African language studies

LOAA000112_00008.pdf

AFRICAN

LANGUAGE

STUDIES

XI

1970

IN HONOUR OF MALCOLM GUTHRIE

SCHOOL OF ORIENTAL

AND AFRICAN STUDIES

UNIVERSITY OF LONDON

Sole Agents: Luzac & Company Ltd.,
46 Great Russell Street, London, W.C.i.




COLLECTED PAPERS IN ORIENTAL AND AFRICAN STUDIES

AFRICAN LANGUAGE STUDIES XI






Malcolm Guthrie

Elliott & Fry



AFRICAN

LANGUAGE
STUDIES

XI

1970

PRESENTED TO MALCOLM GUTHRIE

Professor of Bantu Languages in the University of London
by his colleagues and friends
on the occasion of his retirement

Edited by

Guy Atkins

SCHOOL OF ORIENTAL
AND AFRICAN STUDIES
UNIVERSITY OF LONDON

Sole Agents: Luzac & Company Ltd.,

46 Great Russell Street, London, W.C.i.


© School of Oriental and African Studies
University of London


CONTENTS

Edward Ullendorff Prefatory note .... PAGE 1
Guy Atkins Writings of Malcolm Guthrie 2
Pierre Alexandre Pre-initial elements in Bulu (A.74) nominals 5
Antony Allott Language and property : a universal vocabulary for the
analysis and description of proprietary relationships 12
B. W. Andrzejewski The roobdoon of Sheikh Aqib Abdullahi Jama : a
Somali prayer for rain 21
D. W. Arnott 1st and 2nd person pronominal forms in Fula . 35
Patrick R. Bennett The problem of class in Kikuyu 48
Jack Berry A note on Krio tones .... 60
J. Bynon A class of phonaesthetic words in Berber . 64
Jack Carnochan Categories of the verbal piece in Bachama. 81
Hazel Carter Consonant reinforcement and Kongo morphology 113
David Dalby Reflections on the classification of African languages with special reference to the work of Sigismund
Wilhelm Koelle and Malcolm Guthrie . 147
G. Fortune Some speech styles in Shona .... 172
C. G. B. Gidley Maiwutsiya : the comet myth among the Hausa 183
Lyndon Harries The phrasal predicate in Swahili .... 191
Mervyn Hiskett Mamman Konni : an eccentric poet and holy man
from Bodinga ....... 211
Jan Knappert The origin of the term ‘ Bantu ’ . . . . 230
W. Michael Mann Guthrie’s linguistic terminology and its application to
Bemba ........ 237
Joan Maw Some problems in Swahili clause structure. 257
A. E. Meeussen Tone typologies for West African languages 266
F. W. Parsons Is Hausa really a Chadic language ? Some problems of
comparative phonology ..... 272
C. Rowlands Ideophones in Yoruba ... 289
David Rycroft The national anthem of Swaziland 298
N. V. Smith Repetition of the verb in Nupe .... 319
John Stewart Tongue root position in the Volta-Comoe languages and its significance for the reconstruction of the original
Bantu vowel sounds ...... 340
A. N. Tucker and M. A. Bryan Tonal classification of nouns in Ngazija 351
E. O. J. Westphal Analysing, describing, and teaching Bantu languages . 383
W. H. Whiteley Notes on the syntax of the passive in Swahili 391
W. A. A. Wilson External tonal sandhi in Dagbani 405
F. D. D. Winston Some Bantu-like features of Efik structure. 417




Dear Professor Guthrie,

I hope you will feel that this comprehensive collection of essays, written in
your honour by your colleagues, disciples, and friends, represents a not unworthy
reflection of the teaching, example, and inspiration which you yourself have
evoked in your fellow-workers. In asking me, the least competent of your
colleagues, to write this prefatory note, the editor of this volume was no doubt
mindful of the combination in me of time-worn obsolescence and distance from
Bantu studies.

From this vantage point I have for many years been able to appreciate and
to admire the scholarly rigour and high measure of originality—apparent even
to a student of a different language area—of such works as your classification of
the Bantu languages and your Bantu sentence structure. You are generally
acknowledged as the foremost scholar in the field of African languages in this
country. Your magnum opus on Comparative Bantu, though only published in
part so far, has already enjoyed a world-wide response and has led to your
election as a Fellow of the British Academy, the first one in the sphere of African
languages. In this way your role as pioneer of subject and method has received
national and international recognition.

From a literary and bibliographical point of view the genre of Festschriften
has not been universally fortunate, but the astonishing unity of treatment and
substance, coupled with a most catholic coverage of area, apparent in the present
volume, must surely be seen as a tribute to yourself. Those of us who have
contributed to this volume (and not least the colleague who accepted the
responsibility of editorship) have always been mindful and appreciative of your
qualities as scholar and teacher, of the impeccable judgement you have brought
to bear on your professional pronouncements, and the burdens of administration
and organization which you have carried for so long to the abiding advantage of
those who were members of your department.

In presenting this volume to you we had the threefold aim of acknowledging
the eminence of your scholarship, of offering a token of our friendship and
gratitude, and of wishing you and our subject a long continuance of your fruitful
activity as leader and inspirer in the field of African studies.

Yours sincerely,

Edward Ullendorff


WRITINGS OF MALCOLM GUTHRIE

(Books, articles, translations)

1 Lingala grammar and dictionary. Conseil Protestant du Congo,
Leopoldville, 1935.

2 Grammaire et dictionnaire de lingala, avec un manuel de conversation
fran^ais-lingala. Heffer, Cambridge, 1939. 2nd ed. La Librairie Lvangelique au
Congo, Leopoldville, 1951. Reprinted by Gregg Press, Farnborough, Hants and
Ridgewood, N.J., 1966.

The dictionary and phrasebook were also published as a separate item in
1939 and 1951.

3 ‘ Music in African worship ’, Congo Mission News, 105, 1939, 23-4.

4 ‘ Tone ranges in a two-tone language (Lingala) ’, BSOAS, X, 2, 1940,
469-78.

5 Translation of the New Testament into Lingala : Mokanda na Kondimana
na Sika na Nkolo na biso Yesu Masiya. La Societe Biblique Britannique et
Etrangere, London, 1942.

6 ‘ The lingua franca of the Middle Congo ’, Africa, XIV, 3, 1943, 118-23.

7 ‘ East Africa’s reactions to European culture ’, JRSA, XCIII, 4698, 1945,
488-97.

8 The tonal structure of Bemba. Thesis presented for the degree of Ph.D.,
University of London, 1945 (typescript).

9 ‘ The voice of Africa ’, a Bemba poem by I. Braim Nkonde [English trans-
lation by Malcolm Guthrie], Africa, XVII, 4, 1947, 275-6.

10 Bantu word division : a new study of an old problem. OUP for International
African Institute, Memorandum XXII, London, 1948.

11 The classification of the Bantu languages. OUP for International African
Institute, London, 1948.

2nd ed. Dawsons for International African Institute, London, 1967.

12 ‘ Gender, number, and person in Bantu languages ’, BSOAS, XII, 3/4,
1948, 847-56.

13 Universal decimal classification . . . reclassification of the Bantu languages
[incorporating Dr. Malcolm Guthrie’s classification]. British Standards Institu-
tion, London, 1951 (duplicated).

14 The Bantu languages of western equatorial Africa. (Handbook of African
languages.) OUP for International African Institute, London, 1953.

15 ‘Some features of the Mfinu verbal system’, BSOAS, XVIII, 1, 1956,
84-102.


WRITINGS OF MALCOLM GUTHRIE 3

16 ‘ Observations on nominal classes in Bantu languages BSOAS, XVIII,
3, 1956, 545-55.

17 Linguistic survey of the northern Bantu borderland. Vol. I [supervisors
Malcolm Guthrie and A. N. Tucker]. OUP for International African Institute,
London, 1956.

18 ‘ La classification des langues bantus : approche synchronique, methodes
et resultats ’, Travaux de PInstitut de Linguistique de /’ Universite de Paris, IV,
1959, 73-81.

19 ‘ Problemes de genetique linguistique : la question du Bantu Commun ’,
Travaux de VInstitut de Linguistique de /’ Universite de Paris, IV, 1959, 83-92.

20 ‘ Teke radical structure and Common Bantu ALS, I, 1960, 1-15.

21 Bantu sentence structure. SOAS (agents : Luzac), London, 1961.

22 ‘ Bantu origins : a tentative new hypothesis JAL, I, 1, 1962, 9-21.

23 ‘ The status of radical extensions in Bantu languages ’, JAL, I, 3, 1962,
202-20.

24 ‘ Some developments in the prehistory of the Bantu languages ’, JAH,
III, 2, 1962, 273-82.

25 ‘ A two-stage method of comparative Bantu study ’, ALS, III, 1962,
1-24.

26 ‘ African languages ’, in Trends in modern linguistics. Ninth International
Congress of Linguists, Cambridge, Mass., 27 Aug.-l Sept. 1962. Ed. Christine
Mohrmann, F. Norman and Alf Sommerfelt, Spectrum, Utrecht, 1963, 59-61.

27 ‘ Bantu languages in the “ Polyglotta Africana ” ’, Sierra Leone Language
Review, III, 1964, 59-64.

28 ‘ Some uses of arithmetical computation in comparative Bantu studies ’,
TPS 1964, 1965, 108-28.

29 ‘ Comparative Bantu : a preview ’, JAL, IV, 1, 1965, 40-5.

30 ‘ Language classification and African studies African Affairs, spring
1965 (special number), 29-36.

31 Comparative Bantu : an introduction to the comparative linguistics of the
Bantu languages. Part I, Vol. 1 : The comparative linguistics of the Bantu languages.
Gregg International Publishers, Farnborough, Hants, 1967.

Part I, Vol. 2 : An outline of Bantu prehistory and an inventory of the Bantu
languages (in press).

Part II, Vols. 3 and 4 : A catalogue of Common Bantu with commentary
(in press).

32 ‘ Variations in the range of classes in the Bantu languages ’, in Colloques
internationaux du Centre National de la Recherche Scientifique. Sciences humaines.


4

WRITINGS OF MALCOLM GUTHRIE

La classification nominate dans les langues negro-africaines. Aix-en-Provence,
3-7 juillet 1967. C.N.R.S., Paris, 1967, 341-53.

33 ‘ A tale from western equatorial Africa [Fumu] ’, J. Folklore Inst.
(Bloomington, Indiana), IV, 2/3, 1967, 240-9.

34 ‘ Contributions from comparative Bantu to the study of African
prehistory ’, in Language and history in Africa : collected papers of the London
seminar on language and prehistory in Africa. Frank Cass, London (in press).

35 ‘ The western Bantu languages ’, in Current trends in linguistics, ed.
Thomas A. Sebeok, Vol. VII, Mouton, The Hague and Paris (in press).

36 ‘ Features of verbal structure in south-west Fang ’ (Memorial Volume for
Hans Wolff), IJAL, XXXV, 4, 1969, 356-65.

37 Collected papers on Bantu linguistics. Gregg International Publishers,
Farnborough, Hants (in press).

[Compiled by Guy Atkins]


PRE-INITIAL ELEMENTS IN BULU (A.74) NOMINALS

By Pierre Alexandre

‘ The demonstrative pronoun . . . takes the place of a relative pronoun in the
sense of “ the one who ” or “ the one which . It is probable that there is
a slight vowel sound preceding when used in this sense, but not sufficient to warrant
its writing, as it then becomes too strong. . . . This vowel may appear with nouns
as well showing that it is not a real part of the word.’ 1

Neither is it a real vowel ! In fact the pre-initial element described by A. I.
Good in the quotation given above could well be termed a tone-peg since its
main (or sole) function seems, in fact, to be to support a high tone, which is the
true distinctive trait.

Von Hagen and Bates both failed to notice this prefixal adjunct, and Good
did not expand its description beyond the few lines I quote. Its Ewondo (A.72)
reflex 6 is termed ‘ article ’ by Pichon and Graffin.2 In Fang (A.75) it is rendered
by e in the Protestant spelling, but Galley 3 does not mention it in his grammar
and seems, in fact, to confuse it to some extent with his so-called ‘ preposition ’,
i.e. the locative extra-independent prefix (also termed ‘ preposition ’ by Pichon
and Graffin), which it does indeed resemble.

Shape

As has been suggested before, the shape of this element can be described as
a pre-initial high tone, with various kinds of supporting elements. The use of
the term ‘ pre-initial ’ is to be understood in relation with the ‘ quotation form ’
of independent nominals, the class prefixes of these forms always having a low
tone.

The tone-peg can be :

A short neutral vowel, before prefixes 0 and C (except when C is a nasal), e.g.

cl. 1A tate father state
cl. 2 bingd women abingd
cl. 5 jam affair a jam
cl. 5 di§ eye adis
cl. 7 jdm thing sjdm
cl. 9/10 fa cutlass afa
cl. *16 v6m place avdm

Good, 16.

Pichon and Graffin, 12-13, 94-6.

Galley, passim, esp. 586-8.


6

PIERRE ALEXANDRE

A short homophonic syllabic nasal before nasal prefixes followed by a vowel
stem

cl. 1 mot person mmot

cl. 9/10 nydo snake rinydo

However in the southern dialects this nasal is often replaced by a- as above, e.g.
mot, amot

With non-syllabic nasal prefixes, the prefix is vocalized and becomes syllabic
cl. 9/10 ndd house ddd

With all syllabic prefixes, the tone of the prefixal syllable changes from low
to high :

cl. 1 ngdl wife dgdl
cl. 2 beyal wives bdydl
cl. 3 mfdk bag mfdk
cl. 4 mimfdk bags mimfdk
cl. 5 aid night did
cl. 6 meld nights mdlu
cl. 7 eld tree did
etc.

In proverbs, drumming, songs, and other oral literature, however, there are
occurrences of a- before CV prefixes : abild, abeydl, etc.

With dependent nominals, the shape of this element is conditioned by the
grammatical function or form of the DN, rather than by its shape.

With demonstrative adjective-pronouns it is either a- (as mentioned by Good)
or b-

cl. 1 nyu anyu, rinyu
cl. 2 bd abd, mbd
cl. 3 wu awu, dwu
etc.
With per sonal substitutes
cl. 1 nys iinye
cl. 2 bd mbe
cl. 3 wo dwo
etc.

With the so-called ‘ indefinite ’ adjective-pronoun -bok it is high tone on DP
when -bdk acts as nominal substitute

(m)mota mbdk the other person

mbdk the other (cl. 1)

Other dependent nominals have no corresponding forms.


PRE-INITIAL ELEMENTS IN BULU (A.74) NOMINALS

7

Function

Pre-initial high tone regularly occurs :

When the nominal is followed by a demonstrative of shape DP-(li)/(nA)

When it is subject antecedent to a relative clause

When it is the subject of an auto-predicative sentence (most often in answer
to a question)

Thus :

bot bAngAwu

abota bA (bA16, bAnA) bdngAwu
abota bdngAwh, mbe bAna

(a)j6 ji ? afa

With dependent nominals :
Demonstratives

abA bAngAwu . . .

bA bAngAwu

people died

those (these) people died
the people who died are these

(lit. ‘ they these ’)

what (is) this ? It’s a cutlass

those who died . . .
those died

Personal substitutes

... be vewu
be bAwuya
mbe bAwuya
za Iee ? Anye

(and) they die

as for them they are dead
they are the ones who are dead
who is (there) ? It’s him

Pre-initial high tone may also occur :

In ordinal construction

kAlAta bAa or AkAlAta bAa second book

With possessive adjectives

ndA jam or AdA jam my house

Compare the construction of possessive pronouns (Aa-DP-Stem)

cl. 1 warn my Agwam mine

cl. 2 bArii my embAm mine

etc.

In the genitive construction with connective ya. In this case there seems to
be a difference between

bota ya Alam (some) people of/from the village
and abota ya Alam the people of/from the village


8

PIERRE ALEXANDRE

With adjective-pronoun -b6k. Here again there may be a nuance

bota bevok other people or certain people

abota bevik the other people

cf. b6v6k the others (cl. 2)

With object antecedent in relative construction, when object substitute is
not used after the verb

mmot mengdyen . . . the man I saw . . .

or mot meng&ydna nye . . . the man whom I saw . . .

Nature of pre-initial HT

From all that precedes it would seem that Bulu prefixes with pre-initial high
tone are quite comparable to the double prefixes encountered in so many Bantu
languages in other zones (D, E, F, etc.).

There is, nevertheless, in Bulu a strong possibility of confusion with other
pre-prefixal elements having identical or nearly identical shapes.

Extra-dependent prefixes

In genitival constructions the EDP is, in most cases, reduced to a high tone,
either in the last syllable of the nomen regens or on the first syllable of the nomen
rectum, depending on their shape

anyu edge

anyu fa edge of cutlass

anyu dceij edge of knife

The exceptions occur when the nomen regens belongs to a class with CV
prefix ; one can then use a CV extra-prefix if, and only if, the nomen rectum
itself does not have a CV prefix

bey&l b6-evdt wives of the chief

or beyal 6vet

but beydl bivet wives of chiefs

Locatives

Position in time and space is indicated in Bulu by a pre-initial element whose
shape differs but slightly from that given above

before cl. 9 independent nominals which have become lexicalized and act
as ‘ prepositions ’ or ‘ adverbs ’

si earth Asi under, below, down
yob sky dyob up, on, upon
yat far bank of river dyat beyond, across
nd& house Andd home


PRE-INITIAL ELEMENTS IN BULU (A.74) NOMINALS

9

mvus back dmvus back, behind

ngo?6 evening dngo?d yesterday

etc.

a- before names of various classes with 0 or C prefix

sDiwild in Duala
aYewondo in Yaounde
aFalisi in France
amikit at the market (makit, cl. 1A)
ajal at the village (jal, cl. 5)
asi in the country (si, cl. 9)
etc.

High tone on syllabic prefixes and probably with non-lexicalized cl. 9 with
a non-syllabic n-prefix and a consonantal stem

cl. 3 mfek bag mfek in the bag
cl. 5 afan forest afan in/to the forest
cl. 9 ndi house ndd in/at the house (cf. asi/isi)
cl. 7 eyoi) occasion 6yoi) when
cl. 6 mefub fields mifub in/to the fields
etc.
hesitate to call this locative element an extra-independent prefix.

concord system is quite irregular—that is if there is a locative concord system.4
The demonstrative paradigm DP-(li)/(nd) has quite regular reflexes of CB
classes 16 and 18

va vdna vile

mu muni mull

generally presented as ‘ adverbs but in fact absolutely identical in structure with

cl. 1 nyu nyunA nyuli
cl. 2 bd bind bite
etc.
Yet mu and vi are in most cases used by themselves in sentences such as

za?d mu come here

biseke vile they are not there

and only seldom encountered in phrases such as
aDiwdld muni right here in Duala

4 There is quite definitely one in A.72. Cf. Pichon and Graffin, 28.


10

PIERRE ALEXANDRE

Furthermore the independent nominals keep their regular concord pattern
which can, however, be encased within what looks like a locative concord
ddd nyi in this house 5

iidd nyi mu in this here house

the latter occurrence being indeed rather rare.

Finally the verbal concord, at least in modern usage, is identical with the
general impersonal concord, i.e. cl. 1 or 9, without predictability (v6m place, has
1 /6 concord)

ane mvo?6 (cl. 1), ene mvo?6 (cl. 9)
aDiwAl& ane abeij'l
aDiwal& ene abeijj

av6m mengdtoa mu ambe aberj

it’s all right, Qa va bien

in Duala it’s beautiful

it was beautiful where I stayed or the
place in which I stayed was beautiful

I have some reasons to think—but I cannot check without going back to the
field—that older people would use a locative verbal prefix a- in these sentences.

Instrumental and connective a

Instrumental relationship 6 and plain connection between two things are
marked by a pre-initial d-. This d- could not formerly be used to mark connection
between two persons, for which ba was used, but this usage tends to become
obsolescent and d- is now used as a general connective

mendim a-meyok water and/with wine

es& Mango (formerly esa b& bongd) father and children

There is, however, a difference in shape between the connective and the
instrumental : while the former cannot be elided, the latter is reduced to initial
high tone when preceding a syllabic prefix or a cl. 9 h-

but

aci?i el6 Ma
aci?i el6 ov6n
fa a-ovdii

he cuts wood with a cutlass
he cuts wood with an axe
cutlass and axe

Conclusion

The more one studies the A zone languages, the more one becomes convinced
of their orthodoxy, if I may term it so. The ‘ not quite Bantu ’ fallacy is generally
based upon the lack of elements commonly found in many or most languages of
the other zones: vowel suffixes, double prefixes, locative classes are often-quoted

5 Possibly anda nyi rid& nyi :: in this house this house.

6 The passive extension in -ban is very seldom used ; when it is used the agent is introduced
by a-. A sentence like akdl&ta nyu atilibanaya Mica ‘ this book has been written by a teacher
is ‘ theoretically ’ possible and ‘ theoretically ’ correct, and quite unlikely except in written form.


PRE-INITIAL ELEMENTS IN BULU (A.74) NOMINALS 11

examples. A deeper analysis shows that these elements are not really missing,
but rather submerged and partly coalescent. In the most drastic cases, tones
still subsist as the ghosts of vanished phonemes. Which is indeed a very significant
clue for the reconstruction of Bantu genesis.

BIBLIOGRAPHY

(Sources quoted in the article)
Bates, G. L., Handbook of Bulu, Elat, 1926.

Galley, S., Dictionnaire fang-frangais et frangais-fang suivi d'une grammaire fang, Neuchatel,
1964.

Good, A. I., Bulu handbook supplement, Elat, 1934.

Hagen, G. T. von, Lehrbuch der Bulu Sprache, Berlin, 1914.
Pichon, F., and R. Graffin, Grammaire ewondo, Paris, 1930.


LANGUAGE AND PROPERTY:

A UNIVERSAL VOCABULARY FOR THE ANALYSIS

AND DESCRIPTION OF PROPRIETARY RELATIONSHIPS

By Antony Allott

This paper is provoked by experience over the years in the investigation of
African property systems, and the difficulties to which this gives rise in the analysis
and presentation of the data.

The field may be seen as a set of interlocking systems. For the given society
the total social structure in action is one such system, which may be broken down
into a large number of overlapping sub-systems. One such sub-system is the
agglomeration of institutions which regulate the exploitation of land. The language
of the society is another system, of which the specialized legal vocabulary (if any)
is a sub-system.

Each of these systems or sub-systems has a descriptive and a normative aspect
or dimension. Institutions function; legal and other rules prescribe how they
are to function. Not infrequently there is contradiction or tension between the
way things are and the way they ought to be—laws are broken, ungrammatical
sentences are uttered, etiquette is disregarded, and so on.

I

THE PROBLEMS OF INVESTIGATING AFRICAN PROPERTY SYSTEMS
AND THEIR LANGUAGE

The problem at the level of investigation is to uncover or expose the structure
and functioning of the relevant systems. So far as African customary land law
is concerned, this involves an identification and description of the persons or
institutions involved in the exploitation, or control of exploitation, of the land;
and a specification of the range of permitted action or inaction by each interested
person, and in respect of what subject-matter.

In its turn this requires an examination of the set of verbal categories through
which the chosen society expresses or refers to these systems of exploitation ; in
other words, a study of the indigenous operational language through which
these systems function or are explained.

Lastly the research findings must be put in coherent and acceptable verbal
form for communication to the appropriate community, whether this be the
practical world of judges and lawyers in an African country or the academic
community at large. A language of analysis and presentation is therefore required.
The elaboration of such a language is one of the major preoccupations of this
paper.


LANGUAGE AND PROPERTY 13

Field investigation of African customary land law: the role of the vernacular
language

Anthropological, administrative, economic, geographical and legal investiga-
tions of African land laws have been numerous and effective and the literature
on them is large. This is not the occasion to review these investigations or
criticize their research methods; but one may propose that a useful and under-
employed ancillary technique of investigation is through the indigenous language
associated with the given land law. This can include the recording of the naive
presentations by indigenous informants of the land law system or some part of it,
and the more systematic pursuit of the vocabulary and other linguistic apparatus
by means of which the system is operated. A number of such investigations have
been made by the present author in conjunction with linguists and local in-
formants, notably into Sotho, Akan, Swahili and Yoruba.

Without falling into the error of presuming that an institution cannot exist
in the law unless there is a term to refer to it, one may state that there is generally
a correspondence between the linguistic and legal sub-systems. If there is no
term to identify an interest-holder, or to pick out a given category of interest,
then that kind of interest-holder or that category of interest may not ‘ exist ’,
i.e. is not overtly incorporated or recognized in the legal system. The external
analyst may, of course, identify a category of interest-holders (e.g. corporate
holders) not so identified by uninstructed informants, or debate the presence or
absence of some legal quality or function (e.g. ‘ ownership ’; the loss of interests
by prescription), without there being a verbal parallel in the indigenous language ;
in such an instance he is creating an analytical super-category or meta-rule.

A report on the use of Yoruba legal terminology, based on investigations
conducted by E. C. Rowlands and the present writer, in conjunction with
A. Adeyefa, illustrates some of these points. The purpose of the investigation
was to try to discover whether the Yoruba language has a specialized legal
vocabulary, and secondly how terms are used in a legal context. A specialized
legal vocabulary could imply one of three things:

(1) Terms which have an exclusively legal meaning (i.e. are used only in
legal contexts), and are thus not readily intelligible to the ordinary man
(an English example would be the term ‘ fee simple ’).

(2) Terms which have a specialized meaning, i.e. an exclusive legal function,
but which may be intelligible to the non-specialist (cf. the English word
‘ tort ’).

(3) Terms which have legal and non-legal uses or meanings (there are many
such examples in the English language : ‘ estate ‘ frustration ’ are just
two examples).

It is prima facie unlikely that African vernacular languages operating in
conjunction with unwritten customary laws will have developed an arcane legal


14

ANTONY ALLOTT

language unintelligible to the ordinary members of society, since law is an activity
typically performed in public, involving members of the public both as participants
in legal relations, as parties in disputes and as adjudicators or arbitrators. This is
true even where specialized legal personnel have emerged (as with traditional
legal or judicial personnel in some African societies). It is perhaps only in a
conquest situation that law and its administrators can be alien to the majority
of the population—and this is not the case in recent Yoruba history.

On the other hand, it is quite possible that in African traditional societies of
any degree of complexity there will have developed specifically legal social contexts
and hence the specialized use of vernacular terms in legal situations. Our enquiry
into Yoruba legal language was thus a search for specialized terms with no
non-legal uses or for terms having both legal and non-legal functions.

Investigation of a legal vocabulary is to a large extent investigation of the
law which supports it. The function of terms as defining or classifying factors
must be continually borne in mind. At its broadest and vaguest this classifying
function is seen at work in discriminating between ‘ law ’ and ‘ non-law ’, or
between a ‘ right ’ which is legal and one which is only a moral or social claim.
Investigation of abstract terms, such as the Yoruba equivalents to ‘ law ’,
‘ obligation ’, ‘ ought ’, ‘ right ’, ‘ contract ’, is part of the exercise. There is also
a search for terms with a specialized legal meaning, which may or may not consist
in the restriction or function of a word which has a much more general meaning
in ordinary parlance (an English example would be the difference between the
lawyer’s and the non-lawyer’s use of the word ‘ agreement ’). Another field for
investigation, at least in some languages, is that of legal proverbs and maxims.
Such proverbs and maxims may have an explicit and primary legal reference—
‘ the chief owns the land ’; or no explicit legal reference, but frequent citation in
legal contexts.

Ni and the vocabulary of ownership and possession 1

In many commentaries on or explanations of Yoruba land law, one meets
statements about ownership of land and other property. Thus Lloyd 2 mentions
‘ such oft-recurring and ambiguous phrases as “ the land belongs to the oba ”
Equally one reads: ‘ The paramount chief is owner of the lands, but he is not
owner in the sense in which owner is understood in this country [sc. England—
A.N.A.J. He has no fee simple, but only a usufructuary title ’.3 These statements
involve an analysis of Yoruba property and vocabulary systems, a similar analysis

1 See R. C. Abraham, Dictionary of modern Yoruba, London, 1958, 438-9; P. C. Lloyd,
Yoruba land law, London, 1962, 60 et seq.; G. Sertorio, Struttura sociale, politico, e ordinamento
fondario yoruba, Como, 1967, 287 et seq.

2 Op. cit., 43.

3 Per the Privy Council in Oshodi v. Dakolo [1930] A.C. 667, at p. 668.


LANGUAGE AND PROPERTY 15

of the English property and vocabulary systems, and an attempt to strike an
equivalence between the two.

There are two main ways of dealing with the problem of analysing such
statements: either one can start off from an English term or statement and ask
how it could be translated into Yoruba, or one can start off with the Yoruba
words and consider how they may be rendered into English. In both cases what
is actually involved goes further than the mere establishment of a one-to-one
correspondence between the meaning of a Yoruba term and the English term,
since it is necessary to understand and to be able to compare the property systems
which underlie each of the languages. So what one is doing is really to compare
property systems through the medium of language.

It is worth remarking that one is not restricted to a consideration of nouns
so-called; verbs, possessive adjectives, particles, etc., may all be brought into
operation to express a connexion between a person and an object. In law, the
relationship is more than one of mere connexion or association, since possession
is deemed to be legitimate, i.e. deriving from a title recognized by law; so one
of the things to be investigated in Yoruba is whether there are any words which
discriminate between the mere fact of possession, and possession in pursuance of
a lawful title. It was stated that Yoruba does not distinguish verbally between
questions of the form ‘ who is the owner of this car ? ’ (Ta lo ni motor yi) and ‘ who
is in charge of—e.g., the driver—this car? ’. Apparently ni signifies ‘ to be in
charge of ’, ‘ in possession of ’; that is, as with comparable terms in many African
languages, it expresses a simple association between a person and property
(abstract or concrete) or thing: one may compare the English ‘ have ’. Other
words which are relevant in the description of possession are gbe, lo and ti. Gbe
is translated as ‘ occupy ’, ‘ live in ’ or ‘ inhabit ’. Thus : Emi ni o ngbe ile kan ni
Ibadan sugbon to Mr. A ni ‘ I occupy a house at Ibadan, but it belongs to Mr. A’; and
Ibadan ni mo ngbe ‘ I live in Ibadan Lo is translated as ‘ use ’; thus ‘ I am using
this car, but it is not mine ’ Emi ni o nlo motor yi.

Ti: This is described as a nominal particle or pronoun which is emphatic
and expresses ‘ it is of . . .’ or ‘ it belongs to . . .’. Thus ‘ this cap is mine ’ Ti emi
ni fila yi. There are apparently three ni forms: ni = is, ni = have, and perhaps
another ni = own, but the last is doubtful. (The use of ni in the phrase ni are is
interesting. Are is an abstract noun meaning right or justification; ni are means
to have the right, i.e. win a case.)

One is thus in a position to look again at such a maxim as oba lo ni ile, often
translated as ‘ the Oba owns the land ’. Whatever may be implied by the state-
ment—and what is implied is the Yoruba property system—one cannot get out
of it a precise statement that the Oba is the full and beneficial ‘ owner ’, either
for his own advantage or that of his people. The concurrent claims of the land-
controlling lineages; the substantial rights of individual subjects; the limitation
in practice of the Oba’s powers to those of land control rather than benefit—all


16

ANTONY ALLOTT

these points and complexities cannot be got across by relying on the word ‘ own ’,
with or without qualifications.

II

A SCHEME OF ANALYTICAL PROPRIETARY TERMS

One is left with a feeling of dissatisfaction at the inadequacies of ordinary
English legal terminology, and its ineptness for presentation of the total scheme
of Yoruba, or any other African, land tenure system. The immediate reaction is
to turn to Hohfeld and his fundamental legal conceptions, as Lloyd (among
others) does.4 But this exercise is usually unrewarding and unilluminating so far
as African land tenure is concerned. In particular, the Hohfeldian scheme does
not sufficiently expose the hierarchical and concurrent aspects of African property
systems, or the radical distinction between control and benefit. A new scheme
seems called for.

We are talking about permitted, prescribed or forbidden modes of action in
regard to the exploitation of resources. This is sometimes put as defining the
relation between persons and things; but this is misleading, as legal relations
subsist only between persons, though in respect of things (the subject-matter of
the relationship). Classical Roman law separated the implied rights of ownership
into the jus utendi, jus fruendi and jus abutendi: all these are what I term ‘ claims
to benefit ’, i.e. to profit from resources and their exploitation. What was omitted
from the Roman scheme was the power of control, i.e. to determine who should
benefit and how from the exploitation of resources. Rather artificially this was
later brought in under abusus, which was deemed to include the power to alienate ;
but this was, it is submitted, a basic misunderstanding of the different orders or
levels of interests in property.

At the level of benefit persons are entitled by law to profit from economic
resources ; this claim to benefit includes rights of use, to the fruits, and of abuse.
Powers of control constitute a higher-order system, which specifies how claims
to benefit may be exercised. Control is the grammar of the property system, a
set of meta-norms erected over the exploitative norms connoted by benefit.

An analysis of property systems must therefore be based on the fundamental
dichotomy between benefit and control. Since control is at a higher level than
benefit, an analysis must begin with claims to benefit.

BENEFIT

As Hohfeld correctly saw, one may fragment the claim of a person to benefit
from exploitation of a thing. His atomization distinguished between claims

4 Op. cit., 60 et seq. For Hohfeldian terminology, see W. N. Hohfeld, Fundamental legal
conceptions as applied in judicial reasoning, New Haven, 1923.


LANGUAGE AND PROPERTY 17

(or rights) and privileges (or licences). I prefer to distinguish between actions
which are:

(a) permitted

(b) prescribed

(c) forbidden

An agglomeration of claims (and powers, where relevant) held by a person or
persons in respect of a resource may be termed an interest. The person entitled
to an interest is hereafter referred to as its holder.

A holder of a citizen’s interest in Ashanti is permitted to plant cocoa on land
which he has appropriated from the virgin forest; he is (or was) under a duty to
pay certain customary dues to his superior chief out of the land and its produce;
certain forms of use (e.g. commercial exploitation of mineral resources) were
forbidden to him.

These permitted, prescribed or forbidden modes of action are protected or
sanctioned by remedial claims and powers. Forfeiture of interest, action to
recover tributes due, and disallowance of unauthorized dealings, are examples
of such sanctioning claims and powers.

Total analysis of a claim to benefit in respect of a given resource requires an
answer to the questions : Who ? Over what ? How ? Subject to whom ?

Who?

A claim to benefit may be held exclusively or non-exclusively. Non-exclusive
benefits include the claim to breathe the outside air, to navigate the high seas, etc.
In African systems some common rights which appear non-exclusive, such as the
claim to hunt over certain territory, will be found on further investigation to be
exclusive as regards strangers to the community, though non-exclusive internally
in the community in which they subsist. The fact, then, that there is a large or
indefinite number of co-holders of the claim or interest does not make it non-
exclusive, provided that it is in principle possible to list those entitled.

An interest may be held by a single person, or by a number of persons con-
currently entitled. Interests held by a single person, e.g. the holder of self-acquired
property in Yoruba or Fante law, may be styled individual interests. Plural
interests fall into several sub-categories : (i) Joint interests, where two or more
individuals hold the same title, claim or interest for their separate and joint
benefit: an example is where two brothers join to acquire land in Akan law.

(ii) Corporate or community interests, where an organized group of individuals
holds a single claim or interest in the name of and for the benefit of the entity
conceived as a unit: examples from West African land law include the titles
of a political stool in Ashanti, and of a corporate lineage in Yoruba law.

(iii) Common interests, where the interest or title attaches to a community or
group, but the exercise of the interest is severally and for their individual benefit

c


18

ANTONY ALLOTT

by the members of the community or group (e.g. the right to hunt over com-
munity land).

One interest may be held concurrently with another in respect of the same res.
There are many examples from African land law. The interest of a farmer may be
separated from that of another person to the economic trees on his land;
separation of title between rights to exploit the soil and the right to a building
erected on it is common; interests of benefit are almost invariably concurrent
with interests of control: thus chiefs and family-heads often have land control
powers over land in the beneficial occupation of their subjects or dependants.

For the removal of doubts, it may be stated that the same categories of interest
can subsist in the control as in the benefit sphere.

Over what ?

Most analyses of property rights distinguish between claims to corporeal
and incorporeal (or intangible or intellectual) property. The former is physically
visible and tangible—land, movables; the latter is allegedly not—copyrights,
choses in action. A functional analysis of property rights, it is submitted, will
not support such a dichotomy. The question is what modes of action are
permitted, prescribed or forbidden ; always such action must emerge in perceptible
form. Copyright law protects an author’s claims over his intellectual productions ;
in reality, what it does is to inhibit various forms of contravening action by
those not authorized by the copyright-holder. Such action always takes place
in the physical world, whether it is impressing marks on paper or emitting sound
waves from a musical instrument.

There is always therefore a res, a subject-matter of the interest, manipulative
action in respect of which is controlled by the interest. For purposes of easy
description only, and not because of a fundamental distinction in the conceptual
sphere, the interests affecting different kinds of subject-matter may be grouped
together. English law classically distinguishes between land and other property;
some African property systems echo this distinction, though apparently distin-
guishing between rights to land and rights to things on the land (houses, trees),
the latter being assimilated to the category of movables. A number of the pastoral
societies erect a special property regime for animals, or certain types of them.
These distinctions are of great importance both in the law of enjoyment inter vivos
(e.g. women not to acquire full titles to land or cattle) and in the law of succession.

There are difficulties of classification: is adultery a species of theft ? 5 Are
wives in Yoruba succession law a kind of immovable property ? 6 When customary
law recognizes a tortious action for seduction of an unmarried daughter, is it
protecting a property right, and if so, of whom and in what ?

6 Cf. R. S. Rattray, Ashanti law and constitution (1st ed.), Oxford, 1929, 317.

6 Cf. G. B. A. Coker, Family property among the Yorubas (2nd ed.), London, 1966, 39.


LANGUAGE AND PROPERTY

19

How?

The how of an interest specifies the modes of exploitative action permitted, etc.,
to the holder. Such specification comprises limitations by scope (e.g. agricultural
exploitation only permitted; erecting house on farm-land forbidden; only
subsistence crops to be grown; etc.), duration (at pleasure of grantor; for life-
time of holder; permanently; etc.), and subjection to concurrent interests of
benefit and control (farmer not to extend his farm so as to cut across the line of
advance of his neighbour; planting to begin only when local headman so orders).

Subject to whom ?

Many of the classical debates of African law—is there ‘ ownership ’ in African
law? is the chief’s title one of paramountcy or proprietorship? what is the
nature of a family-member’s interest in family land ?—have been provoked by the
hierarchical character of African land law systems. This aspect is discussed
below under control.

CONTROL

A power of control in respect of property means that the holder of the power
has the legal capacity to prescribe, permit or forbid modes of beneficial/exploitative
action. Normally the holder of a beneficial interest has powers of control also :
he can exclude others from his land, for instance. But he is also subject to superior
control. Thus in African law an individual holder of land may get his land from
a chief or family-head by allocation. The power to allocate land, and where it
exists to control the use thereafter, are examples of land control. A landlord
may control how his tenant uses the land, a husband may control where his wife
farms, and so on.

Control, then, extends to the creation, enjoyment, transfer and termination of
beneficial interests in property. An inferior land-controlling interest may be
subjected to a superior controlling interest. Such hierarchies of control are
common in African land tenure systems (e.g. Ashanti or Tswana).

Much confusion has been caused by this hierarchical system of land control.
Obsessed by the English feudal theory that one can only grant an inferior interest
out of what one possesses himself, some analysts are forced to conclude that the
paramount rulers in such traditional systems as the Lozi, the Ashanti or Buganda
must have possessed the plenitude of proprietary rights in land, out of which
they ‘ carved ’ the subordinate interests of sub-chiefs, lineages or individual
occupiers. This confusion arises, it is suggested, from a failure to separate the
control function and the benefit function, which operate in different orders.
Elimination of the confusion is not assisted by asserting that the paramount
ruler in such instances holds the ‘ allodial ’ or ‘ absolute ’ title, with his subjects
holding ‘ inferior ’, ‘ usufructuary ’ or ‘ possessory ’ titles. The truth is that the
subject has one kind of interest, the ruler another. The subject’s interest is sub-


20

LANGUAGE AND PROPERTY

ordinated to the control function of the ruler; but nevertheless it may be absolute
within its own domain. In my own earlier writing I attempted to separate the
jurisdictional and proprietary rights of chiefs in Akan law;7 Gluckman 8 uses
the terms ‘ administration ’ and ‘ production ’ in his own most recent analysis.
It is submitted that the terms ‘ control ’ and ‘ benefit ’ more exactly express the
different orders of operation of these functions.

CONCLUSION

Two things will not escape the reader’s notice: the first is that the words
‘ own ’ and ‘ ownership ’ are redundant, and indeed misleading, for the description
of African property systems ; the second is that every property system is amenable
to the same analysis and the same terminology. English law equally recognizes
powers of land control in a variety of bodies: compulsory acquisition powers
enable public authorities to dispossess an individual beneficial holder of his land ;
building regulations and planning laws prescribe what and how he may build
on it. Nor is the concurrent or hierarchical aspect of African land laws absent
from other systems. Even the family property system, which might seem to have
no analogue in modern English law, is not entirely absent. As the legislature with
one hand destroys the old family property system, so it, and the courts, proceed
to erect an alternative in its place (with the Matrimonial Homes Act, 1967, and
the judicially evolved rules regarding presumptions about the title of a wife in
her husband’s property).

An extension of this vocabulary to English and other property laws would,
it is hoped, clarify the relations of those interested, and at the same time demon-
strate the essential connexion of those laws with the laws which prevail in Africa.
In other words, just as a universal grammar and general semantic theory have
been developed in the linguistic sphere, so can a general grammar and vocabulary
of property be devised in the sphere of property relations. Viewing as we do
property relations as doing rather than being, as action not state, we must escape
the tyranny of the nominal and conceptual, and turn our attention instead to
the analysis of the verbal, the function.

7 E.g. in my Akan law of property, University of London Ph.D. thesis, 1954, unpublished.

8 M. Gluckman, ‘ Property rights and status in African traditional law’, [in] Ideas and
procedures in African customary law, ed. M. Gluckman, London, 1969, at pp. 252 et seq. See also
J. Vanderlinden, ‘ Reflexions sur l’existence du concept de propriete immobiliere individuelle
dans les droits africains traditionnels ’, ibid., 236 et seq.


THE RO OB DOON OF SHEIKH AQIB ABDULLAHI JAMA:
A SOMALI PRAYER FOR RAIN

By B. W. Andrzejewski

The prayer for rain which is the subject of this paper may be of some interest
to both linguists and students of religion in Africa. It was composed by a prominent
Somali sheikh from the Jigjiga region, who is a champion of Somali oral literature
in all its aspects and in particular in its service to Islam. The word roobdoon means
‘ rain-seeking ’ and is applied to all prayers for rain and the accompanying rites.

Among the Somali people, who with very few exceptions are Sunni Muslims 1
special ceremonies take place in time of drought, according to the instructions laid
down in Arabic manuals of law and religious observance.2

The essence of these instructions is quite simple. The community afflicted by
drought chooses a man of religion known for his piety, who convenes a meeting for
an appointed time and leads the prayers. Everyone is to repent and carefully avoid
sinful actions or attitudes. Old, worn-out clothes have to be put on and people
march in a procession to an open place which has no shade. Old men and small
children must be present, and if possible also the animals afflicted by drought.
They are all exposed to the heat of the sun and suffer it patiently, asking God to
forgive them their sins and send them rain which would bring relief without
causing damage and devastation by flooding.

During these ceremonies Arabic is normally used, as in fact it is in most other
prayers, hymns and religious rites. The reason why Arabic is preferred to Somali
on such occasions is not far to seek. As the language of the Koran, the Tradition
and Muslim theological scholarship, Arabic enjoys enormous prestige and is
sometimes referred to in Somali as ifka Jldaha ‘ the language of God ’. Prayers and
invocations in Arabic seem to many Somali people more appropriate for worship
and more efficacious than their mother tongue.

All men of religion are expected to know Arabic well and some actually
achieve such a high standard that they compose literary works in it, especially
poetry.3 Among the lay public quite a number of people, particularly in towns,
have a good knowledge of Arabic and the new elite educated in government

1 For_ information about Islam in Somali society see Cerulli, 1957, 1959, 1964; Jama'
'Umar ‘Isa, 1965 ; and Lewis 1955, 1955-6, 1961, 1963, 1965(a) and (6), 1966. Note that all
works mentioned throughout this paper are referred to under the name of the author and the
date of publication, as given in ‘ Bibliographical References ’.

2 One of the commonly used manuals is that of Sharaf al-Din, 1933, which is particularly
favoured by Sheikh Aqib. The section referring to prayers for rain is found in pp. 99-101 of
the book.

3 For an account of Somali literature in Arabic see Cerulli, 1957, 150-1, and 187-200. His
bibliographical notes should be augmented by the following items : ‘Abd al-Rahman bin
shaykh ‘Umar, 1954, 1964; Ahmad bin Husayn bin Muhammad, 1945 ; Jama' ‘Umar ‘Isa,
1965. There is also a private collection of manuscripts in the possession of Prof. I. M. Lewis of
the London School of Economics.


22

B. W. ANDRZEJEWSKI

schools in Somalia and universities in Arab countries, usually reach high levels of
proficiency. Yet the vast majority of the population has a very limited knowledge
of Arabic, just sufficient to understand commonly used prayers and hymns, and
this only as a result of having them translated and explained orally in Somali by
men of religion during sermons and as a part of religious instruction.4 It is mainly
as a concession to such lay folk that religious poems are composed by Somali
sheikhs ; they form a compromise with linguistic reality, but what is lost in ritual
dignity is gained in the fervour which the full understanding of the words can
produce.

Although some sheikhs look upon the use of Somali in religious practices as
second best, others encourage it. Among these pioneers of Somali as the language
of worship of God and veneration of saints we find such prominent poets as
H$aji ‘Ali jeerteen,5 Sheekh Uwdys,6 Ismaa'iil Faar&h,7 Sheekh ‘Abdilte Isdaq,8
Mohammed ‘Abdilte Hasan 9 and Sheekh Gaby dw.10

In recent years one of the most ardent champions of using Somali for religious
purposes has been Sheikh Aqib Abdullahi Jama,11 the author of the poem
presented in this paper.

Sheikh Aqib was born in Gqlufa, a settlement about 5 km. east of Jigjiga, some
forty-eight years ago ; he is a member of a well-known clan called Bartire and is
related on his father’s side, seven generations back, to the Somali national hero
Wiilwial.12

4 This is done in prose which is often both imaginative and refined. The preachers constantly
refer to particular passages of the sacred texts in Arabic as their authority.

5 His full name was Haaji ‘Ali ‘Abdirahmdan. He died about 150 years ago. Among his
numerous poems Alif y^nne ‘ The letter Alif said ’ is particularly well known. Each letter of the
Arabic alphabet is personified in it, as it were, and speaks of a particular aspect of Islamic
doctrine and worship.

6 For information about this poet and preacher and his works see 'Abd al-Rahman bin
shaykh 'Umar, 1954, 1964 ; Cerulli, 1964 (117-38); Ibn Muhyi al-Din, 1955 ; Lewis, 1955-6.

7 He is particularly well known for his poem which contains the phrase Qunihdii n^big^enna
yda qiydasi kdrayd ? ‘ Who can gauge the beauty of our Prophet ? ’ See Lewis, 1958.

8 A poet contemporary with Sheekh Uweys, of the Upper Juba region.

0 The leader of the Dervish insurrection of 1899-1920. For information about this remarkable
man and his poetry, see ‘Abd al-Sabur Marzuq, 1964 ; Andrzejewski and Lewis, 1964 ; Jama'
‘Umar ‘Isa, 1965 ; and Lewis, 1965(a).

10 His Full name is Sheekh M3.hamu.ud M^hdmmed Gabydw. He died towards the end of
the nineteenth century. His poems are very well known, not only in the Benadir Region where
he lived and worked, but also in other Somali-speaking territories.

111 have used an anglicized spelling of his name. The Somali pronunciation of the name
is 'Aaqib ‘Abdulldahi Jaamd'. His official first name is Ibrahim (Ibraahiim), but he prefers
Aqib, the name given to him by his mother, according to an old Somali custom. I am greatly
indebted to him for allowing me to translate his poem and for providing me with most of the
background information. I am also grateful to Mr. Omar Aw Nuh (‘Umar Aw Nuuh) of the
Cultural Division of the Somali Ministry of Education for his help in the preparation of
this paper.

12 A Somali chieftain and war leader about whom many legends are narrated and to whom
many wise sayings are attributed. See Laurence, 1954, pp. 106-22, and Shire Jaamac Axmed,
1967.


23

THE ROOBDOON OF SHEIKH AQIB ABDULLAHI JAMA

He became an orphan early in life and at about the age of seven joined a
Koranic school where he did well and reached the middle grade earlier than other
children. He then discontinued his studies and turned to the usual pastoral and
agricultural pursuits of Somali youth. He became an accomplished horseman and
marksman, and had a brief period of service in the Italian armed forces, which
however ended with his desertion when Italy joined the war against Britain, since
he saw no reason to endanger his life for a morally doubtful cause. The with-
drawing Italian forces left behind them stores and equipment which the general
public readily distributed among themselves. With the proceeds of what he
managed to acquire in this way, he opened a cafe in Jigjiga which proved quite
successful. It was there that he came under the influence of a contemporary poet
of distinction, cAmej6 Ydgol Furr6, and discovered that he himself had a poetic
talent. On the advice of ‘AmSje he handed over the cafe to his relatives to look
after and returned to his family farm, where he would have more leisure and could
turn his attention to the art of poetry. The themes which inspired Aqib Abdullahi
were not worldly ones and this led him to resume his religious studies. He joined
an itinerant theological study group 13 and after a few years he became himself
a sheikh.

Towards the end of the Second World War the activities of the Somali Youth
League gave a new impetus to Somali nationalism.14 Sheikh Aqib recognized the
role which both Islam and Somali poetry could play as unifying forces. Endowed
with special ability as a peacemaker, he concentrated in his preaching on the evils
of fratricidal raids among different clans and gathered a large following.

It was about 1948 that he saw in a dream an alphabet for writing Somali.15
Next morning he wrote it down and was able to use it as an aide-memoire device
in his work as a preacher and composer of hymns. The esoteric nature of his
script did not make it suitable for dissemination and general use, but provided him
with a method of recording Somali oral literature, and now he has one of the
largest collections of Somali poems, proverbs and stories, both religious and secular.

In 1962 Sheikh Aqib left his native region of Jigjiga and came to Mogadishu,
where he soon won recognition both from the general public and from the people
especially interested in Somali culture. He was invited on several occasions to
broadcast on Radio Mogadishu.

13 For an account of such groups, called in Somali h6r, see Andrzejewski and Musa H. I.
Galaal, 1966.

14 See Drysdale, 1964 ; Jama' ‘Umar ‘Isa, 1965 ; Lewis, 1965(a); and Touval, 1963.

15 The script of Sheikh Aqib is based on the alphabetical principle and is no doubt inspired
by the Arabic script, from which it differs in that it marks the vowels throughout and has signs
for the Somali vowels e and o. It does not, however, mark length. The actual shapes of the
letters are of Sheikh Aqib’s invention. For similar new alphabets used for Somali, see Lewis,
1958, and Moreno, 1955, pp. 290-7. The motive behind the invention of new alphabets for Somali
seems to arise from the deficiencies of the Arabic script when applied to Somali and the
reluctance among the conservative sections of the public to use the Latin alphabet.


24

B. W. ANDRZEJEWSKI

The poem which forms the main subject of this paper was recited during a
bad drought in 1956, which was a threat both to livestock and crops. When
Sheikh Aqib came with his itinerant students to a group of settlements called
F^rhoodldy, near Jigjiga, the inhabitants requested him to lead them in prayers for
rain. In addition to the usual prayers and scriptural readings in Arabic, he chanted
his poem in Somali, in the traditional (J&anto genre 16 which is often used for
marching songs. Nevertheless, the poem departs in one way from the usual
standards of Somali secular poetry in that the same alliteration is not carried
through the whole poem, but changes in each couplet, to follow the order of the
Arabic alphabet. This practice, which can be found in other religious poems, is
most probably inspired by a certain type of religious poem in Arabic where,
however, instead of alliteration there is rhyme, but where each line begins with a
consecutive letter of the alphabet.17

As some Arabic consonants have no obvious Somali equivalents, the poet
substitutes them by the nearest Somali sounds. The Arabic letters sin and sad, for
example, both correspond to the Somali consonant s in the alliteration of the poem.
To make up for this apparent deficiency Arabic loan-words are used at certain
points which, though they begin with the same consonant in Somali, have etymons
which are differentiated in this respect, e.g. satdar (Arabic sitar) and s&adiq
(Arabic sadiq).

Each couplet begins with an invocation in which the poet uses the praise names
of God current in Muslim worship. It is important to observe that he gives them
in the forms which have been borrowed from Arabic and not in their pure Somali
equivalents.18 These names, which from the point of view of Somali grammar are
all masculine singular nouns, are listed below.

d&akir who remembers macbuud who is worshipped
daa’im who is everlasting mujiib who answers (prayers)
dsjyaan who rewards and punishes khdaliq who creates
fatdab who opens, who initiates ndasir who is victorious, who grants
(everything) victory
haliim who is clement, who is patient qdadir who is powerful
jaliil who is glorious qdni who is rich
jawdad who is generous rabmdan who is merciful
kafiil to whom (one’s cares) can be rahiim see rabmdan
entrusted, who protects sdadiq who speaks the truth
karlim who is generous, who is noble saldam who is peace, who grants peace

16 For an. account of the genres of Somali poetry see Andrzejewski and Lewis, 1964, and
Maino, 1953. For bibliographical information on Somali poetry, see Andrzejewski, 1967; and
Johnson, 1967 ; to which Musa H. I. Galaal, 1968, should be added.

17 For examples of such alphabetical poems see ‘Abd al-Rahman bin Shaykh ‘Umar, 1964,
pp. 86-7 and 208-10.

18 For an account of the praise names of God in Somali, see Lewis, 1959.


THE ROOBDOON OF SHEIKH AQIB ABDULLAHI JAMA

25

sandac who makes, who creates
satdar who is a curtain or shield, who
covers up (sins)

shaakir who is cognizant of merit, who
is appreciative

shakuur as above ; or ‘ to whom thanks
must be offered ’

taabid who is firm, who is immutable

t£ajir the same meaning as qdni

tawdab who accepts repentance or

penance
wdahid who is one

wakiil the same meaning as kafiil;
or ‘ who can act on a per-
son’s behalf’

wahdab who provides (all)

Each invocation is followed by a petition and thus the poem resembles a litany
in structure, except that both parts are recited by the same person who leads the
prayers. The congregation repeats after each pair of lines the following refrain

in Arabic. Allahu Allah Allahu Allah Allahu, aghith lana mataran I Oh God, oh God, oh God, oh God, Oh God, help us with rain !

The Somali text given below has been transcribed by myself under plain voice
dictation from Sheikh Aqib and then checked against his chanted recitation given
to a small group of Somalis interested in oral literature.19

1. pdaha arsaaqd uunkjisdw

Ibdha roobka noo irmaanhe ydey

2. pdaha baddha bjydha ku shubdw
Bjldaddani ndo barwaaqhe ybey

3. pdaha tawaab ee t^ajir ohdw
Tggydasha §ngegdy harhedda ku tuul

4. pdaha tdabid hh ee tawdabka haydw
Sidaan jechllahdy na solansii yhey

5. pdahay jaliil ee jawdad hdq dhow
Jalaad na djl ye, ifrkii rdob hey

6. fldahay haliim ee hdq Loo c§,abuddw
Hayawdanka oomdy barbed u lis bey

7. |ldahay khdaliq ee khdlqiga uumdw
Khardab cfacdy khdyr ku s6o b^ddel hey

19 The system of transcription used is the same as in Musa H. I. Galaal, 1956, and my The
declensions of Somali nouns, 1964.


26

B. W. ANDRZEJEWSKI

8. pdaha daahm ee d^yaanka ohdw
Dadkii cg.bbay rdob ha ndo dspe dey

9. pdahay ddakir dh een dikrigji badinndw
Daruurdha rdob kd sdo d$a ydey

10. pdahay rahmaan ee rahiimka ohdw
Isldamka rafaadsdn rdob u lis dey

11. pdahay sabaan kastd sdadka bahshde

Si doond faldw, na sii madar eey

12. pdahay saldam ee satdarka ohdw
Sidaan jecellahdy na solansii ydey

13. pdaha shdakir ee shakuurka dhow
Dijaamda shiddn harded ku shub dey

14. pdaha sdadiq eh ee sandaca dhdw
Sidaad ti^da ba ku saburray dey

15. pdaha durkiyo ndcfiga bal darbow
Ddlkii kd jdbnay e, daruurta lis dey

16. pdaha c(ulkiyo samdda (psdydw

Md (Jaansdn karrd e, c[ibcdha rdob dey

17. pdaha dalaamka h^bdenka dporiyde
Dardar kd dambaysiiydw, ddeqda lis dey

18. pdaha cadcdedda cadddan kd c[igde
cflmig?eda Qgdw, hardedda cds dey

19. pdaha qdni ydh e qdmmiga faydow
Aghithnaa ghdythan ‘damman, qiydas u Us dey

20. pdaha fatdah dh ee arsdaqda furdw

La kala f jri^ ye rdob f§,ydw kden

21. pdaha qdadir dh ee taldda qummiydw
Arldda mddar qqoya noogu qub dey


27

THE ROOBDOON OF SHEIKH AQIB ABDULLAHI JAMA

22. pdaha kariim ee kafiilka dhow

Abdar mi karri e, kdydka rdob khen hey

23. pdaha lay^a wadi lulaysa (prtow
LaGugu hqmdi yey, ladndan rdob hey

24. Pdaha macbuud ee mujiibka dhdw
Arlada madar gqacja ndo mdri yhey

25. pdaha ndasir eh ee na'mdda baddndw
Nageeyb nusqdan Id’ ndo khen hey

26. pdahay waahid 6h ee wakiilka dhdw
Wahdabow wardabi Aadmiga hey

27. pdaha fad uu hildy hilldac kd bahshdw
Adda hayd rdob hamiiqbdh dh hey

28. pdaha Idas La qoddy bjyaha ku lacjdw
Labeenshdha rdob arldda ku lis hey

29. pdahay tald ydab leh yasiri jjrdw
Addan yusri Kda yaboohsaday hey

1. You who give sustenance to your creatures, oh God,
Put water for us in the nipples of rain !

2. You who poured water into oceans, oh God,
Make this land of ours fertile again !

3. Accepter of penance, who are wealthy, oh God,
Gather water in rivers whose beds have run dry !

4. You who are steadfast and act justly, oh God,
Provide us with what we want you to grant !

5. You who are glorious, truly bounteous, oh God,
Our cries have undone us, grant a shower of rain !

6. You who are clement, truly worshipped, oh God,
Milk water for beasts which are stricken with thirst !


28

B. W. ANDRZEJEWSKI

7. Creator of nature who made all things, oh God,
Transmute our ruin to blessing and good !

8. Eternal rewarder of merits, oh God,

Let that rain come which people used to drink !

9. We have done much Remembrance,20 oh God who remember,
Loosen upon us rain from the clouds !

10. You who are merciful and compassionate, oh God,
Milk rain from the sky for Muslims in need !

11. Giver of victuals at all times, oh God,

Who can do what you want, bestow on us rain !

12. You who are peace and a curtain, oh God,
Provide us with what we want you to grant !

13. Recorder of merit, who requite us, oh God,

Into scorched empty ponds pour us water of rain !

14. You who are truthful, creator, oh God,

We accept in submission whatever you say !

15. You who mete out good and evil, oh God,

In this land we are broken, milk the clouds from above !

16. The earth and the sky you constructed, oh God,
We cannot get water, bring forth drops of rain !

17. The darkness of night you transfigure, oh God,
And make daylight follow ; milk the sky lavishly !

18. You who gave brightness to sunshine, oh God,

And know its principles, give us brown water from rain !

19. You who are rich and ward off cares, oh God,
Milk temperate rain ! help us with rain everywhere !

20. You who open all and give sustenance, oh God,
People have scattered; send forth healthy rain !

20 This is a ceremony called in Somali dikri (from Arabic dhikr) ‘ Remembrance (of God) ’.
It consists of devout invocations repeated for a long time, accompanied by rhythmic movements
of the body, clapping, stamping and sometimes also drumming. It is practiced at the meetings
of the Sufi fraternities. See al-sharif ‘Aydarus bin al-sharif ‘Ali, 1955; and Lewis, 1955-6.


29

THE ROOBDOON OF SHEIKH AQIB ABDULLAHI JAMA

21. Almighty, perfecter of counsels, oh God,
Pour for us rain which would make the land wet !

22. You who are bounteous, the protector, oh God,

We cannot survive drought, send us rain from your store !

23. You who drive the air which sways the trees, oh God,

It is you whom we praised, grant us the goodness of rain !

24. You who are worshipped and answer prayers, oh God,
Make the rain spread over the whole of the land !

25. Bestower of victories, benefactor, oh God,

Bring us faultless rain which makes us dwell where it falls !

26. You who are one and are trusted, oh God,
Provider of all, give water to man !

27. You who spark off lightning from clouds you have loaded, oh God,
It is you who have power over rain which satiates !

28. You who fill water-holes dug in wadis, oh God,
Milk rain on this land, cream-giving rain !

29. Who used to relieve the strangest plights, oh God,
It is to you that I have turned for help !

In the translation of the poem great care has been taken to preserve the
imagery of the original and to avoid embellishments or omissions. At the same
time no attempt has been made at a literal translation, since it is doubtful whether
it would provide any insight into the nature of the original, on account of a vast
difference in structure between the two languages. For readers who might be
interested in identifying the meanings of the individual words, a vocabulary is
provided below. It contains all words not explained so far, which cannot be found
in Abraham’s dictionary or in Bell’s The Somali language or which require
additional elucidation. The letters ‘m’ and ‘ f’ refer to the gender of nouns 21 and
the capital code letters denote the root extension class in verbs according to my
system of classification.22

21 As defined in my The declensions of Somali nouns, 1964. This work was originally prepared
as a Ph.D. thesis under the guidance and supervision of Professor Malcolm Guthrie.

22 As presented in my articles, 1968, 1969.


30

B. W. ANDRZEJEWSKI

Aadmi (m) human being or beings ; the offspring of Adam

aghithnaa ghdythan 'iarnman an Arabic phrase meaning ‘ assist us with rain which is
general ’ (i.e. which spreads over the whole country)

drlo (f) country, land

arsdaq (Z) to give sustenance

arsdaq (f) sustenance

c$abud (Z) to worship

Hamman see under aghithnaa, etc.

cadcded (f) sunshine

barwaaqde (AYN) to render prosperous

ddrab (Z) to prepare (root: darab ~ darb)

daldam (m) darkness

dibic (f) drop

dibco (m) plur. form of dibi ‘

dur (m) evil, misfortune

(Jaansd (SAN) to fetch water for oneself

^ijaamo (m) plur. form of (Jij&an

cfijdan (f) natural pond, reservoir

dey a particle used in poetry, without any specific meaning

fdyd (Z) to bare

f^yow (Zf) to be healthy

firi(| (Z) to disperse

ghdythan see under aghithnaa, etc.

hamiiqbdh (m) that which satiates (poetic)

hil (Z) to fasten a cover on a water or milk vessel before loading it on to a burden
beast; to load

hdmdi (Z) to praise

hayawdan (m) animals

ib (f) orifice of the nipple or penis

irmaande (Zf/AYN) to cause to be (‘ to become ’) in milk ; to cause to have water
Isldam (m) Islam, Muslims
jaldad (m) shouting, clamour
khalqi (m) creation, creatures
khardab (m) ruin, devastation

Idas (m) shallow well, usually dug in a seasonal river bed

labdenshe (m) one which gives cream (poetic)

ladndan (f) being well, prosperity

ld
ldnaa see under aghithnaa, etc.

mddar (m) rain (poetic)

ndcfi (m) profit, benefit

nagdeye (m) that which causes people to settle in a particular place for a time (poetic)


31

THE ROOBDOON OF SHEIKH AQIB ABDULLAHI JAMA

n^'mo (f) favour, benefit, grace
nusqaan (f) deficiency, defect
qammi (m) care, worry
qgo (IN) to wet (trans.)
qiydas (f) measure, moderation
qummi (IN) to straighten, to rectify
rafaadsdn (Zt) to endure hardship
sdbur (Z) to become patient; to accept patiently
solansii (SUN) to grant
shiddn (ANf) to be alight
tawdab (m) reward, recompense
uun (Z) to create (root: uun ~ uum)
uun (m) creation, creatures
yaboohso (SAN) to seek offers of help or gifts
ydsir (Z) to relieve, to ameliorate (poetic)
ydey a positional variant of dey
yusri (m) ease, relief (poetic)

Many of the words in the poem occur in their contracted forms or in their
positional variants which require elucidation. In the list below the technique of
explanation is the same as in the introduction and notes to Musa Galaal’s
Hikmad Soomaali. The -ow component in several contractions is a masculine
vocative suffix, used when addressing a person.

adda = adi + bda
adaan = adi + bdan
'§.abuddw = cg,abudd + ow
baddndw = baddn + ow
badinndw = badinndy + ow
bahshde = bahshd ee = bihiyd ee
bahshdw = bahshd + ow = bihiyd + ow
darbdw = darbd + ow
d§?e dey = d^’o dey
dambaysiiydw = dambaysiiyd + ow
doond faldw = doond faldw = doond fald

+ ow

dporiyde = dporiyd ee
durkiyo = durka iyo
cfigde = (figdy ee
(fijaamda = cpjaamdha

c[irtdw = 4,irta + ow
een = ee aan = ee aannu
dh ee - dh ee
faldw = fald + ow
fayddw = faydd + ow
furdw = furd + ow
haydw = hayd + ow
jjrdw = jjrdy + ow
karrd e = karrd e
lacfdw = lacja + ow
qgdw = $g + ow
dhdw = dh + ow
shubdw = shubdy + ow
uumdw = uumdy + ow
uunkjisdw = uunkjisa + ow

When Sheikh Aqib recited his roobddon, the villagers were very much moved,
apparently reaching a state of religious ecstasy (way ku jjbboodeen). The emotional


32

B. W. ANDRZEJEWSKI

appeal of the poem, quite apart from its artistic value, is no doubt enhanced by
two facts: firstly all members of the congregation could fully understand the
language and, secondly, to many men the poem must have evoked memories of
early childhood.

In Somali culture there is a tradition of prayers for rain recited by women who
go with their small children to gatherings in isolated spots; adult men are never
present. The prayers are in Somali, are very simple in structure, and contain
imagery which is obviously derived from direct experience and observation and
not from Muslim learning.23 Thus for example, in a woman’s prayer from the
Jigjiga region the quality of the rain which they want is described as follows:

Ku rarankii abaareed ba naga reeba noo kden I

Ku r^adkii La qaadaa ba rays yeeshd ndo kden 1

Ku rdmadkii cfala ba subagga Lagd riiho ndo kden !

Ku ceesaam6 giirgiiran g§eskpoda humbeeya ndo kden I

Bring us one which takes away from the ground the hidden embers of drought !

Bring us one which leaves every footprint wet !

Bring us one through which ghee is obtained from beasts that have calved
recently !

One which puts froth on the horns of the spotted young goats !

In the same poem we find a lament on the effects of the drought on the people
around:

Umuli oontay, rdob dey, Allow rdobka ndo kden I

Ardadu gybdtay, rdob hey, Allow rdobka ndo khen I

Aroos d^landdl, rdob hey, Allow rdobka ndo khen !

A woman who has given birth is stricken with thirst. Rain, oh God, give us
rain !

The students of the Faith are scorched. Rain, oh God, give us rain !

The bride and the bridegroom had to trudge away from the feast. Rain, oh
God, give us rain !

23 These prayers are accompanied by certain ceremonies which seem non-Islamic in their
nature. According to Mr. Musa Galaal—a recognized authority on Somali customs and oral
literature—in the Northern part of the Somali-speaking territories there is a rite called ahaddysi:
women select a particularly virtuous and pious young wife, and then dishevel her hair, throw
sand at her, strike her and generally molest her. They tell her that she must pray for rain and
that unless God hears her prayer they will continue tormenting her. The reasoning behind it is
that a good person’s prayers are most likely to be answered. Similarly, small children are
harassed by being made to walk barefoot on sand hot from the sun. Cords are also tied round
the little fingers on their left hands and they are told to pray for rain, or they will not be released.
It may be of interest to note that it is by the little finger of the left hand that spirits are believed
to leave a possessed person during the shar (zar) ceremonies (see Lewis, 1961).


THE ROOBDOON OF SHEIKH AQIB ABDULLAHI JAMA 33

The small children who accompany their mothers also chant their own prayers
in which they describe the plight to which the drought has reduced them:

Allahaydw f&r baa i hicJAn,

Allahaydw, i sdo fur6 1

Allahaydw, harmed roob !
Allahaydw, bakhtdan cunay I
Allahaydw, bjyAan haday !

Oh God, my finger is tied,24
Oh God, release it for me !
Oh God, give us rainwater !
Oh God I have eaten carrion !
Oh God I have stolen water !

BIBLIOGRAPHICAL REFERENCES

‘Abd al-Rahman bin shaykh ‘Umar, c. 1954. Jala" al-'aynayn ft manaqib al-shakhayn :
al-shaykh al-wali hajj Uways al-Qadiri wa al-shaykh al-kamil al-shaykh 'Abd
al-Rahman Zayla'i, Matba'at al-Mashhad al-Husayni, Cairo.

‘Abd al-Rahman bin shaykh ‘Umar, 1964. Al-jawhar al-nafis fi khawas al-shaykh Uways,
Matba'at al-Mashhad al-Husayni, Cairo.

‘Abd al-Sabur Marzuq, 1964. Tha’ir min al-Sumal: al-mulla Muhammad 'Abd Allah Hasan,
Dar al-Qawmiyya lil-Taba'a wa al-Nashr, Cairo.

Abraham, R. C., 1964. Somali-English dictionary, University of London Press.

Ahmad bin Husayn bin Muhammad, 1945. Manaqib al-ustadh al-shaykh Isma'il bin Ibrahim
al-Jabarti, Mu§tafa al-Babi al-Halabi, Cairo.

Andrzejewski, B. W., 1964. The declensions of Somali nouns, School of Oriental and African
Studies, London.

Andrzejewski, B. W., 1967. ‘ The art of the miniature in Somali poetry’, African Language
Review, VI, 5-16.

Andrzejewski, B. W., 1968. ‘ Inflectional characteristics of the so-called weak verbs in Somali ’,
African Language Studies, IX, 1-51.

Andrzejewski, B. W., 1969. ‘ Some observations on hybrid verbs in Somali ’, African Language
Studies, X, 47-89.

Andrzejewski, B. W., and Musa H. I. Galaal, 1966. ‘ The art of the verbal message in Somali
society ’, Neue Afrikanistische Studien, ed. J. Lukas, Hamburg, 29-39.

Andrzejewski, B. W., and I. M. Lewis, 1964. Somali poetry : an introduction, The Oxford
Library of African Literature, The Clarendon Press.

al-sharif 'Aydarus bin al-sharif ‘Ali al-'Aydarus al-NadirI al-‘Alawi, 1955. Bughyat
al-dmal fi taTikh al-Sumal, Stamperia A.F.I.S., Mogadishu.

Bell, C. R. V., 1953. The Somali language, Longmans, London.

Bell, C. R. V., 1968. The Somali language, reprinted by Gregg International Publishers, Ltd.,
Farnborough.

Cerulli, Enrico, 1957, 1959, 1964. Somalia : scritti vari editi ed inediti, Istituto Poligrafico
dello Stato, Rome, vols. I, II and III.

Drysdale, John, 1964. The Somali dispute, Pall Mall Press, London.

Ibn MuhyI al-DIn Qasim al-Barawi, 1955. Majmu'a qasa'id, Mu§tafa al-Babi al-IJalabi,
Cairo,_ 2nd ed.

Jama' ‘Umar ‘Isa, 1965. Ta'rikh al-Sumalfial-'usur al-wusta wa al-haditha. Matba'at al-Imam,
Cairo.

24 See note 23.

D


34 THE ROOBDOON OF SHEIKH AQIB ABDULLAHI JAMA

Johnson, John William, 1967. A bibliography of Somali language materials, Peace Corps,
Hargeysa, Somali Republic.

Laurence, Margaret, 1954. A tree for poverty : Somali poetry and prose, The Eagle Press,
Nairobi.

Lewis, I. M., 1955. Peoples of the Horn of Africa : Somali, Afar and Saho, Ethnographic
Survey of Africa, Part I, International African Institute, London.

Lewis, I. M., 1955—6. ‘ Sufism in Somaliland : a study in tribal Islam ’, Bulletin of the School
of Oriental and African Studies, XVII, 3, 581-602 ; XVIII, 1, 146-60.

Lewis, I. M., 1958. ‘ The Gadabursi Somali script ’, Bulletin of the School of Oriental and

African Studies, XXI, 1, 134—56.

Lewis, 1959. ‘ The names of God in Northern Somaliland’, Bulletin of the School of Oriental
and African Studies, XXII, 1, 134-40.

Lewis, I. M., 1961. A pastoral democracy : a study ofpastoralism and politics among the northern
Somali of the Horn of Africa, Oxford University Press.

Lewis, I. M., 1963. ‘ Dualism in Somali notions of power’, The Journal of the Royal Anthro-
pological Society, XCIII, Part I, 109-16.

Lewis, I. M., 1965(a). The modern history of Somaliland: from nation to state, Weidenfeld
and Nicolson, London.

Lewis, I. M. 1965(6). ‘ Shaikhs and warriors in Somaliland’, African systems of thought, ed.
M. Fortes and G. Dieterlen, Oxford University Press.

Lewis, I. M., 1966. ‘ Conformity and contrast in Somali Islam ’, Islam in tropical Africa, ed.
I. M. Lewis, Oxford University Press, 253-67.

Maino, Mario, 1953. La lingua somala : strumento d'insegnamento professionale, Ferrari,
Occella & C., Alessandria (Italy).

Moreno, Martino Mario, 1955. Il somalo della Somalia : grammatica e testi del Benadir,
Darod e Dighil, Istituto Poligrafico dello Stato, Rome.

Muhammad al-Zuhri al-Ghumrawi, 1933. Al-siraj al-wahdj Mu?tafa al-Babi al-Halabi,
Cairo. A commentary on Matn al-Minhaj by Sharaf al-Din Yahya al-Nawawi
published together with the original text.

Musa H. I. Galaal, 1956. HikmadSoomaali, ed. B. W. Andrzejewski, Oxford University Press.

Musa H. I. Galaal, 1968. The terminology and practice of Somali weather lore, astronomy and
astrology (cyclostyled), available from New Africa Bookshop, Mogadishu (P.O.
Box 897).

Sharaf al-DIn Yahya al-Nawawi, 1933. Matn al-minhaj, Mustafa al-Babi al-Halabi, Cairo,
published together with a commentary ; see under Muhammad al-Zuhri al-Ghumrawi.

Shire Jaamac Axmad, 1967. ‘ Wiil-waal (Garaad Faarax) sheekadiis’, Iftiinka-Aqoonta {Light
of Education), VI, Mogadishu, 1-31.

Touval, Saadia, 1963. Somali nationalism : international politics and the drive for unity in the
Horn of Africa, Harvard University Press.


1ST AND 2ND PERSON PRONOMINAL FORMS IN FULA

By D. W. Arnott

One problem which has to be faced in analysing a class language is whether
pronominal forms in general, and 1st and 2nd person forms in particular, are to
be treated as part of the class system, or as external to it. In many cases there is
a clear correlation between 3rd person pronominal forms and concord markers of
the nominal system, as well as a close syntactical link, so that there is little difficulty
in including them within the system. But the position regarding 1st and 2nd person
forms is often much less clear. This is so in Fula, where a detailed study of the
shapes and behaviour of 1st and 2nd person forms, alongside those of 3rd person
forms, is necessary before a decision can be made on the best way to treat them.

2. The problem is here examined with reference to the Gombe variety of Fula,
but similar considerations would apply for all dialects. After an introductory
section listing the forms and the types of construction in which they occur, they
are compared first from a morphological and then from a syntactical point
of view.

3. Fula pronominal forms occur in five series, exemplified in Table I, and here
labelled respectively Subject Element (S.E.), Object Element (O.E.), Independent
Pronoun (I.P.), Possessive Pronoun, and Possessive Suffix. For Gombe Fula each
series consists of 32 forms, as described in Arnott 1970, chapters 22 and 24-6;
but here a shorter list is adequate for exemplification. The forms in each of these
series are in complementary distribution, and occur in comparable syntactical
situations; this will be clear from the selected examples given below, which
illustrate the various syntactical situations in which these pronominals can occur:

Subject Elements occur in combination with a verbal base (i.e. radical + tense
suffix) to form a minimal verbal complex (Arnott 1970, §§ 28.1-4), e.g.

General Future Relative Past
mi-nodday I will call noddu-mi I called
a-nodday you will call noddu-cfaa you called
o-nodday he/she will call o-noddi he/she called
ngel-nodday he/she/it will call (diminutive) ngel-noddi he/she/it called (diminutive)
min-nodday we (he and I) will call min-noddi we called
en-nodday you and I will call noddu-tfen you and I called
on-nodday you (pl.) will call noddu-cfon you (pl.) called
6e-nodday they will call 6e-noddi they called
kon-nodday they (dim.) will call kon-noddi they (dim.) called
Note on-/-tfon. that some forms vary according to tense, e.g. a-/-daa, en-/-den,


36

D. W. ARNOTT

Table I

Person Subject Element Object Element Independent Pronoun Possessive Pronoun Possessive Suffix
1 sing. mi/-mi -yam/-am miin’ am -am
2 „ a/-aa/-daa -ma/-maa’/-e aan’ maada/maa -a
1 pl. min -min minon amin -iimin
2 pl. incl.1 2 en/-en/-den -’en enen meeden/men -ii’en
2 pl. excl. on/-on/-don -’on onon moodon/mon -ii’on
3 sing, o class o/mo -mo’/-moo- hanko’ maako’ -iiko’
nde „ nde -nde’ hayre’ maare’ -iire’
ndu „ ndu -ndu’ hayru’ maaru’ -iiru’
nge „ nge -nge’ hange’ maage’ -iige’
ngu „ ngu -ngu’ hangu’ maagu’ -iigu’
ngal „ ngal -ngal’ hangal’ maagal’ -iigal’
ngel „ (diminutive) ngel -ngel’ hangel’ maagel’ -iigel’
3 pl. Be class 6e -6e’ hamBe’ ma66e’ -iiBe’
de „ de -de’ hanje’ maaje’ -iije’
di „ di -di’ hanji’ maaji’ -iiji’
kon „ kon -kon’ hankon’ maakon’ -iikon’
(diminutive)

In the above table, a final apostrophe represents Final Glottality, where the form
occurs in pause.

Subject Elements also occur in Subject position in non-verbal sentences, e.g.

mi Pullo I am a Fulani

o Pullo he/she is a Fulani

min Ful6e we are Fulani

6e Ful6e they are Fulani

mi bumdo
a bumdo na ?

min wum6e
on wum6e na ?

I am blind
are you blind ?
we are blind
are you blind ?

1 The en forms, referring to the speaker and the person(s) addressed, are treated as 2nd person
inclusive rather than 1st person inclusive, because their morphological structure is markedly
different from that of the 1 st plural min forms in most respects. Syntactically, too, they resemble
the 2nd person forms rather than the 1st plural, since in the Relative Past, Relative Future and
Subjunctive tenses the ‘ you and I ’ S.E. -en, -cfen follows the verbal base, like the 2 sing, and

2 pl. S.E.s -aa, -cfaa and -on, -don, whereas the 1 pl. S.E. min precedes the base in these tenses
as in all others.

Semantically, too, since en refers to both ‘ you ’ and ‘ I ’ there is no particular reason why it
should be labelled a first person because of the ‘ I ’ element. If in English it is reasonable to
assign ‘ you and I ’ to the 1st person on formal grounds, because of sentences such as ‘ you and
I have finished our work ’, in Fula it is equally reasonable to assign the pronoun to the 2nd person
because formally it behaves more like the other 2nd person forms than like the 1st person
plural forms.


FIRST AND SECOND PERSON PRONOMINAL FORMS IN FULA 37

Object Elements occur in combination with a minimal complex to form an
enlarged complex, e.g.

General Past Relative Past

o-noddii-yam he called me o-noddi-yam he called me
o-noddii-ma he called you o-noddu-maa he called you
o-noddii-mo he called him/her o-noddi-mo he called him/her
o-noddii-ngel he called him/her/it o-noddi-6e he called them
o-noddii-min he called us
o-noddii-6e he called them noddu-maa-mi I called you
noddu-moo-mi I called him/her
Desiderative noddu-mi-ngel I called him/her/it
Alla wall-am God help me noddu-mi-6e I called them
Alla wall-e God help you
Alla wallu-mo God help him/her
Alla wallu-fie God help them

Note that some forms vary according to tense and other factors, viz. -yam/-am,
-ma/-maa/-e, -mo/-moo. For details, see Arnott 1970, §§ 37.5-11.

Object Elements also occur after infinitives and participles, e.g. nodduki mo
‘ to call him ’, nodducfo mo ‘ one who called him ’, noddudo 6e ‘ one who called
them ’.

Independent Pronouns occur in a number of different positions, including
(a) first item in an appositional nominal group (Arnott 1966, §§ 27 ff.; 1970,
§§6.8, 24.14), (Z?) in ‘Prelude’ position in a sentence (Arnott 1970, §§7.10-17,
24.15), or (c) as an ‘addendum’ at the end of a sentence (Arnott 1970, §24.15), e.g.

(а) miin Bello
hanko bumcfo on
minon Ful6e
hanko Pullo o’o
ham Be FulBe Ben
hankon mbaccon kon
hanji pucci cfi’i

also hanko o’o
aan o’o
onon fie’e

(б) miin mi-anndaa
aan a-anndaa na ?
hanko o bumdo
hangel ngel bumngel
ham fie Be wumBe

(c) mi-anndaa-mo hanko
mi-anndaa-ma aan

aan Bello
onon Ful6e

I, Bello
he, the blind man
we Fulani
(he,) this Fulani
they, the Fulani
they, the children
(they,) these horses
this one (lit. he this)
you there (lit. you this)
you there (pl.)
(as for) me, I don’t know
(as for) you, don’t you know ?
(as for) him, he is blind

(as for) him (dim.), he is blind
(as for) them, they are blind
I don’t know him

I don’t know you

you, Bello

you Fulani


38

D. W. ARNOTT

Possessive Pronouns occur as the second item in genitival complexes, while
Possessive Suffixes occur in combination with the stems of certain kinship and
other relationship terms, e.g.
Poss. Pronoun Poss. Suffix
baaba am jawmam my father, my master
baaba maada/maa jawma your father, master
baaba maako jawmiiko his/her father, master
baaba maagel jawmiigel his/her/its (dim.) father, master
baaba amin jawmiimin our father, master
baaba meeden/men jawmii’en our father, master (yours and mine)
baaba moodon/mon jawmii’on your (pl.) father, master
baaba ma 6 Be jawmii6e their father, master
baaba maakon jawmiikon their (dim.) father, master

4. While these five series are complete sets in which all the forms are in com-

plementary distribution, there is an obvious initial distinction between 3rd person
forms and 1st and 2nd person forms, namely in their semantic and grammatical
reference. 3rd person forms in each series refer to one or more person(s) or
thing(s) other than the speaker or person addressed ; and the particular form used

is determined by the class of the noun referred to. (The noun has usually already
occurred in the linguistic context, but it may simply be the Fula word for a person
or thing in the context of situation but not previously mentioned.) Thus the
3rd person forms listed in Table I could refer to nouns such as the following :
Class
0 laamdo chief; Pullo a Fulani; kodo stranger
nde loonde storage-pot; deptere book
ndu waandu monkey ; kutiiru dog ; harmaaru prostitute
nge nagge cow ; wiige heifer
ngu linngu fish; puccu horse ; mbonjokku bag
ngal gertogal hen ; cofal pullet
ngel laamngel petty chief; baccel child
6e laamBe chiefs ; Ful6e Fulani
de gertoode hens ; depte books
di baadi monkeys ; pucci horses; bacci children
kon laamkon petty chiefs ; mbaccon small children

1st and 2nd person forms (which refer to the speaker or person addressed or
both, alone or together with others associated with them), do not refer back to a
noun of any particular class in the same way as the 3rd person forms. Even in the
case of 1st plural and 2nd plural exclusive forms, where a third person or thing
could be included in the reference, there is no grammatical reference to any
particular class. For instance min ‘ we ’ could refer back to miin e hanko ‘ I and


FIRST AND SECOND PERSON PRONOMINAL FORMS IN FULA 39

he ’ (o class), miin e hangel ‘ I and he/she/it ’ (ngel class, diminutive), or miin e
hangu ‘ I and it ’ (ngu class); cf. also

miin e soobaajo am, min-shomii (soobaajo is in the o class)

I and my friend are tired
miin e puccu am fuu, min-shomii (puccu is in the ngu class)
both I and my horse are tired

5. This, then, is one way in which the two groups of forms can be distinguished;
but a more detailed study is necessary, both from a morphological and from a
syntactical point of view.

MORPHOLOGICAL CONSIDERATIONS

6. From a morphological point of view it is clear from Table I that most 1st and
2nd person forms have a different internal structure from the corresponding
3rd person forms.

(а) While 3rd person Subject and Object Elements of any one class have the
same shape (with the single exception of the o class, which in Gombe normally
has Subject Element o, Object Element -mo), 1st and 2nd sing. Object Elements
have different shapes not systematically relatable to the corresponding Subject
Elements. Only in the case of the plural forms, min, en, on,2 is the Object Element
the same shape as the more usual form of the corresponding Subject Element.
Moreover (i) 1st sing, and all 2nd person Subject Elements each have two or three
different shapes, some of which are suffixed to the verbal base, and not prefixed
to it, as is the case with most S.E.s; and (ii) 1st sing, and 2nd sing. O.E.s each
have several shapes, found in different tenses, whereas all 3rd person forms have
only one shape, constant for all tenses (again with the single anomalous exception
of the o class, which has -moo with 1st sing. S.E. in certain tenses—e.g. noddu-
moo-mi given in § 3 above).

(б) 3rd person Independent Pronouns consist of an initial element han- 3/hay-
(in complementary distribution, hay- occurring only before -r-) combined with an
element systematically relatable to the S.E. and O.E. of the same class, and more
closely to the corresponding suffixes in nouns and adjectives; for details, see
Arnott 1970, § 24.1. 1st and 2nd person forms do not have this initial han-/hay-
element; they could perhaps be analysed as consisting of a final -Vn combined
with an element identical with the S.E., the V normally being identical with the
vowel of the 1st element (except in the case of 1st pl.).4 But this is a different
pattern from that of the 3rd person forms.

2 The S.E.s en and on both have an initial glottal stop, i.e., ’en, ’on, although this is not
shown in the orthography used here (which is virtually the same as that recommeded by the
UNESCO orthography meeting at Bamako in 1966).

3 n — homorganic nasal. Most other dialects have kan-/kay-.

4 Other dialects have S.E. min, I.P. minen, or S.E. men, I.P. menen ; this last pair is more
comparable with the enen and onon pairs.


40

D. W. ARNOTT

(c) 3rd person Possessive Pronouns consist of an initial element maa- combined
with an element systematically relatable to the S.E. and O.E. of the same class,
and more particularly to corresponding suffixes in nouns and adjectives (for
details see Arnott 1970, § 24.1). The only exception is ma66e, in which the initial
element can be stated as maG-, where € is a consonant identical with the following
consonant; this is comparable with the ma€- in many possessive forms in other
dialects : makko, maggu, majje, makkon, etc.

The uncontracted 2nd person forms have a somewhat similar structure, but the
initial element is not maa- but mW-, the long vowel being identical with that of
the final element. (Contrast, for instance, meecfen with the 3rd person form
maagel—not *meegel—and moodon with the 3rd person form maakon—not
*mookon.) This final element, it may be noted, resembles the third form of the
S.E. -daa, -den, -don, though the vowel of the 2nd sing, possessive is short, not long.

The two 1st person forms, on the other hand, have quite a different structure,
with no such initial element at all.

(d) Possessive Suffixes. Only in this case is there a close parallel between
1st and 2nd person plural forms and 3rd person forms, all having an -ii- element
combined with a final element systematically relatable to the corresponding S.E.
and O.E. But 1st and 2nd singular forms do not have a comparable structure.

7. Final Glottality. It is noticeable that all the 3rd person forms have Final
Glottality when they are in pause, indicated by the final apostrophe in Table I.
(3rd person S.E.s always precede the verbal base, or the Complement in non-verbal
sentences, so that they never occur in pause.) 1st and 2nd person forms, on the
other hand, do not have Final Glottality, with the sole—and rather odd—
exceptions of 1st and 2nd sing. Independent Pronouns, and the long-vowelled form
-maa of the 2nd sing. O.E.

8. From the foregoing it is clear that on the whole the 1st and 2nd person
forms are, morphologically, markedly different from the 3rd person forms, the
exceptions being the 1st and 2nd person plural forms of O.E. and the Possessive
Suffix, though even these differ from the corresponding 3rd person forms in not
having Final Glottality.

9. In the five categories of pronominals considered above and illustrated in
Table I, there are, as indicated in § 4, distinct 3rd person forms for each of the
nominal classes, although there is no such class distinction in 1st and 2nd person
forms. In addition to the pronominal forms, there are distinct forms for each
nominal class in the six categories of forms illustrated in Table II, viz. Genitive
Element, Relative Element, Interrogative, Far Demonstrative, Near Demonstrative
and Referential; and these forms too are relatable in each case to the S.E. of the
same class, as follows :

Genitive Element)
Relative Element/

are identical with the

S.E.


FIRST AND SECOND PERSON PRONOMINAL FORMS IN FULA

41

Interrogative is identical with the S.E. + -ye/-e 5

Far Demonstrative is identical with the S.E. + -ya/-a 5

Referential is identical with the S.E. + -n/o 6

Near Demonstrative resembles the S.E., with reduplication of the vowel, and
an intervening glottal stop, or a glide with or without glottal creak.

All these forms, varying for class, are relatable to 3rd person pronominals.
It is not surprising that there are no corresponding 1st and 2nd person forms in
these categories ; but the absence of corresponding forms is itself a further point
of contrast between 1st and 2nd person pronominals and 3rd person pronominals.

Table II

Genitive and Relative Element Interrogative Demonstrative Referential
Near Far
3 sing, o class mo moye o’o oya on
nde „ nde ndeye nde’e ndeya nden
ndu ,, ndu nduye ndu’u nduya ndun
nge „ nge ngeye nge’e ngeya ngen
ngu „ ngu nguye ngu’u nguya ngun
ngal „ ngal ngale nga’al ngala ngal
ngel „ ngel ngele nge’el ngela ngel
3 pl. 6e class fie 6eye 6e’e 6eya fien
de de deye de’e deya den
di „ di diye di’i diya din
kon „ kon kone ko’on kona kon

SYNTACTICAL CONSIDERATIONS

10. The essence of the class system, however, is agreement between forms
having the same reference. And in considering whether or not 1st and 2nd person
forms are to be treated as belonging to the class system we must consider how far
they enter into class agreement in the same way as 3rd person forms. The con-
structions appropriate for this purpose are :

A. Non-verbal sentences consisting of S.E. + Complement:

i. S.E. + Noun Complement, with or without I.P. or Demonstrative

as Prelude

ii. S.E. + Adjective or Participle Complement with or without I.P.

or Demonstrative as Prelude

6 -ye, -ya where the S.E. ends in a vowel, -e, -a where it ends in a consonant.
c-n where the S.E. ends in a vowel, zero where it ends in a consonant.


42

D. W. ARNOTT

B. Nominal groups having I.P. as initial item.

For in these constructions 1st and 2nd person S.E.s and Independent Pronouns
can occur in combination with adjectives, participles, nouns and specifiers, all of
which contain elements which clearly show the class to which they belong.

A.i. {Prelude +) S.E. + Noun Complement

11. When the S.E. is a 3rd person form, there is often agreement between the
S.E. and the noun, and where an Independent Pronoun or a Demonstrative occurs
initially as a Prelude, it also belongs to the same class as the S.E., e.g.

o laamcfo 7 he is a chief

hanko o laamcfo as for him, he’s a chief

o’o o laamcfo as for this one, he’s a chief

hangel/nge’el ngel baccel tan as for it/this one, it’s only a child
hange nge wiige as for it, it’s a heifer

nde’e nde Alkura’aaniire as for this (book), it’s a Koran

hangal ngal cofal as for this (hen), it’s a pullet

Very often, however, while the noun is in the appropriate class, the S.E., and
the Prelude, if any, is in the cfum class—the neuter class which is used rather than
one of the other non-personal classes where it is desired to be non-specific, to avoid
reference to any particular class, e.g.

hanjum) .

------> num
cfu’um J -—

deptere
nagge
mbonjokku
gertogal

as for

'a book
a cow
a bag
a. hen

As the nouns are in one of the specific classes (nde, nge, ngu, ngal in the above
examples), there is no formal agreement between the I.P./Demonstrative and S.E.
on the one hand and the noun on the other hand. It could be argued that in view
of the non-specific reference of the dum class, and the fact that a cfum-class form
can often be used in place of a form from one of the more specific classes, there is
not a complete breach of agreement. But agreement is a formal phenomenon,
and in these last examples there is not in fact any formal agreement between the
I.P./Demonstrative and S.E. on the one hand and the noun on the other hand.

12. We may also consider examples such as the following :

(hanko) o baccel

(hanko) o laamngel tan

(hanko/o’o) o harmaaru

(ham6e) 6e bacci tan
and even in a dependent clause

. . . saa’i o mbabba

(as for him) he’s a child

(as for him) he’s only a petty chief

(as for her/this one) she’s a prostitute
(as for them) they are only children

when he (was) a donkey

7 The concord elements marking the class are underlined in this and subsequent paragraphs.


FIRST AND SECOND PERSON PRONOMINAL FORMS IN FULA

43

Here the S.E. and I.P./Demonstrative are in one of the personal classes
(o singular, 6e plural), while the nouns are in other classes—ngel, ndu, cfi, nga
respectively. From these it is clear that cases can occur where formally there is
non-agreement in class between a 3rd person S.E. in a non-verbal sentence and the
noun Complement, though in these cases at least there is agreement in number.

13. In view of the foregoing, it is not surprising that a 1st or 2nd person S.E.
can be followed, in such a sentence, by a noun in any semantically appropriate
class. Most frequently the noun is in one of the personal classes, o or 6e, e.g.

(miin) mi moodibbo/kodo/Pullo (as for me) I’m a malam/stranger/Fulani

(minon) minmoodi66e/Ful6e (as for us) we’re malams/Fulani

a Pullo na ? are you a Fulani ?

But nouns in other classes are not infrequent, e.g.

miin mi laamngel non as for me, I’m just an insignificant little chief

aan a baccel tan as for you, you’re only a child

minon min mbaccon pamaron tan as for us, we are only little children

Similarly, in fables, where animals, birds, trees, etc., are made to speak, one
encounters such sentences as

naa mi mbabba (nga class), mi puccu I’m not a donkey, I’m a horse

A.ii. {Prelude +) S.E. + Adjective /Participle Complement

14. Where the Complement is an adjective or participle, however, there is a
distinction between 3rd person forms and 1st and 2nd person forms. A 3rd person
S.E. (and any I.P. or Demonstrative in the Prelude) and the adjective/participle
Complement are always in agreement, i.e. belong to the same class, e.g.

(hanko) o bumdo/nanaro
(hangel) ngel bumngel /nanarel
(hayru) ndu wumndu/nanardu
(hange) nge wumnge/nanare
(ham6e) 6e wum6e/nanar6e
(hankon) kon mbumkon/nanaron

naa ndu rewru, ndu wordu

(he) he is blind/refractory

(it) it is blind/refractory

(it) it is blind/refractory (e.g. dog)
(it) it is blind/refractory (e.g. cow)
(they) they are blind/refractory
(they) they are blind/refractory
(e.g. children, small animals)

it (e.g. dog) is not female, it is male

Where however the S.E. is a 1st or 2nd person form, the Complement, though
normally in one of the personal classes, may be in some other class. For instance,
one finds not only sentences such as the following

Class

o mi bumdo I am blind

o a nanaro you are incorrigible


44

D. W. ARNOTT

Class
6e on yidaafie you (pl.) are much-loved
6e (onon moodi66e) on hulniiBe (you malams) you are awesome
6e enen en Jeere’en we are of a different kind
but also
ngel wakkatiire nden mi pamarel no at that time I was small
(no = = formerly)
ngel (aan) a nanarel you are (an) incorrigible (little thing)

and

(aan) a gorko na a gorgel ? are you a man or a boy ? (lit. male

person or male child)

To this question there could be two answers, either mi gorko ‘ I am a man ’
(o class) or mi gorgel ‘ I am a boy ’ (ngel class). It must be said that in some similar
cases (e.g. mi pamarel ‘ I am small ’) my informants, while accepting the sentence
with an adjective Complement, would usually prefer to have a noun Complement
with an adjective dependent on it; e.g. they would prefer mi baccel pamarel ‘ I am
a small child ’. Nevertheless it is clear that with 1st and 2nd person forms there is
more latitude than with 3rd person forms.

B. Nominal groups with an Independent Pronoun in first position

15. The pattern of agreement can be seen most clearly and consistently within
certain types of nominal group, viz. appositional groups, in which an I.P. in first
position is followed by a noun, adjective, participle or specifier, or a combination
of these.

16. Where the I.P. is a 3rd person form, there is always agreement between it
and the following nominals, e.g.

hanko Pullo o’o
hangel baccel nge’el
hanji bacci di’i
hankon mbaccon ko’on
hange nagge nge’e
hangu puccu ngu’u
hanko pamaro /noddaado o’o
hangel pamarel / noddaangel nge’el
ham6e famarfie/noddaafje 6e’e]
hanji pamari/noddaadi di’i J
or, more simply

he, this Fulani

it, this little child

they, these children

they, these little children

it, this cow

it, this horse

he, this small one/this one that was called

it, this small one/this one that was called

they, these small ones/these that were called

hanko o’o; hangel nge’el, hanji di’i, hankon ko’on; hange nge’e, hangu ngu’u;
hambe 6e’e, etc.


FIRST AND SECOND PERSON PRONOMINAL FORMS IN FULA

45

17. When the I.P. is a 1st or 2nd person form, the other nominal is again
frequently in one of the personal classes, but it may quite readily be in one of the
other classes, the diminutive class being by no means the only non-personal class
so used, e.g.

Class
0 aan o’o you there ! (lit. you this)
0 aan Pullo o’o you Fulani there !
ngel aan nge’el you little one there !
ngel aan pamarel nge’el you tiny little one there !
ngum aan pamarum ngu’um you there, miserable little creature !
ndu hey aan ndu’u hi ! you there ! (dog)
ngu aan puccu ngu’u you horse there !
be onon 6e’e you there ! (pl.)
6e onon Ful6e 6e’e you Fulani there ! (pl.)
di onon pucci di’i you horses there !

Thus while with any given 3rd person I.P. the following nominal can only be
in the same class as the I.P., with any given 1st or 2nd person form there is a much
wider range of possibilities.

18. It may be added that such appositional groups can occur (a) in a Prelude
and be recapitulated by a S.E., or as an ‘ addendum ’ in apposition (b) to an O.E.
or even (c) to a Possessive Pronoun or Possessive Suffix, e.g.

3rd person

(a) hanko pamaro o’o, o-yamaaki

this small one is unwell

hangel pamarel pge’el, ngel-walaa innde

this tiny little one has no name

hangu ngu’u, ngu yaawngu

this one is fast

1st person

miin nyaawdo ni’i, mi-waawataa-dum

I, sick as I am, can’t manage it

2nd person

aan pamarel nge’el, ko mbaawu-daa?

you, so small as you are, what can you manage ?

enen nayeehon ni’i, en-mbaawataa dogguki

you and I, so old and shrunk as we are, we can’t run

onon moodi66e, on hulnii6e

you malams, you are awesome


46

D. W. ARNOTT

3rd person

(b) mi yirfaa-mo, hanko towdo o’o

I don’t like him, that tall one

mi-yidaa-ngel, hangel pamarel nge’el

I don’t like it, this tiny little one

mi-soodataa-ngu, hangu puru ngu’u

I won’t buy it, this dun one

2nd person

mi-yicfaa-ma, aan towdo o’o

I don’t like you, you tall one

mi-noddaayi-ma, aan pamarel nge’el

I didn’t call you, you tiny little one

3rd person

(c) mi-anndaa innde maako, hanko Pullo o’o

I don’t know his name, that Fulani

mi-anndaa innde maagel, hangel pamarel nge’el

I don’t know its name, that tiny little one

mi-anndaa innde jawmiigu, hangu puru ngu’u

I don’t know the name of its master, that dun one

2nd person

. faan Pullo o’o
mi-anndaa mnde maaaal — —— , ,

------[aan pamarel nge’el

T , , [you Fulani there

I don t know your name-r ..... r ..

J (you little fellow there

faan towcfo o’o
mi-anndaa innde jawmal —..— — , ,

- [aan pamarel nge’el

I don’t know the name of your master f^ou °J®

J [you little fellow

In all these cases there is regular agreement with 3rd person pronominal forms,
but variation of class is possible with 1st and 2nd person forms. Thus S.E.s,
O.E.s and Possessive forms are linked to the same system of agreement, or variation
in agreement, as appositional groups with an Independent Pronoun in first
position.


FIRST AND SECOND PERSON PRONOMINAL FORMS IN FULA

47

CONCLUSION

19. Of the three types of construction just examined, A.i. is inconclusive, since
there is considerable variation in the class possibilities with 3rd person as well as
with 1st and 2nd person pronominals. But in constructions A.ii. and B, while
3rd person pronominals enter into close one-to-one agreement with the other
nominals in question, the 1st and 2nd person forms do not enter into the same
one-to-one agreement, particularly in B. They are in fact not tied to individual
classes in the same way as 3rd person forms. Taking account, therefore, of their
syntactical behaviour, as well as their morphological distinctiveness, discussed in
§§ 6-9, we may conclude that it is appropriate to treat the 1st and 2nd person forms
as being outside the class system, rather than within it.

BIBLIOGRAPHICAL REFERENCES

1966. D. W. Arnott, ‘Nominal groups in Fula’, Neue Afrikanistische Studien (Deutsches
Institut fur Afrika-Forschung, Hamburg—Klingenheben Festschrift), 40-61.

1970. D. W. Arnott, The nominal and verbal systems of Fula. Clarendon Press, Oxford.


THE PROBLEM OF CLASS IN KIKUYU

By Patrick R. Bennett

A principal characteristic of Bantu languages is the ‘ class system The
typically Bantu concord-based system has parallels in many non-Bantu languages.
It may further be considered an analogue (though greatly expanded) of grammatical
gender in certain languages. In all these cases, there exist various sets of nouns.
Such sets are normally, though not invariably, distinguished by overt markers.
Similar markers usually occur with other parts of speech, such as adjectives,
verbs, pronominals, etc., and the choice of marker for each part of speech
corresponds to, or is determined by, the set to which the noun indicated belongs ;
that is, the noun’s gender or class.

In most languages showing such a system, the gender or class of a noun, and
the behaviour of items agreeing or showing concord with it, can be predicted
to some extent from the form or meaning of the noun. Only in very few, if indeed
any, is this completely predictable. In some the number of distinct forms of
adjectival, pronominal, and verbal markers is identical with the number of noun
sets. In others, however, there is no such one-to-one correspondence, and even
where it does exist the forms of the affixes indicating gender and concord reference
often show great differences. In many languages nouns having formal character-
istics of one gender or class show the concord behaviour of another. Even in
cases where there is correlation between class affiliation and meaning, exceptions
exist, names of female beings belonging to masculine or neuter genders, for example.

Besides such anomalies, in Bantu languages, because of the fairly large number
of classes distinguished, there arises also the problem of establishing the various
classes and the inventories of each. The main problem in establishing what classes
exist is the decision whether to consider a class to consist of a singular-plural pair,
or to treat each set of nouns distinguished by prefix and concord behaviour as
a separate class. The latter interpretation has in its favour the fact that the
number of plural forms so distinguished is usually smaller than the number of
singulars. One thus does not double the number of classes by accepting this
interpretation, but rather reduces the number of times one must discuss the formal
and syntactic behaviour of a given plural type. Though both treatments are in
common use, the latter is somewhat more frequent, though for pedagogical
purposes the former may be more popular and perhaps practical. One reason
for the more frequent choice of the latter is the force of tradition. The system of
labelling Bantu classes first set out by Bleek (though often attributed to Meinhof)
has, with very few modifications, been used as standard in comparative treatments,
and even in discussions of single languages, partly to facilitate comparative
interpretation of the class system described, partly as a conveniently pre-existing
and generally acceptable system of numeration.


THE PROBLEM OF CLASS IN KIKUYU

49

The principal objection to the Bleek class numeration is also the principal
hindrance to determining the inventory of noun classes : there are very few, if any,
languages where there is a simple and direct connection between nominal form,
concord behaviour, and traditional classification. The most commonly used
method of determining class affiliation seems to be based on three criteria. If a set
of nouns can be found whose prefix and inventory seem etymologically connected
with those of a Bleek-Meinhof ‘class ’, any noun, regardless of its prefix, showing
identical concord is assumed to belong to the traditional ‘ class ’ in question. If,
however, a set of nouns is found whose prefix can be identified with a traditional
‘ class ’, it is so identified, even if identical in concord with another set as defined
above. Again, if there exists a set of nouns lacking any trace of a prefix, it will
be treated separately and given a class number such as ‘ la ’, the number being
determined by that of the traditional class with which the set agrees in concord.
Though there is much variation in practice, this represents fairly accurately the
usage of most of those who employ Bleek-Meinhof numeration, and approximates
the basic principles of many analyses not following this tradition.

The result is an analysis which is useful to the comparativist, but which is
awkward and leaves many questions unanswered in synchronic description.
This paper will attempt to illustrate some of its shortcomings and suggest a type
of analysis which, though still considering the interests of the comparativist,
more adequately represents the synchronic facts of the language and its differences
from, as well as similarities to, the general Bantu system. This will be accom-
plished by describing the class system in Kikuyu, a Thagicu dialect 1 spoken in
central Kenya. At first sight, the Kikuyu system seems simple, but there are
several complexities.

One must first distinguish between concord and class. Though there is a
correlation, it is not exact. Herein, ‘ class ’ will refer to a set of nouns formally
and functionally distinguished in ways to be discussed. The term ‘ group ’ will
be used of a set of nouns governing identical concords, whatever the formal
properties of these nouns. A maximum of fourteen such groups can be established
in Kikuyu, each governing specific prefixes of three main series (each with sub-
divisions), namely adjectival, pronominal, and verbal. No such series, however,
shows fourteen formally distinct prefixes. The adjectival shows 12, the pro-
nominal 2 shows 13, and the verbal 12. The figure 14 is reached by considering
the set of three prefixes, not the individual items, and overlooking overlaps. Thus,
when some nouns with adjectival prefix N- govern a pronominal prefix I-, while

1 For definition, see my article ‘ Dahl’s Law and Thagicu African Language Studies,
VIII, 1967. The orthographic conventions here agree with those followed there, except that
tone will only rarely be marked, as it is for the most part irrelevant to this problem. In the
tables, only those few items showing low tone will be marked.

2 Though there are various forms in each of the three series, the pronominal shows the
most variation. Here, as in all cases, the preconsonantal form of each prefix is given, as being
the simplest.

E


50

PATRICK R. BENNETT

others take i-, and some nouns whose pronominal prefix is I- take not N- but mi-
as adjectival prefix, one recognizes two morphemically distinct (though formally
identical) prefixes of shape N-, and two morphemically distinct prefixes (with
identical shape) of form i-, and three distinct concord groups. Columns I and II
of Table I show the three basic concord prefixes of each group, together with a
capital letter designating the group and a number indicating the traditional
‘ class ’ it represents. In all cases the latter is determined by the form of the
adjectival prefix, except that where overlaps in adjectival prefixes occur the
pronominal concords are considered. It will be seen that, though ‘ classes ’ 1
through 16 are found, 8 and 14 are lacking. This results from merger of 8 with 10
and 14 with 3, involving phonologic shift in pronominal and verbal prefixes,
identification of the two classes in each case, and loss of one adjectival prefix.
Had this not occurred, one would find two additional groups, 8 with *i-/i-/i-
(all alternating with ci-), and 14 with *u-/u-/u-.

In Table I, groups A and C show identical adjectival prefixes, as do H and I.
Group D shares pronominal and verbal with H. The verbal prefix of B is identical
with that of F. Groups A, B, C, D, E, H, and I show adjectival prefixes differing
formally from the other series. Groups A and B show different pronominal and
verbal. Groups C, D, H, and I show similar though minor differences, indicated
in the notes to Table I. It seems simplest to ignore such overlaps and differences
as irrelevant to discussion of the concord system. Parallel differentiations exist
in many Bantu languages, and many formal overlaps are due to phonologic shifts
and analogic extensions. It seems most reasonable and simplest to treat the three
prefix-forms as representatives of a single morpheme in each group, with one-to-
one correspondence between groups and morphemes.

The problem of establishing classes, and the number of morphemically distinct
nominal prefixes, is rather more difficult. Here one must at least consider the
factors of form, concord, and singular-plural pairing. Derivation and semantic
factors might also be considered. However, at least in Kikuyu derivation seems
not to affect the form or general behaviour of a noun significantly, and cannot
seriously be thought to determine class affiliation. While meaning in some cases
could be said to correlate with class affiliation, it is by no means a one-to-one
correspondence. All nouns in Group A (‘ Class I ’) refer to humans, and the
majority of nouns referring to humans belong to this group. However, nouns
referring to humans are also found in at least groups C, E, G, H, and L, with
plurals in B, D, F, I, and K. The correspondence in other semantic areas is
even less complete. This therefore cannot be taken as a criterion.

As Columns III, IV, and V of Table I show, 49 sets of nouns may be distin-
guished on the basis of the above three factors. This does not mean 49 mor-
phemically distinct ‘ classes ’ must be recognized, but only that no more than 49
can be distinguished by these criteria. If one considers only nominal prefix form,
there are 18 distinctions. Concord grouping alone gives 14. Prefix and singular-


THE PROBLEM OF CLASS IN KIKUYU

51

plural pairing give 25. A different set of 25 distinctions is given by the combination
of prefix and concord group. None of these corresponds to the traditional
comparatively based division of classes, where the maximum number of distinct
‘ classes ’ is 22.

While obviously the traditional figure of 22 is unrealistic, it is equally obvious
that the extremes of 14 (concord groups) and 49 (maximum differentiation) do
not perfectly fit the structural situation. To revise these figures, one must establish
criteria by which to determine morphemic affiliation and hence class inventory.
A few principles are basic to determination of morphologic structure. One of
these principles requires interpretation of two directly contrasting items as
structurally distinct, where free variation and conditioning do not exist. Another
allows interpretation of free or conditioned alternants as members of the same
structural unit, even if dissimilar. These, however, do not suffice, and one must
decide which other factors in the analysis are of primary importance.

It has already been decided to treat the groups of concord prefixes as
morphologically distinct, despite formal overlaps between groups and differences
within groups. Because of this, and because determination of concord may be
said to be the primary, if not sole, function of class, it seems best to treat concord
grouping as one of the main factors in class division. Thus two sets of nouns
showing different concord, though formally identical, must be assumed to show
morphologically distinct, because functionally contrasting, prefixes.

Such an analysis is supported to some extent by the fact that in all groups at
least one set of nouns shows a prefix identical in form with the group’s adjectival
prefix. These are Aa and Ab ; Ba ; Ca ; Da ; Ea, Ed, and Ee 3; Fa, Fb, Fc,
Fd, Fe, Ff, Fg, Fh, and Fi; Ga, Gb, and Gc ; Ha ; Ie, Ig, Ih, and Ii ; Ja, Jb,
and Jc ; Ka and Kb ; La and Lb ; Ma, Mb, and Me ; and Na. Since the formal
identity of the group A and group C adjectival prefixes, for example, is interpreted
as formal overlap of distinct morphemes, the formal identity of the corresponding
noun prefixes may be similarly interpreted. If one considered that the number of
distinct forms, rather than the number of distinct groups, determined the number
of morphologically distinct adjectival prefixes, this analysis would be more
difficult to defend.

The part pluralization should play in determining class affiliation is hard to
decide. Clearly this is an important factor in the noun system. In practice, it is
often easier to determine the group of a noun with prefix mu- by asking for its
plural than to elicit a verbal subjective prefix. The irregularities of stem in Gc/Id
and Jb/Ih must be described, though they in no way affect concord. The
difference in plural formation between Eb and Ec, where the plural of the latter
includes the singular prefix, must also be described. How relevant is this to ‘ class ’ ?

3 Though Ea, Ed, and Ee are not formally identical with one another, all three may be
included here because of the double form of the adjectival prefix, i- before consonant-initial
stems, ri- before vowel-initial.


52

PATRICK R. BENNETT

Table I

I a II b c III IV a V b VI
a b c
A(l) mu- fl- a- 1 a(l) mu- S Ba muturi smith 2 00 3
b(l) mu- S la mwana child 2 1
c(la) 0- S Bb ithe father 4 ‘ 10’8
B(2) a- a- ma- a(2) a- P Aa aturi smiths 2 co
b(2a) ma- P Ac maithe fathers 4 ‘ 10’
C(3) mfl- il- 0 u- a(3) mu- S Da mugunda garden oo
b(14) u- S Fa uta bow 7 00
c(14) gu- S Fb guoya fur 2 8
d(14a) 0- S Fc hinya strength 00
D(4) mi- i- i-8 a(4) mi- P Ca migunda gardens 00
E(5) i-10 ri- ri- a(5) i- S Fd itimu spear11 oo
b(5) ri- S Fe riitho eye 12 ‘ 10 ’
c(5) ri- S Ff riiko hearth 12 ‘ 20 ’
d(5) ri- S Fg riithori tear 12 00
e(5) ri- S Fh riitwa name 12 ‘ 10 ’
F(6) ma- ma- ma- a(6) ma- P Cb* 13 moota bows 7 00
b(6) ma- P Cc* maguoya furs, wool 2 8
c(6) ma- P Cd11 mahinya strengths 00
d(6) ma- P Ea matimu spears 11 00
e(6) ma- P Eb maitho eyes 12 ‘ 10 ’
f(6) ma- P Ec* mariiko hearths 12 ‘ 20 ’
g(6) ma- P Ed maithori tears 12 oo
h(6) ma- P Ee* mariitwa names 12 ‘ 10 ’
i(6) ma- P Ma magflru legs 3 18
G(7) ki- ki- ki- a(7) ki- S lb kirima hill11 oo 10
b(7) ki- S Ic kiura frog 12 oo 16
c(7) ki- S Id kindu thing 1
H(9) N-17 i- i-8 a(9) N- S Ie ndurume ram 00
b(9a) 0- S If bengi bank 00 18
1(10) N- i-18 i_ 20 a(8) ci- P Ab ciana children 2 1
b(8) i- P Ga irima hills 11 oo 18
c(8) ci- P Gb ciflra frogs 12 00 16
d(8) i- P Gc* indo things 1
e(10) N- P Ha ndurume rams 00
f(10a) 0- P Hb bengi banks oo 18
g(10) N- P Ja ndigi strings oo
h(10) N- P Jb* ninimbi flames 1
i(10) N- P Jc* nduui rivers ‘ 10 ’21
J(ll) ru- ru- ru- a(ll) ru- S Ig rurigi string oo
b(ll) ru- S Ih ruririmbi flame 1
c(ll) rfl- S Ii rufli river ‘ 10’21
K(12) tu- tu- ffl- a(12) tfl- P La twana small children 22 00
b(12) tfl- P Lb tumiti small trees 23 ‘ 20’
L(13) ka- ka- ka- a(13) ka- S Ka kaana small child 22 00
b(13) ka- S Kb kamflti small tree23 ‘20’


THE PROBLEM OF CLASS IN KIKUYU

53

Table I (contd.)

I II III IV V VI
M(15) ku- kfi- ku- a(15) ku- S Fi kfiguru leg 3 15
b(15) kfi- S 24 kfirima to cultivate 26 oo
c(17) ku- P Na kfindu area 1
N(16) ha- ha- ha- a(16) ha- S Me handu place 1

EXPLANATORY

NOTES TO TABLE I

Meaning of the vertical columns

I Designation of Concord Group
(traditional number in parentheses)

II a Adjectival Prefix

b Pronominal Prefix
c Verbal Subjective Prefix

III Designation of Noun Set (traditional
number in parentheses)

IV a Prefix of Noun Set

b Number of most nouns in set
(S = singular, P = plural)
c Designation of partner in singular-
plural pairing

V a Example
b Gloss

VI Estimated number of nouns in set

Footnotes to table i

1 The subjective prefix has a variant fi- used in relative constructions ; the objective prefix
is -mfi-.

2 All nouns in these sets refer to human beings, though not all such nouns are in these sets.

3 This indicates the set is large and capable of expansion.

4 All words for human beings, mostly kinship terms.

5 Numbers in quotation marks are approximations only.

6 This has the form gu- in certain cases when preceding o.

7 Many abstract, often adjectivally derived, nouns included.

8 The two nouns differ only in tone : gudyd ‘ fur ’, gu6y& ‘ fear ’.

9 The objective prefix has the form -mi-.

10 This occurs with consonant-initial stems only ; with vowel-initial stems ri- occurs.

11 Only consonant-initial stems are included.

12 Only vowel-initial stems are included.

13 The asterisk indicates the prefix of the plural is not added to the stem of the singular,
but to some other form.

14 It is a matter of subjective interpretation whether one considers this added to the stem
(ma+hiuya) or to the singular including prefix (ma + 0+hinya). This is true also of Bb.

15 All body parts ; the other two are guoko/mooko ‘ arm(s) ’ and gfitu/matu ‘ ear(s) ’.

10 These include many derogatory augmentatives, some retaining the prefix of the singular
base form.

17 This represents a homorganic nasal, in some cases realized only as a lengthening of
preceding vowels.

18 Only very recent loan-words.

10 Preconsonantal only. Before vowels a prefix ci- occurs.

20 The subjective prefix behaves as in note 19 ; the objective shows only -ci- before vowels,
and -ci- and -i- as free alternants before consonants.

21 These often have alternative plurals in Ig.

22 Mostly diminutives, many including the prefix of the singular base form.

23 All diminutives, retaining the prefix of the singular base form in the singular, and that
of the plural base in the plural.

24 No corresponding plural ; pluralization is adverbial only (‘ to cultivate many times ’,
rather than ‘ many cultivations ’).

25 Verbal nouns or infinitives only.


54

PATRICK R. BENNETT

In Kikuyu, as in other Bantu languages, pluralization is comparatively easy
to predict. The only formal change is usually substitution of one noun prefix for
another, with corresponding change in concord group. Except for group M
(which will be discussed) each group is exclusively singular or plural, when
collectives are not considered, and most nouns in one group will have a plural
in a specific one of the other groups. In some descriptions of Bantu languages
the concept of what is called ‘ gender ’ or ‘ series ’ has been evolved (this will
here be termed ‘ pairing ’, to avoid possible misinterpretation). A ‘ gender ’
or ‘ pairing ’ is a set of noun stems belonging to one specified class and group
in the singular and another in the plural. Such analyses are far from new, and
many descriptions equate pairing with class, each ‘ class ’ thus including one
singular form and one plural. In this type of analysis Kikuyu would have only
14 ‘ classes ’, as compared with the traditional Bleek-Meinhof 22. If strictly
interpreted, however, irregularities such as kindu/indo ‘ thing(s) ’ would have to
be given the status of independent pairings, giving 24 basic pairings (half the
49 sets in Table I, minus the unpaired Mb). In addition, one would be forced to
recognize what have been called ‘ one-class genders ’, sets of unpaired collective
or mass nouns ; as these are found in at least 18 of the noun sets listed, the
description would clearly be rather cumbersome and involve much redundancy.
If pairings are to be described at all, it seems best to consider them separate from
class, just as classes must be distinguished from concord groups.

Another argument for this separation is the similarity of pluralization in
Kikuyu to various types of derivation. Set Cb contains many derivative abstracts.
Group G includes a large number of pejorative augmentatives, and group L is
composed almost exclusively of diminutives. In the last two cases (and their
corresponding plurals) there are many instances of retention in the derivative of
the prefix of the base, exactly parallel to the situation in Fa or Ii. The processes
of derivation and pluralization are formally and functionally parallel. In each,
substitution or addition of a prefix is accompanied by semantic change and shift
in group. Even somewhat similar irregularities of formation exist. If derivation is
considered irrelevant to class, pluralization should be so considered as well.
While both are important in the total description of the nominal system, they do
not affect concord in any way. A noun in group F, whether plural or collective,
whatever its singular, if any, whether the singular prefix is present or absent,
whether the plural of a derivative abstract or otherwise, behaves exactly as do all
other nouns of group F. Moreover, most modern works seem to imply that the
pairing is a set of classes somehow connected, rather than a class itself. It seems
best here to consider a class to be a set of nouns of a given concord group showing
a given prefixed morpheme, pairing being irrelevant. It remains to determine
the morphemes in question.

In cases such as group F, there is no obvious reason to recognize more than
one prefix morpheme. Despite differences in pairing and occasional retention of


THE PROBLEM OF CLASS IN KIKUYU

55

singular prefixes, there is but one prefix form and one type of concord. In group I
one finds a case of complementary distribution. The prefix i- (Ib/Id) occurs only
before consonants, ci- (Ia/Ic) only prevocalically.4 These may therefore be
considered to belong to the same morpheme. Both alternants, however, contrast
with N- (Ie/Ig/Ih/Ii) and 0- (If), which contrast with one another in turn. One
therefore considers group I to include three morphemically contrasting classes,
with prefixes i-/ci-, N-, and 0-.

Complementary distribution occurs also in E, but here interpretation is
more difficult. Again, i- (Ea) is preconsonantal only. Both ri- (Eb/Ec) and ri-
(Ed/Ee) are prevocalic. These two contrast directly, and only one can be
morphemically linked with Ea. Comparative evidence and inventory suggest
that the form ri- is original (as well as formally closer to i-) and ri- arose on the
analogy of the pronominal and verbal prefixes. The more productive and closer
to the prevocalic alternant of the adjectival prefix is ri-, however. For these
reasons, though the decision is essentially arbitrary, one will recognize here the
morphemes i-/ri- and ri- as separate.

Complementary distribution once existed in C. The prefix gu- (Cc) arose as
a regular phonologic variant of u- (Cb) before the vowel o. Today, however,
nouns such as woru ‘ rottenness ’ exist, showing the normal prevocalic form of u-.
The two must be recognized as morphemically separate. As in the case of ri-
(Eb/Ec), and for much the same reasons, the historically older form is treated as
separate, to produce a simpler statement in view of the greater productivity of the
newer form.

It is difficult to decide the best treatment of the prefixless classes Ac, Cd,
Hb, and If. These are clearly morphologically distinct from other sets in their
groups, and hence constitute separate classes. However, as they lack prefixes,
can they said to be morphologically distinct from one another? It is in many
ways unsatisfactory to treat nothingness, 0-, as a morpheme, a unit in the
structure of the language, yet there seems no other way to distinguish these
classes morphologically. One is forced to recognize them as distinct on the
basis of concord, but there is no overt or covert structural indication of this
difference.

Group M poses the most difficult problem. The three noun sets are formally
identical. They differ only in meaning and pairing, neither of which is allowed
to affect determination of class. There is no formal reason for recognizing more
than one class. However, the native speaker rejects completely the treatment
of these sets as a single class. The group has been variously interpreted by
investigators of other Bantu languages, in some of which formal differences are
in fact found. Some label all three ‘ Class 15 ’. Some label Ma and Mb 15,

1 Note the parallel to the similar conditioned alternation in pronominal and verbal concords
of this group.


56

PATRICK R. BENNETT

separating Me (‘ Class 17 ’). Barlow 5 treats Mb as a sub-class, ‘ 15a and Ma as
‘ Class 15 ’ ; Westphal6 has called Mb ‘ Class 15 ’ and Ma ‘ Class 15a Ma
includes three names of body parts, and differs in no way from other noun sets.
Mb includes all verbal infinitives. These have certain verbal characteristics 7
and take no plural. Me includes one noun, considered plural 8 and referring to
place. If one recognizes a single class, one must overlook these differences, which
are rather greater than those between subdivisions of other classes, and reject
the analysis of the native speaker ; the difference in number is an especially great
stumbling-block. To recognize three formally indistinguishable classes with
morphemically distinct prefixes, however, is objectively indefensible. Here the
three will be treated as different classes, but it must be realized that this is basically
subjective.

One class of each group is morphemically linked with the concord prefixes of
that group. Normally, only one class shows any formal similarity to the adjectival
prefix of the group, and that class will be assumed to have the same prefixed
morpheme. Group M is again a problem, as all three classes resemble the concord
prefixes. Here only Ma will be assumed to be morphemically identical with the
concord prefixes. This decision is completely arbitrary.

A few problems remain. In most classes are found certain nouns having a
prefix with long, not short, vowel : muugikuyu ‘ Kikuyu person ’, aagikuyu
‘ Kikuyu people ’, giigikuyu ‘ Kikuyu language ’ (all from Gikuyu, name of the
tribe’s ancestor) ; kiiriu ‘ modern slang ’ (from riu ‘ now ’) ; maai ‘ water ’,
tuui ‘ little water ’, muuriu ‘ son aariu ‘ sons ’, etc. This lengthening seems
clearly conditioned by the stem ; in cases such as the first four examples, this
may be a feature of the type of derivation, in others such as maai it may be
phonological, in some no obvious cause can be determined. In any case, there
seems no reason to recognize separate classes or morphemes here. The other
irregularities in derivation and plural formation must be described as well. How-
ever, on the bases established above, it is possible to determine the inventory of
classes with little difficulty.

The best system of labelling must now be considered. As stated, the traditional
system is inadequate. However, it is too well established to be easily replaced
in comparative studies, and the existence of a system facilitating comparison of

5 A. Ruffell Barlow, Studies in Kikuyu grammar and idiom, William Blackwood and Sons,
Edinburgh, 1960, 14a.

6 E. O. J. Westphal, ‘ Oluijhkumbi Vocabulary ’, African Language Studies, II, 1961, 51.

7 They are, for example, capable of taking objects, and indeed subjects, as well as adverbs,
and may to some extent be inflected for tense.

8 The ‘ plurality ’ here is far from the English speaker’s idea of plurality. The difference
between handu (place) and its ‘ plural ’ kundu (place, area) is not so much in number as in size.
The singular may occur with numerals higher than one : handu hatatu ‘ three places ’. The
plural may occur with the numeral ‘ one ’ : kundu kumwe (one place). This being so, one
assumes that no true singular-plural relationship exists. There is, however, a relationship
which is sufficiently similar to that between singular and plural to convince the native speaker.


THE PROBLEM OF CLASS IN KIKUYU

57

Table II

Class number Prefix form Noun sets included
01.010a mu- Aa, Ab
01.011a 0- Ac
02.020a a- Ba
02.021a ma- Bb
O3.O3Oa mu- Ca
03.140a u- Cb
03.140b gu- Cc
03.141a 0- Cd
04.040a mi- Da
05.050a i-/ri- Ea, Ed, Ee
05.050b ri- Eb, Ec
06.060a ma- Fa, Fb, Fc, Fd, Fe, Ff, Fg, Fh, Fi
07.070a ki- Ga, Gb, Gc
09.090a N- Ha
09.091a 0- Hb
10.080a i-/ci- la, lb, 1c, Id
10.100a N- Ie, Ig, Ih, Ii
I0.101a 0- If
11.110a ru- Ja, Jb,Jc
12.120a tu- Ka, Kb
13.130a ka- La, Lb
15.150a ku- Ma
15.151a ku- Mb
15.170a ku- Me
16.160a ha- Na

class systems is so convenient that use of an independent system in description
of Kikuyu is undesirable, however more suitable such a system might be. A few
changes, however, are necessary.

The traditional system, while in many ways unsatisfactory, would be acceptable
in Kikuyu but for one thing. In two cases, an original prefix morpheme has given
rise to two distinct morphemes, these being gu- (Cc) from u- (Cb), and ri- (Eb/Ec)
from i-/ri- (Ea/Ed/Ee). The sets of nouns with these prefixes are not parallel
to those traditionally distinguished by a following ‘ a ’ (la, etc.). However, since
they have developed historically from the morphemes of ‘ Class 14 ’ and ‘ Class 5 ’,
respectively, assignment of new numbers would hinder, rather than help, com-
parative study. To indicate the same type of relationship Tucker 9 has labelled
one such morpheme ‘ 5a ’, for example, another ‘ 5b ’. While this is extremely
satisfactory as a means of distinguishing such items, confusion with the more
traditional derivative labels (la, 2a, etc.) can easily arise.

Further, while there is in Kikuyu a morpheme equivalent to the traditional

A. N. Tucker, ‘ Notes on Konzo African Language Studies, I, 1960, 20.


58

PATRICK R. BENNETT

‘ Class 8 for example, no separate concord group exists. ‘ Class 8 ‘ Class 14

and some others are represented by noun prefixes only. It therefore seems best
to include in the class label indication of the group to which the class belongs.
The system shown in Table II is proposed as an adequate account of the Kikuyu
class system, and suggested as a model for similar revisions of the class labelling
of other languages. Each label gives first the traditional number of the concord
group. This is followed by a modification of the traditional ‘ class ’ number of
the noun prefix morpheme, ‘ Class 1 ’, ‘ Class la ’, and ‘ Class 14 ’ being written
010, Oil, and 140, respectively. To this is added a lower-case letter, which allows
one to distinguish Cb, Cc, and Cd as 03.140a, 03.140b, and 03.141a, respectively,
adequately indicating the difference in relationship. Were it not for the desire
to retain the advantages for comparative discussion of the traditional system,
a much simpler system could be developed, but as it is this must suffice.

A few notes should be given on uses of the concord prefixes other than those
described. Slips in concord are possible and indeed frequent. They can generally
be explained as due to association with a semantically equivalent noun in another
group, and substitution of its concord prefixes. More regular is the use of
groups A and B with nouns used as personal names, regardless of prefix. This is
a clear extension of the use of these with classes including only words for human
beings. The non-adjectival concords of groups C, E, M, and N are often used
adverbially, C of manner, E of time, and M and N of place. In C this may be
due to the presence in class 03.140a of many abstract nouns derived from adjectives
which, like most nouns, may be used adverbially. For example, wega may mean
‘ goodness ’ or ‘ well ’. This usage has been extended to pronominal and verbal
concords. The similar use of E might be explained by the presence of ihinda
‘ time ’ in 05.050a. However, comparative evidence shows that this usage (and,
indeed, that of C) derives from a far earlier period in the history of Bantu. While
it may be that various time-words were found in this class in Proto-Bantu, this
cannot be proven, and other explanations are possible. In M and N the locative
reference is clearly connected with that of classes 15.170a and 16.160a. Here,
though, the nominal usage might be derived from the adverbial, to judge by the
comparative evidence. The subjective verbal prefixes of M and N are also used,
as in many Bantu languages, to express impersonal subjects. Compare the English
‘ there was a man ’ with Kikuyu ni kwari na mundu. Finally, it should be noted
that the non-verbal concords of any group may be used in the first or second
person to refer to the person(s) or thing(s) addressed or speaking, though special
verbal prefixes are used in such cases.

While it might be desirable to discuss semantic correlations of the classes,
this is impossible, in view of the limited scope of this paper. Where such exist,
they seem not to affect division into classes, though of course discussion of such
matters would be necessary in a complete analysis of the nominal system. As in
other class systems, each class includes at least one set of nouns with common


THE PROBLEM OF CLASS IN KIKUYU 59

semantic characteristics, but each also includes enough exceptions to shatter any
generalization.

In the preceding pages the class system of Kikuyu has been discussed and the
inadequacies in this context of the traditional Bleek-Meinhof class numeration
shown. By distinguishing concord group from noun class, and rejecting pluraliza-
tion, derivation, and semantic features as primary criteria, it has been possible
to arrive at a morphologic analysis which seems adequate. It is true that it is
not the only adequate analysis, and that it is perhaps not the most satisfactory
from some points of view. However, of the objectively equal alternatives this
seems best on subjective grounds.

This analysis having been reached, a revised system of numeration was
proposed for Kikuyu. While adequate for Kikuyu, this permits easy comparison
with traditionally-based numeration systems in other Bantu languages. It thus
retains the comparative advantages of the uniform, though inadequate, traditional
system, while far more accurately describing the individual system.

It is suggested that if similar analyses were made of other Bantu class systems,
and similar re-evaluations and reorganizations of the labelling systems undertaken,
it might be possible to arrive at a more satisfactory uniform system of class
numeration and description. Ideally, it seems that the traditional system, which
was based on knowledge of a number of languages far smaller and far less diverse
than the number known and described today, should be replaced. However, this
seems improbable, at least at present.


A NOTE ON KRIO TONES

By Jack Berry

The suprasegmental phonemes of Krio have received little attention in recent
descriptions and about them there still appears to be a strong conflict of opinion.1
In a paper presented to the first Conference on Creole Languages in Jamaica some
ten years ago, I gave a brief indication of what I then believed to be the phonemic
structure of Sierra Leone Krio.2 The topic of the paper and the circumstances in
which it was given precluded detailed consideration of any single aspect of Krio
phonology but I was at pains to adduce evidence I still believe adequate to support
the claim that Krio is a tone language (albeit of a type in which word-related tone
and sentence-related intonation conflict in many and special ways).3 The present
paper seeks to enlarge somewhat on my earlier brief account of the Krio prosodies.

The simplest statement is achieved by recognizing two sets of prosodically
relevant features. Using P and T as cover-symbols for pitches and terminals,
respectively, the full inventory of suprasegmental phonemes in Krio is

P /l, 2, 3/

T /unmarked, #, ||, =/

P(itches) All syllables have pitch. There are three contrasting levels : low/l/,
mid /2/ and high /3/. Transitions to and from these levels give
simple and compound glides. As many as 5 pitch phonemes may
occur in a single syllable : syllables with falling-rising-falling-rising
or rising-falling-rising-falling tone seem about as common, in fact,
as those with level or simple gliding tone.

T(erminals) Internal juncture is not considered in this paper. Other kinds of
transition phenomena associated with pause, however, require some
mention, since in nearly every instance the pitch of the preceding
syllable is affected.

Tx Yes-or-no questions without exception in Krio end in /||/, that

1 In the Introductory course in Krio prepared for the Peace Corps at Indiana University in
1964, for example, the authors neglect tone almost entirely—at some cost, be it said, to their
grammatical descriptions, especially perhaps in their statement of the verbal system.

Some would even deny that Krio is a tone language. In a recently republished paper ‘ Pro-
nunciations of English in West Africa ’, Professor Strevens, for example, suggests that ‘ the
tonal system of Yoruba has disappeared [from Freetown Krio] but a sentence-stress and
intonation pattern broadly like that of Received Pronunciation is present ’ (Papers in language
and language teaching, O.U.P., 1965).

2 The paper was subsequently published as ‘ English loanwords and adaptations in Sierra
Leone Krio ’ in Creole Studies, Vol. II, London, 1961.

3 As well as Africanisms like kanga (‘ magic ’ and/versus ‘ type of dried fish ’) and oka
(‘ gaboon viper ’ and/versus ‘ hydrocephalitis ’), I also cited pairs of words (of unambiguous
English derivation) minimally distinguished by tone.


A NOTE ON KRIO TONES

61

is, with the pitch at first level then rising abruptly to an indeterminate
end-point. As far as I am aware, this terminal occurs only as a
feature of one type of interrogative sentence.

In simple indicative sentences, when pause of whatever length
follows, it is possible to distinguish three types of ending as follows :
T2 (Unmarked): Phonation ceases gradually ; preceding syllable has
falling, rising or level pitch.

T3 /#/ Phonation ceases abruptly ; pitch of preceding

syllable starts level then falls rapidly to well below /l/.

T4 / = / Phonation ceases gradually ; preceding syllable is

drawled and the pitch, if a glide, levels off towards the end at /l/,
/2/ or /3/.*

Examples of the terminals are :

Tx 1 2 /nabrus||/ Is it Bruce ?
t2 1 2 /nabrus/ It’s Bruce
t3 1 2 /nafis #1 It’s a fish
t4 1 2 /nabrus =/ It’s Bruce, [but 1 —]
1 23 /nabrus =/ It’s Bruce, [John and —]

Tx and T4 are unambiguously sentence features {query and comma)
but T2 and T3 are to some extent determined lexically since in certain
sentences, where they occur finally, some words ‘ end in ’ T2, others
in T3, though all words in given circumstances may ‘ end in ’ in
either terminal.4 5

By way of further exemplification of the pitches, transitions between pitches,
and T3,1 give in the table below 63 ‘ tunes ’ of the verb ‘ go ’ in its simple imperative
form. The citation-form of this verb is that labelled normal in the inventory of
intonation at the end of this paper (i.e. /go 212 #/) ; the remaining other 62
patterns available to me at this time from my field-notes are therefore considered
to carry a variety of differential meanings which I have tried in each case to suggest
either by rough contextualization (translations of the rest of the utterance in which

4 Distributionally, terminals are by definition restricted to pre-pause position ; gliding
tones are not. Phonetically, Tj and T3 are distinguished from the somewhat similar glides
/12, 13, 23 ; 31, 32, 21/ by differences in the point of onset of pitch change (middle or end of
syllable v. beginning), in the degree of abruptness of the ‘ rise ’ or ‘ fall ’ (‘ quick ’ v. ‘ gradual ’)
and in the end-point (indeterminate v. /I/, /2/ or /3/).

5 A statement of these circumstances is beyond the scope of the present paper and is there-
fore reserved for a separate publication at a later date.


62

JACK BERRY

they occur in the corpus of texts) or by a brief descriptive comment by native
speakers of Krio identifying the meaning of particular intonations for them.

It only remains here to add that of the 67 ‘ possible ’ patterns of pitch given in
the table, almost all occur in either the ‘ normal ’ or ‘ shifted ’ forms of verbs as
3 1

well as in the citation-forms of nouns. The verb /kam/ ‘ come for example,
exhibits the same range of pitch patterns associated with the same differences of
meaning as the verb /go 212 #/ except, of course, that the individual correlations
of pattern and meaning are different : 31 ‘ go’ (reassuring), ‘ come ’ (normal) ;
312 ‘ go ’ (insulting), ‘ come ’ (coaxing) ; 231 ‘ go ’ (I’m glad for you, — ’),
‘ come ’ (strong command). . . .

INVENTORY

3 ‘ I know you too well, — ’ 2 ‘ I beg you, — ’ 1 Supercilious
3 # ‘ , its your own 2 # ‘ I’m pleased with you, — ’ 1 # Instigation to mis-
affair ’ chief
31 Reassuring 32 ‘ Hurry up, — ’ 21 Polite request
13 ‘ , if you like ’ 23 ‘ —, I don’t care whether you do 12 Simple command
or not ’
13 # ‘ —, get out of my 23 # ‘ —, it will be good for you ’ 12 # Very polite
sight ’
131 ‘ , don’t be afraid ’ 232 Very polite, old- 121 ‘ —,' its a good idea ’
fashioned
231 ‘ , I’m glad for you ’ 132 Insulting
313 ‘ , I don’t need 312 ‘ Please, I beg you, — ’ 323 —
you ’
213 ‘ I don’t care ’ 212 Simple polite
313# Very strong com- 312 # ‘ —, before I lose my 323 # ‘ Hurry up, — ’
mand temper ’
213#‘If you don’t like 212 # NORMAL
what I’m doing,—’
3131 ‘ Whatever it costs 3231 ‘ Don’t be angry, — ’ 2132 ‘ I’m fed up with
you — ’ you ’
3132 Pleading 3232 ‘ Please ! — ’ 2121 ‘ I forgive you, — ’
3121 ‘ —, you ought to be 2131 ‘ —, if you think you can ’
ashamed of your- self ’

1313 ‘ —, get out of my sight ’ 2323 ‘ Don’t give up, — ’ 1213 ‘ —, you know what
you are doing ’
1312 ‘ —, God will bless 2312 * I’ve warned you once already, — ’ 1212 ‘ —, I shan’t tell
you ’ you again ’


A NOTE ON KRIO TONES

63

INVENTORY (contd.)

1323 ‘ , and good rid- dance ! ’ 2313 ‘ —, you have to, whatever you feel about it ’ -
1313 # ‘ , in peace ’ 2323 # ‘ —, and you know why ’ 1213 # ‘ —, I don’t want to see you again ’
1312 # ‘ , you’ve caused me enough trouble ’ 2312 # ‘ —, I beg you ’ 1212 # ‘ I leave you to God’
1323 # Disgust 2313 # Goading
13131 — 23231 ‘ Take no notice of what they say, — ’ 12131 ‘ Nobody wants you, — ’
13132 — 23232 ‘ , but be careful ’ 12132 ‘ I advise you strongly, — ’
13121 Insistent (e.g. at the same time pushing the interlocutor) 23121 ‘ out of my sight, — ’ 12121 Very insulting
13231 ‘ I’m happy for you, 23132 ‘ —, I’m tired of you ! ’
13232 ‘ Don’t be afraid, — ’ 23131 —


A CLASS OF PHONAESTHETIC WORDS IN BERBER

By J. Bynon

While transcribing recordings made in 1964 of women telling folktales in the
Ayt Hadiddu dialect of Tamazight, I noted the fairly frequent occurrence of
words of a category not normally represented in Berber grammars, dictionaries
and texts, but which nevertheless would appear to be characteristic of expressive
narrative style. The words in question are of a type often described by means
of such terms as onomatopoeic, echoic, ideophonic, phonaesthetic, etc., since they
are felt in some way to imitate or mimic the thing to which they are applied or
some aspect of it. Their apparent neglect so far by Berberists is hardly surprising,
for one is unlikely to come across them in the carefully prepared written or
dictated textual material which tended to serve as the principal basis of linguistic
investigation previous to the adoption of the tape-recorder as a standard aid in
data-collecting, and their visibly somewhat marginal position in the language may
also have been a contributing factor in their exclusion. Since, however, I was
attempting to deal exhaustively with the contents of the tapes I could hardly
disregard forms which, whatever their grammatical status, are clearly relevant
to the understanding of expressive style in discourse and narrative and which
may well play a significant role in the sphere of lexical innovation. I was also
struck not so much by the fact of their existence in Berber—which we may be
tempted almost to take for granted since something very similar to them is
probably to be found in the majority of languages, including English—as by, on
the one hand, the degree of pure convention in the relationship between form and
referent and, on the other, the considerable formal restriction and high level
of phonemic integration which they exhibit.

Thus, while an element of mimicry is undoubtedly present in many cases, the
small part played by actual acoustic identity becomes obvious when the Berber
forms are compared with, for example, their corresponding English forms—where
these exist—and even in those instances where the intention is clearly to
‘ represent ’ some real and identifiable sound, this is realized almost exclusively
through the employment of the normal structural elements of the language.

In order to determine something of the principles upon which the forms of
these words depend and of their function and status in the language, I took the
opportunity in April 1968 of asking a young man from the tribe in question,1
who was working with me in London, to make a list of all the words of this type
that he was able to assemble. The resultant document, to which he gave the title
ismawn nna nttini 3-udyac n-isddien, ‘ words that we say in place of noises ’,

1 zayd u-ebbu, from the ayt-cdiddu village of ayt-Iyazi situated near Imilchil in the valley
of the asif mllull, Central Morocco.


A CLASS OF PHONAESTHETIC WORDS IN BERBER

65

consists of seventy-five examples each of which is accompanied by a short explana-
tion and one or more sentences illustrating the manner and context in which it is
employed.2 The procedure adopted is one which has its disadvantages and the
material which I obtained in this way differs in certain important respects from
that obtained under the more natural conditions of spontaneous employment
during narration. These differences, however, are in themselves of some interest
and will be taken into account in the subsequent discussion. The forms provided
by the informant are listed below,3 classed according to the structure of their
stems and numbered serially for ease of reference. Opposite each will be found
not so much a translation as an indication of the context or contexts in which the
term is employed in the examples provided by the informant. Although many
of the forms are probably of fairly restricted application, others no doubt have a
wider range of use and the list of contexts should not therefore be taken as
exhaustive.

Type A : stems consisting of two different consonants, separated or not by a vowel

(i) Stem pattern CxVCa (seventeen examples):

Three are without reduplication.

1. ttaxx a cooking pot falling and breaking (= No. 40)

2. ffaft a child falling over on its face while running

(cf. No. 33)

3. ssaft a slap on the face

Two were given without reduplication when in isolation, but with total
reduplication of the stem in the illustrations.

4. ggabb, ggabb-ggabb stones falling into water (cf. also No. 47)

5. ggaww, ggaww-ggaww blasting with high explosives

A further eleven show total stem reduplication.

6. qqabb-qqabb a mule walking

7. ttabb-ttabb a camel running

2 Anyone desiring, for research purposes, a recording of the text in question should apply
to the author in writing, c/o S.O.A.S., University of London, W.C.l.

3 The phonetic values of the symbols employed in the transcription of the Berber are
approximately those of the International Phonetic Alphabet with the following modifications
and additions : k and g when non-geminate represent the corresponding palatal fricatives,
x and y are voiceless and voiced uvular fricatives, qq is a voiceless uvular plosive, c and E are
voiceless and voiced pharyngal fricatives, h is a voiced laryngal fricative, dd and gg are voiceless
and voiced alveolar affricates, tt 4 § 3 t § gg are emphatics, w and y placed after a symbol
indicate labialization and palatalization respectively, gemination is indicated by doubling the
symbol.


66

J. BYNON

8. ddagg-ddagg

9. ggaww-ggaww

10. mmaqq-mmaqq

11. ggabb-ggabb

12. kkabb-kkabb

13. ddaff-ddaff

14. ddakk-dSakk

15. qqamm-qqamm

16. ttann-ttann

henna being pounded in a mortar; someone
knocking loudly on a door (= No. 41)
sparrows or other small birds chirping
water dripping

children swimming in the river
hens or birds eating

men punching one another in a fisticuff
love-making

human beings eating

a dog eating

And in one case there is perhaps incomplete reduplication (cf. Nos. 34 and 35).

17. bbuE(e)-bbuEE a camel bellowing

(ii) Stem pattern C^VCa (three examples):
All three have total reduplication.

18. mmuh-mmuh

19. ttax-ttax

20. ddah-ddah

a cow mooing

a person carrying out his daily work (plough-
ing, washing clothes, etc.) ‘ until it is
completed ’ (cf. No. 71)

house-building (using rammed earth con-
struction)

(iii) Stem pattern CiVC2 (fifteen examples):
Thirteen show total reduplication.

21. huww-huww a dog, fox or jackal barking
22. saww-Eaww a cat miaowing; a jackal howling; women
singing
23. yaww-yaww women quarrelling
24. hayy-hayy travelling over a distance; the passage of
time
25. samm-samm children eating
26. xiyy-xiyy corn being threshed
27. ^akky-^akky a person crying ibibibibibiw in order to obtain
help
28. hun-hutf a donkey braying
29. farr-farr a bird flying
30. Yacc-yan a child dragging a metallic object (sheet-

metal, empty tin) along the ground
(= No. 61)


A CLASS OF PHONAESTHETIC WORDS IN BERBER 67

31. 4abb-4abb a woman slapping flat cakes of bread into

shape

32. saqq-saqq a person vomiting

33 fafl-fatt a slaughtered beast kicking on the ground

while dying (cf. No. 2)

In two instances the reduplication is incomplete, resulting in an overall pattern

c1vc2-c1vc2.

34. bae-baee

35. mas-maEE

sheep bleating
goats bleating

(iv) Stem pattern QVQ (four examples):
All have total reduplication.

36. hab-hab

37. waE-waE

38. gun-gun

39. yab-yab

dogs barking

a small child crying

a person who spends his whole time sleeping
something being done rapidly

(v) Stem pattern CiC2 (eight examples):
One is without reduplication.

40. ttxx

a cooking pot falling and breaking (= No. 1)

And seven with total reduplication.

41. ddgg-ddgg

42. gg^-gg2}2[

43. ggwrr-ggwrr

44. tttt-ttct

45.

46. ddfE-dcff

47. ggbb-ggbb

henna being pounded; someone knocking
on a door (= No. 8)

trees creaking and groaning in the wind;
a load of wood being transported by
mule-back (= No. 56)

turtle-doves cooing

breaking wind

breaking wind

clothes being laundered by stamping on a
flat stone in the river

stones falling into water (= No. 4. Doubtful.
Originally given as alternative to ggabb-
ggabb in the illustrative sentence, but later
omitted)


68

J. BYNON

(vi) Stem pattern QCa (two examples):

One was given in isolation without reduplication but reduplicated in the
illustration.

48. ttykw, ttykw-ttykw rifle fire

The other with reduplication.

49. ddxw-ddxw milk being churned in a skin

(vii) Stem pattern CXC2 (thirteen examples):
All show total reduplication.

50. hrr-hrr

51. W33-W33

52. xtt-xtt

53.

54. brr-brr

55.

56. gy^-gysra

57. bzz-bzz

58. bxx-bxx

59. gwrr-gwrr

60. 3rr- 3rr

61. Ytt-Ytt

62. 3tt-3lt

a cat purring; a woman grinding corn with
a handquern

a strong wind blowing

a person snoring; someone noisily sucking
in hot gruel through the lips

water flowing in a gulley after rain

thread being spun by means of a spindle and
whorl

a musical instrument being played (whether
by bowing or plucking)

trees creaking in the wind; wood being
transported on muleback (= No. 42)
the buzzing of flies, bees or wasps
heavy rainfall; a person weeping abundantly
children eating chick-peas or roasted barley
grains

grass being mown or corn reaped by means
of a small hand-sickle

a child dragging a metal object along the
ground (= No. 30)

tea, coffee or milk being poured into a vessel
from a height

(viii) Stem pattern CxCg (two examples):
Both show total reduplication.

63. tote

64. ?h-?&

persons whispering or talking together in a
low voice

a person moaning in grief, groaning in pain


A CLASS OF PHONAESTHETIC WORDS IN BERBER 69

Type B: stems consisting of the same consonant repeated, with an intervening vowel

All the members of this type show multiple partial reduplication of the stem;
the vowel is an a in all six examples of the corpus.

(i) Stem pattern CVC, overall pattern (CV)nC (four examples):

65.

66.

67.

68.

dda-dda-dda-ddadd
nna-nna-nna-nna-nnann
qqa-qqa-qqa-qqaqq

thunder; an aeroplane ; a lorry
women chattering together
a hen cackling after having laid an egg

kkya-kkya-kkya-kkya-kkyakky laughter (cf. No. 70)

(ii) Stem pattern CVC, overall pattern (CV)nC (one example):

69. ta-ta-ta-tatt a pot of gruel or water boiling

(iii) Stem pattern CVC, overall pattern (CV)nC (one example):

70. ha-ha-ha-ha-hah laughter (cf. No. 68)

Type C: complex stems

Under this heading may be grouped for convenience five isolated cases which
do not fit into either of the above two types.

In two, both of which exhibit total reduplication, the stem commences with
a cluster.

(i) Stem pattern CxCaVCa (one example):

71. xdam-xdam work being carried out (cf. No. 19)

(ii) Stem pattern CxCgVCa (one example):

72. crabb-crabb a dog lapping up milk

In another two reduplication is only partial, either because of alternance in
the vowels or because of the insertion of a supplementary consonant.

(iii) Pattern C\ V1C2V2-C1V3C2V;i (one example):

73. qqabbu-qqibbi wood being chopped with an axe; black-

smith hammering a ploughshare into
shape

(iv) Pattern CjVCaCa-CiVCa (one example):

74. ddagg3-ddagg a frame drum being beaten


70

J. BYNON

And finally in one unique instance the stem is made up of two entirely un-
related syllables which hold no phoneme in common.

(v) Pattern CxCsCgVC^ (one example):

75. kkrtabb an animal escaping suddenly into its burrow

The sounds employed in the words of the above corpus are shown in the
following table, together with their relative frequency of occurrence as expressed
by the number of different stems in which each is found.

Rank Phonemes Stems
1 a 44
2 b/bb 17
3 h; gy, g/gg, gw/ggw 11
4 y/qq 10
5 t/tt, tty ; x/xx, xw ; d/dd ; r/rr 8
6 E/EE; w/ww; tt; 4/tt 7
7 u ; kk, kky, kw ; m/'mm 6
8 f/fE; ?/« 5
9 3/33 4
10 n/nn 3
11 12 i; gg; 86; yy; c ss; zz; ?fi 2 1

These sounds are, with only two exceptions, the normal ones employed by the
central phonemic system of the dialect. The exceptions are the emphatic voiced
affricate, found in Nos. 9 and 11 and the glottal stop followed by voiced glottal
friction with nasal release, ?fi, of No. 64. The first of these is hardly a major
innovation, resulting as it does from the extension to an existent member of the
affricate series gg, of a process already at work in the language. There are in fact
other examples in Tamazight of the tendency to enlarge the emphatic series of
consonants and the informant’s dialect has, for instance, in addition to the
grooved alveolar fricatives f and 3, the corresponding emphatics § and 4 which,
although of low frequency of occurrence and certainly of secondary origin,1 * * 4
must nevertheless be assigned full phonemic status.5 The second case is perhaps

1 See A. Basset, La langue berbere (in Handbook of African languages), London, 1952,

5-10; L. Galand in Encyclop. Isl., 2nd ed., 1181-2; P. Galand-Pernet, ‘ Emphase et ex-

pressivite : l’opposition z - z en berbere (Maroc du sud) ’ in Communications et rapports du
ler Congres intern, de Dialectologie generale (Louvain . . ., Bruxelles . . ., 1960), III, Louvain,

1965, 39^4-7.

5 Cf., for instance, such minimal pairs as timaj'jwin ‘ flints ’ (pl. of timi/Ja) and timaggwin
‘ female cats ’ (pl. of timiggut), tn3im ‘ you swim ’ (< verb n3y) and (ur-)tn^im ‘ she is not
bald ’ (< verb n^m), etc.


A CLASS OF PHONAESTHETIC WORDS IN BERBER

71

a somewhat special one in that the word in which it occurs represents a sound
which is itself produced by means of the human vocal organs and is perhaps not
so much a formalized representation of moaning as the actual articulatory
‘ bearer ’ of moaning in the culture.6

It is not unusual to find a small proportion of special sounds in the marginal
areas of a language, such as its exclamations, nursery vocabulary, etc.,7 and in
the present instance the fact that in the overwhelming majority of cases the sounds
employed are those of the central system is, I think, of more significance than is
the presence of two deviant sounds. Judged solely from the viewpoint of the
phonemic inventory there is no good reason why the members of the corpus
should not be considered to be genuine lexical items of the dialect.

In the absence of any study concerning the normal combinatory restrictions
affecting the phonemes in the more central areas of the language, nothing useful
can be said regarding this aspect, although certain combinations do give an
unusual impression. An examination of the frequency column in the table does,
however, reveal what appear to be a certain number of deviations from the
‘ norm ’.8 For the consonants the high rank occupied by h, a sound not usually
considered even to form part of the original phonemic inventory of Berber,9 is
noteworthy, as is the total absence of 1 and the very low frequency of the sibilants
and chuintantes. While chance cannot be ruled out in the case of so small a sample,
it is possible that some purely acoustic feature such as the intrinsic energy of the
individual sounds may play a role in the distribution.

Where the vowels are concerned the very high incidence of a by comparison
with i and u is striking, a appearing in forty-four words as against u in six and
i in only two, that is to say it occurs more than five times as frequently as the other
two vowels taken together. This can be compared with a norm for these same
vowels in nominal stems probably somewhere in the region of 3: 3 : 2.10 The
greater intrinsic energy of a as opposed to u and i might again be responsible.

It is clear that oppositions of vowel timbre play a very secondary role in this

0 It would appear that the articulatory tendencies of man are so powerful that even naturally
based sounds are to some degree artificially articulated. Thus, while the vocal expression of
the emotions (pain, grief, fear, surprise, amusement), as well as certain other emissions of air
through the respiratory tract (sneezing, coughing), have a purely physiological basis and are
largely outside the control of the individual as regards the moment and energy of their emission,
they nevertheless appear to have a cultural component in the form of the articulatory pattern
or patterns according to which they are released. The result is that a Berber neither groans nor
laughs nor sneezes like an Englishman.

7 Cf. the click represented graphically as ‘ tut-tut ’, various whistles, bilabial trills, etc., in
English exclamations ; for nursery vocabulary see J. Bynon, ‘ Berber nursery language ’, in
TPS, 1968, 116-17.

8 For the ‘ norm ’, in the absence of anything more comprehensive, see J. Bynon, TPS
1968, 121.

0 See A. Bassett, Lang. Berb., 5-6.

10 See J. Bynon, TPS, 1968, 121.


72

J. BYNON

class of word, the only instance in which it appears to be minimally distinctive
being that of Nos. 17 and 34 while there are four examples of doublets with and
without the vowel a (Nos. 1 and 6, 10 and 41, 32 and 59, 42 and 54). This should
not perhaps be cause for surprise since in Berber lexical meaning tends to be
concentrated in the consonantal framework of the stem, the vowels being relegated
to a mainly grammatical role and we are here dealing with bare uninflectable
stems. It is noteworthy that in those instances where words of this type have,
by a process of derivation, given rise to verbs these have been fully integrated
into the verbal system and internal vocalic inflexion operates in the normal way.11

With regard to consonant gemination, here again the ratio of geminates to
non-geminates is considerably higher than might be expected from comparison
with the other sectors of the vocabulary, the ratio being 65: 35 in our corpus,
whereas it is only 40 : 60 even for nursery stems and a mere 17 : 83 for the stems
of the standard equivalents of these.12 If this latter figure is taken as the norm,
then the percentage of geminates is nearly four times the normal and can be said
to be one of the characteristic features of the class as a whole. Gemination has
phonemic status in Berber and is extensively employed by the morphology, but
the function here seems to be a purely expressive one. At first sight three pairs of
words appear to be opposed solely by this feature (Nos. 6 and 39, 7 and 31,
43 and 59) and, from a strictly phonemic point of view, this is so ; however, what
is not immediately obvious from the transcription, in each of these three cases the
fundamental phonemic opposition of non-geminate: geminate, while basically
one of tension ~ duration, is here accompanied by considerable phonetic differen-
tiation of a secondary nature, resulting in accessory oppositions of voice : voiceless-
ness and plosion: friction, and it is clear that it is these secondary phonetic
features which are basically relevant in the oppositions in question.13 It may be
noted that gemination by its very nature is accompanied by an increase in prom-
inence and it is doubtless this factor which is responsible for its greater
frequency in the class of words under consideration.14

If then the feature of gemination is abstracted, the stem patterns of types
A and B can be represented by the simplified formulae C1(V)C2 and CjVCx
respectively, and these cover seventy out of the seventy-five examples of the
corpus.

As has been stated already these words are invariable in form with the sole
exception that they are subject to reduplication. It will be seen from the corpus

11 Thus the verb sqqiqqy ‘ to cackle ’ has the following stems : simple aorist sqqiqqy,
intensive aorist sqqiqqiy, positive preterite sqqaqqy, negative preterite sqqaqqy ; the corresponding
stems of the verb sxu[[ ‘ to snore, snort ’ are sxun, sxutfu, sxwa[[, sxwa[[.

12 See J. Bynon, TPS, 1968, 121-2.

13 Thus the geminate which corresponds to the voiced fricative y is the voiceless plosive qq,
voiced 4 corresponds to voiceless fi, fricative gw to plosive ggw.

14 See, for example, T. F. Mitchell, ‘ Long consonants in phonology and phonetics ’, in
Studies in Linguistic Analysis, The Philological Society, 1957, 182-205.


A CLASS OF PHONAESTHETIC WORDS IN BERBER

73

that reduplication may be absent or present and in this latter case may be partial
or total. It is clear that the absence or presence of reduplication, in type A stems
at least, is not directly dependent upon their pattern, and examples like Nos. 4,
5, 48 (perhaps Nos. 2 and 33) demonstrate that it is not necessarily an absolute
property of a particular stem. An examination of the ‘ meanings ’, that is to say
the contexts in which they are employed, shows that those stems which have been
given by the informant without reduplication (Nos. 1, 2, 3, 40, 75) all represent
a unitary and completed action, those given both in a simple and a reduplicated
form (Nos. 4, 5, 48) all represent actions which may be either unitary or multiple,
while all the remainder are reduplicated and represent what are essentially con-
tinuous or repetitive actions. This apparent correlation between reduplication
and the character of the action is confirmed by their behaviour at the syntactic
level so that reduplication should probably be treated not so much as a feature
of their morphology as one of syntax.

Turning then from questions of morphology to those of syntax, an analysis
of the illustrations provided by the informant shows these words to be employed
in two quite distinct ways. In the first of these they function as nominals and can
perform, for example, the role of subject of a verb, be governed by a pre-
position or qualified by a possessive pronoun or a genitive construction. Although
there are only twenty-one examples of this nominal use out of a total of some
eighty sentences, there is no obvious reason for assuming that they cannot all
function in this capacity.

When subject of a verb and following it, a position in which nouns capable
of doing so go into the so-called ‘ construct ’ form in Berber, they remain
unchanged and the verb takes the prefix of the third person masculine singular 15
(four examples):

76. inya xdam-xdam n-ku yass ixddamn the work-work of every day has worn

n-flan qqiman ur-xdimn ass-ttx out so-and-so’s workers (and)

they are staying off work to-day 16

They often occur in a negative proposition with xs ‘ except ’ (twelve examples):

77. i fir ran im^yann ur-da-tn-issfcac xs little boys, the only thing that makes

yatt-Yatt n-iqqzdirr d-iyw[[afn them happy is the clatter-clatter

d-imsman of sheet iron, tin cans and bits of

wire, etc.

15 As the unmarked form ; cf. same phenomenon in Berber nursery language, J. Bynon,
TPS, 1968, 130.

18 No serious attempt has been made to find English equivalents for the Berber exemplifiers
and the forms in italics are not intended to be taken as anything other than very approximate
renderings.


74

J. BYNON

When governed by a preposition requiring the construct form they are, of
course, equally invariable (ten examples):

78. ma f-yayn a-mugg allis ixfawn-
-nny s-saww-saww

79. t^i tmttutt-inw t-ti n-eli slliy-nn
i-yaww-yaww-nsnt g-gi3ran

what’s the matter with you, cat, that
you split our heads with (your)
miaow-miaow ?

my wife quarrelled with Ali’s (and)
I heard their bawl-bawl in the
fields

The possessive pronouns which they
nouns (two examples):

80. ar-tynf tmttutt aycum ggbac tsfafa-yi-d
s-c[abb-c|abb-ns

take belong to the series suffixed to

(my) wife was making bread this
morning (and) she woke me up
with her tap-tap

In this nominal use they are frequently qualified by a phrase introduced by
the preposition n- ‘ of ’ (twelve examples):

81. in-asn i-yjirran ad-ur-ttddun s-asif
ad-£umn inya-tn ggabb-ggabb
n-ku yass

say to the children that they are not
to go to the river to bathe, the
glug-glug of every day does them
harm

However, in the majority of examples (fifty-six of the total), these words fulfil
quite another role, for which I would suggest the term ‘ exemplification ’.17
Functioning in this way as an ‘ exemplifier ’ they do not replace any of the normal
constituents of the clause the action of which is being illustrated (or ‘ exemplified ’),
but are situated immediately next to it, usually following, but very occasionally
preceding.

When applied to a perfected action, the verb is in the preterite and the
exemplifier is not reduplicated. There is a single example in which it precedes
the clause it is exemplifying :

82. anniy yan-uwtul bna dduy-nn ^i[-s I saw a hare outside (its burrow and)
irul kkrtabb ikw3m abuhu-ns went towards it, it ran away (and)

plonk it entered its burrow

17 ‘ Demonstration ’ would be as satisfactory were it not for the fact that ‘ demonstrative ’
already has a specialized meaning in grammar ; the term ‘ ideophone ’ has been strictly avoided
because of its highly specific application in Bantu languages (which does not of course mean
that there may not be certain typological similarities between the Berber exemplifier and the
Bantu ideophone).


A CLASS OF PHONAESTHETIC WORDS IN BERBER

75

But usually it follows it (three examples):

83. ta4[-as tbcmitt i-leil 3-ufus tay-nn

aflla n-islli ttaxx ttt?

84. ar-ttazzlan ifirran imic irrdl yuwwn

fatt ikk-3d yif-s ibnyc

85. iwt-i lfqqih s-utjmmis yat-tikklt ssafi

isdu[-i

the pot fell from the boy’s hand (and)
landed crash on top of a stone
(and) broke

the children were running, after a
moment one of them fell down
plonk (and) was covered in dust
the school master gave me a slap
once, smack, he/it sent me spin-
ning

When the action is a repetitive or continuing one the exemplifier is reduplicated,
usually twice in the examples given by the informant,18 and in nearly every
instance 19 the verb is in the intensive aorist. In all the examples the exemplifier
follows the exemplified clause (fifty-one examples):

86. ar-t^ad tmfiutt i43amm hrr-hrr ur- (my) wife was grinding (corn) last
-ax-tuggi yan-n3n s-udida n-uzr3 night grind-grind (and) she didn’t

let us get to sleep with the noise
of the quern

In a large number of cases the exemplifier is followed by a clause introduced by
alli3 (twenty-five examples) or ard (four examples), words which can be approxi-
mately translated as ‘ and thus until, and so until ’:

87. 3ir-as ifirran ayu yi-widi yar-icllb

erabb-erabb alli3 igguwn izri-t

88. da-ssiridn igbbann ku yass g-gwasif

gdff-ddff art^tyly

the children gave milk to the dog, it
lapped, lap-lap, and thus until it
had had enough (and) left it

the launderers did the washing every
day at the river stamp-stamp and
thus until (the sun) went down

One pair of examples was produced by the informant in which, in order to
distinguish two different applications of the same exemplifier, this latter is qualified
by a phrase introduced by the preposition n- in the same way as is frequently the
case when they are used as nominals :

18 With the exception, of course, of all type B stems, but also a solitary example employing
a type A stem (No. 33) :

85a. y[sy-as i-fac|ma yi-wfullus ar-ittf[fi([ I slaughtered the chicken for Fatima (and) it
Jatt-fatt-Jatt alli3 immut struggled kick-kick-kick until it was dead

lu There were a maximum of five possible exceptions in the illustrations provided by the
informant.


76

J. BYNON

89. nddas-aybala yar-yif-nv-tkkat s-un^at

bxx-bxx n-waman ddruy ayd
nn-ur-ngula

90. ufiy-d Ja n-tutmin ar-allant bxx-bxx

n-imftawn eniy immut-asnt Ja
t311inin

we set off for Aghbala (and) it
poured with rain on us splash-
splash of water (so that) we
almost didn’t get there

I came upon some women weeping
splash-splash of tears, perhaps
someone of theirs had died, poor
things

And finally in quite a number of cases the exemplifier has the same root as
the verb that it is exemplifying, although there is no indication as to whether
this is a matter of stylistics or lexical restriction (seven examples):

91. ku yicf da-yshuwwu Ja 3-uqqJJa
huww-huww ard iwcl iddu

92. i3n flan yut-ty ar-isxunu xn-x[t
ur-ggin-n3in alli3 d-yuly wass

every night something barks on the
southern slope of the mountain
woof-woof until it tires and goes
off

so-and-so slept at our place (and) he
snored snore-snore, we never got
to sleep, and thus until the day
came up

The above is the picture that emerges from an examination of the list and
illustrative examples drawn up by a single native speaker. It has the advantage
that, coming from an informant whose phonemic system has already been the
object of some detailed study, it has been possible to establish the forms of the
stems with a relatively high degree of precision.20 The semantic contexts in which
these are employed can also probably be accepted as being adequately representa-
tive of the performance of one individual.

As has already been pointed out, however, a description based uniquely upon
material obtained under such circumstances is likely to be unsatisfactory in a
number of respects,21 and this first corpus-based examination of the subject

20 Perhaps even with excessive precision where certain features such as gemination are
concerned. I am not at all convinced that minimal differences of pattern like that which opposes
No. 17 to Nos. 34 and 35 or No. 2 to No. 33 would necessarily be maintained by other
informants.

21 The most serious theoretical criticism is that the class has been set up in the first instance
on the basis of the linguistic awareness of the native speaker rather than on that of a morpho-
logical or syntactic analysis by the linguist. Further, we may wonder to what extent the
informant’s definition ‘ words that we say in place of noises ’ may not have led him to
reduce the number of examples like Nos. 19, 24, 38, 71, etc. in which noise is either a secondary
factor or even totally absent. Again it is possible that, faced with the somewhat onerous task
of composing numerous illustrative examples, the informant may have tended to use the earlier
sentences that he created as models for the succeeding ones and in this way entire syntactic
patterns may have been missed.


A CLASS OF PHONAESTHETIC WORDS IN BERBER

77

should, therefore, be taken as no more than a starting-point for more detailed
research using material obtained under more natural conditions. Such research
still remains to be undertaken, but a preliminary examination of tapes of women
from neighbouring villages telling folk-tales has resulted in the following
observations:

There appears to be some considerable variation in the extent to which
individual speakers use exemplifiers in story telling (one woman using only
half a dozen different forms over a period of several hours, whereas another used
more than twenty in less than an hour) as well as in the popularity with individuals
of particular exemplifiers. The ones most commonly used are those which refer to
the passage of time or the action of travelling over a distance.

Many forms appear in the tapes which are not to be found in Zayd’s corpus
and it is clear that this latter must be taken as no more than a sample of the total
in use in the dialect. As far as it has been possible to judge, however, all new
examples encountered accord in structure with the patterns already set up on the
basis of the corpus.22

While many of the forms in the tapes are used in exactly the same way as in
the corpus (for example Nos. 20, 23, 24, 39, 75, etc.) this is not always the case.
Thus we find No. 6 reduplicated applied to ‘ a person eating ’ and unredupli-
cated to ‘ persons going quickly into a cavern ’, ‘ a bird entering a house suddenly
through the smoke hole ’, ‘ a man falling into a jar of honey ’, No. 16 to ‘a
woman being larded with stab wounds from daggers ’ No. 73, to ‘ hunters
shooting away at hares ’, and so on.

Certain of our initial findings receive confirmation, such as the fact that
exemplifiers are frequently followed by a clause introduced by alli3 and that
they can be qualified by a genitive phrase introduced by n-, e.g.

ar-tn-tt3ttan ar-tn-tt3ttan ar-tn-tt3ttan they ate them they ate them they ate
kkbb kkbb kkbb kkbb n-tslmin them munch munch munch munch

of little fish

Others, however, clearly require modification and, as might be expected, the
most important differences concern the way in which the exemplifiers are used.
No examples, for instance, of nominal use were encountered in the tapes examined,
whereas there were large numbers of examples of employment for exemplification.
There can be little doubt, therefore, that it is the latter function which is the
fundamental one and that use as a nominal is an extension. The exaggerated
importance accorded it in Zayd’s corpus is presumably the result of the con-
ditions under which the illustrations were composed—used ‘ cold ’ they were given
their most grammatically integrated application.

22 The rapid and often careless articulation of some of the story-tellers makes it impossible
to establish new forms with any degree of precision.


78

J. BYNON

The basic opposition found in Zayd’s examples between non-reduplicated
forms associated with unitary perfected actions and reduplicated forms with
enduring or repetitive actions is confirmed by the tapes, but whereas reduplication
in Zayd’s examples almost invariably means repetition once only, resulting in a
doubling of the stem, the number of repetitions in the texts is invariably greater
than this, and quite frequently reaches a dozen or more. Here again the explana-
tion is probably an over-formalization of the situation by the informant due to
the non-expressive context in which he was operating.

Another thing that is brought out by the recordings is the fact that special
supra-segmental features (of pitch, of intensity, of rate of output) are frequently
associated with these words, resulting in their having much greater prominence
than their immediate surroundings.

Finally, there are a number of cases in which these forms are employed in
ways that are not illustrated in Zayd’s examples. Two of these deserve particular
mention since they demonstrate that exemplifiers can by themselves carry primary
information unaided and thus must be considered to have a degree of lexical,
and not merely expressive, content.

Firstly, an exemplifier can operate without the presence of any verb to express
the action being exemplified, the verbal action being understood from the context
and the nature of the exemplifier. Thus, employing a doublet of No. 12 and the
subject being a bird, we find :

iddu s-islman kkbb kkbb kkbb kkbb
kkbb kkbb kkbb kkbb alli3 asn-d-
-yusy mayd ttattan wi-nnay n-
-i Jinan

it went towards the fish, peck peck
peck . . ., and thus until it carried
back that which those children
(could) eat

Since the exemplifier concerned usually refers specifically to hens or birds
pecking, the nature of the verbal action is unambiguously conveyed.

Secondly, an exemplifier can be introduced by the verb ini ‘ to say ’. Thus,
speaking of women who had drawn water from a well near which a small boy
was waiting:

ur-as-3int tmxibin tismdlt inna-nn they didn’t put back the cover for it,
kkrttab . . . wy-nnay 1-leil the so-and-so’s. He said plonk,

that boy

Here again, since the exemplifier in question is one which is normally employed
for things disappearing suddenly into holes, etc., the meaning is conveyed that
the boy fell down the well. This latter construction is particularly interesting in
view of parallels with languages outside the Berber-speaking area.

In conclusion it may be stated that there exists in the dialect of Tamazight
studied (and, we can safely anticipate, in other Berber dialects as well) a class of


A CLASS OF PHONAESTHETIC WORDS IN BERBER

79

words which we may term ‘ exemplifiers The special features displayed by these
words at the morphological and syntactic levels, their tendency to undergo
reduplication, the obvious phonaesthetic character of many of them, their associa-
tion with special supra-segmental patterns, when taken together provide sufficient
justification for putting them into a special class of their own. Except when used
as nominals, certainly a secondary development,23 they remain largely autonomous
of and external to the clause with which they are associated, usually following it
but occasionally preceding it, and form no essential part of its structure. Their
function would appear to be that of a sort of vocal gesture which takes up and
repeats the idea of the verbal action at a more expressive level, underlining its
sudden or extended nature, and acting as a bearer for certain supra-segmental
features of purely expressive content.

But, although the role of intensifier is undoubtedly important, and while the
opposition non-reduplicate: reduplicate may be considered to carry purely
grammatical meaning, the regular association of specific exemplifiers with par-
ticular sorts of action, or even particular actions, gives them some measure of
content so that they are not limited to merely reinforcing the message of a verb
but may actually substitute for it.

If, then, exemplifiers carry lexical meaning, are they to be considered as
purely arbitrary signs or is there some detectable link between their form and
their content? And if, as it seems impossible to deny, they possess a certain
phonaesthetic quality for the native speaker, upon what does this depend?
While an element of echoism is undoubtedly present in many cases, notably those
which refer to animal cries and other natural sounds, the class as a whole cannot
be dismissed simply as a form of mimicry.24 This is not only because some
exemplifiers are applied to actions with which no obvious sound is associable, but
more seriously because others appear to be closely bound up with the lexical

23 Whereas native Berber nouns normally show inflexion for number, gender and state,
exemplifiers are invariable. In this they are comparable to recent loans and can, in fact, be
regarded as ‘ loans ’ from another sector of the language only partially integrated into the
nominal system.

21 Genuine mimicry is widespread among persons who spend much of their time in close
contact with nature, in particular hunters, shepherds, etc., who employ it to ‘ call down ’ wild
birds, imitate the cries of beasts, etc. But the various articulations employed are quite in-
dependent of those of the performer’s language and, although the manner of their production
may be traditional and therefore part of his culture, the sounds produced do not together form
a system, but each is quite independent of the others. The sole aim of the mimic is to produce
a sound having the maximum degree of acoustic similarity to the model, and in this he is
restricted neither in the range of sounds employed nor in the structural patterns formed ;
furthermore, the sounds that he produces have purely imitative and not symbolic function.

Children at play also use, in their imitations of gunshots, of internal combustion engines, etc.,
a form of vocal mimicry which is often quite highly institutionalized and which, while basically
expressive in function, is at least partly symbolic and draws largely upon the sounds of the
speaker’s language. Such sounds, however, while they may be signals, are embedded not in
the chain of speech but in the play sequence and cannot be considered to be part of language.


80

A CLASS OF PHONAESTHETIC WORDS IN BERBER

stock of the language so that the association is clearly not an acoustic one but
is based upon the native speaker’s feeling for roots as the bearers of semantic
content. This is undoubtedly so in the case of No. 71, which must derive from
the verb xdm ‘ to work ’ since this is itself a loan from Arabic, but it is also at
least highly likely in a number of other cases,25 for example No. 54 (cf. brrm
‘ to spin, turn ’), No. 60 (cf. m3r ‘ to reap, mow ’), No. 38 (cf. sgunfu ‘ to rest ’),
No. 33 (cf. ‘ to struggle ’).

The relationship of exemplifiers to the lexical stock of the language appears,
therefore, to be a complex one and, while they are demonstrably an important
source of innovation for the central system,26 it is also true that the process can
operate in the reverse direction.

25 Since the majority of exemplifiers have a stem pattern based on only two consonants,
whereas most Berber roots contain three or more, the loss of some of the radicals is to be
expected.

20 See footnote 11, p. 72 ; other derived verbs are : shun ‘ to bray (donkey) ’, smmuh
‘ to moo (cow) seiner ‘ to whinny, neigh (horse, mule) ’, sbue ‘ to bellow (camel) ’, sbie ‘ to
bleat (sheep) ’, shuww ‘ to bark (jackal, etc.) ’, smiqqy ‘ to drip


CATEGORIES OF THE VERBAL PIECE IN BACHAMA

By Jack Carnochan

Very little has so far been written about the language usually referred to as
Bachama, and called Kwaa-'Bwiare ‘ language of the people ’ by the ‘Bw&are or
Bachama tribe themselves. This outline of what is a little-known language is
offered as a tribute to Professor Guthrie, who has done perhaps more than any
other scholar to push forward the frontiers of our knowledge of African languages,
particularly in the comparative field.

Before serious work in this field can be undertaken for the West African
area, descriptive studies of many more of the languages are necessary to provide
adequate data for the comparativist. This article is a sketch of part of the
grammatical structure of one of the Chadic languages of West Africa. The
accompanying map shows the area in which it is spoken.

The term ‘ verbal piece ’ is already a commonplace in linguistic studies, and
needs no special comment. It enables me to deal here not only with the verb
word, but also with elements corresponding to terms in categories set up for the
verbal clause, phrase and group used in a technical sense. The exponents of terms
in some of these categories extend beyond the verb word and are of a syntactic
as well as of a morphological nature.

This article is practically the first linguistic study of Kwaa-Bwaare, and at
this stage it is thought preferable to give an account, albeit brief, of the main
categories of the verbal piece rather than to take just one category and to describe
it as exhaustively as possible at all levels of analysis. Because of limitations in
length, the description of the phonology has been kept to a minimum. The
language has by no means been fully investigated yet, and problems such as those
of word division still remain to be solved. This presentation of my findings,
therefore, is not regarded as definitive.

Publications on Kwaa-Bwaare include a schedule of words and phrases of
over one hundred and fifty items in Chapter I, ‘ The Bachama and Mbula ’, in
C. K. Meek’s Tribal studies in Northern Nigeria, Vol. I, Kegan Paul, London,
1931; an incomplete list of phonemes of ‘ Bacama ’ (sic) in Dr. Hans Wolff’s
Nigerian orthography, published for the North Regional Adult Education Office
and printed in Nigeria by Gaskiya Corporation, Zaria, 1954; a short list of words
and phrases with their German equivalents in Zeitschrift der Deutschen Morgen-
landischen Gesellschaft, VI, Leipzig, 1852, ‘ Auszug aus einem Briefe des Dr. Barth
an Dr. Beke ’, written from Kuka, 20 May 1851, 412-13. His list is of the Batta
dialect, and from one item one can imagine him at work, sitting under a tree, with
his informants around him. He points upwards, and gets the response, kdde
‘ branches, trees ’, which he notes against ‘ Baum ’, and then adds a question
mark, and in brackets: ‘ (s. Himmel) ’; did they think that he was pointing at
the sky ? If they had, the response would have been pwi.

G


82

JACK CARNOCHAN

Bachama Language Area (Bachama villages are shown in capital letters)

The Lutheran Church of Christ in the Sudan has published a collection of
hymns, Dyemshi Kwa Bware ka Demsi Kamu Bwatiye, in the Bachama and Batta
dialects, printed by the S. I. M. Niger Press, Jos, undated. The handbook of
African languages, Part II: languages of West Africa, by Dietrich Westermann
and M. A. Bryan, and published for the International African Institute by the
Oxford University Press, 1952, classifies the language as Chadic and gives a few
details, some of them not accurate (for example, the incomplete phonemic in-
ventory available at that time caused the authors to call the people gboare instead
of 6wAare) of its phonological characteristics. In his Languages of Africa, Mouton
and Co., 1963, Joseph H. Greenberg classifies Bachama and the kindred ‘ Bata ’
under ‘ III, Afroasiatic, E, Chad ’. Some details of the language are given in
the present author’s two articles, ‘ Nzeanzo and Won: a Bachama folktale
Journal of the Folklore Institute, Indiana University, IV, 2/3, published by
Mouton and Co., 1967, 230-9 ; and ‘ The coming of the Fulani: a Bachama oral
tradition ’, BSOAS, XXX, 3, 1967, 622-33. There appear to be no items from
Bachama or from Batta in S. W. Koelle’s Polyglotta Africana, Fourah Bay College,
The University College of Sierra Leone, 1963.

The reading transcription used here is phonemic. Seven short vowels,
i, e, a, o, e, u, u, and seven long ones, ii, ee, aa, oo, eo, uu, uh, are differentiated.
Apart from a and aa, short vowels are more open in quality than their long
counterparts. It is important to note that the values of u and u are reversed com-
pared with previous usage, as are those of o and e, and of the four long vowels.
The unrounded close centralized vowels occur very much more frequently than
the rounded close back vowels, and are now written u and uu; the letters u and hu


CATEGORIES OF THE VERBAL PIECE IN BACHAMA

83

are used for the less common sounds. Similarly, o and oo are used for the
centralized half-open to half-close vowels, with slight lip rounding, while e and ee
are reserved for the less common half-open to half-close back rounded vowels.
The enormous saving in diacritics would be a strong recommendation for this prin-
ciple to be adopted in an orthography for Kwaa-Bwaare, should it be acceptable
to the people themselves, even if dotted letters were preferred to barred letters.

Among the consonants, b, p, d, t, g, k, ?, v, J, z, s, J, c, m, n, ny, rj and h have
the qualities normally associated with them in the transcription of African
languages, and for the purposes of this article require no further comment.
Among the rest, gb and kp are voiced and voiceless labial velars respectively;
6 is a globalized bilabial plosive, and d is a globalized post-alveolar plosive;
r is a voiced post-alveolar flap except in final position, when it is a voiced alveolar
rolled consonant. In a few words, rolled e is phonemic, and is written with a
barred letter. This convention in the transcription is the reverse of that used in
the Kwaa-Bwaare text in ‘ The coming of the Fulani ’, but has been adopted in
view of the great saving in diacritical marks. There is a voiced palatal lateral, ly,
as well as a voiced alveolar lateral, 1; the semi-vowels w and y occur with globaliza-
tion, ?w and ?y, as well as without; there are six nasal complexes, mb, mgb, nd,
ijg, nz and nj, each of which functions as a single consonant in which the nasal
element is not syllabic; w is used in digraphs, as in 6wiare ‘ people ’, where there
is an element of rounding in the articulation; similarly y is used in digraphs
where there is a palatal element in the pronunciation, as in myee ‘ persons ’.
The letters y and w are also used as the second element of diphthongs. Capital
letters, used as the initial letters in sentences and proper names, have the same
phonetic values as the corresponding small letters.

Five tones are indicated, high by an acute accent, mid by a vertical accent
and low by a grave accent; rising and falling tones are indicated by a succession
of different marks on the adjacent letters of long vowels and diphthongs and by
* or A over a short vowel. In order to minimize the number of diacritical marks,
a syllable which has the same tone as the preceding one is left unmarked. This
means that the same word may be differently marked in different examples,
according to the tonal sequences of the examples.

The Kwaa-Bwaare material examined includes field recordings of oral tradi-
tions, religious ceremonies, folktales, songs and narrative poems, as well as
conversations collected on a research visit to the Numan area of Northern
Nigeria in 1963. A great deal of additional material has been provided for me by
Mr. E. B. Nadah during his time as assistant at the School of Oriental and
African Studies, London. For reasons of space, it is proposed in general to limit
the units presented here to sentences of one verbal clause only, leaving to another
occasion discussion of the relations of one clause to another. These single clause
sentences alone offer such diverse formal criteria as to require the setting up of
categories of mood, transitivity, aspect, grade, polarity, tense, number, person and


84

JACK CARNOCHAN

gender as well as necessitating the subclassification of the verbs on tonal grounds.
These categories together with the systems of their terms are given at this
point, and the reader may find it helpful on occasion to refer to this summary.

SUMMARY I

CATEGORIES OF THE VERBAL PIECE IN BACHAMA

CATEGORY I Mood

(i) Affirmative ; (ii) Interrogative ; (iii) Imperative

category ii Transitivity

(i) Intransitive ; (ii) Semi-transitive ; (iii) Transitive ; (iv) Ditransitive
category iii Aspect

(i) Aspect 1, Normal; (ii) Aspect 2, Causative; (iii) Aspect 3, Benefactive;
(iv) Aspect 4, Deprivative

category iv Grade

(i) Grade 1, Normal; (ii) Grade 2, Adessive

category v Polarity

(i) Positive ; (ii) Negative

category vi Tense

(i) Tense 1, Perfect; (ii) Tense 2, Past; (iii) Tense 3, Future; (iv) Tense 4,
Continuous; (v) Tense 5, Habitual

category vn Number

(i) Singular; (ii) Plural

category viii Person

(i) First Person Singular. Plural: (a) Exclusive, (b) Inclusive

(ii) Second Person Singular and Plural

(iii) Third Person Singular : (a) Masculine, (b) Feminine. Plural

(iv) Impersonal

category ix Gender

(i) Masculine ; (ii) Feminine

SUMMARY II

SUBCLASSIFICATION OF THE VERBS

1. Monosyllabic Verbs

Class I with four subclasses

Class II with six subclasses

Class III with two subclasses

2. Disyllabic Verbs

Four subclasses

3. Trisyllabic Verbs

Not yet classified


CATEGORIES OF THE VERBAL PIECE IN BACHAMA

85

CATEGORY I: MOOD

There are three terms in this category : affirmative, interrogative and impera-
tive. It is to be noted that affirmative is used here with a different meaning from
positive, which contrasts with negative in the two-term category of polarity.

Imperative mood clauses differ from the other classes in being tenseless, and in
the category of person, the only term applicable is second person. Positive
imperative clauses differ also in having no subject expressed, either by pronoun
or by nominal phrase.

Affirmative, Positive

Hye zumo You ate it 2nd pers. sg.; normal register (NR)

Hana zdmo You ate it 2nd pers. pl. NR

Interrogative, Positive

Hye zumo ? Did you eat it ? 2nd pers. sg.; high register (HR)

Huna zdmo ? Did you eat it ? 2nd pers. pl. HR

Imperative, Positive
Zumi! Eat it!

Zdmdm 1 Eat it !

2nd pers. sg.; extended register (ER)

2nd pers. pl. ER

The range of the pitch of the voice for affirmative clauses is referred to as normal
register. In saying the interrogative clauses, the whole range is higher, and a final
high tone syllable is said with the pitch rising from high to even higher; this is
implied by high register. The pitch range for the imperative clauses is higher and
has greater pitch intervals than in the case of the affirmative, but a final high tone
does not have pitch rising to even higher, as in the interrogative. The intonational
features of the imperative clause are characterized by the term extended register.
Where it is necessary to draw attention to these general intonational characteristics
of a clause, the abbreviations NR, normal register, HR, high register, and
ER, extended register, will be placed against the example. The six sentences
just given are all Aspect 1, and the pitch features referred to above with regard
to final high tone syllables can be heard in the following Aspect 3 (benefactive)
examples.

Affirmative, Positive, Aspect 3

Hye zumdn You ate it for him 2nd pers. sg. NR

Interrogative, Positive, Aspect 3

Hye zumdn ? Did you eat it for him ? 2nd pers. sg. HR

Imperative, Positive, Aspect 3

Ziunon ! Eat it for him ! 2nd pers. sg. ER

Interrogative examples suggesting a Yes or No answer, such as those already
given, sometimes have a final particle yd or &, giving rather greater emphasis.


86

JACK CARNOCHAN

If the question is repeated because no satisfactory response has been obtained,
then both particles may be used to end the clause, in the order yd &.

Affirmative, Negative

Tdd ziuni You didn’t eat it

Hundd zdmi You didn’t eat it

Interrogative, Negative

Tdd ztuni yo ? Didn’t you eat it ?

HanA& zdmi yo ? Didn’t you eat it ?

Imperative, Negative

T36d ztnni! Don’t eat it!

Udd h&nAA zdmi I Don’t eat it!

2nd pers. sg. NR

2nd pers. pl. NR

2nd pers. sg. HR

2nd pers. pl. HR

2nd pers. sg. ER

2nd pers. pl. ER

In the plural form, there are two tonal patterns for *Bee, rising from low to high
as marked above, and high level, *B6e. They seem to be in free variation.

Since affirmative mood clauses will be extensively dealt with in the later
sections below, there is no need to say more about them here. It is convenient,
however, to add further details in this section on both interrogative and imperative
clauses. Apart from the presence of the final particles yd and h, the interrogative
clauses have the same word structure and order as the affirmative. They differ in
intonation. There are other interrogative clauses, however, which are introduced
by one of a set of special words or groups of words, as in

1. Wend dd?

2. Wend nda n&n ?

Who did it? HR
Whom did he see ? HR

In both sentences an alternative intonation is Weno for the first word.

3. Mund nda di?
and Muno nda dd ?

4. Gedi nda Ji ?

5. Ydd nda dd ?

6. A nduwdd nda dd ?

7. Gd mund nda dd ?

What did he do ? HR

When did he come ? HR
How did he do it ? HR
How did he do it ? HR
Why did he do it ? HR

Here, too, there is an alternative with two high tones on muno.

A selection from among imperative mood clauses will show why mood
figures as the highest or most general category of the grammatical hierarchy
presented in this article.

1. Zum ddptd !

2. V6 Pwdddon ta !

3. Wudd kuda!

4. Liiyd duwey 1

5. Vii I

Eat the food ! Transitive

Give the money to Pweddon ! Ditransitive
Go to-morrow ! Intransitive

Get on the horse ! Semi-transitive

Hand it to me ! Give it to me ! Aspect 1


CATEGORIES OF THE VERBAL PIECE IN BACHAMA

87

6. Viiddn I Make a gift of it to me ! Give it to me ! Aspect 2
7. Dumi! Go out! Grade 1
8. Diund! Come out! Grade 2
9. Jik hdra ! Stew the meat! Singular
10. Jdkdm hdra ! Stew the meat! Plural
11. Vfiddn ’ (See 6, above.) The ‘ it ’ refers to a masculine noun
12. Viidd ! Give it to me ! The ‘ it ’ refers to a feminine noun
13. Nzd! Sit down! Positive
14. *Bd6 nzd! Don’t sit down ! Negative

These examples show that the categories of transitivity, aspect, grade, polarity
(but not tense), number, person (in so far as all imperative clauses are second
person) and gender, apply to imperative mood clauses as well as to those which
are interrogative or affirmative mood. And the different tonal patterns of the
disyllabic verbs Lydbi! ‘ Hit it! ’ Puri 1 ‘ Go in ! ’ and Wudd! ‘ Go ! ’ and of the
monosyllables Gd ’ ‘ Sing ! ’ and Bi I ‘ Lie down ! ’ show that a subclassification
of the verbs on tonal grounds is as useful in the description of imperatives as of
interrogatives and affirmatives. Further details of the forms of imperative mood
clauses will be given in the sections on each of the categories, where it is relevant.

CATEGORY II: TRANSITIVITY

Kwaa-Bwaare material shows certain differences in the shape of the verb
and of what follows it in the same clause that clearly correspond to differences in
the grammatical relationship. There are four sets of relations, and these are
dealt with by the four terms intransitive, semi-transitive, transitive and ditransitive
in the category of transitivity. It would be unsatisfactory to classify the verbs
lexically as transitive or intransitive, since the category applies to the verb stem
together with its suffixes and any items that may follow it in the same clause.

1. Intransitive

The text of a recording by Nikodimu Sondo Bukumdi1 begins with the clause
Bdtf-hina nzd g&ard a Geebfr ‘ When we lived over there in Gobir ’, and here the verb
nzd ‘ lived ’ has low tone, and the following gdard ‘ over there ’ is in intransitive
relation with it. If there were a transitive verb-object relation between them,
then the verb would have high tone. In sentence 7 of the same text, Dd Buriye

1 The text of Bukumdi’s recording is published in Carnochan, ‘ The coming of the Fulani :
a Bachama oral tradition op. cit. The numbers refer to the numbering of the sentences in the
text and translation in that article.


88

JACK CARNOCHAN

a 6d nza ‘ Then the Fulani stayed the verb is final and here too has low tone.
This clause also is intransitive. Disyllabic verbs occur in such intransitive clauses as

Duwey a liiyo The horse jumped
Liiyi 1 Jump ! Imp.
Nda purd He went in Aff.
Puri I Go inI Imp.
Hdmon a kdnyo The chief stood up Aff.
Kanyi 1 Stand up ! Imp.
Nzey a gi6d The boy ran away Aff.
Gi6i I Run ! Run away ! Imp.

These examples are all Aspect 1 clauses, and the affirmatives all end with a low
tone -d while the imperatives all end with a low tone -i. The following pair of
examples shows that this is not so for all verbs, and one unresolved problem is
whether or not they are Aspect 1 clauses, with verbs of yet another sub-class:
Nda wudd ‘ He went ’ and Wudo I ‘ Go ! ’ This verb also occurs as the first in a
sequence of two or more imperatives, as in Wuddo keni I ‘ Go and measure it! ’
where it has a long final vowel with falling tone, while another verb Zo ‘ to come ’,
exhibits neither of these features when it has a similar function, as in Z6 purl 1
‘ Do go in ! ’ Further research is necessary in this direction.

2. Semi-transitive

Sentence 47 of the Bukumdi text is Nda wudd a nzd haa kan ‘ He went and lived
in that place with him ’. Here the verb nzd has high tone, which would be said
to go with a transitive relation with the noun following. This sentence, however,
has an alternative, Nda wadd a nzd a hda kan and this is taken as evidence that the
relation of nzd with what follows is not purely transitive; it is considered as
semi-transitive. Similar intransitive and semi-transitive relations are shown by

1. Nda bd

2. Nda bd gdard

3. Nda bd a hdard

4. Nda bd haard

5. Nda bd a Fdre

6. Nda bd Fare

7. Nda tulo

8. Nda tulo a Nomwon
and Nda ttild Nemwon

He spent the night Intr.

He spent the night there lntr.

He spent the night in that place

•He spent the night at Fare Sen

He arrived Intr.

He reached Numan Semi-tr.

Imperative examples are Puro a hddye I ‘ Go into the house ! ’ with its alternative
Puro hedye I Cipo a hd6ye ! and Cfpo ha6ye I ‘ Dive under the water ! ’


CATEGORIES OF THE VERBAL PIECE IN BACHAMA

89

3. Transitive

Transitive clauses have either a nominal or a pronominal object, as

Nda nd homon He saw the chief Nda nan He saw him

Nda gboo kwaa-ha He closed the door Nda gbdord He closed it (/.)

In Grade 1 clauses, disyllabic verbs have no final vowel before a nominal object:

Dga ?us dapto Taa bel zumwey Taa 6ul bdrama Nda ijgdl sdlakey She cooked the food They threshed the corn They killed a hippopotamus He pulled the rope

Many verbs, like those in the four examples above, do not have a pronominal
object in Grade 1 clauses, as

Dga ?usd Taa bold Taa 6ulo Nda ijgdlo She cooked it They threshed it They killed it He pulled it

The Kwaa-Bwaare form is thus intransitive where the English equivalent is
transitive. Clauses with a final -n and a final high tone would have a different
meaning:

Dga ?nson Taa bdlon Taa buldn Nda ijgdldn She cooked it for him They threshed it for him They killed it for him He pulled it for him

and are analysed as Grade 3 (benefactive) clauses. Imperative clauses show
much the same details : ?Us dapto 1 Cook the food ! ?Bsi! Cook it! ?Uson ! Cook
it for him !

4. Ditransitive

In some examples, the verb is followed by two nouns or by two nominal phrases,
and there are usually clear formal distinctions as to whether they constitute one
object or two. By comparison with Taa tdw wey ‘ They wept for the death ’, one
can appreciate that the two nouns after the verb in the next example form a nominal
phrase object: Taa tdw wd bdagdn ‘ They wept for his father’s death ’. In the next
example, the two nouns constitute two objects : Nda vd Pwdddon ta ‘ He gave
Pweddon some money ’. Clauses such as this, where the verb is in relation with
two objects, are classified as ditransitive. Many Aspect 2, causative, clauses are
ditransitive, such as Nda ijgdl kwdareetd sdlakey ‘ He made the donkey pull the rope ’.
This can be compared with Nda ijgdl kwdareetaa homon ‘ He pulled the chief’s


90

JACK CARNOCHAN

donkey which is transitive and has just one object, consisting of two nouns in
genitive relation. One of the two objects of a ditransitive clause can be pro-
nominalized, as in Nda kundn hafiye ‘ He gave him some water or in Grade 2:
Nda ktinAn ha6ye ‘ He fetched some water for him Both the objects can be pro-
nominalized, and this will be illustrated by imperative clauses. The usual way of
saying ‘ Give it to me ’ is Vfi! in which the ‘ it ’ is not overtly expressed. One
also finds the forms, however, Viirdn 1 ‘ Give it (m.) to me ! ’ and Vfiturd ! ‘ Give
it (f.) to me ! ’ The corresponding plural form, used either when addressing
more than one person, or as a polite form when addressing one person only, is
Viimi! ‘ Give it to me, (please) ! ’

CATEGORY III: ASPECT

This category is set up to handle such differences as those between

Taa cfAw kAda
Taa cfAwdi kada

Taa tfAwi kAda

Taa cfAwgi kAda

Aspect 1 They cut down the tree

Aspect 2 They made me cut down the tree

Aspect 3 They cut down the tree for me

Aspect 4 They cut down the tree without my consent

The Aspect 1 example is transitive, the rest are ditransitive, and they are all
affirmative clauses, said with normal register. A similar interrogative set, spoken
with high register, could be given. Imperative contrasts are

V6 Pwdddon ta ’
Vddd Pweddon ta I
Lydbii!

MAcfgi I

Aspect 1 Hand Pweddon the money !

Aspect 2 Give Pweddon the money I
Aspect 3 Play (the drum) for me !

Aspect 4 Get up off my lap !

Aspect 1. Normal

Intransitive and semi-transitive Grade 1 clauses have a final -o vowel, as in
Nda mund ‘ He went back ’, Nda kAgo ‘ He stood up ’, Nda hwddld ‘ He went blind ’.
Transitive clauses have no final vowel in Grade 1 when the object is a nominal
phrase, as in Marka lyeb Pwdddon ‘ Marka hit Pweddon ’. In Grade 2 clauses, there
is a final high tone -A, as in Nda muna ‘ He came back ’, Nda kAgA gaard ‘ He went
and stood over there ’, Nda ggara 6eemey ‘ He picked up a cow’s horn ’.

Two particular cases may be mentioned here. The first is the reflexive form,
where the noun nd ‘ head ’ is used in the appropriate pronominalized form, as
Na yo6 na ‘ I washed myself’, Taa yeb ndgron ‘ They washed themselves ’. ‘ They
washed their heads ’ would require a different form, Taa y66 nyeemigron and the
two previous examples are considered as corresponding to the grammatical
‘ dimension ’ of reflexive forms. The second case is that of the reciprocal or mutual