• TABLE OF CONTENTS
HIDE
 Front Cover
 Half Title
 Title Page
 Introduction
 Identifying the Bantu language...
 Methods of classification
 The Bantu languages classified
 Full classified list of the Bantu...
 Back Cover














Group Title: classification of the Bantu languages.
Title: The classification of the Bantu languages
ALL VOLUMES CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00072644/00001
 Material Information
Title: The classification of the Bantu languages
Series Title: classification of the Bantu languages.
Physical Description: 91 p. : fold. col. map (in pocket) ; 25 cm.
Language: English
Creator: Guthrie, Malcolm, 1903-
International African Institute
Publisher: Pub. for the International African Institute by the Oxford Univ. Press
Place of Publication: London
New York
Publication Date: 1948
 Subjects
Subject: Bantu languages -- Classification   ( lcsh )
Genre: non-fiction   ( marcgt )
 Notes
General Note: "One of a series of publications issued in connexion with the Handbook of African languages which the International African Institute is preparing."
 Record Information
Bibliographic ID: UF00072644
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 00184799
lccn - 49001313

Table of Contents
    Front Cover
        Front Cover 1
        Front Cover 2
    Half Title
        Page 1
        Page 2
    Title Page
        Page 3
        Page 4
    Introduction
        Page 5
        Page 6
        Page 7
        Page 8
    Identifying the Bantu languages
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
    Methods of classification
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
    The Bantu languages classified
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
    Full classified list of the Bantu languages
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
    Back Cover
        Page 93
        Page 94
Full Text














EORGE FORTUNE COLLECT(


~ Ic'-- rC 1









THE CLASSIFICATION OF THE
BANTU LANGUAGES








THE CLASSIFICATION

OF THE

BANTU LANGUAGES


BY

MALCOLM GUTHRIE
PH.D., B.Sc. (LOND.)
Reader in Bantu Languages in the
University of London




















Published for the
INTERNATIONAL AFRICAN INSTITUTE
by the
OXFORD UNIVERSITY PRESS
LONDON NEW YORK TORONTO
1948














Oxford University Press, Amen House, London E.C. 4
GLASGOW NEW YORK TORONTO MELBOURNE WELLINGTON
BOMBAY CALCUTTA MADRAS CAPE TOWN
Geoffrey Cumberlege, Publisher to the University







This study is one of a series of publications issued
in connexion with the Handbook of African Lan-
guages which the International African Institute is
preparing with the aid of a grant made by the
Secretary of State under the Colonial Development
and Welfare Acts, on the recommendation of the
Colonial Social Science Research Council.




























PRINTED IN GREAT BRITAIN









CHAPTER I
INTRODUCTION
atter half of the nineteenth century saw the end of most of the great exploratory
eys into that part of Africa which lies to the south of the Equator. The expansion
.ssionary work which followed on the opening up of the previously unknown
during this period gave rise to intense interest in the languages of the newly
vered peoples. This resulted in the production of a surprising number of diction-
and grammars describing the various languages which were found.
early as 1862 Bleek drew the attention of scholars to the fact that there was a
ing family resemblance between widely separated languages in this area. When
grammars became available it became evident that the Bantu family was indeed
y large one. Moreover, its peculiar characteristics have been of considerable
:st to linguistic students in other fields. Up to this time not a few of the languages
been fairly completely documented, while one or two have received special
tion. This is particularly so in the case of the work of Doke on Zulu, and of
in on Kongo, to mention only two cases in which such studies have reached a
level.
vertheless, there are still two things to be done in this field. No satisfactory
od of classification has yet been developed for this great number of clearly related
ages. In addition there is still no work of reference from which research workers
know where are the principal gaps in our knowledge. The recently published
)graphy of Bantu by Doke has given us a very useful book, in which is set
a considerable detail an account of the works so far published on the various
bers of the Bantu family of languages. As part of the framework of his monograph
used a broad classification based on certain features chosen for the purpose.
:he did not do, because it would have been outside the scope of the work, was
re us any complete picture of what is known about the whole family, or to make
reference to the many languages which have no documentation. To achieve this
necessary to survey the whole field in a different way, and that is what this present
sets out to do.
&e aim of this monograph then is twofold. It is intended first of all to establish
framework which may serve for future reference in identifying and classifying
i languages. Then in the second place, as an important by-product, it will throw
)rominence the places where our knowledge is fragmentary or even non-existent.
.e important thing to bear in mind when consulting this work is that the classifica-
s essentially tentative. In a few areas, where our data are reasonably adequate, the
)ing may lay claim to a certain amount of finality. In many cases, however, where
- sets of languages are known chiefly as names only, the classification is necessarily
mentala, and must not be taken as authoritative in any way.
some cases the conclusion reached from fragmentary data will probably be found
only partially correct. It is hoped, however, that the publication of this tentative
will stimulate any who are in a position to do so to contribute more complete
nation on the subject. In other cases little-known languages have been grouped
hkr r 1 If htAa,1A r1nnr^, 1.p -i"TT '--nnb- --:- ^ 4on-









which show that such grouping is unjustified. Here, too, the co-operation of !
workers will be appreciated, since it is desirable that in any subsequent edition
needful revisions should be made. In this way the setting out of all that cal
known from the available data, however inconclusive, will have served its purpose(

THE PLAN OF THE WORK
There are four principal parts to this study. In the first we shall investigate
questions arising out of the use of the term Bantu. Among other things this invc
establishing and illustrating the criteria to be used in identifying languages as mem
of the Bantu family. The following chapter is then devoted to a discussion of
various aspects of the problems of classification. Chief among these is the methc
be adopted in attempting to classify the Bantu languages. Then in Chapter IV
technique which has been described is applied, and its results shown in the form
series of descriptive classifications of each of the principal areas in turn. Finally t
is a map, together with a key in the form of a complete list of the languages class
in their groups.
One difficulty that had to be faced arose from the fact that some languages have 1
called by more than one name. This has been dealt with in two ways. In genera.
name accepted as correct is that used by the speakers of a language to refer t
Where the name has a class prefix, the usual practice is followed of omitting this pr
and where, as in the case of MWADGA (known as iciinamwaiga), the nan
preceded by the word iciina 'the language of the people of', this is naturally not t
The element -nya- which occurs in some names, e.g. NYADKOLE (called ol
yaikole), is retained, since its meaning is obscure, and sometimes it has bec
universally recognized as part of the word, e.g. it would be of small value to
NYAMWESJ by the shortened form MWES1. In the full list of languages accomp
ing the map, the other names which have been used for a given language are pl
in parentheses, and then at the end of the work there is an alphabetical index t
the names. It would clearly have been neither practicable nor useful to give all
spelling variants of certain names, so a peculiar spelling is only noted if it alters
position of the word in the index.
In the fourth chapter a system of numeration is developed by means of which
language may be referred to by a letter and two figures. Since this also enables
language to be found on the map, the numbering is given in parentheses after
name, each time a language is mentioned, even before the significance of the nun
tion has been explained.
ORTHOGRAPHY
In any study of the Bantu languages in general the problem of orthography becc
very difficult. It is clear that the considerations governing the designing of a prac
orthography do not necessarily operate in this case. On the other hand, to use
system of spelling that did not conform in most respects to the orthography based oi
'Africa' alphabet would be unsatisfactory, while to depart too much from the con
tional spelling of important languages would in itself reduce the usefulness of the v
One of the biggest difficulties in a general study of the Bantu languages a






THiE BANTU LAN(UA. iE 7
the existence of five-vowel languages side by side with the seven-vowel languages.
ae one hand it would be misleading to represent identical pronunciation differently
fferent languages, but on the other the spelling of one language can hardly be
mined by the characteristics of its neighbours. Fortunately the most serious
:t of this problem has been overcome by the recent development of the 'Africa'
ibet to include a system of nine vowel characters. This has meant that the two
characters j and i1 could be used for the extra close vowels of the seven-vowel
ages. As a result of this, those who only know the traditional five-vowel symbol
m in the seven-vowel languages of East Africa will easily recognize the spelling,
11 those who have no acquaintance with other than five-vowel languages. For the
i-vowel languages of the north and north-west, however, the spelling of the words
iad to be adapted. There can be little doubt that the seven vowel characters of
original 'Africa' alphabet are the most suitable for any practical orthography of
Languages, but clearly in a work which covers the whole of the Bantu field con-
icy must mean slight inconvenience in certain cases. Anyone using this work
has been accustomed to the open vowel characters will therefore have to interpret
spelling in this way: | = i, i = e, e = e, p = u, u = o, o = a. It must be
lasized, however, that the use of the new letters implies no suggestion that they
d be suitable for general use in these particular languages. As already stated, it
s reasonably certain that the use of the two open vowel characters is the most
factory thing for them.
other problem in a work of this kind is how to write fricative consonants. It
d not be practicable to use the special symbol in every case, especially as it
gently happens that the exact nature of the articulation is unimportant. For
Lple, in BEMBA (M.42a) there is only one voiced bilabial consonant, and the fact
it is fricative in a word like abantu 'people' would in no sense justify the use of
cial character. When we come to other languages, however, such as MBUNDU
c), there is the difficulty that an identical sound, which has an almost identical
in the language, is written 'v', e.g. ovandu 'people'. Such a spelling could not
sed in this work, but in view of its existence there does arise the necessity for
;ating that the sound in such words is a simple fricative bilabial. Purely as an
dient for our present purpose, therefore, and in no way supporting the undesirable
:ice of using such diacritics in current orthographies, any sound which is known
fricative will be represented by the character for the plosive underlined, e.g.
p (u in the 'Africa' alphabet), p = (f in the 'Africa' alphabet), k = x, g = -.
ie symbols c and j have been used consistently throughout to represent either
al plosives or simple affricates of the type tj, dg. Although this involves the use
e same character for quite different sounds, the fact is that in many cases we do
really know which of the two sounds occurs, and so it is convenient not to have to
aguish them in the spelling.
the case of I and 3 the problem is somewhat similar, except that here we do not
r whether these sounds are essentially distinct from s and z respectively. To use
,honetic symbols for them would frequently mean introducing an extra character
:cessarily, and would also obscure the relationship between words in different
ages. For this reason the symbols s and 7 have been adopted, since they will
cover the possibility of other palatalized fricatives such as q and 7.







The usual method of representing dental sounds presents several problems, ar
the whole seems unsatisfactory for our present purpose. As a tentative measure,
the fact that a sound has a dental pronunciation will be shown by the placing
cedilla beneath it. This means that 0 can be written , and in some cases this is
useful, especially where words with 0 in some languages appear with s in ot
Where 0 is not related to an s in other languages, as in KELE (A.73) the pho
symbol is retained for the dental fricative, but only in these cases.
The character rj for the velar nasal is required in the case of a number of langu
such as FAD (A.66), where it represents a sound that has to be distinguished fron
For this reason it has been used throughout as the nasal in compounds with
consonants, even in those languages where to do so in a practical orthography v
be an unnecessary complication. This is simply because it is desirable that si:
words should be spelt identically in different languages in a study of this kind.
The question of word-division hardly enters into the scope of this present 1
It must be pointed out, however, that the grammatical system which is presupl
throughout the work demands what would be called a 'conjunctive' system of wr
Consequently in none of the examples cited is any regard paid to the current pract
word-division in force in the language in question. This is not to be interpreted a
premature pronouncement on the matter, where a language happens to be written
or less disjunctively, but merely as an attempt to achieve grammatical consiste

DATA
Much of the information needed for this study has had to be derived from obs
tions made by other people. Inevitably this means that the reliability of the
collected in such a way is extremely variable. In some cases, too, further research
show that certain conclusions are invalid through their having been based on !
ments that were misleading rather than incorrect. Such, however, are the limit
due to the inadequacy of our present knowledge of the Bantu field.
In some respects, however, the present work may claim to have a certain distin<
ness, since I have been fortunate enough to be able to gather a very large propc
of the data at first hand. It so happens that the only area where I have had no per
contact with any of the languages is the one I call Zone A. Everything said about
languages of this region therefore has had to be taken from the work of others a
of indeterminate reliability. In much of Zone T also are languages concerning v
I have had to rely on what others have told me or have stated in their published w
The net result of this is that of the languages illustrated in the chapter dealing
criteria the only ones I have not personally studied on the field are FAD (A.66
RODGA (T.24), while in the chapter on differentia every example given has <
been obtained from or verified by native speakers of the language in question.
The map at the end suffers from the same disabilities as the rest of the work. I
effort has been made to check all the available data in order to fix the geogral
limits of each language, but our knowledge is extremely patchy. Some areas, su
Southern Rhodesia, which has been accurately surveyed by Doke, are well knowr
others such as the Portuguese colonies still need much research before we can be
that our linguistic maps are reliable. Taking it over all the map is probably about 5
cent. accurate, a figure which is almost certainly higher than that for any existing i








CHAPTER II
IDENTIFYING THE BANTU LANGUAGES
)RE the problem of classification can be discussed, it is clearly necessary to deter-
e what is meant by the term Bantu. Fortunately from the time this name was first
)duced it has chiefly been applied to linguistic rather than to ethnological facts.
; makes it quite suitable for our present purpose, and justifies our attempting to
ne it linguistically.
is to Bleek that we owe the term, which he spelt Ba-ntu in the first part of his
Iparative Grammar. When he produced the second part of this work seven years
Sin 1869, he dropped the hyphen, but still retained the accent, apparently because
considered that this was necessary for the correct orthography of Xhosa, from which
uage he adopted the term. From that time this name has become accepted for
remarkable family of languages spoken over much of central and southern Africa.
is interesting to reflect that Bleek did not attempt any close definition of the term
tu, contenting himself with what he termed the 'main distinctive features' of the
uages. All he did in fact was to point out that the specific languages he was examin-
had certain features in common, which were, to quote his words (the italics being
'a concord of the pronouns and of every part of speech, in the formation of which
louns are employed (e.g. adjectives and verbs) with the nouns to which they
ectively refer, and the hereby caused distribution of the nouns into classes or
ers'. Had he continued to study this family he would no doubt have given us some
;ria by means of which the languages belonging to it might be distinguished from
rs which do not. In his day, however, there was not enough known about the
uages of central Africa to make this a pressing problem.
ibsequent writers have been principally concerned with typical features rather
. with true criteria. The consequence of this is that few if any of them have defined
t they mean when they say that a given language is Bantu. It would be of small
e to cite in full the lists of the features as given by different people. Instead the
cipal references will be given first and then afterwards some comments.
after Bleek the next outstanding writer to interest himself in the Bantu family was
sius, in the introduction to his Nubische Grammatik, I880. He set out twelve
positionss which were intended to show the peculiar characteristics of the Bantu
ly. These were subsequently quoted in a somewhat condensed form by Cust in
Sketch of the Modern Languages of Africa, 1883. Long after this the propositions
epsius were considered to be an authoritative outline of Bantu criteria, being used
Verner in The Bantu Languages as late as 1919. The last and most important
ion of them is a critical one in Johnstone's great work, A Comparative Study of the
tu and Semi-Bantu Languages, 1919, to which those interested should refer
i 1891 Torrend published his Comparative Grammar of the South-African Bantu
guages, in which he treats at length of what he calls the 'characteristic features' of
family. Of his four principles, however, only the first has much value, and this is
>ly a restatement of the system of prefix concord. He gives nothing at all which
it help towards a delimitation of Bantu.
ve years later there appeared ,tudes sur les langues du Haut-Zambezie by Jacottet,






IO THE CLASSIFICATION OF
a work which contains a considerable introduction dealing with general ques
arising in the field of Bantu language study. Curiously enough, however, he make
attempt whatever to define what he means by Bantu, apparently assuming tha
readers already know.
Another interesting thing is that Meinhof, to whom the subject of Bantu phile
owes so much, does not seem to have attempted any definition of the term B,
Presumably he was content to take its meaning as being sufficiently clear.
Johnstone, in the work already referred to, sets out twelve propositions of his
'to define the special or peculiar features of the Bantu languages', at the same tin
showing why he does not accept the propositions of Lepsius. Although some o
features are of wider application than those of any previous list, yet they have a
weakness which arises from the inadequate grammatical and phonetic technique a
disposal of the writer.
Other writers have also dealt with this question, without making any impo
contribution to the subject, since they mostly reiterate what has been said previo
In 1935, however, Doke published his well-known Bantu Linguistic Terminolog
which he sets out eleven main characteristics of Bantu languages. This is a r
clearer statement than any that had previously been made, but here again the fea,
are not given as criteria, and so do not meet our demand for some means of deci
whether or not a given language is to be taken as Bantu.
Finally Tucker in his Eastern Sudanic Languages, 1940, gives a well-set-out s
ment of what he calls 'criteria... for comparison with Sudanic and Hamitic langua
In point of fact the seventeen characteristics listed have been chosen principal
fit in with the lists the author had already established for Semitic and Hamitic
guages, and as he notes 'there are other Bantu criteria which fall outside these se
+___ __in+r --A 1-11, --1A --,, cA-n- n-0A oonnAr-An tr n--n +- a V


exposition'.
One thing which becomes evident from
so far established any real criteria which can
it would fall within the Bantu family or n
more or less complete statement of the cl
scarcely any one of which is found to app'
accepted as Bantu. Thus, for example,
characteristics such as open syllables, but i
diyal 'stone', and cikas 'hand', while ma
forms as osal 'to work' and osip 'to strike'
that the final vowel has been lost through t
languages with words like these are eithe
Bantu family. If they are to be included, tl
Johnstone also says that 'No two consona
ing vowel, except one of them to be an aspil
in pronunciation'. That this is of little value
words which break these rules, as bgato ',
No. 64), and murro 'fire' in Copi (his I
ezziga) 'tear' in Ganda (his No. 4). Othe
are often mentioned, but it can be shown tl


ly of all these lists is tha
plied to a language to disc
[he most that has been
teristic features of Bant
all the languages which
writers emphasize cert
wunda (L.52) there are s
aguages in Zones A and
Sno answer to this probl(
Eluence of non-Bantu lan
be included in or excluc
ie open syllable cannot I
in come together without
r a nasal, and no consona
shown by the fact that he
' and sxwa 'termite' in
b) and edzigga (this in
netic features such as sti
LUBA-LULUA (L.3i)






PANTTT T.ANT('TTA(r1?B


za) there is no stress of any kind, and that in NY:
) there are no tones.
rious grammatical features are also cited, such as
erb species', but in Zones A and B there are lanl
nized Bantu characteristics, but neither of these
ie problem confronting us then is the establishir
)f the term Bantu can be defined. One of the pi
:al system within which the criteria may operate.
:for me to develop such a system, but the follow
iken to be an integral part of the system. Any E
here in the definitions will be due to this fact, the
spelled when the reader has studied the next ch

THE CRITERIA
is necessary to divide into two groups the criteria
es as Bantu. This is because there are some lanI
tion have to be postulated to such an extent that:
y some of the criteria. These are therefore placed
sidiary'. The use of this term is not, however, to I
lis group are less important, simply that they are
ere then is a bare list of the criteria. This is f
trate each of them in turn.
principall Criteria
A system of grammatical genders, usually at lea
) The sign of gender is a prefix, by means of whi
number of classes varying roughly from ten to
) There is a regular association of pairs of class
plural of the genders. In addition to these two-
class genders where the prefix is sometimes
prefixes occurring in a two-class gender, and
plural prefixes.
) When a word has an independent prefix as the
which is subordinate to it has to agree with it as
prefix.
) There is no correlation of the genders with i
clearly defined idea.
A vocabulary, part of which can be related by fi
mon roots.
Subsidiary Criteria
I A set of invariable cores, or radicals, from which]
agglutinative process, these radicals having the fc
r) They are composed of Consonant-Vowel-Con
') When a grammatical suffix is attached to the
on which words identifiable as 'verbals' are bt


he existence of 'object infixes'
ages which have all the other

of clear criteria by which the
requisites for these is a gram-
here is no space in this present
g suggestions for criteria must
ght obscurity that may appear
gh it is probable that this may
)ter.


to be used for identifying lan-
ages in which contraction and
becomes extremely difficult to
i the second group and labelled
taken to mean that the criteria
ess easy to apply.
lowed by examples chosen to



: five, with these features:
i words may be assorted into a
wenty.
s to indicate the singular and
ass genders, there are also one-
imilar to one of the singular
imetimes similar to one of the

ign of its class, any other word
Class by means of a dependent

x reference or with any other

-d rules to a set of hypothetical


st all words are form
g features:

1 there is formed a






THE CLASSIFICATION


(c) When a non-grammatical, or lexi
formed a 'stem' on which words
nominal belongs to a two-class gel
same in both classes.
(d) A radical may be extended by an e
elements, termed 'extensions', arc
a single vowel.
(e) The only case of a radical occur
verbals used as interjections.
4. A balanced vowel system in the rn
an equal number of back and front vow
EXAMPLES 4
I. The Grammatical Genders
(a) The Classes. It might seem rath
known as this. but even if only for the


ible as nommals are bL
sounds and tones of the

found between it and the
sed either of Vowel-Coi

hout a prefix of any ki

onsisting of one open v,


CRITERIA

rfluous to illustrate a fe
:ompleteness some exam
independent prefix, take


widely separated languages. For clearness the prefixes are shown separated by hyp
A. FAD (A.66). i. mu-r 'person'. 2. bu-nega 'women'. 3. m-bj 'c
4. mj-nlu'heads'. 5. a-kok'stone'. 6. me-ly'days'. 7. i-ku'skin'. 8.
'bones'. 9. n-dam 'width'. o1. u-nyp 'finger'.
B. BUBADGJ (C.zIb). i. mu-yjbj 'thief'. 2. ba-kunzj 'chiefs'. 3. mu-
'night'. 4. mj-luku 'hearts'. 5. 1I-kabu 'gift'. 6. ma-boko 'arms'.
lamba 'cloth'. 8. bj-sagga 'islands'. 9. n-tjna 'root'. o1. n-cete 'i
ii. lu-kulu'leg'. 12. bu-ljtu 'weight'.
C. ILAMBA (F.3I). i. umu-nuna 'brother'. 2. ia-suggu 'wives'. 3. u
gunda 'garden'. 4. imi-kono 'arms'. 5. j-kota 'tree'. 6. ima-lolo 'i
7. iki-muli 'torch'. 8. f|-ndolo 'potatoes'. 9. in-zila 'path'. o1. in-sj
'lions'. Ii. ulu-limi 'tongue'. 12. ika-gii 'small egg'. 13. jpi-tijla '
cloths'. TA_ uU-ta 'how'. iT. kui-6enda 'Poing'.


D. RUGURU (
'tree'. 4. ir
7. iki-sima
sabi 'ropes'.
snake'. 14.
E. BEMBA(M.
peni 'knife'.
7. ici-puna
'laughter'. i
pedes'. 14.
17. umu-nti
F. MBUNDU(
4. obi-tima
'thief'. 8. oi


.35). i. imu-ana 'child'. 2. iwa-lume 'm
li-gunda 'gardens'. 5. -bago 'grass'. 6. g
'well'. 8. ipfi-moka 'potatoes'. 9. im-buli
ii. ulu-kuli 'body'. 12. u-beho 'wind'. I,
iku-sona 'sewing'.
1za). i. umu-kasi 'wife'. 2. aba-londo 'fist
4. imi-fuko 'sacks'. 5. i-bala 'garden'. 6.
stool'. 8. ifi-lamba 'tears'. 9. in-supa'calat
i. ulu-limi 'tongue'. 12. aka-suba 'sun'. 13
ibu-lalo 'bridge'. 15. uku-tui 'ear'. 16. apa-
i 'enclosed place'.
R.II). I. u-lume'man'. 2. oba-pika'slaves
'hearts'. 5. e-limi'tongue'. 6. a-bele'breasts
n-bisi 'fish'. 9. olon-jila 'paths'. o1. olu-n
12. otu-balu 'horses'. 13. o-wuya 'heat'. i,






THE BANTU LANGUAGES 13
LODGA (T.24). i. amu-tiri 'worker'. 2. aba-fambi 'travellers'. 3. an-
iro 'work'. 4. ami-lambu 'rivers'. 5. a-lapi 'rag'. 6. ama-siku 'days'.
Safi-kosi'neck'. 8. apsi-komu 'axes'. 9. am-bilu 'heart'. io. atin-siba
feathers'. i1. ali-bambu 'rib'. 12. abu-kulu 'greatness'. 13. aku-famba
going'.
ice this is not a work on Comparative Bantu, no attempt has been made to
late the numbering of the classes in the different languages. Moreover, where
are other forms of the one prefix, as, for example, with a monosyllabic stem, these
ot shown. Similarly, since we are considering grammatical form and not ety-
gy, it is of no importance that the word otubalu (Mbundu Class 12) is apparently
lural of a loan-word okabalu 'horse' (cf. Portuguese cavallo).
2-Class and Class Genders. This feature has not received the clarity of treat-
in the past that it should have done. For this reason it must be adequately
rated. The following lists of the two types of genders are from the same seven
[ages, but it must not be assumed that they are exhaustive, since other genders
well exist in some of these.
AD (A.66)
2 mu-nega/ba-nega 'woman, women' 3. m-bon 'oil'
4 n-lem/mI-nlem hearts(s' 4. mi-ya 'entrails'
'6 a-lo/me-lo 'ear(s)' 5. a-kpma 'honour'
'8 i-to/bi-to 'cloth(s)' 6. me-kU 'blood'
'6 n-da/me-nda houses(s' 7. i-kfi 'strength'
5 u-non/a-non 'bird(s)' 8. bj-su 'dispute'
9. m-bjla 'speed'
zo. u-son 'shame'
JTRA nT-T (r ,lM\


2 mu-njiga/ba-njgga friends(s'


-----I--- ; _
li-bele/ma-bele
lj-kutu/ba-kutu
lI-ulu (pr. zulu)/i
i-loko/bl-loko 't]
i-yele/ba-yele 'b
n-kjlgu/n-kjqgi
n-kumbu/ma-nl
name(s)'
lu-boko/ma-bok
lu-pusu/m-pusu
bu-tali/ma-tali'


breastss'
smallpox pustule(,
an-ulu 'nose(s)'
ing(s)'
illet(s)'
'neck(s)'
umbu 'nick-

) 'arm(s)'
'skin(s)'
mnife(s)'


2. ba-tybI 'excreta'
3. mu-l}ka 'mercy'
4. mj-tukj 'sweat'
5. 14-kjndi 'aroma'
6. ma-loly 'jelly'
7. i-lunga 'innocence'
8. bj-luk}sa 'rust'
9. m-pjo 'cold'
10. n-golj 'sleep'
11. lu-bjku 'permanence'
12. bu-lulu 'bitterness'


2 umu-tepI/ia-tepI 'thief(s)' 3. umu-lamu 'quietness'
4 umu-gulu/imi-gulu 'leg(s)' 4. imi-garj 'blood'
6 j-klpa/ima-kppa 'bone(s)' 5. J-gulo 'sky'
8 iki-latu/jj-latu 'shoe(s)' 6. ima-kpta 'oil'






14 THE CLASSIFICATION OF
C. ILAMBA (F.3I) (contd.)
7/6 iki-kololo/ima-kololo 'cough(s)' 7. iki-kima 'female habits'
9/10 in-sjme/jn-sjme 'knife(s)' 8. 1J-gao 'bread'
I1/io ulu-tondo/in-tondo 'day(s)' 9. in-zala 'hunger'
i /6 ulu-tpmbi/ima-tymbi 'hair(s)' o1. in-guru 'power'
12/13 ika-nanso/jpj-nanso 'little girl(s)' ii. ulu-nkundi 'dust'
12. ika-uloa 'feeble love'
13. jip-sala 'insufficient wise
14. uu-kata 'laziness'
15. kuj-genda 'going'
D. RUGURU (G.35)
1/2 im-zungu/iwa-zungu 'European(s)' 3. gum-lopa 'blood'
3/4 gum-gulu/imi-gulu 'leg(s)' 4. imi-saiga 'sand'
5/6 -figga/gama-figiga 'egg(s)' 5. -woga 'fear'
5a/4 di-guku/imi-guku 'big fowl(s)' 6. gama-kala 'charcoal'
7/8 iki-goda/ipfi-goda 'stool(s)' 8. ipfi-pfuta 'little oil'
9/o1 igj-guku/tsig-guku 'fowl(s)' io. im-beho 'cold'
I1/Io ulu-limi/tsin-dimi 'tongue(s)' I. u-tulo 'sleep'
11/4 ulu-gogjgo/imi-gojgo 'hill(s)' 13. uku-zegga 'building'
12/6 u-lili/gama-lili 'bed(s)'
13/8 ila-mage/ipfi-mage 'small knife(s)'
E. BEMBA (M.42a)
1/2 umu-lume/aba-lume husbands(s' ia. makanta 'locust(s)'
3/4 umu-bili/imi-bili 'body(s)' 3. umu-lopa 'blood'
5/6 i-sembe/ama-sembe 'axe(s)' 4. imi-pembu 'refreshmen
7/8 ici-londa/ifi-londa wounds(s' 5. i-loba 'earth'
9/o1 in-soka/in-soka 'snake' 6. ama-saka 'kaffir-corn'
9/6 in-noggo/ama-logtgo 'pot(s)' 7. ici-ani 'grass'
Ii/o1 ulu-sato/in-sato 'python(s)' 8. ifi-basi 'leprosy'
11/6 ulu-kasa/ama-kasa 'sole(s)' 9. in-sala 'hunger'
12/13 aka-tende/utu-tende 'heel(s)' io. in-soni 'shame'
14/6 ubu-tanda/ama-tanda 'mat(s)' ii. ulu-bilo'speed'
15/6 uku-boko/ama-boko 'arm(s)' 12. aka-pumpu 'insolence'
13. utu-lo 'sleep'
14. ubu-luggu 'bead(s)'
15. uku-pita 'passing'
16. apa-ntu 'precise place'
17. umu-ntu 'enclosed place
F. MBUNDU (R.i1)
1/2 u-feko/oba-feko 'girl(s)' 3. u-sumba 'fear'
3/4 u-kolo/obi-kolo 'rope(s)' 4. obi-nene 'rubbish'
5/6 e-sala/a-sala 'egg(s)' 5. e-seke 'sand'
7/4 oci-lapo/obi-lapo 'paddle(s)' 6. a-loba 'mud'
8/9 og-golo/olori-golo 'knee(s)' 7. oci-sola 'love'






THE BANTU LANGUAGES 15
:BUNDU (R.ii) (contd.)
9 olu-sapo/olo-sapo 'fable(s)' 8. on-dulu 'gall'
I2 oka-pote/otu-pote 'coat(s)' 9. olon-jele 'beard'
6 o-wato/a-wato 'canoe(s)' io. olu-me 'dew'
6 oku-oko/oba-oko 'arm(s)' i1. oka-soggu 'bead(s)'
12. otu-ma 'clay'
13. o-wisi 'smoke'
ODGA (T.24)
2 amu-yibi/aba-yibi 'thief(s)' 3. an-talo 'abundance'
4 am-pimu/ami-pimu measures(s' 4. ami-saba 'earth'
6 a-boko/ama-boko 'arm(s)' 6. ama-golo 'greediness'
8 afi-lembe/apsi-lembe 'hat(s)' 7. aji-rami 'cold'
io am-buti/atim-buti 'goat(s)' 9. am-bere 'dew'
io ali-bala/atim-bala 'plain(s)' zo. atin-tjalu 'kindness'
6 abu-gamu/ama-gamu 'end(s)' ii. ali-sima'price'
12. abu-lolo 'laziness'
Nill be noted that the one class genders cannot be equated with any notional
ory like 'abstract' or 'substance', but that countable words like 'bead(s)' and
.t(s)' may be found in them.
)m the examples given it is clear that no attempt can be made to associate one
prefixes with the idea of 'singular' and another with 'plural'. This is not only
se of the one-class genders but also because of the fact that one class may be
I in one gender and singular in another, as in FAD 5/6 and o1/5.
Agreement by Dependent Prefixes. To illustrate this criterion two sentences are
from each of the seven type languages. The English equivalent is the same in
case, and one of the sentences is the plural corresponding to the other. The
ing is 'his other knife is lost' and 'his other knives are lost'. In this case also the
es are separated by hyphens simply to throw them into relief, though they are
ally an integral part of the word.
D (A.66) u-keij by-a m-box u-jgazime
a-keg dy-a a-vox e-ggazime
BADGI (C.2ib) bu-tali bu-yiyi bu-sjsu bu-uljmbana
ma-tali ma-yiyi ma-sisu ma-ulimbana
AMBA (F.3I) in-sime y-akwe y-iggi i-11milile
in-sjme z-akwe z-iggi zi-lilimele
fGURU (G.35) gum-mage gw-ake m-yage gw-agire
imi-mage y-ake mi-yage y-agire
MBA (M.42a) umu-peni u-akue u-mbi naa-u-luba
imi-peni i-akue i-mbi naa-i-luba
3UNDU (R.ii) om-moko y-ahe yi-kwabo y-anyelela
olom-moko by-ahe bi-kwabo by-anyelela
IDGA (T.24) amu-kwa w-akwe wu-jwana wu-lalekile
ami-kwa y-akwe yi-nwana mi-lalekile






16 THE CLASSIFICATION OF
The most important feature of this prefix agreement is simply that there
dependent prefix corresponding to each independent prefix. The term 'allitel
concord' has frequently been used, but from these few examples it will be seer
this is hardly an adequate description of the facts. It is therefore preferable not t
it, but simply to speak of agreement by prefixes.
(d) Absence of Correlation of Genders and Ideas. Since this is a negative crit
it is difficult to illustrate. Here, however, are some examples from BEMBA N
will demonstrate that there is not necessarily a correlation even between gende
the idea of person.
Here are a few typical examples chosen from the large number that might be g
tumba umutali 'a long skin bag'
baatumba abatali 'long skin bags'
This shows that the prefixes umu-/aba- are not confined to personal reference
isilu ilitali 'a tall madman'
amasilu ayatali 'tall madmen'
icibambe icitali 'a tall hunter'
ifibambe ifitali 'tall hunters'
These show that other genders may also refer to persons, and so cannot be said
confined to things and non-personal living creatures.
2. The Related Vocabulary
To illustrate this criterion adequately would require a treatise on Compal
Bantu, but its importance and application may be seen in the following examples
MFINU (B.4I). This language is remarkable in having upwards of twelve si
vowels, including three u-sounds and three o-sounds. Here are examples of I
containing these sounds, together with the starred forms of common Bantu to N
they can be related.
With very close u ndu 'pepper' *-LUNGU (cf. BUBADGJ: ndurJI
bbu 'beach' *-BUNGU (cf. BUBADGI: ljbur
With open u mpu 'rat' *-PUKU (cf. KODGO: mpuku)
ndu 'brother' *-LUGU (cf. KODGO: nduku)
With very open u ntu 'chest' *-TULU (cf. BUBADGI: ntulu)
ndu 'bile' *-LULU (cf. BUBADGI: ndulu)
With close o oko 'to pull' *-KOK- (cf. KODGO: koka)
olo 'to bewitch' *-LOG- (cf. BUBADGI: loko)
With open o obo 'to rot' *-BOL- (cf. BUBADGj: bolo)
+tn 'l1ppn' *-TnT.n f-f T-rnn '.( tln\









e first and last words of the singular sentences given above may be taken in order
istrate the first of the subsidiary criteria. To throw the radicals into relief they
it in the upper case, while the grammatical suffixes are separated by a hyphen.
tensionss are in the lower case but may be known in the verbals in that they are
:parated from the radical.
FAD (A.66) uKED ujgaZjMe (i.e. Ext. -e)
BUBADGI (C.21b) buTALi buuLJMBan-a (i.e. Ext. -an-)
ILAMBA (F.3I) inSIMe iLIMil-Ile (i.e. Ext. -il-)
RUGURU (G.35) gumMAGe gwAG-ire
BEMBA (M.4za) umuPENi naauLUB-a
MBUNDU (R.II) omMOKo yaNYELel-a (i.e. Ext. -el-)
RODGA (T.24) amuKUa wuLALek-ile (i.e. Ext. -ek-)
: example from FAD shows the difficulty of applying subsidiary criteria, since
r of the words in this language has a suffix. In spite of this, however, the radical
structure similar to those of the other languages.
illustrate still further the way in which the radical occurs in words of both
here are some groups of words from five of the languages.
D: -L"JK 'marry' (with grammatical suffix, -LVG-a)
-LUGe 'arrange marriage of' aLI_K/meLLK marriages(s'
BADGI: -BAL- 'marry' -BALjs- 'give in marriage'
-BALil- 'assist at marriage of'
1jBALa/maBALa marriages(s'
muBALili/baBALil4 assistants() at marriage'
MBA: -TOOL- 'marry' -TOOLu- 'be married'
-TOOLjsj- 'arrange marriage of'
umuTOOLj/iaTOOLI 'bridegroom(s)'
umuTOOLua/iaTOOLua 'bride(s)'
uuTOOLj 'marriage'
1BA: -UP- 'marry' -UPu- 'be married'
-UFi- 'arrange marriage of'
iciUPo 'marriage' ubuUFi 'married state'
JNDU: -KUel- 'marry' -KUelis- 'arrange marriage of'
-KUeliu- 'be married'
oloHUela 'marriage'
main thing to note from these examples is that either of the consonants of the
may be variable within the one group of words built from it. Thus in BEMBA
il consonant is sometimes -P- and sometimes -F-, while in MBUNDU the
given has -K- or -H- as its first consonant.
example from FAD demonstrates that though the subsidiary criteria may be
t to apply, they are not entirely inapplicable in a language of this kind. In such
is usually possible to determine the radical, but the large number of words
t a suffix often creates problems in the distinguishing of verbals from nominals.






'TIHE CLAbbiHUlIC1UIN Uk


from suffixes, as may be seen in the above example.
The importance of the criterion of the invariability of the stem of nomin
two-class genders can best be illustrated in a negative way. There is an obscure
guage called Ndabe spoken near the Bantu frontier in the Cameroons. It has
features which resemble those usually laid down for Bantu languages, but many
words behave like these:
nyu/nyol 'knee(s)' mfo(-)/mfo (-) 'slave(s)'
fo/fal 'head(s)' ke( )/ke (-) fingers(s'
Thus in spite of the fact that there is a kind of grammatical agreement be
these words and the numerals, the language cannot be accepted as Bantu.
Some difficulty may arise with respect to the criterion that only verbals u:
interjections are entirely without prefix. This occurs in a number of langua
Zone A which have zero prefix in some cases. Thus in DIMA (A.72), for exm
the personal prefixes in the plural are l1- for the first and second persons and b
the third person. In the singular, however, all three persons have zero prefix,
might be mistaken for a case of a word with no prefix at all: e.g. mwana yiml
child is singing'.
Up to the present no language accepted as Bantu has been found in which v
in simple sentences are always without a prefix, and indeed this would appeal
completely foreign to Bantu structure. In those languages where it might be ,
that the sign of agreement with its subject is a self-standing 'pronoun', their
increasing tendency always to use this, in such a way that it is debatable w
or not it should be separated from the verbal. For example, in BASA (A.44) m(
or mepam 'I arrived'.

4. The Balanced Radical Vowel System
To illustrate the Bantu vowel system here are two complete series of radical
from a seven-vowel language and one from a five-vowel language. The tonal beh
of the radicals is identical throughout, and no other radicals exist with these cons
and this tonal behaviour.
BUBADGI (C.zib): -BAL- 'count', -BJL- 'follow', -BIL- 'border', -BE:
annoyed', -BVL- 'be numerous', -BUL- 'break', -BOL- 'rot'.
BEMBA (M.42a): -LAL- 'crack', -LIL- 'cry', -LEL- 'rear', -LUL- 'become 1
-LOL- 'have the eyes open'.
What was said about MFINU in the paragraph on the related vocabulary i
sufficient to show how it may be practically impossible to apply this criterion t
languages.
LANGUAGES WHICH ARE INCOMPLETELY BANTU
As might have been expected, there are languages in which some of the c
we have established hold good, but some do not. An example of one kind o
a partly Bantu language is seen in BJRA (D.32). In this language the voca
relationship referred to in the second criterion is quite clear, as also the stn
features and vowel system of the third and fourth criteria. The grammatical






THE BANTU LANGUAGES 19
and gender system, however, is distinctly fragmentary. There is, for example,
one type of prefix alternation between singular and plural nominals, mbuhu/
hu persons(s), qkama/bakama chiefs(s). The greater part of the nominals are
able, although the element ba- may be prefixed as a sign of the plural, e.g.
/bakima things(s), kiboko 'arm(s)', ganj words(s). The system of agreement
:ans of dependent prefixes is equally incomplete. Words like mbuhu govern
nents of this kind, mindo marjkina 'this other', while those like babuhu
i these agreements, bendo bagkina 'these others'. The remainder of the
, however, govern the same kind of agreement, whether they are singular or
,and whether they have the prefix ba- or not: e.g. kjma lake lando maijkina
father thing of his', bakima lake lando maqkina 'these other things of his'.
are only two verbal prefixes, a- which is usually singular, and ba- which is
y plural.
hough languages of this kind cannot be called Bantu owing to their not having
Implete prefix system we have described as a criterion, their relationship to the
Languages is sufficiently close for them to be taken into account. We shall
ore call them'Sub-Bantu'. Some of the lingua franca languages like MA! GALA
1) are in this category, since they have well-defined two-class genders, like the
lantu languages, but little or no prefix agreement in dependent words.
second type of language which obeys only some of the criteria is found in the
roons and south-eastern Nigeria. These languages obey the first criterion but
Le others. That means that while they have a system of grammatical genders and
nents operated by means of prefixes, they show little or no relationship of
ulary with full Bantu languages. In addition they do not display even the rudi-
; of the structural features laid down in the third criterion; moreover their vowel
a is frequently complicated. An example of this may be seen in BAFUT, a
age spoken in the British Cameroons near the Bantu frontier. There are such
rs as mujgw/bugjw 'knife(s)', azo/njo things(s), ati/iiti 'tree(s)'. There are
agreements as mugjwj kpla 'this knife', buqwj bpla 'these knives', niiljhj npla
-ye', mijh myla 'these eyes', jy gyla 'this man', bo bpla 'these men'. This
of grammatical behaviour is definitely reminiscent of what happens in the lan-
s we have accepted as Bantu, but it cannot be used to establish any clear relation-
such as exists in the case of the Sub-Bantu languages. We shall therefore adopt
:rm 'Bantoid' to describe any language that has a system of prefix genders and
ments of this kind without any other Bantu features.
view of the great difference between these two kinds of partially Bantu languages,
Natural to treat them differently. The Bantoid languages are therefore not dealt
it all in this work, but the Sub-Bantu languages are placed within the scheme of
fiction, being distinguished by the use of italic type.









CHAPTER III
METHODS OF CLASSIFICATION
WE now have to consider the answers to the question, 'How are the Bantu langi
to be classified?' For our present purpose we may ignore the method used by (
which has little or no linguistic basis, being merely geographical. There are, howi
other systems of classification which are based on linguistic facts. These ma
roughly divided into three types: (I) The Historical, (2) The Empirical, (3)
Practical. We shall take each of these in turn.
I. THE HISTORICAL METHOD
It was Meinhof who advocated this method of classification, principally as an
come of the technique of comparative Bantu phililogy he originated. Briefly it w
involve the establishing of a genealogical table for the language family. From tl
would then be possible to assert that the members of a given group had sprung J
a common ancestor, which was itself a late descendant of the common parent c
Bantu.
There is no need for us to discuss here the implications or the merits of this t
nique, since the likelihood of its being able to produce results is so remote. S
useful deductions may be made about the sounds of the hypothetical common BE
but in a field with practically no historical records, true historical study, as dis
from comparative study, is impossible.
From what Meinhof himself says in the chapter on classification in his Introdu
to the Phonology of the Bantu Languages, one may suspect that he realized the na
of the problem involved in the technique he advocates. In fact this is the least
vincing chapter in the whole work.
2. THE EMPIRICAL METHOD
The essence of this method consists in the drawing of isoglosses on the ma
order to show the distribution of various linguistic features. Then if several ol
isoglosses coincide, this may be taken as the boundary between different language a
In practice, however, it means that certain differentia are chosen, and then used i
attempt to divide up the area covered by the Bantu languages. Since this meth<
the one that has been implied in most of the classifications made so far, it mus
considered in some detail, both as to its implications and as to its results.
First then as to the method itself. What are the linguistic differentia that ca
used for plotting isoglosses? The following list includes every type that may L
used.
(a) Lexical, i.e. as to differences of vocabulary.
(b) Grammatical, i.e. as to differences of form and sentence structure.
(c) Phonological, i.e. as to differences of distinction between sound units.
(d) Phonetic, i.e. as to differences in the actual sounds of speech.
(e) Tonal, i.e. as to differences in the tone system.
We must then study each of these types of differentia, and try to show what re
rnn h1 h rhPVxpr hU their Ii q






THE CLASSIFICATION OF THE BANTU LANGUAGES 2z

,exical Differentia
iere are two kinds of lexical differentia. One is concerned with the distribution
tual words and the other with the distinctive features of the words in the voca-
7.
ie first of these kinds was used chiefly by Johnstone, who has shown that some
interesting results can be achieved by means of it. If, for example, we plot the
rrence of these five words for 'house', *NYUMBA, *NGANDA, *NJUBU,
AGU, *NJO, we get well-defined areas. Unfortunately, however, as had
dy been pointed out by Meinhof, isoglosses obtained in this way from one set of
s rarely coincide with those obtained from another set. Instead of plotting the
where the different words for one idea occur, it is also possible to plot the limits
e distribution of certain widely occurring words which can all be related to one
non starred form. Thus, for example, the isoglosses for the occurrence of
,ET- 'bring', *-LEK- 'leave', *-LIM- 'cultivate', *-TUUL- 'put down',
JAL- 'wear', *-BJAL- 'give birth to', *MULJ 'village', all show a remarkable
arity in that none of these roots is represented in the languages north-west of a
from Benguella through Leopoldville to Stanleyville. Experiment proves that
second method of plotting isoglosses is much more useful than the first.
ie of the difficulties in any system of classification based merely on words is that
is rarely a sudden break in the distribution of vocabulary, apart from occasional
like the one just given in the previous paragraph. A study of the fragmentary
-lists given by Johnstone shows that adjacent languages frequently have vocabu-
s that are similar, and it was this fact that led him to base his classification largely
ords. It is also an unfortunate fact that all too often the sum total of our know-
- of a language is contained in a list of words, and this frequently of doubtful
)ility. The important thing for our present purpose, however, is that any system
ossification based on words has an essential weakness. This inheres in the fact
vocabulary can so easily be borrowed by one language from another without
lively bringing the two languages closer together.
iere is only one satisfactory way in which the closeness of the relationship between
ages can be determined by means of word-lists. This involves the application
e technique of comparative study based on the use of hypothetical starred forms.
ly it consists in the compilation for each language of a 'standard' vocabulary of
t a thousand distinct words of common occurrence. That the words should be
ict simply means the avoidance of duplication, thus if -lul- 'become bitter' and
a 'bitter' both occur in a language, only one of them may be put into the standard
bulary.
hen such vocabularies have been made for many languages, it will usually be
d that just over three hundred of the words in a given language, i.e. about 30 per
, can be related to corresponding words in at least two other languages by means
e comparative method, by which starred forms may thus be established.
Hiking then one such standard vocabulary, and comparing it with that of another
iage spoken at a considerable distance, we shall usually find that the proportion
e vocabulary common to the two falls far below the 30 per cent. By taking other
IaPAQ !t I;'tonrP Q hPrnm;rnr r1rM rPnlTr PA nXrP QnmP.timp + inA- tFhA+ alnn-rr .-







22 THE CLASSIFICATION OF
certain line there is a sudden jump in the percentage of related words. This line
then be taken as part of an isogloss. The important feature of this method
insistence on fixed rules of relationship to the starred forms in all cases, since
reduces the likelihood of misleading figures due to loan-words.
Lexical differentia of the second kind, which have scarcely been considered
up to this time, consist mainly of features which have a phonological or tonal si
chance. The isoglosses provided by these features give much more interesting r
than those obtained by plotting individual words. Since, however, we shall be s
ing characteristics of this kind in sections (c) and (e), it will be preferable to
them till then. There are a few other minor features which can be used, such
following.
(i) The existence or not of any partial correlation of certain genders with notii
relative size. For example, in LUBA-LULUA (L.3 ) there is a gender ka/tu
contains such words as kasoko/tusoko 'small villagess' related to musoko/mi
villages(s), even though a large number of words in this gender have no referee
small things, e.g., kapia/tupia 'fire(s)'. In BUBADGI (C.zib), on the other
there is nothing corresponding to this.
(ii) The existence or not of regular types of nominals related to the radic
verbals. For example, in KODGO (H.i6f) there are many kinds of nominals r
to -sumb- 'buy', including nsumbi/basumbi 'purchaser(s)', nsumba 'man
purchasing', nsumbo/nsumbo 'act(s) of buying'. MFINU (B.4I), on the other
has no related nominals of this kind.

(b) Grammatical Differentia
These are some of the most useful differentia for classification, but the only
who has made any serious use of them is Doke, and even he introduces then
subsidiary way. There are naturally many things that could be included under
heading and it is found that the isoglosses given by the more important of
frequently tend to coincide. Here are some of them, including those which cou
have been used previously through lack of data.
(i) The existence or not of double nominal prefixes. For example, BEMBA (1
has umuntu 'person', while LENJE (M.61) has muntu 'person'.
(ii) The existence or not of the extra independent prefixes *PA-, *KU-, *]
For example, LUBA-LULUA (L.3i) has panzubu 'on the house', kunzubu'to
the house', and munzubu 'in the house', while TETELA (C.71) to the imm
north-west has nothing corresponding to this type of grammatical form.
(iii) The existence or not of nasal consonants in the dependent prefixes.
example, in KODGO (H.i6f) there is ntima mieto 'our hearts', but in NDO
(H.2I) mijima ietu 'our hearts'.
(iv) The method, if any, of modifying the prefix of a nominal when it is use
sentence. For example, in each of the three following languages 'oil' is amafut:
'it is oil' in NYIHA (M.23) is mafuta, in TOTELA (K.4I) maafuta, and i
(D.66) nimafuta.
(v) The use or not of the suffix *-E in dependent tenses. For example, 'let us
in MillnnC'n)nTZTITN TT (C 6,r ia tntiump hut in RTTRADCT I(C 9h) it is tut






I 1112, 1)i IN I L) L111iiN1'u.T L Lik-2.J z 3
) The use or not of suffixes other than *-A in principal affirmative tenses. For
iple, KIKUYU (E.5 ) has -jre and -iite, whereas WUNJO (E.6zb) has nothing
*a.
i) The system of tense signs. These vary enormously over the whole Bantu
but similar features may be found within a group. For example, in Group E.30
is the unusual tense sign *-AKA- -E used to refer to events in the immediate
e.
ii) The structure of negative statements. There are certain well-defined types
rmal negatives in Bantu languages; for example, BUBADGI (C.zib) uses an in-
ndent negative particle at the end of the sentence, while DGOMBE (C.31) has
ial negative tenses containing negative elements.
.) The use or not of infixes as substitute objects of verbals. For example, 'he
me' in TIO (B.35) is abara me, but in MFINU (B.4I) ambara.
1 The method of constructing relative clauses. There are several of these; for
iple, in SUKUMA (F.21) we find kinhu ikiagua'the thing which has fallen', but
OGO (G.Ii) icinhu ciono ciagua 'the thing which has fallen'.
Phonological Differentia
atures of this kind have not previously been used for purposes of classification.
r are, however, very useful and produce some interesting isoglosses. The defini-
of this type of differentia as given above was: 'differences in the distinctions
een sound units'. Since, however, the term 'phonology' has been used in more
one sense, it is perhaps important to illustrate fairly fully the characteristics to
h it refers here.
ie phonological differentia are not concerned with the actual sounds used in
:h, nor with the deduced 'sound-changes' studied in comparative Bantu philology.
example, in Swahili -kerjg- 'deceive' and -cegg- 'lop' are different radicals,
-eas in Bemba -cerjg- 'treat unfairly' is not distinguished from -kegg-. At this
t we are not interested in the fact that the -c- in the Swahili radicals corresponds
i -s- in Bemba (as Swahili -cek- and Bemba -sek- 'laugh') since that belongs to
parative study. Neither does it matter whether the -c- of Swahili is identical in
unciation with the -c- of Bemba, since that is a phonetic question. Phonologically
significant thing is that there is an alternance c/k before -e- in the first consonant
ie radical in Swahili but not in Bemba.
other illustration of this kind of characteristic may be seen in the occurrence
1 alternance 1/r. In TIO (B.35) -kal- 'dwell' must be distinguished from -kar-
ied', whereas in KODGO (H.i6f) -kal- 'dwell' is not distinct from -kar-. This
is that there is an alternance 1/r in the second consonant of the radical in Tio but
n Kojgo.
milarly in NYANJA (N.3 a) there is an alternance l/d in the first consonant of
radical, but in YAO (P.zi) there is not. For example, in Nyanja -lul- 'froth up'
o be distinguished from -dul- 'cut off', but in Yao -lul- 'froth up' is not distinct
S-dul-.
part from the alternances in the consonants of the radical, which are too numerous
it, there are also other types of phonological differentia. Of the selection given in
following list some relate to alternances in the sounds of lexical elements, and






24 H1 CL0Ab~rA1tILAII I UOI Uo
others to those of grammatical elements, but it is simplest to deal with them all ui
this heading.
(i) The existence or not of alternances i/j and u/p in the radical vowel.
could be expressed differently as an alternance of seven as against one of five vi
sounds in the radical. For example, RUHIJ (P.12) distinguishes between the vo
in the radicals -jmb- 'swell' and -imb- 'sing', whereas YAO (P.2I) -imb- 'swe
not distinct from -jmb-.
(ii) The existence or not of an alternance of vowel quantity in the radical.
example, BEMBA (M.42a) distinguishes the vowel in -fum- 'go out' from thl
-fuum- 'drizzle', but MWADGA (M.22) does not distinguish -fum- 'go out' I
-fuum-.
(iii) The existence or not of alternances i/e and u/o in suffixes. For exan
in KIKUYU (E.5I) ndjhu 'spray' has to be distinguished from ndjho 'big c
whereas in BUBADGI (C.zib) Jkjkjgu 'neck' is not distinct from qjkjgo.
(iv) Whether or not the alternances in the consonants of the extensions are sir
to those in the consonants of the radicals. For example, the alternance s/z occu
the first consonant of the radical in both GANDA (E.15a) and KODGO (H.I6f),
Ganda distinguishes -sin- 'nauseate' from -zin- 'dance', and Ko1go distingu
-sin- 'be deep' from -zin- 'be burnt'. On the other hand, Ganda has a si
alternance in extensions, while Kongo has not; thus in Ganda -lamus- 'greet' h
be distinguished from -lamuz- 'bargain with', 'whereas in Koigo -sekes- 'wh
not distinct from -sekez-.
(v) Whether or not there are alternances in the vowels of extensions similar to 1
of the radical and those of the suffix. For example, LUNDA (L.52), which ha
alternances i/e and u/o in radicals but not in suffixes, has similar alternanco
extensions. Thus though mukonu 'leg' is not distinct from mukono, -saluk- 'l
out in a rash' must be distinguished from -salok- 'be restless in sleep'. Other
guages like LUBA-LULUA (L.31) have no such alternance in the extensions.
(vi) Whether or not there is a similar alternance of quantity in the vowels o
extensions and those of the radicals. For example, both BEMBA(M.42a) and KO1
(H.I6f) have an alternance of quantity in radicals, since Bemba distinguishes -
'grow up' from -kuul- 'extract' and Koijgo distinguishes -kul- 'grow up'
-kuul- 'liberate'. In extensions, however, Bemba has a similar alternance of qua
whereas Koggo does not; thus -pelel- 'sow broadcast' in Bemba is distinct
-peleel- 'almost arrive', but in Kojgo -kelel- 'filter for' is not distinguished
-keleel-.

(d) Phonetic Differentia
Characteristics of this kind have been used more than once in attempts at class
tion, but these attempts have frequently failed through the inaccuracies in the ava
phonetic data. Moreover, data of this kind are not of much value in certain cases
similar phonetic processes are to be found in many languages outside the I
family. Thus, for example, the fact that the syllables *KI and *KE are pronot
ci and ce in the languages of Group M.50 means little, as the same rules holds gc
some European languages, such as Italian.






THE BANTU LANGUAGES 25
:re, however, are a few phonetic features which occasionally produce some useful
)sses.
The existence or not of vowel length at an internal vowel junction, or before a
compound. For example, neither LENJE (M.6I) nor NYANJA (N.3Ia) has an
iance of quantity in the vowel of the radical, but whereas in Lenje muana 'child'
bounced mwaana (when it is the subject of the sentence), in Nyanja a similar
is pronounced mwana.
The existence or not of phonetic prominence on certain syllables, such as radical
sity, or penultimate vowel length. For example, in BUBADGI (C.zib) the two
wing words both have radical intensity, which means that in the first the promi-
a is on the first syllable, and in the second on the second, LOKolo! 'emulate!'
oKOLo 'cock's comb'. In BEMBA (M.42a), on the other hand, the pronuncia-
of the two following words is identical, showing that there is no radical promi-
e, ukuFISulula 'to bring out of hiding', ukufiSULula 'to invert them (i.e.
,s)'.
) The behaviour of junctions involving a nasal consonant. There are many
rent types of such behaviour, including even a reduction of the number of
dances represented in the pronunciation. For example, YAO (P.21) has identical
unciation of nTEKe 'let me draw (water)' and nLEKe 'let me leave', both being
J a 1- -- L 1 '- 'I WDA/ A /1%/r .--\ .1 ,--- :- ^l' 1:1-- 4.1-:- T-, "D -- I,- --









>f tonal behavior


(ii) The existence or not of an alternance of tone on the nominal suffix. For exa
in BUBADGI (C.zib) nominals with a dissyllabic stem fall into four tone-gi
which may be represented by these type words, mbindu 'dirt', rjgandu crocc<
mpamba 'nothing', nduijgu 'pepper', whereas in KODGO (H.i6f) all such
fall into two tone-groups only.
(iii) The existence or not of an alternance of tone on extensions in nominals
example, in BUBADGI (C.2ib) there is a difference in the tone-patterns of the,
words, mukilill ( __-) 'agent', and muliggil (__-_) 'calmness', whereas in VE
(S.II) a difference of this kind does not exist.
(iv) The existence or not of a tonal alternance in dependent prefixes. Thel
several types of such alternances, so it is only possible here to select one as an illi
tion. In SUKUMA (F.2I) the tone-pattern of the verbal is different in the
following cases, burj ikufpma (-_ ---_ ) 'the goat will go out', burj jikul
(-- ---) 'the goats will go out'. In HA (D.66), however, there is no difference
tone-pattern of the corresponding verbals.
(v) The use or not of tone-patterns determined by the syntactical relations
words. For example, in MFINU (B.4I) the tone-pattern of the first word is dif
in the two following cases, leburu lemo (_ -_) 'one tribe', leburu leso (__
'that tribe', whereas in TIO (B.35) a similar word libura 'tribe' has only one
pattern whatever its context.

The Results of Applying the Differentia
From the list of differentia just given it might be thought that if these were
for plotting isoglosses, the classification of the Bantu languages would be a s
matter. Unfortunately, however, this is far from what is found when the attend
made. A number of the differentia in the list were described as 'useful' and this
no means incorrect. The real difficulty lies in the fact that none of the different
be applied consistently over the whole of the Bantu field. This is not through 1,
data, nor is it because the differentia may sometimes be inapplicable, as, for exa
when languages like BUBADGI (C.2Ib) do not use nominals as sentences and ,
one in section b (v) cannot be applied. What actually happens is that each c
characteristics gives good results in at least one area but not in another.
If the empirical method of classification is to be used consistently each c
differentia should be equally valid wherever applied, otherwise the method coll,
An example or two will make clear the impossibility of the regular application c
method. The extreme case of its failure occurs when an isogloss provided by a
feature coincides with others in indicating a useful division between languages i
part of the field and then cuts right through the middle of a single language else
For example, the use of prefix ku- in place of the prefix tu- for the first person
agrees with several other differentia in dividing the languages of Group E.3o
those of Group E.io. When we try to apply the same distinction farther south,
ever, we find that it marks a boundary right through the middle of NYAMN






THE BANTU LANGUAGES 27
) where ku- is used in the south and tu- in the north, but in this case the isogloss
itary, and so cannot be used.
nilarly the existence of a seven-vowel system in radicals usually distinguishes a
age from a neighboring one with a five-vowel system in such a way that the
ages are found to belong to different groups, e.g. Group C.zo has seven vowels,
1.30 and B.4o five, F.zo has seven and E.zo five, S.zo has seven and S.30 five.
he isogloss arising from this difference is almost the only one separating NYIHA
3) from MALILA (M.24).
r conclusion on the empirical method then is that it cannot be used without some
fiction, and this question will be discussed in the next section.

3. THE PRACTICAL METHOD
ie present work follows a practical system of classification. This means that the
nee of some arbitrariness is admitted as an essential modification of the empirical
od. It is presumably because others have not made this admission that no
actory classification has yet been achieved. Let us at this point then see exactly
we are trying to do.
.e relationships of the languages we accept as Bantu by applying our criteria are
iently clear to make some classification possible, while the number of the languages
i in this family makes classification almost essential. There are two ways in which
:an be undertaken. We may begin with the whole field and try to put in some
daries to divide it up into a number of large sections. Then by a process of sub-
.on we may go on until we reach the smallest useful unit. This classification by
lentation has been the technique adopted by almost all those who have previously
ipted to classify the Bantu languages. There is, however, a quite different
)ach which may be used. Suppose we take the individual language as our
ng-point, and then move outwards from it. It is possible in this way to group
her with the language from which we started other adjacent languages which
ay similar characteristics. At some point we shall decide that we have moved into
ier group. The decision will, however, have an element of arbitrariness in it,
ise although we shall be able to assert that the group displays a certain set of
non characteristics, the distribution of any of these may not be coextensive with
roup. The arbitrariness lies in the exact set of characteristics we choose. If a
restricted set were chosen the group would be larger, and conversely if a wider
rere chosen the group would be smaller. Moreover, it may sometimes happen
although a given set of characteristics is displayed by a group, in one member
lost important of these is missing, yet in all other respects it is clearly necessary
it should be included in the group.
the classification described in the next chapter it is this second technique which
*e followed. This means that the whole of the Bantu field is shown as consisting
number of groups of varying size and closeness of relationship. Since there are
y eighty of these groups, some larger unit is clearly necessary for ease of reference.
lace the groups in sets, a similar method is employed in which an arbitrary blend
aracteristics is made. Naturally the validity of these larger units is less than that
e groups, but still the prevailing consideration in the establishing of such sets is
istic. It nroves convenient to make sixteen of these .ete which arp r all-d 7nnoa






28 'IlH, ULAObbltlUAlIUIN UP ItEi UAINIU LiAINjUAUt,0
By zone, therefore, is to be understood primarily a set of groups which have a ce
geographical contiguity and which display a number of common linguistic feature
well.
The use of the term 'arbitrary' must not lead the reader to think that the classic
tion is capricious. As already stated, the arbitrariness consists in the choice o:
differentia to be used in each case, and not in dispensing with the use of differe
Thus, for example, at the meeting of the three Zones B, C, and H there is a bunc
of so many isoglosses that any system of classification would have to place impo
boundaries here. But the net result of this practical system is simply that in citin!
characteristics of any group or zone we can only state them broadly, and that in a
fact there will usually be some important exceptions. This, however, can hard]
held to be a defect of the system, but rather an inevitable outcome of the facts.
are faced with a situation in which we either have to introduce some element of
trariness or give up all attempts at classification.
What is claimed for the present work is that by taking into account as many fea-
as possible the arbitrariness is reduced to a minimum. Moreover, since it is avov
practical in its intention, similarities between widely separated languages are of
importance, and we shall avoid the kind of grouping suggested by Torrend
Jacottet which places together SUTHU (S.23), MAKUA (P.3I), and MPUD(
(A.7ia).









CHAPTER IV
THE BANTU LANGUAGES CLASSIFIED
ow remains simply to set out some of the results achieved by the application of
practicall method just outlined. Before we do so, however, something must be said
it the use of the terms 'language' and 'dialect'.
ere again we are faced with a situation in which no clear decision can be reached
urely scientific grounds. Moreover, there are added difficulties arising both from
ical considerations and from demographic data. Thus from a purely linguistic
t of view there is no real reason for treating ZULU (S.32) as distinct from Xhosa
r), they could easily be regarded as a cluster of dialects. Yet to do so would
n ignoring the fact that the speakers of these two forms of speech have come to
rd themselves as speaking two different languages. Similarly there might be some
fiction for treating SUKUMA (F.zi) and NYAMWESJ (F.2z) as a dialect
;er, but it happens that the speakers of Sukuma are far more numerous than those
yamwesi, and that for political and demographic reasons we have to consider them
parate languages. Thus in deciding what is to be regarded as a distinct language,
what as a mere dialect, not only have we no watertight linguistic test to apply, but
ave to bring in other considerations which are entirely non-linguistic.
this way it is quite likely that the part of the following classification which will
most revision is that relating to the distinction between languages and dialects.
)me cases the test of inter-intelligibility may be applied, but even this cannot be
without arbitrariness since one has first to decide the nature of the topics to be
with in such a test. Thus it may easily happen that a speaker of one language
no difficulty in conversing with the speaker of another when they confine them-
s to simple trading affairs, but yet these same two would be quite unable to under-
1 each other in a discussion of some point of difference in their social customs.
NUMERATION
mentioned in the introductory chapter the method of numbering the languages
been presupposed by the use of the references whenever a language has been
ioned in the preceding chapters. The system worked out enables any language
referred to by a letter and two figures. Each group is indicated by a figure, and
lumber of the language within the group by a second figure. Thus 42 means
second language in the fourth group. When the group as a whole is referred
;ro is used in place of the second figure, so 40 means the fourth group in a
Zone.
here there is a dialect cluster, this also is given a number like the single languages,
:hen the individual dialects are distinguished by additional small letters. For
ple, MYENE (A.71) is a cluster of three dialects, MPUDGWE (A.71a), Ruggu
b), and GALWA (A.71c).
lere are sixteen zones in all, so that it would not have been practicable to use a
y geographical method of referring to them. Instead, each zone has been given
er, and this is put immediately before the figures denoting the language in each
Thus to say that BEMBA is M.42 means that it is the second language in the







30 THE CLASSIFICATION OF
fourth group of Zone M, and consequently is in some measure related to a lan
like FIPA (M.x3).
AUTHORITIES
In a work of this size it is not practicable to cite all the authorities for the data
As has been stated in the first chapter, much of the information has been gathe
first hand, but there are still some very large gaps in our knowledge of the
field. It is clearly desirable to show where such gaps exist, so some of the lang
are given an initial capital only, but others are put entirely in the upper case.
only with respect to the latter that there are sufficient data to make the classifii
reasonably reliable. This is not to say that all the data used are equally trustwort
that the grouping is in any way authoritative, but simply to indicate that in such
it is at least based on something beyond mere word-lists.
Where a language is referred to by a name written with an initial capital only, nc
is known, apart from what may be given in Johnstone's work, beyond the prob,
that there is such a language. When dealing with languages like these, one of twi
unsatisfactory expedients has had to be adopted. Either the languages have
grouped on the basis of the meagre information contained in the word-lists
relationship asserted by some earlier writer has perforce been used. It will the
be very likely that when more data become available a considerable modificati
the classification may be necessary in such cases. It does not seem probable, ho\
that it will be necessary to make any more groups, but rather a reassortment
languages between the groups. The numeration of the well-known language
therefore be able to stand, and in this way we are provided, for those language:
some documentation, with a means of simple reference which can be used in g
Bantu language studies or in the cataloguing of linguistic works.

CLASSIFIED LIST
Since the map only bears the reference numbers of the languages, it was nec
to provide a key to it in the form of a complete classified list. For this reason i
list is given at this point, but instead the groups in each zone are set out as the
are studied in turn.
As was shown in the preceding chapter, the zones are not made on purely linj
grounds. This means that in some cases the groups placed in one zone display a
closer linguistic relationship than those placed in others. Clearly the only satisf
development of the technique adopted would have been a description of the linI
characteristics of each of the groups. That, however, would have been far beyo:
scope of this monograph. Instead, therefore, the zones are described in some
which throws into relief the nature of the relationship between the groups i
given zone.
In some cases the features noted are divided into two sets. There are wh
termed 'common features', which are the ones not common to the whole of the
field but nevertheless to be found to some extent in adjacent zones. Then the
the 'peculiar features', which are not necessarily confined exclusively to the z,
question, but which do not appear to occur in any of the languages immediately a
ing it.






BANTU LANGUAGECS


GROUP IO GROUP 40 GROUP 60
11 Dgolo 41 Bati 61 YATJNDE
12 KUNDU 42 BO 62 BJULIT
13 Mboige 43 Koko 63 Ntum
14 Lue 44 BASA 64 Maka
15 LUNDU 45 Siki 65 Zimu
46 DGUMBA 66 FAD
47 Gbea 67 Make
GROUP 20
21 Mbuku
22 KWIRI GROUP 50 GROUP 70
23 SUBU 51 NOHU 71MYENE
24 DJALA 52 Naka 7Ia MPUDGWE
53 Laigi 7Ib RujIgu
pc GALWA
54 Dgumbi 7C GALWA
GROUP 30 55 BEDGA 72 DIMA
31 BVBI 56 eke 73 KELE
74 KUTA

Characteristics of the Zone
:his zone there is considerable linguistic relationship between the different
i, and it proves possible to list in some detail the features which characterize it.
nmon Features
'he absence of any genders regularly containing words which indicate smallness
less. (Also in Zones B and C.)
'he absence of an extension -u-. There are passive verbals of one kind or
:r in a number of the languages of this zone, but only in BEDGA (55) has any-
)een noted which even resembles the -u- of other zones. For example, uluma
d', ulumakue 'to be sent'. (Also in Zones B and C.)
'he use of single independent nominal prefixes only. (Also in Zones B and C.)
'he use of particles rather than extra independent prefixes. For example, in
(31) u-ite 'at the stone', a-ite 'to the stone', or in KELE (73) pe-djkokj 'on
ne'. An interesting isolated exception occurs in Group io, where LUNDU (15)
extra prefix u-, as in undabu 'in the house', which can govern an agreement,
idabu ubuki 'in the house it is bad'. (Also in Zone C.)
'he use of nominals as sentences without any prefix modification. For example,
TENE (71) aramba 'roots', mano aramba 'those are roots', or in LUNDU
kaka binene 'big mats' or 'the mats are big ones'. (Also in Zones B and C.)
he occurrence of dependent tenses without the suffix *-E. This is by no means
it exception, even within one group, e.g. MPUDGWE (7ia) -gend- 'go',
,ende '(that) they should go', KELE (72) -lum- 'send', balumiki '(that) they
send', KUTA (74) -pik- 'do', bapikakye '(that) they may do'. Apart from
Examples just given, the suffix -e in dependent tenses only appears to occur in
s Io and zo. (Also in Zones B and C.)







32 THE CLASSIFICATION OF
7. The absence of the verbal suffix *-JLE which occurs in many other zones
in Zones B and C.)
8. The rarity of true negative tenses. FAD (66) and MYENE (71) apparent
a difference of tone-pattern as the sole sign of the negative in some tenses. Ii
languages the negative element is simply added to affirmative sentences, either
verbal as affix, or elsewhere in the sentence. For example, in LUNDU (15) t
the infix -sa-, nalaggaka 'I am reading', nasalaggaka 'I am not reading',
DIALA (24) the -sj-, nalumj 'I sent', nasjlumj 'I did not send'. In BAS.
on the other hand, there is a self-standing negative word bi at the end of the ne
clause, e.g. agatimp bi 'he will not return', cf. agatimp 'he will return';
DUMA (72) there is a double sign ka .. ve, e.g. bisu llvovi 'we have spoken
kaljvovy ve 'we have not spoken'. The principal exceptions to this are in BI
(55), where the negative of hukabapandj 'we shall carry' is the shorter word hua
and in KELE (73), where the negative of meelag 'I will count' is the distinct
mecilaij. (Also in Zones B and C.)
9. The substitute object rarely an infix, but usually a self-standing word
example, in LUNDU (15) bauki sj 'they have heard us'. The principal excep
this is in BVBI (31), e.g. tutapj 'we have shown', tubutapj 'we have shown
(Also in Zone B.)
o1. The absence of an alternance k/g in radicals except as the first radica
sonant preceded by a nasal consonant. For example, in FAD (66) ilak 'to sa
large 'say !' (there are no non-fricative velar consonants in this position in this lang
cf. ikum 'bellows', igum 'hedgehog'; or in MPUDGWE (7ia) ikamba 'to E
but gamba 'speak!', cf. ikola 'shell', ggola 'whirlpool'. (Also in Zones B ai
11. A seven-vowel system which appears to be characteristic of the whole
(Also Zone C.)
12. A single quantity only in radical vowels. (Also in Zones B and C.)

II. Peculiar Features
i. The small percentage of words in the standard vocabularies which can be r
to those in languages of other zones. For example, in MYENE (71) and KEL
there are only 8 per cent. and in FAD (66) only 5 per cent. On the other hand, altl
FAD is not very closely related in other ways to the languages of Group 70, it 1
additional 20 per cent. of words which can be related to the standard vocabulai
that group.
2. Unusual types of relationship between extended radicals and simple ra(
For example, in BO (42) there are these typical series of radicals, -kag- 'bind', -
'get bound', -keges- 'cause to bind'; -bom- 'hit', -bumi- 'get hit', -buma-'
(by)'.
3. The existence of two different classes of nominals with distinct indepe
prefixes and governing different dependent prefixes, but both having their plu
the class with the prefix ba-. For example, in BEDGA(55) ukalj myne'this spe
pl. bakall bane; mutu money 'this person', pl. batu bane. Also in KUTA
e.g. nlujgg muayibi 'the builder knows', pl. baluggj bayibj; musjkj ayib
child knows', pl. basiki bayibi.






ANTU LANGUAGES
the independent prefixes of the


a vuvvwl, V.r. vfvlJ/li
bi class appears as g
'my baskets'.
y in the formation ol
lere are four bases ir
-bue- 'break', -lag-
-laI, -lagme, -lai
juent occurrence of
racteristic of Groups
k 'he cut'; in BO (42
roups the occurrence
jnun 'bird', KUTA
,A (55) and MYENI
may be heard in sor
irrence of the combi
mas been given in sec
'river'.

ges of this zone are
)ects they appear sui
Itu', but from the il
iteria laid down in tI
t described enables
rom the classified lisi
on to be more than 1




GROUP IO
ii NZABI
T'7 (;Pbn


uicLripckbJ wiL
before a vowel, e

bal bases by me;
nmon use, and
int', and -lum-
-luma, -lumi,
es and stems w
and 60. For exa
og 'elephant'; ir
final consonants
) guj 'ten'. TI
), but even in tl
positions as bays
on nl in the spe
.3, and in BO (

Summary
r.ont n nI.r r


-oio k44) me aepenaent
A1 bi-nan 'our baskets',

uffixes. For example, in
iow they appear for the
-bu0a, -bu0j, -bume,
me, lumjka.
nal consonant. This is
n BASA (44) abjpot 'he
sDE (61) ateb 'refusal'.
ly limited to nasals, e.g.
ure is much less notice-
er a word like bayamu

minds. An example from
re is nlu 'head', and in



m those of other zones.
vhich have been loosely
Ive been seen that they
tribution of the various
ted into groups, but as
ps in our knowledge for


13 Tsogo 33 Boma 43 TIENE
14 Cira 34 YAKA 44 SAKATA
15 Punu 35 TIO 45 YANZI
16 LUMBU 36 DEE 46 Dgoli
37 WUMU 47 Dirja
GROUP 20 48 MBUNU
21 MBEDE
22 Mbamba
23 Tsaya
Characteristics of the Zone
i the case of Zone A, here too there is a sufficient measure of linguistic relation-
make it useful to describe the features of the zone in detail.







34 THE CLASSIFICATION OF

I. Common Features
i. The absence of any genders regularly containing words which indicate sm
or bigness. (Also in Zones A and C.)
2. The absence of an extension -u-, and, in most of the languages, of ar
passive verbals. SAKATA (44) is an exception to this last statement, having
related radicals of the type -ful- 'open', -mfumful- 'be opened (by)'. (Also in
A and C.)
3. The use of single independent nominal prefixes only. (Also in Zones A a
4. The use of extra independent nominal prefixes. For example in LUMB
gomikaba 'to the villages' can govern the agreement of a word with a stem like
'all', as gootso 'everywhere towards'. In Group 30 there are three prefixed ell
of this kind, e.g. in YAKA (34) kunzo 'to the house', munzo 'in the house',
'at the house', but whereas the first two can govern an agreement the third
(Also in Zone H.)
5. The affixing of extra dependent prefixes immediately to the nominal withe
use of -a- as in some zones. For example, in LUMBU (16) tsinzubu tsiba
'the houses of the fishermen', or in DEE (36) leyimu lemokeo 'the song
woman'. (Also in Zones A and C.)
6. The occurrence of dependent tenses without the suffix *-E. This appeal
without exception in this zone, where dependent tenses use a base similar to
the principal tenses. In most cases there is a distinct tone-pattern for the dep
tense, e.g. in MFINU (41) bamana (\__) 'they will finish', bamana (---)
they should finish'. (Also in Zones A, C, and H.)
7. The fewness of the tense signs, some languages only using one base
number of tenses is sometimes increased by the use of different tone-patter
example, in MFINU (41) there are only four possible shapes for verbal bases, y
these ten distinct tenses are made. (Also in Zone C.)
8. The rarity of true negative tenses, most of the languages using attached
standing particles. For example, in YAKA (34) the particle pe is the sign of th
tive statement, as in ataqgi mukanda pe 'he did not read the book'; and in M
(41) it is we, which does not come at the end of the sentence but immediately
the verbal, as in bakee: we yyiio 'they did not watch the spear'. LUMBU (
the other hand, does appear to have some negative tenses, e.g. atsefwa 'he is
asafwa go 'he is not dead', though even here the self-standing negative w
usually comes at the end of the sentence as well. (Also in Zones A and C.)
9. The use of an infix as a substitute object. Group 30 is peculiar in that 1
no infix for the first person singular, e.g. in TIO (35) bamubere 'they hit hih
babere me 'they hit me'. (Also in Zones B and H.)
o1. The absence of an alternance k/g in radicals except in first position pi
by a nasal consonant. For example, in MFINU (41) makaa 'charcoal' is not
from magaa, but qkana 'craw-craw' is distinct from rgana 'crocodile'. (.
Zones A, C, and H.)
1i. A five-vowel system, throughout the whole zone. (Also in Zone H.)
12. A single quantity only in radical vowels. In some of the languages of Gr
however, there is a peculiar kind of vowel quantity in verbal bases which I






THE BANTU LANGUAGES 35
consonant. For example, in MFINU (41) oka 'to be', and oka: 'to refuse to
this latter has to be distinguished from okaa: 'to fry'); or in SAKATA (44)
wash' and ozo: 'to learn'. (Also in Zones A and C.)
Stress on the radical syllable. (Also in Zones C and H.)
An alternance of tone on the radical. MFINU (41) is exceptional in only having
ssible tone-pattern for each tense, thus whereas in TIO (35) ofura 'to pay' and
to descend' have different tonal behaviour, in MFINU this could not happen.
i Zones C and H.)

Auliar Features
he almost equal proportions of the standard vocabularies related to those from
* the three adjacent zones, A, C, and H.
unusual vowel sequences in extensions and suffixes. For example, in LUMBU
lemisi 'to wound', usugulu 'to wash'; in WUMU (37) obirisi 'to say',
ozo 'to enter'; and in BOMA (42) osikene 'to surpass'.
he prefix to nomino-verbals, which is u- or o-. This is distinct from any
dent nominal prefix, and governs agreements which are not the same as
or the extra independent prefix ku-. For example, in LUMBU (16) uyaba
>w'; in YAKA (34) usala 'to work'; and in MFINU (41) oba onde 'his
ing'.
he occurrence of double dependent prefixes in certain types of nominal, e.g.
LA (34) miti mimibwe 'good trees', manzo mamabwe 'good houses'; or in
U (41) legiko lilinene 'a big banana'.
he anomalous behaviour of the stem for 'two'. For example, in LUMBU (16)
mimioli 'two arms', but malu mamueli 'two legs'; and in WUMU (37)
niele 'two legs', but mako molo 'two arms'. This does not always occur
up 40, e.g. in MFINU (41) mitana mie 'two valleys', manjo mue 'two

he fusion of extensions and suffixes producing abnormal verbal bases, parti-
in the eastern half of the zone. For example, in TIO (35) obie 'to ripen', obio
ct'; in MFINU (41) osibi 'to whet', osiie 'to squeak', oseu 'to sit', osio 'to
"; in SAKATA (44) otui'to lack', ozie 'to spread', otou 'to try', okuo 'to pull';
MBUNU (48) okue 'to go out'.
he existence of some unusual alternances, particularly in alveolar consonants in
n with -i-. For example, in LUMBU (16) there is an alternance 1/r/d, e.g.
'lips', biriri 'grass', badidi 'small people'; and in SAKATA (44) there is an
ace z/3/j, e.g. ozila 'to enter', o3iba 'to know', ojiga 'to bury'.

Summary
tough this zone has some peculiar characteristics which are hardly to be found
-re in Bantu languages, yet on the whole it seems to occupy an intermediate
a between the three neighboring zones, A, C, and H. Nevertheless, not only
:essary to retain it in order to avoid overloading Zones A and C, but there is a
nt linguistic distinction shown by the bunching of the isoglosses along its
ries to make Zone B a very useful set of groups.






36 THE CLASSIFICATION OF
ZONE C
GROUP IO GROUP 30 GROUP 60
11 BUDGILI 31 DGOMBE 61 MODGO-DKIN]
12 Bukongo 32 Buela 6Ia MODGO
13 Kaka 33 Bati 6ib DKINDU
14 Gundi 6ic Pa)ga
15 Pande 6Id Tity
16 Nzeli GROUP 40 6ie Buuli
17 Kota 41 BUA 6If Bukala
42 AIBA 6Ig Yailima
62 Lalia
GROUP 20 63 DGANDU
21 BADGI-LOJ GROUP 50
21a LOI 51 Mbesa
2Ib BUBADGE 52 SO GROUP 70
2Ic NunV 53 PUKI 71 TETELA
22 SEDGELE 54 LUMBU 72 Kysy
23 Tymba 55 KILI 73 DKITI;
24 Bulia 56 Foma 74 Yela
25 NTUMBA, &c. 75 KELA
25a NTUMBA
25b Waggata
25c Mpama GROUP 80
26 LUSEDGO 81 Deigese
26a POTO 82 SoDgomeno
26b MPESA 83 BUBODO
26c MBUDZA 84 Lele
26d MAI)GALA 85 WODGO
26e BULUKI
26f Kaggana
26g LIKU
27 BVTJA
Characteristics of the Zone
Although this zone covers a very large area and is composed of eight group,
contain nearly forty languages, yet these languages display remarkably close r
ships. On the west and south the limits of the zone are well defined, but the
boundary is somewhat arbitrary, although the languages just over this bound
sufficiently different to justify their being placed in another zone.
I. Common Features
I. The existence of genders regularly containing words which indicate sm
These do not occur in Groups o1 and 20, but in most of the others, e.g. in TE
(71) there is the i/tu gender, as amba (for samba) 'small village', pl. tusaml






THE BANTU LANGUAGES 37
the mu/ba gender for this purpose, e.g. mullbuki 'small parcel', pl. bama-
where the 'embedded syllables' -1|- and -ma- have no grammatical function,
long to the stem of the word. (Also in Zones A and D.)
'he absence of an extension -u-, and in many cases of any true passive verbals.
O3 (85) does, however, have such an extension, which requires the suffix -o
other radicals have -a, e.g. ubela 'to cure', ubeluo 'to be cured'. BUBADGJ
Las the extension -jbu-, e.g. atumj 'he has sent', atumjbul 'he has been sent',
BUDGILI (II) has -jb-, e.g. Ikamba 'to bring', jkambjba 'to be brought'.
is, however, an extension -u- in a number of these languages, but this forms
s which express the neuter of those with the extension -ul- (or -un-) and so
- held to correspond to the -uk- of other zones. For example, in NTUMBA
Imun- 'waken (tr.)' -ymu- 'wake up'. (Also in Zone A.)
'he use of single independent nominal prefixes only. There is one exception
that has been noted in MPESA (26b) where nominals with monosyllabic
have a double prefix, e.g. umutu 'person', pl. babatu. (Also in Zones A, D,
)
'he absence of extra independent prefixes. In most of these languages elements
cannot govern an agreement are used where languages in other zones use extra
ndent prefixes. For example, in BUBADGI (zib) the u- in u-ljkulu ljndaku
house' cannot control any agreements. Similarly in TETELA (71) there is the
It la- as in la-igelp 'to the village', and lu- as in lu-luydy 'in the house',
. of which can govern any agreement. (Also in Zone A.)
'he affixing of extra dependent prefixes immediately to the nominal in Groups
S30. For example, in BUBADGI (zib) bjlamba bjmukunzi 'the chief's
'; in DGOMBE (31) mijo mjkymi 'the chief's affairs'. In most of the other
this does not happen, e.g. in BUA (41) ibalj iakpmy 'the chief's house', and
[ELA (71) lukuki lalupdy 'the door of the house', kuki jamvudy 'the doors
houses'. KELA (75) is exceptional in using -nda- to link the extra dependent
:o the nominal, e.g. isala indaasaggu 'a garden of maize', buca bundakpmy
ief's head'. (Cf. the adjacent zones for both types of behaviour.)
'he occurrence of true negative tenses in most of the groups. On the whole
s little regularity in the formation of negative statements in this zone, but here
- or two examples. BUDGILI (iI) uses the self-standing particle ka at the
negative sentences in some cases but not in others, e.g. babuyiba ka 'they did
3w', but even here there is distinction in form in the tense, cf. bayibakj 'they
DGOMBE (31) affixes a negative element to the tense, but this varies from
o tense, e.g. bupalaka 'we liked', buipalaka 'we did not like', bupall 'we
upaljt 'we do not like'. DKJTI3 (73) and WODGO (85) both have special
'e tenses together with a negative particle at the end of the sentence, e.g. in
7rl tumpeya 'we know', tupeyl ve 'we don't know', and in WODGO the same
ntences are bjtu cpmayiba, bjtu cyayiba bo. (Cf. the adjacent zones for
types of negative construction.)
'he absence of an alternance k/g in radicals except in first position preceded by
consonant. In all the groups of this zone there are cases similar to this example
lUBADGI (2Ib) where ljkambu 'affair' is not distinguished from ljgambu,
*r l 1 -----(?rt-' I ; A-+ n-+ f___r -A-1f-1{ (1_ .-:__ 1A1__ :_ 17-.---. A J- T) N






38 THE CLASSIFICATION OF
8. A seven-vowel system throughout the whole zone. (Also in Zones A and
9. A single quantity only in radical vowels. (Also in Zones A, B, and D.)
io. Stress on the radical syllable. (Also in Zone B.)
i1. Lexical tone on both radical and suffixes. (Also in Zones B, D, and L.)
12. Absence of any tonal distinction in dependent prefixes. There are one <
exceptions to this similar to that found in DGOMBE (31), where the depe
prefix i- which is the singular corresponding to bj- often has a tone which is d:
from that of the prefix i- the singular of jj-, e.g. bjpundu bjkymy (_-_ --
chief's axes' sing: ipundu ikymy (_-_ ---); ndaku jjkpmy (_-_ ---) 'the (
houses', sing: ndaku ikmp ( -_ _--). (Also in Zones B and D.)
13. Regularity of tone-pattern in all syntactical relationships. In general there
tonal modification either to characterize or to indicate syntactical relationship
the tone-pattern of a word is established in any context it is found to be the same
other contexts. (Also in Zones D and L.)

II. Peculiar Features
i. An abnormally high proportion of the standard vocabularies related to th
other languages within the zone. In some cases two languages, such as BUD
(II) and BUBADGI (2zb), have as much as 60 per cent. of the standard vocal
related, but a more average example may be taken from SO (52) which has ab<
per cent. of its vocabulary relatable to that of other languages within the zon
only 15 per cent. to languages in other zones; in addition, of this 15 per cent.
mere 3 per cent. is peculiar to SO within its group.
2. A regular system of extensions in which the vowel of the extension -js- :
ferent from that of -il-, e.g. in BUBADGI (2Ib) -tum- 'send', -tumjs- 'cai
send', -tumil- 'send to', -kom- 'be adequate', -komjs- 'make adequate', -kc
'be adequate for'.
3. The use of a prefix, both dependent and independent, as the singular corres]
ing to the prefix bj-, which consists of a vowel only. This appears to have onl
exceptions in the whole zone, in SEDGELE (22) where the prefix is ki- (ke-:
in SO (52) where the prefix is ki-, or in some forms of the language hi-, e.g. ki
bjtPty 'wall(s)'. In every other language in the zone the corresponding prefix al
to be i-, e.g. in BUBADGI (21b) ibuka ine 'a large pounding mortar', pl. bj
bjne. (This is in direct contrast to what happens in some of the languages of Zc
where the independent prefix is i-, but the dependent prefix is ki-, e.g. in
(B.35) iju kinene 'a large pounding mortar', pl. biju binene.)
4. The occurrence of uncommon prefixes in nomino-verbals. Unlike the lang
of Zone B, these have a variety of prefixes, but none appears to have the common
For example, in BUDGILI (11) jbumba 'to hide', in DGANDU (63) ljtun
send', and in KELA (75) jkenda 'to go', all of which behave like words in the sil
of the j/ma or lj/ma gender. In DGOMBE (31) there is bubala 'to speak
in DK3TIJ (73) ntuka 'to draw water', while WODGO (85) and BUA (41) ha'
special prefix u- like the languages of Zone B, e.g. WODGO, ulika 'to pass';
upaga 'to say'.
5. A similarity in the shape of the dependent and the independent prefixes.
is one of the most striking features of this zone, where, for example, the depe






THE BANTU LANGUAGES 39
:s of the mu/mi gender are usually mu/mi, as in DGOMBE (31) mukanda
i mubp~ gj, 'his book is lost', pl. mjkanda mjndi mjbyrgj.
'he impossibility of using a nominal as a sentence. Unlike the languages of
either zones, these almost always use some kind of copula in similar cases. Here
'ew examples, in BUBADGI (2zb) mubimbf aJga ntaggl 'the traveller is a
nan', pl. babimbi baiga bantaggj; in DGOMBE (31) jmu mpdi mukanda
a book', pl. jmu mjdj mjkanda; in DGANDU (63) line iku ljsala ljnami
Smy garden', pl. bane gku basala anami; in KELA (75) jie ayadj kpmp
their is chief'. In most cases the copula has a dependent prefix, but iku in
NDU is an exception.
The use of a suffix -i (distinct from -J) in dependent tenses. Here are some
les of it, from BUDGILI (ii) bataigi '(that) they should count'; from
MBE (31) tusoni '(that) we should write' (cf. tusonl 'we have written'); from
TDU (63) bukambi '(that) you should work' (cf. bukambi 'you have worked');
FETELA (71) katuuki '(that) we should hear'. There are some exceptions to
Groups 20 and 80, e.g. in BUBADGI (2zb) nakita '(that) I may fall'; and in
GO (85) buyiba '(that) they may know'.
Che use of the two verbal suffixes -I and -akj. (The common suffix of other zones,
loes not seem to occur anywhere in this zone.) There are a few exceptions to
)ut the following examples will show approximately the distribution of the
:s: BUDGILI (11) -lub- 'say', alubakj 'he said'; NTUMBA (25) -yjn- 'hate',
akj 'they hated'; DGOMBE (31) -bal- 'say', bubal| 'we have said', bubalakj
-re saying'; By(A (41) -men- 'see', bamenj 'they saw'; SO (52) -lyk- 'paddle',
'I have paddled', lilpkaki 'I paddled'; DGANDU (63) -uk- 'hear', aukakj
!ard'; DKITI( (73) -kits- 'descend', tukjtsakj 'we descended'; WODGO
)ul- 'strike', abull 'he struck'. KILI (55) on the other hand uses -Ijk and -aka,
at -aki, e.g. -kil- 'do', tukiljki and tuakilaka 'we did', tukill 'we have done'.
rhe regular occurrence of the 'inverted' relative construction. Since this is
r in most languages, one example will suffice. In WODGO (85) the following
pical relative clauses, mukanda mumalomba bjnu 'the book you asked for',
nda mjmalomba bjnu 'the books you asked for'. In these and all similar
the verbal agrees with the antecedent only, and the subject immediately follows
rbal.
A simple consonant system with an almost complete syllabary. In a number of
aguages of this zone many of the words which can be related to those in other
iges occur with sounds which are almost identical with those used in the starred
of common Bantu.

Summary
e principal features of the languages of this zone are a simpler grammatical
are than is found in many others, coupled with a simple phonological and tonal
i. This may, in fact, be taken to be one of the important areas of Bantu, dis-
g as it does fairly homogeneous linguistic characteristics which are different in
ways from those of other zones.






1THE ULAbb1k1iUAT1iU


GROUP 10
S1 Mbole
z1 Leqgola
13 Mityky
14 Genya


GROUP 20
21 Bali
22 Amba
23 Kumu
24 Sorgola
25 LEGA
26 Zimba
27 Bangubai


/I L
GROU]
31 P
32 B
33 H


GROUI
41 K
42 N
43 N


GROU:
51 H
52 H
0 53 N


54 tBIVIMLBI
55 Buyi
56 Kabwari

Characteristics of the Zone
Unlike the three zones already described, this one is of little linguistic signific
There are reasons for not placing any of these groups in the neighboring zone
few, apart from geographical contiguity, for making a zone out of them. Mor
apart from Group 60, our knowledge of the languages of this zone is so fragmentar
even the grouping is in most cases very tentative. For these reasons a mere outl
the distribution of some of the characteristics is all that can be attempted at pre,
I. There is insufficient data for the establishing of standard vocabularies for
languages. From the scanty word-lists which are available some of the words 1
occur here appear to be related to those in languages to the west rather than 1
east, e.g. in BEMBE (54) mbuka 'village', mtuba 'six', and -bpnd- 'figl
correspond to similar words found in Group C.2o. In the Sub-Bantu language.
(z2) the vocabulary is almost the only thing which entitles it to be put in th
but this agrees as to both sounds and tones with many words found in the lang
of Zone C.
2. Most of the languages appear to have a gender which regularly contains
indicating smallness, usually ka/tu. BEMBE (54), however, seems to have
usual one hi/bu, e.g. hibuka/bubuka 'small villagess' (cf. mbuka/ma
villages(s)'.
3. In some of the more northerly languages of this zone there are some ur
genders. For example, Bali (21) in addition to the commoner -/ba, li/ma,
ku/ma has others containing words like these, mnzj/kpzj villages(s), ljbu/r
rivers(s), kplsly/mulply 'knee(s)', bilj/bebj 'rat(s)', Jikul/burjkul 'wife(s]
4. In Groups o10, 20, 30, and 50 independent nominals have single prefixes


GROUP (
61 NYI
62 RU]
63 FUI
64 JUI
65 HAl
66 HA
67 Vim






THE BANTU LANGUAGES 41
rroup 40 there are double prefixes in which the first part is o-, e-, or a-, e.g. in
ANDI (42) omubiri/emibiri 'body(s)'. In Group 6o there are double prefixes
u-, i-, or a- as the first part, e.g. in HA (66) umugezi/imigezi rivers(s).
In most cases the prefix of nomino-verbals is ku- (or uku-), but in Group 40 it
*i-, e.g. in KONZO (41) erihjka 'to arrive'. In BEMBE (54) there is prefix u-
:h governs its own special dependent prefix u-, e.g. utenda ube 'your speaking'.
Most of the languages in this zone have three extra independent prefixes, as
in HOROHORO (28) haqkoje 'at the river', kuqkoge 'to the river', muirkoige
ae river'. In LEGA (25) there is the less usual prefix ga- where others have ha-,
galupzj 'at the river'. The languages of Group 60 also have a fourth prefix of this
, i-, which is of limited usage, e.g. in HA (66) ibuami 'in the realm'.
The extra dependent prefix is added to the nominal in Groups 40 and 60 with-
my modification of the double prefix, but there is the element -a- in Group 60,
areas in Group 40 this is not always used. For example, in NDANDI (42) ebisandu
nuntu 'the man's feet' (cf. omuntu 'man'), ebjsandu bjwe 'his feet', but in
(66) ibitebe biaumuntu (pr. byoomuntu) 'the man's stools', ibitebe biage 'his
Is'. In the Sub-Bantu languages of Group 30, since there is no clear system of
ies of nominals, there are naturally no dependent prefixes, e.g. in PERI (31) Jsu
e or ae 'his eye(s)', miima ndae or ae 'his heart'.
Nominals are used as sentences in some of these languages. For example, in
O1HORO (28) rjgoji gonsoga 'the rope is a good one', pl. migojj yemisoga,
Igoji nsoga 'a good rope', pl. migoji misoga. The languages of Groups 40
60, however, prefix ni- to the nominal stripped of the first part of its double
ix, e.g. in KONZO (41) nilunyoggo 'it is a big pot'.
The dependent tenses mostly have a suffix -e, e.g. in LEGA (25) gulole '(that)
may look'. The Sub-Bantu languages of Group 30 have -i, which is distinct from
:.g. in PERJ (31) aupi '(that) he may know', aupj 'he knows'.
). The suffix *-JLE appears to occur in most of the groups, apart from the Sub-
tu languages. For example, in LEGA (25) aatendjle 'he said'; and in SUBI (64)
zimie 'we got lost'.
:. Apart from Group 40, most of these languages have a fairly simple tense
am, and this is one of the most important features which distinguishes those in
up 60 from Zone E. For example, HA (66) only has three main tenses, by means
rhich it is possible simply to refer, without time words, to two different times in
past and one in the future, as tulaabonie 'we saw', tulaabona 'we have just seen',
Lbona 'we shall see'.
2. There are true negative tenses in most languages in this zone outside Group 60,
in KONZO (41) mutuagonyjre 'we slept (yesterday)', neg. muatutegonya.
rroup 6o, however, a negative element may be prefixed, or infixed, to every affirma-
tense, e.g. in .UBI tuatemie 'we cut', nhituatemie 'we did not cut'.
3. The relative clause is usually constructed without any linking word. A typical
nple may be seen in the following sentence from LEGA (25), bjbjla bjgua-
nine bjakolokjle 'the palm-trees you saw fell down'.
.. In contrast to the languages of Zone C, there is usually an alternance g/k in
cal consonants in these. For example, in many of the languages -gul- 'buy' is
.nct from -kul- 'grow', in BEMBE (k4), however, this alternance is missing.






42 THE CLASSIFICATION OF
probably owing to the fact that whereas the -g- of other languages is represent
-k-, -k- is represented by zero, and these two radicals appear as -kul- 'buy
-ul- 'grow'.
15. There is a seven-vowel system in Groups 10-50 and a five-vowel system
Group 60.
16. There are two quantities of vowel in the radical in Groups 20, 40, and 6(
in HOROHORO (28) igoko 'chicken' has a quantity distinct from that of tjg
'shore', while in HA (66) the quantity of -kul- 'grow' is different from that of -1
'take out'.
17. There are two quantities of vowel in extensions in some languages, e
HOROHORO (28) the quantity of the vowel in the extension of -kusuuk- 'le
distinct from that in -guruk- 'jump'.
18. There appears to be an alternance u/o in nominal suffixes throughout
whole zone, which is in contrast to the whole of Zone C, except Group C.50.
example, in Kumu (22) the suffix of ndabo 'house' is distinct from that of ml
'knife'.
19. In Group 60 there are some peculiar combinations of consonants in the
nunciation of syllables which contain a consonant and two vowels. For examp
NYARUANDA (6i) diumuana '(isuka "hoe") of the child' is heard as dgumr
while ibiatsi 'grass' is pronounced ibjatsi.
20. In Groups 40 and 60 the alternances between voiceless and voiced plosive
masked in junction with a nasal consonant. For example, in NDANDJ (42) oluki
'cloth', pl. esjoigimba; or in FULIRO (63) tukagira 'we did', rjgagira 'I
tutakagira 'we did not', ndakagira 'I did not'.
21. In Group 60 -k- immediately preceding a radical commencing with a voii
plosive is pronounced -g-, e.g. in HA (66) ikintu ikito 'small thing', pr. it
igito.



ZONE E
GROUP IO GROUP 20 GROUP 30
11 NYORO 21 NYAMBO 31 MASABA
12 TORO 22 HAYA 3ia GISU
13 NYADKOLE 22a ZIBA 3ib KISU
14 CIGA 22b Hamba 31c BUKUSU
15 GANDA, &c. 22c Haggiro 32 HADGA
i5a GANDA 22d Nyakisaka 32a WADGA
I5b Sese 22e Yoza 32b Tsotso
16 SOGA 22f Endargabo 33 NYORE
17 GWERE 22g Bumbira 34 SAAMIA
i8 NYALA 22h Mwani 35 NYULI
23 DZINDZA
24 KEREBE
25 JITA






I RANTIT T.AN0TTAC'-T7Q


GROUP 40 GROUP 50 GROUP 70
41 RAGOLI 51 KIKUYU 71 POKOMO
42 G3VSJ 52 EMBU 72 NIKA
43 KURIA 53 MERU 72a GIRYAMA
44 ZANAKI, &c. 54 SARAKA 72b KAUMA
44a ZANAKI 55 KAMBA 72c CONYI
44b ISENYI 56 DAISO 72d DURUMA
44c Ndali 72e RABAI
44d Siora GROUP 60 73 DIGO
44e Sweta 61 RWO 74 TAITA
44f Kiroba 62 CAGA 74a DABIDA
44g Ikjzi 62a HAI 74b SAGALA
44h Giraigo 62b WUNJO
44k Simbiti 62c ROMBO
45 NATA 63 Rupa
46 Sonjo 64 KAHE
65 GWENO
Characteristics of the Zone
ie placing of the limits of this zone has been done on a linguistic basis, but it is
:ult to describe exactly the features which are peculiar to the zone, since there are
options to almost every one. There are many important languages in these groups,
there is a considerable amount of reliable data available, so the grouping is much
;entative than in some other zones. It is most convenient then to take in turn the
rentia which determined the grouping and describe their occurrence.
The standard vocabularies contain a large proportion of words which can be
ed to those found in languages of other zones. In the case of KIKUYU (51), for
iple, it is about 20 per cent.
Apart from Groups 60 and 70, each language makes use of genders which
larly contain words indicating smallness or bigness. For example, DZINDZA
has akahuli/utuhuli 'small egg(s)', (cf. ihuli/amahuli 'egg(s)'), and idzoka/
.dzoka 'big snake(s)' (cf. indzoka/indzoka 'snake(s)'); KISU (31b) has kabono/
)no 'small knife(s)' (cf. kumubono/kimibono 'knife(s)'), and kuusaala/kimi-
a 'big tree(s)' (cf. siisaala/biisaala 'tree(s)'); NATA (45) has akabuhi/ibjbuhi
11 stoness' (cf. rjbuhi/amabuhi stones(s)', and ugusjr}/amasjrl 'big rope(s)'
trusjrj/cas}jr 'rope(s)').
In Groups 10-40 independent nominals regularly have double prefixes, mostly
e type with o-, e-, and a- as the first part. For example, in KEREBE (24) ekintu
g', pl. ebintu. In Group 40 the vowels in certain prefixes are indeterminate,
g heard as i or e (and u or o) according to the vowel of the radical, e.g. in GUSII
ikirugu/ibjrugu chairs(s), ekenene/ibjnene'big one(s)', umurimu/imirimu,
i(s)', omogeka/emegeka 'mat(s)'. In the other groups the independent pre-
are always single.
The prefix of nomino-verbals is ku- (or uku-, &c.) in all the groups of this zone
,t 60, where it is i-, which behaves like the singular prefix of the i/ma gender,
RWO (61) illa 'to look at', or GWENO (60 iruma 'to send'.






44 THE CLASSIFICATION OPY
5. Extra independent prefixes occur in Groups o10-40, but not in Groups 5(
which have an extra nominal suffix. Here are examples of the suffix, in KIKUYU
mbembe-ini'among the maize' (cf. mbembe 'maize'), in KAHE (64) numbe-n
the house' (cf. numba 'house'), CONYI (72c) cisima-ni 'at the well' (cf. cis&
'well').
6. Nominals are used as sentences throughout the zone. In GANDA (15) the
a heavier prefix, e.g. gyemikeka 'they are mats' (cf. emikeka 'mats'); in ZIBA (
a shorter prefix, e.g. mahuli 'they are eggs' (cf. amahuli 'eggs'); in NATA (45]
first part of the prefix is replaced by a nasal consonant, e.g. mbjtymbi 'they
stools' (cf. ibitymbi 'stools'). In many cases ni- is prefixed to the nominal, e.1
WADGA (32a) as in niemisaala (pr. neemisaala) 'they are trees' (cf. emis&
'trees'); and in GIRYAMA (72a) nimacuirba 'they are oranges' (cf. macu
'oranges'). In others the nominal has the same form as in other sentences, e.1
BUKUSU (31c) kuno kumukunda kuaqge 'this is my garden' (cf. kumuku
'garden'); or in DIGO (73) higa majembe 'these are hoes' (cf. majembe 'hoe
7. A suffix -e is characteristic of dependent tenses but by no means confine
them. For example, in KURIA (43) turente '(that) we should bring', turaare
'we are going to bring' (cf. turaarenta 'we are bringing'); or in SAAMIA
kutandule'(that) we should tear', kunatandule'we are going to tear', kuakatan,
'we shall tear (tomorrow)'. GUSJI (42) is exceptional in having an indetermi
vowel as a suffix in its dependent tenses; this is heard as -e if the radical has -(
-o- but otherwise as -i (distinct from -1), e.g. tuguli (-_-) '(that) we should buy
ntuguli (--_ -) 'we shall buy'), and tutebe (---) '(that) we should say' (cf. ntul
(___-) 'we shall say'). The principal exception to this is in Group 60o, where the
of the dependent tenses is sometimes an infix with the suffix -a.
8. The suffix *-JLE occurs in most of the zone, but is absent in Group 60 and n
of 70. RAGOLI (41) is unusual in this zone in having -j but not -ile, e.g. -
'forge', kutyl} 'we have forged'. In Group 50 there are the two suffixes -jle
-iite, e.g. in EMBU (52) -bul- 'hit', nitubuliite 'we have hit', nitubuljle 'wl
(earlier)'.
9. In much of this zone there is an almost unparalleled wealth of tenses. In Gr
10-50 it is frequently possible to refer, without the use of time words, to four diffi
periods of past time and an equal number of future time, e.g. in NYORE (33) t
are the following eight tenses of -sab- 'ask': kuasaba (---) 'we asked (long a
kuasabire (----) 'we asked (yesterday)'; kusabire (___-) 'we asked (this morni
(cf. kusabire (---) 'we have asked'); kuakasaba (---_) 'we have just asl
kulaasaba ( ----) 'we are just going to ask'; nakusabe (__--) 'we will ask (
to-day); kuakasabe ( -_---) 'we will ask (to-morrow)'; kulisaba (----) 'we wil
(after to-morrow)'.
Io. Apart from Groups 30, 40, and 60 there are true negative tenses in :
languages. For example, in JITA (25) -ta- is the negative sign, but the negati.
ecikora 'we are working' is citakukora, whereas there is no form like cikukol
use. Here are two examples from languages which have no negative tenses, in K
(3ib) the negative sign is si ... ta, e.g. kulikula kamaki 'we shall buy eggs',
sikulikula kamakita; andinWUNJO (62b)lulewona q guku'we sawachicken',
lulewona qguku pfo, shows the sign of the negative to be a self-standing word






THE BANTU LANGUAGES 45
Few of the languages in this zone make use of a copula -11 in the formation of
pal tenses. The chief exception to this is in Group 30, where forms like this
GISU (3 Ia) kuli kutema 'we are cutting', but even here SAAMIA (34) agrees
[ADGA (32) and NYULI (35) in not using -li in the corresponding tense, which


Ilmtfi 'IxrT


re one or two Unl
in all positions,
ebts'), katiti 'sn
I 40 there is ij
old in detail', -de
positionn in junctic
', ejiono 'mark
alternances of fl
62b) irika 'to cl
:onal behaviour.
a seven-vowel sy
except in Group I
'e-vowel system.
e two quantities
oups 60 and 70.
rnance between
sonants in Grou]
:s)', olukoba/tsil
agreeing with ul
reeing with the s
.e second radical
:d plosive and its


at consonant alternances IT
g. KISU (3ib) katiti 'sma
one' (agreeing with kabe
in second radical position:
k- 'stroll'; and also the unc
vith a nasal consonant, e.g..
animal's sleeping-place'. ]
'ed and lateral consonants
I', ilka 'to clothe', ijika

n in Groups 40 and 50. ElI
here languages 11-14 hav<

idical vowel in the language

ced and voiceless plosives
;0-50. For example, in W
ba 'belt(s)'; or in RAGOLI
aho/tsimbaho boards(s)',
words).
sonant is a nasal compound
responding nasal, injunction
:cept 60 and 70. For exan
rom indeligo.


s zone. In Grc
ies' (agreeing -
'small knife').
.g. RAGOLI(
non alternance !
}OLI(4I) eggc
rroup 60 there
junction with -
1. 11


In Groups 40 and 50 -k- in a preradical syllable is heard as -g- if the following
nant is a voiceless one, e.g. in NATA (45) ikikulu 'big one' (agreeing with
so 'knife') is heard as igikulu. Similarly in KIKUYU (5 ) the very name of the
age is pronounced gikuyu.
Apart from the two languages JITA (25) and NATA (45), all the languages of
mne make use of lexical tone, and many of them, including even JITA, use a dis-
on of tone-pattern to differentiate tenses which are otherwise similar, e.g. in JITA
:ialiga (_ -) 'we were (yesterday)', cialiga (_--) 'we were (before yesterday)'.
In Groups 50-70 the tonal system is often very complicated, and it is frequently
ult to relate the speech-tones to the essential tones of the language.
In Group 50 the extra dependent prefixes agreeing with nominals in the
lar of the mu/a gender, the plural of the mu/mi gender, and the singular of the
render have a different tonal behaviour from those agreeing with nominals in
classes. For example, in KIKUYU (51) mur}go uamugenj ( - -) 'the
ger's load', mirigo iamugenj (__- ----) 'the stranger's loads'.






46 THE CLASSIFICATION OF
Summary
Although these seven groups have very little in common, yet they do form a
venient zone for reference. The one which has least in common with the oth(
Group 60, but it would be still less suitable to put it into either of the adjacent z


ZONE F
GROUP IO GROUP 20 GROUP 30
11 TODGWE 21 SUKUMA 31 NILAMBA
12 Bende 22 NWAMWESJ 32 RIMI
2za NYANYEMBE 33 LADGI
22b Takama 34 Mbugwe
22c Kiya
22d Mweri
23 SUMBWA
24 KIMBU
25 BUDGU

Characteristics of the Zone
This zone is made up of three fairly closely related groups, and most of its I
daries are sharply defined by the coincidence of several isoglosses. Since, hom
there are few features which are really peculiar to the zone it is simplest to descril
distribution of the most important characteristics in turn.
I. There are genders in each language which regularly contain words indict
bigness or smallness. For example, in TODGWE (I1) kanyonyl/tunyonyl '
bird(s)'; in LADGI (33) kalufjo/tulufjo 'small knife(s)'; RIMI (32) ijoka/mE
'big snake(s)' (cf. njoka/njoka 'snake(s)'.
2. Independent nominals have double prefixes in most of these languages
when determined. For example, in SUMBWA (23) amagyta matimbu 'the
good', but tuagula magyta 'we bought some oil'. BUDGU (25) is an excel
e.g. unti/imiti 'tree(s)'.
3. Each language uses three extra independent prefixes, except RIMI (32)
appears to have only one u-, e.g. umoggo 'in the river', unjia'on the path', unyu
'to the house'.
4. Usually ni- is prefixed to nominals used as sentences, but this is by no r
unexceptional. For example, in TODGWE (ii) nimakala or makala 'it is char
SUKUMA (21) ulu lugoye 'this is a rope'; RIMI (32) Iji mburj 'these are g
LADGI (33) ulu niludjhi 'this is a rope'.
5. The singular class which has the dependent prefix li- has j- as its indepe
prefix. For example, in TODGWE (11) jbala 'garden'; KIMBU (24) 1gi
LADGI (33) Ikyfa 'bone'.
6. A suffix -Ire or -Ile occurs in each language of the zone.
7. The element -ag- or -gga which is used in tenses referring to actions in pr,
also occurs in tenses which do not have this kind of reference. For examl






THE BANTU LANGUAGES 47
)GWE (Ii) -bumb- 'fill', tuakabumbagga 'we filled'; tuabumbagga 'we
just filled'; or in SUKUMA (21) tuahambaga 'we have just planted'.
In most of these languages there is a complicated but unbalanced tense system,
sample, LADGI (33) has four distinctions of past time expressed by its tense
m, but only one future.
The copula is used as a tense formative in a way different from that noted in
E. For example, in KIMBU (24) kuali kuabonjle 'we saw (long ago)'; in
GI (33) kutaha turi majj 'we will draw water'; in SUKUMA (21) tutaali
rgga 'we are still building' (where -ta- is a 'negative' element).
There are true negative tenses in most of the languages; SUKUMA (21)
cteristically has seven simple affirmative tenses but only two simple negative
s. Here are two other examples, in LADGJ (33) tuatpuJgjre 'we have just built',
sjtukutp1ga (note the form tukutygga does not appear to be used in the
native); in RIMI (32) nakurema 'we shall cut', kurfurema 'we shall not cut'
there is no affirmative tense with the sign -u- -a).
The consonant alternances of these languages do not present many peculiarities,
It in the case of RIMJ (32), which has the unusual alternance kl/r/u in which the
iound is flapped, the second a voiceless one-rap 'r', and the third a voiced uvular
ive. This occurs in all positions.
There is a seven-vowel system in every language of this zone.
There are two quantities of vowel in the radicals throughout the zone, e.g. in
'GWE (i) -teel- 'throw', -tek- 'cook', LADGJ (33) -loot- 'dream', -lok- 'pass'.
There are unusual types of vowel coalescence in these languages, where neither
ior -i- in junction with another vowel is heard in speech. For example, in
UMA (21) naaiba 'I forgot' is heard as niiba.
Apart from BUDGU (25) the junction of a nasal consonant and a voiceless
re does not correlate with any masking of the alternances. For example, in
GWE (I ) -kulu 'big', mbusj gkulu 'big goat'; in RIMI (32) -kppi 'short',
e kppi 'short ropes'; but in BUNGU (25) qJguku 'chicken', cf. akakuku 'small
en'.
Tone is used lexically in all the languages except LADGI (33), but even here
ised grammatically. There are frequent examples of tenses which are only dis-
ished by a difference of tone-pattern, e.g. in SUKUMA (21) tuabalaga (_-_.)
mounted (earlier to-day)', tuabalaga (---) 'we used to count'; or in LADGI
uasakjre (---) 'we sought(to-day)', tuasakjre (_-_) 'we sought (yesterday)'.
Verbal prefixes do not all have the same tonal behaviour. Those agreeing with
(sing. of ba-), mi-, and n- (sing.) have a behaviour different from all the others.
example, in KIMBU (24) nyuggu ialjmilaga (-- \) 'the pot is lost',
gu jjalmilaga (- _----\) or in NILAMBA (31) mugunda ualimjlue
- ---) 'the garden is cultivated', migunda ialimjlue ( --- ---) 'the gardens
iltivated'.

Summary
already noted, the boundaries of this zone are well defined, but it will be seen
mparing its characteristics with those of the adjoining ones that no one set of
entia can operate on all sides. Thus it is sharply distinguished from Zone F by






4" Inrr Uj.Ao5 LLA I11 AIUiN Ut
its seven-vowel system, its use of two-vowel quantities, and its use of lexic;
in radicals, as well as by certain grammatical features. From Zone E, however,
to be distinguished by such features as the use of single nominal prefixes, and
characteristics of the tense system.


ZONE G
GROUP 10 GROUP 40 GROUP 60
11 GOGO 41 Tikulu, &c. 61 SADGO
12 KAGULU 41a Tikulu 62 HEHE
4ib Mbalazi 63 BENA
42 SWAHILI 64 PADGWA
GROUP 20 42a AMU 65 KEDGA
21 TUBETA 42b MVITA 66 Wanji
22 ASU 42c MRIMA 67 Kisi
23 JSAMBAA 4zd UNGUJA
24 BONDEI 43 PEMBA, &c.
43a PHEMBA
43b TUMBATU
GROUP 30 43c HADIMU
31 ZIGULA 44 KOMORO
32 DHWELE 44a DGAZIJA
33 ZARAMO 44b Njuani
34 DGULU
35 RUGURU
36 Kami GROUP 50
37 KUTU 51 POGOLO
38 VIDUNDA 52 Ndamba
39 SAGALA
Characteristics of the Zone
Some of the groups in this zone are more closely related to each other th,
others; in particular Group 60 is one on its own, but this is the most convenient
into which to put it. Since there are few if any features peculiar to this zone
simplest to describe the differentia one by one.
I. Although there is insufficient data for the compiling of many standard vo
laries, it seems very likely that there is a rather high proportion of vocabulary coi
to some of the groups.
2. Genders regularly containing words which express bigness or smallness
in almost every language, but the form of the prefixes varies greatly. For exam]
TUBETA (21) kasuke/tusuke 'small cloth(s)' (cf. suke/suke 'cloth(s)'); in
(22) kabuji/bubuji 'small goats' (cf. mbuji/mbuji 'goat(s)'; in ZIGULA
kagola/wagola 'small knife(s)' (cf. ggola/ggola 'knife(s)'; in RUGURU
ilatsoka/ipfitsoka 'small snake(s)' (cf. intsoka/intsoka 'snake(s)'); and in F
(62) akafugu/utufugu 'small pot(s)' (cf. ikifugu/ififugu 'pot(s)').
3. In Groups io, 30, and 60 independent nominals have double prefixes, altl






THE BANTU LANGUAGES 49
oup 30 this is not the invariable rule. For example, in KAGULU (12) imusehe/
ehe 'old personss); in KUTU (37) uluzabi/zinzabi (or luzabi/inzabi)
s)'; in SADGO (61) iljpjsj/amapjsj 'egg(s)'.
There are extra independent prefixes, except in parts of Groups 20 and 30. For
ple, in TUBETA (21) nyumbeni 'at the house' (cf. nyumba 'house'); in
BAA (23) nyumbai 'at the house' (cf. nyumba 'house'); and in DGAZIJA(44a)
- there is a prefixed element o- as well as a suffix, as in osindoni 'at the market'
indo/zindo markets(s)) and in onyuiguni 'in the pot', (cf. nyujgu 'pot').
The nomino-verbal prefix is ordinarily ku- (or uku-), but in DGAZIJA (44a)
i-, e.g. uhula 'to buy'; and in SADGO (61) it is ki- (distinct from the singular
:of the ijj/jfj gender), e.g. kiseiga 'to build'.
The first person plural prefix which is very useful in some groups as a dis-
ishing feature is useless here since it varies so greatly. For example, in SAMBAA
BONDEI (24), and SADGO (61) it is ti-, in DGAZIJA (44a) it is ri-, but in
p Io and in DHWELE (32) and DGULU (34) it is ki- or ci-, whereas in most
cases it is tu-.
Nominals are used as sentences in most languages of this zone, usually with some
ication of the prefix. For example, in KAGULU (12) a single prefix is used, as
Ssuke inoga 'this is a good cloth' (cf. isuke 'cloth'), as also in KIDGA (65), e.g.
nasyta 'this is oil' (cf. amasyta 'oil'). In DGAZIJA (44a) there is a prefixed
:nt, e.g. ggomro mhu 'it is a big river' (cf. mro 'river'); while sometimes the
Sis unchanged, e.g. in KUTU (37) gano gamafigga gaigu 'these are my
(cf. gamafigga 'eggs'), or in PHEMBA (43a) nti ule lie 'this tree is a long one',
ti lie 'a long tree'). The regular use of the prefixed element ni- is apparently
led to Group 20 and SWAHILI (42).
The suffix *-ILE occurs in most languages except those in Groups 20-40.
uliar feature of some of these, however, is that although this suffix does not occur
irmative tenses, it does in negative tenses, e.g. in VIDUNDA (38) hatukolile
id not work', where the base -kolile is not used in any affirmative tense. HEHE
s unique in this zone in using -ile and -ite almost interchangeably, e.g. -lorJg-
i', tualorgite or tualonzile 'we spoke'.
A suffix -e is used in dependent tenses in all the languages, but its use in
ipal tenses is very rare.
In parts of Group 40 an indeterminate vowel occurs as a suffix, e.g. in TUM-
U (43b) -tambuy- 'understand', nitambuyu 'I have understood', -toggoy-
i', nitoigoyo 'I have spoken', or in DGAZIJA (44a) -som- 'read', risomo
ave read', -fuljg- 'shut', rifuigu 'we have shut'.
There are negative tenses in most groups, e.g. KAGULU (12) ciaporjhola
ierced', neg. cisapoqhole; TUBETA (21) tuhira 'we shall work', neg. setu-
-e; DGAZIJA(44a) ggariwahao 'we are building', neg. ka'riciwaha; KLDGA
ptujkona 'we are going to lie down', neg. sjtukakone. The principal exception is
O(I i) which regularly prefixes si- to affirmative tenses as the sign of the negative.
In many languages of this zone there is no formal sign for relative clauses,
order alone indicating the nature of the syntactical relationship, e.g. in DGAZIJA
hawono esio nahula 'he has seen the book I bought' (cf. nahula esio 'I
it a book'). ASU (22) uses a difference in tone-pattern to characterize relative






50 THE CLASSIFICATION OF
clauses, e.g. mugeni eneza ( __ _) 'the stranger who will come' (cf. muj
ene.a (--- ) 'the stranger will come'). KAGULU (Iz) is unusual in ha
tenses in relative clauses which do not occur in main clauses, e.g. gano ma
gonihandile 'these are the trees I planted' (go- is a special relative prefix, and
base -handile is apparently not used in principal tenses). The relative construe
used in SWAHILI (42) is not characteristic of this zone but of Group E.7o.
13. Most of the languages of this zone have simple consonant alternances.
14. There is a five-vowel system throughout the zone, except in SADGO
and KIDGA (65), which have a seven-vowel system.
15. Apart from Group 60, which has two quantities of vowel in radicals, a si
quantity is characteristic of the zone.
I6. There is a tendency to some form of penultimate prominence in cei
languages of Groups o2 and 30. This feature, which is somewhat rare in Bantu
guages, is, however, regularly present in the form of stress only in SWAHILI(
even in the PEMBA Dialect Cluster (43) it is by no means the general rule, e.1
TUMBATU (43b) he'neneza 'he has not replied', ako'za 'he will sell'.
17. In Groups 30 and 40 the junction of a nasal consonant with a voiceless plc
is sometimes heard in speech without the plosive, but with strong aspiration, e.1
RUGURU (35) -kulu 'big', iijguwo ighulu 'a big cloth'; occasionally a voice
nasal is heard instead of the aspirated nasal. In Group 60 there is frequently nei
a plosive nor any aspiration in the pronunciation of such junctions, e.g. in BI
(63) -tali 'tall', indege inali 'a tall bird'.
18. Lexical tone on the radical occurs only in Groups 20 and 60, while non
suffixes have a lexical tone only in Groups o1 and 20. There is grammatical tor
each of these three groups; e.g. in GOGO (I i), where the fact that any given ten!
a certain shape can only have one tone-pattern shows that there is no lexical ton
the radical, there are three tenses distinguished by tone-pattern alone, as ciaw
(-_-) 'we returned (long ago)'; ciawuya (_--) 'we returned (yesterday)'; ciaw
(_--) 'we have just returned'. In Groups 30 and 40 there is neither lexical
grammatical tone in most cases.
Summary
There is a clear boundary between parts of this zone and the adjacent ones, as
example, between Group G.io and Group F.3o, or between Group G.40 and Gi
P.io. In other cases, however, the relationship across the zone boundary is ir
closer, as between Group G.2o and Group E.7o; nevertheless a considerable mea
of linguistic homogeneity is achieved by the formation of this zone.


ZONE H
GROUP IO
II Vili 16 KODGO i6e N.E. KODGO
12 Kunyi i6a E. KODGO I6f KODGO
13 Bembe i6b YOMBE i6g S. KODGO
14 Ndingi i6c SUNDI I6h ZOMBO
15 Mboka i6d BWENDE






THE BANTU LANGUAGES 51
GROUP 20 GROUP 30 GROUP 40
21 NDODGO 31 YAKA 41 Mbala
22 Mbamba 32 Suku 42 HUDANA
23 Sama 33 Huigu
24 Dgola 34 Tembo
25 Bolo 35 Mbangala, &c.
26 Soqgo 35a Mbargala
35b Yoigo
36 Sinji
Characteristics of the Zone
e difficulties of this zone are great, in view of the peculiar one-sidedness of the
ble data. On the one hand the KODGO (16) dialects are well known, as to both
ulary and structure, but apart from NDODGO (21) we are faced on the other
with an almost complete lack of information. This means that the grouping is
able, and also that any description of the zone must be scrappy. From what we
.ow about these languages it seems reasonable to put these four groups into one
and here are some of the features which may be said to characterize it.
There is a high proportion of vocabulary peculiar to these languages.
There are many series of related nominals; an example from KODGO (16) was
in Chapter II, and here is one from NDODGO (21): -sal- 'sieve', musari/
i 'sifter(s)', musalu/misalu 'sieve(s)', risarilu/masarilu 'sifting-place(s)'.
Lone B.)
Extended radicals are of very frequent occurrence, in most cases being much
common than simple radicals. (Cf. Zone K.)
In most languages there are genders which regularly contain words indicating
ness or bigness, e.g. in NDODGO (21) kanzo/tunzo 'small housess' (cf.
izo houses(s)'. As in this example it is usual for the stem of these words to
nence with an element similar in shape to that of the prefix of another class. In
)GO (i6f) there is the peculiar prefix fi- which forms a gender with no plural,
inzo 'small house'. (Cf. Zone B.)
The independent nominals have a single prefix in all these languages except
3DGO (I6g) where we find forms like ediaki/omaaki 'egg(s)'.
There are extra independent prefixes in this zone, e.g. YAKA (31) has ha-, ku-,
nu-. KODGO (16) is an interesting border-line case, since it uses the prefixes ba-,
and mu-, as in bantu 'on the head' (cf. ntu 'head'), but it also forms more com-
y peculiar compounds which behave like one word, e.g. bana-ntu 'on the head',
a bana- is identical in shape with the self-standing word bana 'that place there'.
Groups io, 30, and 40 frequently have nasal consonants in dependent prefixes such
i-, e.g. in YAKA (3 ) miinda miama mimi 'these lamps of mine'. (Cf. Zone K.)
The second person plural prefix is lu- or nu- throughout the zone, e.g. in
)GO (i6f) lutaiga 'you read'; in NDODGO (21) nuaniana 'you have stolen';
AKA (31) luzayi 'you know'. (Cf. Zones B and R.)
Nominals used as sentences have invariable particles prefixed to them, e.g. in
)GO (16) ibata diami 'it is my village' (cf. bata/mabata villages(s)'. In
")DGO (2TV. however, the invarinhle nominal crn qstnrl nan Q entpnrtp withinlt nr







52 THE CLASSIFICATION OF
change of shape, e.g. ina ialu iami 'those are my chairs' (cf. ialu 'chairs',
HUDANA (42) there is a modification of the prefix of nominals used as senate
e.g. musigi wu aamukufi 'this string is a long one', pl. misiqi mi miamiku
musiqi mukufi 'long string', pi. misigi mikufi).
io. The suffix *-E does not occur in dependent tenses in Groups o1 and 30, 1
found in Group 20. For example, in KODGO (16f) luavutuka '(that) you sl
return', and in YAKA (31) tuakota '(that) we should enter'; but in NDODG(
tukune '(that) we may plant'. (Cf. Zone L.)
ii. The suffix *-.LE occurs in all parts of this zone, and is heard as -ile,
-ele, or -ene according to the vowel of the radical. (Cf. Zone B.)
12. In general true negative tenses do not occur in the languages of this zon
KODGO (i6f) the negative sign is ka . ko, e.g. katusumbidi ntumbu k
have not bought a calabash' (cf. tusumbidi 'we have bought'); in NDODG(
the negative sign is kii . ee, and this is sometimes associated with a differed
the dependent prefix, e.g. kii kaasuririee 'he has not forged' (cf. uasurire 'l
forged'); and in YAKA (31) the sign of the negative is the extra suffix -ko
tuzayi 'we know', neg. tuzayi-ko.
13. The consonant alternances are not unusual, but, as in some other zones,
is an alternance k/g only injunction with nasal consonants. For example, in KO
(16f) -kamb- 'speak' does not have to be distinguished from -gamb-, but lk
'kindness' is distinct from iganga 'medicine-man'.
14. There is a five-vowel system throughout the zone.
15. An alternance of vowel quantity appears to occur in most of these langl
e.g. in HUDANA (42) -beet- 'strike' has a different quantity of vowel from
'sell'. (Cf. Zones B and R.)
16. There is radical stress in these languages.
17. The tonal systems of this zone are fairly complicated, syntactical tone beinl
common. One characteristic of Groups Io and 40, at least, is that although th
lexical tone on the radical, there is none on the nominal suffix. This means th
nominals of any given shape there are never more than two possible tone-patte
a given context.

ZONE K
GROUP 10 GROUP 20 GROUP 40
11 CIOKWE 21 LOZI 41 TOTELA
12 LUIMBI 42 SUBIA
13 LUCAZI
14 LUENA GROUP 30
15 MBUNDA 31 LUYANA
16 Nyeigo 32 MBOWE
17 Mbwela 33 Mpukusu
18 Dkaigala 34 MaPi
35 Simaa
36 janjo
V --lnll '2






THE BANTU LANGUAGES 53
Characteristics of the Zone
some respects this zone is half-way between G and L, but there is still good reason
making it, in spite of its curious geographical distribution. One of the weak points
e classification is Group 30, where the available data are altogether inadequate.
ever, apart from this, the remainder is fairly reliable. It proves most convenient
to describe first the features which are common to all the groups, and then those
ire peculiar to one or more of the groups.
*atures common to all the Groups
Most of the languages of this zone appear to have a gender ka/tu which
arly includes words indicating smallness.
There does not seem to be an extension -u- in these languages, or even true
ve verbals.
Three extra independent prefixes ha- (or ba-), ku- and mu- occur in all
languages.
Nominals are used as sentences, but there are various types of prefix occurring
ch words. For example, in LUIMBI (12) aa mazi 'this is oil' (cf. mazi 'oil'); in
I (21) ze kilitipa 'these are the knives' (cf. litipa 'knives'); in LUYANA (31)
isamu ecile 'this is a tall tree' (cf. ecisamu 'tree'); in SUBIA (42) aa makonde
e are bananas' (cf. amakonde 'bananas').
A suffix -e occurs in affirmative dependent tenses.
The range of tense signs is very varied, and there is little that can be said to be
non even to one group. For example, in LUYANA (3 1), which only has one verbal
there are these four simple tenses, tunookayupa 'we heard', tunakuyupa 'we
i (to-day)', tuliakayupa 'we shall hear (soon)', tunambakuyupa 'we shall hear
* to-day)'; but in MBOWE (32) there are these quite different tenses on a similar
: tunakuyuva 'we heard', natukuyuva 'we heard (yesterday)', tuayuva 'we
heard', kamatuyuva 'we are hearing', matukayuva 'we shall hear'.
There are true negative tenses, and in any given language the negative sign is
lly constant. For example, in MBUNDA (15) the negative has the prefixed
ent ku-, as in kututuggu 'we shall not build' (cf. tutugga 'we shall build'); in
'ELA (41) the negative sign is ta-, but the negative tense corresponding to
:atenda 'we worked' is tatunakutenda.
There is a five-vowel system.
The radical has lexical tone.
Features peculiar to some Groups
The independent nominals have single prefixes in Groups 10-20, but double
xes in Groups 40-50. For example, in MBOWE(32) esitondo/eyitondo 'tree(s)';
.n TOTELA (41) ecikumba/ezikumba 'skin(s)'.
In parts of Group 20 there is a double dependent prefix in nominals, e.g. in
JNDA (15) moko iayihi 'a long knife', pl. bimoko biabihi.
The suffix *-ILE occurs in Groups o1 and 20 only, where it often forms part
e base of simple past tenses, e.g. in MBUNDA (15) -kok- 'pull', tuakokele 'we
d', -an- 'call', tuaanine 'we called'.
T_ -~ln r +1--- ;o r -4:9- _4 \ nrn~r;,e r,.,l.~~nl:






54 THE CLASSFICATIUN U1
distinct from -a. For example, in CIOKWE (11) -lim- 'cultivate', tunalim
have just cultivated', -tumb- 'plant', tunatumbu 'we have just planted' (cf. m
lima 'we are cultivating', mututumba 'we are planting'); in LUCAZI (13) -
'bring', tuanehe 'we brought', -hit- 'pass', tuahiti 'we passed' (cf. tuanehs
have just brought', tuahita 'we have just passed'); or in MBUNDA (15) -t4
'sew', tukatuigu 'we usually sew', -7ol- 'laugh', tukaoolo 'we usually laugh
tucitugga 'we are still sewing', tucigola 'we are still laughing').
5. There is an alternance of two quantities of vowel in radicals in Groups 3(
40 only. For example, in TOTELA (41) the quantity of the vowel in -bool- 're
is distinct from that in -bon- 'see'.
6. Although none of the languages of the zone has any stress, there is a ,
lengthening of the penultimate vowel in some languages of Group io, partict
CIOKWE (i1) and MBUNDA (15).
7. There are one or two features to note about the pronunciation of junctio
nasal consonants with voiceless plosives. In Group io the nasal is not usually
nounced in this case, but the plosive is aspirated, e.g. in LUCAZI (13) likor
khombo 'broom(s)'. In LUYANA (31) a voiceless plosive in junction with a
consonant is voiced, which means that the alternance voiceless/voiced does not
in plosives in this position, e.g. -cana 'small', umboggo unjana 'a small goat'.
8. In Group o1 there is no lexical tone on nominal suffixes. (Cf. Zone H.)
9. Grammatical tone is used to characterize the different forms, but rarely t(
tinguish them. Here is an exceptional example from MBOWE (32), katual
(_-__) 'we did not cultivate (before yesterday)', katualima (----) 'we did
cultivate (yesterday)', where the corresponding affirmative tenses are differed
shape, tunakulima (_-___) 'we cultivated (before yesterday)', natukulima (--
'we cultivated (yesterday)'.


ZONE L
GROUP IO GROUP 30 GROUP 40
II PENDE 31 LUBA-LULUA 41 KAONDE
12 Samba 31a LUBA-KASAI
13 KWESE 3ib LULUA GROUP 50
31c Laijge 51 SALAMPASU
32 KANYOKA 52 LUNDA
GROUP 20 33 LUBA-KATADGA 53 LUWUNDA
21 Kete 34 HEMBA
22 Binji 35 SADGA GROUP 60
24 LUNS A 61 Mbwera
24 LUNA62 DKOYA

Characteristics of the Zone
There is a striking similarity between the languages in these groups, although
less marked in the case of those in Groups 50 and 60. In general the gramm;
features of these languages are those which are usually considered to be typic







a languages. Here again it proves to be simplest to describe the characteristics in
sets, taking first those which are common to all the groups.

!atures common to all the Groups
There is a single prefix in independent nominals.
Extra independent prefixes pa- (or ha-), ku-, and mu- are in general use,
-egularly govern both nominal and verbal agreements.
Nominals are rarely used as sentences, some kind of copula being used in most
. For example, in LUNA (24) there is the element -i, which takes a dependent
i, as in bai baefi (pr. bee beefi) 'they are thieves'; or in SALAMPASU (51)
iponyi 'they are thieves'.
There is the suffix *-ILE throughout the zone, though it appears to be missing
)DGE (23).
There is no alternance g/- in these languages, since -g- only occurs in junction
a nasal consonant, while zero consonant only occurs in junction with vowels. The
ice of the alternance is clearly seen in the following pair of words from LUBA-
AI (3Ia) lueeso/gjgeeso 'pot(s)', where the prefixes can be shown to be lu/n.
symbol a stands for an indeterminate nasal consonant.)
There is a five-vowel system in all languages.
There is no stress or other form of word prominence in any language of this

When the second radical consonant is a simple nasal, the alternance 1/n in some
sions is obscured. For example, -tumin- 'send to' is not distinct from -tumil- in
if these languages.
There is an alternance of tone on radicals right through the zone.
There is a difference in the tonal behaviour of the dependent verbal prefixes and
for the Ist and 2nd persons. For example, in KAONDE (41) uapitile ( __ ) 'he
d' (where u- agrees with muntu 'person'), uapitile (-_--) 'you (sing.) passed'.

featuress peculiar to some Groups
An extension -u- occurs in most of the groups, but it is not found in 60. In
p io it occurs in the form -eu- (-iu-), e.g. in KWESE (13) -val- 'give birth',
wu- 'be born', -tum- 'send', -tumiu- 'be sent'; in Group 20 it occurs as part of
S(-ebu-), e.g. in SODGE (23) -lel- 'give birth', -lelebu- 'be born', -tum-
', -tumibu- 'be sent'; in Group 60 it appears as a long vowel, e.g. in DKOYA
-hem- 'give birth', -hemuu- 'be born'.
In Groups 50 and 60 there are double dependent prefixes in some nominals.
example, in LUNDA (52) mutondu uawuwahi 'a good tree', pl. mitondu
vahi; or in DKOYA (62) mutondo wautali 'a tall tree', pl. bitondo biabitali.
Extra suffixes to verbals, such as -ko and -mo, occur regularly in Groups 50
o, and here and there in Group 30. For example, in DKOYA (62) uaikala-mo
veekalamo) 'he sat in it'.
Dependent tenses are formed with a suffix -e (-i in Group 50) in all groups
it 20, where a suffix -a is used. Other tenses rarely make use of a suffix -e
.s zone.
The tense systems of most of these languages are simple. Usually there are not






56 THE CLASSIFICATION OF
more than two distinctions of past and two of future time expressed by me
tense signs. For example, in KAONDE (41) we find tuapotele 'we bought (
to-day)', tuapota 'we have just bought', tusakupota 'we shall buy (to-day)', tuk
'we shall buy (after to-day)'. This is by no means without exceptions, as in LI
(52), where there are four distinct past tenses referring to simple actions.
6. Negative tenses occur in Groups 10-30, but not in 40-60. For exam
KWESE (13), where the negative sign is -ko, the negative sometimes correspc
form to the affirmative, as in jgajiyile 'I knew', neg. qgajiyile-ko, but in
tenses there is a special negative form, e.g. mbagguvutuke 'I will return', gg
qguvutuka 'I will not return'. The principal exception occurs in LUBA-KATI
(33) and SADGA (35), where the sign of the negative is ke which may be use
any tense. In the other languages of Group 30, although the sign of the neg.
ka- there are true negative tenses which have no corresponding affirmative
KAONDE (41) the negative sign is keci ... ne which is apparently not even at
to the verbal. In LUNDA (52) the negative sign is hi- . -ku, where tl
element is affixed to the verbal and the second to the last word in the claus
tukuzata mudimu 'we will do the work', neg. hitukuzata mudimu-k
DKOYA (62) the special base -fua-ko receives the dependent prefix, and is fo
by the nomino-verbal in ku- to express the future negative, e.g. tukulaba 'w
count', neg. tufua-ko kulaba; but in other tenses there is the negative sign k
-ha, e.g. tualaba 'we counted', neg. kitualaba-ha.
7. Relative clauses are often constructed by means of dependent suffixes, ex(
Groups 20, 30, and 50. For example, in LUNA(24) aakamona'he saw', biakar
yi'when he saw'; or in LUWUNDA (53) asadil 'they do', yisadila-u '(things)
they do'.
8. There is an alternance of quantity in radical vowels in all of these lang
except those in Group 60.
9. In Group 50 there is no alternance i/e or u/o in suffixes. For exam
LUWUNDA (53) ipepu 'wind' is not distinct from ipepo.
io. In each group except 60 there is an alternance of tone on nominal suffix
11. Although there are tone-patterns to characterize the tenses in all these lang
it is rare for the grammatical tones to be the sole distinguishing feature. One e)
of such a distinction does occur, however, in LUNDA (52), e.g. uakama (_.
went to sleep (yesterday)', uakama (--_) 'he is asleep'.


ZONE M
GROUP 10 GROUP 20 GROUP 30
11 PIMBWE 21 WANDA 31 NYIKYUSA
12 Ruggwa 22 MWADGA
13 FIPA 23 NYIHA (Nyika)
14 RUDGU 24 MALILA
15 MAMBWE 25 SAFWA
26 Iwa
27 Tembo






THE BANTU LANGUAGES 57
GROUP 40 GROUP 50 GROUP 60
41 TAABWA, &c. 51 BIISA 61 LENJE
4ia TAABWA 52 LALA 62 SOLI
4Ib Sila 53 SWAKA 63 ILA
42 BEMBA, &c. 54 LAMBA 64 TODGA, &c.
42a BEMBA 55 Seba 64a TODGA
42b Dgoma 64b Toka
42c Lomotua 64c Leya
42d Nwesi
42e Lembue
Characteristics of the Zone
zone is much less homogeneous than the preceding one, but some of the,
in it may equally be said to display most of the typical Bantu features. For
tive purposes it is most convenient to take various differentia and indicate the
ition of each of them in turn.
Most all of these languages appear to have a gender such as aka/utu (or ka/tu)
regularly, though not exclusively, contains words indicating things of small size.
n extension -u-, which expresses a passive, occurs throughout the zone.
ouble independent nominal prefixes occur in all groups except 60.
he extra independent prefixes pa-, ku-, and mu- appear to be used through-
zone, and to control both nominal and verbal agreements.
extra suffixes, such as -po, -ko, -mo, are used with verbals throughout the zone,
Groups 30 and 40 they are used with nominals also. For example, in NYIK-
(31) pabutali 'at a distance', pabutali-po 'at a distance from it'.
ominals are regularly used as sentences. In Group o the nominal has the same
.s when used with a verbal, e.g. in RUDGU (14) icisu 'a knife', cii icisu 'this
fe'. In Groups 20 and 30 a single prefix is used instead of a double one, e.g.
LILA (24) ulukusa 'a rope', lukusa 'it is a rope'; the principal exception to
MWADGA (22), which replaces the first part of the double prefix with a-,
tala 'bed', acitala 'it is a bed'. In Groups 40 and 50 there is a single prefix
long vowel, e.g. in TAABWA (41a) ubusansi 'mat', buusansi 'it is a mat'.
up 60, where there are no double prefixes, another element is often prefixed,
ILA (63) bantu 'people', mbantu 'they are people'.
here is a verbal suffix -e throughout the zone, and normally this is the sign
native dependent tenses. Rarely it also occurs in principal tenses, e.g. in
'62) nitukalime 'we shall cultivate'.
suffix such as -ile occurs in all groups except 60, but its actual nature varies.
ups 10 and 20 it is -ile (or -Ile), but bases formed with it have fewer alternances
second radical consonant, e.g. in MWADGA (22) -let- 'bring' and -lek-
both have the same -ile base, -lesile. In Group 30 the suffix is -jle, but there
fference in the alternances of the radical consonants. In Groups 40 and 50 the
as an indeterminate vowel, being heard as -ele in sequence with -e- or -o-,
erwise as -ile, e.g. in BIISA (51) -pet- 'bend' has -petele, and -pat- 'hate'
Ltile.
ense signs tend to be numerous in these languages, though BEMBA (42a)






58 THE CLASSIFICATION OF
is probably an extreme case with about thirty by means of which distinct on4
affirmative tenses may be formed. One striking feature is the rarity of any el
like -jqga regularly indicating actions in progress. There is such an element in
(62), e.g. tulalimi 'we have cultivated', tulaliminga 'we are cultivating', but
be seen from this example its use is peculiar, since it appears to be added n,
simple tense, but to one indicating a completed action.
io. In most of the groups there are special negative tenses, but in Io the
negative element, such as -ta- or -si-, which appears to form negative tenses
spending to the affirmative. Elsewhere the negative tense is often quite distinc
the affirmative, e.g. in NYIKYUSA (31) afjkjle 'he has arrived', neg. akafil
in LENJE (61) tulaakulima 'we will cultivate', neg. teetukaliime.
ii. There is an alternance g/- in radical consonants in Groups 10-3o but
the others, the chief exceptions being that it is missing in MAMBWE (15) and
and is present in TODGA (64). For example, in SAFWA (25) -gog- 'kill' is d
from -og- 'wash', but in a language like BEMBA (42a) there is nothing like thi
12. There is no alternance f/v or s/z in Groups 30 and 40 (or in most of Gr
and 61, 62). For example, in NYIKYUSA (31) -sjmb- 'write' is not distinct
-zjmb-, but in Group 20 these might be different radicals.
13. Other alternances which are absent throughout the zone are l/d, l/r, s/J
14. There is a five-vowel system in radicals in Groups 40-60, and a seven-
system in Group 30. In the two remaining groups there is a mixture; thus the
seven-vowel systems in I1-14, but a five-vowel system in 15, and 21-3 hal
vowels but 24, 25 have seven.
15. There is an alternance of quantity in radical vowels in Groups 30-50 bu
in Groups 2o and 60. In Groups io there is a mixture, since PIMBWE (i) ]
such alternance but 13-15 have.
16. In Groups 40-60 there is an indeterminate alveolar consonant in some
radical syllables, but in the other groups there is not. For example, in LENJ
-tey- 'prepare', -teyel- 'prepare for', -tern- 'cut', -temen- 'cut for', which:
that this extension has a consonant which is heard as -n- or -1- according to w]
the second consonant of the radical is a nasal or not.
17. In general these languages have no form of word prominence, but although
appears not to be used anywhere, WANDA (21) and MWADGA (22) have a slit
crease in length in the penultimate syllable, e.g. in WANDA (21) tuakala 'wi
just bought' is pronounced twa:ka-la, and tuakazile 'we bought' twa:kazi-le
18. The syllable arising from the junction of two vowels usually contains ;
vowel, even in those languages which have no alternance of vowel quantity in ral
For example, in LENJE (61) muana 'child' is pronounced mwa-na. Sin
vowels in junction with nasal compounds are always pronounced longer, e
TODGA (64a) -samb- 'wash' is heard as -sa-mb-.
19. NYIKYUSA (31) makes no use of tone, either lexical or grammatica
elsewhere there is an alternance of tone in radicals, except in FIPA (13) and RU
(14), and on nominal suffixes except in PIMBWE (11). In LAMBA(54) and the
of Group 60, however, there are only three tone-patterns for dissyllabic nominal
instead of the four that might have been expected if there were a full double alten
20. Grammatical tone is freouentlv the only wav of distinguishing tenses







entical shape, although Group 60 appears not to have this feature. For example,
ADGA (22) tuaiza (---) 'we have just come', tuaiza (-_-) 'we came (before
ay)', tuaiza (--_) 'we shall come (soon)'; or in FIPA (i3) tualimjle (--__
tivated (before yesterday)', tualimfle (-_-) 'we cultivated (yesterday)'.
dependentt verbal prefixes, agreeing with a nominal, have a different tonal
iur from those of the first and second persons. For example, in MAMBWE
aalola (_-__) 'they will look' (a- agreeing with antu 'people'), tumalola
we will look'; or in LENJE (61) baatafuna (---) 'they will chew', tuatafuna
'they will chew'.
Summary
tough some isoglosses which are important in separating groups elsewhere cut
trough the groups in this zone, yet the boundaries are well defined. In some
s too, as will have been noted, there is much in common between these groups,
of an apparent lack of homogeneity. For all that, this zone probably illustrates
early than others the basic fact that the sorting of groups into zones is primarily
)hical, though with as much linguistic justification as possible.


ZONE N
DUP 10 GROUP 20 GROUP 30
MANDA 21 TUMBUKA, &c. 31 NYANJA, &c.
DGONI z2a TUMBUKA 3Ia NYANJA
MATEDGO 2Ib POKA 31b CEWA
MPOTO 2Ic KAMADGA 31c MADANJA
TODGA 2Id Seqga 32 Mbo
z2e Yombe 33 Mazaro
2If Fulgwe
2Ig Wenya
2ih Lambia GROUP 40
21k Wandia 41 NSEDGA
42 KUNDA
43 NYUDGWE
44 SENA
45 Rue
46 Podzo
Characteristics of the Zone
groups which constitute this zone have many similarities to one another, and
cases they are quite different from the neighboring ones in other zones. In
this it proves to be of little value to attempt to distinguish the features which
unon to every group from those which are not. Instead the distribution of
differentia will be described in turn.
. Group 1o the independent prefix which serves as the singular of mi- is an
ninate nasal consonant. For example, in MPOTO (14) rjkogjgo/mikojgo






O0 Ir, .l.-AOO1 Ir I IIUiN Ur
'tree(s)'. In the other groups this prefix is usually m-, e.g. in TUMBUE
mlomo/milomo 'lip(s)'.
2. The independent prefix governing the agreement li- is itself lj- in Gi
e.g. in MPOTO (14) l4h|ba ili 'this broom'. In the other groups it is usually 2
not infrequently gives rise to modifications in the pronunciation of the firs
consonant. For example, in NYUDGWE (43) phiri/mapiri 'hill(s)', ti
masomba 'fish(es)'.
3. An extension -u- occurs in each group, but it is used by itself in Gr
and 20 only, e.g. in TUMBUKA (zia) -kom- 'kill', -komu- 'be killed'. Ii
30 it occurs in the longer form -iu- (or -eu-), e.g. in NYUDGWE (43) -gui
-guriu- 'be bought'. In Group 30 it is part of the compound extension -i
-edu-), e.g. in NYANJA (3 a) -majg- 'tie', -marjgidu- 'be tied'.
4. Independent nominals have single prefixes throughout the zone.
5. There are extra independent prefixes capable of governing both nom
verbal agreements, such as pa-, ku-, and mu-, in each language.
6. Double dependent prefixes occur here and there, but without any rel
For example, in NYANJA (31 a) certain stems behave like -tari in these exam
others do not, cintu cacitari 'a long thing', pl. zintu zazitari.
7. The first person plural verbal prefix is ti- right through the zone. I
exception occurs in MATEDGO (13), where the prefix tu- is used with future,
but even here ti- is used with past tenses, e.g. tusegga 'we shall build', ti
'we have built'.
8. Nominals are used as sentences, but usually have ni- or ndi- prefixed
in this case. In Group o1 the use of ni- is by no means without exception
MPOTO (14) lugoye 'a rope', nilugoye or lugoye 'it is a rope'. In TU1\
(21) a nasal consonant is prefixed in most cases, e.g. cijaro 'a door', ncijar
door'.
9. A suffix -e is the usual sign of the dependent tense, and in some ca:
also used to form the base of principal tenses.
o1. There is no suffix like -ile in Groups 20-40. In Group io there is a
of suffixes, but -ile only occurs in a few bases in MPOTO (14). In MAN:
there is a suffix -Iti, e.g. -tot- 'sew', titotiti 'we sewed'; in DGONI (12) the
e.g. -jegrg- 'build', tijeigj 'we built'; while in MATEDGO (13) there is I
-Ite, e.g. -phal- 'pull', tiphalite 'we pulled'.
i 1. In most of these languages there are few distinctions of time express
tense signs, the principal exception being MANDA (II), where there are four
past tenses and four future.
12. Actions in progress are mostly referred to by means of the nomino-\
ku- together with a copula -li, e.g. in NYANJA (3ia) -fun- 'search', tinaf
searched', tinali kufuna 'we were searching'.
13. True negative tenses are uncommon, elements either being affixed to tl
or occurring elsewhere in the clause as self-standing words. NYANJA (3ia)
examples of the former type, e.g. tinapita 'we passed', neg. sitinapita. The
type occurs in MANDA (ii) tiletijt kibiga 'we brought a pot', neg. tiletit]
lipa; and in TUMBUKA (2za) tikawona zovu 'we saw an elephant', n
tikawona zovu cara.






inr DtAIN I U IN -7AjUAU1-T 01
'here is usually no special word or element to indicate the relative construction,
in some languages there is a special tone-pattern for tenses in relative clauses.
lternances of the type p/ph occur in first radical position in most of these
es. For example, in TUMBUKA (zia) -par- 'scrape', -phar- 'tell', or in
[A (3 a) -pit- 'pass', -phik- 'cook'.
in alternance 1/d is found in this zone, as, for example, in NYANJA (31a),
lul- 'froth up' is distinct from -dul- 'cut across'.
Radicals commencing with nasal compounds are found in verbal bases in some
, languages. This is rare in Bantu languages. For example, in TUMBUKA
igir- 'enter', or in NYANJA (3 a) -mver- 'obey'.
7UMBUKA (21a) is peculiar in having the alternances p/b/b and k/g/g,
ir- 'scrape', -bab- 'give birth to', -bab- 'irritate'.
n Group io there are seven-vowel systems, but elsewhere only five vowels are
n radicals.
[here is no alternance of quantity in any of these languages.
'he languages of this zone are notable for the different voiced labial sounds
:ur in them. In MANDA (11) there is a labio-dental semi-vowel, e.g. -uik-
ihere the first consonant appears to be distinct from -w-. In TODGA (15)
a labio-dental plosive which is distinct from the bilabial plosive, e.g. -cbar-
-bar- 'give birth to'. In POKA (azb) there is a 'v' without friction, which is
from both w and v, as in -vur- 'lack'. In NSEDGA (41) there is a fricative
ich has to be distinguished from the pure w, e.g. -wir- 'proclaim', -wir-

Ihere is no form of word prominence in Groups Io-zo, 40, but in Group 30
penultimate vowel length, which is confined to the last word in the sentence.
simple, in NYANJA (3Ia) tadula 'we have cut', pr. tadu-la; tadula qkuni
,e cut firewood', pr. tadula ikhu-ni; tadula ikuni kumudzi 'we have cut
d by the village', pr. tadula ikhuni kumu-dzi.
In the languages of this zone syllables arising from the coalescence of two or
owels do not usually contain long vowels, e.g. in TUMBUKA (21a) muana
is pronounced mwana. This is not without exception, especially when the
wels are similar and both are grammatical elements, e.g. in NYANJA (3 a) the
ng two words have the same number of syllables, but the first syllable in the
distinctly longer than that in the second: aaseka 'they have laughed', aseka
iugh'. On the other hand, vowels in junction with nasal compounds are never
nced with increased length in these languages; thus the first vowel in these
irds has the same length in MPOTO (14) njoka 'snake', irkorgo 'tree'.
A nasal compound containing a voiceless consonant is usually aspirated in these
ges. For example, in NYANJA (3Ia) mpasa 'mats' (pl. of lupasa) is pro-
d mphasa. In Group io this also happens in the singular of the n/mi gender,
MANDA (II) iJkoggo 'tree' (sing. of mikoigo) is pronounced gkhoggo.
[n TODGA (15) and TUMBUKA (21) there are some special speech sounds
arise from double junctions of the following kinds, as in TUMBUKA (21a)
'to dry up', pr. kupxa; kupia 'to get burnt', pr. kuppa.
There is an alternance of tone on the radical in MANDA (i ) only in this zone.
ANJA (3 Ia) there is an alternance of tone on nominal suffixes, giving rise to two







64 THE CLASSIFICATION OF
'neck(s)', lugoji/jgoji 'rope(s)', where igosi and jgoji are really akosi an
respectively.
zo. The only form of word prominence occurs in the languages of Group I1
there is a slight stress on the radical.
21. There are no lexical tones in Groups o1 and 30. In Group 20 there is a
nance of tone on radicals and also on nominal suffixes.
22. Grammatical tone is used to characterize tenses in Groups io and 2o,
case has yet been observed where it serves to distinguish them.
23. In Group 20 there is a correlation between tone-patterns and syntactical r
ships in some cases, e.g. in YAO (21) saasu sijaasiice (-- ---) 'the fire
lost', acila saasu cila sijaasiice ( -- -- -- ---) 'that firewood is lost'.


ZONE R
GROUP IO GROUP 20 GROUP 30
11 MBUNDU 21 KUANYAMA 31 HERERO,
12 Ndombe 22 NDODGA 3ia HERERC
13 Nyaneka 3Ib Mbandiei
3ic Cimba

GROUP 40
41 YEEI
Characteristics of the Zone
This zone is sharply distinguished from its neighbours, but it is not easy to
the features which are peculiar to it. This is largely because the characteristic
separate it from the languages on the north (i.e. in Zone H) are different fro
which separate it from those on the east (i.e. Zones K and S). For this re
attempt is made to divide up the differentia into two sets.
1. In every language there is a gender which regularly, though not exc
contains words indicating small things. In MBUNDU (i1) it is oka/otu, in
20 and 30 it is oka/ou, and in YEEI (41) it is ka/tu.
2. In most of these languages extended radicals are commoner than simple
Thus in the standard vocabularies of Groups io and 20 there are less than 20o
of simple radicals among those used for forming verbals, against the more 1
per cent. Examples of these may be seen in the following radicals which do r
to occur in the unextended form: in MBUNDU (i i) -pitahal- 'pass', -talabs
3. An extension -u- appears to occur in all of the languages of the zone
express the passive. For example, in KUANYAMA) (21) -dal- 'give birth to'
'be born', or in HERERO (3Ia) -hind- 'send', -hindu- 'be sent'.
4. Double independent prefixes occur throughout the zone. In Groups I
first part of the prefix is usually o-, but there are the following exceptions.
case the class which governs the dependent prefix li- has e- as its independer
e.g. in NDODGA (22) eyego lioye 'your tooth'. In MBUNDU (11) the sir
the omu/oba and the omu/obi genders and the plural of the omu/oba






THE BANTU LANGUAGES 63
illy accompanied by an extra suffix -ni, e.g. in LOMWE (32) muhice 'river',
Ice-ni 'in the river'.
ie nomino-verbal prefix is ku- in Groups Io and 20, but in 30 it is u- or o-.
ie verbal prefix for the first person plural is tu- in Groups io and 20, but
0.
)minals are frequently used as sentences in these languages. Sometimes there
litional element, as in MAKONDE (23), e.g. citale 'iron', ncitale 'it is iron';
ies the nominal has the same shape as in other cases, e.g. in MAKUA (31)
?a ikina 'this is a small house', inupa ila ikina 'this house is a small one'.
suffix -e is the sign of the dependent tense in all the languages of this zone.
Lently occurs, however, in other tenses too, e.g. in MWERA (22) situtote
'we shall sew the cloth'.
'here is a suffix -ile in Group 20, although in some cases as MAKONDE (23)
occurs in negative tenses. In Group 30 there is only the one base, formed with
I in principal tenses. In Group o1 -ile does not occur, but -jte is common,
RUJHJ (12) uses the very unusual -jke, e.g. -son- 'sew', tusonike 'we
and in some cases even uses -e as an alternative, e.g. -pit- 'pass', tupitike
te 'we passed'.
'here are negative tenses in most of these languages. Their form is sometimes
to that of the affirmative, but any relationship varies from tense to tense, and
there a negative corresponding to each affirmative tense. Here is an example
WERA (22) which may be considered as typical: situcegge 'we shall build',
kacegga; tuaceigile 'we built', neg. tukanaacegga.
nfixed elements serve as substitute objects in each group, but in 30 they are
i to the m/a gender, e.g. in LOMWE (32) yamphwanya 'they found him',
ohwanya ela 'they found it' (i.e. 'house' empa).
elativee clauses are usually identical in shape and tone-pattern with principal
the word order alone indicating whether the clause is relative or not.
n Groups io and 20 there is no alternance s/z, e.g. in YAO (zI) -sito 'heavy'
t have to be distinguished from -zito. In MABIHA (25) and in the whole of
30 neither s nor z occurs.
n Group 30 only is there an alternance 1/r in radicals, e.g. in MAKUA (31)
ry' is distinct from -rik- 'draw (water)'.
Ln unusual alternance t/t occurs in parts of Group 30, as well as the aspirated
in which the affricates are really predental. For example, in MAKUA (31)
,arth', it4ala 'hunger', ithala 'veranda', itghapa 'trap'.
n Group io there are seven-vowel systems in the radical, but in Groups 20
only five-vowel systems.
'here is an alternance of vowel quantity in the radical in Group 20 only, for
e, in YAO (21) -jim- 'refuse' is distinct from -jiim- 'stand'. The only excep-
DGINDO (14) which also has an alternance of quantity, cf. litoosi 'banana'
;oji 'rope'.
'he junction of a nasal consonant with the first radical consonant involves the
ig things in most of Groups io and 20. The nasal is not heard before -s-,
fAO (21) lusasa/sasa 'wall(s)', where sasa is really osasa. The alternance
I a voiceless and a voiced plosive is masked, e.g. in YAO (21) lukosi/igosi








necK(sy, mgoji/rigoj 'rope(s), where jgosi and ggoji are really rkosi and
respectively.
20. The only form of word prominence occurs in the languages of Group ic
there is a slight stress on the radical.
21. There are no lexical tones in Groups io and 30. In Group 20 there is ai
nance of tone on radicals and also on nominal suffixes.
22. Grammatical tone is used to characterize tenses in Groups io and 20,
case has yet been observed where it serves to distinguish them.
23. In Group 20 there is a correlation between tone-patterns and syntactical re
ships in some cases, e.g. in YAO (21) saasu sijaasiice (-- __-_) 'the fire
lost', acila saasu cila sijaasiice (-_ - --- -) 'that firewood is lost'.



ZONE R
GROUP IO GROUP 20 GROUP 30
II MBUNDU 21 KUANYAMA 31 HERERO, 8
12 Ndombe 22 NDODGA 3Ia HERERO
13 Nyaneka 3Ib Mbandier
31c Cimba

GROUP 40
41 YEEI
Characteristics of the Zone
This zone is sharply distinguished from its neighbours, but it is not easy to i
the features which are peculiar to it. This is largely because the characteristics
separate it from the languages on the north (i.e. in Zone H) are different fror
which separate it from those on the east (i.e. Zones K and S). For this rea
attempt is made to divide up the differentia into two sets.
i. In every language there is a gender which regularly, though not exch
contains words indicating small things. In MBUNDU (ii) it is oka/otu, in (
20 and 30 it is oka/ou, and in YEEI (41) it is ka/tu.
2. In most of these languages extended radicals are commoner than simple r:
Thus in the standard vocabularies of Groups o1 and 20 there are less than 20 pi
of simple radicals among those used for forming verbals, against the more u
per cent. Examples of these may be seen in the following radicals which do n<
to occur in the unextended form: in MBUNDU (i ) -pitahal- 'pass', -talaba:
3. An extension -u- appears to occur in all of the languages of the zone,
express the passive. For example, in KUANYAMA) (21) -dal- 'give birth to',
'be born', or in HERERO (3Ia) -hind- 'send', -hindu- 'be sent'.
4. Double independent prefixes occur throughout the zone. In Groups I0-
first part of the prefix is usually o-, but there are the following exceptions. Ii
case the class which governs the dependent prefix li- has e- as its independent
e.g. in NDODGA (22) eyego lioye 'your tooth'. In MBUNDU (I1) the sink
the nmn/nhn and the nmui/nhi aenlder and the nlural of the omu/oba v


rTITTT7 -IT AO 'TT-'Td r",AHTT







i1. B AIN U LAIN UAUPs 05
enders only have double prefixes when the stem is monosyllabic or commences
owel, e.g. omuine/obiine 'finger(s)'but utima/obitima hearts(s). In Group
: are some classes with double prefixes consisting of two identical vowels with
evening consonant, e.g. in KUANYAMA (21) onjila/eenjila 'path(s)', or in
GA (22) osinima/iinima things(s). In YEEI (41) double independent pre-
- apparently used only with monosyllabic stems or with those commencing
rowel, and even then the same vowel is used in both parts of the prefix, e.g.
/imiya 'thorn(s)'.
:tra independent prefixes, pa-, ku-, and mu- are used in each language. In
o1-3o they are added to the double prefix, e.g. in HERERO (3ia) ondundu
uondundu (pr. kondundu) 'to the hill'. In YEEI (41) in addition to the
prefix there is also a compound form, e.g. pikali 'chief', kusikali 'to the
)ut muzi 'village', kuokumuzi 'to the village'.
Groups o1-30 extra dependent prefixes also are added to the double prefix,
/IBUNDU (I I) obiti 'trees', obianja biobiti 'branches of the trees'. In YEEI
lich does not ordinarily use double prefixes in independent nominals, a similar
word occurs through the use of an indeterminate vowel to link the extra
ent prefix. This vowel is heard as a, e, or o according to whether the
if the prefix is -a- (or zero), -i-, or -u- respectively, e.g. murumi/barumi
en', pipuna piomurumi 'the man's stool', pl. zipuna ziabarumi (pr.
imi).
double dependent prefixes in certain kinds of nominals regularly occur in Groups
4o. For example, in HERERO (3Ia) omuti omusupi 'a short tree', pl. omiti
pi, or in YEEI (41) pipuna sisikuru 'an old stool', pl. zipuna zizikuru.
ise they only occur in NDODGA (22), e.g. olutu olunene 'a big body', pl.
tu omanene.
ominals of one type or another are regularly used as sentences in each group.
simple, in MBUNDU (11) obiti 'trees' or 'they are trees'; or in NDODGA
gulu 'house', oggulu yianje 'it is my house'; and in YEEI (41) dipamba
'it is a hoe'.
here is a suffix -e which serves regularly as the sign of the affirmative
ent tense, but which is rarely used in principal tenses.
A suffix -ile occurs in each language, but sometimes, as in YEEI (41)
itly, it is more characteristic of relative than of principal clauses.
kn indeterminate vowel suffix occurs as a tense formative in some of the
e.g. in KUANYAMA (21) ohatuloggo 'we work' (-logg- 'work'), ohatu-
'we build' (-tugg- 'build').
L copula -li is rarely used as a tense auxiliary in these languages.
'he tense signs of the languages in these groups are not numerous, it usually
possible to refer only to one past time and one future time without the use of
:ime words.
Negative signs are fairly consistent in each language, but the form of the
e tense is often different from that of the affirmative. For example, in MBUNDU
- ... -ko is the sign of negative, as in katuakokele ukolo-ko 'we did not pull
ie' (cf. tuakokele ukolo 'we pulled the rope'), but the negative of tukoka
'we are pulling the rope' is katukoki ukolo-ko, and the base -koki does not






66 THE CLASSIFICATION OF
occur in any affirmative tense. Similarly in HERERO (3 ia) ka- is the negative
as in tumunine 'we found', neg. katumunine, but the negative of matumul
find' is katumuna.
15. Relative clauses in Group io may or may not be introduced by a linking
e.g. in MBUNDU (Ii) esala (elina) tuasagjga litito 'the egg we found is
elsewhere there are extra prefixes agreeing with the antecedent used with i
verbals. The chief exception to this is in NDODGA (22) which uses a speci
word formed from the stem -oka with n- prefixed to the dependent prefi
osiloiga sioka omuhoggi esiniggi 'the work the teacher has done', pl. ii
mbioka aahoggi yeyiniggi.
16. There are some unusual alternances in radical consonants in these lang
In Group 20 there is l/d, e.g. in KUANYAMA (21) -lil- 'weep', -dil- 'be tab
well as the rare d/nd, e.g. -dudum- 'growl', -ndudum- 'thunder'. In Gr(
there is t/t (and n/p), e.g. -tak- 'shake', -tar- 'look out'.
17. The consonant alternances in radicals are markedly different from tl
prefixes in some cases. Forexample, in MBUNDU ( i) b is distinct from both p
in radicals, as in -banj- 'look at', -pal- 'run away', -mal- 'finish', but ba
(pr. bonjila) is not distinct from paonjila 'on the path', similarly abanu 'pec
not distinct from amanu.
18. There is a five-vowel system in the radicals of each language.
19. There is no alternance of quantity in radicals in these languages, but
vowel of a radical is in junction with a similar vowel in an extension, this may si
a long vowel. For example, in KUANYAMA (21) where -fu-ul- 'strip' app
have a different quantity from -ful- 'rub', but has in fact one syllable more.
20. There is a slight lengthening of the penultimate vowel in MBUNDU (1
no stress.
21. Tonal data are only available for MBUNDU (i i), where there is an alterni
tone on radicals, so that -kul- 'plant' and -kut- 'tie up' have different tonal beh
There is, however, no tonal alternance on nominal suffixes, with the resu
nominals have only two possible tone-patterns. Those with disyllabic st(
represented by the typical words onjila (-__) 'path' and onjila (-- ) 'bird'.


ZONE S
GROUP 10 GROUP 30
II VENpA 31 XHOSA
32 ZULU, &c.
3za ZULU
GROUP 20 32b DGONI
21 TSWANA 33 SWAZI, &c.
2ia ROLOD 34 NDEBELE
2ib Kgatla
2ic Maggwato
22 PIDI
23 SUTHU








Characteristics of the Zone
.s zone is different in many ways from most of the others, since it contains only
groups, and these are much better documented than most. The most useful
f dividing up the features would have been to deal first with those common to
;t two of the three groups, and then with those peculiar to one group. To do
However, would have meant duplicating some of the data, so instead the occurrence
h of the features is indicated in turn.
[n these languages there are a number of radicals consisting of a single consonant
but no vowel. For example, -n- 'rain' occurs in almost every language. Such
Is are not unknown in other zones, but elsewhere there are rarely more than
of them, whereas the average number in this zone is ten. VENpA (Ii) is
tional in having about sixteen, while some of the languages of Group 20 have
ix in common use.
:n Groups 20 and 30 there are no genders which regularly include words in-
ig small or large things. In VENpA ( i) there is a gender ku/zwi in which are
words to refer to small things, e.g. kugi/zwi4i 'small villagess' (cf. muqi/miqi
e(s)'). A type of word-building which is not common in the other zones
bed so far gives rise to words of the following type in Groups 20 and 30. In
(22) maruana 'little clouds' (cf. maru 'clouds'), or in XHOSA (31) indluana
house' (cf. indlu 'house'), are examples of what are loosely called 'diminutives',
ke all other cases where the relationship between words is on the lexical level,
luite impossible to predict what the so-called diminutive of any given word

here is an extension -u- in each group, but in VENpA (11) it usually occurs
t of the compound extension -iu-, e.g. -to4- 'seek', -to4iu- 'be sought'. The
ar way in which junctions containing -u- are heard in Groups 20 and 30 will
erred to in a later section.
Extensions -el- and -is- do not have indeterminate vowels as they do in so
other groups outside this zone. For example, in ZULU (32) -thurg- 'sew',
gel- 'sew for', -qgen- 'enter', -ggenis- 'cause to enter'.
Double independent prefixes occur in Group 20 only, e.g. in DGONI (32b)
i/imithi 'tree(s)'.
Extra independent prefixes are not used with any regularity in the languages
Zone. In VENpA (11) they do not occur at all, the extra suffix -ni being used
ress a similar meaning, e.g. masimu-ni 'in the gardens'(cf. masimu'gardens').
oup 20 there is usually an extra suffix -rj, but in some cases an extra prefix is
.s well, e.g. in ROLOD (z2a) luapi 'sky', muluapi-rj 'in the sky'. The agree-
governed by such words is peculiar in that it is similar to that governed by the
.o-verbals, as muti-q kapjtsa 'inside the pot' (cf. pjtsa 'pot', kutlala kapjtsa
Ailing up of the pot'). In Group 30 the first part of the double prefix is usually
ed by e- when the extra suffix -ni (or -ini) is used, e.g. in DGONI (33b)
raleni 'in the path' (i.e. emgwala-ini, cf. umgwaya 'path').
)ouble dependent prefixes occur in Group 30 only, but the vowel of the first
f the prefix differs from that of the corresponding independent prefix, e.g. in
J (32) umuthi omude 'a tall tree', pl. imithi emide.







70 THE CLASSIFICATION OF
22. There is a difference in the tonal behaviour of the dependent verbal p
and those for the first and second persons in each of these languages. For ex:
in VENPA (1i) bafula (---) 'they forge', rifula (_--) 'we forge'; in SUTH
utlabona (---) 'he will see', utlabona (_--) 'you (sing.) will see'; or in X1
(31) BaBalile (-_) 'they have counted', siBalile ( ) 'we have counted'.
23. In Group Io the nominal suffix has an alternance of tone but extensii
not; this means that there is a maximum of four patterns for nominals of all le
In the other groups there is a tonal alternance both on extensions and on suff
nominals, which gives rise to a larger number of possible tone-patterns the long
stem.


ZONE T
GROUP IO GROUP 20 GROUP 30
II KOREKORE, &c. 21 TSWA, &c. 31 COPI
iia .SaJgwe 2ia Hleggwe 32 TODGA
IIb KOREKORE 2ib TSWA
iic Tabara 22 GWAMBA
IId Budya 23 THODGA
12 ZEZURU 23a HLADGANU
13 MANYIKA, &c. 23b Tsojga
I3a MANYIKA 23c Joijga
i3b Tebe 23d Bila
14 NDAU 24 RODGA
15 KARADGA
16 KALADA

Characteristics of the Zone
In some ways there is a fairly close relationship between Group 2o and the lani
of the previous zone. Since, however, the arranging of the groups into zones is .
dictated by convenience of reference, it is preferable to put these three groul
a zone by themselves. The following description of the characteristics follow'
same plan as that used in the previous zone.
I. As in the languages of Zone S, there are about ten radicals in most ol
languages which consist of a consonant only, as -n- 'rain'. In Group 30, ho
there are only seven or eight of these radicals, although even this is greatly in
of the two or three of other zones.
2. There are genders which regularly include words referring to small thinly
number of these languages. In 11-13 there is a ka/tu gender, and in KARj
(15) swi/bu, e.g. swiggurube/burgurube 'small pig(s)' (cf. ggurube/jg
'pig(s)'); while in some of the languages of Group zo there is fi/swi (or zwi),
TSWA (21) Jimutana/zwimitana 'small villagess); comparing these with
miti villages(s), it will be seen that the stem of the first pair is different as we
3. There is an extension -u- in each language. It serves to express the passii
I I 1r 1 A _7 T TTTTT /I.._ -\ .







THE BANTU LANGUAGES 69
roup spoken in northern Nyasaland, this particular series does not occur. Instead
is one in which the clicks have a sort of double sound, being released from a
Rex position and then flapping against the lower teeth.
SThere are seven-vowel systems in Group 20, but five-vowel systems in the
Groups. In most cases the number of vowels in speech is two in excess of that
e alternance.
There is no alternance of quantity in radical vowels in any of these languages.
. A striking feature of Groups o1 and 20 is the masking of the alternances of
rst radical consonant in junction with certain prefixes. For example, in VENDA
lternances such as t/r and k/h are obscured in junction with the indeterminate
consonant, as in lutagga/thagga 'reed(s)' (where the plural is really ftarjga
s heard as thagga) luragga/thajga 'pumpkin plantss' (where the plural is
Srajgga but is also heard as thaiga). In junction with zero prefix, which forms
s that serve as the singular of others with ma-, other alternances such as r/h
Iz/1 are masked, as in Jaho/maraho 'buttock(s)', JaQa/mahaQa shoulders(s'
.e Jaho and fa4a are really fraho and ha4a respectively, using the symbol
ie zero prefix). An example of the masking of the alternance p/b in junction
the indeterminate nasal infix is seen in these radicals from SUTHU (23) -pat-
', -mpat- (i.e. -npat) 'bury me', -bat- 'strike', -mpat- (i.e. -nbat-) 'strike me'.
In Groups 20 and 30 double junctions in which the middle sound is -u- are
times heard with quite different consonants. For example, in SUTHU (23)
ua 'to be paid' (cf. hulifa 'to pay') is heard as huliJwa; and in XHOSA (31)
umua 'to be bitten' (cf. ukuluma 'to bite') is heard as ukulunywa.
There is a marked type of word prominence in the languages of this zone, which
sts in the lengthening of the penultimate vowel. In Group 20 this is normally
aed to the last word in the sentence, but in others it usually occurs in each word.
There is an alternance of radical tone in each of these languages, but in Groups
Ld 30 there is a type of alternance which appears to be without parallel in other
i languages. Although these are two-tone languages, there are three possible
patterns for verbals with simple radicals in certain tenses. Thus in VENPA
lir- 'hoe' and -rum- 'send' have quite distinct tonal behaviour, and in addition
- 'carve', which is usually similar tonally to -lim-, has its own patterns in three
ir tenses. In XHOSA (31) also there are three kinds of radical from the point of
of tonal behaviour, thus -lim- 'cultivate' and -thum- 'send' are tonally quite
.ct, but -6oph- 'bind' which usually behaves like -thum- has different patterns
me tenses. This type of threefold tonal alternance distinguishes these groups
all others.'
ais curious fact may be explained historically by assuming that radicals like the Venda -bal-
e Xhosa -8oph- originally had long vowels, as indeed related radicals do in those languages which
n alternance of vowel quantity. It is found that the tone-patterns of verbals with these radicals
ntical with those that are obtained by telescoping similar patterns for radicals with one extension.
n Venla robaga (--_) 'we carved' has a tone-pattern distinct from that of rolima (---) 'we
but if the second and third tones of the patterns of rolimela (---) 'we hoed for' are merged,
similar pattern is obtained. If then -bad- originally had a long vowel and behaved tonally like
al with an extra syllable (as actually happens in many languages of other zones), it is easy to see
ie present tonal behaviour arose. An exactly similar explanation will account for the cases where
laviour of radicals like -boph- in Xhosa differs from those like -thum-. This means that though
mrnnce of nuantltv dlisannenarpd 1t some timn thl- nnf -1. a 4 n -n proi Al.a .- ;.






70 THE CLASSIFICATION OF
22. There is a difference in the tonal behaviour of the dependent verbal p
and those for the first and second persons in each of these languages. For ex
in VENPA (11) bafula (---) 'they forge', rifula (---) 'we forge'; in SUTH
utlabona (----) 'he will see', utlabona (__--) 'you (sing.) will see'; or in X:
(31) 6a6alile (- __) 'they have counted', siBalile ( ---) 'we have counted'.
23. In Group o1 the nominal suffix has an alternance of tone but extensi
not; this means that there is a maximum of four patterns for nominals of all 1
In the other groups there is a tonal alternance both on extensions and on sufl
nominals, which gives rise to a larger number of possible tone-patterns the lon;
stem.






THE BANTU LANGUAGES 71
-rasu- 'be thrown away'; or in COPI (31) -wogg- 'deceive', -woggu- 'be deceived'.
In Group 20 the extension usually occurs as part of the compound -iu-, e.g. in
HLADGANU (23a) -kum- 'find', -kumiu- 'be found'.
4. Extensions -el- and -is- occur in Groups 20 and 30 with these vowels, e.g.
in RODGA (24) -yis- 'carry', -yisel- 'carry for'; -bon- 'see', -bonis- 'show'. In
Group io, on the other hand, both of the corresponding extensions have the same
indeterminate vowel, e.g. in NDAU (14) -par- 'scrape', -parir- 'scrape for', but
-pet- 'bend', -peter- 'bend for'; and -kur- 'grow', -kuris- 'cause to grow', but
-pon- 'get well', -pones- 'cure'.
5. Independent nominals have single prefixes throughout the zone.
6. Extra independent prefixes occur regularly only in Group io, where they
govern both nominal and verbal agreements. In the other two groups nominals with
the extra suffix -ni govern the same agreements as nomino-verbals, which have the
prefix ku- or gu, e.g. in TODGA (32) nyumba-ni guamuntu 'in the house of the
man'. COPI (31) in addition to the extra suffix sometimes has the extra prefixes ha-
and mu-, but these make no difference to the agreements, e.g. munyumba-ni
kuakue 'in his house'.
7. Double dependent prefixes appear to occur in COPI (31) only, e.g. mndogga
wawunene 'a good tree', pl. mindogga yayinene.
8. In Groups 20 and 30 the extra dependent prefix is linked to the nominal with
the common -a-, but in Group o1 by an indeterminate vowel. This is heard as a, e,
or o according as the vowel of the prefix is a, i (or zero), or u respectively, before the
zero prefix which serves as the singular to ba- it is heard as with a, e.g. in ZEZURU
(12) musue uetsoko 'the monkey's tail' (tsoko/tsoko monkeys(s)', rutsoka
ruomunhu 'the person's foot', pl. tsoka dzabanhu, but rutsoka ruatenzi 'the
master's foot', pl. tsoka dzabatenzi. The principal exception to this is in MANYIKA
(13) where the linking vowel is usually -e-.
9. Nominals are used as sentences in Group io with no modification of the prefix,
e.g. in ZEZURU (Iz) rukoba rupami 'a wide river' or 'the river is a wide one'. The
principal exception in this group appears to be NDAU (14) which apparently prefers
the copula -ri in such cases, e.g. iyi iri mhatso yaggu 'this is my house', pl. idzi
dziri mhatso dzaggu. In the other groups there is usually an additional element
prefixed to the nominal, e.g. in RODGA (24) bafambi 'travellers', ibafambi 'they
are travellers'; in COPI (31) the element is different for independent and dependent
nominals, e.g. mndoiga 'a tree', imndogga 'it is a tree', cilo ncacinene 'the thing
is a good one', pl. silo nsasinene.
10. A suffix -e is the sign of the dependent tense in the affirmative in Group Io,
and in TODGA (32) also. In Group 20, however, this suffix is not used, as in COPI
(31), e.g. -dzib- 'know', micidziba '(that) you should know', i.e. -ci- -a is the sign
of the dependent tense.
i1. There is a suffix -ile in Groups 2o and 30, but in COPI (3 ) -ite is used with those
radicals which consist of a consonant only, e.g. -pf- 'hear', hipfite 'we have heard'.
12. The tense systems of these languages are not very complex, and there are rela-
tively few tense signs. At most a double distinction of past time appears possible and
a single future time reference, without the use of time words, e.g. in ZEZURU (12)
-pind- 'enter', tapinda 'we entered (to-day)', takapinda'we entered (before to-day)'.








-hum- 'come out',
go out', i.e. -na- -E
I4. There are ne


ch of 1
L-, as i:
tense s

in Gi
ree wi


special fixed


clause is introduced by link v
some of the tense signs used
in TODGA (32) nadiwona'
sign -rjga- -a is peculiar to :
16. There are very large ,
guages, TODGA (32) having
position.
17. There are five-vowel s
different vowel qualities are J
I8. In Groups to and 30
masked in junction with certa:
s/ts is obscured in junction
tsero/masero baskets(s), bi
really jsero and ftsero resp
are masked in junction wit
likhokho/magokho 'coconi
19. In Groups o1 and 30 t'
involves the use of special cc
'child' is heard as mrjana,
gubgwa.


sometimes is the only c
ticafamba (_---) 'we


Many of the gaps in
are due to the income
owing to the limitation
large amount of inforn
In making this selei


ve ver
dijgai
rerbals
conso


out', i.e. -di- -ile; h

the groups of this zoi
n hinabala 'we shall
sign as well in the neg

group Io by a special I
th the antecedent, e.
le did', pl. zwinu zv
uffix to the verbal in
uliarity in many of th
bals do not occur in
vona-go '(which) I s

nant alternances in r;


teams in each of the languages of this
:quently heard in speech.
ie alternances in the first radical cons
prefixes. For example, in ZEZURU (I:
ith the zero prefix which has ma- as
tsara/matsara lines(s), where tser<
tively. Or in TODGA (32) alternanc,
the prefix li-, e.g. likhoha/makho
s)' where likhokho is really ligokho
pronunciation of double junctions wit]
sonant clusters. For example, in NDI
while in TODGA (32) gubua 'to dry

nly available for Group Io, where the
nominal suffixes. Grammatical tone .
shing feature in tense formation, e.g. ii
avel', ticafamba (--_) 'we are still

GENERAL CONCLUSION
descriptions of the characteristics of th(
of the available data. Others, howe
ed by the size of this work. This mea
electedd careful selection had to be ma
-'1---- __*


described from zone 1
The main intention


This reason correspondii
possible.
ons and examples, howev(






THE BANTU LANGUAGES
)f the classification. From those which have been given it shnr


is not. Moreover, they have t]
-e of differentia which is inesc:
e looked in vain for some indica
.-. --A ^- L -I- ---i 1 1 -- -'- -


. .--.-----------.......-----........
mental. This means chiefly that it lays
idation on which something more pei


'.. AAX^LllJLr3 U lOllllC%,/C, %
ief the element of arbitrari
grouping of languages. Ai
seness of the relationship b
-e is no standard against w
cessary to avoid expressil
ily objective basis.
rork, it is avowedly tentati
im to finality, but is to be
+ m--, :- +m- l-n u, 1:.









A,1iMiIr ,L10 1 UV I rM tilN I


the speakers of that language as far as is known. The others given in paren
include the most important of those names which have been used at some ti
other to refer to it.


r
1)


e

I
.1


vvnere me prenx to oe
laced immediately after
iterpreted as meaning th
evidence on the point.

GROUP IO
.x 1 Dgolo
,.12 KUNDU
L.I Mbonze (Rombi)


tne to
that ii
sed, bi

ZON


&.-. LU% k TV. LUIlu UUI
A.15 LUNDU (Rond


GROUP
A.2I Mbuku
A.22 KWIRI
A.23 SUBU, j-
A.24 DIALA (Wuri)


GROUP
A.31 B;BI, i- (Ediya)


GROUP
A.4I Bati (Cerga)
A.42 BO (Bojkej)
A.43 Koko
A.44 BASA (Mvela)
A.45 Siki
A.46 DGUMBA
A.47 Gbea



GROUP :
B.I NZABI, bi-
B.Iz $ebo (W. Kota)
B.I3 Tsogo, u-


rm or te name snown nere is
I some cases no prefix is given
it simply that up to the present

E A
GROUP 50
A.5I NOHU (Limba)
A.52 Naka (Puku)
A. ; Lanai


..5,4 guiiui. (k^O nuii
A.55 BEDGA
A.56 jeke (Bulu)


GROUP (
A.6I YAINDE (Eunc
A.62 BVIJL
A.63 Ntum
A.64 Maka
A.65 Zimu (Njiem)
A.66 FAD
A.67 Make


GROUP
A.7I MYENE, u-
A.71a MPUDGWE
A.71b Rungu, u- (E
A.71c GALWA (DI
A.72 DI;MA, li-
A.73 KELE, dj- (Dgo:
A.74 KUTA, j- (Kota


ZONE B
GROUP ]
B.i4 Cira, i- (3ango)
B.15 Punu, yi-
B.I6 LUMBU






rION OF THE BANTU LAN


GROUP 20
[BEDE, le- (Mbete, N.E. Teke)
Ibamba, le-
saya, le-


GROUP 30
UMU, i- (Dguggulu, E. Teke)
ege, i- (W. Teke)
oma, i-
AKA
1IO, i- (S.W. Teke, Lali)
1EE, esi-
7UMU, e- (Wumbu, Mbunu)



Z(
GROUP IO
UDGILI
ukoggo
,aka (Yaka, Yariga)
?undi
ande
Izeli, li- (Ndzali)
Lota


GROUP 20
ADGI-LOI
a LOi
b BUBADGI (Rebu)
c Nyny
EDGELE, ke-
'ymba
lulia
UTUMBA, &c.
a NTUMBA, lu-
b Waqgata
c Mpama
XUSEDGO
a POTO, li-
b MPESA
c MBIDZA, li-
d MAI)GALA (Lijgala)


GROUP 40
IFINU, e- (Funika, Mfunulg;
OMA, e- (Buma)
IENE, ke- (Tende)
AKATA, ki- (Lesa, Tete)
ANZI, ki-
)goli (Dgulu)
)iiga (Dzir)
IBUNU, gi- (Mbunda)










GROUP 20
e BULUKI
f Kaggana
g LIKU, i-
3VJA, i-


GROUP 30
3GOMBE, li-
luela (Lirgi)
lati (Beige)


GROUP 40
3UA, li- (Bali, Barggo)
IJBA, li- (Dgelima, Beo, Tui
1)


GROUP 50
VIbesa
50, hi- (Soko)
'UKI, tu- (Topoke)
LUMBU, tu- (Turumbu)
CILI, i- (Lokele)
'oma, 1i-







THE CLASSIFICATION OF


E.7i POKOMO, ki- (Pfokomo)
E.72 NIKA
E.72a GIRYAMA, ki-
E.72b KAUMA, ki-
E.72c CONYI, ki-
E.72d DURUMA, ki-


GROUP IO
F.II TODGWE, ki-
F.z2 Bende


GROUP 20
F.2I SUKUMA, ki-
F.22 NYAMWES1, ki-
F.22a NYANYEMBE, ki-
F.22b Takama
F.22c Kiya
F.22d Mweri



GROUP IO
G.II GOGO, ci-
G.z2 KAGULU, ci- (N. Sagara)


GROUP 70
E.72e RABAI, ki-
E.73 DIGO, ki-
E.74 TAITA
E.74a DABIDA, ki-
E.74b SAGALA, ki-


ZONE F
GROUP 20
F.23 SUMBWA, ki-
F.24 KIMBU, ki-
F.25 BUDGU, iki-


F.31
F.32
F.33
F.34


GROUP 30
NILAMBA, iki- (Ilamba)
RIMI, ki- (Nyaturu)
LADGI, ki- (Irargi)
Mbugwe


ZONE G


G.37
G.38
G.39


GROUP 30
KUTU, ki-
VIDUNDA, ci-
SAGALA, ki-


GROUP 20
G.2I TUBETA, ki- (Taveta)
G.22 ASU, ci- (Pare)
G.23 3AMBAA, ki- (Sambara)
G.24 BONDEI, ki-




GROUP 30
G.3I ZIGULA, ki-
G.32 DHWELE, ki-
G.33 ZARAMO, ki- (Dzalamo)
G.34 DGULU, ki-
G.35 RUGURU, iki-
G.36 Kami, ki-


GROUP 40
G.41 Tikulu, &c.
G.41a Tikulu, ki-
G.4Ib Mbalazi, ki-
G.42 SWAHILI, ki-
G.42a AMU, ki-
G.42b MVITA, ki-
G.42c MRIMA, ki-
G.42d UNGUJA, ki-
G.43 PEMBA, &c.
G.43a PHEMBA, ki-
G.43b TUMBATU, ki-
G.43c HADIMU, ki-
G.44 KOMORO
G.44a DGAZIJA, ki-
G.44b Njuani, ki-









GROUP IO
E. i NYORO, oru- (Guggu,
E.I2 TORO, oru-
E.I3 NYADKOLE, olu- (Hir
E.I4 CIGA, olu-
E.I5 GANDA, &c.
E.i5a GANDA, olu-
E.I5b Sese, olu-
E.I6 SOGA, olu-
E.I7 GWERE, olu-
E.I8 NYALA, olu-


GROUP 20
E.z2 NYAMBO, eki- (Karagv
E.22 HAYA, eki-
E.22a ZIBA, eki-
E.22b Hamba
E.2zc Haggiro
E.22d Nyakisaka
E.22e Yoza
E.22f Endal)gabo
E.22g Bumbira
E.22h Mwani
E.23 DZINDZA, eci- (Jinja)
E.24 KEREBE, eki-
E.25 JITA, eci- (Kwaya)


GROUP 30
E.3x MASABA
E.3Ia GISU, lu-
E.3Ib KISU, ulu-
E.3Ic BUKUSU, ulu-
E.32 HADGA, olu- (Luhya)
E.32a WADGA, olu-
E.32b Tsotso
E.33 NYORE, olu-
E.34 SAAMIA, olu-
E.35 NYULI, olu-


)NE E
GROUP 40
E.4i RAGOLI, ulu-
E.42 GIVSII, iki- (Kisii)
E.43 KURIA, iki-
E.44 ZANAKI, &c.
E.44a ZANAKI, iki-
E.44b ISENYI, iki-
E.44c Ndali
E.44d Siora
E.44e Sweta
E.44f Kiroba
E.44g lkizy
E.44h Giraggo
E.44k Simbiti
E.45 NATA, iki- (Ikoma)
E.46 Sonjo (Sonyo)





GROUP 50
E.5i KIKUYU (Gikuyu)
E.52 EMBU, ki-
E.53 MERU, ki-
E.54 SARAKA
E.55 KAMBA, ki-
E.56 1AISO, ki-




GROUP 60
E.6I RWO, ki- (Meru)
E.62 CAGA
E.62a HAI, ki- (Mopi, M
E.62b WUNJO, ki- (Mar:
E.62c ROMBO, ki-
E.63 Rupa
E.64 KAHE, ki-
E.65 GWENO, ki-


I'MV, BAIN"IU LANUiUAUE'l






THE CLASSIFICATION OF


E.71 POKOMO, ki- (Pfokomo)
E.72 NIKA
E.72a GIRYAMA, ki-
E.72b KAUMA, ki-
E.72c CONYI, ki-
E.72d DURUMA, ki-


GROUP 10
F.I TODGWE, ki-
F.z2 Bende


GROUP 20
F.2I SUKUMA, ki-
F.22 NYAMWESI, ki-
F.22a NYANYEMBE, ki-
F.22b Takama
F.22c Kiya
F.22d Mweri



GROUP IO
G.II GOGO, ci-
G.I2 KAGULU, ci- (N. Sagara)



GROUP 20
G.z2 TUBETA, ki- (Taveta)
G.22 ASU, ci- (Pare)
G.23 AMBAA, ki- (Sambara)
G.24 BONDEI, ki-




GROUP 30
G.3i ZIGULA, ki-
G.32 DHWELE, ki-
G.33 ZARAMO, ki- (Dzalamo)
G.34 DGULU, ki-
G.35 RUGURU, iki-
G.36 Kami, ki-


JP 70
E.72e RABAI, ki-
E.73 DIGO, ki-
E.74 TAITA
E.74a DABIDA, ki-
E.74b SAGALA, ki-


EF
GROUP 20
F.23 SUMBWA, ki-
F.24 KIMBU, ki-
F.25 BUDGU, iki-


GROUP 30
F.3I NILAMBA, iki- (Ilamba
F.32 RIMI, ki- (Nyaturu)
F.33 LADGI, ki- (Irargi)
F.34 Mbugwe



SG
GROUP 30
G.37 KUTU, ki-
G.38 VIDUNDA, ci-
G.39 SAGALA, ki-

GROUP 40
G.41 Tikulu, &c.
G.41a Tikulu, ki-
G.4ib Mbalazi, ki-
G.42 SWAHILI, ki-
G.42a AMU, ki-
G.42b MVITA, ki-
G.42c MRIMA, ki-
G.4zd UNGUJA, ki-
G.43 PEMBA, &c.
G.43a PHEMBA, ki-
G.43b TUMBATU, ki-
G.43c HADIMU, ki-
G.44 KOMORO
G.44a DGAZIJA, ki-
G.44b Njuani, ki-








GRC
G.51 POGOLO, (
G.52 Ndamba


THE BANTU LANGUAGES
GROUP 60
G.6I SADGO, eqi-
G.6z HEHE, eki-
G.63 BENA, eki-
G.64 PADGWA, eki-











L.34 I
L.35 S



L.4I I






I"tE I BAINI U LA NiUAli-ht


GROUP 50
M.5 BIISA, ici- (Wisa)
M.52 LALA, ici-
M.53 SWAKA, ici-
M.54 LAMBA, ici-
M.55 Seba





GROUP IO
N.x i MANDA, ci-
N.I2 DGONI, ci-
N.13 MATEDGO, ci-
N.x4 MPOTO, ci-
N.15 TODGA, ci- (Siska)


GROUP 20
N.2I TUMBUKA, &c.
N.2za TUMBUKA, ci-
N.2Ib POKA, ci-
N.2ic KAMADGA, ci- (]
N.2id Segga
N.2ie Yombe
N.zIf Furjgwe
N.2Ig Wenya
N.z2h Lambia
N.2zk Wandia



GROUP IO
P.Ix NDEDGEREKO, ki-
P.I2 RUIHI, ki- (Rufiji)
P.x3 MATI}MBI, ki-
P.14 DGINDO, ki-
P.I5 MBUDGA


GROUP 20
P.2z YAO, ci-
P.22 MWERA, ci-


GROUP 00
M.6I LENJE, ci- (Ciina Muku
M.62 SOLI, ci-
M.63 ILA, ci- (Sukulumbwe)
M.64 TODGA, &c.
M.64a TODGA, ci-
M.64b Toka
M.64c Leya


EN
GROUP 30
N.3x NYANJA, &c.
N.3ia NYANJA, ci-
N.3xb CEWA, ci- (Peta)
N.3Ic MADANJA, ci-
N.32 Mbo
N.33 Mazaro


GROUP 40
N.4I NSEDGA, ci-
N.42 KUNDA, ci-
N.43 NYUDGWE, ci- (Tete)
N.44 SENA, ci-
N.45 Rue, ci-
N.46 Podzo, ci-





EP
GROUP 20
P.23 MAKONDE, ci-
P.24 NDONDE, ci-
P.25 MABIHA, ci- (Mavia)


GROUP 30
P.31 MAKUA, i-
P.32 LOMWE, i-
P.33 DGULU, i-
P.34 Cuabo, ci- (Cuambo)






THE CLASSIFICATION OF THE BANTU LANGUAGES
ZONE R


GROUP IO
R.II MBUNDU, u- (Nano)
R.12 Ndombe
R.I3 Nyaneka


GROUP 20
R.21 KUANYAMA, oci- (Humba)
R.22 NDODGA, oci- (Ambo)


GROUP 30
R.3I HERERO, &c.
R.3xa HERERO, oci-
R.3Ib Mbandieru
R.3ic Cimba


GROUP 40
R.41 YEEI (Yeye)


ZONE S


GROUP IO
S.II VENPA, ci-


GROUP 20
S.21 TSWANA, si- (Cwana)
S.z2a ROLOD, si-
S.21b Kgatla, si-
S.2ic Maggwato, si-
S.22 PIDI, si- (Pedi)
S.23 SUTHU, si-


GROUP 30
S.3x XHOSA, isi- (Xosa)
S.32 ZULU, &c.
S.32a ZULU, isi-
S.32b DGONI, isi-
S.33 SWAZI, &c.
S.34 NDEBELE, isi- (Tebele)


GROUP IO
T.II KOREKORE, &c.
T.iia 3ajgwe
T.ixb KOREKORE, ci-
T.Iic Tabara, ci-
T.IId Budya
T.z2 ZEZURU, ci-
T.I3 MANYIKA, &c.
T.i3a MANYIKA, ci-
T.i3b Tebe, ci-
T.i4 NDAU, ci- (Sofala)
T.15 KARADGA, ci-
T.I6 KALADA, ci-


ZONE T
GROUP 20
T.2I TSWA, &c.
T.2Ia Hleigwe, pi-
T.z2b TSWA, pi-
T.2z GWAMBA
T.23 THODGA
T.z3a HLADGANU ($aigaan)
T.23b Tsoqga
T.23c Jogga
T.23d Bila
T.24 RODGA, si-


GROUP 30
T.31 COPI, .i- (LeDge)
T.32 TODGA, gi-









INDEX TC
e following alphabetical inde:
iges has been indicated by re
is a list of the abbreviations u!
Angola
Bechuanaland
Cameroons
1. Belgian Congo
.French Middle Congo
Gaboon
Kenya
Nyasaland
L. Northern Rhodesia
(C.B., U.) D.2z
(S.W.) R.22
(K.) G.42a
(C.B.) C.42
sija (Comoro Is.) G.44a
(T.T.) E.63
T.T.) G.2z
(N.R.) M.42

a (C.B.) H.41
azi (Somaliland) G.4Ib
A., N.R.) K.24
C.B.) D.z2
nba (C.F.) B.22
nba (A.) H.2z
idieru (S.W.) R.3Ib
Igala (A.) H.35a
i (C.B.) C.21b
S(C.B.) D.z2
ibangu (C.B.) D.27
(C.) A.44
C.) A.4I
C.B.) C.33
C.) A.47
le (C.F.) B.z2
)a (C.B., N.R.) M.42a
)e (C.B.) D.54
>e (A.) H.13
(T.T.) G.64
e (T.T.) F.z2


NGUAGES

mate geographical locati4
e territory where they ar

Portuguese East Africa
Rio Muni (Spanish Guil
Ruanda-Urundi
Union of South Africa
Southern Rhodesia
South-West Africa
Tanganyika Territory
Uganda Protectorate
Zanzibar Protectorate
i (R.M.) A.55
: (C.B.) C.33, C.41
C.B.) C.4z
a (C.B.) C.5I
e (C.F.) B.2z
P.E.) T.23d
l.B.) D.24
(C.B.) L.22
:C.B.) D.32
(N.R.) M.51
See also Bu-
!.) A.42
(P.E.) N.32
,a (Cabinda, C.B.) H.I5
,o (C.) A.2I
e (C.B.) D.Ix
(A.) H.25
L (C.F.) B.33
i (C.B.) B.42
ei (T.T.) G.24
jge (C.) A.i3
er) (C.) A.42
ve (N.R.) K.32
'C.B.) C.41
S.E. (C.B.) D.2I
ba (C.B.) D.33
rgi (C.B.) C.zIb
(Fernando Po) A.3i
a (S.R.) T.IId
Iza (C.B.) C.26c






THE CLASSIFICATIC


Mbugwe (T.T.) F.34
Byja (C.B.) C.27
Bukala (C.B.) C.6if
BukoDgo (C.F.) C.I2
Mbuku (C.) A.21
Bukusu (K.) E.3ic
Bulia (C.B.) C.24
Bulu (G.) A.56
Byly (C.) A.62
Buluki (C.B.) C.26e
Buma (C.B.) B.42
Bumbira (T.T.) E.22g
Mbunda (A., N.R.) K.I5
Mbunda, gi- (C.B.) B.48
Mbundu, ki- (A.) H.2I
Mbundu, u- (A.) R.i i
Mbugga (T.T.) P.I5
Buggili (C.F.) C.I
Buigu (T.T.) F.25
Mbunu (C.B.) B.48
Mbunu (C.B.) B.37
Buru (C.B.) C.42
BupoIo (C.B.) L.24
Buuli (C.B.) C.6Ie
Buyi (C.B.) D.55
Bvanuma (C.B.) D.33
Bwari (C.B.) D.56
Bwela (C.B.) C.32
Mbwela (A.) K.I7
Bwende (C.B.) H.i6d
Mbwera (N.R.) L.6I

Caga (T.T.) E.62
Cazi (A., N.R.) K.I3
Cerga (C.) A.4I
Cewa (N., N.R.) N.3xb
Ch- See C-
Ciga (U.) E.I4
Cimba (S.W.) R.3ic
Ciokwe, Cioko (A., C.B.) K
Cira (G.) B.I4
Cokwe (A., C.B.) K.II
Conyi (K.) E.72c
Copi (P.E.) T.3
Cuabo, Cuambo (P.E.) P.34
Cwana (B., S.A., S.R.) S.z2


Dabida (K.). E.74a
pI)aso (T.T.) E.56
Ndali (T.T.) E.44c
Ndali (T.T.) M.2z
Ndamba (T.T.) G.52
Ndandi (C.B.) D.42
Ndau (P.E., S.R.) T.:
Ndebele (S.R.) S.34
NdeiJgereko (T.T.) P
Deqgese (C.B.) C.8i
Digo (K., T.T.) E.73
Diia (C.B.) B.47
Ndiigi (Cabinda) H.1
Ndombe (A.) R.12
Ndonde (T.T.) P.24
Ndogga (S.W.) R.3i
Ndoqgo (A.) H.2I
Dyala (C.) A.24
Dyma, Ndumu (G.).
Dunda (T.T.) G.38
Duruma (K.) E.7zd
Dyumba (G.) A.7Ib
Dzalamo (T.T.) G.33
Ndzali (C.F.) C.I6
Dzindza (T.T.) E.23
Dzii (C.B.) B.47

Ediya (C.) A.3I
Eleku (C.B.) C.26g
Embu (K.) E.52
Ena (C.B.) D.I5
Ena (A., N.R.) K.14
Endaggabo (T.T.) F.
Enya (C.B.) D.I5
Eundu, Ewondo (C.)

Fai (G., R.M.) A.66
Fiji (T.T.) P.I2
Mfinu (C.B.) B.4i
Fiote (Cabinda, C.B.
Fipa (T.T.) M.13
Foke (C.B.) C.53
Foma (C.B.) C.56
Foto (C.B.) C.26a
Fuliro (C.B.) D.63
Fuluka (C.B.) C.73






ANTU LANGUAGES


:C.F.) B.3i
e (N.R.) N.zif
, Mfunurjga (C.B.) B.4I

C.B.) C.z6d
(G.) A.7Ic
(U.) E.i5a
L (C.B.) C.63
a (Comoro Is.) G.44a
C.) A.47
La (C.B.) C.42
(C.B.) D.15
S(K.) E.51
) (T.T.) P.i4
o (T.T.) E.44f
C.F.) C.II
la (T.T.) E.72a
J.) E.3Ia
B.) K.33
T.T.) G.II
(A.) H.24
C.B.) B.46
(C.) A.zI
i (C.B.) M.42b
)e (C.B.) C.31
S(G.) A.73
(T.T.) N.12
(N.) S.32b
B.) K.33
C.B.) D.28
(T.T.) G.34
(N., P.E.) P.33
(C.B.) B.46
ba (C.) A.46
bi (R.M.) A.54
(C.F.) C.I4
L (U.) E.I
ulu (C.F.) B.3I
T.T.) G.z2, G.35
(T.T.) G.34
Guzii (K.) E.42
ba (P.E., S.A.) T.24
S(T.T.) E.65
(U.) E.17


r.) D.66


S(Z.) G.43c
T.) E.62a
L.) H.25
(T.T.) E.22b
(C.B.) D.22
'K.) E.32
a (T.T.) D.65
) (T.T.) E.22c
C.B.) D.52
r.T.) E.22
r.T.) G.63
(C.B.) L.34
(N.) N.2IC
(S.W.) R.31a
R.U.) D.6I
U.) E.I3
C.) E.3z
nu (S.A.) T.z3a
re (P.E., S.R.) T.2Ia
>ro (C.B., T.T.) D.28
(C.B.) H.41
C.B.) D.33
L (A., S.W.) R.zi
(C.B.) D.5I
i (C.B.) H.42
(A.) H.33

T.T.) E.44g
(T.T.) E.45
3.) A.74
R.) M.63
(T.T.) F.3I
k.) K.I2
(C.B.) C.6ic
(T.T.) F.33
(T.T.) E.44b
:C.) A.23
.R.) M.26

G.) B.II
.) K.I3
(C.) A.65
r.T.) E.23
.T.) E.25
P.E.) T.23c
(Comoro Is.) G.44b






86 THE CLASSIFICATION OF
Kabwari (C.B.) D.56 Koggo, N.E. (C.B.) H.i6e
Kagulu (T.T.) G.I2 Koggo, S. (A., C.B.) H.i6g
Kahe (T.T.) E.64 Koggo, Bu- (C.F.) C.I2
Kahonde (C.B.) L.4i Koggo, tu- (C.B.) C.85
Kaka (C.F.) C.I3 Koggola (C.B.) C.72
Kakoxgo (Cabinda, C.B.) H.i6a Konzo, Konjo (C.B., U.) D.41
Kala (C.B.) C.6if Korekore (S.R.) T.iIb
Kalaga (S.R.) T.I6 Koria (K., T.T.) E.43
Kamaiga (N.) N.21b Kota (C.F.) C.I7
Kamba (K.) E.55 Kota (G.) A.74
Kami (T.T.) G.36 Kota, W. (C.F.) B.z2
Kande (G.) B.i3 Dkoya (N.R.) L.62
Dkaggala (A.) K.I8 Kuamba (C.B., U.) D.23
Kaqgana (C.B.) C.26f Kuanyama (A., S.W.) R.z2
Kanyoka (C.B.) L.32 Kuba (C.B.) C.83
Kaonde (C.B., N.R.) L.4i Kuba (B.) R.4I
Karagwe (T.T.) E.2I Dkucu (C.B.) C.73
Karaiga (S.R.) T.I5 Kukwe (T.T.) M.3I
Kauma (K.) E.72b Kumu (C.B.) D.23
Kela (C.B.) C.75 Kunda (N.R., P.E.) N.42
Kele (G.) A.73 Kundu (C.) A.I2
Kele (C.B.) C.55 Dkyndu (C.B.) C.6Ib
Dkel3 (C.) A.42 Kunyi (C.F.) H.z2
Kerebe (T.T.) E.24 Kuria (K., T.T.) E.43
Kete (C.B.) L.z2 Kusu (K.) E.3Ic
Kgatla (B.) S.zib Kysy (C.B.) C.72
Kikuyu (K.) E.5I Kuta (G.) A.74
Kili (C.B.) C.55 Kutu (T.T.) G.37
Kimbu (T.T.) F.24 Dkutu (C.B.) C.8I
KijQga (T.T.) G.64 Dkuty (C.B.) C.73
Kioko (A.) K.II KwaIgwa (N.R.) K.37
Kiroba (T.T.) E.44f Kwaya (T.T.) E.25
Kisi (T.T.) G.67 Kwese (C.B.) L.I3
Kisii (K.) E.42 Kwiri (C.) A.22
Kisu (U.) E.3Ib Kyopi (U.) E.II
Kiya (T.T.) F.22c
Koko (C.) A.43 Lala (N.R.) M.52
Dkole (U.) E.I3 Lali (C.B.) B.35
Kololo (N.R.) K.3i Lalia (C.B.) C.62
Kombe (R.M.) A.54 Lamba (N.R.) M.54
Dkomi (G.) A.7Ic Lambia (N.R.) N.zIh
Komoro (Comoro Is.) G.44 Lagge (C.B.) L.3Ib
Konde (T.T., N.) M.3I Laggi (C.) A.53
Konde (T.T.) P.23 Laqgi (T.T.) F.33
Koqgo (C.B.) H.I6f Lega (C.B.) D.25
Korggo, E. (Cabinda, C.B.) H.i6a Lega, N.W. (C.B.) D.27






THE BANTU LANGUAGES


Lega, S.W. (C.B.) D.26
Lele (C.B.) C.84
Lemba (C.B.) C.75
Lembue (C.B.) M.42e
Leijge (P.E.) T.3 i
Le3ggola (C.B.) D.Iz
Lenje (N.R.) M.6I
Lesa (C.B.) B.44
Leya (N.R.) M.64c
Liku (C.B.) C.26g
Limba (C.) A.5I
Limi (T.T.) F.32
Ligi (C.B.) C.32
Loj (C.B.) C.zIa
Loki (C.B.) C.z6e
Lolia (C.B.) C.62
Lolo (C.B.) C.6Ia
Lomotua (C.B.) M.42c
Lomwe (P.E.) P.32
Lozi (N.R.) K.2I1
Luba, E. (C.B.) L.34
Luba, N.E. (C.B.) L.z4
Luba, N. (C.B.) L.23
Luba, S. (C.B.) L.33
Luba-Lulua (C.B.) L.3I
Luba-Kasai (C.B.) L.3Ia
Luba-Katalga (C.B.) L.33
Lubale (A., N.R.) K.I4
Lucazi (A., N.R.) K.I3
Lue (C.) A.I4
Luena (A., N.R.) K.I4
Luimbi (A.) K.Iz
Lujazi (A., N.R.) K.I3
Lulua (C.B.) L.3ib
Lumbu (G.) B.I6
Lumbu (C.B.) C.54
Luna-Ijkolrgo (C.B.) L.23
Lunda (A., C.B., N.R.) L.52
Lundu (C.) A.I5
Lurjgu (N.R., T.T.) M.I4
Luwunda (C.B.) L.53
Luyana, Luyi (N.R.) K.3I

Mabiha (P.E.) P.25
Macame (T.T.) E.62a
Maka (C.) A.64


Make (G.) A.67
Makonde (T.T.) P.z3
Makua (P.E.) P.3x
Makua, W. (P.E.) P.3
Malila (T.T.) M.24
Mambwe (N.R., T.T.) M.I5
Manda (T.T.) N.II
Maianja (N.) N.31c
Maggala (C.B.) C.z6d
Maggwato (B.) S.2ic
Manyema (C.B.) C.7o, D.zo, 50
Manyika (P.E., S.R.) T.I3a
Maraggu (T.T.) E.62b
Marutjgu (C.B., N.R.) M.4I
Masaba (U.) E.3I
Mapi (N.R.) K.34
MateDgo (T.T.) N.I3
Matimbi (T.T.) P.I3
Mavia (P.E.) P.z5
Mazaro (P.E.) N.33
Mb- See B-
Meru (K.) E.53
Meru (T.T.) E.6i
Mf- See F-
Mihavani (P.E.) P.33
Mityky (C.B.) D.I3
Moiggo (C.B.) C.6Ia
Mosi (T.T.) E.62a
Mp- See P-
Mukuni (N.R.) M.6I
Mwani (T.T.) E.zzh
MwaIJga (N.R., T.T.) M.22
Mwera (T.T.) P.22
Mweri (T.T.) F.22d
Mv- See V-
Myene (G.) A.7I

Naka (C.) A.52
Nano (A.) R. i i
Nata (T.T.) E.45
Nc- See C-
Nd- See D-
Ng- See G- or D-
Nika (K.) E.72
Nilamba (T.T.) F.3I
Nj- See J-






THE CLASSIFICATION OF


Noho, Nohu (C.) A.5 I
Ns- See S-
Nt- See T-
Nujgo (A.) H.36
Nyny (C.B.) C.zic
Nwesi (C.B.) M.4zd
Nyabuigu (C.B.) D.53
Nyakisaka (T.T.) E.zzd
Nyala (U.) E.I8
Nyambo (T.T.) E.z2
Nyamwaigga (T.T.) M.22
Nyamwesi (T.T.) F.2z
Nyagga (C.B.) D.43
Nyanja (N.) N.3Ia
Nyagkole (U.) E.I3
Nyanyeka (A.) R.I3
Nyanyembe (T.T.) F.zza
Nyara (U.) E.I8
Nyari (C.B.) D.33
Nyaruanda (R.U.) D.6I
Nyaturu (T.T.) F.32
Nyejgo (A.) K.I7
Nyiha, Nyika (T.T.) M.23
Nyikyysa (T.T., N.R.) M.3
Nyoka (C.B.) L.32
Nyore (K.) E.33
Nyoro (U.) E. I
Nyuli (U.) E.35
Nyuigwe (P.E.) N.43
Nz- See Z-

Danja (N.) N.31c
Dee (C.B.) B.36
Dg- See G-
Dhwele (T.T.) G.32
Dk- See K-

Oli (C.B.) C.6Ie

Mpama (C.B.) C.25b
Pande (C.F.) C.15
Paqga (C.B.) C.6Ic
Parjgwa (T.T.) G.65
Paigwe (C.) A.66
Pare (T.T.) G.22
Pedi (S.A.) S.22


Pemba (Z.) G.43
Pende (C.B.) L.I
Perf, Pere (C.B.) D.32
Mpesa (C.B.) C.26b
Peta (N., N.R.) N.3xb
Phemba (Z.) G.43a
Pidi (S.A.) S.22
Pimbwe (T.T.) M.ii
Pindi (C.B.) L.I3
Pindi, Pinji(C.B.) L. i
Pinji (G.) B.I3
Podzo (P.E.) N.46
Pogolo (T.T.) G.51
Poka (N.) N.2ib
Poke (C.B.) C.53
Pokomo (K.) E.7I
Ponda (A.) K.I3
Mpoirgwe (G.) A.7Ia
Poto (C.B.) C.26a
Mpoto (T.T.) N.I4
Mpovi (G.) B.I3
Puki (C.B.) C.53
Puku (R.M.) A.52
Mpukusu (A.) K.33
Mpuigwe (G.) A.7Ia
Punu (G.) B.15
Putsu (P.E.) T.26

R- See also L-
Rabai (T.T.) E.72e
Ragoli (K.) E.4I
Rebu (C.B.) C.2ib
Rega (C.B.) D.25
Mrima (T.T.) G.42c
Rimi (T.T.) F.32
Rolog (B.) S.2Ia
Rombi (C.) A.I3
Rombo (T.T.) E.62c
Rondo (C.) A.15
Rojga (P.E., S.A.) T.24
Rori (T.T.) G.6I
Rotse (N.R.) K.3I
Ruanda (R.U.) D.6I
Rue (P.E., S.R.) N.45
Rufiji (T.T.) P.z2
Ruguru (T.T.) G.35






THE BANTU LANGUAGES


Rujhi (T.T.) P.12
Rumbu (C.B.) C.54
Rundi (R.U.) D.62
Ruggu (G.) A.7ib
Ru1qgu (T.T., N.R.) M.i4
Ru1gu (C.B., N.R.) M.4I
Ruggwa (T.T.) M.I2
Rujsa (T.T.) E.63
Rwo (T.T.) E.6I

Saamia (U.) E.34
Safwa (T.T.) M.25
Sagala (K.) E.74b
Sagala (T.T.) G.39
Sagara, N. (T.T.) G.iz
Sakata (C.B.) B.44
Sake (G.) A.74
Salampasu (C.B.) L.5I
Sama (A.) H.23
Samba (C.B.) L.I2
Sambaa, Sambara (T.T.) G.23
Sanga (C.B.) L.35
palgaan (P.E., S.A.) T.23a
Saqgo (T.T.) G.6I
Saggo (G.) B.I4
Saiqgwe (S.R.) T.iia
Sanjo (N.R.) K.36
Saraka (K.) E.54
Sapi (T.T.) F.2I
Seba (N.R.) M.55
Sebo (G.) B.I2
Seke (G.) A.56
Sena (P.E.) N.44
Serga (N.R.) N.2id
Nseqga (N.R.) N.4I
Seggele (C.B.) C.22
SeIggeju (T.T.) E.56
Sergo (C.B.) C.26
Sese (U.) E.i5b
Sese (C.B.) D.32
Sh- is written 3- and indexed as S-
Siki (C.) A.45
Sila (C.B., N.R.) M.4ib
Uilele(C.B.) C.84
Simaa (N.R.) K.35
Simbiti (T.T.) E.44k


Sinji (A.) H.36
Sjora (T.T.) E.44d
Sira (G.) B.I4
Siska (N.) N.I5
So (C.B.) C.52
Sofala (S.R., P.E.) T.I4
Soga (U.) E.i6
Sokili (T.T.) M.3i
Soko (C.B.) C.52
Soli (N.R.) M.62
Sona (S.R.) T.II-I5
Sorjge (C.B.) L.24
Sorgo (A.) H.26
Soqgola (C.B.) D.24
SorJgomeno (C.B.) C.82
Sonjo, Sonyo (T.T.) E.46
Sogo (C.B.) C.83
Su (C.B.) D.42
Subi (T.T.) D.64
Subia (B.) K.42
Syby (C.) A.23
Suku (C.B.) H.32
Sukulumbwe (N.R.) M.63
Sukuma (T.T.) F.2I
Sumburu (C.B.) D.32
Sumbwa (T.T.) F.23
Sundi (C.B.) H.I6c
Suggu (C.B.) C.7I
Suthu (S.A.) S.23
Sutu (T.T.) N.I3
Swahili (K., T.T., Z.) G.42
Swaka (N.R.) M.53
Swazi (Swaziland) S.33a
Sweta (T.T.) E.44e

Tabara (S.R.) T.iie
Taabwa (C.B., N.R.) M.4ia
Taita (K.) E.74
Taveta (T.T.) G.2I
Tebe (P.E.) T.i3b
Tebele (S.R.) S.34
Tege (C.F.) B.32
Teke, E. (C.F.) B.3i
Teke, N. (C.F.) B.23
Teke, N.E. (C.F.) B.2i
Teke, S. (C.B.) B.37






THE CLASSIFICATION OF


Teke, S.W. (C.F., C.B.) B.35
Teke, W. (C.F.) B.32
Tembo (A.) H.34
Tembo (N.R.) M.27
Tende (C.B.) B.43
Teqgo (T.T.) N.13
Tete (C.B.) B.44
Tete (P.E.) N.43
Tetela (C.B.) C.71
Th- (for 0-) see $-
Thoiga (P.E.) T.23
Tiene (C.B.) B.43
Tikulu (K.) G.41
Tio (C.F., C.B.) B.35
Tity (C.B.) C.6Id
Toka (N.R.) M.64b
Tomba (C.B.) C.25
Toiga (N.R., S.R.) M.64a
Tojga (N.) N.15
Togga (P.E.) T.32
Toggwe (T.T.) F.II
Topoke (C.B.) C.53
Toro (U.) E.Iz
Totela (N.R.) K.4I
Tsaya (C.F.) B.z3
Tsh- See C-
Tsogo (G.) B.13
Tsojga (P.E., S.A.) T.23b
Tsotso (K.) E.32b
Tswa (P.E.) T.2Ib
Tswana (B., S.A., S.R.) S.2i
Tubeta (T.T.) G.21
TukoIgo (C.B.) C.85
Tuku (C.B.) D.I3
Ntum (C.) A.63
Tymba (C.B.) C.23
Ntumba (C.B.) C.25
Tumbatu (Z.) G.44
Tumbi (T.T.) P.I3
Tumbuka (N.) N.2Ia
Tuqgu (C.B.) C.42
Turu (T.T.) F.32
Turumbu (C.B.) C.54
Tusi (R.U., T.T.) D.6I, 66


Ulj (C.B.) C.61e


Unda (C.B.) L.53
Unguja (Z.) G.42d

Mvela (C.) A.44
Venta (S.A., S.R.) S.II
Vili (C.F.) H.xI
Vinza (T.T.) D.57
Mvita (K.) G.4zb

Wanda, Wandia (T.T.) M.z2
Wandia (N.R.) N.zIk
Wagga (K.) E.32a
Walgata (C.B.) C.25b
Wanji (T.T.) G.66
Wemba (N.R.) M.42
Wenya (N.R.) N.2Ig
Wisa (N.R.) M.5I
Woggo (C.B.) C.85
Wumu, Wumbu (C.B.) B.37
Wuggu (T.T.) F.25
Wunjo (T.T.) E.62b
Wuri (C.) A.24

Xhosa, Xosa (S.A.) S.3I

Yailima (C.B.) C.6Ig
Yaka (G.) B.34
Yaka (C.B.) H.3i
Yaka (C.F.) C.13
Yarga (C.F.) C.I3
Yanzi (C.B.) B.45
Yanzi (C.B.) C.2I, B.z6
Yao (T.T., P.E., N.) P.21
Yaynde (C.) A.6I
Yeei (B.) R.4I
Yela (C.B.) C.74
Yembe (C.B.) L.24
Yeye (B.) R.4I
Yombe (C.B.) H.I6b
Yombe (N.R.) N.2ie
Yorgo (A.) H.35b
Yoza (T.T.) E.2ze

Nzabi (G.) B.ii
Zanaki (T.T.) E.44a
Zaramo (T.T.) G.33








Nzeli (C.F.) C.I6
Zezuru (S.R.) T.I2
Ziba (T.T.) E.22a
Zimba (C.B.) D.z6
Zimba (C.B.) D.15


THE BANTU LANGUAGES
Zimu (C.) A.65
Zinza (T.T.) E.23
Zombo (A.) H.I6h
Nzuani (Comoro Is.) G.44b
Zulu (S.A.) S.32
















PRINTED IN
GREAT BRITAIN
AT THE
UNIVERSITY PRESS
OXFORD
BY
CHARLES BATEY
PRINTER
TO THE
UNIVERSITY







University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs