H. Kisbey added to our knowledge of Zigula (Zanzibar coast), and the Rev. Herbert Barnes illustrated the Nyanja language of Lake Nyasa in its northern form.

As regards the actual work of Bleek — the general survey of the Bantu tongues — the first forward step since Bleek's time was taken by Father J. Torrend, S.J., in his Comparative Grammar of the South African Bantu Languages, originally published in 1891. But the most noteworthy advance on Bleek was made by a fellow German, Carl Meinhof, a Lutheran missionary, who had probably studied the Bantu languages for some time before his first noteworthy publication, Grundriss einer Lautlehre der Bantusprachen, published at Leipzig in 1899, and added to subsequently by his Grundziige einer vergleichenden Grammatik der Bantusprachen ('Materials for a Comparative Grammar of the Bantu Languages '), besides many articles in German linguistic periodicals, which are referred to by me elsewhere in this work. One to whom all Bantu students must remain deeply indebted is A. Seidel, a great German philologist, whose studies of African languages have gone far outside the Bantu field (as have those of Meinhof), but who, from the early 'nineties onwards, has given to the world a series of admirable grammars or grammatical sketches, of dictionaries and vocabularies of east, south-west, and north-west Bantu languages.

Nor in any record of Bantu discovery should the modern Portuguese be left out of account. Serpa Pinto, Capello, and Ivens, in their works of the 'seventies and 'eighties of the last century had, as already related, given vocabularies ofAfigola and north-west Zambezia Bantu languages which have been of assistance to investigators; but in the 'nineties, Henrique de Carvalho pub- hshed that really important study of the Lunda language which is referred to in the Bibliography ; and various Portuguese missionaries (especially Ernesto Lecomte and Affonso Maria-Lang) sub- sequently wrote excellent accounts of the northern members of the Herero group in southern Angola (Umbundu and Nyaneka).

Bishop Steere had a noteworthy linguistic successor as philologist in a lay member of the Universities' Mission, Mr. A. C. Madan, who, beginning in the 'nineties with the illustration of Swahili, has brought out during the last few years works of considerable value, not only of specific languages such as those of Northern Rhodesia (Seiiga, Bisa, Lala-Lamba, Lenje), but on the structure and phonology of the Bantu languages in general.

C. W. Hobley, almost the oldest now in seniority of service amongst the British East African officials, had in 1899 discovered the northernmost colony of the Bantu in eastern Africa — the Konde orHasa people of North-west Elgon, and had published a short vocabulary of their language in the Journal of the Royal Anthropological Institute. In later years he contributed much information regarding the speech of the Kavirondco Bantu and of the Basuba (Kcosoava), as well as the Akamba of British East Africa.

The Luganda language " received its first efficient treatment by a member of the Church

' Into northernmost Zambezia (N. Rhodesia) the widespread Luba language enters, and the Bantu student was at one time mainly indebted to the missionary C. A. Swan for his knowledge of the southern Luba speech. Mr. Swan published in 1892 Notes on the Construction of Chiluba. But since then (1909) we have had the Kiltiba (Sangd) Dictionary of fimile Jenniges.

'^ This Sanskrit of the liantu was, I believe, first put on paper by Stanley in the vocabularies of his

la A HISTORY OF RESEARCH INTO

Missionary Society, P. J. Pilicington, who was killed in 1898 in the mutiny of the Sudanese soldiers. Other members of the Church Missionary Society, the Revs. Blackledge, O'Flaherty, Walker, Hattersley, Purvis, H. E. Madox, and Crabtree,' and Dr. A. R. Cook, from this time onwards pub- lished works on Luganda, Ru-nyono, and on the interesting Masaba speech of West Elgon (originally discovered by Hobley). Mrs. Hildegarde Hinde wrote on the Kikuyu and Kamba languages. A. C. Mollis, Secretary to the British East African Administration during the early years of the twentieth century, materially helped our comprehension of the different dialects of Swahili by illustrating them in the Journal of the African Society. Swahili in the later years of the nineteenth and the commencement of the twentieth century had received additional treat- ment (additional to the still useful works of Steere and Madan) at the hands of Pere Sacleux and other French missionaries. Dr. G. C. BQttner, Carl Meinhof, A. Seidel (German philologists), and the Rev. W. E. Taylor, of the Church Missionary Society, the last-named being a leading authority on Swahili. [An excellent description of the Giryama dialect of the Nika language had at an earlier date been published by the same writer (W. E. Taylor).] An adjoining language, the ^ambala, received in 191 1 the amplest treatment from the point of view of scientific philology by the German missionary, Karl Roehl. In fact, with the uprise of the German power in East Africa went hand in hand an investigation of the East African languages, chiefly, but not always, by missionaries Catholic and Protestant : C. Velten (Nyamwezi, Kami, Hehe, Swahili), Ferdinand WOrtz (Pwkojmoj and Tikuu), Ernst Brutzer (Kamba), A. Worms and M. Klamroth (Dzalamco), Bernhard Krumm (Matumbi), Fathers Cassian Spiss and Johannes Hafliger (Sutu and Mateiigco), R. Wolif (Kinga), C. Schumann (Nkonde), Dr. Otto Dempwolff'and Dr. Eduard Sachau (Pcogojru)).

About the middle of the 'nineties, Miss Alice Werner commenced those studies of the Bantu languages which have made her subsequently so noteworthy, the more so as she writes with equal facility in German and French, as well as English, and has therefore contributed some of her most remarkable essays to French and German periodicals. Miss Werner's work in its critical insight into the structure of a language recalls that of Carl Meinhof, and of an equally great Bantu scholar, Bernhard Struck, whose illuminating articles have appeared at intervals in German and English publications during the last ten years. The Rev. W. A. Crabtree has also attracted attention as one able to deal with the Bantu languages from the general point of view of scientific philology by his essays in Xhe Journal of the African Society; while from Paris have recently been issued several notable studies on the Bantu languages by Mile L. Homburger. The most important of these is her i.tude sur la Phomtique historique du Bantou. Another interesting French contribution to Bantu literature is from the pen of Captain Avelot. It deals with the little-known OOgcowe and Gaboon languages, and is entitled Recherches sur I'histoire des migrations dans le bassin de VOgoue (see Bibliography).

In the South African field there are curious gaps in our knowledge, and much remains to be done here. Unhappily, whilst we are waiting for it to be done languages and dialects are fading out of existence in the whirlpool of race movements which is going on. The most interesting of all the South African languages— Venda of the Northern Transvaal — had still a most imperfect vocabularj- until the Rev. T. Schwellnuss amplified it in a private communication for my information. The Kafir-Zulu dialects are insufficiently known, though thanks to the kindness of the Hon. R. Coryndon of Swaziland, of Lady (Florence) Phillips and their friends, the Swazi dialect has been set

Through the Dark Continent, published in 1878. Probably Burton and Speke had first shown that it was a Bantu language by their works published in 1859 and 1864. Its general character and affinities were not however made clear till the publication in 1882 of the Outline Grammar of the Luganda Language of the Rev. C. T. Wilson, C.M.S.

' For details, see Bibliography. The Rev. W. A. Crabtree has worked on the Masaba dialects, in addition to Luganda, but his material, if it has been printed, has not come under my notice.

THE BANTU LANGUAGES 13

forth in writing for the purposes of this book. The Zulu language has at last been fully exemplified in the Zulu-English Dictionary of A. T. Bryant ; a Kaftr {pusa)-English Dictionary was published at Lovedale Mission in 1899 by the Rev. Albert Kropf; an excellent Kafir (pwsa) Grammar by I. McLaren in 1906 ; and an exceptionally remarkable WOrterbuch der Sot/to- Sprac/ie {Siid-Afrika), by K. Endemann (Hamburg, 1911), gives us a full illustration of the Sutco or Sothco speech.

Emin Pasha, when Governor of Equatoria, compiled a number of vocabularies, which, however, did not do more than assist to limit on the north the extent of the Bantu field ; but when he returned to take service in German East Africa he inscribed some interesting vocabularies of the Nyanza and Ituri forest languages, which after long concealment were brought to light and published recently through the energy of Bernhard Struck. In this connexion, though of much earlier date, may be mentioned the vocabularies published by Stanley in his Darkest Africa, the importance of which — scanty as they are— was not appreciated till within the last few years. Honourable mention should also be made of the work of Father Auguste van Acker in his Dictionnaire Kitabwa-Fran^ais ei Franfais-Kitabwa, published by the Congo Museum, Tervueren ; and the sumptuously produced Notes ethnographiques . . . sur les Bushongo, &c., by Emil Torday and T. Athol Joyce, which includes much linguistic information. Mr. Emil Torday, within the last fifteen years, has worked hard to place before the world the remarkable culture of south-central Congoland, and incidentally has collected almost the only information we possess regarding the curious Bantu languages north-east and north-west of the Luba field, besides illustrating several dialects of Luba. His researches have even extended as far north as the Ababua language group on the northern Bantu frontier. His work, published and unpublished, appears in the vocabularies of the present book.

German work of recent years in the Kamerun Protectorate has scarcely resulted in those great gains to philological knowledge we had hoped for, seeing what has been accomplished in East and South-west Africa. The most splendid monument to Teutonic activity in this quarter lies almost outside the Kamerun region — the Monograph of Giinter Tessmann on the Fang people and their languages. Die Pangwe is deserving of international recognition (in common with a work by Dr. Leonhard Schulze, of earlier publication, on Namakwaland). Carl Meinhof, twenty years ago, collected and published much interesting information on the Bantu languages spoken in the basin of the Cameroons (Wuri) river (the West Kamerun or Duala group, as it might be called). Father Haarpaintner gave the first accurate account of the Yaunde speech (northern Pafiwe) ; Fathers G. A. Adams, P. H. Skolaster, H. Nekes, E. Schuler, H. Dorsch, and Captain Lessner have contributed valuable information about the tongues of the West Kamerun group between 1898 and 1910. In addition to their published articles, I am personally indebted to the English naturalist, Mr. Geo. L. Bates, for six valuable vocabularies of the leading Pariwe or Fang dialects, and for the only vocabulary extant of the Njiem tongue, which in some way connects the Pafiwe group with the corrupt Bantu languages of the Kadei-Lobai region. As regards these outlying jargons of the Upper Safiga, the Kadei, and the western bend of the great Mubangi river, our sole information comes at present from Mr. Geo. L. Bates, from the German explorer, Strumpell, and from Dr. Outilleau, a French medical officer who explored in 1910-11 the ethnology of the Sanga and Mubaiigi basins.

Dr. Freiherr von Danckelmann and Dr. E. Zintgraff, in common with Captain Lessner, already mentioned, contributed much exact geography in map-making to enable us to locate Bantu tribes and languages in the western Kamerun, especially the invaluable vocabularies of Koelle. It is largely thanks to them and to Lieutenant Hutter (who with Dr. Hoesemann travelled over the northern versant of the Cameroons river basin in 1900-4) that we have been able to fix with some precision the boundary in West Central Africa between Bantu and Semi-Bantu. Several French officials, notably the administrator Clozel, the explorer-botanist Auguste Chevalier, and the far- travelled and industrious PereCalloc'h have thrown some light on the structure of these Semi-Bantu

14 A HISTORY OF RESEARCH INTO THE BANTU LANGUAGES

languages, south-west of the Central Sudan, between the Mubangi basin and the water-parting of the Kamerun coast-lands ; and Pfere Calloc'h and Captain M. R. Avelot have added to our know- ledge of the languages spoken in French Congo and the Gaboon. German explorers have dealt (slightly) with the interesting Semi-Bantu Bali group, between the Cameroons river and the Upper Cross river basin ; the English administrator and ethnologist, P. Amaury Talbot, has furnished us with full information regarding the ' almost- Bantu ' tribes (chiefly the Ekoi) between the Old Calabar region and the Rumbi mountains, information which has been supplemented recently by Mr. North- cote Thomas's Specimens of Languages from Southern Nigeria. [In this region also the author of this book travelled extensively some twenty-six years ago, to collect vocabularies for this Comparative Study of the Bantu Languages.]

In concluding this historj' of the rise and progress of research into the Bantu speech family, I may perhaps be permitted to summarize my own credentials as a humble follower of the great philologists to whose memory my book is dedicated. It was the writings of Richard Burton, more especially, which first rivetted my attention on this subject. I knew something in general about the Bantu languages before I set foot in Bantu Africa in the year 1882. In the spring of that year I accompanied an exploring expedition under the Earl of Mayo, which was to investigate Portuguese South-west Africa. From Aiigola I proceeded to the River Congo, and with the help of H. M. Stanley penetrated into the then almost unknown region of the Upper Congo, transcribing languages in preparation for a continuance of Bleak's unfinished Grammar. In 1884 I learnt more or less thoroughly the Swahili language at Zanzibar, and shortly afterwards studied the dialects of East Africa and Kilimanjaro. Between 1885 and 1888 I was Vice-Consul in the Kamerun and Acting Consul in the adjoining regions of Southern Nigeria, and for three years gave myself up with ever- growing interest to linguistic studies in those directions, much of the results being privately printed by the Foreign Office. Between 1889 and 1896 I wrote down vocabularies and grammatical notes of the leading languages and dialects of eastern Zambezia, Nyasaland, and south Tafiganyika; and in two visits to South Africa made some acquaintance with piosa and Zulu. From 1899 to 1901 I worked in situ at the Bantu languages of the Uganda Protectorate and the adjoining Congo forest. In 1904-6 and 1907 I took advantage of visits to West Africa to obtain (through correspondents or from travelling natives) additional information regarding Kamerun or Luafigo) languages. I had been asked to contribute the first article on the Bantu languages to the Encyclopaedia Britannica (tenth edition, 1903), and rewrote and enlarged that article for the eleventh edition. To further my Bantu studies, I paid several visits to Germany (the goal of all students of Africa before this unhappy war broke out). But in the pursuit of my investigation of the Bantu languages I have derived nearly as much information from Paris, Brussels, and Lisbon, as from Berlin, Hamburg, and Stuttgart. I should mention also that a good deal of my work has been checked in course of preparation by reference to natives of various parts of Bantu Africa, who, when coming to England, have been good enough to visit me and supply me with information and criticism. Many of the vocabularies here printed, though compiled (in some cases before I was born) by other travellers, have thus been tested by me in England, as well as locally in Africa, and have received in conse- quence certain corrections, additions, and explanations.

In addition to the Bantu family, I have made some research into the North African dialects of Arabic and the Berber spoken in Southern Tunis, into Gala and Somali, the Nilotic and Masai tongues of Equatorial East Africa, the languages of the Central Sudan, of Southern Nigeria, of the Kru coast, and of the Mandifigiu peoples.

CHAPTER II

THE DISTRIBUTION AND CHARACTERISTICS OF THE BANTU LANGUAGES

What are the Bantu languages ? And why should they have been preferred as a special subject of interest in philological research to a degree far exceeding that of other language families of purely African location ? They constitute a very distinct type of speech which, as contrasted with others amongst the groups of negro tongues, is remarkable as a rule for the Italian melo- diousness, simplicity, and frequency of its vowel sounds, and the comparative ease with which its exemplars can be acquired and spoken by Europeans. The Bantu languages are attractive to the explorer not only from the harmonious adjustment of vowels and consonants, but from the logic of their grammatical structure, which, in the majority of these tongues, provides for a wide range and a nice discrimination in the expression of ideas.

This one negro language family now covers the whole of the southern third of Africa, with the exception of very small areas in the south-west (still inhabited sparsely by Hottentot and Bushman tribes) and a few patches of the inner Congo basin. The Hottentot-Bushman region at the present day is limited to Great Namakwaland and the adjoining districts of Cape Colony and the western part of the Kalahari Desert, Kacokcoland, south of the River Kunene, and undefined portions of the plains between the Upper Kunene and the Okavaiigo). With this exception, and the now Europeanized portion of Cape Colony, Bantu Africa extends from Port Elizabeth and Kaffraria on the south to Fernando Pd and the vicinity of Old Calabar on the north-west, and to Lamu and the Tana river on the north-east. There are a few small enclaves of non-Bantu-speaking negroes on the north banks of the northern Congo, and patches of Hamitic, Nilotic, or unclassified tongues between the south-east corner of the Victoria Nyanza and Usambara in East Africa. But for these relics of pre-existing or invasive tongues, it might be said that the whole of this southern third of Africa contained but one indigenous speech family, the BANTU, and that it was only necessary to define the northern boundary of its range.^

The northern boundary of the Bantu field is still a little uncertain and not easy to delineate geographically. It may be said to start on the west coast of Africa in the Bight of Biafra (due north of the island of Fernando Po), at the mouth of the Rio del Rey in the southern portion of the Bakasi peninsula, which flanks the estuary of the Old Calabar river. From the eastern bank of the Rio del Rey the boundary is carried to the Ndiaii river, and thence with zigzags to the western flanks of the Manefiguba mountains of western Kamerun ; then to the junction of the Sanaga and Mbam rivers and eastwards to the Lower Kadei, and down stream to its junction with the Safiga ; from this point again eastwards to the Lobai river and down that stream to near its junction with the Mubangi. Then the border-line retreats west and south again, but once more makes for the Mubangi

' VVe may suspect that in the Abongoi or Akwa of the Gaboon and the other Pygmy forest negroes of the Congo basin we have vestiges of peoples only recently converted to a corrupt form of Bantu speech ; but their jargon is nevertheless of Bantu type.

i6 THE DISTRIBUTION AND CHARACTERISTICS

river in its lower course, crossing it above Ibefiga and extending eastwards to the Lower Moiigala, which it reaches at Lilcimi. Allowing for the enclave of the Sudanic Bondunga or Ndoiiga people who reach to the banks of the northern Congo at Lisala (near Bwpiotio), the Bantu border-line may be said to follow the water-parting between the Moiigala-Dua basin and that of the Ebaila, and then to cut across— east-north-eastward — to the Wele-Mubangi, near the falls of Bango) and the post of Jabir. Then another loop southward across the Likati and back to the Wele, after which the border-line declines south-eastwards to the Upper Aruwimi, below the confluence of the Nep<oka). Thence in little zigzags to the Lindi river, and abruptly west through the dense Congo forest back to the Congo main stream below the Lindi confluence (some distance west of Stanleyville). Next, the Congo is followed up stream past Stanleyville ; and then with a western loop for the Sudanic Bamafiga the Bantu frontier crosses the Lualaba-Congo once more and serpentines eastward to the extreme Upper Ituri and the south-west coast of the Albert Nyanza. The east coast of the Albert Nyanza is all Bantu as far north as Maguiigco ; and thence, in a south-easterly direction, the frontier keeps close to the west bank of the Victoria Nile till it enters the Busioga country near the northern shores of Lake Victoria. Crossing the Sito river, the Bantu boundary is carried (with an eastern loop) to the north-west slopes of Mount Elgon. From this point the boundary goes south-east and then abruptly south-west to the entrance of the Nzoia river into the Victoria Nyanza. The north-eastern coast-lands of the Victoria Nyanza are excluded from the Bantu sphere, and are populated by large colonies of the Gej^a or Ja-Iuco Nilotic negroes ; but from the southern entrance to Kavirondw Bay the frontier between Bantu and non-Bantu is carried south-east and south and is determined mainly by the rise of the lofty Equatorial plateaus. A great wedge of non-Bantu peoples comes from east and north, and touches the Victoria Nyanza at the head of Speke Gulf. From this point the Bantu border- line passes southwards in zigzags to the Luwambere river which flows into Lake Eyasi (throwing off sastward the Uraiigi peninsula of Bantu speech) and skirting Ugfogco attains to the Usagara and Nguru hills, thence curling east and north to the Rufu river. Crossing the Rufu the northern boundary of Bantu Africa encloses a north-eastern projection to the western, southern, and eastern flanks of Kilimanjaro). From eastern Kilimanjaro) the boundary-line returns coastwards, then once more describes a huge north-western loop up the valley of the Athi river to the Kikuyu country, and north again to beyond Mount Kenya and the sources of the Tana river. Along the Equatorial coast of the Indian Ocean there is a narrow strip of Bantu territory northward which includes the Swahili settlements of Witu, Lamu, and Patta,' and passes up the valley of the Tana river for a considerable distance. There are runaway, Swahili-speaking slaves on the Juba river, but they do not represent any ancient extension of Bantu speech in that direction. Yet it is evident that Equatorial Africa was at one time almost continuously Bantu in speech (save for the enclaves of Hamitic or Sudanic tongues such as MbuluBge and Mbugu or the click-using Useria and Sandawi) from the east coast of the Victoria Nyanza to the Indian Ocean ; and that into this Bantu sphere there came, several centuries ago, hordes of invading Nilotic Masai, Dorcobo), Nandi, and Ja-luo). These invasions reduced the area of the Bantu languages in Equatorial East Africa to islands and peninsulas.

The islands of Zanzibar and Pemba are, of course, Bantu ; Zanzibar being the metropolis of

' South of this eastern boundary-line between the Victoria Nyanza and the Indian Ocean— a line which terminates on the sea coast at Kiuiiga in about i° 45' South latitude— there are several islands of Bantu speech— notably the Sunyu on Lake Natron, the Mbugwe on Lake ^L'^nyara, the Wa-meru (Caga) on Mount .Meru. Within the eastern Bantu area there are small enclaves of non- Bantu speech —TAtU^A (Nandi-Nilotic) on Speke Gulf and in eastern Unyamwezi ; Nege, Mbitlunge, and Wasi (Kushitic-Hamite) ; Sandawi (resembling JJushman) and UsMa ( IVangoinwia), between Unyamwezi andUgmgoi; and the perhaps ' Sudanic ' Mbugu in a small enclave in'northern Usambara. Besides, of course, the roving Masai and Donobci). A good deal of light is thrown on this complex ethnographical question in Ueber die Sprachen der Tatoga-und Iraknleute, "by Bernhard Struck, Berlin, 191 1. [In Honderabdruck aus ' Das Hochland tier RiesenkraUr ', by Ur. Fritz Jaeger.]

OF THE BANTU LANGUAGES 17

the great Swahili language. The Komoro) islands — at any rate the Grand Komorto, Mtohila, and Johanna— come within the Bantu sphere, as their people speak ancient dialects of Swahili.

The importance of this language family to Africa at the present day and in the near future is unquestionable. Throughout Africa, north of the Bantu border-line, you meet with a bewildering number of languages widely differing from the other and mutually quite incomprehensible ; and although there may exist linguae /rancae of Arabic (in its different dialects), of Pigeon-English, Creole- French, Hausa, or Portuguese, linguistic difficulties are relatively enormous, and have long delayed missionaries, traders, and representatives of Government from getting into close touch with the indigenous tribes. Moreover, most of these African languages, north of the Bantu field, are exceed- ingly difficult to pronounce and to acquire, whereas the leading features of the Bantu tongues are an Italian-like phonology and a relative simplicity of structure. Further, when once one Bantu language is acquired it is not very difficult to understand the structure and even the vocabulary of others. Assuming that a European knows nothing but Kipi-kongu) on the west or Swahili on the east, he can go far and wide over Bantu Africa with either of these tongues to help him, so universal in form are many of the word-roots indicating the simple concepts of everyday life. One could probably select twenty-five words of everyday intercourse which were common (or almost identical in form) to all the leading Bantu languages of the Cameroons, Congo, Angola, British South Africa, Portuguese East Africa, German East Africa, and Uganda. If, say, one started (as many an early pioneer like Livingstone did) with a band of porters or canoemen from the Zambezi, one might go far and wide to the north, west, and east, and with the slow rates of travel in pre-railway days, find one's porters accustoming themselves by degrees to the changes of dialect and language, till at length the caravan arrived at Tafiganyika, the Victoria Nyanza, or the Upper Congo, with its men still able to ask their way, or to understand the drift of what was said to them. Who that in earlier days explored Equatorial East Africa or Equatorial West Africa, or the very heart of the continent, has not realized the shock and interruption in easy relations when his expedition at last left Bantu Africa and arrived in a district where the languages belonged to another family — Nilotic, Sudanic, or Nigerian ? How completely at a loss he and his men felt, just as one who had more or less grasped the principles of the Romance languages of the present day, with no previous know- ledge of German, would feel on entering Alsace-Lorraine or southern Germany from France or Switzerland. Though, ol course, the differences would be far greater, since German, after all, is one of the many Aryan tongues and has an easily recognized community of origin and interrelationship with the Romance languages.

There is no mistaking a Bantu language for a member of any other African speech family. A momentary glance at the numerals, at a dozen word-roots with their prefixes or suffixes, deter- mines the fact whether it is or is not a member of the Bantu family. The phonology also is as a rule distinctive, though appearances may be deceptive in the case of a few languages of the north- western part of the Bantu field. The semi-Bantu languages on this north-west borderland have a vocabulary which contains a greater or smaller amount of Bantu roots, and farther north and west there are other language families which display obvious resemblances and affinities with what may have been the Bantu mother tongue ; but outside the Bantu family there is no known speech group in Africa which displays all the characteristic features of Bantu word-construction and syntax and at the same time shows unmistakable affinity in word-roots. Curiously enough, there are languages in southern Kordofan, in Nigeria, at the back of the Gold Coast, or in the Sierra Leone region, the syntax or construction of which frequently recalls the Bantu idiosyncrasy ; but the word-roots of the vocabulary would be found wholly dissimilar. Or there are others, again, in West Central

17S« C

i8 THE DISTRIBUTION AND CHARACTERISTICS

Africa that exhibit a decided likeness to Bantu in their word-roots, j'et in syntax and word- construction are quite unlike the Bantu.

The following propositions may be laid down to define the special or peculiar features of the Bantu languages :

(i) The vowels of the Bantu languages are always of the Italian type, and no pure Bantu lan- guage includes obscure sounds like 6, a, a, and m. Each syllable must end in a vowel, though in some modern dialects in eastern and western Equatorial Africa, in Congoland, and South Africa, the terminal vowel may be elided in rapid pronunciation or may be dropped, or absorbed in the terminal consonant, generally in such cases a guttural or a nasal. No two consonants can come together without an intervening vowel, except one of them be an aspirate or a nasal (m, n, »), and no consonant is doubled in pronunciation. Apparent exceptions occur to this last rule where two nasals, two r's, two d's, or two ^s, come together through the elision of a vowel or a labial, or where the ordinary aspirate, nasal, or dental is reinforced so that it sounds like a double //, «, or /.

(2) They are agglutinative ' in their construction, the meaning of the root-word being defined or enlarged and the syntax formed by the addition principally of prefixes, and secondarily of suffixes added to the root; but no infixes (that -is to say, no mutable syllable) is incorporated into the middle of the root-word.

(3) The root, excepting very rarely its initial vowel or consonant or its terminal vowel, is practically unchanging ; though its first or penultimate vowel or consonant may be modified in pronunciation by the preceding prefix, or the last vowel of the root similarly, by the succeeding suffix. In the great majority of cases, however, the root-word remains absolutely unchanged through all syntaxial combinations. It is frequently monosyllabic and very seldom of more than two syllables in its original and underivative form. But it is impossible to trace back Bantu con- struction to a monosyllabic basis, as some writers have attempted to do.

(4) Substantives are divided into classes or genders indicated by the pronominal particle (' prefix ') which precedes the root of the fully-expressed noun.' -These prefixes in an ordinary way have attributed to them, definitely, either a singular or a plural sense. With the exception of the abstract prefix Bu- (No. 14), no singular prefix can be used as a plural ; and although the and (Bd-), the 6th (Ma-), the 8th (Bi-), and the 12th (Tti-) are in some languages used in an honorific or collective sense, which causes them even to be applied like prefixes of the singular number, this use is easily traced down from a plural origin. There is a certain degree of correspondence between the singular and plural prefixes. Thus, No. i prefix (Mu-) takes invariably as a plural the 2nd Bd- (Wa-, Va,-, A-), though Ba.- may also in the less orthodox languages serve as plural to the singular of No. 9, or to nouns which have not been endowed with or have dropped their prefix. No. 4 (Mi-) nearly always serves No. 3 (Mu-) as a plural, though it may also be used with No. 14. No. 8 (Bi-) is practically the only plural of No. 7 (Ki-). The number of prefixes common to the whole family is perhaps seventeen, though there may be two other classes, Nos. 18 and 19 (Gu-, sing., Ga-, plural) which, if they cannot be shown to be ancient variants of Nos. 3 and 6, would raise the total number to nineteen. These seventeen or nineteen pronominal particles, three of which (15, 16, and 17)

' ' Agglutinative ' is a classifying term which has lost its value ; for all languages are agglutinative in the sense that as a rule when a concept increases in complexity the word used for it enlarges in volume, the language being equally agglutinative whether the added syllables are tacked on to either end of the word or inserted in the middle. Even inflexions were originally due to additional syllables which .iffected and modified the root vowels by their proximity. To some extent, indeed, the Bantu languages are not strangers to inflexion. For instance, verbal roots may become ' inflected ' to express changes of tense, mood, and meaning.

' In numerous Bantu languages certain classes of prefix may fall into desuetude and the root-word appear without a class-indicator. But in such case the noun's class is always present to the native's mind, and is indicated by its pronoun and its concord or adjectival particle. Class prefixes in Bantu, after all, are analogous to our ' class prefixes ' ' he ' and ' she ' in ' he-goat ', ' she-bear '.

OF THE BANTU LANGUAGES 19

often decay into mere prepositions, are attached as prefixes to the roots of the nouns and verbs. Other monosyllables may be used as qualifying prefixes before noun-roots— chiefly in a masculine or feminine or diminutive (' father ' or ' mother ' or ' child ') sense ; but they have no ' concord ', no distinctive pronominal particle.

(5) Bantu languages are especially characterized by the principle of the concord. In close con- nexion with the prefixes are similar monosyllables associated with each class of noun, which are prefixed to the adjective or numeral in correspondence with the noun, or serve the noun in the sentence as pronoun or as pronominal particle in the verb. They combine with a vowel (most commonly -a) to form a genitive copula ; and as an identifying suffix or prefix fuse with any interro- gative, relative, collective, or demonstrative pronoun, with adverbs or prepositions which refer to the noun they represent. The forms of these concord-particles or pronouns are usually like the prefix with which they are associated ; yet there is occasionally a dissimilarity showing diversity of origin.' From these concord-particles seem to have arisen, early in the development of the Bantu languages, ' preprefixes ', or additionally demonstrative articles (answering somewhat to the English 'the'). These, when definiteness of utterance was desired, preceded the actual prefix.' In the most archaic forms of Bantu speech these preprefixes are syllables of consonant and vowel identical with the pronominal concord : thus the full form of the conjoined preprefix and prefix in Class i of the north-west Elgon dialects is Gii-mu, Gu- being the preprefix oi Mm- in both Classes i and 3. Ba-ba is the full form of the 2nd Class, and Ga-ma. of the 6th Class. But in all but the most archaic types of Bantu language, the prefix, if retained at all, is a mere vowel— i/-mu-, ^-ba-, A-ma-, /-mi.

(6) With the aid of these pronominal and adjectival particles a complete ' concord ' is main- tained between the noun and its dependent adjectives, numerals, demonstratives, pronouns, verbs, prepositions, and (sometimes) also adverbs.

(7) Suffixes of an adjectival, locative, or prepositional sense may also be applied to nouns, and, of an adverbial and postpositional quality, to verb-roots.

(8) No sexual gender is directly recognized in the classes or categories represented by the nineteen prefixes with concord particles, or by the personal pronouns. These last make no distinction between ' he ' and ' she ', ' they ' (feminine), and ' they ' (masculine). But a distinction of sex may be conveyed to nouns and adjectives by sex-prefixes and suffixes (usually Na-, Nya-j Se-, Si-, Sa- ; -kazi and -rume). Of the seventeen or nineteen categories of nouns, the prefixes and con- cords of Classes i and 2 are chiefly reserved for human beings. Classes 3 and 4 have a great deal to do with trees, vegetables, and implements (especially of wood), and members of the body. Class 5 comprises many round things like fruits, eggs, stones, the sun, the eye, a tooth, a horn ; and is also associated with gigantic objects and augmentatives. Class 6, its plural, has a collective sense and refers to liquids, assemblages of people, animals, and inanimate objects; Class 7 (plur. No. 8) indicates tools, furniture, methods, 'kinds of, 'manner of doing things or of speaking'; Classes 9 and 10 were often associated with beasts, birds, reptiles, and fish ; Class 11 {Lu- or Du-) with ' long things ', rivers, tongues and languages, long stretches of country or ranges of mountains. Classes 12 and 13 were usually diminutives, but Class 13 (Ka-) could also be given a specially feminine, tender, and finally honorific sense. Class 14 {Bn-) was generally the prefix of abstract qualities, answering to our English suffix ' -ness' ; Class 15 was applied specially to certain

' Thus, in many Bantu languages the concord-particle of the ist, 3rd, 4th, 6th, 9th, and loth prefixes, when it is used in a pronominal rather than adjectival sense, differs totally from the prefix. The 1st prefix is Mu- ; its concord-particle when adjectival is niu-, so also may be its pronoun in the objective case ; but its pronominal particles in other connexions may be a-, gu-, yu-, ye-, u-, ka-, -kwe, &c. The 6th prefix, J/a-, has an adjectival concord ma-, but a pronominal ji.'a- (ya-, a-).

' Similar ' preprefixes ' are present in the Temne language group of Sierra Leone, that West African speech family which offers such a strong analogous resemblance in its syntax to Bantu.

20 THE DISTRIBUTION AND CHARACTERISTICS

members of the body like ears, anns, and legs, but was mainly associated with ' doing', things, with verbs as an infinitive, with direction as a locative—' to ' or ' from '. Class i6 represented only ' place ', ' here ', ' on ' ! or had to do with time. Class 17 meant ' in-ness ', ' inside '. Classes 18 and 19, where they existed, were given the signification of ' great size ', ' unusualness ', ' awfulness '.

(9) Numeration in the original Bantu was certainly decimal. The root lor ten — kitmi— is virtually common to more than nine-tenths of the Bantu languages, and so in a lesser proportion are words for ' hundred ' {-kan* or -kdina).

(10) Demonstrative prefixes and pronouns are often preceded by a directive '«-' {m- before labials) which emphasizes attention. Pronouns in the nominative case always precede the verb- root and the objective or accusative pronoun ; which last likewise ordinarily takes its place before the governing verb-root.'

(11) The root-word of the verb, free from prefixed or suffixed particles, is always the 2nd person singular of the imperative. The sense, application, and complexity of verbal roots as concepts can be modified, enlarged, affected adverbiallj' in many ways by changing the terminal vowel and adding suffixes, some of which almost lead to a process of inflexion by the carrying back of their dominant vowel. In normal Bantu languages verb-roots end in -a. almost invariably. The verb can be given a passive instead of an active signification by means usually of a suffix. No changes of the verb-root are attempted at the initial ; but preceding the verb-root come the particles which indicate tense and mood, the pronouns, prepositions, and the syllables of negation. On the other hand, adverbial particles and occasionally particles with a relative or ' refer-back ' sense are tacked on to the termination of the verb-root. Negation is most commonly conveyed by prefixial particles — Ka-, Sa-, Ta-, Ki-, Si-, or Ti- (or variants of these), but also by negative suffixes or adverbs, and sometimes (additionally) by changing the terminal -a of the verb-root to -/. With very rare excep- tions, the prefix which turns the simple verb-root into an infinitive is Ku- (No. 15 prefix).

(12) Prepositions are used, not postpositions ; the only exception to this rule being the locative 'Hi (-nyi, -n, -«) meaning ' in ', which often accompanies the 17th prefix and concord, Mu-. Adverbs, on the other hand, when not treated as independent qualifying nouns or adjectives are suffixes, following the verb-roots or pronouns they govern. The most characteristic and widespread of the Bantu prepositions — in some cases not to be distinguished from Class prefixes- are «a-, ui- (with, by, and), -a (of), ku- (to), pa- (at, on), and mu- (in).

These twelve groupings ^ of Bantu characteristics must be regarded as defining the features of

' Sometimes in a fuller form (like the French moi) the objective pronoun is placed a/ier the verb in the corrupt north-west Bantu languages. There is also a doubtful exception in orthodox Bantu to the rule that the nominative pronoun particle shall precede the verb. This is the instance of the 2nd person plural of the imperative mood, wherein, in most Bantu languages, the pronoun particle (a very archaic one in form) -ni, follows the verb-root as a suffix {Pouia ! = love thou ! ; Penda-ni ! — love ye ! ).

' C. R. Lepsius, in his introduction to his Nuhische Cratnmalik (I?erlin, 18S0), also gave at some length twelve propositions to define the peculiar character of the Bantu language family. These for a long time were regarded as the standard by which the Bantu features were to be appraised in setting them apart from other language types. But his definitions are not in all cases accurate, adequate, or intelligible, when a complete survey of the forty-six Bantu groups is effected — an impossibility in Lepsius's day. Summarized, Lepsius's propositions were as follows :

(l) The prefixes of the nouns are the most outstanding features in the Bantu languages, and may be contrasted with the similarly characteristic suffixes of the Hamitic tongues. There are about eight different prefixes in the singular. [Lepsius was wrong here ; in his own day eleven had been computed; and even if he had argued that the three Mu- prejixes are to be counted as one — though their concords show them to be inherently different— we should still be left with nine, twelve being the probable total.'\ They may be most easily compared with the genders or sex class-divisions of the Hamitic ; but the Bantu prefixes have nothing to do with sex. Oh the other h.nnd, they fall into classes which distinguish humanity from what is not human— animals, trees and plants, implements, constructions, and other things, especially everything that is extraordinary, outstanding. Some stress should be laid on the tendency in the prefixes to distinguish between conscient beings and inconscient things. [This feature in Bantu is somewhat exaggerated by L.] The meaning of these prefixes, which must once have been

OF THE BANTU LANGUAGES ai

nearly all the Bantu languages. But to scarcely one of them cannot some exception be quoted — solitary or unusual, it may be— if careful research is made. In the same way, though all these

independent root-words, is now no longer to be traced, and this fact alone would imply that the origin of the Bantu languages goes far back into Time. [The origin, yes j but not necessarily the 7>iigration- dispersal over Central and South Africa.\

(2) Sex differences appear unimportant to a primitive people, and perhaps from this arises the fact that in the whole array of negro languages there is no sex discrimination. [An incorrect and over-stated remark, in view of the indication of sex classification in the pronouns, articles, and concord particles of the Bofigcu, Bari, Masai, Musgu, and oilier obvious 7tegro languages.^ Lepsius then illustrates the extent to which sex discrimination penetrates the Hamitic, especially the Kushite group of those tongues.

(3) Outside the limits of the Bantu languages, there exist no class prefixes. [How about Temne ? There are class prefixes also in other Semi-Ba,ntu tongues not known to Lepsius, like Avativie of SW. TctigOdland.\ But nominal prefixes, which have lost their original use, are found frequently in the second zone [/'. e. the Sudan and West Africa], and are also thus characteristic for their origin. [' Her- kunft ' — all this and much else in Lepsius' s twelve Propositions is very vague and of no clear application to the definition of Bantu salient features.] These nominal prefixes may be compared with the nominal suffixes in the Hamitic.

(4) Personal pronouns are prefixed to the verb-root. There are no personal suffixes in Bantu [Lepsius has overlooked exceptions to this general rule]. This last condition is the opposite of what occurs in the Hamitic languages wherein personal suffixes permeate the speech.

(5) In his fifth proposition Lepsius defines the principle of the concord. [Not altogether correctly, for he lays too much stress on the ' euphonic ' and ' alliteration ' theory. Euphony in vowel sympathy and in initial consonants undoubtedly infiuences the phonetics of the concord-particles, but does not prevent them sometimes from being dissimilar to the prefixes with which they are associated.] Traces of this concord or vowel harmony may be met with in certain Sudanese languages (' languages of the second zone '), sometimes in the noun, occasionally in the conjugation. In the last mentioned it is to be seen in the south Bantu. Traces of a vowel harmony, which is certainly more euphonic than anything else, since it arises from the vowel of the word-root, may nevertheless have some ancient connexion with consonantal alliteration.

(6) In the Bantu languages, owing to the predilection for prefixes, there are no postpositions, only prepositions. [Incorrect : how about the locatives -m,-Xim, -la?] The Hamitic languages are divided in this respect : Libyan and Old Egyptian favour prepositions ; Kushite languages employ only postpositions.

(7) The noun in the genitive case in the Bantu languages is placed after its governing noun, and is linked with the governing noun by a particle which is compounded from the latter's governing noun's prefix. Here again the practice is dissimilar in the two groups of Hamitic languages : Old Egyptian and Libyan follow the Bantu method ; the Kushitic tongues place the genitive before the governing noun.

(8) The south Bantu languages arrange the pronouns and the verb so that the nominative comes first, then the verb, and then the object ; so that the verb comes between subject and object. In the Hamitic languages it is the rule that the verb either begins the sentence or closes it, so that subject and object come into immediate contact.

(9) This word-placing in the Bantu is however modified when the object is represented by an abbreviated pronoun. This is inserted before the verb-root, but after the pronoun of the subject in the nominative case. The abbreviated pronoun representing the object in the sentence resembles [Lepsius thinks] an infix. In any case there is no parallel to this arrangement in Hamitic speech.

(10) It is characteristic of the Bantu languages that every syllable should end in a vowel or a nasal consonant ; a difference without importance. This syllabic character tends through the isolation of the syllables to give a clearness of expression which does not exist in Hamitic speech.

(11) Associated with this feature is the fact that, in contrast to the terminal enunciation, the initial utterance of words or syllables in Bantu may be frequently strengthened, particularly by the apposition of a nasal. Such, as an initial consonant has quite a different meaning and effect to what it would have at the termination of the word or syllable. While the nasals are of no more importance than other conso- nants in the final syllable, they are much more potent as an initial sound if followed by another consonant ; because in such case their tone element is almost equal to a syllable, probably because they are the remains of an earlier full syllable. [Lepsius' s meaning in this eleventh clause is far from clear, and the whole proposition is valueless. Perhaps he is alluding to the ' directive n ', frequently referred to in the present work.]

(12) In his last proposition Lepsius descants on the influence of tones, of intonation in ancient African forms of speech, employed as in Chinese, Burmese, Siamese, and one or more groups of West and West Central African languages. He classes intonation as among the characteristics of the Bantu tongues. [ True, it plays an important role in Secuana and SesutU, and perhaps to a less degree in Zulu- Kafir. It is also an essential element in the pronunciation of Mpongwe and of the Fang or Panwe lan- guages of the north-west J- but— as a tneans of discrimination between monosyllabic words — roots worn down by attrition to a common semblance it is far from being one of the distinctive features of the Bantu languages

22 THE DISTRIBUTION AND CHARACTERISTICS

points taken together constitute — with an array of numerals, pronouns, prepositions, noun-roots, verb-roots, and syntaxial features held in common— the outstanding marks that distinguish a Bantu from a non-Bantu language, it must not be supposed that anyone of these qualities or distinctions is peculiar to the Bantu family. Bantu phonology may be matched in some of the Sudanic tongues, and even (to a remarkable degree) among the Papuan and Melanesian languages. The Fula, the Wolof, the Temne, some of the Nilotic tongues and the Hottentot possess the principle of the con- cord;' in the Fula, in the languages of south Kordofan, of north Togoland, of Sierra Leone, nouns are ranked in a great variety of classes independent of sexual gender, each class being marked by a special initial sound, a prefix, or a suffix. Change suffixes into prefixes — a revolution which may take place somewhat quickly in a language (as witness the difference between Tudor English and Victorian English)' in the placing of prepositions— and you would have in Fula a form of speech very reminiscent of the Bantu family. The Bantu languages, therefore, do not strike one as foreign in their origin to Negro Africa, less so, in fact, than the Fula. Probably the parent speech was a prefix-using tongue of ' West African ' features spoken originally in the very heart of Africa, some- where between the basins of the Upper Nile, the Bahr-al-ghazal, the Mubafigi, and the Upper Benue, one of a chain of similar prefix and concord languages stretching from westernmost Africa to Kordofan. The first invasive move of the archaic Bantu seems to have been eastward, toward the Mountain Nile and the Great Lakes. The proto-Bantu folk were certainly once — many centuries ago— settled in the Nile valley north of the Albert Nyanza. One can trace their place-names still, in countries long since colonized by Nilotics or Sudanese. In this direction (the south-western part of the Anglo-Egyptian Sudan) they must have remained till at least as late as three or four hundred j'ears before Christ — late enough to have been in full possession of goats and oxen and to have received the domestic fowl ' from Egypt or Abyssinia. Then they embarked on their great career of conquering and colonizing the southern third of Africa.

' This principle of the concord between the class prefix or termination of substantives and the adjectives, pronouns, and verbal pronominal particles dependent on them is by no means confined to the languages of the African negro and negroid. It may, indeed, like so many other ideas, have originated in Europe or Eurasia, and have spread to Africa. The concord is present in the Aryan tongues (Ilia, bona femina, &c.), in the Semitic and Hamitic languages. In these last the concord of the genders extends even to the personal pronoun ' thou ', which is rendered by feminine as well as masculine forms. In the Bantu, Fula, Temne, and other African ' class ' languages, the classes are many and do not refer to sexual gender ; in the Dravidian and certain of the Caucasus languages the categories are few, but likewise have no sexual reference. In the Aryan, Semitic, and Hamitic (and Nilotic and some Sudanese) speech families, the classes may be three — masculine, feminine, and neuter — or merely two : masculine and feminine.

' Teutonic English, like modern German and Dutch, was largely prefixial in its qualifying preposi- tions. Our ancestors said ' uprise ', ' uptake ', ' understand ', ' offset ', ' enfold ', where we, especially during the nineteenth century, would prefer the more analytical locution of ' rise-up ', ' take-up ', ' set-off', ' fold-in ', &c.

' I have employed this argument several times before, and so far it has not been successfully com- bated : namely, that the great dispersal of the Bantu from their first common area of development in the very heart of Africa cannot have taken place before they had received the domestic fowl and given it a well-marked generic name — Z-j/X-;/— which, by the by, is very like the early Persian names for 'fowl'. Now the fowl was first domesticated from the wild Callus ferrugineiis in India (or from the allied G. battkiva in Burma), about four thousand years ago. [It reached China as a domestic bird about 1400 B.c] But it was totally unknown to the ancient Egyptians, or to the Greeks before the seventh century B.C. It did not reach Mesopotamia till about seven hundred years before the Christian Era, nor Egypt till after the Persian invasion of 525 B.C. Even supposing it spread rapidly up the Nile valley as a domestic bird, it could hardly have reached Central Africa for another hundred years— if so soon. Con- sequently for the fowl to have become so well-established amongst the Bantu as to have received a lasting and almost universal name amongst them, much time is required between the Persian invasion and the period of the great Bantu migrations : say 300 years. Amongst those negro races which do not speak Bantu languages, though they may be living in closest proximity to the Bantu, the word for ' fowl ' is quite different from the Bantu term (though this last may extend to the Semi-Bantu languages). Nor is it likely

OF THE BANTU LANGUAGES 23

The most archaic Bantu languages at the present day are those of Bunyorio, Bukonjo), Ruanda, Buganda, the islands of the Victoria Nyanza, and the region of the north-west slopes of Mount Elgon (Masaba). It is reasonable to suppose that after the Bantu language-type came into existence — perhaps between the Shari basin and the Bahr-al-ghazal (its nearest relations are found in the Niger and Cross river basins) — its first great concentration lay in the region of the Great Lakes, in Equatorial Africa. Here it was shut off for a while from westward extension by the dense wall of Congoland forest. Southward, down the shores of Tafiganyika and thence south-west and west across the mountains and plateaus of southern Congoland to the Atlantic ; eastward and south- eastward to the Indian Ocean and the Zaiigian coasts ; southward across the Zambezi and Mashonaland to temperate South Africa swept the Bantu invaders, armed, it may be, with novel iron weapons and led by a Hamiticized aristocracy. They progressed, no doubt, as rapidly as the Zulu or Basuttti hordes overran Central Africa under the white man's observation in the nineteenth century. Here our forefathers or contemporaries have been able to testify to the spread of the Zulu clans and the Zulu tongue from the 35th to the 3rd degree of South latitude in about fifty years.

The original Bantu invaders found no empty Africa before them. We may be certain from abundant evidence that Central and South Africa have been inhabited by man for many thousands of years, and probably by some black-skinned negro — forest negro, Sudanese negro, or Congo Pygmy — for an indeterminately long period, as far south as the Zambezi river. But South Africa proper, between the high plateaus of northern Becuanaland and the sea-coast of Cape Colony and Natal may possibly have been the domain exclusively of the Bushman and of the Hottentot hybrid down to the first Bantu invasions from across the Zambezi in (at a guess) about 700 a.c. South Africa then or earlier may have been partially depopulated from some cause — some epidemic of germ disease. Anciently— thirty, fifty thousand years ago, farther back still, perhaps— there existed in southernmost Africa a remarkable cave-dwelling type, the ' Strandlooper', whose skull bore less of a resemblance to the Bushman or the negro than to some round-headed, semi-Caucasian stock of the Mediterranean. This Strandlooper either co-existed alongside the Bushman or preceded and was followed by this specialized desert negro. In time the Straridloopers died out, leaving perhaps some traces of their presence in varied Bushman strains, and bequeathing to the Bushmen their wonderful gift of drawing and engraving, so reminiscent of the cave and rock pictures of the Solu- trian and Magdalenian Cave men of Europe or the Palaeolithic and Neolithic nomads of Algeria and the Sahara.

We may imagine, however, some two thousand years ago, a South Africa beyond the Zambezi and Kunene rivers given up for a time to the dwarfish, steatopygous, yellow-skinned, click-using Bushman.

The next disturbing element may have been the Hottentot ; a hybrid between negro and

that the fowl was earlier introduced into East Africa by seafaring Arabs, thus reaching the Bantu home by another route long anterior to two hundred or three hundred years before Christ. The fowl was seemingly not introduced from Java or Sumatra into Madagascar by the early Malay colonists of that island ; but within the Christian Era by the Arabs ; and in Madagascar it bears no native name resembling the Bantu root-word -kul;u. It may nevertheless be argued that the fowl, like tobacco and the Brazilian Casarca duck, might have been introduced to the coast regions of Bantu Africa quite recently, long after all Central and South Africa had been ' Bantu-ized ', and have rapidly spread over the southern third of the continent, carrying its name with it. I5ut in that case why did not its name similarly reach the negro languages across the Bantu border-line.' Tobacco and the 'Muscovy' duck are nowhere known by universal names throughout Bantu Africa, though their Portuguese names of 'Taba' or 'Tabaco', and ' Pato' may still survive here and there and be shared by Sudanic, Nilotic, Nigerian, and Bantu languages. The Bantu, it is to be noted, have, like the Aryans, remained faithful long after their dispersal to the word-roots referring to the animals, plants, implements, and abstract concepts known to them in their original home.

24 THE DISTRIBUTION AND CHARACTERISTICS

negroid and some more northern Bushman race, which seems to have migrated from Equatorial East Africa south-westwards to the Central Zambezi and thence to the Atlantic coast near Walfish Bay, and on again southwards till the Hottentots entered (what is now) Cape Colony and displaced the Bushman. Later, it may be, there came other invaders from the north-east, by sea, rather than by land, the mysterious gold-mining, stone-building, monolith and phallus-worshipping people who originated the Zimbabwe walls, temples, and emblems. These are amongst the unsolved mysteries of Africa ; but the most striking of these stone cities, together with this whole impulse and procedure of gold-mining, most certainly were foreign to the Bantu arts and pursuits. The Bantu have seldom any word for gold. The tribes oldest in legendary history in South and East Africa use for gold an Arab word meaning 'moneys'. Zimbabwe and its like originally, perhaps, and cer- tainly later, early in the Christian Era, became associated with the exploring and trading voyages of the south Arabians ; and among the early writers of Islamic times there were to be gathered traditions that when the Arabs first visited the south-east coasts of Africa they found them peopled with ' Wakwak ' savages, who from the brief allusions made strike one as more like Bushmen than Bantu. The arrival of the Bantu hordes (the ancestors of the Karaiia; Becuana, Zulu-Kafir, Mapiigane, and Roiiga peoples), may have been the cause of the abandonment of the stone fortresses and mining depots by the sea-people whose forerunners designed or built them. After a time, however, the Arabs got into friendly relations with the Bantu colonizers, who in a much clumsier fashion imitated their mining and their stone-building. The Arabs, indeed, by their trade, their influence, and their slave-markets seem to have moulded the conquests of some early Bantu warrior-chief into that empire of the Mwenemutapa' — Lord of the mine— which the Portuguese found still a potent state when they reached South-east Africa at the commencement of the sixteenth century.

The Bantu were quite possibly settled on the more northern coast of the Indian Ocean— the land of Zanj— at the beginning of the Christian Era. The early Arab traders from Mokha, Aden, Mokalla, and Maskat cultivated friendly relations with them. A hybrid type sprang up, the Swahili, or men of the ' Suahil ' (coast-lands). With the aid of the Arab daus (sailing vessels) many of these Swahili from the land of Zanj or Zafig (the Persian or Arab name for Black man's country, and the origin of ' Zanzibar ') colonized the Komoro) islands and even reached Madagascar, supplying the Malagasy dialects with numerous Swahili words.

In the Congo basin the Bantu invaders often appeared as solitary huntsmen, boldly attacking the big wild beasts with their iron spears. Bantu culture throughout the Congo basin is closely associated with the iron spear— ^ongu, Kongu, Liangs. They probably found the Congo forests peopled already with forest negroes speaking tone languages of West African type or with a still more primitive people, the Congo Pygmies. There would even seem to be portions of the inner- most Congo basin which the tall Bantu have not yet penetrated and where there are only Pygmies. On the Northern Congo there still remain patches of non-Bantu territory inhabited by negroes speaking languages of an as yet unclassified type, vaguely styled ' Sudanic '. Somewhat similar to these are the Forest Lendu to the west of the Albert Nyanza.

The first great Bantu migrations undoubtedly emanated from the vicinity of the Victoria Nyanza and north Tanganyika, and were directed round and not through the Congo forests. But in course of time, the Bantu communities founded in the region of the Albertine Nile and the Nyanzas, or farther still to the north-west between the sources of the Wele and the Aruwimi, broke through the wall of forest, and sent streams of migrants across the northern parts of Congoland to the Gaboon and the Cameroons ; reaching as their final effort the island of Fernando P6, which was not

» Afwette = ■ lord ', ' monarch ' ; MuUpa. = ' mine ', ' digging ', ' excavation ', in the dialects of the Lower Zambezi. This the Portuguese corrupted into Monomotdpa.

OF THE BANTU LANGUAGES 25

very anciently severed from a Cameroons promontory by some volcanic rift. Such an east-to-west propulsion might, in contact with the Nigerian peoples of the Cross river and Benue basins, have created bj' fusion the Semi-Bantu languages ; and no doubt one or two groups of such arose in this way and not very anciently. But it is more probable, taking all known facts into account, that a large section of the Semi-Bantu speech forms is either descended from sisters of the Bantu parent tongue {born in that central region north of the Benue, east of the Niger, and west of the Shari river), or has been originated by early, very early, south-westward migrations of the proto- Bantu before the great eastward move into the Nile basin and before the exact shaping of Bantu features had taken place.

One is led irresistibly to deduce from the linguistic, ethnological, and anthropological evidence before us that at some such critical period in their career the negro speakers of the early Bantu languages were brought under the influence of a semi-Caucasian race from the north or north-east. Perhaps it was a gradual drifting into Central Africa of Egyptian or Gala adventurers coming up or across the basin of the Mountain Nile ; an infiltration of a superior type of man rather than a forceful invasion. Descendants of such ancient civilizers of Central Africa are undoubtedly to be seen at the present day in the Bahima, Ruhinda, Batutsi aristocracies of the Nyanza regions, the Mangbettu and Azande 'royal' families of the Nile-Congo water-parting, the Bafi-bupongoj of central Congoland, the Luba chieftains, and the many handsome-featured pale-skinned castes and ruling clans in so many of the Bantu peoples. Such good-looking ' negroid ' types may be encountered among the Zulus, the Becuana, the Hererto, the Alunda, Baluba, Manyuema, and the northern Congo riverain tribes, the Fang peoples, and the Duala of the Cameroons. Livingstone, Burton, Stanley, as well as later travellers, were all struck with the Egyptian-like features of the aristocratic families in the big Bantu states and kingdoms or among the warlike tribes of Central Africa, Similar aristocracies were noted by the pioneer missionaries and traders in southernmost Africa.

The Bantu-speaking peoples of Africa, it might be here stated, do not constitute a race apart from other negroes or offer any homogeneity of physical type. But on the whole they represent so much the average negro type that ' Bantu ' is still in favour as a physical definition among crani- ologists. In reality, they are just fifty millions of negroes whose speech belongs to one of the many language families of 'negro type; only in this case the language family instead of being confined in its range to a hundred villages or two hundred square miles, is spread over the southern third of Africa— say over 3,500,000 square miles— from the Cameroons, the Northern Congo, the Nyanzas, and the Mombasa coast to Cape Colony and Natal. Bantu languages are spoken by Congo Pygmies and forest negroes of marked prognathism and stunted stature, by fuzzy-haired brown- skinned Fernandians still in the stone-and-wood age of culture, by tall-statured, handsome pale- skinned negroids with Egyptian profiles — the Bahima and BaCutsi, by great, burly, coarse- featured, downright, ' West Coast' negroes like the Wanyamwezi, the Baganda, the COvambco, and the folk of the Angola coast-lands ; by tribes with the yellow skins, broad, wrinkled faces of the slit-eyed Bushmen, by the very black-skinned, large-eyed comely Atoiiga of Nyasa, the semi-Arab Swahili or Karaiia, the stately Zulu, the mean-looking Batwa of Bangweulu, the Assyrian-like Baluba and Bupofiga), the simiesque Banandi of the Ituri forest or Babofigco of the Gaboon, and the all-round, moderately ugly, black negroes of average stature and average negro characteristics throughout the rest of South and Central Africa.

Yet about the Bantu speech and the culture which accompanies it (ordinarily) there is a suggestion, strengthened by the association of these languages with metal-working (iron more especially), with agriculture, cultivated plants, and cattle-keeping, that adds to the impression derived from their legends, their religious beliefs, games, and weapons. It is that the Bantu language family was finally moulded by some non-negro incomers of possibly Hamitic affinities, akin at any rate in physique

26 THE DISTRIBUTION AND CHARACTERISTICS

and culture, if not in language,' to the dynastic Egyptians, the Galas, and perhaps most of all to those ' Ethiopians ' of mixed Egyptian and Negro-Nubian stock that down to one thousand years ago inhabited the Nile basin south of Wadi Haifa and north of Kordofan. Such a race may even have been akin to the Tibu farther west, the Tibu of Fezzan, the Eastern Sahara, and the Libyan desert. We know that some of the weapons of the Central Congo are to be traced northwards to Tibu weapons and implements of ancient date.

In spite of the suggestion of Egyptian influence in the domestic animals and plants of Bantu Africa (as also of the Western Sudan) and of the Egyptian profiles among the Bahima, I cannot but think this ' Egyptian ' influence over the Bantu was wrought indirectly through Gala, Ethiopian, and perhaps Tibu, through more or less Hamitic peoples influenced by Egyptian civilization of an early type. If much direct Egyptian influence had found its way to Central Africa from the Lower Nile it must surely have imported into Darkest Africa that deep attachment to Stone— for building and for worship — which emanated from the Mediterranean and south-west Asiatic peoples. But no skill in stone-quarrying, stone-carving or stone-building ever reached the Bantu, or for the matter of that, the Fula, the Mandingo), or the Hausa. Of all such arts the Bantu culture and languages are ignorant.

The Baganda and Banyorco legends of the incoming strangers of remote antiquity, the wander- ing demi-gods (' Bacwezi '—ghosts, spirits) who came with long-horned cattle, trained dogs, iron weapons, religious theories, and the elements of civilization generally, make no mention of building in stone or of building at all. The very similar Ful shepherd aristocrats of Western Africa likewise had the ' Bantu ' culture, the herdsmen's craft, a simple agriculture with the hoe,'' an acquaintance with iron and copper (though copper plays a very secondary role in Ful and Bantu thought, legend, and speech) ; but no notion of stone-building. With the exception of the Bube or Fernandian indigenes, no Bantu people has been found living in an age of stone implements, though there are abundant evidences to show that nearly all Negro Africa (except perhaps the innermost forests of the Congo basin) went through ages of using flaked, chipped, bored, and polished stone weapons and tools. The Bushmen and Hottentots had remained in this stage, without knowledge of metals, using stone and also horn, bone, stick, thorn, and shell. Most of the forest negroes apparently adopted— or reverted to — wood before they were introduced to the use of copper. An industry in smelting and hammering copper, in exporting copper in the ingot or in manufactured form seems to have arisen a long while — many centuries— ago in Katanga, where the copper deposits are singularly rich ; and this trade in copper to have spread from south Congoland to north and west

' It is curious that, but for a few obvious loan-words in the east of Africa, there is absolutely no Hamitic impression or affinity about the Bantu languages. Lepsius seems to have been impressed by the complete dissimilarity between the two language families. I cannot see myself much evidence of Hamitic influence on the structure, syntax, or origin of the Nilotic negro languages and the Bongto of the Egyptian Sudan— not such as there is for example in the Hausa and even Musgu. But one can detect in the numerals of the more eastern and south-eastern of the Nilotic tongues, and in a number of their nouns, evidence of Kushite (Gala and Somal) having long been in contact with the Nilotes, Masai, and Nandi. But one finds no such evidence in the Bantu languages, not even those of the Nyanzas, which are the most archaic. Yet in the regions where the purest Bantu speech exists you meet with these cattle-keeping aristocracies such as Bahima, Batutsi, and Ruhinda, whose physical resemblance to Galas, Somalis, even Abyssinians and Pharaonic Egyptians, is most striking. One can trace in the local folk- lore, religion, domestic animals, &c., the influence of the Caucasian or semi-Caucasian races of North- east Africa, but not a trace of their Hamitic languages. There is perhaps a little influence of the Nilote in the shaping of the Bantu mother language. Can the handsome negroids from the north and east who would seem to have been the leaven that stirred the Bantu dough more than two thousand years ago and who urged these Sudanic negroes to spread over and occupy the southern third of Africa have been derived from some stock like the Tibu which, though semi- Caucasian in blood, had received no language from the Asiatic Hamites ?

' Though the Gala, even far inland in Equatorial East Africa, knows and uses the plough, no plough has ever been known to the Negro or the Fula.

OF THE BANTU LANGUAGES 27

Congoland ; and southward as far as Nyasaland. Was it pre-Bantu ? We have no means of telling ; but there are many indications showing that the invasion of the Bantu was facilitated by the use of iron weapons, and that the working of iron ore and the fabrication of iron weapons and tools in Central and South Africa dates only from the Bantu conquest of the southern third of the continent.

The spread of the more aristocratic ' negroid ' Bantu is also associated with the ancient Egyptian or Gala long-horned ox {Bos taunts aegyptiacus). This breed possibly had a west Asiatic origin, and it is evidently nearly related to the Indian humped ox. No trace of any wild progenitor has yet been found in its ancient habitat, Egypt and Ethiopia. It was obviously the first type of domesticated cattle in Africa. Later there entered East Africa the Indian Zebu breed, which may have also replaced the aegyptiacus type in southern Arabia. The Zebu, or humped type, crossed Africa from the east coast to western Congoland, the western Sudan, and western Zambezia, and travelled down the south-east coast of Africa to Zululand. The Hamites, I should say, had already intro- duced the long-horned aegyptiacus cattle to the regions south of the Victoria Nyanza. And here they were brought by that mysterious race, the Hottentot hybrid, to central South Africa and down the south-west coast to Cape Colony. At some unlcnown period afterwards the Bantu peoples round the Nyanzas received these long-horned cattle from the north and conveyed them in their southward migrations as far as Damaraland. Probably in all but the elevated regions of Central Africa these straight-backed, long-horned cattle died out from germ-diseases, which is why so many Bantu tribes at the opening up of Africa by the modern white man were without cattle or had adopted the humped cattle of Indian origin. Cattle, both of the aegyptiacus and indicus breeds, seem to have been introduced from Bantu Africa into Madagascar by Arab intermediaries and slave- dealers ; which is why a Bantu word for 'ox' is the chief term employed in Malagasy.

One of the difficulties attending the acceptance of the theory that the ' Bantuizing ' of one-third ol Africa has been a long process in unwritten history and cannot be compressed within a period of about two thousand years, is the splitting up of the mother tongue into more than two hundred and twenty distinct languages, and the improbability of these languages with their marked idiosyncracies having sprung into existence in so short a period as is assigned by writers like myself. By searching Portuguese records in regard to the Koiigto or Karafia tongues we do not find a great difference between the speech of the sixteenth and seventeenth centuries and the speech of to-day ; the numerals of the Bakwiri or Barundco at Ambas Bay (Cameroons) written down by some French or Dutch trader at the close of the seventeenth century are almost identical with the modern form.

But with the evolution of languages, as of species. Nature— I believe— proceeds ^^rsmZ/ww/, alter- nately with slow progression. A great jumble of events, and lo !— new languages spring suddenly into existence. Those that suit the altered circumstances remain and continue their course for centuries with slow modifications. Not more marvellous would have been the rapid differentiation and specialization of the leading Bantu languages in Central and South Africa, between — let us say— the year One of the Christian Era and the twelfth century, than was the coming into existence of the Romance languages, engendered by the impact of the Goth, German, and Slav on the vulgar Latin of the decline and fall of Rome. No Romance language existed in the sixth century A.C. But, by the close of the twelfth century, Italian, Sicilian, Provencal, French, Castilian, Portuguese, and Rumanian were distinct, mutually unintelligible tongues; analytic where the parent Latin was synthetic, possessing an individual character which has not greatly altered in the slow subsequent changes. Dante's Italian of the thirteenth century would be perfectly intelligible in the streets of Florence to-day, and is not very different from the dialect spoken in Central Italy from the tenth century onwards. Yet in the sixth century no such language existed. We might almost say, judging from the Oaths of Strasbourg and other scanty evidence, that the French language was born and shaped in its essential features between the sixth and the tenth centuries. English which

28 THE DISTRIBUTION AND CHARACTERISTICS

would have been almost unintelligible to an orthodox Anglo-Saxon of looo a.c. was created between about iioo and 1350.

I am disposed to agree with Lepsius in the belief that the Bantu language-type, like the parent speech of the Aryans, has taken a very, very long time to shape out of some negro speech in the heart of Africa ; I am only arguing that the commencement of the sudden and rapid invasion of central and southern Africa by the Bantu cannot be referred back much earlier than the second century b.c. ; and that the differentiation of the more than two hundred distinct forms of Bantu speech occurred subsequently and rapidly. Six thousand years ago the Aryan Schwarmerei was probably just beginning, somewhere in eastern Europe. And there were then— I suggest— but the slightest dialectal differences in tribal speech amongst the Arj'an Russians to indicate that one group of clans would become the progenitors of the Aryan-speaking Asiatics — would in Asia generate the Tokhari, the Pisacha, the Sanskrit, Prakrit, and Zend languages ; another tribe Aryanize the Lithuanians ; yet another section (as they marched north-westward, westward, south-westward, or southward) become ancestors of the Slavs, or the Goths and Germans, the Kelts and Kelti- berians, the Itali, the Dakians, Thrakians, and Greeks.

But if close resemblance in structure and syntax, if similarity of numerals and pronouns, and the possession in common of a great number of root-words of nouns and verbs, adjectives and preposi- tions are to be given full value as evidence of near relationship and of a recent origin from a common source, then we must regard the expansion and differentiation of the Bantu languages as a much more recent and rapid process than that which brought about the Aryanizing of all Europe and much of South-west Asia.

The map of Bantu Africa will show the main directions taken in presumed history by the different streams of Bantu migration ; and an examination of the groups of Bantu languages will, after detaching true and widespread Bantu roots from their vocabularies, leave a residue which must represent the assimilated fragments of prior languages spoken by the peoples whom the Bantu armies conquered and fused with ; just as in the Keltic tongues, the Armenian, Albanian, and the modern Romance dialects there are words retained from languages of utterly different affinities which were in occupation of the land when the Aryan or Aryanized invaders came with their overpowering influence. Some very slight influence of Gala (a Hamitic 'white man' speech) can be traced through the East African Bantu, about four hundred miles south of the present Gala range in East Equatorial Africa. The Sudanic tongues, the Nilotic and Masai, even the Nubian and Kordofan languages have sent words from the north — chiefly of domestic animals, wild beasts, and trade goods— circulating through northern and western Congoland, through the Nyanza Bantu, the speech of Kikuyu, Kilimanjaro), Usambara, and the coast-belt of Equatorial East Africa. The unclassified, somewhat monosyllabic tongues of the Niger delta and the Cross river have influenced the worn-down Bantu languages of the West Cameroons. The influence on Bantu of Arabic, Portuguese, English, French, and Dutch is so palpable and (excepting that of Arabic and Persian) so comparatively modern that it hardly comes within the range of philological studies. It needs no explaining. Considering how long the Arabs and the Gala Hamites have been in influential contact with the Bantu of East and South-east Africa, and how powerfully the Aryanized Persians of the tenth, eleventh, and twelfth centuries a. c. influenced the Zafigian coasts and islands, it is surprising how comparatively few Arabic and Persian words have crept into Bantu speech. Swahili is by far the most Arabized of the Bantu tongues ; yet except in refinements of vocabulary often ignored by the common people, it remains essentially and very typically a Bantu language. The fact that the ancient and mediaeval Bantu invaders of eastern Zambezia and South-east Africa knew and cared nothing for gold till some incoming foreign people from across the seas taught them the value of that metal, is shown by the absence of any true Bantu root-word for ' gold '. The word by which 'gold' is rendered in the Zambezian and South-east African Bantu languages is— if not

OF THE BANTU LANGUAGES 29

a variant of the Portuguese ouro or the Enghsh 'gold '— derived from Arabic, and usually a corruption of dir/tsm^ (iicidraiun, ndaUmit).

Subtracting the foreign, the pre-existing elements from the modern Bantu tongues, we are left in nearly every group with a surprisingly large proportion of words traceable back to a common inheritance from Old-Bantu , from the pristine Bantu of North Equatorial Africa. Most of the defeca- tions from this rule are due to that human love of metaphor and trope, that dislike in timid or refined minds to calling a spade a spade, a devil a devil, a urinal a urinal. Totemism, superstition, would start a practice of referring to dreaded snakes not as ' nyoka ', ' the snake ', but as ' the long animal ', 'the hisser', 'the coiler'. The Bantu tongues have as many paraphrases for alluding shame- facedly to the genital organs or the functions of generation and defecation as the most modest speech of Europe and America. The variations of Bantu vocabularies, especially in the south, have been mainly caused by Hlodnipa. This word— a verb in Zulu meaning ' to have shame ', to be ashamed ' — covers the practice of not calling sorhe common object, some beast, bird, fish, utensil, or geographical feature by its proper name if that name happens to be the cognomen of a near relative, a husband, father, brother, mother, great friend, or respected chief. The custom refers more to women's utterance than men's ; unhappy Woman throughout the long history of the ascent of the human species having always been regarded as unlucky, as an Eve doing the wrong thing and bringing down ill luck by offending the vague Powers of Nature and Fate. Consequently, in parts of Bantu Africa, if a woman marries a man named ' Lion ', though she may continue to call her husband shyly and whisperingly by his name, she will henceforth call the real lion, the ' roarer' the 'slayer of beasts ', the ' big cat '. Farther north, if the totem animal of the clan, or the most striking local example of divine or demoniac power is a crocodile, it might not do to be heard calling ' crocodile ' (on the principle of letting sleeping dogs lie). So the crocodile —to the confusion of the inquiring philologist — is not called Ngwena or Ngandti, by one of its widespread Bantu names, but ' the long one ', ' the sly one', 'the snapper-up ', or some other roundabout nickname. Evidently, however, as with family slang in our own land, a fashion in misnomers seldom starts from the vagaries of a few fantastics; otherwise the Bantu languages would not have remained after some two thousand years of dispersal so astonishingly true to type.

Before the reader's attention is turned to the long series of illustrative vocabularies which I have thought it best to supply as a prelude to the discussion in detail of the existing Bantu lan- guages, it might assist him or her if I outlined the most striking features of the presumed Bantu mother speech. We can, of course, only deduce these from an analysis and comparison of those Bantu languages which have preserved the greatest proportion of widely-distributed word-roots in their vocabulary, or have similarly retained archaic forms of prefix, adjective, pronoun, preposition, adverb, and verb, which, from their recurrence and re-emergence in so many parts of South and Central Africa, would seem to have been part of the common stock of ideas in the original Bantu centre of dispersal. I shall thus have supplied some standard of comparison by which the diver- gency and the degree of divergency of the modern Bantu languages may be gauged, in the outcome of their development from the parent tongue of North Equatorial Africa, two to three thousand years ago.

The following are examples of some among the many noun-roots of universal or considerable distribution in Bantu Africa ; given in the shape which they probably assumed in the original language or group of dialects (the Ur-Bantu of German writers, or as I prefer to style it in English, Old-Bantu) from which the two hundred and seventy-six Bantu languages and dialects of to-day are descended.

' In its turn, no doubt, derived from the Greek drachma.

30 THE DISTRIBUTION AND CHARACTERISTICS

Noun-roots

Father (Se, Sa, Si ; Tata) ; mother (Nya, Na, Nyina ; Mama) ; brother (-ina, -dugu, -pangi) ; sister (-lumbu) ; maiden, virgin (-wali, -wala, -isika, -dumba) ; grandparent (Kaka, Kukoi) ; child, son (-keke, -ana, Meona) ; wife (-ka, -kaai) ; twins (-pasa or -longco).

Chief (-kama, -ami, -fumu,' -keosi, -ini (-ene)) ; doctor, medicine-man (-fumu,' -ganga) ; witch, sorcerer (-Icoki, -sawi, -sawa, -lemba) ; thief, stealer (-ibi, -iba) ; man (-ntu, whence Muntu = a man, Bantu = men) ; male (-lume) ; female (-ka, -kazi, -kat>i) ; body (-wili, -biri, -zimba) ; skin (-gojba (-gcovi, -gcozi), -kanda) ; head (-tu, -twe) ; hair (-bale, -bili, -suki, -kisi) ; nose (-gulco, -puleo, •pula, -pomboj, -indto) ; eye (-isoo) ; ear (-tu, generally Ku-tu) ; lip (-Icomco) ; mouth (-nwa) ; tooth (•inco, -gegu, -songa) ; tongue (-limi, -laka) ; chin, beard (-devu) ; saliva (-te, Mate); neck (-kingca, -kcosi) ; shoulder (-bega) ; buttock (-takco) ; back, backbone (-gongoj, -uma, -tana, -busa) ; chest (-fuba,' -adi, -tulu) ; breast, nipple (-bele) ; heart (-tima— this word is sometimes used also for chest and liver) ; navel (-kcjfu, -kuvu, -tcotco, -kundi, -kombco) ; belly, stomach (-vumu, -fu, -nda, -mimba) ; bowels (-tumbto, -da or -ra).

Arm (-bcokeo, -koaiMo) ; armpit (-apa) ; hand (-anda, -anja) ; finger or toe (-nwe, -nye, -kumu) ; finger-nail, and sometimes finger (-ala, -dala, -jala) ; penis (-bcolco) ; thigh (-berw) ; leg (-gulu, ■rundcoj ; foot-sole (-ayco) ; vein (-sipa, -sisa) ; blood (-rcopa, -gazi, -nga, -kame or -gama) ; milk {•ziwa, -bele); fat, butter, oil (-fata'); bone (-fupa and -kupa,' -bisi) ; tear {-scozi, -ziga, -bezi) ; breath, life, soul (-conyu, -oya, -oyco) ; death (-fu, -fua >) ; ghost, spirit (-zimu, -limu, -banda, -dumba).

Sky, and often God and rain (•g:ulu or -iguru, -julu, -lungu, -ingu) ; rain (-bula, -lungu) ; sun, sunlight, daylight (-juba, -ise, -sana, -laiiga or -tanga) ; moon, moonlight, month (-ezi from -ela, -era, to be white, shining, -gondco, -sungi) ; star (-ezi, -eti— perhaps from the same source as moon— and •nyenye, -iiini, -sonda, -tanda, -data or -jata) ; year (-aka, -vu) ; night or day (-lo> or -ileo— leloj is a widespread root for 'to-day' — naku, -siku, -tuku, or -fuku) ; darkness (-giza, -sise, -ilima, -rima, or -zima) ; wind, cold (-pepco) ; smoke (-ojsi, -isi, -iki, -ika, -dinga) ; fire (-lilw, -otco, from -ota, to warm, -pia, -bazu) ; charcoal (-kala) ; earth, ground, country (-si or -ti) ; stone (-bwe, -tadi, -ala or •bale, -manya) ; salt (-nyu, -ngtia, -kele) ; iron, iron ore (-ela or -bela, -uma, -tadi, -londw) ; water (-zi, -diba or -ziba) ; river, stream (-geda, -gazi or -giji, -iteo, -donga) ; lake, broad river, sea or (-ziba, -anza or -anja, -zadi, -sabi); mountain, hill (-scazi, -lima, -pili); river-valley (-rambco or •dambu) ; road, path (-dila, -gila, -handa, -bcoka).

Tree, stick (-ti) ; dry wood, firewood (-kuni, -kui) ; thorn (-iba) ; grass (-ani, -asi) ; banana- tree (-gomba, -konda) ; banana-fruit (-tuki, -konde, -dizi) ; yam (-lungu, -coma, -kua, -rail) ; wine, mead, or beer (-alwa, -labu, -lafu or -Iwfu) ; mushroom (-btoa) ; forest (-bira, -itu, -zitu).

Bee, honey (-uki) ; termite (-swa) ; fish (-swi) ; frog (-ula) ; snake (-coka, -piri (viper), -mamba (cobra)); crocodile (-kwena, -gandu, -mamba); egg (-ki, -landa) ; bird (-uni) ; wing, membrane (•papa) ; guinea-fowl (-kanga) ; francolin (-kwala) ; domestic fowl (-kuku).

Beasts of all kinds — flesh (-ama) ; rat (-beba, -puku, -kusw) ; porcupine (-nungu, -gumba) ; elephant (-dcobco, -joofu, -damba, -dembto) ; hippopotamus (-gubu, -vubu) ; rhinoceros (-pala. -pem- bela) ; pig {-gulu, -guluba) ; buflalo (-ati, -aka) ; giraffe (-twiga) ; ox, cow, cattle (-gombe or -kcomto) ; milch-cow or breeding-goat (-gcoma) ; horn (-mbe, -pembe, -sengu, -iga) ; goat (-buzi, -budi, -peeni) ; dog (-bwa) ; cat (-paka, -simba '') ; leopard (-gwe, -koi, -sui) ; lion (-simba, -tare, -tambco) ; hyena (•pisi,' -mb wi, -bungu, -puru) ; monkey (-kima) ; baboon (-yani, -kolcobe, -ponibo)) ; ape (-zike, -scokto).

Sorcery, fetish, religion, witchcraft, to bewitch (-luka or -duga, -sawa; -banda; -lemba, -kisi,

' A marked correspondence in Bantu phonology exists between/ and /•, 7' and^, j and/, 2 and 7': so that we find especially in the West, but also in East Africa, -kumu alongside -fumu, -guia, alongside •vuta. and -futa., -kuba and -fuba, -fua and -kfta ; -suniu corresponding to -fumu and -suta to -vuta.

" The range in meaning of this root is considerable. Sometimes it rises to ' lion', sometimes sinks to 'genet '. ' Query, cognate with old root for ' bone ' ?

OF THE BANTU LANGUAGES 31

•flti) ; magic, medicine (-ganga) ; name (-ina or -zina, -itto) ; song, to sing (-imba, -imbu) ; fear, noun and verb (-coga, -copa, -tina, -tia) ; shame (-sconi) ; hunger (-zala, -ala) ; sleep (-Ico, Tulco, from the verb-root -la, -lala, to extend oneself) ; dream, noun and verb (-Itota, -duta)) ; dance, noun and verb (-kina, -bina) ; love, desire, noun and verb (-penda, -kunda, -tanda) ; intelligence, and verb ' to know ' (-manya, -manyi) ; goodness, good (-ema, -wa, -into, -nuga, -ntofu, -scoga, -bote, -bota) ; bad, badness, excrement (-bi) ; rottenness, rotten (-covu, -ora or -coza, -vunda) ; fierceness, bitter- ness (-kali) ; hardness (-gumu, -guma) ; smallness, little, young (-toj or -tu, -dongu, -ke, -nandi, ■titi, -nini, -goinco, -ana) ; greatness, large (-kulu— from -kula, to increase — nene) ; oldness (-kulu, •nuna or -duna, -kuka, -kuta, -dala) ; young, new (-pya) ; whiteness, white (-ela, -eru, -tuka, -tuba) ; sickness (-dwala) ; shortness, (-fupi) ; length, distance, height, ' there ' (-le, -la, -de, -rare) ; ' within- ness ', inside (-da, -mu, -te) ; ' betweenness ' (-kati) ; place, ' hereness ' (pa-, -nco) ; house (-ndu, -zto, -zcobco, -dagco, -dabco, -umba— clay- walled); home (-ka, -kaya ; -anda, -banda, -ganda) ; canoe (-atoa) ; paddle (-kapi, -kasi) ; knife (-ale or -bale, -siu, -hamba, -panga, -paka, -pcokco, -mage) ; spear (-onga, -ofigeo, -furau) ;bow (-ta, -tta) ; arrow (-fwi, -soiiga, -bamba, -katto, -sala, -tegw) ; axe (-pasa, -bazoi or -bagu, -zuka, -gembe, -gimbu, -temeo) ; shield (-gabco, -guba) ; clothing— anciently ' skin ', ' pelt'— cloth, blanket (-gubco) ; door, doorway (-kuki, -ugi, -ibi, -belcd, -pitu ; -liangco) ; drum (-gcoma) ; thing, property {-ntu, -ma or -uma).

Prefixes and Preprefixes The original forms of the prefixes alone and of the preprefixes combined with them were probably as follows. It should be remembered that the preprefix is virtually identical with the 'article ' of Aryan and Semitic languages. It is also the concord-particle and pronoun of the class to which the prefix belongs : —

Prefix. Preprefix and Prefix.

Class No. I. Mu- (singular) ] usually reserved for human Gu-mu- (Wu-mu-).

r beings, and almost invari-

,, „ 2. Ba- (plural) ) ably for men and animals Ba-ba-.

„ „ 3. Mu- (sing.) Gu-mu-.

„ „ 4. Mi- (pi.) Gi-mi-.

„ „ 5. Di-, Li-, or Ri- (sing., sometimes used as an Di-di-, Ri-ri-, Li-li (I-ri-, E-ri-).

augmentative)

„ „ 6. Ma- (pi. and collective) Ga-ma-.

„ „ 7. Ki- (sing.) Ki-ki-.

„ „ 8, Bi- (pi. and collective) Bi-bi-.

„ „ 9. N- (M-), if, Ni- (sing.) Gi-ni-, Yi-ni-, In- (Im-), In-, Ini-.

,, „ 10. Zi-, Xi-, Ti- (pi.) Zi-zi, Ibi-, I-ti-, I-ti-n-.

„ „ II. Du-, Lu-, Ru- (sing., a good deal associated Du-du-, Lu.lu-, U-lu-, Ru-ru-.

with ' length ')

„ „ 12. Tu- (pi. and collective diminutive) Tu-tu-, U-tu-.

„ „ 13. Ka- (sing., nearly always used as a diminu- Ka-ka-, A-ka-.

live)

„ „ 14. Bu- (sing., collective, and sometimes dimi- Bu-bu-.

nutive as plural to Ka-)

„ „ 15. Ku- (sing., often used as preposition mean- Ku-ku-, U-ku-.

'to', 'towards ')

,, „ 16. Pa- (a locative, meaning 'place', ' at 'i ' °" ') Pa-pa-.

„ „ 17. Mu- (a locative, meaning 'in ') Mu-mu-.

„ ,, 18. Gu- (sing., augmentative) ? Gu-gu-.

,, „ 19. Ga- (pi. to Gu-, augmentative) Ga-ga-.

32 THE DISTRIBUTION AND CHARACTERISTICS

Associated with these preprefixes and prefixes, when they assumed the r6le of demonstratives, was the ' directive « ', a prefatory nasal which usually has the effect of ' this is so-and-so ' when preceding a noun or a pronoun, and which may be connected with either the gth prefix or with an old root meaning 'is', ' it is '. This initial «- («- before a guttural, nt- before a nasal) in many of the Bantu languages gives a more insistent character to personal pronouns or demonstratives, in the sense of't/iis, this one '. With the ' directive n ' affixed, these foregoing prefixes became in their archaic forms (and in this shape reappear in numerous existing groups) as follows ; — Class i. figu- mu ; 2. Mba-ba ; 3. Ngu-mu ; 4. Ngi-mi ; 5. Ndi-H, Ndi-di ; 6. Ifga-ma ; 7. Nki-ki ; 8. Mbi-bi ; 9. Ngi-ni, Nyi-n' ; 10. Nzi-zi ; 11. Ndu-lu; 12. Ntu-tu ; 13. Nka-ka; 14. Mbu-bu ; 15. Nku-ku ; 16. Mpa-pa. (No such nasalized demonstratives are traceable for Classes 17, 18, and 19).

The most archaic forms which can be deduced for the Numerals are the following :

1. -nwe, -nye, -guma, -dala (meaning ' finger ' or ' thumb ') ; -mu, -mto, -mwe, -musi, -mcusa.

2. -ball, -bale, -bili.

3. -tatu, -satu.

4. -ne, -nai.

5. -tanco, -sanco.

6. -sambu; -kaga ; -sasatu, -sasaba ; -tandatu or Ntatu ndatu (3+3) ; -tupa, -tuba.

7. Sambco, Sambco-bali ; Mpungati.

8. -nana.

9. -enda, Kenda ; Ifuka, Ivua.

ID. -kumi ; -longco, Mu-longco ; -kama. (Plurals of tens — twenty, thirty, &c. — were Gama-kumi, Gimi-longco, &c. In the Bantu mother language there were no special terms for ' twenty ' as in the Semi-Bantu and Sudanic tongues, nor was reckoning by scores, as it is in them).

100. Gama or Kama ; Dili-gana, Eri-gana.

The original forms of the Personal Pronouns in the Bantu mother speech were probably these :

I Mi, N, Ne, Ni. (Other modern forms like Ngi, Ngu, Ndi, Nze, Mbi, Mba are pos-

sibly the original pronoun N- prefixed to particles meaning ' am ' or ' here '.) me -mi, -n-, -ni-, -iigi-.

my -ngu, -iige, -m, -mi, -ni, -ne.

Thou Gu, Owe (We), Ku-, Wu-.

thee -ku-.

thy -kto.

He (she) Gu, Gwe (Yu, Ye), Ka-, A-.

him (her) -mu- ; -ge, -ke, -ye.

his (hers) -kwe, -ke.

We Tu, Su (Si), Twe, Cwe, Swe, Hwe, Fwe.

us -tu-.

our -itu.

Ye Nu, Nyu, Nwi (Ni), Mu, Mwe.

you -ni, -nyu, -nu-, -mu-.

your -inyu, -inu.

They Baba, babco.

them -ba-, -bu.

their -babu.

The reflexive pronoun 'self was probably -li- (-ji-, -zi-, -ri-, -i-, -e-).

OF THE BANTU LANGUAGES • 33

' All ' was rendered by -onse, -onte or -onke ; and in some senses by -cona and -ama. ' Many ' was -ngi or -ingi, -vula or -bula, -bama.

Adjectives were scarcely distinguishable from consonantal forms in Old-Bantu, and their more striking examples in persistence of word-root are cited with the nouns in the foregoing list. Several suffixes were probably employed in ancient as in modern Bantu to give an adjectival and adverbial sense to noun-roots, notably the widespread suffix -fu or -vu, which perhaps may be traced to an older form -bu, and be related to the 14th Class prefix. Adverbs were allied to nouns and pronouns, and prepositions likewise. Amongst these last, roots that are characteristically persistent and must therefore have been present in Old-Bantu are : pa-, on, at ; ku-, towards, to, there ; si- or -nsi, down ; -la, -le, -li, -di, -dia, far, distant, there ; nda, -da, inside ; -tei, -kati, between ; -eru, -nja, -nze, outside ; -pi, which ? where ? how much ? ; -nca, much, exceedingly, properly ; -mbele, before ; -pi or -fupi, near.

The adverbial particles of negation, usually when employed with the verb, were prefixed or in- fixed in conjunction with the nominative pronoun particle which precedes the verb-root. They were :

Ka- (becoming -kco, as a relative adverbial suffix) ; Sa- or Ta- ; Ki- ; Si- or Ti-.

The following suffixes seem to have existed in the Bantu mother speech for incorporation with the verb-root at its terminal to modify, reverse, and extend its meaning. Already, perhaps before the divergence of the main groups occurred, it was becoming customary to change the pristine -a of the verb-root termination into -i to express negation or doubt, -e to give a subjunctive or potential sense to the verb,' -u or -u to indicate other changes of meaning. But in addition to this inflexion of the terminal vowel of the verb-root, a variety of suffi.xes came into use which have remained in force amongst nearly all the descendants of the Old-Bantu.

These were in all probability :

(i) -ba or -wa,^ probably one of the word-roots meaning ' be ', and used to turn a verbal root from the active voice to the completely passive ; that is to say, passive with the suggestion of an agent causing passivity. Example : Vu-leka, to leave ; \m-lekwa, to be abandoned (by some one). '

(2) -ka. This particle also turns a transitive verb-root into an intransitive, but indicates passivity without an agent directly causing it. The comparison of the -ka suffix with the -ba or -•wa, passive and the ordinary transitive verb-root used in an active sense, might be illustrated in English by the following example : ' He cools, chills ' ; this would be equivalent to the ac/ive verb in Bantu : ' He is chilly or cold ' : this would answer to the sense of the -ka suffix : ' He is cooled or chilled ' (by some one or something) : this would be implied by the -ba or -wa suffix. Thus if -pasua (a develop- ment oi-pasa) meant ' to split ' in Old-Bantu, -pasu'ks. would mean ' to be in fragments ', and -pasuvrtk. ' to be split' (by some one).

(3) -ga or -iiga. This suffix conveyed either a sense of thoroughness in action, or still more, continuity in the sense of the English participle termination -ing. It also inferred habitual or recurrent action. Example : -pula, ' beat ' ; -pulaga., ' beat thoroughly, continuously '.

(4) -sa (za, jsa, sia ?) was causative. Thus if -lia was ' to eat ', -/«sa or -//sa would be ' to cause (some one else) to eat '.

(5) -la, -ila or -ira and -ia (and -ela or -era or -ea, by absorption of the terminal -a ' ; also -ina or -ena) indicated doing a thing to, for, or by another person : as for example -fa = ' to die ' ; but -fia and -fira would mean ' to die for (or on behalf of) or through ' another person.

(6) The vowel u (or in conjunction with the terminal a, oa) combined with -la or -ra, expressed a sense of reversal. Thus if -fuiiga meant ' to shut ', -fungxAa. would mean ' to open '.

' Examples of the change to -/ and -e would be : Ni-pangB. or MpangK, ' I make ' ; Ni-pangt, ' that I make ', ' let me make ' ; and N-ki-pangi, ' I not make '.

' In the Nyanza group and some ofthe western Bantu languages the passive suffix is -ibwa.

' A +i constantly resolves itself into e in Bantu ; a + u into a). In these suffixes /, r, and ti, I and n frequently permute ; as well as s and^, j and z, b and w.

34 " THE DISTRIBUTION AND CHARACTERISTICS

(7) -pa was connected perhaps with the root meaning ' place ', ' location ', ' at ', but came to mean as a verbal suffix ' becoming ', ' developing into ', ' acquiring certain characters *. Thus from the adjective -nene, ' fat ' or ' large ', was made the verb -nenepa., ' to become fat ', from -bi arose -bipa., ' to become bad ' ; from -ugco ' fear ' was constructed -cogupa., ' to be afraid '.

(8) -ana indicated reciprocal action : Ku-penda would mean ' to love ', but Ku-pendaaa. ' ' to love one another '.

(9) -ama and -ta probably had a reflexive sense, meaning to do certain things for one's self. Thus -ima would be lengthened into -imsma, ' to erect one's self ; -pa, ' give ', became -/>Ata, ' to give one's £elf' or ' to get '.

(10) -ile or -ide (with the variant -ine), perhaps also -isi (and, from their fusion, -itsi, -idzi, -ije) conveyed a sense of past or preterite action ; and in fact -ile (-ine, -ide, -ire) suffixed to the verb- root has become the chief form of preterite tense throughout the Bantu languages. The alternative iorm -isi (extended into -itse, -idzi, -ije, &c.) has to a great extent disappeared as an independent suffix, and is only traceable with difficulty in the composition of words. Thus, assuming -kunda or -gonda to mean ' love ' or ' desire ' in the Old-Bantu, it would become in most of the modern Bantu languages -kundiXt in the preterite sense of I (Thou, he, &c.) loved', or -gonzi (-gond-i%i), a form in which it actually exists in some of the Nyanza languages. More often the -itsi suffix united the other suffixes and verbal forms. It may really have arisen from an older combination of the -ide preterite with the -sa causative. The present form in Luganda (-idza) rather suggests this.

In addition to these suffixes, the Old-Bantu, like some of its descendants, probably affixed other 'prepositional', locative, adverbial particles to the verb-root after the manner of -pa (No. 7 in this list. In this connexion -mu would mean ' in ', ' inwardly ' ; -kw, ' to ', ' towards ' (as well as from quite another root, ' not ') ; -ge (-je) would be interrogative ' how ? '—in modern Swahili u-td-fany»-'^e means 'how wilt thou manage it', or literally, 'thou— will— do— how.''

Probably in Old-Bantu, as in its modern descendants, the verb was conjugated by prefixed as well as suffixed particles ; and these last partook of the nature of prepositions (like ku, ' to ' ; pa, p«a, ' here ', ' now ', ' there ' ; -a, ' of ' ; na, ni, ' with '), adverbs and pronominal particles, related to the noiin prefixes— ka, ma, gu, la, da, nga, ngi— and mostly by auxiliary and abbreviated verbal nouns, such as li (di) and ba, meaning ' being ' ; ta, ya, ' doing ', ' going '.

The most strikinglj' widespread verbal roots in the modern Bantu languages probably assumed the following forms in Old-Bantu :

abuse, insult	-tuka.
beat	-pula, -btila, -tera, -beta.
bite	-luma.
breed	-zala, -ala, -biala.
bring	-leta.
build	-lunga, tunga.
bury	-zika, -jika.
buy or sell	-gula, -suma, -tenga, -landa.
come	•iza, -pika or -fuka.
cut, chop	-tema, -baza, -senga, -kata.
dance	-kina or -bina.
deny	■kana.
die	-fu, -fua, -fa (-kwa).
dig, hoe	•lima.
drink	■nyiia, -niia.

' This was sometimes extended to -sand, -ngdna. by fusion with other suffixes.

OF THE BANTU LANGUAGES

35

eat

evacuate, defecate,rain,

give birth fall fear

fight, quarrel fish, catch fish give

go

heat, warm, cook hunt know laugh

leave, cease love, want, desire press, milk rot

say, speak see send sing

sit, remain, abide sleep, extend the body, snore

■la or -da, -lia or -dia.

■nya (this root is related to the ' mother' or feminine prefix, nya-).

-gwa, -bwa. -tia, -tina. •rwa, -rwana. •vuba, -luba. -pa, -nika.

-genda, -gya (-ja, -ya), -tamba, -hamba (march, walk), -pita (pass). -5ta, -aka, -pi, -pika. -winda, -winga.

•manya, -dzi, -dziwa, -jiwa, -azi. •seka or -seba. -leka, -sika.

■panda, -kunda, -gonda, -funa, -tanda. -kama.

-bora (-bwza), -bunda. -ti, -gamba. •bona, -lola, -laba. ■tuma. ■imba.

-ikala, ■sala. ■lala, -kcona or -gona.

smell, stink

stand, stop, be erect

steal

strike, kill

watch, tend

weep, cry, mourn

•nuka, ■nunka.

-g^ma, -sima.

■iba.

-ita, -tta, -ta ; -kuba, -buba, -bula.

•linda.

•lUa.

The original concept of the verb in Old-Bantu was no doubt participial, of the ' verbal-noun ' order, the noun of action. But these verb-roots seem in Bantu to be older than most of the nouns that do not, like the verb-roots, terminate in -d ; as though actions were described before all but the most prominent oBjects. The original verb-roots were largely monosyllabic and never more than disyllabic ' (unless onomatopoeic) ; but I cannot agree with some authorities who have argued that the whole of the Bantu roots, substantival and verbal, can be traced back to utterances of a single syllable. Yet they tend that way, and perhaps, in the far past of African peoples, might have been traced to an ultimate affinity with a monosyllabic parent stock, which afterwards left behind more direct and monosyllabic tone-using descendants in the regions between the Cross river and the Volta.

The simple verb-forms in Old-Bantu obviously ended in -a (with the exception of two or three irregular, interjectional verbs, such as Ti, ' say ' ; Njco ! ' come ! '; Li, ' be ' ; Pi, ' cook ', and various other meanings) ; and the majority of the nouns formed from verbs ended in -&) (probably from -au), -!, or -u. The -e termination of noun-roots may, like those that end in -a, have been very ancient and indicate independence of origin from verb-roots, or may — as so often occurs in verbal suffixes and prefixes— arise from the fusion of a- and -i.

* It is noteworthy that many archaic word-roots in Bantu begin with a vowel and are disyllabic such as -d/i (buffalo), -dfs (heat), -isCt) (eye), -a/a) (canoe).

D a

36

THE DISTRIBUTION AND CHARACTERISTICS

Indeed, in reconstructing the phonology of Old-Bantu, and especially of the still earlier African parents of that language family, one is led to assume a greater simplicity of vowel sounds even than exists to-day in the very Italian-like Bantu languages, and to postulate only o, a, i, and u for the original Bantu vowel equipment. So many of the existing u's and unstressed o's, as well as the e's, can be traced back to fusions of a and «, a and i, or to the broadening of u and i.

As far as we can trace the phonology of Old-Bantu by comparing, one with another, the pro- nunciation of root-words in the more archaic Bantu tongues of the present day— notably the Nyanza- Taiiganyika-mid-Zambezian groups — we may surmise that it made no use of tones or clicks to imple- ment its resources in differentiating sounds ; and that, besides a simple vowel system — o, a,i, u, and perhaps u — it possessed the following consonants (interesting deficiencies being also noted) : —

m, b, p, possibly w.

V, f. These, however, may well have been absent in the pristine days, and have arisen subsequently, just before or just after the great dispersal, principally from b.

Is, s, s. "G was doubtfully present, as a variant of / and s. ■£> almost certainly was absent; and is still of very uncommon occurrence compared with 6.

d, t, I, n. R is doubtful as a radical letter, though often arising recurrently from / and d, and in turn producing z.

i. J, if it had already appeared, arose from a palatalizing of d (dy, d) or g (gy) ; a lingualizing ofy, or a hardening of z. C, f, / were probably absent from Old-Bantu.

y. In the primitive Bantu speech it is doubtful if y existed as a distinct pro- nunciation" from I and gl.

n. This ringing nasal sound was probably heard when n preceded a guttural, and there may have been in Old-Bantu some of that general nasality of utterance which so strongly characterizes most of the negro languages of West Africa as to amount almost to a physical and racial trait— a distinc- tion still more to be remarked in Bushman and Hottentot and in the Semi- Bantu. But the « as an unsupported consonant (like the «^ in German or in the English words ' king ' and ' ringing ')— so common a feature in modern Bantu— can always be traced back to a fusion between n and g or k.

The guttural y was probably absent from Old-Bantu, not having as yet arisen from the slurring of^g- or the faucalizing of w.

k, g. But no X, X, or q.

h was probably absent, though it was soon to arise from p, s, k, or /.

Labials Labial-dental

Dental-lingual

Lingual-alveolar

Palatal-lingual

Palatal Palatal-faucal

Faucal Aspirate

Even in the Old-Bantu home there must have been a tendency among the dialects of the mother tongue towards an abrupt and wide-ranging variation between certain consonants— a psychic disturbance, so to speak, such as must often arise in aggregated genera and species of animals and plants or full-blown human languages, when they are about to produce great variations from type. And this sudden impulse towards varied pronunciation caused transitory or permanent permutation in consonants. Yet the permuted and the previous forms of word-root must have co-existed in the same tribe as they often do at the present day. These alternations were particularly marked as between labials and dentals— « changing to z,j to s, and /to 6 ; between labial and faucal— w and g,fa.Tid k, V and g ; between the lingual and the alveolar — / and d (together with r, which last is scarcely, distinguishable in Old or Modern Bantu from / and d) ; / and n, t and r ; and between the palatalized d and_/', d and z, t and s, / and c, s and f. P permuted easily to /and w, and still more readily and inexplicably to A, as soon as the great Bantu dispersal commenced. The aspirate, now so prominent

OF THE BANTU LANGUAGES 37

in the phonology of the majority of Bantu languages, was almost certainly absent from Old-Bantu. No root-word traced back to its most primitive form discloses the presence of h among the Old- Bantu consonants. The modern Bantu aspirates represent an older p,f, s, or k, or the faucal gasp of a vanishing click in the pre-Bantu languages.

In conclusion, I might state that I do not share the theory of certain German philologists that we should attribute to the Old-Bantu some degree of vagueness in consonantal utterance, resulting in y being the parent of the modern^ or w, 7' the antecedent of 6 and w, v and/; x of /^, and so forth.' The comparative study of the existing two hundred and twenty-six languages derived from Old-Bantu leads, on the contrary, to the conclusion that this mother speech of innermost Africa had broad, simple vowels, and distinct, well-defined consonants, almost limited in range to m, b, s, z, d, t, I, n, k, and g. In this respect, in the development of intermediate vowels and consonants between those of more pronounced type, the modern Bantu languages have only followed a parallel course to the Aryan, Semitic, and Hamitic languages. It may be argued that the early articulate utterances of Man were vague in sound : undecided in vowel quality, and in the influence of tongue, of teeth, nose passage, and epiglottis on consonantal out-breathings. The extraordinary Bushman language family may be pointed to as an illustration of a very primitive form of speech surviving, and offering a similar vagueness in utterance. But the Bushman dialects may quite well be, like the Bushman race, examples of speech forms that have greatly degenerated and have been much specialized by isolation in South African deserts. And the vowel sounds of Bushman and Hottentot appear to me to be simple, few, and clearly marked, except for nasality of tone. It is their clicks which often obscure the values of the consonants. But even here we have no great range of subtle consonantal sounds as in the modern languages of Europe, northern Africa, and western Asia. The Bushman lan- guage, no doubt, like the race that uses it, is exceedingly ancient and shows the severe attrition of long usage. The simplification of vowel and consonant values (like the sudden simplification of customs after long slavery to tradition) tends to be one of those revolutionary occurrences which accompany the birth of a new language species or a new people : for instance, the Arabic dialects of the Sudan and of East Africa — rapidly becoming two new languages — are softening the more exasper- ating of Arab consonants and enlarging considerably the use of broad, distinct vowels ; Swahili Arabic, with the regular intercalation of vowel and consonant, is at the opposite pole to the Maghrib dialect of Morocco, which is mainly a collocation of choking consonants. One has only to compare the phonology of modern Arabic with that of the ancient Semitic tongues of Mesopotamia and Syria to realize (as also in the case of Polish and Russian, as compared to Yugo-Slav and Old Lithuanian) how languages tend as they spread, develop, and are more and more spoken, to differentiate their vowels and their consonants with ever-increasing complexity ; so that at last the speech becomes unspeakable by a newer generation or unpronounceable by an invading usurper. Then one of those radical changes takes place which gives birth to some new and more harmonious jargon : Latin becomes Romance, Greek changes into Romaic, the cumbrous Anglo-Saxon is modulated into the pleasantly Frenchified English of Chaucer, which we still speak. And I should not be astonished if, during the next twenty or fifly years, several languages of Africa and Western Asia — even Armenian, Lesghian, and Albanian— were scrap-heaped, and forms of speech with simpler, more melodious phonologies were adopted in their place.

It is because the majority of the Bantu languages are so simple and clear in their phonology, so logical in their syntax that they are learnt with comparative ease by the stranger. The majority of them are much easier to master and to retain in the memory than Hausa, Ful, or Arabic. Those

' Still less do I agree with their ' pooling ' problematic vowel sounds by the device of a heavily cir- cumflexed / ii). They cannot themselves suggest how this is to be pronounced. The circumflex is a shallow and out-worn mystification.

38 DISTRIBUTION AND CHARACTERISTICS OF THE BANTU LANGUAGES

languages will survive — certainly Hausa and Arabic will, though I doubt whether the world or even the heart of Africa will tolerate for long the innumerable plurals and countless verb intricacies of the fantastic Fulde speech— but they will not spread much beyond their present limits. Already in northern Equatorial Africa Hausa and Sudanese Arabic are giving way to the Bafigala of North Congoland as a trade and administrative language, because it is a harmonious clear-cut, easily acquired, sufficiently expressive form ' of speech. Swahili is carrying all before it as the trade language of the regions between Galaland and the Zambezi. Zulu-Kafir will become the second language of South Africa if its exponents are wise enough to eliminate the silly clicks which at present mar its phonology and cause the European to take lip instead the ugly and stupid jargon known as ' kitchen Kafir '.

The political importance ol the Bantu languages in the future will be as great as the political importance of the Indian vernaculars. This has been an additional inducement to me to study them. But it is perhaps the unwritten history of Negro Africa enshrined in their word-roots which has proved for me the strongest lure in this long-sustained work of research.

CHAPTER III

ILLUSTRATIVE VOCABULARIES OF 276 BANTU AND 24 SEMI-BANTU LANGUAGES AND DIALECTS

NOTES FOR THE GUIDANCE OF READERS IN RESPECT TO THESE

VOCABULARIES

The orthography, both in regard to African geographical names and the rendering of African languages, is that adopted throughout the volume and explained in detail elsewhere. A summary of its principles for convenience is given here in apposition to the vocabularies.

M, b, V, p, / and w, are pronounced as in English. When any of these or other sounds are aspirated, the aspiration, if pronounced and distinct, is indicated by the letter h ' : and where it is less emphasized, by the Greek aspirate '. In numerous Bantu languages there is considerable indecision on the part of the individual speaker or the tribe as to the utterance of b and w, so that to the European listener the labial consonant is heard sometimes as a b, at others as a w, and occasionally as bv or vh. I have preferred to write it either w, b, bv, or v, according as the local or predominant pronunciation caught my own ear, rather than to impose accents on the w, which in reality give little help to the English student, who is sure to be understood by the native if he uses either a b, w, or v. Nevertheless this indeterminate, labial is represented by several writers well qualified to hold an opinion, as tu. Y is used by me for the consonantal t (the Germany). There is a great tendency to palatalize consonants in Bantu ; that is to say, to follow them by an f or a jv sound {dy, ty, ny, ky, gy). There is no other logical method of rendering this pronunciation than by adding the y to the palatalized consonant. Where, however, this palatalization (a very faint i or y) follows the terminal consonant and the employment of a _y might mislead the eye, the effect is better given by the vertical accent '.

G and k are pronounced as in English, the g being always as in the words ' get ', ' give ', ' go '. The faucal k (the Arabic j) is represented by q. The Greek gamma (r, y) stands for the velar g (the Arabic c), which is often rendered in English by gh and which is pronounced like the French r grasseye, or like a lower utterance of x- This last— the Arabic r- and Scotch cli — is represented by x, which is to be taken as the equivalent of the Greek x. This is the value of g and ch in Dutch and in many German words ; but the more palatalized sound of the terminal ch, or cit before a consonant, so often heard in German in ichand licht, is represented by x. H is pronounced as in English or German, but the strong h of Arabic (^) is rendered by /;', or sometimes by the double h {hli). The faucal contraction of the Arabic ain ( ^) is represented by _L ; the mere elision of a vowel or consonant by the apostrophe, ' ; and the strongly marked hiatus (the Arabic /tam««)— which is the gap in pronunciation occurring between distinct words in English (such as 'still ill')— is shown by ;. The nasal consonant sounding like ng in 'ringing' is indicated by «, or the merely nasalized consonant or vowel by ~. A'' is pronounced as in English. When

' There is much aspirating of consonants in numerous groups of Bantu languages. Wherever ph, th, dh,gh, kh, vh, &c., are encountered in this work they must be pronounced separately and with their true values : p-h, t-h, &c.

40 NOTES FOR THE GUIDANCE OF READERS

it is an initial consonant preceding another consonant it is slightly vocalized, something like the pronunciation of tm in the English word 'unto'. When it is doubled, like all other double consonants it is doubly pronounced.

The Polish or dental / (f) is hea#d in some Bantu languages (as it is in Portuguese and Brazilian pronunciation) but not so prominently or with such etymological importance as to be worth discrimination. The ordinary Bantu / is alveolar, like the English. It interchanges with r to such an extent that little distinction can be made in transcribing certain languages, except under the circumstances mentioned below. One person may use /, another r in pronouncing the same word in the same tribe or clan. The ordinary Bantu r sounds very much Uke the r in the pronunciation of Germans or of educated English people. The rattled r of Spain or the southern half of France is present in a few Bantu tongues (such as the Nyoroj dialects) and is best and most logically represented by a doubled ;-. The cerebral r (r) of provincial English and of many South Asiatic tongues is virtually unknown in Bantu Africa. The velar r (r) is scarcely distinguishable from y, and should be thus expressed unless it can be shown to proceed from an older alveolar or dental consonant. R permutes most easily with / and d. In some languages, however, it is necessary to discriminate between r and / in writing and in pronunciation, because r has taken the place of/, as / may have of rf, in the prefix forms and word-roots.

With regard to the dentals, s is always the upper or light hissing sound, as in the word ' hiss ', or 'saint'; z similarly is always the lower and heavier sibilant as in ' hazard ' and 'zeal'. The English sh and zh (French c/i and 7) are rendered by ^ and y ; the English //; in 'think ' and ' bath ' by t> ; and d/i (tli in ' that ' and ' bathe ') by d. T and d, when printed without aspirate, accent, or modification are to be pronounced as in English and most other languages. But the / in Bantu is often aspirated or pronounced explosively, somewhat as English people use the word ' utter ' when they speak of 'ut-ter astonishment'. The aspiration— as already mentioned— is indicated by the /; or the ' following the /; and the explosive quality by the accent ' (=/')• The / in Africa on the lips of natives often glides into an r sound. When it is most like an r I spell it thus, when most like a / I render it by / (or d, as the case may be). I do not think it necessary, from any point of view, to discriminate further between the various slurred 'or eccentric pronunciations of / by the tongues of Africans, as these variations are — at any rate in Bantu— of no etymological importance. The d in Bantu is occasionally (as is the /) heard as though doubly pronounced, and is in such cases doublj' written. The dental, alveolar, palatal, and lateral clicks in the Zulu-Kafir languages are represented respectively by tf, f, f, and;?.

C stands for the combination of//, a fusion of consonants that is more conveniently represented by one symbol, the Italian c (as before ; and e). It often results from a palatalized k, which in some tongues is best rendered by ky; and is pronounced like the English ch in 'church'. Similarly j is used for the combined sounds of d^. Where these compound consonants appear to be doubled in pronunciation they are more logically rendered by tc and dj, as they are thus pronounced. Just as the c comes very near k^', so the j sound is scarcely distinguishable from the palatalized ct {dy). In some of the south-eastern Bantu languages there are palatalized b's, v's, and ^'s, which are a source of much unnecessary fuss and complication of writing by some who have rendered them on paper. They should be written in accordance with their local pronunciation as by, vyy py ; bz, vz, ps ; or b^, v^, and pf.

The vowels of this phonetic alphabet are, of course, given their ' Continental ' values rather than those of the eccentric English spelling which grew up in the sixteenth and seventeenth centuries. O unaccented is pronounced as o in 'not' or 'doll': 0 represents the sound of o in ' store ' or aw in ' gawd ' ; o is a little deeper than the average German pronunciation of that vowel and represents the sound of the English m in ' hurt ', or i in ' bird '. A, * (similarly shaped throughout, even in italics) always stands for the » in 'father', 'rather', 'gala'. The short

IN RESPECT TO THESE VOCABULARIES 41

sound of this vowel (similar to the English u in ' but', or the o in 'worry'), is represented by U, a. y£, a: is the equivalent of the English * in ' fat', ' gather' or the French e mfemme. Unstressed e has the value of «■ in ' met ', ' rend '. E, e is the e in ' fete ', or « in ' hate '. E, e gives the short sound of e in men' and e in ' there ' (really a fused diphthong between e and 6). I is the English i in ' hit ' ; /, /, the e in ' cedar', and ee in ' feet '. 0, u unstressed represents the u in ' pull ', ' put ', ' bull ' ; and u, ii is the u in ' ruler ', and the 00 in ' mood ' (the French ou, German «). U,u = the French u. The sounds of o in 'bone' (French an and German oh) is represented by the Greek u. I have not thought it necessary to discriminate further between the various pronuncia- tions of the single symbol o (o, o, u), as to do so is quite needless for the philological understanding of my subject, and only an unnecessary puzzle to the reader's eyes. In some of the South and South Central African forms of Bantu there is a rather indeterminate vowel which varies in pronunciation between the short u in 'put' and the o in 'hope'. This is particularly observable in the Secuana group and in the tribal name ' Basutto '; which is ordinarily rendered ' Basoth<o ' by specialists and is actually pronounced more like Sutco than anything else, though to some it seems more akin to ScoQthoi. There is a vacillation also between the sound of i in 'pit' and « in 'met' in many Bantu tongues; but in such cases I prefer to write it as i or e, according as the sound of the word struck my own ears or those of some trustworthy colleague in African philology ; for in the etymology of Bantu there is considerable permutation between e and / when they are unstressed.

The stress mark is ", and the unstress ". Neither is employed unless the pronunciation of the vowel is exceptionally deep or light. The accent or pitch of the voice is indicated by '. This is so normally and frequently on the penultimate syllable that the accent is only employed when an exception to the prevailing rule in Bantu transfers the voice-pitch to the first, last, or other syllable than the penultimate. Monosyllables are not accented unless uttered with vehemence.

The ordinary high and low voice tones in speaking J — ;- are best shown by employing

the accent ' for the high tone and ^ for the low. The other tones of the speaking voice so prominent in the East Asiatic or West African languages are unrecorded in connexion with Bantu (wherein the manner of speaking more resembles the European mean than the voice production of Hottentot and Bushman, of West Africa, or Eastern Asia), and it is not necessary to provide for them in this work. The use of the high and low tones of the voice for purposes of etymological distinction is not common in Bantu and is only observable (perhaps) in the Becuana group and most markedly in the Panwe languages of the north-west Bantu area. Even here they scarcely come within the scope of the present work, which is not so much a pronouncing dictionary as a treatise of comparative philology. In the case of the southern Nigerian tongues, the discrimination of tones in language transcription is of great importance.'

The conventional symbols S and j stand for ' male ' and ' female '. The essential and (in nouns) virtually unchangeable root of the word is printed in heavier type where it is necessary to hold it in view; consequently the reader will regard the lighter-printed syllables as changeable excrescences on the main concept— preprefixes*, prefixes, suffixes, and infixes. When there are articles or preprefixes present in a language they are usually, but not always, given in these vocabularies, so that the Bantu word takes its fullest form. But when in ordinary pronunciation the preprefix is dropped, the word is printed with only its customary class prefix (if any).

The plural forms of the nouns are separated from the singular by a semicolon and begin with

' A reference on this point might be made to my Phonetic Spelling (Cambridge University Press). The tones of the South Nigerian languages have been accurately described in Mr. Northcote Thomas's recent works (Harrison & Sons, and ' Man ', R.A.I.).

' The preprefix is the vowel or syllable which precedes the actual prefix.

42 NOTES FOR THE GUIDANCE OF READERS

a minuscule letter. Frequently only the plural prefix is given, the rest of the word being the same as in the singular. When this plural prefix replaces a detachable prefix in the singular, it is followed (unless the whole word is printed) by .a hyphen (-) ; where the plural prefix is super-imposed on the singular form of the noun (which may be the bare root-word or retain its singular prefix), the plural prefix is followed bj' a plus sign ( + ), unless for the reader's special convenience the whole of the plural form is printed. When it is necessary to draw special attention to the plural form of the word, which may — verj- rarely — differ in slight degree from the root of the singular, the abbreviation pi. is prefaced. The hyphen preceding or following a word means that the root form must be preceded or followed by some varying detachable prefix or suffix. Variants of what is substantially the same term are separated from one another by a comma only ; quite distinct words by a full stop. Figures and letters in brackets are the distinctive number of a dialect, a peculiar word or two of which may be quoted as a variant, if it is not sufficiently distinct from the dominant type of language to be represented in a separate column. But figures may also indicate noun classes when connected with demonstrative pronouns. As many equivalents as possible are given for the one English term or concept, provided their composition bears on the etymological connexions between the Bantu languages. Ordinarily the most commonly used word is printed first; but because the reader from a partial knowledge of any of these languages here illustrated does not recognize an equivalent of the English test word he must not be in a hurry to conclude that this equivalent does not exist ; for it maybe scarce, recondite, or archaic, and thus have escaped his notice. Great care has been taken to attribute to no language or dialect words the existence of which cannot be attested ; but it may often be only among the wise men or old women of the tribe that the word is remembered or rightly applied. These vocabularies are intended primarily for etymological purposes, and do not necessarily represent the most modern and current form of speech, but all the words known to be or to have been in these Bantu languages since they first came under observation ; just as English or French might be illustrated by selections from the recorded speech of the three last centuries.

The gaps in the columns mean that the author has been unable to find an equivalent native word, either because of the still very faulty nature of our records or because the concept is lacking where the language is spoken. In a very few cases the whole of the columns attributed to a form of speech may be blank. This provision of space serves a double purpose : it draws attention to the exact lacunae in our knowledge and may perchance stimulate research ; and it provides the necessarj' room for filling up these vocabulary forms completely in course of time and in later reprints. The name of the scarcely known language or dialect at the head of the column records its attested existence as a Bantu tongue ; and one object in the setting forth of these 276 languages and dialects has been to place before the reading world as complete a list of known forms of Bantu speech as possible.' It is probable, indeed, that scarcely one distinct Bantu language now remains unlisted in this book under some fairly accurate and locally recognizable name, even though from deficient information I may not have enumerated all its dialects. As regards naming the languages, I have got as near as I could to the local designations. Where this was not possible, or it was vexatious to set aside some name long-established in European mouths, I have chosen the term most likely to lead the inquirer to the spoken tongue or to its printed records in literature.

On referring to such for verification 01 my own version, a reader maj' chance to observe that the original European or American authority spells the word differently to its form in my book.

' The largest number illustrated in previous works is fifty-six. My total is approximate, as it is not always easy to decide the separation between a language and a dialect. In all about 276 forms of speech are catalogued in the following vocabularies, and about 270 are illustrated. The specimens of Semi- Bantu are additional.

IN RESPECT TO THESE VOCABULARIES 43

In such cases I have departed from the original spelling In favour of my own phonetic system, in order not to puzzle my readers by perpetually varying methods of sound-representation ; or because I knew the other author's version to be incorrect in the light of later research. It must be borne in mind that I have travelled most extensively over Bantu Africa between 1882 and 1907, and have put many orthographies to the test as regards local native pronunciation. Yet I respect other people's version of a word which may differ from my own (unless palpably an error in hearing, writing, or printing), and give, if need be, the alternate version. Similarly, should any very critical person peruse these vocabularies and find minute differences between the words here recorded and those he has been accustomed to hear pronounced in Africa, let him not too rashly conclude that I am wrong : we may both be right, but the critic may have listened to a different dialect or to a more precise or more careless speaker. Some of my own early work of the 'eighties in Bantu or Nilotic transcription was at one time hastily condemned as inaccurate by other philologists following in my tracks, and for many years I sat meekly under their rejection of my version, until fuller information came along (or previously buried work was unearthed), and I was found to have been right after all : it had merely been the case of a different dialect or a speech wrongly named, but genuine under another designation. Similarly, I have in this book resuscitated and rehabilitated the work of half-forgotten pioneers by transferring the vocabularies they had wrongly named or wrongly placed to their right titles and geographical allotment.

The Bibliography which follows these language illustrations recounts in detail the sources ot my information. For convenience of classification I have arranged the Bantu languages into forty-six groups distinguished numerically by letters of the old alphabet. The formation of the groups is guided mainly by interrelationships, and a little by propinquity ; their sequence is geo- graphical, and commences in the north-east of the Bantu field with the most archaic and primitive examples of this speech family— the languages round the Edward, Albert, and Victoria Nyanzas. The course followed thence is southward and eastward down through East ^nd East Central Africa to Zululand and Cape Colony, and back northwards into the westeirn parts of Central Africa till the survey ends with the Island of Fernando P6.

The following is a list of the groups for convenience of reference : «

The Nyanza Languages

The Wunyamwezi Languages

The British East Africa Languages

The Kilimanjaro) Languages

The ^ambala Languages

The Zangian Languages

The Usagara-Ugo)go) Languages

The Upper Rufiji Languages

The Lower Rufiji-Ruvuma Languages

The North Ruvuma-North-east Nyasaland Languages

The Ukinga Languages

The Tanganyika-Bangweulu Languages

The North-west Nyasa Languages

The Yato-Ngindoj Languages

The Mozambique Languages

The South Nyasaland-Southern Rhodesia Languages

The Pungwe-Sabi Languages

The Mapafigane-Ronga Languages

BANTU :
Group	A.
)j	B.
u	C.
)»	D.
11	E.
»J	F.
»)	G.
)»	H.
»J	I.
J»	J-
' »»	K.
n	L.
11	M.
11	N.
11	0.
li	P.
11	Q.
1}	R.

44 NOTES FOR THE GUIDANCE OF READERS

The Becuana-Transvaal Languages

The Zulu-Kafir Languages

The West Central Zambezia Languages

The Western Zambezia Languages

The North-west Zambezia Languages

The South-west Africa Languages

The Aiigola Languages

The Koiigu) or Westernmost Congoland Languages

The South Congoland (Luba-Lunda) Languages

The Upper Kwangto Languages

The Kwafigco-Kasai Languages

The Central Congoland (Luange-Lwmami) Languages

The Middle Lwmami Languages

The Elila-Lowa-Lualaba (Bulega) Languages

The Ruwenzori-Semliki Languages

The'Upper Ituri Languages

The Wele-Aruwimi Languages

The Aruwimi-L(omami Languages

The North Central Congoland Languages

The Kwa-Kasai-Upper COgiuwe (Teke) Languages

The Central OOgcowe Languages

The OOgcowe-Gaboon Languages

The Spanish Guinea-West Cameroons Languages

The Manenguba Languages

The Middle Sanaga Languages

The Pafiwe (Fang) Languages

The Kadei-Sanga-L(obai Languages

The Fernandian Languages

The Cameroons-Cross River Languages

The Cross River-Calabar Languages

The Benue Languages

The Bauci Languages

The Central Nigeria Languages.

The Tcogojland Languages

The Senegambian or Guinea Languages ' '

For the convenience of readers I append in an abbreviated form the scheme of phonetic system of spelling adopted in this book.

Consonants : m, b, v, p,f, and w ; «, d, t, s, s, r, l,j, andy ; /;, g, k, as in English ; c like English ch ; X like ch in ' loch ' or in German machen ; i- like German cit in ich, licht (almost English 5//) ; y (Greek gamma) like g in modern Greek or the French rgrasseye or the Arabic gliaiit ; ^ = English sh; ^= English sli or French j\ d= English th in ' that ', and 6 = English th in ' think ' ; and « — the ng in ' ringing '. The dental click in Zulu-Kafir is d, the alveolar f, the palatal f, and the lateral;?.

Vowels : a = English * in ' father ' ; a, the sound of 11 in ' but ' ; ce, the sound of <t in 'gather', 'hat'; e = « in 'met', ? = « in 'fete*, i=e in 'there'; r= / in 'hit'; i,t = i in 'ravine'; o = o in ' not ', o = 0 in ' store ', u = o in ' bone ', o = « in ' hurt ' ; u = u in ' pull ', u = 00 in food ', « = French u.

Group	s.
If	T.
ij	U.
»>	V.
'* »	W.
>»	X.
»	Y.
jy	Z.
)>	AA.
»>	BB.
>»	CC.
»)	DD.
»	EE.
99	FF.
>J	GG.
if	HH.
J>	n.
JJ	JJ.
J>	KK.
)»	LL.
»	MM.
>»	NN.
»	GO.
»	PP.
))	• QQ.
>»	RR.
)J	ss.
JJ	IT'.
SEMI-BANTU :
Group	A.
j>	B.
»	C.
>i	D.
»	E.
ii	F.
)»	G.

ILLUSTRATIVE VOCABULARIES OF BANTU LANGUAGES

TOGETHER WITH DEFINITION OF GEOGRAPHICAL ALLOTMENT OF EACH TONGUE AND EXHIBITION OF PREFIXES AND CONCORDS

GROUP A

THE NYANZA LANGUAGES

Sub-group A i Konjoj I. Olu-konjco

Sub-group A 2 Nyorco 2. Uru-nyorco 2 a. Ru-gufigu

2 b. Ru-kycopi

2 c. Uru-tcrw (Ru-soilgora, Ru-ireo, d^c.)

z d. Oi)ru-hima

	I.	2.	2 a.	2 b.	2C.	2d.
English	Olu-konjco	Uru-nyorco	Ru-gungu	Ru-kycopi	Uru-torco (Ru-songora ctntiR{i-iTa,&'c.)	COru-hima

Adze

Animal, wild beast

Ant

Ant, white (termite)

Ape (chim- panzi or gorilla)

Arm

Arrow

Axe ...

Baboon Back ...

Banana Beard...

Bee ... Belly ...

E-suka; suka Eny-ama

M-bali; p/.

esim-bali.

N-scoki Emi-lungulu

Eki-tera ; ebi- M-pundu

OOku-bcokco

(jOmu-swa ; ami-. Eki-korco; ebi-

Em-basa ; //. esiom-

Efi-gerebe ;

esiofi- OOmu-gongco

CObu-karaata ;

ama-kamata Esion-deru.

(one hair of =

OOnu-leru)

En-cwaiiku. Em-baizi

Eny-ama. Eny-emera. Eki-sorco

OOru-hazi ; em- pazi. OOniu- zigya ; ba-

En-swa. (Ki-swa = a termite hil- lock). En-kubebe

I-sike. Eki-kuya; ebi-

OOmu-kconco ;

emi- (jOmu-tegco :

emi-.

OOmu-scogi.

E-kimara.

OOmw-ambi Eny-anzi.

Em-pangco.

En-demu Efi-kcobe

(jOmu-gongco. Eki-bega

Eki-tcoke ; ebi-.

Eii-konje OJmu-leju,

Ebi-reju

Eny-ama. Eny-ama

Ny-emera

N-kwiri

N-kuyege

Eki-kuya

COmu-kconco OOmu-kconco Bu-scogi Efi-gcowe

M-pangco

Ki-guri ; bi-

(x)niu-gongco

En-jcoki

En-zcoki ;

esion-ziuki In-daor /En-da. En-da

Bu-kcole.

Olu-isira

Ma-vega.

Mu-gongco,

Ki-buni (loins) Ki-tcoke ; bi-. Eki-tcoke

N-gonja Mu-redu, COmu-ledyu

Mu-redju.

(Ji-miri juiika

= one hair of

beard) K-eokyee ; //. En-jcoki

b-cokyee. Mu-da En-da

En-cwanku

Eny-emera. Eny-ama Efi-kura

Em-pazzi

En-swa

Ki-tera. Eci-kuya

(x)mu-ko3nco

OOmu-scogi

Eny-emera. Eny-ama

Em-pazzi

En-swa

Yi-siki

OOmu-kconca OOmu-ambi

E-hangu ; pi. Em-pangco m-pangu

Efi-kerebe. Eii-kcobe

Em-pundu OOmu-goiigco OOmu-gongco

Eki-tcoke OOmu-ledyu

En-jcoki Em-buri En-da

Eci-tcoci Ebi-rezu

En-zcoki En-da

46

ILLUSTRATIVE VOCABULARIES OF BANTU LANGUAGES

	I.	2.	2 a.	2 b.	2 c.	2d.
English	01u-konjo>	Uru-nyorti)	Ru-gungu	Ru-kywpi	Uru-torco (Ru-songora antiRu-iTu>,crc.)	Oi)ru-hima
Bird	Eny-wnyi ; esiony-. Uny-uni. Mu-isundi	Eny-o>nyi	Kiny-uni ; bi-	Eny-unyi	Eny-eanyi •	Eciny-conyi
Blood	OOmu-sasi	Esigama. E-samaga	Mu-sagama <>rMu-samaga	E-sagama	E-sagama. E-samaga	E-sagama
Body	OOmu-biri	U)mu-biri	Mu-biri	OOmu-biri	(jl)niu-biri	COmu-biri
Bone	Eri-kuha; ama-	E-gufwa, I-gufa ; ama-	Ki-gufwa; bi-	...	I-gufwa; ama-	I-guffa
Borassus palm	Eci-keoga	Aka-tugu;o>bu-. Eki-kcoga	Ka-tugca	...	Aka-kcoga	Aka-tuga
Bow	OObu-ta; ama-	CObu-tta ; ama-tta	Bu-ta ; ma-ta	OObu-ta	OObu-tta	(jObu-ta
Bowels	Aman-da	Ma-ra {J>1.)	. . .	...
Brains	COb-ongto. Esion-cuere	CObw-ongco	Bw-ongeo	CObw-ongca	OObw-ongoa	COw-ongcd
Breast (man's) Eki-kuba	Eki-fuba	...	. ..	Eki-fuba	Eci-fuba
Breast	Eri-bere ; ama-	1-were ; ama-	I-weri;ma-veri	I-bere	I -were	I-bere
(woman's)
Brother ...	(ji)mu-hara or	Omw-ana-wa-id<a Mw-ana-wa-ici	Omw-ana-mau	Omw-ena-we-itca	Gmw-ene-w-esM
	Omw-ana-w-etu OOmu-genzi.
		COwa-nyina.
		Mu-nyanya.
		B-ene (;*/.)'
Buffalo ...	Em-bcogca ; esiom-	Em-bcdgco	Em-bugcD	Em-beogco	Em-bugco	Em-bcdgco
Bull	En-de e-nume	E-nimi, Numi	Gi-numi	...	(Ente)eci-nume	E-nimi
Buttocks ...	Ebi-kalero> (//.),Eri-takeo a.ms,-(o/ beasts	Ebi-bunco	Bi-nyo> (//.)		Ebi-bunu	Ebi-buncd
Canoe	(jObw-atco; am at<o	CObw-atw	Bw-atoi ; pi. g-ateo'	Bw-ato>	Obw-atto	U)bw-ateo
Cat	Aka-jaiigo>	En-jangu	...	...	En-jangu	En-zangu
Charcoal ...	• 1*	Ama-kara	...	...	Ama-kara	Ama-kara
Chief	(jOmu-kama; aba-	Mu-bitco.tOmu- kamwa; aba-. OOniu-nyoreo. U)mu-lemi. OOmu-kungu	Mu-kama		tOmu-kamwa	OOmu-kama. OOmu-bitoj
ChUd	Omw-ana ; ab ana. CO-lume keke J. Aka	• Omw-ana - „ com-cojo>=o ■ „ mu-tci> = 5	Mw-ana „ ob'-saiza=S „ mu-teo = 5	Omw-ana	Omw-ana	Omw-ana
	mu-keke = in	Mu-hal<d
	fan/
Cloth	ODmu-tanda.	COrau-tanda	Ng-oyi (cotton)	Ku-buku	(Omu-tanda.	COmw-enda
	Omw-enda	(cotton) .	Lu-vugoa		Oru-goye
	{cot/on)	Oru-goye (bari)	(bari)		(bar/.-). Oru-kcozi
Cold	Em-beheo. COmu-rombe	Embehci	...	...	Em-beheo	Em-behw
Country . . .	Eki-taka	En-si	Ky-alco	• . •	En-si	En -si
Cow	En-de esigija	En-te e-zigija	En-te gi-za- gidya	En-te	En-te e-bugu- ma	En-te e-zigiza. N-gabe, K-gube. Ki-gaboj

' B-ene is the equivalent of the B-ena {ha-ina.) further south, meaning^ kinsmen ', ' brethren '. ' Perhaps this is a slurring of Gama-tco : vide Lu-masaba.

GROUP A: NYANZA LANGUAGES

47

English

Olu-konjco

Uru-nyorco

2 a. Ru-gungu

3 b. Ru-kytopi

2C.

Uru-torto

(Ru-songora

antiRu-ira>, &'c.)

2d. COru-hima

Crocodile

Day

Devil, spirit

evil

Doctor (medi- cine man) Dog

Donkey Door ...

Dream

Drum

Ear

Egg

Elephant . ,

Excrement

Eye

En-dicoka

(snake).

I-hondue Ki-rco.

Bu-hingi.

Omw-esi,

Omw-ise ( =

daylight) (Omu-limu;

aba-

Mu-kumo>

Em-bwa ; esiom-

Bu-sSku.

OOl-uyi,

R-uyi; pi.

esiony-uyi En-dcdtu

En-gcoma OOku-twe; ama-

twe Ere-ye; ama-yi,

Eri-hui En-zcogu ;

esicDn- Ama-gedzie. Ama-vezi. Ama-idzai Er-isoj

Gi-nyunigim

£n-sambia. Em-pyeo

Eki-rco. I-zuba Mu-ki-rco. (jOmu-sana Ki-roj; ebi-rco.

Hangwe

OJmu-cwesi ; ... .„

aba-.

COmu-zimu ;

aba- COmu-fumu

Em-bwa. M-bweni Em-bwa

Eki-bwa (« big

dog)

En-kaina (jOru-higi ; //. Mu-zigeo

em-pigi

En-dcotco. N-d(osiri

Eki-rutci> En-gcoma En-gcoma En-gcoma

(jOku-tu; ama-tu Ku-tu ; ma-tu COku-tu

Ei-huri ; ama- I-huri ; ma- huri Eii-jcajea N-zuzco; N-jej<o N-jcoju

Face, forehead Wbu-su

Fat Eki-sawu.

Ama-guta {oil,

butter) Father ... Tata. I-se.

E-sa

Ama-zi

Er-ijSco ; ama-isco

Ma-bi

R-isco ; ma-isco Er-isco OObu-sco

Fear .. Finger

Fire Fish

Foot .. Forest

Eri-saga

(jOmu-nwe ;

eme-.

OOmu-rcotco.

Eki-kumco COmu-rirco Eyi-swe ;

esi-swe. I-sui.

1-sumbi ; esi- Eki-sandco.

OOku-gulu OOmu-situ.

Olu-loiige

(jObu-s<o Bu-sco

Eki-paju ; ebi-. Ma-savu

Ama-gita (oil,

6-<r.) Tata. 'Sco. I-se. Tata Tata. Baba

l-se-nya)we

(my). I-si.

i-s<o

En-kise. A-tinyiri

OOku-tina COru-kumco; en- Lu-galco. Kalu OOmu-kumco

Eki-ara; eby-

(Eki-saiza =

thumb)

OOmu-rirco Mu-lirco Mu-rreo

En-cui N-tui En-cii

Eki-renge. Ki-renge.

Eki-gere ; ebi- Ku-gulu Eki-bira Ki-bera

E-kigere

En-sambia

En-sambia

Eki-reo, Eci-rco Eky-iru ; ebi-

COmu-cwesi (Omu-cwesi

COmu-fumu

Em-bwa. Em-bwa.

Eki-bwa (a big Eki-bwa (a big

dog) dog)

Oru-higi ; eny- Oru-higi ; eny-

igi igi

En-doatco

En-dcato0

Eii-gcoma (Jl)ku-twe or (jOku-tuitu I-huli Ei-huri

Efi-guma COku-tu; ama-tu

En-deaba, En-dasba

Ama-izi. Ama-tutorco

Er-is<o

En-zcozu Ama-zi

Er-isu

CObu-sco (jObu-sco

En-sazu. Eki-sadzu.

Ama-vutta, Ama-zita

Ama-futa

Tata. I-se, I-sco Tata. I-se, I -sco

(Oku-tina

Oru-kumco or OOdu-kumco

OOmu-rreo

OOku-tina

(jOru-kumco ; pi. efi-kumco. Eki-ara

tOmo-rrco En-cu. En-fwi En-tchu nr En-c5 or Em-fu Eki-gere ; ebi- Eki-gere ; ebi-

Eki-bira

Eci-bira

E 2

48

ILLUSTRATIVE VOCABULARIES OF BANTU LANGUAGES

	I.	2.	2 a.	2 b.	2C.	2d.
English	01u-konj<o	Uru-nyorco	Ru-gfungu	Ru-kyeopi	Uru-toroj (Ru-soiigora rtWRu-irco.dft.)	COru-hima
Fowl	En-gookco '	En-kcokco	En-kcoku	Eii-kcakaa	En-kcokco	Eti-koakco
Frog	Eki-tere	Eki-kere	...	...		Eki-kere
Ghost	OL)mu-limu ;	COmu-zimu ;	M u-zimn	...	OOmu-zimu ;	COmu-zimu ;
	aba-	ami-. Eki-turu			emi-. (jOmu- zummco	emi-
Giraflfe	...	N-tuiga	...	...	En-tuiga	.••
Girl	...	Mu-isiki	•••			...
Goat	Em-buri, Em- buli ; esiom-. Em-bene } ; pi. esem-bene	Em-buzi	M-buli	Em-buzi	Em-buze. Em-penne 5	Em-buzi
„ (he) ...	Esa-mban (S). Em-faya	Em-berabuzi, Em-paya (S).	...	...	...	...
„ (she) ...	...	Buguma (})	...	...		...
God	Ru-hanga or	Du-hanga.'	Ka-tonda	Ru-hanga	Ru-hanga" or	Lu-gaba or
	Du-hanga	Ka-tonda			Du-hanga	Nyam-hanga
Grandparent	I-sekuru	Is-en-kuru. Nyin-en-kuru	...	•••	...
Grass	Om-wata.	Om-wata.	Bu-sibbi.	• ••	Eki-suki.	Om-wata.
	Eki-suki	E-tete. CObun-yansi, CObu-yansi	Lu-saka		E-scojco. E-tete. Em-burara. OObu-nyasi	CObu-nyadzl
Ground	(x)mu-taka	I-taka	I-taka	I-taka	I-taka. Ahansi	I-taka
Ground-nut	...	Eki-nycobwa		...	Ama-kere. Eki-nycobwa	Eci-nycobwa. Ama-kerre
Guinea-fowl	En-ganga	Eki-tajumba	Nku-jumba	• •■	En-tajumba	Eci-kanga
Gun	Em-bundu	COmu-gaiigco	Mu-duku	• ••	COmu-gaiigco	Em-bundu
Hair	COlu-yuwiri ; esion-ziwiri. Se-suere. Bu-eya {on body)	I-scoki ; ama- suki	I-scoke (Bo-ya, on limbs)	I-so>ke	I-scoke	I -scoke
Hand ... ...	Eki-ganza.	Eki-ganja.	Kiganja.	Ki-ganza	Eci-ganza.	Eci-ganza.
	Eme-nue [i. e	Ebi-ara ( =	Mu-kwnco ;		COmu-kconco	COmu-kconco
	fingers)	fingers). En- garto, (Ji)mu- kunu (also lower arm)	mi-
Head	COmu-twe ; eme-twe	tOmu-twe ; emi	Mu-tue	COmu-twe	COmu-twe	COmu-twe
Heart	(jOmu-tummeo. (jOmu-tima	COmu-tima. tOmu-tuma	Mu-tima	OOmu-tima	COmu-tummco. COmu-tima	COmu-tima
Heel	Aka-sinziru ; cobu-	Eki-sinzirco. Eki-kongcoijco	Ki-sinzirco	...	Eki-sinzirco	Eki-tsintsineo
Hide	Eki-ani. En-gcoba	Eki-satco. Ol)ru-hu.OL)mu guta. En-gcoz	Ki-satu. N-gcozi	Ru-hu. M-pirn	Eki-satco	COru-hu ; em-pu
		{for slinging
		infants on back
Hill	Eki-cwa ; ebi-	Eki-swa. Em-bara	...	-	Oru-scozi	...
Hippopotamu!	3 Eyi-sere ; esi-	En-jubu	M-vuvu	...	Em-vuvu. En-sere	En-zubu
Hoe	I-suka	En-fuka. Eni-funi	...	...	...	...

Etnin Pasha adds the word vutte : perhaps in error.

' Ru-hanga also means ' skull '.

GROUP A: NYANZA LANGUAGES

49

English

Olu-konjco

Uru-nyorco

2 a. Ru-gungu

2 b. Ru-kyupi

Uru-toreo (Ru-songora

andRu-ira, &'c.)

2d. COru-hima

Ojbw-coki U-coki

Ei-hembe ; Ei-hembi

ama-. En-kule En-jii. Ka; //. N-umba

ama-ka.

{0-mbsi= aiiAe

house of) En-Jara N-zala

I-ba,Ba (Barco) Em-pisi M-pici

Iron Eky-coma, Eki-coma. Bi-uma

Ec-coma. (Ji)bu-tale.

M-seke OOmu-singa

Island OOku-itsinga tfr Ei-zinga Ki-singa

Eri-tsinga ;

ama-

Honey Horn	(jOb-ukyi Eri-hembe ;
	ama-
House	Eny-umba ;
	esiony-
Hunger	En-zala.
Husband . .	En-jera
Hyena	Em-piti

CObw-cDCi OObw-oaci

Yi-hembe; ama- Yi-hembe;ama- Eki-kuli

En-ju

Eny-umba ; ama-. En-ju