10 Language in South Africa
10 Language in South Africa
10 Language in South Africa
Edited by
Rajend Mesthrie
University of Cape Town
The Pitt Building, Trumpington Street, Cambridge, United Kingdom
http://www.cambridge.org
This is a thoroughly revised and updated version of Language and Social History first
published in 1995 by David Philip Publishers (Pty) Ltd © Rajend Mesthrie and the
authors.
Contents
Introduction 1
v
vi Contents
Index 476
Maps
viii
Contributors
ix
x List of contributors
xi
xii Acknowledgements
1 Vowels
The vowel chart with IPA (International Phonetic Association) symbols:
2 Consonants
/ dental click
// lateral click
=| palatal click
! alveolar click (palato-alveolar/(pre-)palatal in Nguni)
’ glottal stop (Khoesan)
ʔ glottal stop (English)
x voiceless velar fricative (Khoesan)
x lateral click (Bantu; spelling form)
kx voiceless velar affricate
c palatal stop (Khoesan)
c dental click (Bantu; spelling form)
q palatal click (Bantu; spelling form)
ŋ velar nasal
ʃ voiceless alveopalatal fricative
voiced alveopalatal fricative
voiceless alveopalatal affricate
voiced alveopalatal affricate
xiii
xiv List of phonetic symbols
ɹ postalveolar approximant
voiced glottal fricative
j voiced palatal fricative
θ voiceless dental fricative
ð voiced dental fricative
(Because of the different traditions of scholarship some variation
is unavoidable.)
3 Diacritics
4 Non-phonetic symbols
∗ proto form
→ is rewritten as
< is derived from
> becomes
< > spelling form
/ / phonemic form
[ ] phonetic form
() optional element
Abbreviations
xv
xvi List of abbreviations
FT Flaaitaal
FUT future tense
FV final vowel
GEAR Growth, Employment and Redistribution
HAT Verklarende handwoordeboek van die Afrikaanse taal
HG High German
HSRC Human Sciences Research Council
IMP imperative
INF infinitive
IP inflectional phrase
ISAE Indian South African English
KZNED KwaZulu-Natal Education Department
L low tone
L1 first language
L2 second language
LANGTAG Language Plan Task Group
lit. literally
LOC locative
LWC language of wider communication
MCE manually coded English
ML matrix language
MLF matrix language frame
Mod. modifier
MOI medium of instruction
N (or n.) noun
N. Ng. Northern Nguni
NED Natal Education Department
NEPI National Education Policy Investigation
NGO non-governmental organisation
NLP National Language Project
NP noun phrase/National Party
NS Northern Sotho
nsA non-standard Afrikaans
nsE non-standard English
NZE New Zealand English
ODA Overseas Development Administration
OE Old English
+ OE feature of other L1 varieties of English
ON Old Norse
PAC Pan Africanist Congress
PANSALB Pan-South African Language Board
List of abbreviations xvii
PASS passive
PB Proto-Bantu
pl. plural
PNK Proto-Niger-Kordofanian
PRAESA Project for Alternative Education in South Africa
PRES present tense
PRP pre-prefix
PSB Proto-Southern Bantu
PSEB Proto-South-eastern Bantu
PWV Pretoria–Witwatersrand–Vereeniging
RDP Reconstruction and Development Programme
REL (or rel.) relative
RO rights and obligations
RP received pronunciation
S sentence
SABh South African Bhojpuri
SAE South African English
SAG South African German
SB Southern Bantu
SBE Southern British English
sE standard English
SEB South-eastern Bantu
sg (or sg.) singular
SS Southern Sotho
Sw. Swati
Tsw. Tswana
UNISA University of South Africa
USAID United States Agency for International Development
V vowel
v. verb
v.i. intransitive verb
VN verbal noun
v.t. transitive verb
VOC (Verenigde Oostindische Compagnie/Dutch East India
Company)
VP verb phrase
WSAE white South African English
Xh Xhosa
Z Zulu
ZE Zulu English
Introduction
This volume is the fifth in a group of books which aims to present a detailed
overview of the languages and language-related issues in specific territories. The
previous volumes, on the USA, the British Isles, Australia and Canada, have
successfully attained these aims, and have served as well-referenced introduc-
tions to those areas for students trained in linguistics as well as for general
readers. It is hoped that, despite the complexities of South African history and
language politics, the present volume will prove as useful a reference. It is my
brief in this introduction to make comparisons with previous volumes in the
series, and to outline the issues that make language a concern of the wider public
in South Africa.
1
2 Introduction
r Although the number of speakers of English as an additional language con-
tinues to grow, when speakers give up their language under pressure from
another language, it is not always towards English that the shifts occur. For
example, in some urban areas Tsonga and Venda speakers shift to the dominant
African language of the area, like Sotho. Chapter 15 by R. K. Herbert details
an ongoing shift from Tsonga to Zulu in some parts of the country.
r Dell Hymes’ lament (1981: vi) in his foreword to the American volume in
this series that there was not a single chair in the United States devoted to
the study of native American languages does not hold true in South Africa,
where departments of African languages are relatively large and numerous.
(However, many departments of African languages currently face a large
decline in enrolment.) Hymes’ remark does resonate for Khoesan languages,
which are not taught as subjects at South African universities. The number of
linguists acquainted with Khoesan structure is accordingly minuscule.
r There is greater pressure on other groups of people in South Africa to learn
an indigenous language than is the case in the UK, the USA, Canada or
Australia. Speakers of English and Afrikaans in rural areas often do learn an
African language ‘naturally’ from childhood, in some cases even before they
learn English or Afrikaans. Gough (1996) records the positive associations
that speaking Xhosa has for a white eastern Cape farming community, whose
vernacular English, especially among males, is peppered with Xhosa words,
phrases and ideophones. However, Kaschula (1989) believes that generally the
farming register of whites in the eastern Cape is a limited one that precludes
serious bonding with Xhosa employees.
r Some newspapers in African languages are quite successful in having a large
circulation, e.g. the Xhosa newspaper Imvo and the Zulu newspaper Ilanga.
Overall, though, the rate of functional literacy in South Africa is not high.
Harley et al. (1996) put the number of adults who have not completed primary
education at 7.45 million. Equating illiteracy with this level of seven years
of formal schooling, and with the total adult population estimated to be
26 million, this constitutes an adult illiteracy rate of 29 per cent.1
Deciding on a format for this book has not been straightforward. Indeed, looking
through the previous four volumes in this series, it is clear that there is no
overarching formula that will present the complexities of language distribu-
tion, description and function in the territories concerned. Ferguson and Heath
settled upon a simple formula for their USA collection: ‘American English;
Languages before English; Languages after English, Language in use’. Such
a formula would be highly controversial in the South African context, since it
would impose a misleading Anglocentric view of the country. Trudgill’s volume
Introduction 3
on the British Isles has as its major partitions ‘English; Celtic languages; Other
languages; and The Sociolinguistic Situation’. The volume on Canada begins
with a collection of chapters dealing with the most important current language
and language-related matters in a thematic way and then switches focus to its
ten provinces and two northern territories. This seems to work well in giving an
overview of language in the Canadian context. For South Africa it is doubtful
that this success can be repeated, since – with few exceptions – regional de-
scriptions of language in the nine provinces have yet to be done systematically.
The nine provinces themselves are only a few years old; and as maps 1.1, 1.2
and 1.3 show, the provincial boundaries of South Africa in the three periods – the
nineteenth century, the apartheid period and the post-apartheid era – differ quite
drastically. The format of the present volume comes closest to the Australian
volume which has the following headings: ‘Aboriginal and Islander languages;
Pidgins and creoles; Transplanted languages other than English; Varieties of
Australian English; Public policy and social issues’.
The division of this volume is partly historical and partly thematic. Part 1
comprises eight chapters on the main language groupings in the country:
Khoesan, Bantu, Afrikaans, English, Sign Language, German (as a represen-
tative of European languages, other than the two official ones) and Indian
languages (as representing some of the changes undergone by multilingual
Asian communities that came to South Africa). Part 1 thus may be considered
the foundations of the modern South African language mosaic, though it cannot
claim to be exhaustive.
Part 2 covers the theme of language contact in thirteen chapters. The focus
falls on the following:
(a) borrowing, mixing and switching between languages as well as on intercul-
tural communication norms and misconceptions;
(b) language change and shifts from one language to another in some commu-
nities, with particular reference to the role of gender;
(c) a closer study of the characteristics of two new varieties of English, which
owe their distinctiveness in no small measure to the particularities of colo-
nial and apartheid policies;
(d) the rise of new township codes, based on Afrikaans and/or the Bantu
languages of the country.
Part 3 deals with language planning, policy and education, with a special
eye on recent developments. In the early and mid-1990s planning and policy
were the key areas that occupied the attentions of sociolinguists. Part 3 is thus
a fitting way of rounding off this book by testing the heat generated at the
linguistic fireplace. It deals further with the rationale for the most multilingual
state policy in the world; the problems and obstacles associated with the policy;
and the vision required to put the policy into effective practice.
4 Introduction
3 TERMINOLOGY
Terminology pertaining to languages and social groups in South Africa – as in
some other countries – can be a minefield. In this respect language use clearly
reflects and replicates struggles over various kinds of political inequality, chiefly
involving gender, class and ethnicity. Readers in South Africa have become
accustomed to quotation marks, variant spellings and epithets like ‘so-called’,
‘officially classified’, and – now – ‘formerly classified’ in much academic
writing describing specific communities. These labels reflect the desire of many
academics not to ‘naturalise’ a largely arbitrary division among people, made
in the interests of apartheid. There is no consensus among contributors to this
volume about the appropriateness of the scare-quotes and the lack of capi-
talisation for the term coloured (which were meant to signify opposition to
the apartheid labels). For the sake of internal consistency and after much de-
bate we have settled on coloured, white and black with no further punctuation.
(Terms pertaining to forms of identification other than colour are given the
usual capitalisation: thus Afrikaner, Zulu or Indian.) This solution is by no
means perfect, since some political writers prefer to draw a distinction between
Black (a positive term for people of indigenous African descent) and black
(a positive term that embraced a sense of unity amongst Blacks, Indians and
coloureds against apartheid). Fortunately context usually makes it clear whether
the broader or the more usual narrower sense is intended. Synonyms for the term
‘black’ are numerous and have all run foul of the process of semantic derogation.
An early term, used without denigration by the missionaries of the nineteenth
century for the Nguni-speaking people, was Kaffir, based on the Arabic for
‘unbeliever’. The term eventually attained disrepute in popular parlance and is
considered highly offensive today. (In one of the library copies at my univer-
sity of the Dictionary of South African English, the pages containing a detailed
entry for this item were conspicuously crossed out – presumably by an enraged
student.) Other terms like native came to be used officially and colloquially in
the early twentieth century, but these too eventually became quite offensive.
Even today a linguist has to be wary of the connotations of the term ‘native
speaker’, especially ‘native speaker of an African language’. The more cir-
cumspect ‘mother-tongue speaker’ is the usual phrase one encounters in South
African sociolinguistic writing. Other synonyms were tried out by the apartheid
regimes, notably Bantu (from aba-ntu, the Nguni word for ‘people’, made up
of the plural prefix aba plus the root ntu for ‘person’). Because repressive
apartheid policies frequently contained this word (e.g. Bantu Administration
Board, Bantu education) and because it sounded grammatically incongruous
to hear it used as a singular form (a bantu), the word itself became associated
with apartheid, and went the same way as its predecessors. So strong was the
stigma attached to the word that linguists were in the uncomfortable position
of being just about the only ones using it, since it already denoted a particular
Introduction 5
sub-family of the Niger–Congo family, the largest in Africa. For a time the
term Sintu was promulgated as a more acceptable term for linguists, which
would do away with bantu altogether. This term (containing an appropriate
prefix si- for a language, and the root -ntu for ‘person’) never fully caught
on; though it is safe to say that Bantu is still a term one employs with care.
In this text it is used only as a technical term within historical linguistic dis-
cussion. However, we can take heart from a call from one academic (N. Maake
at a conference in 1998) that it is time people reclaimed the positive aspect of
the term bantu. (A student of mine, M. Ntleki, has reminded me, too, of the
names of prominent political figures like Bantu Stephen (Steve) Biko and Bantu
Holomisa.)
Our unholy grail does not end here. For a while in the 1970s apartheid
ideologues stressed the plurality of cultures and advocated the term ‘plural
development’ for their discriminatory homeland policy. Some wags began
referring to black people as ‘plurals’, and there was the linguistic joke en-
quiring whether Kaizer Matanzima, who was the first person to be installed
as a homeland leader, should be described as ‘the first person plural’. The
Dictionary of South African English on Historical Principles contains some
wonderful citations for ‘plural’:
1978 Drum (magazine) June 2 Just imagine overseas readers of South African newspa-
pers rolling on the floor in fits of laughter when read something like ‘The Dube hostel
is built to accommodate 10 000 single male Plurals’ . . .
1978 Sunday Times July 16: . . . Every Government Department has received a letter from
the Secretary for Plural Relations which says: ‘The Honourable the Minister of Plural
Relations and Development has indicated that the word “plural” must please under no
circumstances be used as a noun to mean “Bantu”.’
At about the same time, proponents of Black Consciousness were proposing
new terms like Azanian for the people of South Africa (from the root -zan,
found in words like Tanzania and Zanzibar) and Azania for the country. The
Azanian People’s Organisation remains part of the political landscape of what
is still ‘South Africa’.
The term African is a positive one that has many connotations and deno-
tations. In one sense it is used as a slightly more favourable term than black
(in the narrow sense). However, it can sometimes clash with the other sense
pertaining to people from the entire continent of Africa. It is also sometimes
contested as being too exclusive: one letter to the editor of the Cape Argus in
1998 complained that it was racist to limit the term to black people: African, it
argued, should mean any person born in Africa, not just a black person. In this
parlance black African would not be tautologous.
Related to the contested polysemy of African are the meanings of the terms
Afrikaans and Afrikaner, respectively ‘language of Africa’ and ‘person of
Africa’. Nowadays it is becoming quite common to hear claims that Afrikaans
6 Introduction
is an African language and an indigenous one at that. At stake here are ques-
tions of continued access to resources and support in educational institutions.
In one sense of ‘indigenous’, Afrikaans may well qualify, since its speakers
believe it to be a unique creation within Africa, which is not spoken out-
side southern Africa. How different Afrikaans is from Dutch and whether it is
really a separate structural entity, rather than a modification of Dutch, is not a
straightforward issue (see Roberge, chap. 4, this volume). In another sense of
‘indigenous’ and ‘African’, with all the connotations of not having had access
to resources previously and not being developed for use in higher education,
Afrikaans clearly falls on the other side. Finally, African, meaning ‘belonging
to Africa’, should not be confused with the technical linguistic sense of a com-
posite of the four families of Africa: Hamito-Semitic, Khoesan, Niger–Congo,
Nilo-Saharan.
The terminological problems do not end with the synonyms for ‘black’. The
colonial terms ‘Hottentot’ and ‘Bushman’ are also (mostly) in disrepute, an-
thropologists and linguists for a time preferring ‘Khoi’ and ‘San’ respectively.
Khoi was differentiated into ‘Khoi’ (for the language) and ‘Khoikhoi’ for
the people. However, since Khoikhoi etymologically means ‘men of men’
and San is a word that the San themselves did not use (and may well be
derogatory) there is much reason to tread warily. (One positive etymology is
the root sa-, ‘to inhabit, dwell, be located’, suggesting their primordial status.)
Archaeologists are gradually reverting to the term ‘Bushman’ in recognition
that ‘San’ might be no better in its connotations, and on the explicit preferences
of one group, the Ju/wasi (Parkington 1994: 209). Furthermore, Traill (chap. 2,
this volume) argues for the spellings Khoe and Khoekhoe, accepting Nienaber’s
arguments that this is the best representation of the phonetics, and is the form
preferred in Nama orthography. ‘Khoesan’ is a convenient term of reference
for the composite group of Khoekhoe and San, though it might misleadingly
imply a historical and cultural unity. See Traill’s important note 1 on a further
linguistic distinction between ‘Khoe’ and ‘Khoekhoe’.
There is ongoing debate about the use of prefixes for denoting African lan-
guages, and contributors to this volume have made their differing preferences
clearly known to me. For reasons set out by Herbert (1992: 6–7) and Bailey
(1995: 34–5) language names in this book will generally be used without pre-
fixes (Zulu rather than isiZulu). (One exception is the spelling Iscamtho favoured
by Ntshangase in this volume, for a variety that has not otherwise been commit-
ted to writing.) See further Herbert and Bailey (chap. 3 in this volume, note 3).
Finally, although it has been customary for two decades to refer to ‘South
African Black English’, ‘South African Coloured English’ and ‘South African
Indian English’, but just ‘South African English’ for the L1 variety of whites,
I follow de Klerk’s (1996) lead in opting for ‘South African English’ as a general
cover term, which can be prefaced by any ethnic or other descriptive label as
Introduction 7
necessary. Unfortunately the acronyms no longer roll off the tip of the tongue
(e.g. ISAE versus the older SAIE). The use of ethnic descriptors should not be
taken as unqualified acceptance of old apartheid labels – though few linguists
would dispute that the sociolects described here are very much still in existence.
However, we should be equally alert to the possibility of new non-ethnic forms
of English that might be developing, as seems to be happening with young
urban people at some schools, colleges and universities.
In concluding this section on disputes and changes in terminology, I am
struck by the aptness of Edwards’ (1998: 1) remarks in the previous volume on
the Canadian situation: ‘In some settings, disputes over language and culture
are largely symbolic; deeper problems between groups lie elsewhere, usually
in political or economic domains, and language, or religion, or tradition act
mainly as team jerseys.’
4 EDITORIAL NOTE
This book had its first incarnation as Language and Social History: Studies
in South African Sociolinguistics, published in Cape Town by David Philip in
1995. The present volume is a revised and updated version of that book. For
reasons of space and to accommodate some new research, six chapters of the
previous volume had to make way for five new ones. (The remaining chapters
have been revised and updated to varying degrees, some quite considerably.)
The editor wishes to stress that the six chapters from the previous volume not
included here are well worth study and are equally valid today. For reasons of
space, certain new topics could not be accommodated in the present volume.
For example, the status of Afrikaans in post-apartheid South Africa is a topic
of immense interest generally, and of pressing concern to some sectors of the
South African population. (On this issue the reader is referred to van Rensburg
1999. In this volume the status of Afrikaans has been discussed as part of the
unfolding new language dispensation.)
note
1 The authors defined an adult as someone over fifteen.
bibliography
Bailey, R. 1995. ‘The Bantu languages of South Africa: towards a sociohistorical
perspective’. In R. Mesthrie (ed.), Language and Social History: Studies in South
African Sociolinguistics. Cape Town: David Philip, pp. 19–38.
de Klerk, V. 1996. Focus on South Africa (Series: Varieties of English around the World).
Amsterdam: Benjamins.
Edwards, J. 1998. Language in Canada. Cambridge: Cambridge University Press.
Ferguson, C. A. and S. B. Heath 1981. Language in the USA. Cambridge: Cambridge
University Press.
8 Introduction
Gough, D. 1996. ‘The English of white eastern Cape farmers in South Africa’. World
Englishes, 5, 3: 257–65.
Harley, A., J. Aitchison, S. Land and E. Lyster 1996. A Survey of Adult Basic Education
in South Africa in the 1990s. Cape Town: Sached Books.
Herbert, R. K. 1992. ‘Language in a divided society’. In R. K. Herbert (ed.), Language
and Society in Africa: The Theory and Practice of Sociolinguistics. Johannesburg:
Witwatersrand University Press, pp. 1–19.
Hymes, D. 1981. ‘Foreword’. In Ferguson and Heath, pp. v–ix.
Kaschula, R. 1989. ‘Cross-cultural communication in a north-eastern Cape farming
community’. South African Journal of African Languages, 9, 3: 100–4.
Parkington, J. 1994. ‘San’. In Saunders (ed.), pp. 208–9.
Romaine, S. 1991. Language in Australia. Cambridge: Cambridge University Press.
Saunders, C. C. 1994 (ed.). An Illustrated Dictionary of South African History.
Johannesburg: Ibis Books.
Silva, P. 1996 (ed.). A Dictionary of South African English on Historical Principles.
Oxford: Oxford University Press.
Trudgill, P. 1984. Language in the British Isles. Cambridge: Cambridge University Press.
van Rensburg, C. 1999. ‘Afrikaans and Apartheid’. International Journal of the
Sociology of Language, 136: 77–96.
Part 1
R. Mesthrie
1 LANGUAGE PROFILE
South Africa has been the meeting ground of speakers of languages belonging
to several major families, the chief ones being Khoesan, Niger–Congo, Indo-
European and Sign Language.1 (It is surely time to include Sign languages in
our genealogies of language, and to devote as much space to them as to any
other language family in our sociolinguistic surveys.) The Khoe (formerly called
‘Hottentot’) and San (a.k.a. ‘Bushman’) languages, thought to be historically
unrelated (and in fact divisible into three families) are now, with very few
exceptions, close to extinction. The Bantu languages (belonging to the wider
Niger–Congo family) are the numerically predominant languages of the country,
comprising essentially the following:
r the Nguni cluster (Zulu, Xhosa, Swati, Ndebele);
r the Sotho cluster (North Sotho, South Sotho, Tswana);
r Tsonga;
r Venda.
(See map 15.1 for the main distribution patterns of these languages.) The term
‘cluster’ denotes a set of varieties that are closely related along linguistic lines
(though in terms of socio-political status the varieties may be quite independent).
In addition to these official languages a number of Bantu languages are spoken
in smaller numbers by migrant mineworkers from neighbouring countries, and
by more recent immigrants. Such languages include Chopi, Kalanga, Shona,
Chewa, etc. Still other special cases exist: Phuthi, for example, is a minority
language of the eastern Cape, more widely represented in the neighbouring
country, Lesotho (Donnelly 1999); Makhuwa and Yao are languages spoken in
Durban by the descendants of ex-slaves from Mozambique dating back to the
1870s (Mesthrie 1996).
The Indo-European family in South Africa has members of the Germanic
branch (English and Afrikaans, and, to a lesser extent, German), the Indic branch
(Hindi, Urdu, Gujarati and Konkani among others) and the Romance branch
(chiefly Portuguese, spoken to varying degrees by immigrants from Angola,
11
12 R. Mesthrie
2 LANGUAGE STATISTICS
The 1996 census showed an improvement in its language question over its
predecessors, since it attempted to elicit whether respondents ‘spoke more than
one language at home, and if so, what was the next most often spoken language’
(Census Database 1996). However, even this does not go far enough in parts of
the country where many individuals are proficient in several languages. In urban
areas like Gauteng it is quite common to receive answers like the following from
students from Gauteng about the languages they are proficient in:
My father’s home language was Swazi, and my mother’s home language was Tswana.
But as I grew up in a Zulu-speaking area we used mainly Zulu and Swazi at home. But
from my mother’s side I also learnt Tswana well. In my high school I came into contact
with lots of Sotho and Tswana students, so I can speak these two languages well. And
of course I know English and Afrikaans. With my friends I also use Tsotsitaal. (Twenty-
three-year-old male student from Germiston)
Nguni languages
Ndebele 586,961 1.5
Swati 1,013,193 2.5
Xhosa 7,196,118 17.9
Zulu 9,200,144 22.9
Sotho languages
North Sotho 3,695,846 9.2
South Sotho 3,104,197 7.7
Tswana 3,301,774 8.2
Tsonga 1,756,105 4.4
Venda 876,409 2.2
Afrikaans 5,811,547 14.4
English 3,457,467 8.6
other 228,275 0.6
unspecified 355,538 −
TOTAL 40,583,573
neighbourhoods. In the course of moving from area to area with her family,
she had attended schools in which the dominant African languages were: North
Sotho (up to Standard 1 = grade 3); Tswana (up to Standard 5 = grade 7);
South Sotho (up to Standard 6 = grade 8); North Sotho again (up to Standard
8 = grade 10) and Tswana again (up to Standard 10 = grade 12).
At that time only bilingualism in English and Afrikaans was taken seriously
by the apartheid censuses. For example, the census figures for African languages
in the 1991 census excluded speakers from the Transkei, Bophutatswana, Venda
and Ciskei homelands. The presentation of language demographics in abbre-
viated tabular form should not be allowed to conceal the essentially dynamic
nature of language use in any society. Language statistics must always be in flux
with large-scale movements in and out of the country, with shifts in language
preferences, and above all the very fluid multilingual nature of communication
(with changing preferences and the birth of new codes) within countries like
South Africa.
3 SOCIOHISTORICAL PROFILE
for three distinct families of languages within this traditional designation (see
chap. 2, n. 2). Khoesan peoples may have originated further north – the ar-
chaeological and linguistic evidence suggests northern Botswana; and two San
languages (Sandawe and Hadza) are still to be found as far north as Tanzania.
Khoesan and Bantu contacts in southern Africa were extensive, as suggested
by Parsons: ‘The Bantu-speaking peoples of southern Africa are inheritors of
Khoisan ancestry and culture, which may be seen not only in their physical
appearance but in their religions and medical ideas and in their folk tales about
wild animals’ (1982: 19).
Relations between Khoesan and later southern African settlers varied: the
above quotation suggests that relations must have been mostly peaceful (if sub-
servient) with the Bantu-speaking peoples (see Herbert, chap. 15, this volume).
Relations with European settlers were less benign, leading to the ultimate de-
struction or radical transformation of Khoekhoe and San society. There are
no Khoe languages spoken in South Africa today; Nama – still spoken in
Namibia – may be described in colonial parlance as the last of the Hottentot
languages. San languages do survive in Namibia, Botswana and elsewhere, and
in ever-shrinking numbers in South Africa. Their speakers may have largely
shifted to Afrikaans, but they often retain a distinctive identity.
The Bantu languages of South Africa are classified as part of the Niger–
Kordofanian family, spoken over two to three thousand years ago in what is
today the Cameroon–Nigeria region. Iron Age civilisation was brought south
of the Zambesi and Limpopo by small numbers of Bantu-speaking farmers who
first appeared a few centuries ad (see chap. 3).
A key event in modern South African history was the establishment by the
Dutch, the richest European trading nation of the time, of a trading station at
the Cape in 1652. Prior to this there had been stopovers by Portuguese and
English sailors for the purpose of refreshment and recuperation. As a result
a jargon form of English with words from Portuguese and Dutch came to be
known by the locals well before 1652 (den Besten 1989). Although the Cape
was initially regarded as only a refreshment post, it soon developed into an
extensive colony, and as such required government. The settlement at the Cape
came to include in time a large proportion of Germans and Huguenot French
refugees, and other Europeans in small numbers, all of whom formed a new
Cape Dutch community, for convenience simply labelled ‘Dutch’ here. Strife
soon followed between Dutch and Khoesan over land and cattle. The Dutch
had to look elsewhere for labour for the new colony: they imported slaves in
large numbers from 1658 onwards from Madagascar, Mozambique, the East
Indies and India. It is one of the ironies of history that at about the time that
large numbers of African slaves were being forcibly exported out of Africa into
the New World, the southern tip of Africa was itself stocking up on slaves from
the East. The slave population of the Cape was possibly one of the most diverse
South Africa: a sociolinguistic overview 15
in the world in terms of origins, religion, culture and language. The roots of
the large coloured population of the western Cape go back to this period, with
a multiple ancestry that involves the Khoesan, Eastern and African slaves, and
the offspring of European and non-European. The Khoesan were to a large
extent reduced in numbers because of conflicts with the Dutch and the effects
of European diseases, notably the smallpox epidemic of 1713.
Eastward expansion took Dutch farmers away from the small colony at the
Cape, and into conflict with the Xhosa in the late eighteenth century. In 1795, at
the time of the Napoleonic wars, British forces captured Cape Town, and took
over the colony as a naval base. With the ensuing peace of 1803, the colony
was handed back to the Dutch, but not for long; Cape Town was recaptured
by the British in 1806. From this time European missionary activity became
significant, with the first schools for black and coloured people being set up on a
small scale. The first purely civilian British population came later in 1820, with
poorer sections of British society being settled in the eastern Cape, far from
the polite society of Cape Town. These eastern Cape settlers became embroiled
in frontier wars with the Xhosa. The roots of South African English go back
largely to this settlement (see chap. 5). The British followed an Anglicisation
policy in the Cape, replacing Dutch with English as the language of government,
education and law. This was one of the causes of Dutch discontent. Feeling their
religion, culture and language under threat, and with their right to keep slaves
eroded with the emancipation of 1834, as well as for other economic reasons,
Afrikaners trekked further into the interior with the intention of escaping British
influence. By this time Afrikaans had evolved as a colloquial variety of Dutch,
with admixture from other languages. As early as 1707 Hendrik Bibault had
declared, Ik ben een Africaander – ‘I am an Afrikaner’ (Prinsloo 1994: 7–9).
Afrikaans culture, which had evolved out of the Dutch and slave experience
in Africa, gelled as people moved away from Cape Town. (This was the same
period in which Europeans were expanding – also via wagons – into the interiors
of South America and Australia.)
The period from the 1820s onwards is regarded as one of great flux in political
alignments among the indigenous Bantu-speaking peoples. Traditional history
recounts the rise to power of Shaka in consolidation of a Zulu empire in Natal.
He was both a powerful and shrewd military leader, and a despot according
to most accounts. The consolidation of a Zulu unity led to conflicts with other
chieftains, and this is known as the period of the Mfecane (an Nguni word for
‘great wandering, dispersion of people’). Of particular note is the trek of the
Ndebele people away from Zulu territory to the highveld, and subsequently
away from Afrikaner firepower into what is now south-western Zimbabwe.
Ndebele is today spoken in Zimbabwe and northern parts of South Africa
(especially the former Kwa Ndebele homeland). Another victim of the Mfecane
are the Mfengu, believed to have fled from Zululand to the eastern Cape to live
16 R. Mesthrie
as clients of the Xhosa and the colonists. Their language tends to be classified as
a social dialect of Xhosa, rather than Zulu. The historian Julian Cobbing (1983)
has criticised the Mfecane thesis, arguing that it was popularised by colonial
historians, as a legitimisation of white conquest. The upheavals of the time, he
argues, were not so much related to Shaka’s rise to power as to the penetration of
commercial capitalism, including covert slave-trading. Critics of the Cobbing
thesis are unhappy about the lack of substantial evidence in its favour.
The 1820s onwards was the period when African languages were being writ-
ten down for the first time by missionaries, in conjunction with local consultants.
It was an exciting and taxing time for linguists among the missionaries, who
battled to come to terms with the unfamiliar structures of African languages. For
example, the principle of alliterative or euphonic concord – elaborate agreement
between prefixes of subject nouns with verbs and other entities like adjectives,
and genitival and relative nouns – was only discovered over thirty years after
the first missionaries arrived in the eastern Cape. Reverend John Bennie pub-
lished a monograph in 1826 entitled A Systematic Vocabulary of the Kaffrarian
[= Xhosa] Language in Two Parts; To Which is Attached an Introduction to
Kaffrarian Grammar, in which he came close to discovering the principle, with-
out actually hitting on it (Doke 1959). Reverend William Boyce, who arrived
in South Africa in 1830, published his discovery of the principles of concord
in 1834, eight years after Bennie’s work.
The earliest Xhosa written texts were translations of the Gospels. In many
territories the dialect selected by the missionaries for writing came to have
prestige because of this association. The rise of African languages thus did not
follow from the more familiar bases of standardisation familiar in the West:
urbanisation and the prestige accruing from the economic and social status of
certain groups of speakers. Rather, it was based on the external force of mis-
sionary influence. This has developed into a modern-day paradox: the standard
varieties of African languages are associated with the rural areas, which are
no longer centres of prestige. High-status blacks are more likely to be urban-
wise ‘modern’ people, who speak English and non-standard urban varieties of
African languages, showing extensive borrowing of vocabulary, code-switching
and neologisms. The question can thus be raised whether the standardisation of
African languages via the mission presses, sermons and nineteenth-century dic-
tionaries may have taken place too early to be effective as a norm representing
black social and political aspirations.
From the late 1840s onwards a second British settlement took place, this
time in Natal, which had been annexed from the Afrikaners by the British in
1843. Described as largely ‘impecunious aristocrats’ by the historian Hattersley
(1940), they were of different regional and social origins from the earlier
settlers in the eastern Cape, and more mindful of the social symbols and
system of Victorian England (Lanham 1978: 158). Lanham locates many of the
more prestigious phonetic developments of twentieth-century South African
South Africa: a sociolinguistic overview 17
English as emanating from this group. Although many British children born in
Natal learnt Zulu, a new pidgin form, then called Kitchen Kaffir (modern-day
Fanakalo), stabilised in Natal out of contacts between the English, Zulus and
Afrikaners. The colonists needed to find a cheap labour source other than among
the local Zulus, whose men initially resisted cheap manual labour. The Natal
government looked to India as a source of cheap labour, and between 1860 and
1911 over a hundred and fifty thousand Indian people were brought to Natal.
For the greater part of the twentieth century the population of Indians exceeded
that of whites in Natal province.
The trekking Afrikaners eventually established the republics of the Transvaal
and Orange Free State in the 1850s. Although they had chosen to escape British
domination, and had installed Dutch as the official language of the republics,
the influence of the English language was still strong. For example, one of the
trekkers, Anna Steenkamp, kept a diary in English. The Bloemfontein news-
paper of the time, The Friend of Sovereignty, continued to be published in
English (Parsons 1982: 119). Lanham (1978: 119) mentions that parents in
Pretoria were demanding more English and less Dutch in their schools, up
to the 1890s. Meanwhile, in the 1870s in the Cape a tradition of writing in
Afrikaans rather than Dutch was emerging, with the formation of the Fellow-
ship of True Afrikaners in Paarl, outside Cape Town. It is another irony of
history that Afrikaans was first substantially written by the descendants of
Muslim slaves, who used Arabic script in writing Afrikaans religious texts.
According to Davids (1990: 1) seventy-four such texts are extant, the bulk of
them produced between 1868 and 1910.
The 1860s are better known as the period of the discovery of enormous
deposits of precious metals in the interior. The scramble to gain possession
of the new wealth brought Britain into conflict with the Afrikaner republics.
The Transvaal was annexed as a British colony in 1877. Afrikaner nationalism
gelled in this period with the resentment at British rapacity. Two wars were
fought over control of the land and its wealth, in 1881, when the Afrikaners
won back the Transvaal, and between 1899 and 1902 (in what is now called
‘the South African War’) when they were heavily defeated and maltreated. In
1879 a British force had invaded Zululand to protect its new Transvaal colony
from a supposed Zulu threat. This was the offensive that brought about the
final subjection of black people in the nineteenth century. The late nineteenth
century saw urbanisation on a large scale, with a large influx of Europeans of
Christian and Jewish faith. It also saw a large-scale movement of black people
into the mining areas. Parsons (1982: 148) cites a visiting British historian’s
description of Kimberley, the centre of the diamond industry, in 1895: ‘Here
in the vast oblong compound, one sees Zulus from Natal, Fingoes, Pondos,
Tembus, Basutos, Bechuanas, Gungunhanas, subjects from the Portuguese
territories, some few Matabili and Makalaka, and plenty of Zambesi boys
from the tribes on both sides of that great river. There were 2, 600 workers
18 R. Mesthrie
in the compound from as far north as Lake Tanganyika.’ There were also
Indians (who were not legally permitted to venture into the interior), Chinese,
and people from many parts of Europe, the USA and Australia. People of
Khoesan ancestry also did not escape the lure of the mines, in particular the
Korana and the Griqua, by then bilingual in Afrikaans and Kora/Gri. In this
great babel the pidgin Fanakalo, which had originated in the eastern Cape and
Natal, was particularly useful. The mining industry must have also sown the
seeds for new mixed urban varieties of African languages that were to become
more prominent in the twentieth century. Lanham argues convincingly that
the mining industry brought three different strands of English together (Cape
English, Natal English and, to some extent, RP), in ways that laid the foun-
dations for the twentieth-century continuum of (white) South African English
varieties.
Alfred Milner took over the administration of the conquered Boer republics
and ruled South Africa from Johannesburg between 1901 and 1905. One of his
aims was to anglicise the Afrikaners and bring them into the fold of the British
Empire. He emphasised English over Dutch in the schools. State education
was aimed at whites; the education of black people was left to the churches
and mission schools. In the wake of the atrocities of the South African War,
Afrikaners resisted Milner’s anglicisation policy. The status of Afrikaans as
bearer of local cultural values and the identity of an Afrikaner nation began to
gain prominence.
The rapid growth of capitalism in the early twentieth century drew increas-
ingly more rural people into wage labour. There were vastly disparate wages
for white and black workers (Parsons 1982: 225). The Union of South Africa
was formed in 1910, combining the two former Boer republics and the British
colonies of the Cape and Natal into one state. The state oversaw the further
dispossession of black people from their land. The Land Act of 1913, which set
aside most of the country’s land for control by whites, destroyed the economic
independence of black people. The official languages of the Union were Dutch
and English. Afrikaans was not recognised as an official language until 1925,
when it replaced Dutch in that capacity.
The apartheid governments of 1948 onwards enforced separation of peoples
along the lines of colour, with the Group Areas Act of 1950 and the pass laws.
The latter were aimed at channelling black male labourers to where they were
needed (industries and white farms), while keeping their families in the rural
areas. The 1940s saw the rapid growth of townships like Moroka, which later
formed a central part of Soweto.
The Bantu Education Act of 1953 tried to create a permanent underclass of
black people by placing rigid controls over syllabi and the media of instruction.
Equally cynically, it enforced the closure of the mission schools which offered
quality education (albeit in small numbers) to black people, often on non-
racial lines. Such sociopolitical arrangements clearly influenced the course of
South Africa: a sociolinguistic overview 19
itself unsound. Problems lay in the way the policy was implemented, and in the
manner in which the wishes of parents were ignored. A UNESCO document of
1953, entitled ‘The use of vernacular languages in education’, was, at about the
same time, stressing the value of mother-tongue education in the early years of
schooling. The humanist orientation of the UNESCO document was, however,
sadly lacking in Bantu education policy.
Resistance to Bantu education and the language policy it attempted to impose
led to the Soweto uprisings of 1976. The 1970s and 1980s became a period of
intense struggle against white domination, in which schoolchildren played a
prominent role. It is worthy of note that the event that led to the eventual arrival
of democracy to the country in 1994 should have its inspiration in a linguistic
protest against Afrikaans as a medium of instruction.
Since it was being widely used by the anti-apartheid political leadership,
English became the language of unity and liberation. Although black schoolchil-
dren had pride in their home languages, the latter had become too closely
connected with the divide-and-rule policy of apartheid to be considered as
languages of educational and economic progress. With the negotiations that
led to the first democratic elections of 1994, it was English that was the de
facto lingua franca. The African National Congress (ANC) leadership seemed
at one time to be heading for a policy with English as the only official lan-
guage. Language was not a great priority for the ANC in the way it was for
parties representing the Afrikaner power bloc. The position of Afrikaans be-
came an important negotiating chip during negotiations (Crawhall 1993). At the
same time many educators and sociolinguists put their weight behind cultural
and linguistic pluralism. Empowering the majority of South Africans meant
empowering their languages too. A policy with English as the only official
language would have been anathema to many Afrikaans speakers. However,
having English and Afrikaans as the official languages would have given off
signals to the majority of the population that nothing had changed. Clearly
if English and Afrikaans were to remain as official languages, there was a
strong case for some African languages to be given the same status. The classic
dilemma of multilingual colonised societies then presented itself: which of the
African languages should be chosen? The politicians’ solution was to opt for
all nine of the major African languages (listed below). Whether this was an
enlightened decision or one of political and symbolic expediency, taken in the
hope that English would become the de facto working language of state, will be-
come clear in the years ahead. One possible solution that generated a great deal
of debate was a proposal by Neville Alexander (and made earlier by a politician,
Jacob Nhlapo) that a new standard Nguni language be enhanced, made up of the
‘cluster’ of Zulu, Xhosa, Swati and Ndebele; as well as a new Sotho standard
based on North Sotho, South Sotho and Tswana. This would have the satisfying
outcome of having two major African languages (plus the smaller Venda and
South Africa: a sociolinguistic overview 23
(2) Recognising the historically diminished use and status of the indigenous lan-
guages of our people, the state must take practical and positive measures to elevate
the status and advance the use of these languages.
(3) National and provincial governments may use particular official languages for the
purposes of government, taking into account usage, practicality, expense, regional
circumstances, and the balance of the needs and preferences of the population
as a whole or in respective provinces; provided that no national or provincial
government may use only one official language. Municipalities must take into
consideration the language usage and preferences of their residents.
(4) National and provincial governments, by legislative and other measures, must
regulate and monitor the use by those governments of official languages. Without
detracting from the provisions of subsection (2), all official languages must enjoy
parity of esteem and must be treated equitably.
(5) The Pan South African Language Board must –
(a) promote and create conditions for the development and use of
(i) all official languages
(ii) the Khoi, Nama and San languages; and
(iii) sign language
(b) promote and ensure respect for languages, including German, Greek,
Gujarati, Hindi, Portuguese, Tamil, Telugu, Urdu, and others commonly used
by communities in South Africa, and Arabic, Hebrew, Sanskrit and others
used for religious purposes.
However, as the major public sectors are discovering, social change within
this broad vision is not easy to achieve in the short term, especially within a
troubled local economy and global economic pressures. The key question for
linguists and educators is the extent to which the new constitutional flexibility on
language can be put into effective practice. In some respects language policy and
practice are in flux in the post-1994 era, with many sectors still experimenting
with the most effective and the least divisive language options. A vivid picture
of the transition in the defence force in one eastern Cape centre is given by
de Klerk and Barkhuizen (1998), where although Afrikaans is being overtaken
by English, there is still room for its use and new spaces are being opened
for Xhosa in the eastern Cape. As far as education is concerned institutions at
school, college and university level previously employing an ‘Afrikaans-only’
medium have had to rethink their policies in terms of the constitution, and
post-apartheid economic realities.
As with other public sectors energies in language education are now be-
ing focused away from negotiation and planning to ‘delivery’. Two important
language initiatives in this regard are the Pan-South African Language Board
(PANSALB) and the Language Plan Task Group (LANGTAG). PANSALB is a
permanent body established in terms of the constitution as a proactive agent for,
and watchdog over, linguistic rights. LANGTAG was a short-term initiative of
the Department of Arts, Science, Technology and Culture (DACST). Its brief
South Africa: a sociolinguistic overview 25
was to advise the minister (then Ben Ngubane) on planning for policy making
within the language guidelines of the new constitution.
LANGTAG brought together a broad range of language practitioners (includ-
ing sociolinguists) enabling comprehensive consultations with different com-
munities and sectors, intensive discussions and some new research. The task
groups presented reports on the following areas: language services; language
equity; language as an economic resource; heritage and sign languages; edu-
cation; and the position of African languages. The consolidated final report
and individual reports have been published by the Human Sciences Research
Council (HSRC) in conjunction with the DACST. The LANGTAG dossier thus
forms an important foundational set of research documents for the (macro)
sociolinguistics of post-apartheid South Africa. Other major resources include
the submissions to PANSALB (upon general invitation) by a range of cultural,
educational, political and language organisations. Furthermore, in response to
its call for written submissions the Constitutional Assembly had by March 1996
received over a thousand responses from members of the public expressing their
wishes regarding the country’s language policy. It can safely be said that plan-
ning and policy was the aspect of language study most in the public eye in the
1990s.
notes
1 Strictly speaking, Khoesan is not a ‘family’ but a ‘phylum’. That is, it is a loose
cover term for a group of families showing cultural and geographical cohesion, but
for which no linguistic unity has been proven (see Traill, chap. 2, this volume). Some
linguists (including Herbert, chap. 3, this volume) feel safer describing Niger–Congo
as a phylum rather than a family.
2 Describing North Sotho as ‘Pedi’ may have been an error, as Pedi is but one di-
alect of the language. Nowadays the term ‘North Sotho’ is being increasingly used
officially.
bibliography
Census Database 1996. http://www.statssa.gov.za.
Davids, A. 1990. ‘Words the slaves made: a socio-historical–linguistic study’. South
African Journal of Linguistics, 8, 1: 1–24.
Cobbing, J. 1983. ‘The case against the mfecane’. Seminar paper, Centre for African
Studies, University of Cape Town.
Crawhall, N. 1993. ‘Negotiations and language policy options in South Africa’. Cape
Town, National Language Project (unpublished document).
1996. ‘Alien tongues’. Bua, 10, 2: 4–7.
de Klerk, V. 1996 (ed.). Focus on South Africa. Amsterdam: Benjamins.
de Klerk, V. and G. Barkhuizen 1998. ‘Language attitudes in the South African National
Defence Force: views from the Sixth South African Infantry’. Multilingua, 17, 2–3:
155–80.
26 R. Mesthrie
den Besten, H. 1989. ‘From Khoekhoe foreigner talk via Hottentot Dutch to Afrikaans:
the creation of a novel grammar’. In M. Pütz and R. Dirven (eds.), Wheels Within
Wheels. Frankfurt: Peter Lang, pp. 207–54.
Doke, C. M. 1959. ‘Language pioneers of the nineteenth century’. African Studies, 18,
1: 1–27.
Donnelly, S. 1999. ‘Southern Tekela Nguni is alive: reintroducing the Phuthi language’.
International Journal of the Sociology of Language, 136: 97–120.
Finlayson R. and S. Slabbert 1997. ‘ “We just mix” – codeswitching in a South African
township’. International Journal of the Sociology of Language, 125: 65–98.
Hartshorne, K. 1995. ‘Language policy in African education: a background to the
future’. In R. Mesthrie (ed.), Language and Social History: Studies in South African
Sociolinguistics. Cape Town: David Philip, pp. 306–18.
Hattersley, A. F. 1940. Portrait of a Colony. Cambridge: Cambridge University Press.
Lanham, L. 1978. ‘South African English’. In L. W. Lanham and K. P. Prinsloo
(eds.), Language and Communication Studies in South Africa. Cape Town: Oxford
University Press, pp. 138–65.
Mesthrie, R. 1996. ‘Lessons in survival: 120 years of Makhuwa and Yao in South Africa’.
Bua, 10, 2: 14–16.
Mohamed, Y. 1997. The Teaching of Arabic in South Africa – History and Methodology.
Cape Town: University of the Western Cape (Department of Arabic Studies).
Parsons, N. 1982. A New History of Southern Africa. London: Macmillan.
Population Census 1996. ‘The People of South Africa – Census in Brief’. Report
No. 03-01-11. Pretoria: Statistics South Africa.
Prinsloo, D. 1994. Afrikaners. In Saunders (ed.), pp. 7–11.
Saunders, C. C. 1994 (ed.). An Illustrated Dictionary of South African History.
Johannesburg: Ibis Books.
UNESCO 1953. ‘The use of vernacular languages in education – a report’. Paris:
UNESCO.
van Rensburg, C. 1999. ‘Afrikaans in post-apartheid South Africa’. International Journal
of the Sociology of Language, 136: 77–96.
van Wyk, E. B. 1978. ‘Language contact and bilingualism’. In L. W. Lanham and
K. P. Prinsloo (eds.), Language and Communication Studies in South Africa. Cape
Town: Oxford University Press, pp. 29–52.
2 The Khoesan languages
A. Traill
1 INTRODUCTION
The sociolinguistic story of the South African Khoesan1 languages is one of
language death (Dorian 1989), and finds its place in the discussion of language
death in Africa (Dimendaal 1989, Brenzinger 1992, Brenzinger et al. 1991). In
the case of many of the Cape Khoekhoe languages or dialects, historical and
other records have been rich enough to permit some quite specific sociolinguis-
tic reconstructions of the circumstances attending their death. However, there
is not much of a sociolinguistic texture that can illuminate the well-known his-
torical record of the holocaust that finally obliterated the speakers of the /Xam
Bushman dialects in the space of forty-odd years, between 1875, when W. H. I.
Bleek and Lucy Lloyd worked with the rich (albeit threatened) language, and
about 1911, when Dorothea Bleek visited the last few speakers in Prieska and
Kenhardt. Although a contributing factor to the death of /Xam was undoubtedly
the extermination of many of its speakers, it is generally possible only to spec-
ulate about other conditions that destroyed the language. This applies to the
other Bushman languages of South Africa, with the added difficulty that many
of them were so inadequately documented that we cannot even be sure about
their exact linguistic status.
Thanks to the extensive surveys of Köhler (1981), Westphal (1971) and Winter
(1981), we have detailed surveys of most of the Khoesan2 languages that
are extinct or extant. In South Africa itself, the Khoesan languages are rep-
resented today only by speakers of a Nama dialect in the Richtersveld, and
along the Orange river in the northern Cape3 and by a handful of speakers
of /’Auni and =| Khomani, closely related Southern Bushman languages of
the Kalahari Gemsbok Park, Gordonia district in the Northern Cape Province.
The Richtersveld Nama speakers are bilingual in Afrikaans and Nama, and it
appears that Nama is the dominant language for the ‘Boorlinge’ (not recent im-
migrants) speakers. Until the 1950s children were monolingual in the language,
27
28 A. Traill
but the effect of compulsory school education in Afrikaans may have led to
a change in language loyalty. A recent survey reports other first and second
language Nama-speaking communities, all bilingual in Afrikaans, from Port
Nolloth on the Atlantic eastward to Pella on the Orange river and into Gordonia
(Crawhall 1997: 22). The number of speakers (more accurately ‘semi-speakers’)
of /’Auni and =| Khomani (perhaps a dozen) is very small, and the language
is on the verge of extinction. Probably all the surviving speakers are more
or less trilingual, to some degree in /’Auni or =| Khomani, and in Nama and
Afrikaans.
Richtersveld and Orange river Nama are all that is left of the Khoekhoe
linguistic tradition that included the many Cape Khoekhoe dialects as well as
Nama spoken up the west coast to Namaqualand and beyond into Namibia, and
!Ora and Gri spoken to the east along the Orange, Vaal, and Harts rivers. /’Auni
and =| Khomani are the closest linguistic relatives of /Xam: their imminent disap-
pearance will complete the extinction of the !Kwi group of Southern Bushman
languages (Köhler 1981: 469). These Southern Bushman languages were once
spoken over Bushmanland and the Karoo, from the Orange river in the west
The Khoesan languages 29
to Lesotho and the Orange Free State in the east, with the outlying language
//Xegwi found at Lake Chrissie, in the eastern Transvaal.
In the early seventeenth century there were about eleven closely similar Cape
Khoekhoe varieties spoken from the Cape of Good Hope in the west, along the
southern Cape coast and its hinterland as far east as the Fish River (Elphick
1985: 51). Estimates of the number of all South African Khoekhoe (including the
Nama) in 1652 vary between 100,000 (Elphick 1985: 23) and 200,000 (Wilson
1969: 68). Within sixty years of that date ‘the traditional Khoekhoe economy,
social structure, and political order had almost entirely collapsed’ (Elphick
1985: xvii), and smallpox epidemics in 1713, 1735 and 1767 had ravaged the
population, wiping out virtually all the western Cape Khoekhoe. And within
100 years of 1652, the western Cape Khoekhoe language had begun to disappear,
being gradually replaced by Khoe-Dutch (Nienaber 1963: 97ff.), and the Eastern
Khoekhoe varieties had been absorbed by Xhosa through political incorporation
of the Khoekhoe chiefdoms (Marais 1968: 111).
This is the dramatic background to the extinction of the Cape Khoekhoe and
the death of their language. However, far from vanishing without a trace, the
Cape Khoekhoe have had a profound effect on the genetic features of many
South Africans. Their language has exerted an influence on the development of
Afrikaans and has extensively restructured the phonological systems of Xhosa
and Zulu, greatly enriching the lexicons of these two languages in the process.
It is these influences that allow one to reconstruct aspects of the sociolinguistic
situation that led to the death of the Cape Khoekhoe languages.
Two distinct areas can be identified in this process, the first in the east between
the Kei and Keiskamma rivers, where the Khoekhoe were ‘incorporated by the
expanding Xhosa chiefdoms during the early 1700s’ (Harinck 1972: 158), and
the second in the west, where the Khoekhoe language was replaced by pidgin
Dutch or Dutch (Elphick 1985: 210ff, Nienaber 1963: 97–8). In the east ‘contact
and interaction between Xhosa and Khoe was facilitated by the fissiparous ten-
dency in the Xhosa social structure and by similarities in their respective social
organisation’ (Harinck 1972: 158). This resulted in assimilation of Khoekhoe
into Xhosa lineages and Xhosa into Khoe chiefdoms. In the latter case this gave
rise to the Gonaqua (=| gona) and later the Gqunukwebe and, from a linguis-
tic point of view, to language mixture in which the Khoekhoe language was
dominant (Harinck 1972: 157). Although Gona was a Khoe language it had
changed sufficiently to present Khoekhoe from the western Cape who heard it
in 1772 with difficulties of understanding (Wilson and Thompson 1969: 103).
The well-known result of this intimate and long contact was the incorporation
of a large vocabulary containing adapted click consonants, which led to the
30 A. Traill
This hybridization can only be accounted for by reciprocal marriage between the in-
coming Xhosa and . . . [the] Khoe. The children of polygynous marriages between Khoe
males and Xhosa females learned Xhosa from their mothers and incorporated it into the
Khoe language while participating in Khoe society external to the immediate family.
Khoe prevailed as the predominating element of the Gonaqua’s language because the
Xhosa language was not incorporated as much by offspring of marriages between Khoe
women and Xhosa men. (Harinck 1972: 158)
(Sales 1975: 29). The fact that Van der Kemp felt it was necessary at the time
to produce a Khoekhoe catechism, Tzitzika Thuikwedi miko Khwekhwenama
(Principles of God’s word for the Khoekhoe), would indicate that the Khoe
language still had some vitality even then (Sales1975: 29).
However, it seems clear that a process of rapid language shift to Dutch (i.e. a
form of early Afrikaans) or Xhosa had begun. The social situation in which
this was taking place was chaotic, and hastened the death of the language:
Khoekhoe, Xhosa, Boers and British were caught in the struggle to establish
control over the eastern frontier. The Khoekhoe were doomed in this violent
conflict and by 1809 the Earl of Caledon’s ‘Magna Carta of the Hottentots’
and the destruction of the last independent Khoekhoe territory at the Gamtoos
river reduced virtually all Khoekhoe to the status of servants of the colonists
(Mostert 1992: 350–1). However, the language was still spoken to some extent
in 1820 when Thomas Pringle visited the Bethelsdorp community and heard the
‘uncouth clucking sounds of the Hottentot language spoken by some of them
to each other’ (Sales 1975: 84). The Gonaqua surface again in the historical
record in 1829 when those Gonaqua who had remained with the Xhosa moved
to the Kat river settlement. We have no idea of the form in which Khoekhoe was
spoken at this stage but it can be safely assumed that this period marks the last
stages of the language. Six years prior to this, in 1823, the first written record
of Xhosa appeared in the form of John Bennie’s Incwadi yokuqala ekuteteni
gokwamaXosa (The first book in the language of the Xhosa). As the click
words show, the process of click incorporation which had begun some two
centuries before had been consolidated by the time the donor language was
dying, and today these clicks remain as a vivid remnant of that Khoekhoe
language.
In the south-western Cape, the linguistic contraction of the closely related
Khoekhoe varieties spoken there – Hesse (Hai-se), Chainou, Cocho, Guri,
Goringhai (!uri-//’ae) and Gorachou (!ora-//xau) – was rapid (Elphick 1985: 53;
Elphick and Malherbe, 1989: 5), and by 1750 they had begun to disappear with
the shift to Khoekhoe-Dutch (Elphick 1985: 211). Nienaber (1963: 98) notes
that between 1773 and 1797 travellers such as Thunberg, Sparrman and Barrow
‘kon die taal [Khoekhoe] nog net aan die uithoeke van die Kolonie beluister,
veral aan die Oostelike grens’ (could still hear the [Khoekhoe] language only
in the outlying districts of the colony, particularly on the eastern border).
It has been argued that the processes that destroyed the social, political
and economic structures of the western Cape Khoekhoe were far advanced
only sixty-one years after van Riebeeck landed in Table Bay (Elphick 1985);
the smallpox epidemic of 1713, which virtually wiped out the Khoekhoe in
the western Cape, merely consummated this breakdown. The result for the
Khoe language spoken there was that within a hundred years of van Riebeeck’s
arrival in 1652 it too had largely succumbed, and was replaced by Afrikaans.
32 A. Traill
But whether or not this Khoekhoe Afrikaans was a pidgin in its initial stages is
still debated (see chap. 4, this volume).
Elphick describes the linguistic response to the rapidly changing situation at
the Cape as the emergence of a pidgin Dutch spoken by the Khoekhoe (Elphick
1985: 211). Elphick gives a number of examples of Khoekhoe-Dutch from as
early as 1673 and the early part of the eighteenth century. Some of the examples
show the use of the pronoun ons, ‘we’, which anticipates Afrikaans usage:
Duitsman een woordt calm ons u kelem
(Dutchman [if you] speak a word, we [will] slit your throats)
Rademeyer provides other examples and emphasises that this Dutch spoken by
the Khoekhoe in the immediate vicinity of the Castle in 1666 was in fact a very
‘gebroke Hollands’ (broken Hollands) (1938: 33). However, the suggestion that
this was indeed a pidgin would require more convincing sociolinguistic evi-
dence than these brief accounts provide. In fact, the swift collapse of Khoekhoe
society, the rapid shift to a variety of Dutch, the dramatic effect of smallpox on
the numbers of speakers and the precipitate contraction of the Khoe language
make it unlikely that a pidgin would have crystallized in the western Cape.
To add to the social, economic and physical onslaught on the Khoekhoe,
their language itself faced two intimidating problems. The first was extreme
linguistic prejudice: from the first contacts between Europeans and Khoekhoe
there had been a persistent attitude on the part of the Europeans that the lan-
guage was utterly bizarre, unpleasant, inarticulate and not human. Nienaber
(1963: 76ff.) quotes many such opinions from the late sixteenth century on.
These prejudices fed the second problem, namely the view that the language
was unlearnable, and from as early as 1663 this led to official government pol-
icy that the Khoekhoe should learn the colonial language (Wilson 1969: 66).
By 1700, therefore, it was possible to make do with some version of Dutch
for fifty miles east and north of the Castle at Table Bay. At first there was a
need for Khoekhoe interpreters, but during the eighteenth century this soon
diminished and eventually disappeared as the Khoekhoe shifted to Dutch.
Henry Tindall described how even missionary work relied on interpreters until
the Khoekhoe language had been replaced (Nienaber 1963: 97–8). When the
Moravian missionary Georg Schmidt first started his work among the Khoekhoe
in 1737 at Baviaanskloof (later to become Genadendal) he found monolingual
Khoekhoe speakers. He tried to learn the language but found he could not imi-
tate the clicks. He ‘soon perceived that the language was too difficult . . . to
master and . . . therefore commenced to teach them to speak Dutch’ (du Plessis
1965: 54), relying in the meantime on an interpreter. Within three years he was
distributing Dutch New Testaments to those who had learned to read. How-
ever, the language partly survived the assault, because sixty-four years later
The Khoesan languages 33
in the same place, it was necessary to repeat the sermon after each service
in the Khoekhoe language for the benefit of a number of older people who
‘understood only the Hottentot language’ (Kruger 1966: 89). These pockets
of Khoekhoe survival must have been the exception; children were not being
taught the Khoekhoe language and everywhere it was being replaced with a
version of Khoekhoe-Dutch.
A further source of linguistic pressure on remnants of the Khoekhoe language
must have come from the variety of Dutch spoken by the slaves who lived in
some intimacy with Khoekhoe labourers on farms (Marais 1968: 13). Whatever
the extent and nature of this influence, it must have given a strong impetus to
the process whereby Khoekhoe-Dutch was already engulfing Khoekhoe. As a
result of all these pressures the Khoe language of the western Cape was simply
overwhelmed. However, it did not disappear without trace: it survives to this day
in the Afrikaans lexicon in the form of many plant, animal and bird names
(Scholz 1940), and in both Afrikaans and English in numerous place names
(Raper 1972; Nienaber and Raper 1977).
The social and other forces that consumed the western Cape Khoekhoe
dialects gradually spread to the remaining Khoekhoe languages of South Africa:
Nama, Kora (!Ora) and Gri (Xri). At the beginning of the nineteenth century,
Little Namaqualand, to the north of the Kamies mountains, was beyond the
official north-western border of the Cape Colony. The aridness of the area and
its isolation for a while from events to the east and south served to slow slightly
the inexorable advance of white and mixed-race settlers, the Trekboers and
Basters (or Bastards). As a result the Khoekhoe and their language flourished
for some time in this region. Significantly, missionaries to these parts did learn
the Khoekhoe language: their work here and in Great Namaqualand across the
Orange river led to the first grammar, dictionary, Bible translation, catechism
and hymn-book in a Khoekhoe language (Haacke 1989; Strassberger 1969:
63, 69). Ironically, however, it was the missionaries together with the encroach-
ing Basters who have been identified as the ‘alien elements . . . effecting the
disintegration of the Khoi Khoin’ (Carstens 1966: 205) through their acquisi-
tion of political power in the region.
It is significant that one of the only places in which a vital Khoe language
still survives in Little Namaqualand is in the Richtersveld where ‘no missio-
nary ever achieved political power’ (Carstens 1966: 208) – although the people
of the Richtersveld were easy converts to Christianity – and where a major
influx of Basters only took place in 1936. In the rest of Little Namaqualand
the Khoekhoe shifted to Khoekhoe-Dutch/Afrikaans. However, this shift must
have been more gradual than it had been earlier in the western Cape because a
Khoekhoe linguistic remnant is still to be found among the much older members
of the communities around Kharkhams (Leliefontein district), in the form of
34 A. Traill
Khoekhoe plant names, animal names, place names and domestic terms. Many
of these words still contain a [!] (alveolar) or [//] (lateral) click (Links 1989:
61ff.). In the Richtersveld, the Boorlinge (the native Khoekhoe inhabitants as
opposed to more recent immigrant groups) have a variety of Nama as their
mother tongue, and children were monolingual until the 1950s when offi-
cial policy required them to learn Afrikaans as a second language in school
(E. Boonzaaier, personal communication). Although there is bilingualism
among the Richtersveld Khoekhoe, the absence of any significant shift to
Afrikaans among them suggests that the language has a good chance of sur-
viving as the last Khoekhoe language of South Africa. This will undoubt-
edly be reinforced by the very recent emergence of pride in Nama identity
(E. Boonzaaier, personal communication).
To the east, along the Orange and Vaal rivers, the Kora and Gri dialects were
moribund (Krauss 1992: 4) even before Beach studied them in 1926 (Beach
1938). A few older people who knew a few words and phrases could still be
found around Douglas, Prieska, Campbell and Griekwastad as late as the 1980s
(van Rensburg 1984: 669), but the account Beach (1938: 183) gives vividly
documents the end of the language: ‘Finding a pure representative of the Korana
tribe is like finding a rare gem. And sorting out a few old Korana (still able
to speak Hottentot) from a community of Griqua . . . is like sifting diamonds
from sand. There are a few . . . [Griqua] left in Kokstad but less than half-a-
dozen of these can speak Hottentot; the others all speak Afrikaans.’ And when
Beach visited Kimberley ten years later nine of his ten Korana informants were
dead.
Beach considered Kora and Gri to be closely related dialects which were
difficult to distinguish; indeed, their linguistic history suggests very similar
origins. On the basis of vocabulary recorded by the eighteenth-century explorer
Le Vaillant, he concluded that the ‘language spoken by the Cape Hottentots
was essentially the same as that of the present-day Korana and . . . considerably
different from present-day Nama’ (Beach 1938: 181).
This assessment was based on a purely linguistic comparison, but it is sup-
ported by the fact that a good-quality vinyl record of a Kora speaker made by
Pierre de Villiers Pienaar in about 1938 was minimally intelligible when played
to a native Nama speaker in Namibia in 1991 (W. H. G. Haacke and E. Eiseb,
personal communications). The linguistic affinity between Kora and the eastern
and western Cape Khoekhoe dialects, as opposed to Nama, has also been noted
by Nienaber (1963: 535). This is consistent with the traditional origins of the
Kora and the Khoekhoe component of the Gri in the two groups of western
Cape Khoekhoe, the Gorachouqua and Chariguriqua or Grigriqua.
From a sociolinguistic perspective, the history of the two groups is one of
more or less rapid shift to Khoekhoe-Dutch from which developed a distinc-
tive variety of Afrikaans. In the case of the Gri, the shift seems to have been
The Khoesan languages 35
far advanced by 1801 (Marais 1968: 34): in the case of the Korana, whose
Khoekhoe identity was still entrenched then, it was more gradual. When the
Berlin Missionary Society first began to work among the Korana in 1834
at Bethany in the Southern Orange Free State, there were 20,000 nomadic
Korana between the Orange and Vaal rivers. Together with the fact that the
missionary Wuras used a Kora interpreter and prepared a catechism, grammar
and vocabulary in Kora, this suggests that a vital Khoekhoe language and
significantly monolingual speech community was still in existence at the time
(van der Merwe 1985: 40; Beach 1938: 182; du Plessis 1965: 213).
But the Korana and Griqua were not of equal status on the northern frontier.
The ethnonym ‘Griqua’ replaced ‘Bastard’ in 1813 at the insistence of the mis-
sionary Campbell, who persuaded the Basters that the latter term was offensive
(Ross 1976: 16). From a sociolinguistic perspective this is significant because
the replacement name may convey a stronger Khoekhoe linguistic affiliation
than ‘Bastard’ might. However, it is difficult to estimate precisely what the dom-
inant language was among the Griqua. Even in its origins the group was not
linguistically homogeneous, incorporating both speakers of Khoekhoe-Dutch
and Khoekhoe (Marais 1968: 32). Evidence concerning a group of Basters who
migrated from the northern frontier to what is now southern Namibia sheds
some light on this: despite conventional wisdom that the Basters were primar-
ily or even exclusively speakers of ‘Dutch’, Khoekhoe Afrikaans was certainly
not the dominant language of this group, but was their second language; Nama
was their first language and the fact that children still spoke it shows that it
was not yet moribund (Cluver n.d.a: 113–14). Whatever the linguistic situation
among the Griqua, they were also proficient in Khoekhoe-Dutch and gradually
the Khoekhoe language was replaced.
The Griqua were the dominant political group on the northern frontier in the
early part of the nineteenth century, and they sought to reduce the Korana and
Bushmen to a dependent status as labourers (Ross 1976: 15). This exploitation
contributed to the destruction of the Korana and the Bushmen, and it must have
had a strong impact on the changing linguistic affiliations of the area, in the
shift to a variety of Afrikaans.
However, both Griqua and Korana societies collapsed in the course of the
political developments on the northern frontier. This is reflected linguistically
in the death of both Khoekhoe varieties (notwithstanding the remnants of
Khoekhoe spoken today by a few older people in Kakamas, Pella and Keimoes
(Hoff, personal communication). The variety of Afrikaans that replaced Kora
and Gri was distinctive and has been labelled Orange River Afrikaans by van
Rensburg (1984: 514–15), who characterises it as follows:
die nie-standaard Nederlands . . . wat veral aan die Oranje Rivier maar ook op ander
plekke in die binneland, vanaf sowat die begin van die agtiende eeu deur ’n noe-
menswaardige aantal sprekers gebruik is. Dié form van Afrikaans is sterk beinvloed
36 A. Traill
deur Hottentots. Talle sprekers van Oranjerivier Afrikaans vas vroeërs ook ’n vorm van
Hollands magtig. (van Rensburg 1984: 514–15) (the non-standard Dutch . . . that was
used by a significant number of speakers especially on the Orange river but also at other
places in the interior, from around the beginning of the eighteenth century. This form
of Afrikaans had been strongly influenced by Hottentot [i.e. Khoekhoe]. Earlier, many
speakers of Orange River Afrikaans were also fluent in a form of Hottentot.)
Cluver (n.d.a) describes this Orange River Afrikaans as ‘strongly creolized’ Cape
Dutch interspersed with Khoekhoe words, a characterization of Khoekhoe-
Dutch that may well apply to the type of Dutch that had replaced Khoekhoe at
the Cape a century before.
The San languages of South Africa were all members of the !Kwi group of the
Southern Bushman language family (D. F. Bleek 1929; Köhler 1981).
The geographical spread of this group in historical times covered virtually
the whole of what is modern South Africa, from the eastern border of Swaziland
in the north-east to the mouth of the Orange river in the north-west, and from
the Natal midlands in the south-east to the western Cape in the south-west. It is
reasonable to assume that the !Kwi languages or dialects had been spoken over
most of this area for some 8,000 years (Wright 1971: 1). Today, with the excep-
tion of a handful of speakers of the moribund /’Auni (i.e. /’Auo) and =| Khomani
languages of Gordonia and the Kalahari Gemsbok Park, all these languages are
dead, their speakers having been exterminated or their remnants absorbed into
the Bantu-speaking or (what are now) Afrikaans-speaking coloured communi-
ties. Wright estimates that there could have been 10,000 to 20,000 Bushmen in
South Africa before this process began; the extinction was complete in about
three hundred years.
The !Kwi languages fell into three or four groups. The linguistic affiliations
between, and even within, the groups are not always clear from the available
material; some of the main varieties are given below. (For a detailed list of all
of them, see Winter 1981.)
The largest and most extensive was /Xam or /Kham, the language of the
so-called Cape Bushmen. This was recorded in a number of more or less
closely related varieties in the whole of the former Cape Province south of the
Orange river from the Colesburg and Burgersdorp area in the north-east to
the Katkop hills north of Calvinia in the north-west and from the Achterveld in
the Fraserburg district in the south-west through Oudtshoorn to the Graaf-Reinet
area in the south-east. W. H. I. Bleek examined the differences between a num-
ber of the /Xam varieties from this area in 1857 and stated that ‘the different
Bushmen dialects spoken within this colony vary little from each other . . . one
language . . . is spoken by all these Bushmen’ (1873: 2).
The Khoesan languages 37
There can be little doubt that the first Bushmen encountered in the south-
western Cape by Europeans were also speakers of /Xam. The variety of the !Kwi
group known as //Ng !k’e was recorded much later by Dorothea Bleek at Mount
Temple in the area of the Langebergen near present-day Olifantshoek between
1911 and 1915. She described this as ‘a language . . . very like the /Kham tongue,
but much too distinct to be classed as a dialect’ (1927: 56). This language was
formerly spoken from the Vaal river in the east to the Molopo in the north and the
west. Bleek found a few speakers on the right (i.e. east) bank of the Vaal and on
the lower Molopo in Gordonia (1929: 1). Elsewhere, she gives the distribution
of //Ng !k’e as ‘Griqualand West and Southern Gordonia’ (1942: 5).
A further group of !Kwi varieties was found by Dorothea Bleek in 1911
in what is now the Kalahari Gemsbok Park. These were /Auni (/’Auni) and
the Xatia (Kattia or =| Keikusi), dialects of the same language. She described
the language as less closely allied to /Xam than //Ng !k’e. To the languages
of this area should be added =| Khomani (Doke 1937) and Ku /khaasi (Story
1937), both reported in 1937. Further east was =| Kunkwe of the Warrenton area
(Meinhof 1928–29), //Ku//e, spoken near Theunissen (D. F. Bleek 1956),
Seroa, spoken near Bethany in the Orange Free State and in what is now Lesotho,
and !Gã !ne spoken near Tsolo in the Transkei (Anders 1934/5). The easternmost
!Kwi language was //Xegwi, spoken at Lake Chrissie in the eastern Transvaal
(Lanham and Hallowes 1956a, 1956b).
The more recent history of the speakers of /Xam is well known, with ‘their
societies shattered by warfare, starvation and disease; the women and children
enslaved; the men all but exterminated by the genocidal hatred of their enemies’
(Penn 1991). From a sociolinguistic perspective this situation satisfies virtually
every requirement for language death (Brenzinger 1992: 290). Indeed, this had
happened within about 170 years of the first clashes between the /Xam and the
frontier farmers of the Cape Colony in about 1740. There is no clear picture of
the linguistic situation before this time. Speakers of /Xam and Khoekhoe had
been in social contact for centuries, and there is a limited amount of evidence
that this led to some degree of bilingualism (Penn 1991; Wilson and Thompson
1969: 63–4). The /Xam recorded by W. H. I. Bleek showed very little Khoekhoe
influence. On the basis of this evidence one may assume that the bilingualism
affecting this variety of the language had not involved language shift. He did
notice that a number of words for ‘abstract concepts’ appeared to be of common
origin in /Xam and Khoekhoe, but he concluded that these had probably been
taken over from Khoekhoe into /Xam ‘in consequence of the contiguity of the
two nations’ (1873: 8).
The linguistic situation changed, however, after 1740 when /Xam society
faced continuous pressure from war, dislocation and extermination, which also
spilled over into Khoekhoe and Baster communities. One can only guess that
this had linguistic repercussions which set the scene for the eventual death of
38 A. Traill
/Xam. But these effects took some time to emerge, and the evidence for the
changes is extremely thin. When the missionaries Johannes Jacobus Kicherer
and John Edwards landed in Table Bay in 1799 they met two Bushmen and a
Korana, who had Dutch names: Vigilant, Slaparm (‘Weak Arm’) and Oorlam
(‘Knowing One’). Clearly, some linguistic force had begun to stir in the interior.
However, when Kicherer and Edwards set up the first mission to the /Xam on
the Zak (Sak) river in the same year, they found themselves in a vital /Xam
community. Kicherer remarked that ‘their language is so difficult to learn that
no one can spell or write the same’, and none of the missionaries succeeded
in mastering it (du Plessis 1965: 104–5). Penn (1991) notes that at first these
missionaries relied on the services of one Gerrit Visser, son of the frontier farmer
Floris Visser, who could speak /Xam. This gives a fascinating, if frustratingly
meagre, glimpse into the linguistic dynamics of the area. Later the missionaries
relied on another /Xam speaker as principal interpreter, but they had already
begun a daily routine of instruction in Dutch for the children (Penn 1991).
After 1754 the Trekboers of the frontier began their retaliatory commandos
against the /Xam. Over a period of forty-four years thousands were killed;
surviving women and children were distributed as slaves among farmers; and
some women were given as wives to Khoekhoe members of commandos (Penn
1991). This destroyed the basis of /Xam society, immediately creating condi-
tions under which language maintenance was impossible: ‘Those San who grew
up on farms, either as captive children, or as the descendants of clients, were
absorbed into the mixed Coloured community . . . they mingled in race with ne-
groid and Indonesian slaves, with whites, as well as with herders who resembled
them physically’ (Wilson 1969: 72). Surviving groups of /Xam speakers were
driven into remote areas around the Hartebeest river and further west where
they struggled to survive in the context of diminishing resources and the con-
tinual encroachment of farmers on their land. Here too, attempts were made by
Trekboers, Basters, Korana and Xhosa to exterminate them (Marais 1968: 28).
Within one year, between 1858 and 1859, they had virtually disappeared from
the neighbourhood of the Hartebeest river. It is precisely from this area that
Jantje Tooren or //Kabbo (‘Dream’), one of W. H. I. Bleek and Lucy Lloyd’s
main informants, came.
Bleek began his work with the /Xam in 1870. Although the material he and
Lloyd collected necessarily focuses on the /Xam language, it is nevertheless
possible to piece together an outline of the broader linguistic situation of the
/Xam at that time. The most significant fact is that there was bilingualism among
the informants. Most of them had worked for farmers and could speak ‘Dutch’.
Indeed, it is clear that Bleek and Lloyd relied on this fact in their linguistic
work. Their original manuscripts contain many notes in ‘Dutch’ translating
a /Xam word or a sentence, and in Bleek’s report he lists four texts, ‘Lion
and Bushman’, two versions of ‘Woman transformed into lion’, and ‘The lost
The Khoesan languages 39
child’, as translations from the Dutch (1873: 5). There was also bilingualism
in Kora. In the turmoil of the times, the /Xam had formed alliances with these
Khoekhoe and had even been absorbed by them. Bleek records how he spoke
(in Dutch?) to a ‘Bushman’ prisoner in Cape Town who told him he had been
brought up by the ‘Korannas’ (sic) since he was a child (Bleek and Lloyd
1911: 436). =| Kásin, another of the informants used by Bleek and Lloyd, had a
father who was a Korana chief and a mother who was a /Xam; he was fluent
in both languages (W. H. I. Bleek 1875: 5; Deacon 1996). One may assume
that these were not isolated cases. Despite this bilingualism, it seems that /Xam
was being maintained at least among the adults Bleek and Lloyd recorded.
There are no signs that the /Xam of the Bleek and Lloyd texts was seriously
influenced by Khoekhoe and not at all by Dutch; as mentioned above, from
this evidence and for these speakers it appears that the shift to other languages
had not yet taken place. However, there is no record of the transmission of the
language to children. There is the possibility, though, that the language of the
/Xam that Bleek encountered was in fact beginning to show the first symptoms of
shrinkage. Bleek referred to //Kabbo as ‘our best informant . . . [who] was nearly
sixty years of age . . . and [who] was picked out from among twenty grown-up
Bushmen as one of the best narrators’ (Orpen 1874: 12). The implication is
clearly that older speakers had better command of linguistic skills.
Forty years later, the shift had begun to take effect. In 1910–11 Dorothea
Bleek visited the few remaining speakers of /Xam at Prieska and Kenhardt;
they worked as shepherds or labourers on farms or as servants in the villages.
One of them, the old Janikie Achterdam, had been with W. H. I. Bleek forty
years before; she sang some songs and told the story in /Xam of the moon and
the hare. It is poignant that she ended with the words ‘nu is ik klaar’ (now I am
finished) (Treble Violl 1911: 9).
Bleek’s biographical sketches of members of this group provide brief insights
into their history and linguistic situation. Some still spoke /Xam fluently, but
knew no folklore; others did not speak the language at all. Some had a clear
memory of their personal histories over a period of sixty or seventy years,
but one, Roman Titus, did not know his parentage. Perhaps the most dramatic
case involved Guiman and his wife Rachel, daughter of /ogən-aŋ. Rachel had
been taken as a young girl by a farmer’s wife and had grown up speaking
Afrikaans. She learned /Xam from Guiman, who in turn learned Afrikaans
from her (D. F. Bleek 1936: 201–3).
The fate of the remaining !Kwi dialects or languages differs only in some
of the details. In 1857 W. H. I. Bleek used Lichtenstein’s short comparative
vocabularies of Bushman and Kora to discover that the version of /Xam spoken
in the Colesberg and Burgersdorp district differed very little from the varieties
further west (W. H. I. Bleek 1875: 2). In 1814, missionaries to Tooverberg
ministered and taught entirely through the services of a Khoekhoe interpreter
40 A. Traill
from Tulbagh named Cupido, who had been a farm worker in Graaf-Reinet
and had learned /Xam from the Bushmen farm workers. But within ten years
Tooverberg had become the white farming town of Colesberg, and the /Xam
and their language had begun to disappear. Bleek’s limited investigation thirty
years later of a few speakers from this area unfortunately tells us nothing about
the general state of the language.
The same sequence of events affected the mission stations among the
Bushmen at Bethulie and Philippolis, which were founded between 1820 and
1830 (Sales 1975: 62–3). It is likely that a different language in the !Kwi group
was spoken here (possibly //Ku //e, Seroa, //Ng !k’e).4 But the identity of the lan-
guage was irrelevant. The fact that ‘not even one missionary ever understood the
Bushman language’ (Sales 1975: 63) repeated language attitudes encountered
100 years previously among the Cape Khoekhoe, and did nothing to slow the
demise of any of the languages. The Philippolis Bushmen faced a more daunting
problem than indifference to their language. It came in the form of the Griqua,
who took them as labourers or drove them out. By 1835 all the survivors ‘were
reduced to the level of labourers or had fled for instance to the Orange River
valley’ (Ross 1976: 24–5).
The flight to the east would have brought fugitives into contact with speakers
of the !Kwi language, Seroa, and Southern Sotho. There is no useful record of
Seroa, but it was evidently spoken in what is now Lesotho and in adjacent areas
of the Orange Free State. When the missionary Arbousset travelled through the
eastern Orange Free State in 1836 he remarked that Seroa was the most widely
spoken language. In his remarks on Joseph Orpen’s paper on the mythology of
the Maluti Bushmen, W. H. I. Bleek concluded, on the basis of thirteen words
(including six proper names), that the language later to be called Seroa was
‘essentially the same as, although dialectally differing from, that of the more
western Bushmen’, i.e. the /Xam (Orpen 1874: 12).
One of the words, tsha, ‘eland’, is the same in Dorothea Bleek’s S11e, !Gã
!ne of the Transkei, and the word cagn, ‘deity’ is the same as /kaggen, ‘mantis’
in /Xam (Bleek 1956). W. H. I. Bleek’s knowledge of the relationships of Seroa
to other languages of the !Kwi group could not have been based on much more
than these two words.
In 1870, Bushmen were still numerous in the Quthing district of Lesotho,
despite attacks elsewhere in the region from the Sotho, slave raiding by the
Korana and attempted extermination by imperial troops under Colonel Bowker
(How 1962: 53, 57–8). However, it seems that the language had disappeared
by the 1880s and one must infer that there had been a rapid shift, in this case to
Sotho or the Nguni varieties spoken in that area.
It is most likely that the Bushmen of Natal were also speakers of Seroa, at
least in historical times. Wright (1971: 189) describes how almost all the bands
engaged in raiding Natal between 1840 and 1872 were from East Griqualand
The Khoesan languages 41
and south-eastern Lesotho, and at least some of these operated from the Quthing
area under the protection of Chief Moorosi; this is precisely where the largest
number of surviving Lesotho Bushmen was found in 1879 (How 1962: 58; Jolly
1994). These raiders had close alliances with the Bhaca and the Mpondomise
of the present-day Transkei and were most likely bilingual in these Nguni
varieties. Wright quotes the statement of one Jacobus Uys, who spoke to a
group of Bushmen in southern Natal in 1840 through an interpreter, a ‘Hottentot
named Jan’ (Wright 1971: 54–5). Since there was no Khoekhoe language spo-
ken in those parts it is likely that Seroa was being used. At least one can
tell from this encounter that there had been no shift to Dutch, as there had
been in the rest of South Africa. Evidence that Seroa survived to some ex-
tent until 1873 in the Qacha’s Nek area comes in the form of Qing, a young
Bushman from that area who acted as Joseph Orpen’s guide in Basutoland.
Orpen used a number of different interpreters to communicate with Qing,
and he described his bilingualism as follows: ‘the language he spoke best be-
sides his own was that of the Baputi, a hybrid dialect between the Basuto
and the Amazizi languages’ (Orpen 1874: 2). Qing’s bilingualism is likely to
have been typical for the period, and it gives an idea of at least one of the
directions of language shift which rapidly culminated in the disappearance of
Seroa.
Everything known about the !Kwi language !Gã !ne, once spoken in the Tsolo
district of the Transkei, comes from the material collected by Anders from two
middle-aged semi-speakers in about 1931. The mother of one of them was a
‘true Bushwoman’; the other had come to the Tsolo district from the Umtata
district and he had spoken the Bushman language with his uncle some forty-five
years before. Since then, he had lived among the Mpondomise, and the sounds
of !Gã !ne ‘were . . . like far off memories of other times. Patience and time
were required to allow his memories to wake up after long dormancy’ (Anders
1934/5: 82). This resuscitation of a language that was almost dead yielded some
140 words. Anders concluded that !Gã !ne was most like the //Ng !k’e recorded
by D. Bleek in Gordonia, Griqualand West and the Vaal River area in 1911
and 1915 (Anders 1934/5: 85). If //Ng !k’e represented a continuum of dialects
spoken in this line through to the Transkei, it is plausible to suggest that Seroa
could have been one of them. Unfortunately, the Seroa material collected by
Orpen is so limited that only the word for ‘eland’ referred to above has a cognate
in !Gã !ne.
Further east, at Lake Chrissie in the eastern Transvaal, the !Kwi variety known
as //Xegwi was spoken. In the 1950s there were fewer than thirty-six speakers
left. They were described as still knowing their language ‘fairly well’ and being
bilingual in ‘Swazi-Zulu’ and Afrikaans (Potgieter 1955), though thirty years
earlier they evidently did not speak the ‘taal’ (i.e. Afrikaans) (D. F. Bleek
1929: 1). Potgieter remarks, however, that children were showing signs of
42 A. Traill
forgetting their language, and he makes the interesting observation that the
//Xegwi were ‘not inclined to speak their own language in the presence of
Swazi or Europeans’ (1955: 7). This probably reflects a strong stigma at-
tached to speaking //Xegwi, a situation that would contribute to language
shrinkage.
According to tradition, the earlier speakers of //Xegwi spoke Sotho as well.
There are in fact a number of borrowed forms in //Xegwi from Sotho, Zulu,
Afrikaans, English and even Tsonga (Lanham and Hallowes 1956b). However,
it is not possible today to estimate where and when the contacts with Sotho and
Tsonga took place. In this small community there also appear to have been such
wide differences in pronunciation that Winter has claimed it is not possible to
decide whether the available descriptions of the language represent only one
dialect (Winter 1981: 342). But this variation probably means the informants
were ‘terminal speakers’ (Tsitsipis 1989: 119). These factors all describe a
language in its last stages.
It has been suggested that the //Xegwi were refugees from Basutoland
(Potgieter 1955). The presence of a Sotho influence in the language may lend
some plausibility to this claim. But given the lack of any linguistic details about
Seroa itself, the claim remains speculative. Nevertheless, the existence of a
number of //Xegwi lexical items with clear reflexes in =| Khomani of Gordonia
(Lanham and Hallowes 1956b) lends some support to the suggestion already
made that there may have been a certain integrity to the !Kwi languages spo-
ken from Gordonia in the west through Griqualand, the Orange Free State and
Basutoland to Lake Chrissie in the east.
Ultimately, the death of //Xegwi seems to have been caused by the death of
all its speakers rather than by a shift to Swazi or Zulu. In 1975 I interviewed
Jopi Mabinda, the last //Xegwi speaker. He was able to reproduce perfectly
the linguistic material he had given to Lanham and Hallowes (Lanham and
Hallowes 1956a) and he was fluent in Zulu. He told me he was the only speaker
of the language and that he spoke it to his sister and brother-in-law, who only had
a passive knowledge of it. He was murdered at Lothair, in the eastern Transvaal,
in 1988 (Boekkooi 1988).
In 1911 Dorothea Bleek visited the Lower Nossop and Auop rivers in
Gordonia (the area that is now the Kalahari Gemsbok Park), to study the
Bushman language spoken there. She found speakers of a !Kwi language called
/Au or /Auo; the people called themselves /Auni (D. Bleek 1937b: 208). (She
erred in her transcriptions, which should have been /’Au, /’Auo, /’Auni respec-
tively.) There was also a closely related dialect named Xatia. She described the
situation as follows:
They were in their natural state, living in bush screens and clothing themselves with
skins. They collected wild vegetables and hunted when they could, chiefly with guns
owned by the ‘Bastaards’ who had made themselves their overlords . . . interpreters were
The Khoesan languages 43
difficult to find, therefore only a small amount of linguistic material could be collected.
Yet that shows the place of the language among the others. (D. Bleek 1929: 2)
She placed the language as a somewhat distant linguistic relative of /Xam.
Since these Basters would have mainly spoken a variety of Afrikaans and prob-
ably some Nama, we may infer from her remarks that the /’Auni were not
yet bilingual in Afrikaans and were maintaining their language and lifestyle
despite their relationship of clientship. Twenty-five years later the University
of the Witwatersrand’s research expedition to Tweerivieren at the junction of
the Nossop and Auob rivers found a different situation. After a great deal of
effort, seventy Bushmen were collected for anthropological, linguistic and phys-
ical study. Of these, forty-three spoke a hitherto unrecorded language called
=| Khomani, twenty-six spoke /’Auo, one was a Vaalpens5 and the remainder
spoke Nama (Bleek also noted the presence of a ‘Vaalpens’ woman who turned
out to speak Ku/haasi: D. F. Bleek 1937a; Maingard 1937). Maingard makes
the interesting observation that adults and children were speaking =| Khomani,
but he also notes that there was one ‘best’ speaker of =| Khomani who was
the most conversant with the language, Ou Abram or !gurice; that others
were not speakers of pure =| Khomani, and that Ou Abram’s children Malxas
and /Khanako, who were also his informants, had forgotten the lore of their
forefathers. Malxas was also fluent in Afrikaans and translated all Ou Abram’s
folktales into that language (Maingard 1937: 237, 261). Within a year, Ou
Abram was dead.
I interviewed Malxas’s son at Nossop camp in the Kalahari Gemsbok Park in
1973; he spoke only Afrikaans and any remaining =| Khomani speakers had dis-
persed. The fate of =| Khomani follows a classic course: bilingualism, shrinkage
of the language, shift. In this case the shift was to Nama and/or Afrikaans.
One can see that when Maingard conducted his study, the process of obsoles-
cence was well entrenched. It is interesting to read how both Maingard and Doke
attributed phonetic imprecision, morphological variation and the stylistic im-
poverishment they found in =| Khomani to the conceptual style of the Bushman,
in which he ‘is quite content with relative approximations, so long as he is
understood by his fellows’ (Maingard 1937: 253, 260; Doke 1937: 87). Yet all
these features are well-known symptoms of language decay.
Maingard provides evidence of =| Khomani’s relationship to /Xam in the form
of fifty-two words and other shared features. It is worth noting that this list far
exceeds, in quality and quantity, those inadequately transcribed and limited
sources that have so frequently been used to judge that a certain !Kwi variety
or language is merely a version of some other !Kwi variety or language, and
that on this basis there must have been mutual intelligibility between !Kwi
languages and dialects. While it is certain that the two /Xam varieties of the
Flat and Grass Bushmen6 recorded by W. H. I. Bleek and Lloyd hardly differed,
and that /Xam and =| Khomani could not have been mutually intelligible, it is
44 A. Traill
not at all clear from the meagre evidence available what degree of mutual
intelligibility existed between these and any of the other !Kwi languages. We
will never know the answer to this, but the question is worth noting because
of its sociolinguistic importance. For instance, !Xóõ, spoken in south-western
Botswana, and /’Auni share a number of related linguistic features and some
common vocabulary because they are genetically related. But the !Xóõ do not
understand a word of spoken /’Auni and would have to become bilingual or use
a lingua franca in order to communicate. If this situation applied to any extent
between the !Kwi languages of South Africa it would have had an impact on
patterns of bilingualism and language shift and contributed to the death of the
languages.
In 1973 at Nossop camp I also interviewed /Okos, a woman who claimed to be
the last speaker of what she called /Nuhci (i.e. /’Auo). This was probably close
to the truth. I went through all the grammatical and lexical material Dorothea
Bleek had published on the /’Auni language in 1937 and I found that /Okos
had maintained the language in every detail. This is extraordinary. As with Jopi
Mabinda, the last //Xegwi speaker, it seems that the language, having ceased to
be a vital means of communication, must have assumed a powerful symbolic
value which maintained the speaker’s identity in defiance of the forces that had
consumed all the other !Kwi languages of South Africa. Recently, a few more
elderly individuals with some knowledge of /’Auni or =| Khomani have been
found in Gordonia. Independently, they have preserved a strong sense of their
Bushman origins, but Afrikaans is their first language and the only language of
their children.
There are still many vital Khoesan languages spoken in southern Africa. These
are to be found in Namibia, where Ju still flourishes, and in Botswana, where
the greatest variety of Khoesan languages is found. But the attrition continues.
The language of the ‘Masarwa’ studied in 1913 by Dorothea Bleek at Khakhea
in southern Botswana is dead; Eastern =| Huã spoken in the Kweneng district of
Botswana is shrinking and severely threatened; the Tyua dialect of Sepako in
north-eastern Botswana and adjacent parts of Zimbabwe is moribund; Deti of
the Rakops area is dead; the !Xóõ of the Aminuis Reserve in Namibia, whom
Dorothea Bleek (1929: 2) studied in 1913 (she called the people /Nu //en),
is moribund. In Namibia, there has been such a dramatic shift from Nama to
Afrikaans and English that the vitality of the language is seriously threatened
(Cluver n.d.b; Haacke 1989).
Other surviving Khoesan languages are shrinking as a result of a lack of offi-
cial interest, language education policy, and the economic and social conditions
of speakers. There is also wholesale bilingualism in local varieties of Tswana.
The Khoesan languages 45
These languages are therefore threatened by pressures only slightly less dra-
matic, but no less severe, than those that led to the disappearance of the Khoesan
languages further south.
notes
1 This spelling of the more familiar ‘Khoisan’ has been adopted in this chapter following
Nienaber’s (1990) discussion and rejection of it on linguistic grounds. The conven-
tion not only affects the term ‘Khoesan’, but is extended to other familiar, related
terms. Thus ‘Khoikhoi’ becomes ‘Khoekhoe’ and ‘Khoi’ becomes ‘Khoe’ in all uses.
In this chapter ‘Khoe’ and ‘Khoekhoe’ frequently have a special linguistic sense:
‘Khoe’ denotes a family of languages, one branch of which includes the ‘Khoekhoe’
languages of South Africa (Nama, Gri, !Ora) and Namibia (Khoekhoegowab); the
other branch of the ‘Khoe’ family consists of the non-Khoekhoe languages, none
of which is indigenous to South Africa. In its non-linguistic sense, ‘Khoekhoe’ is
also applied to the people who speak or spoke one of the ‘Khoekhoe’ languages. In
this latter usage, ‘Khoekhoe’ replaces the traditional and now discredited ethnonym
‘Hottentot’. The name ‘San’, derived from the Khoekhoe word saan, is a popular
replacement for the ethnonym ‘Bushman’, which is widely perceived to be offensive.
‘San’, however, lacks any linguistic validity and it may even be confusing when used
as an ethnonym. Thus, there is no valid family of ‘San’ languages, and some ‘San’
speak Khoe languages. Equally, there is no valid linguistic family of ‘Bushman’ lan-
guages; the people commonly referred to as ‘Bushmen’ speak languages from one of
the three distinct Ju, Khoe or Southern families.
2 The term ‘Khoesan’ is linguistically misleading because it does not refer to a single
family of languages. In fact, it is applied to three genetically unrelated groups of
languages, which may be referred to for convenience as the Northern (including Ju,
!Xung etc.), Central (including Nama, Cape Khoekhoe, Naro, etc.) and Southern
(including /Xam, !Xung etc.) families (D. F. Bleek 1929). The Khoe languages do
constitute a genetic unity, but not the so-called ‘San’ languages (Northern and Southern
families), either with one another or with the Khoe languages.
3 There are currently 4,000 Bushmen living at Schmidtsdrift near Douglas in the north-
ern Cape. They represent a proportion of the South African Defence Forces’s Bushman
Battalion that was deployed in northern Namibia, and their families. The group con-
sists of about 1,200 speakers of Kxoe, a Khoe (Central family) language, and !Xũ
(!Xung), a Northern Bushman language. These languages are mutually unintelligible,
and Afrikaans is used as a lingua franca among males and as a medium of instruction
in school. Females are largely monolingual in either Kxoe or !Xũ. This large num-
ber of Bushmen represents the majority of Angolan Khoesan (L. P. Voster, personal
communication), but obviously they are neither historically nor linguistically South
African Khoesan.
4 The Bushmen at Bethulie had links with those from the Colesberg and Aliwal North
areas. An interview with one in 1877 offers a rare piece of evidence concerning the
lack of mutual intelligibility between certain neighbouring !Kwi varieties. The man,
Toby or Kwa-ha, said: ‘I can speak Bushman language well, but I cannot under-
stand the Bushmen of Riet River; their language is “too double”’ (Orpen 1877: 85).
This presumably refers to a variety spoken about a hundred kilometres north in the
46 A. Traill
Reddersburg area of the Orange Free State. I am grateful to T. Dowson for this
reference.
5 Vaalpens was an ethnonym used by various commentators and Bushmen themselves
to refer to groups of Bushmen from the south-western corner of the then Bechuanaland
Protectorate.
6 The Grass Bushmen lived in the Katkop hills between Kenhardt and Brandvlei. The
Flat Bushmen lived at various waterholes between Kenhardt and Van Wyk’s Vlei
(Deacon 1986).
bibliography
Anders, H. 1934/5. ‘A note on a southeastern Bushman dialect’. Zeitschrift für
Eingeborenen-Sprachen, 25: 81–9.
Beach, D. M. 1938. The Phonetics of the Hottentot Language. Cambridge: Heffer.
Bennie, J. 1823. Incwadi yokuqala ekuteteni gokwamaXosa. Grey Collection, South
African Library, Cape Town.
Bleek, D. F. 1927. ‘The distribution of the Bushman languages in South Africa’. In
Festschrift Meinhof. Hamburg: Augustin, pp. 55–64.
1929. Comparative Vocabularies of Bushman Languages. Cambridge: Cambridge
University Press.
1936. ‘Speech of animals and moon used by the /Xam Bushmen: notes on pho-
tographs’. Bantu Studies, 10: 163–203.
1937a. ‘Grammatical notes and texts in the /Auni language’. In J. D. Rheinallt
Jones and C. M. Doke (eds.), Bushmen of the Southern Kalahari. Johannesburg:
Witwatersrand University Press, pp. 195–200.
1937b. ‘/Auni vocabulary’. In J. D. Rheinallt Jones and C. M. Doke (eds.),
Bushmen of the Southern Kalahari. Johannesburg: Witwatersrand University Press,
pp. 201–20.
1942. ‘The Bushman tribes of southern Africa’. In A. M. Duggan-Cronin (ed.), The
Bushman Tribes of Southern Africa. Kimberley: Alexander McGregor Memorial
Museum, pp. 1–14.
1956. A Bushman Dictionary. New Haven, Conn.: American Oriental Society.
Bleek, W. H. I. 1873. Report of Dr. Bleek concerning his Researches into the Bushman
Language and Customs, Presented to the Honourable the House of Assembly. Cape
of Good Hope Official Publications A17-’83.
1875. Second Report concerning Bushman Researches by W. H. I. Bleek, Pre-
sented to the Houses of Parliament. Cape of Good Hope Official Publications
G 54-’75.
Bleek, W. H. I. and L. C. Lloyd 1911. Specimens of Bushman Folklore. London: Allen.
Boekkooi, J. 1988. ‘Murdered: the last of the Mountain Bushmen’. Sunday Tribune,
4 December.
Brenzinger, M. 1992. ‘Language shift in East Africa’. In R. K. Herbert (ed.), Language
and Society in Africa. Johannesburg: Witwatersrand University Press, pp. 287–303.
Brenzinger, M., B. Heine and G. Sommer 1991. ‘Language death in Africa’. In
R. H. Robins and E. M. Uhlenbeck (eds.), Endangered Languages. Oxford: Berg,
pp. 19–44.
Carstens, P. 1966. The Social Structure of a Cape Coloured Reserve. Cape Town: Oxford
University Press.
The Khoesan languages 47
1 INTRODUCTION
The present domain of the Bantu languages extends in an eastward progression
from the Cameroon–Nigerian borderlands through the equatorial zone to the
Kenyan coast and then southwards to the Cape. The geographic expanse is
thus enormous, occupying fully one-third of the African continent, as is the
degree of linguistic diversity. On account of the well-known problem of distin-
guishing languages and dialects, a precise count of the Bantu languages is not
possible; their number is conservatively reckoned at about four hundred. Some
250 million people speak one or more of the Bantu languages as mother tongues
today.
This chapter considers the linguistic sociohistory of southern Africa, with par-
ticular attention to the Bantu languages. The term ‘Bantu’ (Bâ-ntu) was coined
by W. H. I. Bleek in 1857 or 1858 (Silverstein 1993 [1968]), and popularised
through his Comparative Grammar (1862). Bleek noticed certain recurrent
patterns among widely distributed languages on the African continent, and
he happened upon the composite term Bâ-ntu to name these languages and
their speakers. The prefix ba-, the so-called class 2 prefix, is the plural marker
for many noun stems with human referents in these languages.1 The stem *-ntu
names representatives of the given class; hence Bantu is conveniently translated
as ‘people/persons’. (Cf. Zulu abantu; Northern Sotho batho; Tsonga vanhu;
Venda vhathu, etc.) Bleek’s coinage follows the frequent onomastic tradition
where a group self-identifies itself as ‘(true/real) people’, reserving ethnonyms
for outsiders.2
It is not possible to date with any certainty the arrival of the first Bantu-
speaking Africans into the territory of present-day South Africa. It is clear,
however, that their arrival preceded the arrival of European settlers by many
centuries. The notion, long promulgated by many settler historians, that the first
Bantu speakers crossed the Limpopo around 1652 is convenient fiction. To the
contrary, archaeological research shows conclusively that there were Bantu-
speaking groups who kept livestock and practised cultivation by at least 300 ad
(Maggs 1991: 37). The precise relationship between these prehistoric groups
50
The Bantu languages: sociohistorical perspectives 51
le
Ni
er
Nig
NON-BANTU NIGER-CONGO
Congo
BANTU
A
E E
C
B D
F
L
G
H
L
M P
K
N
R
hereby caused distribution of the nouns into classes or genders’. Apart from
the misidentification of prefixes as pronouns, Bleek’s statement is a reasonable
description of that structural feature which continues to figure prominently in
characterisations of the Bantu languages. Guthrie (1948: 11) lists two criteria
which are determinative in the identification of a language as belonging to the
Bantu family:
1 a system of grammatical genders, usually at least five . . .
2 a vocabulary, part of which can be related by fixed rules to a set of hypothetical
common roots.5
The linguistic label Bantu is thus reserved for a group of languages exhi-
biting marked similarity in structure and vocabulary, both of which are pre-
sumed to derive from common ancestry. Guthrie (1948) developed a referential
scheme for the Bantu languages, which divided them into geographical
zones labelled A–T (later revised as A–S); subdivisions within the zones were
grouped numerically, e.g. S.30 names the Sotho-Tswana languages, S.31
Tswana, S.33 Southern Sotho, etc.; each of the latter may name a dialect cluster,
e.g. S.31b Kgatla, S.31d Kgalagadi, etc. Following the tradition of historical
The Bantu languages: sociohistorical perspectives 53
rber
Be
A r a b i c
Nubian
gay Beja
Son
Ful Bamba
ra Hausa Kanuri
Amharic
Oromo
Akan Yoruba Zande
Dinka Somali
Igbo
Sango
Luo
Lingala
Kongo
Luba Swahili
Hadza
Sandawe
Bemba
Phyla Makua
Afroasiatic Shona
Nama
Nilo-Saharan Tswana
ho
Niger-Congo Sot
Zulu
Khoisan
Xhosa
linguistics, the term Proto-Bantu is reserved for the hypothetical ancestor lan-
guage, the Ursprache, of the modern descendants distributed throughout the
subcontinent.
Greenberg (1963) was the first to establish the clear relationship between the
Bantu languages and related languages called Benue-Congo, most of which are
spoken in Nigeria and Cameroon. Prior to Greenberg’s classification, Bantu was
generally taken to be an independent language family. The Benue-Congo family
of languages is in fact a subfamily within a much larger phylum generally known
as Niger-Congo. Grimes (1996) lists 1, 436 languages for Niger-Congo, making
it the largest of the world’s phyla. The other major, independent language phyla
of Africa include Afrasian (Afroasiatic), Nilo-Saharan6 and Khoesan.7 Only
the latter has significant historic presence south of the equator; indeed, much of
54 R. K. Herbert and R. Bailey
Bantoid
Tikar South
Mambiloid
Ekoid-Mbe Nyang
3.1 Bantoid language relationships (source: Williamson and Blench 2000: 35)
There was much looseness in this framework, and it has been subjected to con-
siderable debate and elaboration over the past three decades. The most pressing
issue is the relationship between Bantu and adjacent languages, a problem that
is acute in south-western Cameroon where there are a number of languages
that are transparently related to Bantu, though ‘not Bantu’. The label Bantoid
is now often used to name Bantu and these closest relatives, i.e. Bantu is today
seen as a subgroup within a larger Bantoid unit.
A number of scholars have offered opinions and models on the question
of Bantu’s closest linguistic relatives within the Benue-Congo sub-branch of
Niger-Congo. The major dispute is about the exact placement of Bantu within
the larger Bantoid branch; the delimitation of Bantu from the other Bantoid
languages is certainly not straightforward (Watters 1989: 404ff.; Blench 1997:
94; Maho 1999: 40–5; Williamson and Blench 2000: 34–6). The details of this
debate are not relevant to present purposes; the integrity of (Narrow) Bantu
56 R. K. Herbert and R. Bailey
within the Niger-Congo tree is generally not disputed, though see Bennett and
Sterk (1977) for a contrasting view. Williamson and Blench (2000: 35) pro-
pose that there is a basic split within (Narrow) Bantu between the North-west
(Guthrie’s Zones A, B, C and part of D) and ‘Other Bantu’, though they recog-
nise that there are also some grounds to separate East and South Bantu from all
of the remaining Bantoid languages. Piron (1998) gives a detailed description
of the analytical problems in subgrouping within Bantoid.
3.3 The spread of Bantu languages: out of the forest and beyond
The notion of a ‘Bantu expansion’ originated in the late nineteenth century. The
idea, originally promoted by Sir Harry Johnston (1858–1927) and his contempo-
raries, was that marauding Bantu armies marched forward, conquering land and
populations, with an unstoppable military might. The march of Bantu-speaking
invaders was generally accepted as the explanation for the spread of Bantu
languages throughout the present domain. In its most common form, the in-
vaders comprised males alone; presumably, women and children would have
slowed the march. There were few details regarding chronology, but it was
generally believed that the ‘Bantu hordes’ arrived in South Africa during the
period of European settlement at the Cape. Werner (1933) suggested that local
men were killed, and that the invaders married the local women. As Herbert
(chap. 15, this volume) notes, the ‘myth of invading Bantu males’ has been
seriously overplayed in the literature and is deficient in important conceptual
and analytical details. In place of militaristic ‘invasions’, ‘conquerors’, ‘armies’
and ‘hordes’, it is perhaps more accurate to think of opportunistic agricultural
migrants.
It was only in the 1960s that Roland Oliver’s (1966) work in African
history attempted a synthesis of the material from linguistics, history and
archaeology. Advances in archaeology, coupled with increased sophistication
in historical linguistics, allowed scholars to set aside the myth of the invad-
ing Bantu and replace it with another scenario. In this latter view, the spread
of the Bantu languages was linked to the spread of metalworking, agricul-
ture and village life, and (perhaps) the replacement of small-statured hunter-
gatherers with larger (presumed) speakers of Bantu languages (Phillipson 1977,
1985).
By the 7th century ad, related people had spread throughout subequatorial Africa.
Since these later occurrences coincide with an introduction of agricultural economies,
new forms of society and metallurgy, and domesticated plants and animals previously
unknown, even in wild form, in the subcontinent, it is safe to conjecture that this
constellation of traits was brought into the region by people who had not previously
lived there. Since these early settlements are all found in regions now populated by
The Bantu languages: sociohistorical perspectives 57
Bantu speakers, it is also reasonable to believe that the arrivals of agricultural societies
and Bantu speakers are synonymous. (Vogel 1997: 436)
That is, population growth replaced conquest as the modus operandi of Bantu
language spread. By this time, radiocarbon dates had yielded sites in the far
south from 1,600 years ago. The original outward movement of peoples and
languages (Eastern Bantu) from the purported homeland was assumed to have
taken place in the past 2,500–3,000 years.
Certain difficulties in the dispersal model became apparent in the 1980s.
Most notably, there was a mismatch of data from Iron Age archaeology, which
was concentrated in central and southern Africa, with the crucial missing data
from historical linguistics, especially in southern and central Cameroon, the
hypothesised homeland for the Bantu languages. The chief critic of the expan-
sionist viewpoint was Jan Vansina (1979, 1980). In more recent work, however,
Vansina (1995) has accepted the idea of an expansion from a homeland located
somewhere in north-west Cameroon. Notably, he rejects the idea of a single
great expansion caused by population pressure from the adoption of agriculture
and iron working.
The terms ‘South-eastern Bantu’ and ‘Southern Bantu’ are often used inter-
changeably to refer to the languages of Zone S, though the latter term is
sometimes used to refer to this set of languages, excluding Shona. Ehret
(1999: 53) excludes Shona from a ‘Southeast-Bantu’ subgroup which includes
S.20–60 plus Lozi (K. 21), which he describes as ‘a nineteenth century creole
of an S.30 language’ (1999: 49). In Ehret’s scheme, there are four co-ordinate
branches on an intermediate family tree of which South-east Bantu is one:
Kaskazi Kusi
The evidence for this particular structure is based upon shared innovations,
though the published data do not address the integrity of the South-east Bantu
subgroup itself, which seems to be taken for granted.
There are regular sound correspondences in much inherited lexical material
for the South African Bantu languages.
Pedi Tsonga Venda Zulu gloss
/f/ /h/ /fh/ /ph/
-fála -hala -fhálá -phála scrape
-féla -hela -fhélá -phéla to end (v.i.)
leswafó hahu fhafhú iı́phaphú lung
lefele hele (beté) iı́phela cockroach
whether differentiation into the distinct language groups occurred in East Africa
before the southward movements of peoples or during/after the southward
spread. The nine officially recognised South African languages are conveniently
subgrouped:
S.40 Nguni languages (Ndebele, Swati, Xhosa, Zulu)
S.30 Sotho languages (Northern Sotho, Southern Sotho, Tswana)
S.53 Tsonga (classed with Tshwa, Ronga, etc. in Mozambique)
S.21 Venda (isolate, though possibly linked to Shona in Zimbabwe)9
The relations among the various Southern Bantu languages are usually repre-
sented along the lines sketched by Doke (1967):
Strictly speaking, only the lower-level groupings were asserted by Doke to have
any genetic meaning; the higher level arrangement of five language groups was
exclusively referential. Differentiations within subgroups Nguni and Sotho are
best viewed as local phenomena, though there are some outstanding questions
in this regard. One such question concerns the precise relationship of clus-
ters of languages within Nguni. Mainly on the basis of phonological facts,
it has been traditional to distinguish Zulu and Xhosa forms from a disparate
collection of clearly related languages. This separation goes back to the work
of nineteenth-century scholars, most notably A. T. Bryant. The term zunda is
sometimes used for the former; the latter, often called tekela or tekeza languages,
include Swati, Northern Ndebele, Hlubi, Baca, Phuthi and others. Among the
phonological features setting Tekela apart is a distinctive affrication of alveolar
stops (Z. -thathu, ‘three’; umuntu, ‘person’; thina, ‘we’ vs. Sw. -tsatfu, umuntfu,
tsina) and the correspondence of zunda /z/ for tekela /t/. The more common ci-
tation name for the Swati language is in fact the Zulu form Swazi; compare also
Zulu izimbuzi, ‘goats’ with Swati timbuti. Further, Zulu and Xhosa have three
distinct points of articulation for click consonants whereas most of the Tekela
dialects have a single position:
Zulu Swati gloss
-xoxa -coca chat (lateral click [||])
-qala -cala begin (prepalatal click [!])
-cela -cela ask for (dental click [|])
The Bantu languages: sociohistorical perspectives 63
Doke’s (1967) representation of the relationship among the various Nguni lan-
guages raises an important issue. In particular, the status of the Tekeza group
as an ancestral unit opposed to Zulu and Xhosa has never been demonstrated;
indeed, there is little reason to class these languages together other than on
the basis of shared retentions that distinguish them from Zulu and Xhosa. The
current linguistic situation in South Africa is complex on account of pressure
from the standard languages, but there is little or no evidence to support the idea
that Tekeza represents a linguistic subgroup. More likely, the so-called Tekeza
group is a collection of languages spoken by peoples who were not subjugated
and assimilated to Zulu. In cases where Xhosa and Tekeza share features, it is
most safely assumed that Zulu has innovated.
The larger issue for language historians is whether the collection of Zone S
languages diverged prior to their arrival in southern Africa. There is no convinc-
ing evidence for a Proto-Southern Bantu (‘Proto-Zone S’) language from which
the present-day languages descend. The demonstration of such a unit would de-
pend on a set of innovations which characterise this group of languages and
distinguish it from other Bantu languages. In the absence of such evidence,
the genetic ‘unity’ of the Southern Bantu languages needs to be called into
question.10
Huffman (1989), using ceramic evidence, argued that Sotho-Tswana and
Nguni movements are reflected in separate migrations and ceramic paths, which
he terms Moloko and Blackburn, respectively.
The earliest Iron Age sites in southern Africa are all in areas either still or
comparatively recently occupied by Sotho-Tswana. The Nguni seem to have
been later arrivals. Following Louw and Finlayson (1990), Janson (1991/2) has
suggested that Makua (Guthrie’s P.31), a Bantu language presently spoken in
Mozambique, and Sotho-Tswana share a period of common development in
present-day Zimbabwe; Janson’s hypothesis is based on similar developments
in the two sets of languages, which he argues must be shared innovations.
These innovations include the evolution of the Proto-Bantu prenasalised voiced
stops into voiceless unaspirated stops, e.g. *mp > p; this unusual change does not
occur elsewhere in Bantu. Bailey (1995b: 47) suggested that this change might
be due to a Khoe or San substratum. There are other similarities between Makua
and Sotho languages, especially the Sotho varieties that show less evidence of
contact with Nguni speakers. The possibility of historical links between Sotho
languages and Makua was first noted by van Warmelo (1927), though this
observation is generally not cited in the later literature.
Janson suggests that the Sotho-Makua community was displaced in the
eleventh century by incoming Shona, Chewa and Sena groups with the results
that Makua was removed to the north and east, separated from the other Southern
Bantu languages, and Sotho-Tswana moved to the south and west, where it came
64 R. K. Herbert and R. Bailey
WESTERN
B ANTU
HEA RTLAND Early Iron Age
Nucleus
(Ng wana)
uni)
-Ts
tho
(S o
KO
MOLO
N
UR
KB
AC
BL
into contact with the Nguni. This hypothesis suggests that Makua is historically
a Southern Bantu language, which has undergone change in contact with other
Bantu languages; in certain regards, it is not typical of the languages of the area
in Mozambique where it is today located.
Apart from the low-level subgroups identified above, it is not possible to as-
sert anything definitive regarding possible relations within the class of Southern
Bantu (Zone S) languages. In part, the difficulty arises from language con-
tact and diffusion over the past millennium. From certain limited perspectives,
Nguni and Tsonga seem closely related. Baumbach (1987: 2) suggested that
Tsonga is properly viewed as part of the Nguni cluster commonly called Tekeza
or Tekela, which is co-ordinate with Zulu and Xhosa.
Thus, the Tsonga group is, according to Baumbach, co-ordinate with the
so-called Nguni Tekeza dialects (Swati, Bhaca, Phuthi, etc.) and, ultimately, a
descendant of Proto-Nguni. However, Baumbach’s proposal has met with little
enthusiasm.11 One cannot ignore the possibility that any linguistic similarities
The Bantu languages: sociohistorical perspectives 65
Tekeza Tsonga
between Tsonga and Nguni are the result of longstanding contact, or shared
Southern Bantu heritage. There are few, if any, linguistic innovations that are
shared by Tsonga and Tekeza, and the demonstration of such shared innovations
is generally taken as a prerequisite to the postulation of common ancestry. As
Herbert (chap. 16, this volume) notes, Zulu incursions into Tsonga-speaking
territory are of longstanding duration, and Zulu was the prestige and dominant
group. How such ‘despotic domination’ might have affected language practice
is a topic for investigation. Louw and Finlayson (1990: 403) noted ‘strong
affinities’ between Nguni and the Tsonga of South Africa and neighbouring
Mozambique, but they also reported that more northern forms of Tsonga show
‘strong influences’ of Shona.
Prehistoric language contact is, of course, one of the most confounding fac-
tors in reconstructing the linguistic history of the region. We do not know, for
example, how historic movement of peoples displaced during the Mfecane (see
chapter 1) might have affected language patterns. In part, the problem becomes
all the more acute when we recognise the non-tenability of early models of
monolithic movements, e.g. the notion that any whole, bounded speech com-
munity (‘a tribe’) relocated itself with no effect on language practice. The more
likely scenario is that some populations were absorbed and that some displaced
populations co-mingled to form new speech communities. Further, the data
available to us are not ideal since language variation has been dramatically
reduced on account of language standardisation and the promulgation of stan-
dard languages over the past seventy-five years in South Africa.
For contact among the Bantu languages, there is the further problem that
the languages themselves, precisely on account of their shared ancestry, are
broadly similar in structure and shared vocabulary. The analyst is on firmer
ground in reconstructing the effects of contact with Khoesan speakers, the
earlier inhabitants of the region, on inmigrating Bantu languages (see chap. 15,
this volume).
At a general level, there is a rather poor understanding of historical rela-
tions among the various Bantu language clusters (Guthrie’s ‘groups’). To a
66 R. K. Herbert and R. Bailey
large extent, linguistic subgrouping within these groups (e.g. within Nguni, or
Sotho-Tswana, or Tsonga) has not progressed significantly. Nicolaı̈ (1998) pro-
vides a useful review of the many problems presented to historical linguists
working exclusively with non-text-based data.
This section is concerned with the traditional four clusters of Bantu lan-
guages in South Africa: Nguni, Sotho, Tsonga and Venda. As noted above,
there is no convincing evidence for in situ differentiation of these clusters. There
are so-called mixed languages, e.g. Phuti and Northern Transvaal Ndebele, but
these are both more appropriately viewed as Nguni languages that have been
Sotho-ised relatively recently rather than points on a linguistic continuum.
Phuthi is relatively well described (Mzamane 1949; Donnelly 1999), whereas
Northern Transvaal Ndebele (Ndrebele) is now virtually extinct under the influ-
ence of Northern Sotho. Wilkes (1999) provides a useful statement of Northern
Sotho/Tswana influences on Southern Transvaal Ndebele.12 Although clearly a
Nguni language, Southern Transvaal Ndebele shows important Sotho-Tswana
influences in lexicon, phonology, morphology and syntax. The topic of language
contact has been woefully understudied in South African linguistics, with the
exception of urban vernaculars.
political and education discourse was the location of ‘real’, ‘true’ or ‘proper’
Africans in the traditional homeland. So, for example, Zulu speakers resident
in South Africa’s cities were ‘out of place’.13 Within the education arena,
adherence to rigidly conservative rural language standards worked to the se-
vere disadvantage of urban schoolchildren, many of whom consistently failed
‘mother-tongue’ matriculation examinations.
The delimiting of South African indigenous languages has traditionally been
associated with a delimitation of population groups. For a variety of reasons,
central government sought a small, manageable number of population groups.
Eventually, the language = cultural group equation was extended as a justifi-
cation of the failed homeland policy of the apartheid government: language =
culture = homeland. On the basis of this extension, the government sought to
deny citizenship and residence rights to its African populations.
Language census data are notoriously unreliable since there is little interest
in defining what counts as ‘a speaker’. In South Africa, it is often taken to
be axiomatic that Zulu persons speak Zulu, Xhosa persons speak Xhosa, etc.
Bearing in mind their inherent unreliability, population numbers for the relevant
language/population groups according to recent census data are:14
These data are asserted to reflect the number of mother-tongue speakers. Bearing
in mind that the majority of South Africa’s population is urban, multilingualism
is widespread.
There have been several calls to ‘harmonise’ the Nguni languages into a sin-
gle written standard, the most recent by Alexander (1989). Since mother-tongue
speakers of the various Nguni languages account for approximately 45 per cent
of the national population, the implications and potential education and literacy
consequences would be considerable. Alexander also called for a similar har-
monisation of the Sotho-Tswana languages, which are spoken by an additional
24 per cent of the population. The two remaining African languages, Venda
and Tsonga, which are spoken by 2.2 per cent and 4.4 per cent respectively, are
language isolates within South Africa.
Harmonisation was never seriously considered, and there was little popular
support for the idea, which speakers perceived as a threat to their ‘traditional’
ethnic identities. Understandably, there was little appreciation of the relative
recentness of the creation of standard languages, and for the government’s role
in the selection, identification and reification of those ‘traditional’ identities.
Whether harmonisation was linguistically practicable is an open question. See
Msimang (1998) for a useful discussion of related issues in the harmonisation
debate.
68 R. K. Herbert and R. Bailey
(The Southern Sotho examples have not been marked for tone in the above
examples.)
Tswana was originally known as Western Sotho, and the indeterminacy of
naming the language and its speakers are once again instructive. The Kgatla
dialect is the basis of the South African written standard. Some of the varieties
included within the scope of Tswana, e.g. Sekgalagadi, are sufficiently diver-
gent to warrant consideration – on linguistic grounds – as separate languages.
Schapera and van der Merwe (1943: 3) noted that Sekgalagadi was no closer
to Tswana than it was to Pedi or South Sotho. Janson (1995: 401) reported that
Tswana and Kgalagadi are not mutually intelligible. Present-day speakers of
Kgalagadi are dispersed over a large part of Botswana, in the Kalahari desert
or around the fringes. There is some suggestion that Kgalagadi represents a
‘purer’ form of the language, uninfluenced by surrounding languages. This is,
of course, not a tenable linguistic position to adopt. The promotion of a single
identity follows from the use of Standard Tswana in the educational context.
Janson (1991/2) argued that the phonology of Kgalagadi is more conservative
than the rest of Sotho-Tswana.
70 R. K. Herbert and R. Bailey
The case of Northern Sotho is also instructive. The ‘ethnic group’ Northern
Sotho was demonstrably invented by the Nationalist government to unify a
diverse set of people, who formerly were called ‘the Transvaal Sotho’, some-
times distinguishing Northern Sotho and Eastern Sotho. Pedi, the language of
one prestigious group, was selected as the basis for the standardised language.
The range of linguistic and cultural diversity within the Northern Sotho group is
very wide, so wide that van Warmelo declared that the ‘Northern Sotho language
is a fiction’ (1974: 76). Van Warmelo also noted (1974: 72) that it was difficult
to draw any real boundary between Tswana and the Northern Sotho cluster on
linguistic grounds, and that the basis for differentiation was entirely political
and administrative. The Northern Sotho peoples lack any traditional endonym,
i.e. a name used internally to refer to the group of people, and were known as
maAwa, based on a common form for the word meaning ‘no’ (NS awa; SS che;
Tsw. nyaa) (Herbert 1996: 1346).
In addition to the influence of North Nguni on South Sotho there is con-
siderable evidence of contact from Nguni (perhaps North Nguni again) on the
Sotho-Tswana group of languages as a whole. Laterals are absent from peri-
pheral Sotho languages and dialects such as Kgalagadi, Phalaborwa, Lobedu,
Dzwabo, Kgaga, Hananwa, Tlokwa etc. Bailey (1995b: 45) noted a relation
between the distribution of laterals in Nguni, Sotho-Tswana, and Tsonga, and
the spread of the Nguni central cattle pattern. Laterals are absent from the other
Zone S languages, and these do not show direct evidence of Nguni contact. At
another level, it is worth noting that the standardised forms of Sotho-Tswana
languages are based on the speech of the largest and most successful groups
(Kgatla, Ngwato, Hurutshe, Pedi, Southern Sotho), and it is these groups that
were most affected by Nguni culturally and linguistically, absorbing other Sotho
groups. The so-called relic languages such as Kgalagadi, Pai, Phalaborwa and
Dzwabo are perhaps better sources for data on the pre-contact character of
Sotho-Tswana.
by Swiss missionaries early in the twentieth century, who bestowed the name
Thonga, a Zulu form, upon the group (Harries 1988). Most of the people are
now content to call themselves vaTsonga and their language xiTsonga. How-
ever, there is an alternate name for part of this group, Shangaan, which is an
eponym for one of the Zulu chiefs, Soshangane, who subjugated many clans
in the nineteenth century. This label is rejected by those clans that were never
subjugated, but preferred by many who were. The analyst is thus presented
with a group of people who are demonstrably similar in language and custom,
with some sense of shared history, who variously self-label as vaTsonga and
maShangana and call their language either xiTsonga or xiShangana. As noted
above, a few analysts, most notably Baumbach (1987), have suggested a his-
torical affiliation between Tsonga and Nguni, but this idea has met with little
support. Bailey (1995b: 45) noted: ‘Impressionistically, the Tsonga group as
a whole shares more phonological and grammatical features with Nguni than
with any other Bantu language group. It may be that the relationship of genetic
differentiation between Nguni and Tsonga occurred in situ.’ However, one must
also allow that there has been considerable Nguni-isation of Tsonga varieties
over several centuries, and in particular as a result of the Mfecane disturbances
in the nineteenth century. The notion that Nguni and Tsonga (and other lan-
guages of Mozambique such as Ronga and Tshwa) differentiated in the present
domain poses a significant challenge to the historical linguist.
The other Tsonga group in South Africa is the so-called Tembe Thonga
of KwaZulu-Natal, most closely related to the Ronga of Mozambique. This
language is virtually extinct, though there are some older speakers, particularly
women, who have full facility in the language. The label Gondzze is sometimes
used for this variety of speech.
the ruling lineage in most chiefdoms came ‘from the north’, i.e. north of the
Limpopo in present-day Zimbabwe. This would explain the higher frequency
of Shona lexical items in the courtly language. Equally striking to the Shona
influence is the absence of Nguni influence, which is pervasive in neighbouring
Sotho languages; again, this suggests that the Venda were within a protective
orbit when the Nguni were penetrating elsewhere in southern Africa.
Venda is well described (Poulos 1990; van Warmelo 1989), and there is also
a description of the musanda courtly language by Khuba (1993). Despite the
lexical borrowings from Shona, there is no good evidence for Shona morpho-
logical or phonological influence on Venda. It should be noted that the influence
of Shona on other neighbouring languages, e.g. Northern Sotho and Tsonga,
has also confounded historical investigation.
5 CONCLUSION
A consideration of the socio-history of Southern Bantu languages reveals that
there are more questions than answers available to scholars. While there is no
question about the Bantu character of all nine of the officially recognised in-
digenous South African languages and the linguistic classification of these
into four distinct groups (Nguni, Sotho-Tswana, Tsonga and Venda), it is
not possible to demonstrate whether two or more of the latter groups repre-
sent a valid (higher level) linguistic subgroup. Indeed, the historical unity of
the Southern Bantu languages (with or without Shona) remains an empiri-
cal question for investigation. It is worth noting that this characterisation is
largely true for the vast majority of Bantu language groups. The extent to which
prehistorical contact, bilingualism, movement and so forth obscured speech-
community boundaries and eventually confounded linguistic inheritance is an
obvious complication for historical linguistics. At the same time, linguists are
on firm ground in recognising an ‘eastern quality’ for all of the Southern Bantu
languages.
Prior to standardisation, the linguistic situation in South Africa no doubt con-
sisted of a chain of language varieties rather than recognisable, homogeneous
speech communities. The creation of ‘tribal groups’ is clearly, in some very
large measure, a product of colonialism and its residue (see e.g. Harries 1988).
One needs to recognise that the number nine is simply the output of histori-
cal accidents and design perpetrated by missionaries and government agents.
Of course, the successful promulgation of these nine ‘identities’ among the
indigenous population has been variously, though largely, successful.
The challenges for sociolinguists working in South Africa are manifold.
These include the uncovering of linguistic history and relationship with a view
to reconstructing the lineage of the nine indigenous languages with official
status. Equally daunting is the challenge of discovering the effects that language
standardisation, under the aegis of the former language boards, has had on
linguistic diversity. In the present dimension, the challenges are to document
patterns of language use and change. Within the scope of the latter topic is
the constitutional directive to provide for the development and protection of
the country’s linguistic resources. The sociolinguistic future of South Africa’s
indigenous languages will depend on the creation of conditions and incentives
for their maintenance and promulgation throughout the citizenry.
notes
1 Doke (1993[1960]: 80) notes that, according to Alice Werner, the original coinage
may have been Sir George Grey’s. The form builds upon an earlier suggestion by
Barth that they be called ‘the Ba-languages’.
2 Indeed, the form bantu and its cognates reveal such exclusive marking even in the
modern-day languages. Forms deriving from muntu (sg.) in the various languages
74 R. K. Herbert and R. Bailey
name a person like the speaker; Europeans and other non-Bantu speakers are excluded
from the scope of the term. Bantu are ordinary people, ‘true people’. Whites cannot be
bantu since they lack *ubuntu, ‘the quality of personhood’. This observation is not to
argue that the Bantu languages or their speakers are inherently racist. Rather, the point
is that Bleek’s original coinage nicely captures the scope of his intended distribution
since outsiders are excluded.
3 There is some ongoing debate within South Africa as to whether African language
names should be cited with or without the language appropriate prefix, e.g. Zulu
or isiZulu, Tsonga or Xitsonga/xiTsonga. Within this chapter, languages names are
cited in their most common forms within the scholarly literature, which are usually
prefixless. Bailey (1995a: 34–5) identifies many of the structural problems inherent
in the proposal that native forms, including class prefix, be the citation form in other
languages.
The other onomastic controversy surrounding language names in South Africa is
the family name Bantu, which despite its genealogy was applied to racist discrimina-
tory policies during a long period in recent South African history. As an ethnonym,
the form is highly offensive in South Africa. Its usage is restricted to languages
(Bantu languages, Bantu-speaking peoples). Khumalo (1984) suggested that the term
Sintu be used in its place, based on the Zulu/Xhosa prefixal form isi- (< Proto-
Bantu *ki-), e.g. isiZulu, and Sotho-Tswana se-, e.g. Setswana. This class 7 prefix
precedes most language names in South Africa, e.g. isiZulu. More recently, Maho
(1999: 264) makes a similar proposal and suggests that Proto-Bantu be appropriately
named Kintu, which would presumably have the meaning ‘the (true) language’. The
data on Bantu language names are complex; lu- and li- prefixes are common outside
the south and east, and many ki- (and its derivatives) names show an initial l- that
may be a remnant prefix. Venda is unique among the southern languages in hav-
ing two endonyms, Tshivenda (< *ki-) and Luvenda (<*lu-). The latter form has a
more restricted scope than the former, being used exclusively to name the language,
whereas Tshivenda also refers to custom, to ‘the Venda way’. The latter is the more
common form today, although Luvenda may be used to refer to ‘very good’ forms of
speech.
4 The other two co-official languages are the settler languages, Afrikaans and English.
5 Guthrie adopts the term ‘Sub-Bantu’ for languages in which the agreement system is
fragmentary or missing; he used ‘Bantoid’ for languages meeting his first criterion
but not the second, i.e. languages exhibiting prefixal agreement but lacking cognate
vocabulary with Bantu (1948: 19). It is important to note that the current use of the term
Bantoid is quite different; it is used to name a larger unit of which Bantu languages
form a subgroup. This usage is attributed to Greenberg (1963).
6 A few scholars give credence to the idea of a genetic relationship between Niger-Congo
and Nilo-Saharan. This idea was suggested by Gregersen (1977) under the heading
Kongo-Saharan. In present schemas, Niger-Congo is seen as most closely related
to Central Sudanic, with which it is a co-ordinate branch within a Niger-Saharan
macrophylum (Blench 1997: 99).
7 Depending on the inclusion of Madagascar within the African continent, a fifth lan-
guage family is sometimes noted: Austronesian (Malayo-Polynesian). Malagasy, the
national language of Madagascar and the only representative of this language family
in Africa, was brought to Africa from insular Southeast Asia less than two thousand
The Bantu languages: sociohistorical perspectives 75
years ago. Good arguments are made for a Bantu stratum in Malagasy (Dahl 1954),
but these do not affect the central concerns of the present discussion.
8 Nurse (1997: 168) refers to ‘the languages of southern Africa (Zone S)’ as a linguistic
subgroup that is ‘tacitly assumed by many but not really proved’.
9 A possible link between Venda and Shona is a controversial topic. The strongest
evidence for such a relationship comes not from linguistics but rather archaeology.
Huffman (1996) provides the relevant evidence. From a linguistic point of view,
there are intriguing occurrences such as the use of some Shona terms in the courtly
musanda language (Khuba 1993). It may be, however, that the pre-Shona exercised
a kind of overlordship relationship with the Venda.
10 There are, however, interesting data here. For example, the Proto-Bantu form for
‘ear’, */-tú/ or */-kutú/, is not attested in the southern region. All of the Zone S
languages have a form based on the root */-njebé/ (class 9). This may be evidence
for a Proto-Southern Bantu innovation; however, one cannot rule out diffusion as an
explanation for such sporadic examples.
11 One of the present authors (Bailey) believes that there was a more significant Tsonga
presence in the historical region now known as KwaZulu-Natal. According to Bailey,
pre-Tsonga speakers were overwhelmed by incoming Southern Nguni speakers, and
this Tsonga substratum is cited as the source for the Tekela accent. Vocabulary within
the Tekela group is, however, overwhelmingly of Nguni origin.
12 There is some confusion surrounding the language name Ndebele, which is ap-
plied to three distinct entities: Zimbabwean Ndebele, Northern Transvaal Ndebele
and Southern Transvaal Ndebele. There is occasional confusion in this regard, e.g.
Grimes (1996) notes that Southern Transvaal Ndebele is sometimes called a dialect of
Northern Sotho. Despite contact influences from Northern Sotho, Southern Transvaal
Ndebele is unambiguously a Nguni language. Northern Transvaal Ndebele, described
by Ziervogel (1959) is now extinct, having been replaced by Northern Sotho.
The zeal with which the Nationalist homeland policy was implemented led to the
elevation of several minor dialects to the status of official languages, e.g. Ndebele
(Southern Transvaal Ndebele) and Swati, both North Nguni dialects. Until the 1980s,
these languages were classed as dialects of Zulu, and Zulu materials were used
in education without major difficulty. Ndrebele (Northern Transvaal Ndebele) was
spoken over too dispersed an area for a homeland to be consolidated while a history
of widespread bilingualism with Pedi rendered it bureaucratically unnecessary. Thus,
the ‘tenth’ indigenous Bantu language of South Africa was rendered unnecessary
and obsolete.
13 The failed homeland policy of the Nationalist government sought to legislate this
sense of order by stripping Africans of their South African citizenship and replacing
it with citizenship in one of the bantustan creations.
14 Data for South Africa are extracted from the 1996 Population Census Report
(http://www.statssa.gov.za/census96). Data for other countries are from Ethnologue
(Grimes 1996).
15 It is often claimed that Zulu functions as a lingua franca for 70 per cent of South
Africa’s population (Government Gazette, vol. 407, no. 20098, 28 May 1999), but
the empirical basis for this claim is uncertain. However, it is important to stress that
the Population Census numbers reflect home-language status only, not patterns of
language use or knowledge.
76 R. K. Herbert and R. Bailey
bibliography
Alexander, Neville 1989. Language Policy and National Unity in South Africa/Azania.
Cape Town: Buchu Books.
Arbousset, Thomas 1991 [1840]. Missionary Excursion into the Blue Mountains, ed.
and trans. D. Ambrose and A. Brutsch. Morija, Lesotho: Morija Archives.
Bailey, R. 1995a. ‘The Bantu languages of South Africa: towards a sociohistorical per-
spective’. In R. Mesthrie (ed.), Language and Social History: Studies in South
African Sociolinguistics. Cape Town: David Philip, pp. 19–38.
1995b. ‘Sociolinguistic evidence of Nguni, Sotho, Tsonga and Venda origins’. In
R. Mesthrie (ed.), Language and Social History: Studies in South African Socio-
linguistics. Cape Town: David Philip, pp. 39–50.
Baumbach, E. J. M. 1987. Analytical Tsonga Grammar. Pretoria: University of South
Africa.
Bennett, Patrick and Jan P. Sterk 1977. ‘South Central Niger-Congo: a reclassification’.
Studies in African Linguistics, 8: 241–73.
Bleek, W. H. I. 1862. A Comparative Grammar of South African Languages. Cape Town:
J. C. Juta & Trübner.
Blench, Roger 1997. ‘Language studies in Africa’. In J. O. Vogel (ed.), Encyclopedia of
Precolonial Africa. Walnut Creek, Calif.: AltaMira Press, pp. 90–100.
Dahl, Otto C. 1954. ‘Le substrar bantou en malgache’. Norsk Tidsskrift for
Sprogvidenskap, 17: 325–62.
Doke, Clement M. 1931. Report on the Unification of the Shona Dialects. Government
Blue Book, Government of Southern Rhodesia.
1967. The Southern Bantu Languages. London: International African Institute.
1993 [1960]. ‘The growth of comparative Bantu philology’. In R. K. Herbert
(ed.), Foundations in Southern African Linguistics. Johannesburg: Witwatersrand
University Press, pp. 71–96.
Donnelly, Simon 1999. ‘Southern Tekela Nguni is alive: reintroducing the Phuthi lan-
guage’. International Journal of the Sociology of Language, 136: 97–120.
Ehret, Christopher 1997. ‘African languages: a historical survey’. In J. O. Vogel
(ed.), Encyclopedia of Precolonial Africa. Walnut Creek, Calif.: AltaMira Press,
pp. 159–66.
1999. ‘Subclassifying Bantu: the evidence of stem morpheme innovations’. In
J. Hombert and L. M. Hyman (eds.), Bantu Historical Linguistics: Theoretical
and Empirical Perspectives. Stanford: CSLI Publications, pp. 43–147.
Fortune, George 1993. ‘The contribution of C. M. Doke to written Shona’. African
Studies, 52: 103–29.
Greenberg, Joseph H. 1963. The Languages of Africa. Bloomington: Indiana University.
Gregersen, Edgar A. 1977. Language in Africa: An Introductory Survey. New York:
Gordon & Breach.
Grimes, Barbara F. 1996 (ed.). Ethnologue, 13th edn. Dallas: Summer Institute of
Linguistics.
Guthrie, Malcolm 1948. The Classification of the Bantu Languages. London: Oxford
University Press.
1967–71. Comparative Bantu: An Introduction to the Comparative Linguistics and
Prehistory of the Bantu Languages, 4 vols. Farnborough: Gregg.
Harries, Patrick 1988. ‘The roots of ethnicity: discourse and the politics of language
construction in South-East Africa’. African Affairs, 87: 25–52.
The Bantu languages: sociohistorical perspectives 77
Oliver, Roland 1966. ‘The problem of the Bantu expansion’. Journal of African History,
7: 361–76.
Phillipson, David W. 1977. ‘The chronology of the Iron Age in Bantu Africa’. Journal
of African History, 16: 321–42.
1985. African Archaeology. Cambridge: Cambridge University Press.
Piron, Pascale 1998. Classification interne du groupe bantoı̈de, 3 vols. Munich:
LINCOM Europa.
Poulos, George 1990. A Linguistic Analysis of Venda. Pretoria: Via Afrika.
Schapera, Isaac and D. F. van der Merwe 1943. ‘A Comparative Study of Kgalagadi,
Kwena and other Sotho Dialects’. Communication No. 9, School of African Studies,
University of Cape Town.
Silverstein, R. O. 1993 [1968]. ‘A note on the term “Bantu” as first used by W. H. I.
Bleek’. In R. K. Herbert (ed.), Foundations in Southern African Linguistics.
Johannesburg: Witwatersrand University Press, pp. 17–18.
Vansina, Jan 1979. ‘Bantu in the crystal ball. Part 1’. History in Africa, 7: 287–333.
1980. ‘Bantu in the crystal ball. Part 2’. History in Africa, 8: 293–325.
1995. ‘New linguistic evidence and “the Bantu expansion”’. Journal of African
History, 36: 173–95.
van Warmelo, N. J. 1927. ‘Die Gleiderung der südafrikanischen Bantusprachen’.
Doctoral dissertation, Hamburg. (Published in Zeitschrift für Eingeborenen-
sprachen, 18, 1: 1–54 and 18, 2: 81–127.)
1989. Venda Dictionary: Tshivenda–English. Pretoria: J. L. van Schaik.
1974. ‘The classification of the cultural groups’. In W. D. Hammond-Tooke (ed.), The
Bantu-speaking People of Southern Africa, 2nd edn. London: Routledge & Kegan
Paul, pp. 56–85.
Vogel, Joseph O. 1997. ‘Bantu expansion’. In J. O. Vogel (ed.), Encyclopedia of
Precolonial Africa. Walnut Creek, Calif.: AltaMira Press, pp. 435–8.
Watters, John R. 1989. ‘Bantoid overview’. In J. T. Bendor-Samuel (ed.), The Niger-
Congo Languages. New York: University Press of America, pp. 401–20.
Werner, Alice 1933. Myths and Legends of the Bantu. London: G. G. Harrap.
Wilkes, A. 1999. ‘Language contact and language change – the case of Southern
Transvaal Ndebele, a Nguni language spoken in the Republic of South Africa’.
In P. F. A. Kotey (ed.), New Dimensions in African Linguistics and Languages.
Trenton, N. J.: Africa World Press, pp. 243–51.
Williamson, Kay 1971. ‘The Benue-Congo languages and Ijo’. In T. A. Sebeok (ed.),
Current Trends in Linguistics, vol. IV: Sub-Saharan Africa. The Hague: Mouton,
pp. 245–306.
Williamson, Kay and Roger Blench 2000. ‘Niger-Congo’. In B. Heine and D. Nurse
(eds.), African Languages: An Introduction. Cambridge: Cambridge University-
Press, pp. 11–42.
Ziervogel, Derek 1959. A Grammar of Northern Transvaal Ndebele. Pretoria: van
Schaik.
4 Afrikaans: considering origins
Paul T. Roberge
1 INTRODUCTION
The three groups primarily responsible for the formation of Afrikaans –
European settlers (from 1652), the indigenous Khoekhoe and enslaved peoples
of African and Asian provenance (from 1658) – were quite distinct during
the first decades of the Cape Colony. This distinctness was defined by physi-
cal appearance, culture, religion and language. By the end of the Dutch East
India Company (VOC) era in 1795, a number of processes had eroded these
boundaries, inter alia: the incorporation of the Khoekhoe into the European-
dominated society as wage-labourers subject to Dutch law; conversion of slaves
and free blacks to Christianity or Islam; and miscegenation and intermarriage
among groups (cf. Elphick and Shell 1989: 184). Descendants of these groups
had further come to share in a common vernacular that was unique to southern
Africa.
During the VOC era (1652–1795), the language of European settlers in southern
Africa reflected not the emerging standard Dutch of the metropole but rather the
popular and regional varieties of the rank and file. Kloeke (1950) concluded that
the Netherlandic base of Afrikaans must lie in the southern part of the modern
province of South Holland. Scholtz (1963: 232–56) acknowledged Hollandic
affinities, even though he disputed the idea that the metropolitan base could
be located in one specific region in Holland. There is reason to believe that
Afrikaans has historical links to an inchoate koine that formed in Amsterdam
and other urban centres in Holland during the seventeenth century due to internal
immigration and an influx of refugees from Germany and French-speaking
regions. Because the cities were not able to absorb all the immigrants into the
mainstream economy, these groups must have been well represented in Dutch
colonial populations (cf. Ponelis 1993: 122, 127–9; Buccini 1996). We know, of
course, that the Dutch-speaking population at the Cape of Good Hope reflected
a variety of dialectal backgrounds (Utrecht, Zealand, Brabant, Flanders, the
79
80 P. T. Roberge
eastern provinces of the Netherlands). The link between Hollands and Afrikaans
may be attributable to a strong ‘founder effect’ exerted by the Dutch outpost’s
first commander – Jan van Riebeeck (1619–77) – and his entourage, as Kloeke
thought. Alternatively, the link may reflect an inchoate seventeenth-century
koine that had formed in the cities of Holland (especially Amsterdam) among
speakers who were constitutive of the founder population of the Dutch colony
in southern Africa.
The Cape Colony included significant numbers of Europeans to whom Dutch
was not native, namely speakers of Low German dialects (which constitute a
segment of the dialect continuum that stretches from the Netherlands through
northern Germany), High German dialects and French, with the arrival of
Huguenot refugees at the Cape in 1685.
By 1800, there were few Khoesan in the colony who were not in the service
of the Europeans as farm and domestic labour. From 1775, the offspring of
female Khoekhoe and male slaves – known as Basters or Bastaard hottentots –
were legally indentured until the age of twenty-five. The inboek system was
later construed by farmers to apply to all Khoesan children (Elphick and
Malherbe 1989: 32). Along the northern frontier, the class of Cape Dutch-
speaking Khoekhoe who had been in service came to be known as Oorlams; one
such group pushed into present-day Namibia at the beginning of the nineteenth
century. The Basters were of mixed European, Khoekhoe and slave parentage.
From this class there emerged in the early nineteenth century a series of Cape
Dutch-speaking communities along and to the north of the Orange river, known
collectively as Griqua. In the mid-nineteenth century a group of Basters settled
in Rehoboth in Namibia.
Only in the urban milieu of Cape Town was there sufficient opportunity for
the emergence of a slave ‘community’, abetted especially by the growth of
Islam.
The slave population also increased naturally by procreation. From the 1760s,
the percentage of the Cape slave population that was locally born was at or
near 50 per cent (Shell 1994b: 16–17). The children of liaisons between slave
women and European or Khoekhoe men were de iure slaves (Elphick and Shell
1989: 202). By 1834, when the institution was abolished at the Cape, as in
other British colonies, the slave population had risen to 36,169 (Armstrong and
Worden 1989: 109).
The ethnic diversity of the Cape slaves meant linguistic diversity as well.
While some slaves were proficient at several European and/or Asian languages,
others brought only their own languages, which were of little utility for com-
munication among themselves and with their masters and indigenous South
Africans. Two lingue franche gained currency: some slaves used a variety of
Creole Portuguese, which persisted throughout the VOC era (cf. Valkhoff 1966:
146–91; den Besten 1997), and/or a more-or-less koineised South African
variety of Malay (cf. den Besten 2000). But most new arrivals used jargonised
forms of Dutch to converse with their masters, with indigenes and with one
another.
Gender was an important sociolinguistic variable at the old Cape. Edith Raidt
(1994: 175–257, 1995) has examined a corpus of fifty-seven Cape Dutch texts
dating from between 1710 and 1805 either written by women or containing
speech attributed to them. She attempts to reconstruct the social networks
of each of these women, to the extent possible, with a view toward reveal-
ing patterns of variation. Women are shown to have been on the vanguard of
change in the direction of Afrikaans and at the same time conservative in their
preservation of Dutch dialectisms. However, upper-class women seem to have
been instrumental in the preservation of Dutch at the Cape. In 1751, Hendrik
Swellengrebel (1700–60) retired as governor of the Cape of Good Hope and
returned to Holland with his children. His eldest daughters, Helena Johanna
(1730–53) and Johanna Engela (1733–98) – both born at the Cape – kept a jour-
nal during their voyage from Cape Town to the Dutch Republic that is written
in a very good Dutch (Barend-van Haeften 1996). The pietistic diarist Susanna
Catharina Smit (1799–1863), born in Uitenhage, put her religious experience
to paper between 1843 and 1851 in a less elegant but still quite passable Dutch
(Puddu 1996).
Afrikaans: considering origins 83
late eighteenth century, and subsequently in the former Orange Free State and
Transvaal. Standard Afrikaans developed between roughly 1900 and 1930, and
is drawn mainly from the eastern frontier variety, with adlexification from Dutch
in learned vocabulary.
‘If we go back in time, the problem of what Afrikaans is becomes more and
more difficult’, wrote Valkhoff (1972: 2), and notwithstanding a far better
understanding of the material facts, his words remain true today. Exactly how
extra-territorial Dutch was transformed into Afrikaans has been a warmly dis-
puted question for more than a century. Currently, there are three basic positions
on the formation of Afrikaans, with varying degrees of overlap and difference
in emphasis. They are not necessarily incompatible (cf. Kloss 1978: 151); and
it is important to bear in mind that the questions asked are often not the same.
European Dutch dialect or dialect group’ (Thomason and Kaufman 1988: 255).
The superstratist hypothesis asks us to believe that any number of non-standard
dialects in the Low Countries contributed rules and features to Afrikaans (the
cafeteria principle in its original sense). The Flemish dialectologist J. L. Pauwels
(1958, 1959) claimed that the Brabantine dialect of Aarschot preserves the
model for the etymologically opaque double negation (Sy het nie gesê dat sy
môre gaan wen nie, ‘She didn’t say (that) she is going to win tomorrow’)
and the neologistic demonstrative pronouns hierdie, ‘this’, daardie, ‘that’ (lit.
‘here, there + that’) replacing Dutch deze/dit, ‘this’, die/dat, ‘that’; in a sim-
ilar vein on the demonstratives, see Raidt (1994: 161–74). Such claims are
undermined by the fact that both features are not attested until quite late in
our Cape Dutch source material, and that Afrikaans does not otherwise show
strong affinities to southern Netherlandic dialects (i.e. Brabants, Flemish and
Zeeuws). In fact, Buccini (1996) has shown that the European base of Cape
Dutch/Afrikaans shares essentially the same demographic and dialectal pro-
file as New Netherland Dutch in North America, where Afrikaans-like double
negation and demonstratives (among other features) are entirely unknown.
According to Raidt (1978: 119, 1983: 24–8 and 191, 1991: 124–31, 176–7),
native-language (L1) ‘interference’ and imperfect approximation of Dutch re-
sulted in ‘broken language’ on the part of the large number of non-native
speakers – European as well as non-European – in a multilingual society –
but not outright pidginisation, much less creolisation (similarly, Pheiffer 1980:
1–11; Conradie 1998). Speech ‘errors’ that were initially random and unsystem-
atic eventually coupled with parallel internal changes in progress (most notably
the deflective tendency). Yet it is hardly likely that German- and French-based
inter-language varieties played a critical role in the restructuring of Dutch at the
Cape. As Buccini (1996) points out, the European population of New Netherland
was no less heterogeneous, yet New Netherland Dutch shows nowhere near the
same degree of deflection and restructuring as Afrikaans. The different linguis-
tic outcomes are surely due to radically different sociolinguistic conditions in
the two former Dutch colonies.
To be sure, superstratists do acknowledge some limited substrate influence.
Raidt derives reduplication in Afrikaans (staan-staan, ‘stand-stand’, i.e.
‘standing’) from Malay (1983: 169–72, 1991: 225–6, 1994: 148–60) and the
object marker vir (Hulle ken vir ons baie goed ‘They know us very well’) from
Creole Portuguese (1983: 183–7, 1991: 226–7, 1994: 116–47). (Further to these
features see now, respectively, den Besten et al. forthcoming; den Besten 2000.)
1983: 138–9, 1985, 1989). Competition among, and selection of, linguistic
variants imported from the metropole is an ongoing process in the history of
Afrikaans. To cite but one example: Afrikaans reflexive pronouns are identical
to the oblique forms of the corresponding personal pronouns: Hy was hom, ‘He
washes himself’; Hulle besin hulle, ‘They change their minds’, beside Dutch
hij wast zich, zij bezinnen zich. Non-standard varieties of Dutch frequently use
the oblique forms of the third-person pronouns as reflexives; Standard Dutch
zich is borrowed from German and did not become common in Holland until
the sixteenth century. Both variants are in competitive alternation in our Cape
Dutch source material to the end of the eighteenth century, by which time the
latter had become recessive. To the extent that traditional diachronic formu-
lations are translatable into terms of synchronic variation and selection, the
variationist component of this hypothesis qualifies mainly as a shift in perspec-
tive. As such, the problems associated with the superstratist hypothesis may
present themselves as before in less clear-cut cases.
Combrink (1978: 72–77) explains the demise of personal agreement in the
Afrikaans verb as the linguistic consequence of mixing between similar but non-
identical inflectional systems in the Netherlandic and Low German dialects
represented at the Cape. Because the exigencies of efficient communication
implied greater reliance on syntax and lexical roots, verbal inflections became
completely redundant and thus dispensable. In this way Cape Dutch could be
morphologically stripped even while retaining its Continental West Germanic
syntactic typology. It is true that koineisation can produce inflectional sim-
plification while leaving intact more complex grammatical systems (Holm
1988: 10). But these same exigencies of perceptibility and ease of decoding also
underlie the loss of inflectional morphology during pidginisation and creolisa-
tion. While verbal inflection was reduced in New Netherland Dutch (especially
in the last stages), this reduction was largely phonologically motivated; ‘the
principle remained until the end’ (Buccini 1992).
According to the interlectalist component of the hypothesis under discussion,
language shift within the Afro-Asian substratum was preceded by spontaneous,
untutored approximations of Dutch (imperfect code-switching) on the part of
adult language learners in the early years of the colony, with succeeding gener-
ations acquiring Cape Dutch natively (as bilinguals for an indeterminate period
of time). The contemporary Afrikaans of people of colour still bears the im-
print of the interlanguages of their forebears, even though there was, according
to this view, no pidgin or creole ancestor in the conventional sense of these
terms (cf. van Rensburg 1985: 138–54, 1989: 137–8; Kotzé 1989; Webb 1993).
Ponelis (1988, 1993: 27–30, 1994) provides the most coherent articulation of
the interlectalist position: (1) The Cape Colony was a heterogeneous, multilin-
gual society in which Dutch was a minority first language in the early years and
was approximated in a haphazard, untutored way on account of its extensive use
Afrikaans: considering origins 87
that Afrikaans evolved from ‘broken’ forms of Dutch that emerged during the
first fifty years of Dutch occupation as the vernacular of slaves, Khoekhoe and
their descendants of mixed race. It was during this time also that the speech of
European children came under the influence of these varieties (1953: 26, 95,
202–3). Thus Franken followed Hesseling in stressing contact with people of
colour, even while de-emphasising somewhat the latter’s construct of a mixed
‘Malayo-Portuguese’ lingua franca (cf. 1953: 43). Hesseling’s thesis was revis-
ited by Marius Valkhoff, according to whom Cape society from the second half
of the seventeenth century and still in the first half of the eighteenth century was
so much integrated that there was a very close intercourse between Europeans,
indigenes and slaves. Valkhoff assumed the emergence of a ‘proto-Afrikaans’
among the latter groups during the first fifty years of Dutch occupation (1966:
204–7, 1972: 48–9). During these ‘linguistic encounters’ Creole Portuguese
provided the flux in the semi-creolisation of Dutch, though Malay gradually
overtook it as a lingua franca in the East and slave language in southern Africa
during the eighteenth century and left its mark as well (1972: 72, 83).
Nowadays, the creolist hypothesis is perhaps most closely identified with the
research programme of Hans den Besten (e.g. 1978, 1986: 224 et passim, 1987,
1988, 1989, 1997). From as early as 1590, local Khoekhoe used jargonised
forms of Dutch and English in their contacts with Europeans who called at
the Cape of Good Hope. The period 1652–8 saw the emergence of a ‘fort’
situation (1989: 226–7); that is, a situation in which Europeans established a
permanent outpost on the shores of continents and developed complex rela-
tions with indigenes. According to den Besten (1989: 219–20), the Khoekhoe
‘could develop a pidgin of their own without interference of other groups of
non-native speakers of Dutch. The resulting pidgin was called Hottentots-
Hollands or Hottentot-Dutch.’ From 1658, the slaves (re)pidginised Dutch
in their encounters with the Khoekhoe and Europeans, and contributed their
own modifications. Creolisation occurred in the western Cape around 1700
following the withdrawal of Khoekhoe into the interior to escape European
domination and in the aftermath of the 1713 smallpox epidemic. The Cape
Dutch pidgin(s) may have become a native language for at least some speakers
among locally born slaves and the mixed offspring of Khoekhoe who remained
in the western Cape. The Khoekhoe who withdrew from the western Cape,
however,
took their pidgin (creole?) with them, and probably influenced the other Hottentots in
the north and in the east, so that those Cape farmers who – from about 1700 onward –
started to colonize the future eastern districts of the Cape Colony could again meet with
Khoekhoen who spoke some kind of Dutch. Things were different in the north, in the
Orange River area, since whites appeared there relatively recently, i.e. in the 19th century.
(Den Besten 1989: 224)
Afrikaans: considering origins 89
Modern Cape and Orange River Afrikaans are traceable to the pidgin and creole
Dutch formerly spoken widely by people of colour in these regions. Euro-Cape
Dutch exerted a ‘decreolising influence’ on these varieties, although decreolisa-
tion was ‘counterbalanced by “creolizing” influences exerted upon Cape Dutch
by the Dutch Creole (or Creoles)’ (1989: 225). By about 1850, one may speak
of an ‘Afrikaans koine with dialectal differentiation’ (1989: 226).
As a class, fort creoles typically differ less radically from their lexifier lan-
guages than do plantation creoles. That Afrikaans has remained linguistically
much closer to Dutch than the Caribbean Dutch creoles is in den Besten’s view
(1989: 227) attributable to three factors: (1) The population of the Cape Colony
comprised a high percentage of Europeans. (2) The Cape Dutch pidgin/creole
was a second or third language, for many slaves could avail themselves of creole
Portuguese and/or Malay. The availability of these lingue franche limited
somewhat the importance of Cape Dutch pidgin for inter-ethnic communica-
tion. (3) The legally free Khoekhoe were in a better position than the slaves to
improve their performance in the direction of the superstrate by virtue of their
greater access to that language.
Ponelis challenges den Besten’s assertion that ‘Hottentot Dutch’ supplied
the foundation for subsequent developments: ‘[Den Besten] considers no so-
ciohistorical evidence . . . [and] his position is based entirely on shaky linguistic
evidence’ (1993: 33–4). Although the latter assessment seems unduly harsh,
it is improbable that ‘Hottentot Dutch’ could have developed beyond unstable
and highly variable jargons and interlanguages into what one could reasonably
construe as a pidgin sensu stricto, that is, a code that is characterised by social
norms and some measure of grammatical fixity.
The length of time between the beginnings of immigration and what Baker
(e.g. Baker 1993: 137–8) calls ‘Event 1’ – the point at which the slave popula-
tion surpasses the slave-owning European population – is crucial. The longer
this period, the greater the exposure of newly arrived slaves to the superstrate
language. In the Cape Colony, the pre-Event 1 period was roughly fifty-two
years, that is, 1658–1710 (cf. Armstrong and Worden 1989: 121), and thus of
sufficient length to produce a linguistic variety much closer to the superstrate
language than in other slave societies (cf. Corne’s contribution on Réunionnais
in Baker and Corne 1982).
Subsequent to Event 1, the rate of dilution of the superstrate language is
determined by the rate of increase in the slave population. In the Cape Colony,
the slave population never greatly exceeded the settler population. Moreover,
there was no subsequent formation of a plantation society. The Cape was poorly
suited to plantation agriculture, and there were no large slave holders save for the
VOC itself and a few of the bigger farms in the western Cape. The demographic
event corresponding to Baker’s Event 2 – when the number of locally born slaves
90 P. T. Roberge
surpasses the slave-owning population – did not take place in southern Africa.
These facts suggest that L2 acquisition on the part of subsequent arrivals could
be more directly targeted toward the language of Europeans than in other slave-
labour systems in which creole languages have formed – as den Besten and
others (above) have rightly stressed.
At the same time, however, the scenario above makes implausible the for-
mation of a Cape Dutch creole in the conventional sense of the term. A Cape
Dutch pidgin cannot have been nativised in the sense that it provided the pri-
mary input for L1 acquisition; there was, after all, no actual withdrawal of the
superstrate language. Furthermore, we find no evidence whatsoever to substan-
tiate the view that decreolisation (loss of basilect) represents a developmental
stage in the history of Afrikaans.
Acrolectal Cape Dutch is the variety closest to the metropolitan language at the
end of the VOC era. It is preserved in the diary fragment of Johanna Duminy
(1797, published in Franken 1938). Limitations of space do not permit more
than a cursory overview of important linguistic affinities and divergences.
Although gender in the noun had virtually disappeared by 1797 (de huijs,
Dutch het huis, Afrikaans die huis, ‘the house’), acrolectal Cape Dutch does not
92 P. T. Roberge
show the same degree of verbal deflection as Afrikaans, which, as a general rule,
has done away with inflectional oppositions entirely (Dutch werken, ‘work’: ik
werk, ‘I work’; jij werkt, ‘you (sg.) work’; wij werken, ‘we work’, etc. beside
Afrikaans werk: ek, jy, ons werk).
Duminy consistently distinguishes between finite and non-finite forms of
the verb. As concerns personal agreement, the direction of change in acrolec-
tal Cape Dutch seems to be towards invariant finite inflection, the result of
which would be a simple binary opposition between finite and non-finite forms
of the verb. It is the singular (the exponents of which could be either zero
or -t) that has encroached on the plural termination -e(n) and not vice versa. In
the plural Duminy’s usage vacillates between inflected and endingless forms
(wij sliep beside wij sliepe, Dutch wij sliepen, ‘we slept’); one finds neither
deflected infinitives (sij liet haar wage inspannen/*inspan, ‘She had her wagon
inspanned’) nor intrusions of plural -e(n) into the singular (Ik gaf, ‘I gave’,
but not *ik gave(n)). Final cluster reduction is evident in the Duminy diary
(direk: Dutch direct, Afrikaans direk), and one would think that it brought ad-
ditional pressure to bear on second- and third-person singular verb forms in -t
and on the weak past participle (gewerk for gewerkt). However, cluster reduc-
tion may not have been as general in acrolectal Cape Dutch as it is today in
Afrikaans. Several idiosyncrasies of Duminy’s usage are hardly consonant with
the assumption of a fully diffused rule, namely, the presence of ahistorical -t
in the present-tense first-person singular (ik komt, Dutch ik kom, ‘I come’)
and plural (wij komt, Dutch wij komen), and in the strong preterite ik gaft
(for ik gaf ).
Acrolectal Cape Dutch preserved the pan-Germanic distinction between
‘strong’ (ablauting) and ‘weak’ (dental suffixal) inflection in preterital con-
jugation (laten/liet, ‘let’; bestellen/bestelde, ‘order’) and in the past participle
(krijgen/gekreegen, ‘get’; opbrengen/opgebragt, ‘bring up’). It also maintained
both hebben, ‘have’, and zijn, ‘be’, as auxiliary verbs in periphrastic tense
formation:
(1) a. hij see niet minder als die ander man heeft gekreegen
‘He said not less than that other man got’
b. ik bin buyte geweest
‘I was outside’
c. ik hat ook een groote caatel gekogt
‘I had also bought a large “caatel” ’
d. sij ware de voorige dagt al na de vandiesie gereeden
‘they had already driven off to the sale the previous day’
Afrikaans has retained only ‘have’ as the tense auxiliary. It has eliminated the
preterite and pluperfect tense (1c, 1d) altogether and has regularised the past
participle (kry/gekry, bring/gebring).
Afrikaans: considering origins 93
The Duminy diary is no less important for the hallmark Afrikaans features that
it does not show: the double negation (nie . . . nie), the demonstrative pronouns
hierdie, ‘this’/daardie, ‘that’, reduplication, subjectival ons for wij, ‘we’, etc.
In the Cape Dutch pidgin a preverbal particle ge, together with a phonological
variant ga (thus, gesien/gasien), marked events situated in the past. The fact that
Afrikaans developed in a multilingual contact situation raises the possibility of
multilevel syncretism, in which phonological, syntactic and semantic properties
of morphemes can be traced to multiple sources. The use of ge/ga as a past-
tense marker closely corresponds to the Dutch past particle prefix ge-. There
is also evidence to suggest that Khoekhoe preverbal preterital particles with a
similar canonical shape may have reinforced the observed usage; Nama gye,
go (Kroenlein 1889: 101, 106); kò (recent past), kè (remote past) (Hagman
1977: 62).
94 P. T. Roberge
As concerns modality, the Cape Dutch pidgin employed maskie, ‘never mind,
perhaps, (even) if’ (Creole Portuguese maski, Portuguese mas que) to indi-
cate that the action or state of the predicate is uncertain and has not (yet)
become part of reality. Consider the utterances attributed to Khoekhoe (3a) and
slave (3b–c) speakers:
(3) a. Duytsman altyt kallom: ‘Icke Hottentots doot makom: Mashy doot,
Icke strack nae onse grote Kapiteyn toe.’ (ten Rhyne 1686, in Schapera and
Farrington 1933: 140)
‘Dutchmen always say: I Hottentots dead make [I will kill Hottentots]: never
mind dead [i.e. if I die], I soon [go] to our great chief.’
b. Maski ik wil dat bloed ook wel drinken, dan word ik sterk. (1707, cited from
Franken 1953: 48)
‘Maybe I (will) also want to drink that blood, then I become strong.’
c. Seijde dien slaaf teegens hem: ‘Maskij jouw, komt maar hier.’ (CJ 344 1739:
371)
‘The slave said to him: “Never mind you, just come here!”’
The morpheme kam(m)e is attested several times in our Cape Dutch pidgin
corpora, being attributed to Khoekhoe (4) and slaves (5).
(4) a. kamme niet verstaan (Kolbe 1727: I, 504)
‘truly do (will/would) not understand’
b. Ey Vrouw die Tovergoeds ja zoo bytum, ons ik kame niet verdragen. (Kolbe
1727: I, 528)
‘Oh, woman, this/that medicine stings so, we I (?) truly do [shall] not endure
it.’
c. Vrouw, jou Tovergoeds bra bytum, dat is waar, maar jou Tovergoeds ook weer
gezond makum, dat is ook waar. Ons Tovermanns kame niet helpen, maar die
Duits Tovervrouw ja bra, die kame helpe. (Kolbe 1727: I, 528)
‘Woman, your medicines sting very much, that is true, but your medicines
also make healthy again, that is also true. Our medicine men truly do/will not
help, but the Dutch medicine woman [is] indeed good, she truly helps (will
help).’
(5) a. kammene Kumi, Kammene Kuli (Mentzel 1944 [1787]: III, 99)
truly-not food, truly-not work
‘If I have nothing to eat, I do not work.’
b. Kammene Kas, Kammene Kunte (Mentzel 1944 [1787]: III, 99)
truly-not money, truly-not cunt
‘If you have no money, I shall give you no sex.’
Den Besten resolves kam(m)e as kan, ‘can’, plus a verbal marker -me. Although
the contexts in (4)–(5) do allow interpretation as ‘can’ (indeed, kame in (4c)
might be better parsed as kan me [ mij], ‘can me’), the obscurity of the second ele-
ment militates against this analysis. I make so bold as to etymologise kam(m)e
as the root contained in the Khoekhoe form that Sparrman recorded as kammasa
and glossed as ‘truth, it is true’ (1977 [1786]: II, 265). Cognates are to be found
Afrikaans: considering origins 95
in Nama ama-b, ‘truth’; ama-se, ‘truly’ (Nienaber 1963: 519), Kora kx’ama,
‘true’; kx’ama-b, ‘truth’, Gri k’ama-se, ‘truly’ (Meinhof 1930: 143, 148). The
basic meaning ‘truly’ seems to have been preserved in the utterances in (4)–(5).
But in the latter data set, the semantic range of kam(m)e subsumes future time
reference, prediction, and even counterfactuality. It is possible that syncretism
occurred in the Cape Dutch pidgin between kamma(sa) and what Lichtenstein
(1930 [1815]: II, 473) recorded as Khoekhoe t’2 kamüh, ‘lie’ (//kamüh or
=| kamüh, according to Nienaber 1963: 373; cf. Nama =| hòmi, ‘lie’).
Completion of an action in the Cape Dutch pidgin appears to have been
expressed by (al ) gedaan (lit. ‘(already) done, finished’) within the middle
field before the main verb or adjective (gedaan being the past participle of the
Dutch verb doen, ‘do’). Den Besten (1987: 19–20, 22, 1989: 238) cites (6)a–c
in support of this reconstruction:
(6) a. Ons soek kost hier, ons al gedaen wegloopen . . . (slave, 1706, cited from
Franken 1953: 89)
‘We seek food here, we have run away.’
b. de Clercq heeft gesegt jij mijn Cameraat gedaan vast maken . . . (slave, 1720,
cited from Franken 1953: 50)
‘De Klerk said you have tied up my comrade.’
c. Die Gift al gedaan dood, wie kan hy meer wat schaden (Khoekhoe speaker,
cited from Kolbe 1727: II, 114).
‘This/that poison has died, whom can it harm any more even a little?’
active – knowledge of the Cape Dutch pidgin. Acrolectal Cape Dutch would
have become more and more diluted with increasing social and geographic dis-
tance from centres of power. The extent of dilution would naturally be greatest
in the rural areas along the frontier (where speakers were simply not as familiar
with prestige norms), within the slave community generally by virtue of multiple
inputs, and among the Khoekhoe and Basters. In other words, the Dutch lan-
guage at the Cape of Good Hope formed a continuum from the most basilectal
varieties within the Afro-Asian substrate to the extra-territorial Dutch of the
European superstrate. The speech of individuals took on or avoided pidgin fea-
tures depending on the interlocutor, the nature of their communicative networks
and the sociolinguistic circumstances (code-switching: cf. Roberge 1994b). L1
acquisition of acrolectal Cape Dutch simultaneously with the Cape Dutch pidgin
resulted in a number of functional convergences.
Convergent deflective tendencies having different origins, motivations and
degrees of intensity are presumed to have triggered the stripping of verbal
morphology that is a defining feature of Afrikaans today.
Convergence between acrolectal Cape Dutch and the pidgin has left indirect
reflexes or residue in Afrikaans. During this process the tense auxiliaries hebben,
‘have’, and zijn, ‘be’, were reintroduced. Yet not all inflectional categories were
fully restored even though their exponents managed to survive. The result is a
residue of allomorphs for ‘have’ that are used more or less interchangeably in
Orange River Afrikaans: het (< Dutch heeft), had (< Dutch had ):
(7) Die goue pondtjie baas? Ja, ek had hulle gakjen baas. Ek het die tiensielings
ook gakjen. (Van Rensburg 1984: II, 275)
‘The gold pound, sir? Yes, I was familiar with it, sir.
I was familiar with the ten shilling piece, too.’
The same holds true for vestiges of ‘be’, which is also to be discerned in Orange
River Afrikaans:
(8) a. Die boere was baie laat hier gakom. (Van Rensburg 1984: II, 40)
‘The boers came here very late.’
b. want ek is maar op hom [die plaas] grootgeword (van Rensburg 1984: II,
219)
‘because I grew up on [this farm]’
Standard Afrikaans, by contrast, uniformly uses het as the sole tense auxiliary.
(See Roberge 1994a: 80–2 for details.)
A rather more direct legacy of the Cape Dutch pidgin is manifest in the
adverbialisation of kamma. Although we do find vestiges of the canonical mean-
ing ‘truly’ (9a), kamma in colloquial Afrikaans bestows a nuance of pretence
or ostensibility, as we see in (9)b:
Afrikaans: considering origins 97
Den Besten (1989: 238) points out that Dutch klaar, ‘ready, finished’, can
also mean ‘already’ in Afrikaans. In colloquial speech klaar often occurs in
combination with al (cf. Donaldson 1993: 206, 232).
(12) a. Brand die vuur al? Ja, ek het dit klaar (= al/alreeds) gemaak. (Donaldson
1993: 266n.)
‘Is the fire going? Yes, I’ve already got it going.’
b. Heule had klaar gedôd. (Rademeyer 1938: 76)
‘They have already died [i.e. they are all dead].’
c. Kietie sy was eintlik al klaar doot gewies, want sy was ‘n plaasmeisie.
(Small 1965)
‘Kitty, she was actually already dead because she was a farm girl [trying to
survive in the city].’
d. en lôp sê vir hulle . . . lat hulle hom regmak want hulle-t al klaar gebetaal
(van Rensburg 1984: II, 222).
‘and go tell them that they have made him right because they have already
paid’
The capacity of (al ) klaar to express completive aspect in Afrikaans almost cer-
tainly continues the pattern (al ) gedaan in the Cape Dutch pidgin (cf. den Besten
98 P. T. Roberge
11 CONVERGENT HYBRIDISATION
12 CONCLUSION
Four factors have been shown to be pivotal in the formation of Afrikaans:
(i) the emergence of an extra-territorial variety of Dutch in course of the
seventeenth and eighteenth centuries, namely acrolectal Cape Dutch; (ii) the
existence of a fully developed system in contact with developing systems such
that the degree of basilectalisation of colonial speech was far less drastic than
in creole communities where there was significant attrition of speakers of the
lexifier language; (iii) the development of a stable Cape Dutch pidgin for
inter-ethnic communication within the Afro-Asian substratum; (iv) linguistic
convergence between the various segments of colonial society, with attendant
hybridisation.
bibliography
Armstrong, J. C. and N. A. Worden 1989. ‘The slaves, 1652–1834’. In Elphick and
Giliomee (eds.), pp. 109–83.
Baker, P. 1993. ‘Assessing the African contribution to French-based creoles’. In Salikoko
S. Mufwene (ed.), Africanisms in Afro-American Language Varieties. Athens:
University of Georgia Press, pp. 123–55.
Baker, P. and C. Corne 1982. Isle de France Creole: Affinities and Origins. Ann Arbor:
Karoma.
Barend-van Haeften, M. L. 1996 (ed.). Op reis met de VOC: De openhartige dagboeken
van de zusters Lammens en Swellengrebel. Met medewerking van E. S. van Eyck
van Heslinga. Zutphen: Walburg Pers.
Bruyn, A. and T. Veenstra 1993. ‘The creolization of Dutch’. Journal of Pidgin and
Creole Languages, 8: 29–80.
Buccini, A. F. 1992. ‘The colonial Dutch dialects: new evidence for the origins of
Afrikaans and the development of standard Dutch in seventeenth century Holland’.
Paper presented to the Germanic Linguistics Roundtable, University of California
at Berkeley, April 10–11, 1992.
1996. ‘New Netherland Dutch, Cape Dutch, Afrikaans’. Taal en Tongval,
Themanummer 9: 35–51.
C J 344 1739. Crimineele Justitiën (reports of the Criminal Court), Government Archive,
Cape Town.
Combrink, J. G. H. 1978. ‘Afrikaans: its origin and development’. In L. W. Lanham
and K. P. Prinsloo (eds.), Language and Communication Studies in South Africa.
Oxford and Cape Town: Oxford University Press, pp. 69–95.
Conradie, C. J. 1998. ‘Preteritumverlies in vroeë Afrikaans’. Tydskrif vir Geestesweten-
skappe, 38: 6–20.
Davids, A. 1991. ‘The Afrikaans of the Cape Muslims from 1815 to 1915’. MA thesis,
University of Natal, Durban.
den Besten, H. 1978. ‘Cases of possible syntactic interference in the development of
Afrikaans’. In P. Muysken (ed.), Amsterdam Creole Studies II. Amsterdam: Instituut
voor Algemene Taalwetenschap, Universiteit van Amsterdam, pp. 5–56.
100 P. T. Roberge
1985. ‘Die doppelte Negation im Afrikaans und ihre Herkunft’. In Norbert Boretzky,
Werner Enninger and Thomas Stolz (eds.), Akten des l. Essener Kolloquiums über
‘Kreolsprachen und Sprachkontakte’ vom 26.1.1985 an der Universität Essen.
Bochum: Studienverlag Dr N. Brockmeyer, pp. 9–42.
1986. ‘Double negation and the genesis of Afrikaans’. In P. Muysken and
N. Smith (eds.), Substrata versus Universals in Creole Genesis. Amsterdam: John
Benjamins, pp. 185–230.
1987. ‘Die niederländischen Pidgins der alten Kapkolonie’. In N. Boretzky,
W. Enninger and T. Stolz (eds.), Beiträge zum 3. Essener Kolloquium über
Sprachwandel und seine bestimmenden Faktoren. Bochum: Studienverlag
Dr N. Brockmeyer, pp. 9–40.
1988. ‘Universal-Grammatik und/oder Zweitsprachenerwerb: Der Fall Afrikaans’.
In N. Boretzky, W. Enninger and T. Stolz (eds.), Beiträge zum 4. Essener Kollo-
quium über ‘Sprachkontakt, Sprachwandel, Sprachwechsel, Sprachtod’. Bochum:
Studienverlag Dr N. Brockmeyer, pp. 11–44.
1989. ‘From Khoekhoe foreignertalk via Hottentot Dutch to Afrikaans: the creation
of a novel grammar’. In Pütz and Dirven (eds.), pp. 207–49.
1997. ‘Kreolportugiesisch in Südafrika: Malaio- oder Indoportugiesisch?’ In R.
Degenhardt, T. Stolz and H. Ulferts (eds.), Afrolusitanistik – eine vergessene
Disziplin in Deutschland? Dokumentation des 2. Bremer Afro-Romania Kolloqui-
ums vom 27.–29. Juni 1996. Bremen: University of Bremen, pp. 317–51.
2000. ‘The slaves’ languages in the Dutch Cape Colony and Afrikaans vir’.
Linguistics, 38: 949–71.
den Besten, H., C. Luijks and P. T. Roberge forthcoming. ‘Reduplication in Afrikaans’. In
S. Kouwenberg (ed.), Twice as Meaningful: Reduplication in Pidgins and Creoles.
London: Battlebridge.
Deumert, A. 1999. ‘Variation and Standardisation: The Case of Afrikaans (1880–1922)’.
Ph.D. thesis, University of Cape Town.
Donaldson, B. C. 1993. A Grammar of Afrikaans. Berlin and New York: Mouton de
Gruyter.
Elphick, R. 1977. Kraal and Castle: Khoikhoi and the Founding of White South Africa.
New Haven and London: Yale University Press.
Elphick, R. and H. Giliomee 1989a. ‘The origins and entrenchment of European domi-
nance at the Cape, 1652–c. 1840’. In Elphick and Giliomee (eds.), pp. 521–66.
1989b (eds.). The Shaping of South African Society, 1652–1840, 2nd edn. Middletown,
Conn.: Wesleyan University Press.
Elphick, R. and V. C. Malherbe 1989. ‘The Khoisan to 1828’. In Elphick and Giliomee
(eds.), pp. 3–65.
Elphick, R. and R. Shell 1989. ‘Intergroup relations: Khoikhoi, settlers, slaves and free
blacks, 1652–1795’. In Elphick and Giliomee (eds.), pp. 184–239.
Franken, J. L. M. 1938 (ed.). Duminy-dagboeke. Cape Town: Van Riebeeck Society.
1953. Taalhistoriese bydraes. Amsterdam and Cape Town: Balkema.
Hagman, R. S. 1977. Nama Hottentot Grammar. Bloomington: Indiana University.
Hesseling, D. C. 1899. Het Afrikaansch. Leiden: E. J. Brill.
1923. Het Afrikaans, 2nd edn. Leiden: E. J. Brill.
Holm, J. 1988–9. Pidgins and Creoles, 2 vols. Cambridge: Cambridge University Press.
Afrikaans: considering origins 101
Valkhoff, M. F. 1966. Studies in Portuguese and Creole. With Special Reference to South
Africa. Johannesburg: Witwatersrand University Press.
1972. New Light on Afrikaans and ‘Malayo-Portuguese.’ Louvain: Editions Peeters
Imprimerie Orientaliste.
van Rensburg, M. C. J. 1983. ‘Nie-standaardvorme, variasiepatrone en Afrikaans uit die
vorige eeu’. In G. N. Claassen and M. C. J. van Rensburg (eds.), Taalverskeidenheid:
‘n blik op die spektrum van taalvariasie in Afrikaans. Pretoria, Cape Town and
Johannesburg: Academica, pp. 134–61.
1984 (ed.). Die Afrikaans van die Griekwas van die tagtigerjare, 2 vols. Bloemfontein:
University of the Orange Free State.
1985. ‘Taalverskeidenheid, taalversteuring en Afrikaans’. In H. J. Lubbe (ed.), Fokus
op die taalkunde. Acta Academica, Nuwe Reeks B. Bloemfontein: University of
the Orange Free State, pp. 123–66.
1987 (ed.). Finale verslag van die groter navorsingsprojek ‘Gesproke Afrikaans:
Die Afrikaans van die Richtersveld’. Bloemfontein: University of the Orange Free
State.
1989. ‘Orange River Afrikaans: a stage in the pidgin/creole cycle’. In Pütz and Dirven
(eds.), pp. 135–51.
van Selms, A. 1979 (ed.). Abu Bakr se ‘Uiteensetting van die godsdiens’. Amsterdam:
North Holland.
WAT = P. C. Schoonees et al. 1970– (eds.). Woordeboek van die Afrikaanse taal.
Stellenbosch: Buro van die WAT.
Webb, V. N. 1993. ‘Die herkoms van Oranjerivierafrikaans’. In Linguistica: Festschrift
E. B. van Wyk. Pretoria: J. L. van Schaik, pp. 161–71.
Worden, N. 1985. Slavery in Dutch South Africa. Cambridge: Cambridge University
Press.
5 South African English
Roger Lass
1 HISTORICAL BACKGROUND
1.1 Introduction
According to the 1996 census figures, English is the mother tongue of some 3.45
million people in South Africa. In terms of the old racial classifications, about
1.71 million of these are white, 0.58 million coloured, 0.97 million Indian and
0.11 million ‘African’. Broadly speaking, white, coloured and Indian English
in South Africa are distinct ‘ethnolects’. This fact, however unpalatable its
sociopolitical implications and however unsavoury its origins, is nevertheless
historically significant. English was brought to this country from England, and
was in its early days an instrument of English (= white by default) hegemony.
Because of the education system then (as now), and the contingencies of inter-
group relations, English must be seen primarily as a language that diffused
from white European (specifically British) mother-tongue speakers to other
communities.
The whole history, and the particular kinds of diffusion that occurred, have
an important bearing on the structural properties of all varieties of English
spoken in South Africa. Communities that shift from one language to another,
whatever they ultimately make of the language shifted to when it becomes a
mother tongue, are severely constrained by the properties of the input. To put
it crudely but usefully, if South Africa had been settled mainly by Scots, and
Scottish English had been the main input, and taught in the schools, all varieties
of South African English (SAE) would now pronounce postvocalic /r/ (in far,
mother), would not distinguish the vowels of foot and food, and would have
three distinct vowels in bird, heard and word. If on the other hand the main input
had been from West Yorkshire and Lancashire, SAE would not distinguish the
vowels of cut and put, but would distinguish won and one.1 In fact there was a
large settler input from both Scotland and the north of England (not to mention
Wales and Ireland); but virtually nothing has survived of this heritage except a
few words and usages.2
So it should not be construed as racist or insensitive to take white SAE as a
kind of reference point for all other varieties; this is simply a matter of history.
104
South African English 105
Indeed, as I will show below, all mother-tongue varieties spoken in South Africa
(and even second-language varieties such as Afrikaans English) are not only
autonomous dialects of English, but specifically dialects of Southern British
English (SBE), with a distinctly eastern rather than western cast.3
A comment on the sense of ‘dialect’ is necessary here. To say that SAE is
‘a dialect of SBE’ is not to say that in some way it ‘deviates’ from a modern
SBE norm; rather, all modern SBE varieties are what they are because they
share a common ancestor or set of closely related ancestors; and that because
of this (and in many cases because of subsequent contact with the descendants
of their own ancestors), they are clearly recognisable in a wider perspective
as southern. Thus a ‘dialect’ is a member of a cluster of (historically) related
varieties that normally share a common name (and whose speakers normally
consider themselves to speak varieties of ‘the same language’). In particular, I do
not use the term in the lay sense: ‘dialect’ as opposed to ‘standard’. Standards
are dialects too, in the technical usage.
This is important, because of the frequent use in South Africa of ‘SAE’
and ‘Standard English’ as opposed terms. This is historical and sociolinguistic
nonsense. Virtually every regional variety of English has its own sociolinguis-
tic continuum from ‘standard’ (= educated, non-stigmatised, favoured by the
schools, normal for public discourse) to vernacular (non-standard, stigmatised
by at least part of the standard-speaking community). As we shall see below,
SAE constitutes just such a continuum, though with the special complication
that there are in fact two standard types: one local and unmistakably South
African, and one (perhaps nostalgically and ignorantly) looking to an image of
the speech of southern England as its source of norms.
And so on. These categories should be self-explanatory. (For the most part
they reflect particular historical vowel phonemes, but this is not always relevant
here.)
The most important southern features are:
(1) [æ] (or a higher vowel) in trap. This is a seventeenth-century development
of older fully open [a], which is retained in the north and Midlands of
England and in Scotland, and to a large extent in Ireland. In the Southern
hemisphere Englishes (see the next section), and to some extent now in
London vernacular and certain posh (‘Morningside’) Scots varieties, it tends
to raise still higher, to [ε].
(2) strut/foot split. All southern and south-Midland English dialects (and
their descendants), and in this case Scots as well, have distinct vowels in
these categories. foot usually has something in the vicinity of [u], and
strut a large range, from lower mid back [∧] to something much fronter,
e.g. central [–a] to centralised front [ä] or even raised [ε].
(3) Lengthening I. In this (seventeenth-century) southern change, /æ/ length-
ened before the voiceless fricatives /f, θ, s/ and often /nt, ns/, so that
for typically southern dialects trap will have a short vowel and bath
(= bath, pass, dance . . . ) a long one, usually different in quality. The
quality change dates from the eighteenth and nineteenth centuries. Thus
Australian (AusE) and New Zealand (NZE) English have [æ] or [ε] in
trap, and low front [a:] in bath; SAE has the same qualities in trap,
but usually centralized back [ä:] or back [a:] in bath; and most of the
USA has more or less the same quality in both, but short trap and long
bath.
(4) Lengthening II. This later change lengthened /æ/ before voiced stops and
nasals except /ŋ/: so typically [æ] in trap, [æ:] or a slightly raised version
in bad, bag, man. This affects all southern dialects, and the United States,
South Africa, Australia and New Zealand as well.
In short, any variety of English that has [æ] or a higher vowel in trap, distinct
strut and foot, a distinction of length or length and quality between trap
and bath, and a length contrast in cat versus cad is southern. And obviously
all varieties of SAE fall into this group.
(1) Northern hemisphere ETEs. Mainly American and Canadian, though Irish
English (if not in all its features) also belongs to this group. The primary
input to these varieties is from the seventeenth to the early eighteenth
centuries (e.g. the USA first in 1607; Canada in part from 1583, but mainly
from 1713).
(2) Southern hemisphere ETEs. SAE and its offshoots such as Zimbabwean
(historically more correctly, Rhodesian) English, and Australasian English
(AusE, NZE). These derive from later colonisations (Australia in 1788;
New Zealand (from Australia) beginning in 1792, but established as a colony
in 1840; South Africa first in 1795 (Cape Colony), again in 1806 and 1814,
but with the first really large input in 1820, followed by major settlements
in the 1840s and 1870s).
We might expect that the earlier an ETE was established, the more ‘archaic’ its
features will be. Thus the stereotypical American will have [æ] in trap and [æ:]
in bath (the quality change of the lengthened vowel is mid-eighteenth century);
Australia and New Zealand have [a:] in bath (a typical late eighteenth-century
value), and SAE has the nineteenth-century backer vowel. In addition, the
Southern hemisphere ETEs are typically non-rhotic (do not pronounce postvo-
calic /r/: see section 3.3.4 below); /r/ loss begins seriously in the eighteenth
century, and is not complete until the early nineteenth.
There is another north–south divide as well. The southern ETEs are in most
ways typically ‘British’, as opposed to ‘American’: not only in pronunciation,
as we have seen, but equally striking in vocabulary. A few examples:
BRITISH SAE USA
petrol petrol gas(oline)
bum, arse bum, arse ass
dustbin dustbin garbage-can
chemist chemist drugstore
silencer silencer muffler
dinner-jacket dinner-jacket tuxedo
There are other (non-phonological) ‘British’ features as well, such as got as
past participle of get as opposed to the American distinction between got and
gotten (I’ve got some = I have some, I’ve gotten some = I’ve obtained some).
This is of course due to the USA having cut itself off from the larger British
community through the rebellion (or as the Americans call it, the revolution) of
1776, while South Africa, Australia and New Zealand retained their ties with the
mainland throughout the nineteenth and most of the twentieth centuries. (One
can still find out more about the British royal family from the Cape Times than
from the Washington Post.) These distinctions (at least grammatical and lexical)
are beginning to erode; the dominant world-wide role of American popular
culture is making American English increasingly familiar, with American terms
108 R. Lass
the face of it this should have led to a very strong Scots influence on the new
emerging English, at least in the Cape; but if the sociolectal picture among
the Scots clergy then was anything like that among older ones today, their
speech was probably quite ‘Anglicised’. In any case, their main contact was
with Dutch speakers, and the institutionalised norms for Afrikaans English
(AE) are still southern English, not Scots (e.g. Afrikaans English has distinct
trap and bath, and a goose/foot distinction, neither of which occur in
Scots).
Even after the establishment of the Boer republics (Orange Free State and
Transvaal), English remained among much of the Dutch-speaking population
the language of geleerdheid (‘well-educatedness’), and continued to be used in
commerce and major aspects of public life (see Lanham and Macdonald 1979:
9–18 for a more detailed account).
A second English-speaking influx arrived with the Natal settlers of the 1840s
and 1850s; these, unlike the predominantly rural or urban working-class input
of 1820, were largely standard speakers, e.g. retired military personnel and
financially hard-pressed aristocrats. They brought into the eastern part of the
country another (later) standard variety, and there is still a kind of ‘hyper-
English’ stereotype associated with Natal.
The final major wave, of very diverse origins, came about 1875–1904, after
the discovery of gold on the Rand. This was probably the most dialectally het-
erogeneous lot of them all, but it (like the Natal input) seems not to have had
a major effect on the subsequent development of SAE as a distinct type; the
seeds of that development were already sown in 1820. What the Natal settle-
ment may have done is to entrench more deeply a particular set of ‘colonial’
(i.e. nostalgic) attitudes toward British norms; and this, in all parts of the
country, has been of considerable importance in defining the lay notion of
‘standard’. Lanham has claimed that the Natal input is particularly important as
the source of certain variables such as ‘glide-weakening’ (i.e. monophthongi-
sation) of certain diphthongs, but these processes are so widespread in English
that they cannot be argued to have a specific regional input; they could come
from anywhere, unlike the specifically southern features mentioned above.
(For the type of argumentation involved in showing – or not showing – that
a particular South African feature has a specific regional origin, see Lass and
Wright 1986.)
‘East Cape’ makes social class and gender less indicative; this does
not seem to be the case nowadays, certainly not in the Western Cape,
where stratification is equally clear for both genders, regardless of
Eastern Cape connections.
While these social characterisations may have had some weight in the 1960s
and 1970s, two or three decades later only the bare outlines remain acceptable;
the correlations with gender and class (or better, gender stratification within
social classes)7 are still operative, but in a rather different way, the ‘East Cape’
variable plays no role, and there are many subvarieties of interest (if poorly
studied) within the three groups.
Without going into the details of Lanham and Macdonald’s quantitative stud-
ies, many of whose results are uncertain, we can characterise the lectal hierarchy
in a loose qualitative way, which in the present state of our knowledge is about
the best that can be done, and given present social fluidity is probably safer.
Conservative SAE: The type of speech least distinguishable from
Southern English, at its highest end (what I would call ‘Extreme
Conservative’) virtually RP of a rather archaic type. The most
familiar examples are reflected in the SABC announcer hierarchy
up to the early 1990s.8 Conservative accents were the ones typ-
ical of ‘serious’ news announcers, especially of anchor-persons.
Such speech is also common among the ‘first families’ of older
urban areas such as Cape Town, schoolteachers (especially English
teachers), and in general upper-middle-class people of a normative
disposition.
Respectable SAE: The local standard, that range of accent types asso-
ciated with all other white standard speakers, e.g. Democratic Party
and English-speaking National Party politicians, university lectur-
ers, teachers, physicians, accountants, lawyers (attorneys and advo-
cates both, though some of the latter may tend to be conservative).
Extreme SAE: The range of accent types associated with relatively low
socio-economic status, lack of education, and less skilled or non-
professional (‘blue-collar’) work, and the lower end of the ‘white-
collar’ scale. The more extreme a variety is, the harder it becomes
to distinguish it from second-language Afrikaans English.9
This trichotomy is cross-cut by the results of South Africa’s unfortunate social
history; in particular, the mother-tongue varieties of various ‘non-white’ com-
munities (Indian and coloured) have their own internal varietal stratification,
though speakers may ‘cross over’ in complex ways into the white hierarchy
(see Mesthrie and McCormick, chaps. 11 and 17, this volume). Ironically (see
section 3.2.4 below on goose), at least one of the distinctive characteristics of
112 R. Lass
the speech of those who have so long been excluded from the centre of South
African social life is in fact hyper-conservative.
diphthongs are excluded, for example, before the velar nasal /ŋ/: sing, sang,
song, sung, but not *boung, *beeng, *boong, etc.
3.2.2 The KIT split, TRAP and DRESS: the SAE chain shift
Southern ETEs have (at least in Type 2 and 3 lects) rather higher vowels
in the trap and dress classes than are found elsewhere. SAE also has a
very centralised nucleus (with complex allophonic distribution) in kit (see
below). According to one interpretation, referred to by Lass and Wright (1985)
as ‘atomistic’, these three phenomena are unrelated. Lanham and Macdonald
(1979: 46) say of ‘raised e’ and ‘raised æ’ that ‘origin is unknown’. The central-
ized [ï] in kit (for phonetic details see below) on the other hand ‘originates in
Afrikaans’. They claim this on the basis of qualitative similarity (the vowel in
Respectable and Extreme sit is virtually the same as that in Afrikaans sit). But if
this were in fact the source, it would present the rather extraordinary and highly
unlikely case of a language borrowing one single vowel quality (in the same
etymological category) from another; fortunately, there is another explanation,
which ties together both raised trap and dress and centralised kit.
The three qualities are related because they originate in a single (if very com-
plex) process: a nineteenth-century vowel shift (the ‘South African chain shift’:
see Lass and Wright 1985, 1986), in which raising of /æ/ toward [ε] and raising
of original /ε/ to [e] seems to have forced (most of ) original /i/ to centralise:
I → ï
↑
ε
↑
Or, using lower case for the older (and still mainland British) values, and small
caps for the newer (SA) ones, we can visualise the pattern this way:
kit → kit
↑
dress
↑
dress
↑
trap
↑
trap
At least this was the end result; the shift, however, was apparently not just
a matter of a simple raising at the outset. It seems to have arisen out of an
extremely complex and variable situation, where the vowels in these categories
114 R. Lass
had a fairly large set of realisations, including coexisting raised and non-raised
dress and trap, as well as raised, unaffected and lowered dress. The history
is extremely complex and controversial (see Lass and Wright 1985 for a detailed
discussion); but I can sketch out what the probable developments were.
We are fortunate in having a very good piece of evidence for the state of a char-
acteristic 1820 Settler variety: the enormous Chronicle of Jeremiah Goldswain,
a sawyer from Buckinghamshire who was one of the original immigrants.
Goldswain has left us some 528 foolscap pages in a ‘naive’ (i.e. non-standard)
and apparently partly self-invented spelling, which provide important clues to
the kind of English one would have found in the 1820 input. Among the features
are:10
(a) raised trap: contrector, ‘contractor’; atrected, ‘attracted’; lementation,
‘lamentation’;
(b) lowered dress: amadick, ‘emetic’; hadge, ‘hedge’; sant, ‘sent’;
(c) raised dress: git, ‘get’; kittle, ‘kettle’; liter, ‘letter’;
(d) lowered kit: presner, ‘prisoner’; deferent, ‘different’; sleped, ‘slipped’;
(e) retracted kit: buld, ‘build’; busket, ‘biscuit’; contunerd, ‘continued’.
Such complex developments of single historical categories are not unusual
in rural dialects of mainland English (see Lass 1987b for discussion); but in
South Africa they led in the end to a shift. It seems likely that Goldswain’s
spellings represent at least in part not fixed categorical values, but ‘zones of
convergence’, in which, say, a good part of the trap and dress classes ((a)
and (b) above) would occupy more or less the same part of the vowel space,
and the same with dress and kit ((c) and (d) above). Retracted kit, however,
moved into a ‘free zone’, since there were no other short vowels in its immediate
vicinity (the closest would be strut, and this was at the bottom of the vowel
space). What seems to have happened is that over time the four categories
involved ‘spaced themselves out’ by raising, and kit became more and more
centralised.11
SAE proper (that is, Respectable and Extreme)12 can in fact be defined by the
behaviour of kit: here (and nowhere else in the English-speaking world), the
words it and sit do not rhyme. Initially and after /h/ (it, hit), in velar environments
(kit, sick) and often before /ʃ/ ( fish), kit is closer and fronter; elsewhere it is
more centralized [ï]. (Though this is a good overall description, the pattern may
not always be quite this neat, since the split of fronter and backer allophones
may still be in progress, and there are therefore lexical exceptions. For details
of two speakers’ distributions, see Lass and Wright 1985: §5.)
This kit split (as Wells calls it) is one of the more striking social variables; in
Conservative SAE the contrast is lacking or nearly so; in Respectable (see Lass
and Wright 1985; Lass 1990b) the fronter value is [], as in all environments
in RP and most Conservative SAE; while in Extreme the fronter allophones
South African English 115
are raised and fronted further, to around [i] (often identical in quality to the
f leece vowel). Thus [] in it, sit marks Conservative (or more formal styles
for some Respectable speakers, especially female); [] in it and [̈] in sit is
the Respectable norm; and [i] in it and [̈] in sit characterises Extreme (and
generally Afrikaans) SAE. In Extreme varieties (and for many Respectable
speakers as well) there is considerable further retraction before syllable-final /l/
and after /w/, leading to merger or near merger with foot: this produces what
are perceived as non-standard homophone pairs like women/woman, bill/bull,
will/wool.13 In Extreme this retraction may also occur before and after /f/ as in
fifty, fit (which then becomes a near-homophone of foot).
The dress vowel is usually half-close front [e], and is not an important
variable, though it tends to be closer in female than male speakers in non-
Conservative varieties, and often quite centralised, approaching the fronter kit
allophones (hence ah big yaws ‘I beg yours’ in Malan 1972; on ah see below).
The pre-/l/ allophones in Extreme SAE often have a preceding [j], especially
initially and after /h/: help [(h)jεlp]. In some Respectable and Extreme, this
vowel lowers and retracts before dark /l/, to around [ε] or even [].
trap, on the other hand, is an important social marker. Both Conservative and
Respectable have [], sometimes a bit higher than RP //, but never approaching
[ε], which is the Extreme value, and used as an imitative stereotype. Some
Extreme speakers apparently perceive trap as so close to (the opener versions
of Respectable) dress that they write it that way: I have seen takkies spelled
tekkies on shop signs. Before dark /l/ in syllable codas it tends to lower and
retract, except, curiously in some Respectable pronunciations of the word Natal,
where it raises ([ntε]).
both Respectable and Extreme it is backer, even fully back [ɑ:] (generally
in Respectable backer for men and younger speakers). It becomes acutely
significant when, as in Extreme SAE, it may round to [ɒ:], and even raise
toward [ɔ:], which is a common stereotype (gimmia chorns, ‘gimme a
chance’, in Malan 1972). There is some evidence that at least weak rounding
is beginning to become less stigmatised now.
Some words with the apparently correct environment for this vowel
may have trap for some speakers (e.g. plaster, transition, substantial,
Flanders); it is unclear whether this is socially significant. Some items
apparently always have trap (masturbate, massive, gas).
frequently correlates with a rounded bath (see section 3.2.4 above), but
in any case the two normally remain distinct. This monophthongisation is
another stereotype: Malan (1972) gets a lot of mileage out of things like
laugh’s larkatt ‘life’s like that’, etc.
Some Extreme varieties have a characteristic triphthong for mouth, or
perhaps a diphthong with a palatal onglide, i.e. [ju], especially after /n/
and (variably deleted) /h/, e.g. house as [(h)jus].
(3) choice. This appears to be virtually the same in all varieties, with the first
element a little lower than the speaker’s thought, and the second the
higher version of kit. Some older Conservative speakers may, as in RP,
have the first element as open as lot.
(4) goat. Another important marker. In Conservative SAE, it is a diphthong
ending in [-u], with the first element centralised half-open [ε] or unround-
ed mid-central, in the general [ə] area; the centralised front realisations
may have some rounding; a general Conservative value would be [εu]
or [!u].
In Respectable SAE, unrounded first elements do not appear, the lip
rounding is stronger, and the normal onset is [!]. The second element may
be central [] or unrounded [¨"], and monophthongisation is common, espe-
cially in younger speakers. This can create a minimal contrast with nurse
([œ:] vs. [ø:]), as in boat, Burt). Outsiders often have trouble distinguishing
monophthongised goat from nurse, which may produce interesting con-
fusions: I recall once hearing an English-speaking politician refer to what
I was convinced was the turtle (total) onslaught.15 The monophthongisa-
tion is commoner among younger speakers (e.g. university students), and
does not seem to be linked with gender, as so many ‘advanced’ Respectable
features are.
In Extreme SAE, the first element of the diphthong is unrounded and
retracted, often in the vicinity of strut, thus making a back-gliding coun-
terpart to face: [∧u] matching [∧].
(5) square. In Conservative SAE, as in RP, this is typically a diphthong of the
general type [ε]. In Respectable and Extreme it monophthongises, more in
the latter than the former, and with a closer articulation. To put it rather sim-
ply, Claire would be pronounced only as [klε] in Conservative, [klε] or [klε:]
in Respectable, and [kle:] by younger Respectable and Extreme speakers.
(In Extreme the vowel may be closer, i.e. raised towards but not merging
with fleece.) This is a highly salient variable, and many Respectable
speakers (even those who monophthongise) stigmatise non-diphthongal
variants. Monophthongisation here then has a quite different value from
that which it has in price. This illustrates the important point that it is not
the actual phonetic nature of a linguistic object that gives it a social status,
but (probably arbitrary) evaluation.
South African English 119
Only SAE (all varieties), Scots and some Jewish varieties of English have a
phonemic velar fricative /x/. Except in Scots it does not occur in native English
words, but this is a purely historical matter; words that do have it (e.g. gogga,
gatvol, chutzpah) are normally so well integrated that they can be assumed
to have this extra phoneme. (On the bracketed voiceless /w / see section 3.3.3
below.)
3.3.4 Rhoticity
Some English dialects allow /r/ to appear in all positions: initially, between vow-
els, before consonants, and finally. Others allow /r/ only initially and medially,
never before consonants, and finally only in connected speech if the following
word begins with a vowel. The former type (/r/-pronouncing) are called ‘rhotic’,
the latter (/r/-dropping) ‘non-rhotic’.
A rhotic dialect then will pronounce /r/ in rat, very, cart, far; a non-rhotic
dialect will pronounce /r/ in rat, very, never in cart, and in far only if the word
following it begins with a vowel (/r/ pronounced in far off but not in far from,
etc.). This is called ‘linking /r/’. Some non-rhotic dialects have an ‘extended’
122 R. Lass
linking /r/, which appears not after certain words, but after certain vowels,
regardless of whether the word in question has a historical (orthographical)
<r>. This (as in Africa-r-and Asia) is called ‘intrusive /r/’. This is rare in any
form of SAE (some speakers with linking /r/ appear not to have it at all), and
tends to be stigmatised by Conservative speakers (even though typical of many
varieties of RP and similar mainland lects.) A third logically possible type,
with only intrusive and no linking /r/, does not appear to exist. There is an
implicational relation: intrusive implies linking, but not vice versa.16
SAE of all kinds, like the other southern ETEs, is generally non-rhotic, but
not always categorically so. Conservative SAE is fully non-rhotic in precisely
the same way as RP; Respectable is as well, but with some differences in detail,
and occasional sporadic rhoticity, especially in /r/-final words before pause or
hesitation, and in the name of the letter <r>. Anything beyond very sporadic
rhoticity (regardless of the quality of the /r/) is an Extreme marker.
(if not commonly) with native speakers as well, e.g. cre[h]ate, li[h]aison. This is
presumably in origin a weakened glottal stop; but it may have some social mo-
tivation as well, as a hypercorrection (the result of teachers correcting apparent
‘h-dropping’).
4 MORPHOSYNTAX
The morphology and syntax of L1 varieties of SAE have not been well studied,
and it would be premature to try and give any kind of detailed account. A
selection of miscellaneous grammatical South Africanisms can be found in
Branford’s Dictionary of South African English. There are some variables that
do seem (impressionistically) to be socially relevant, aside from those that
distinguish all varieties of SAE except perhaps the most Conservative (e.g. now
and just now with future meaning, loss of obligative force in must,17 lexical
archaisms such as robot for traffic-light, bioscope and of course Afrikaans and
other local loans).
One salient variable is the allomorphy of the definite and indefinite articles.
The common rule whereby the is /ðə/ before consonants and /ði:/ before vowels,
and the indefinite article is a /ə/ before consonants and an /æn/ before vowels,
holds for Conservative and most Respectable varieties; but for some Respectable
speakers the may (variably) be /ðə/ before vowels (normally with a glottal stop
preceding the vowel), though the indefinite article is never a before vowels. The
latter is characteristic only of Extreme (a apple, etc.).
Another feature that seems to be developing some importance as a marker
of Extreme is an extension of the busy V-ing construction to an increasing
range of stative verbs (on this construction see Lass and Wright 1986). In
most varieties of SAE (unlike other Englishes), busy V-ing can be used as a
relatively unemphatic progressive marker with some stative verbs (I’m busy
relaxing, etc.); but in Extreme it seems to have recently extended its domain
to die (I have heard Extreme speakers on television newscasts say things like
When I got to the car he was busy dying). Judging from the reactions I’ve heard
to this usage, it is developing into a strong social marker.
Most other Extreme markers of this kind are scattered and unsystematic, and
often of Afrikaans origin, apparently: familiar and stereotypic ‘errors’ castigated
by schoolteachers include I threw him with a stone = ‘I threw a stone at him’;
bring/come with = ‘bring/come with (someone)’, for example, bring it with for
‘bring it with you’; and the use of by in the sense ‘at somebody’s house’, such
as I had lunch by him.
One stigmatised variable is generally mistakenly interpreted as morphosyn-
tactic but is in fact phonological: this is what is perceived by many speakers
as ‘dropping’ forms of the verb be in (apparent) we going, you going for
we’re/you’re going, etc. A little reflection shows that these do not represent
124 R. Lass
systematic copula loss: *he going for ‘he’s going’ does not occur. What happens
here is simply a consequence of (a) non-rhoticity, and (b) monophthongisation
of near and cure (section 3.2.5). That is, while we has fleece and you has
goose, we’re has near and you’re has cure (the ‘underlying’ /r/ is realised as
[ə]). Deletion of this [ə] leaves behind monophthongs very close (but usually
not quite identical) to fleece and goose respectively.
These scattered examples reflect the state of the art; there has as yet been no
really detailed investigation of morphosyntactic variables of L1 SAE of the sort
there has been for phonological ones. One useful historical study of the grammar
of Settler English of the 1820s is that of Mesthrie and West (1995). They
examined a corpus of letters written by eastern Cape settlers to the governor
in Cape Town between 1820 to 1825. The letters contain a wealth of material
that give a fair idea of the dialect grammar of the period. Mesthrie and West
were particularly interested in (a) those grammatical features that have survived
to the present day (e.g. the use of an ‘adjective with infinitive’ construction,
as in incapable to provide for themselves); and (b) those features that were
eventually lost among Settler descendants but which survive in varieties that
started out as L2s, chiefly Afrikaner English and Cape Flats English (spoken
largely by coloured speakers). The authors provide the example of the dative of
advantage which was fairly common in the Settler corpus (e.g. I likewise dug
me a garden). No longer in use among white South Africans, the construction
survives in the Cape Flats, where it is ironically stigmatised as an incorrect use
of the reflexive. (On Cape Flats English see further McCormick, chap. 11 in
this volume.) The Mesthrie and West study thus offers a historical framework
for dialect syntax in South Africa.
notes
1 Scots has no /u/ vs. /u:/ contrast, but has /u/ or /–u/ in foot, food; bird, heard, word have
respectively/ir, ər, ∧r/. In the north-west of England cut, put both have /u/, as does
won, but one has /þ/ as in top.
2 Among the fragments of Scots heritage are pinkie, ‘little finger’ and timeously; the
north of England is represented by stay in the sense of ‘live in a place’, and the
secondary-stress on the prefixes con-, com- (as in cònfı́rm, còmpúter). Final stress in
educáte and similar words may be an Irish heritage, or an English archaism (since
such stressings were common in the late eighteenth century), or even an indigenous
development; the picture is not clear.
3 This property of Southern hemisphere Englishes (see section 1.3) was noticed as early
as the nineteenth century; A. J. Ellis in his massive dialect survey (1889) classified New
Zealand English as a sub-variety of ‘Southeastern’, along with the dialects of Kent
and Essex. It seems that no matter what the demography of the original settlement,
colonial Englishes turn out to be southern; on this phenomenon of ‘swamping’ see
Lass (1990a).
South African English 125
4 For more detailed treatments see Lass (1987a, chap. 5; 1990a); the latter is somewhat
technical, but has some maps that may be useful.
5 This is the type often called (somewhat erroneously) ‘BBC’ or ‘Oxford’ English.
For the history and description of RP, see Wells (1982: I, sections 1.1, 1.4, 2.1, 4.1).
6 Lanham and Macdonald (1979: 24) fail to define a white working-class population
(except for certain mining towns), on the odd grounds that ‘there is no labouring
class’, since manual labour ‘correlates almost totally with ethnic identity’ and is
minimally represented in the white population. To the eyes of an observer who spent
over a decade in the UK, this is simply false; there is a white working class as
clearly defined as that in England (both in terms of social attitudes and speech),
though it is not (and this holds for Britain as well) restricted to ‘manual labour’:
many white English-speaking policemen, railway workers, shop-assistants, etc. are
clearly not members of the Type 2 community, and speak Extreme SAE, just as their
counterparts in the UK speak recognisable working-class vernaculars.
7 The tendency of female speakers (especially working class and lower middle class) to
be somewhat ‘posher’ than males, and for lower-middle-class males to have ‘covert’
working-class norms, has been familiar since Labov’s work in New York in the 1960s
(e.g. Labov 1966).
8 This chapter was originally written at a time of major transition in South Africa;
some of the institutional identifications have changed, e.g. SABC is now SAFM (on
radio) and SATV. The varieties, however, will remain, and we need some descriptive
anchor point.
9 The stereotypical speaker of Extreme SAE (Sow Theffricun Innglissh) is the WUESA
(white English-speaking South African) parodied in Ah Big Yaws? (Malan 1972).
10 The only complete edition of Goldswain’s Chronicle is Long (1946–9); for a pio-
neering and still immensely valuable study see Casson (1955).
11 It might seem at first that kit too should have raised; but the original shift seems
to have been of the type that prohibited merger of categories, and raised kit would
intersect the short allophones of f leece (e.g. sit would become a near-homophone
of seat). Later on, however, (see below) some allophones of kit did indeed move
towards f leece.
12 ‘SAE proper’ includes Zimbabwean mother-tongue English as well, which appears
to be a Transvaal offshoot. Conservative SAE is of course ‘SAE’, but because
of its often rather archaic British character not entirely ‘proper’ in the relevant
sense.
13 Actually the homophony is often not total; women, bill may have a slightly less
rounded vowel than woman, bull. But the difference is so subtle that it is stigmatised
as a merger.
14 In some (especially male) Respectable speakers, there is some crossover, with the
first element of mouth at [a] and that of price a centralised back vowel, but fronter
than bath.
15 Judging from irritated letters I have been receiving as a panelist on what used to be
SABC’s ‘Strictly Speaking’ programme (now ‘Word of Mouth’ on SAFM), it now
seems that this monophthongisation has become salient, and stigmatised by older
Conservative speakers.
16 Not only is intrusive /r/ rare in SAE; in Respectable, linking /r/ is less common than
in British varieties, often being replaced by a glottal stop.
126 R. Lass
17 Most South Africans seem not to realise that must in other varieties is a strong
obligative verb; a simple neutral direction such as ‘Passengers must proceed to
Gate 2’ is perceived by non-South Africans at first as an order rather than a request
or piece of information, and they often find it offensive and ‘bossy’.
bibliography
Branford, J. (with B. Branford) 1991. A Dictionary of South African English, 4th edn.
Cape Town: Oxford University Press.
Casson, L. 1955. The Dialect of Jeremiah Goldswain, Albany Settler. UCT Lecture
Series 7. Cape Town: University of Cape Town.
Ellis, A. J. 1889. On Early English Pronunciation, with Especial Reference to
Shakespeare and Chaucer. Part V, Existing Dialectal as Compared with West Saxon
Pronunciation. London: Trübner.
Labov, W. 1966. The Social Stratification of English in New York City. Washington:
Center for Applied Linguistics.
Lanham, L. W. 1967. The Pronunciation of South African English. Cape Town: Balkema.
1978. ‘South African English’. In L. W. Lanham and K. P. Prinsloo (eds.), Language
and Communication Studies in South Africa: Current Issues and Directions in
Research and Inquiry. Cape Town: Oxford University Press, pp. 138–66.
Lanham, L. W. and Macdonald, C. A. 1979. The Standard in South African English and
its Social History. Heidelberg: Julius Groos Verlag.
Lass, R. 1987a. The Shape of English. Structure and History. London: J. M. Dent.
1987b. ‘How reliable is Goldswain? On the credibility of an early South African
English source’. African Studies, 46: 155–62.
1990a. ‘Where do Extraterritorial Englishes come from? Dialect input and recod-
ification in transported Englishes’. In S. Adamson, V. Law, N. Vincent and S.
Wright (eds.), Papers from the 5th International Conference on English Historical
Linguistics. Amsterdam: John Benjamins, pp. 245–80.
1990b. ‘A “standard” South African vowel system’. In S. Ramsaran (ed.), Studies
in the Pronunciation of English. A Commemorative Volume in Honour of A. C.
Gimson. London: Routledge, pp. 272–85.
Lass, R. and Wright, S. M. 1985. ‘The South African chain shift’. In R. Eaton, O. Fischer,
W. Koopman and F. van der Leek (eds.), Papers from the 4th International Confer-
ence on English Historical Linguistics, Amsterdam, 10–13 April, 1985. Amsterdam:
John Benjamins, pp. 137–62.
1986. ‘Endogeny vs. contact: “Afrikaans influence” on South African English’.
English World-Wide, 7: 201–24.
Long, U. 1946–9. The Chronicle of Jeremiah Goldswain, Albany Settler of 1820, 2 vols.
Cape Town: Van Riebeeck Society Publications.
Malan, R. [pseud. Rawbone Malong] 1972. Ah big yaws? Cape Town: David Philip.
Mesthrie, R. and P. West. 1995. ‘Towards a grammar of proto South African English’.
English World-Wide, 16, 1: 105–33.
Mitchell, A. G. and A. Delbridge 1965. The Pronunciation of English in Australia.
Sydney: Angus & Robertson.
Wells, J. C. 1982. Accents of English, 3 vols. Cambridge: Cambridge University Press.
6 South African Sign Language: one language
or many?
1 INTRODUCTION
In this chapter we discuss the signed language used by the Deaf community
in South Africa, and examine the historical conditions for its emergence. We
describe the legal and actual situation of South African Sign Language in South
Africa today, particularly in relation to schooling. We investigate the different
factors that underlie the claims that there is more than one sign language in
South Africa, and we spell out the practical consequences of accepting these
claims without further examination.
We assume without argument that Deaf1 people in South Africa, far from
being deficient, or disabled, are a linguistic minority, with their own language,
South African Sign Language, and their own culture, South African Deaf
culture.2 Like everyone else in this post-modern world, Deaf people have dif-
ferential membership in many cultures, on the basis of, for instance, religion,
lifestyle, daily practices, political beliefs and education. However, what they
all have in common is membership in a community that uses signed language,
and socialises with other people who do the same.3
Thus, the model we adopt is non-medical. We are not interested here in de-
gree of hearing loss, the remediation of hearing, audiological measures, speech
therapy or any other medical views of deafness. We regard deafness only as
the sufficient, but not necessary, precipitant of signed-language development,
and our concern here is to examine certain sociolinguistic issues that come into
play in the consideration of the status of the signed language used in South
Africa.
of expressing the entire range of human experience that spoken languages are
able to express; they have as many registers, and as much complexity as any
other human language. Signed languages have phonological, morphological,
syntactic and semantic levels of representation. These have been shown to be
exactly the same as those proposed for any other human language. The distin-
guishing feature of signed languages is that they are made through the medium
of space, not sound, and that they use the hands, face, head and upper torso for
their realisation.
There is no one universal signed language. Signed languages, just like other
languages, arise naturally, through use by a community of users in a context
of natural use, and they evolve and develop over time as they are passed down
from generation to generation. They differ from most spoken languages in the
important respect that only 10 per cent of Deaf children are born to Deaf parents,
and thus, Deaf children tend to learn signed language from other Deaf children
and adults, and not usually from birth, in their own homes. This, added to the
fact that signed languages are not written down, probably leads to a slightly
higher degree of variability in the signed language of a community. However, in
general, the signed language used in one country is identifiably distinct from the
signed language used in another country, particularly where these countries are
geographically and historically unrelated. Thus, for instance, Namibian Sign
Language and Thai Sign Language are mutually unintelligible.
Furthermore, signed languages are not related to the spoken language of the
geographical area in which they occur. Although English is the primary spoken
language in Britain and the USA, the signed languages of these two countries are
not related. If we compare American Sign Language and British Sign Language,
we discover that these two languages are mutually unintelligible. Historically,
American Sign Language is related to Old French Sign Language, since the first
teachers of the Deaf in the USA came from France. In any event, the indigenous
signed language in the USA did not evolve from the indigenous signed language
in Britain. Thus, although the official spoken language of both Britain and the
USA is English, British Sign Language and American Sign Language are not
mutually intelligible. South African Sign Language (as the case in point) can
trace some of its influences to Irish Sign Language, but less so to British Sign
Language.
In certain countries of the world – for example, Sweden – the natural signed
language used by the Deaf (in this case, the Swedish Deaf) is one of the offi-
cial second languages of Sweden, and users of Swedish Sign Language, as a
result, have all the language rights accorded to users of an official second lan-
guage. Deaf people thus have a legal right to receive their schooling in signed
language, and to have signed-language interpreters provided for all their official
interactions with the hearing public. This accords them full access to the life
of the country.
South African Sign Language: one language or many? 129
Some natural signed languages, such as American Sign Language and British
Sign Language, have been fostered and developed. As a result, their oral
tradition has spawned a body of signed language literature, which is now cap-
tured on videotape and is studied and analysed. Thus, just as users of other
languages keep a more permanent record of their artistic creations by writing
them down and studying them, users of signed languages, with the help of video
technology, have taken the opportunity to do the same.
These are a few examples of signed languages in countries that have recog-
nised the language rights of their citizens and made provision for the devel-
opment of these languages. More common is a deep and unfounded prejudice
against signed languages, and a consequent marginalisation of Deaf people and
their human rights.
It was not until the new constitution of the Republic of South Africa in 1996 that
education was declared compulsory for deaf children. It should be noted that
there were more Deaf children in South Africa (before 1994) who had never
been to school, than those who had attended a school at some time.
The history of sign language in South Africa is, of course, deeply intertwined
with the history of apartheid schooling and its complicated language policies.
For this reason, we present some of the details of the history of schools for the
Deaf in South Africa, with particular reference to the role of different churches,
and apartheid racial and ethnic classifications. Additionally we highlight the
different communication practices that were prescribed or emerged in the dif-
ferent schools for the Deaf.
To help the reader find a way through the morass of detail, we provide a
guiding generalisation: schools for the white Deaf insisted on oralism, whereas
schools for the other races allowed some measure of manualism (in most cases,
not a natural signed language, but a mixture of speech and some signs). It is
clear that speaking was perceived by the authorities as the prestigious form of
language. Hence there was an insistence on oralism in schools for the white
Deaf, while, based on pigmentation, manualism was permitted increasingly in
schools for the Deaf of other racial groups.
The churches most deeply involved in establishing and running schools for
the Deaf in South Africa were the Dominican Catholics and the Dutch Reformed
Church. The first school for the Deaf in South Africa was established in Cape
Town in 1863 by the Irish Dominican Order, under the leadership of Bishop
Grimley.4 This school, from its inception, catered for all race groups, and used
signed language as a medium of instruction. The Dominican nuns, who came
from Ireland, had been influenced by the policy of signed language instruction
originating in France in the eighteenth century, as a result of the work of the
Abbé Charles-Michel de l’Epée. In contrast, the policy in Deaf education in
Britain, and in Germany, was strictly oral; that is, Deaf children were taught
to lip-read, and made to speak. In Ireland, however, probably owing to Deaf
education being in the hands of the Catholic Church, the French policy of
manualism was entrenched.
A landmark event in the history of Deaf education world-wide was the Con-
ference of Milan, in 1880. All Deaf delegates were excluded from voting, and
the World Congress of Educators of the Deaf voted for a policy of strict oralism
in schools for the Deaf. This effectively excluded Deaf teachers from teaching
Deaf children and led, in most Deaf schools of the world, to signed language
going underground. It should be noted that Deaf people, wherever they were,
did not stop signing to one another. However, signed language world-wide was
frowned upon as a medium of instruction, and in many cases was forbidden. The
use of signed language also became stigmatised, and Deaf people, particularly
those who wanted to consider themselves educated, did not sign in public.
132 D. Aarons and P. Akach
By the time of the 1904 census, however, the Dominican Grimley Institute
in Cape Town (also known as St Mary’s) still embraced a policy of manua-
lism in the school. At that time two other schools for the Deaf had been
established in South Africa. These schools served only white Deaf children.
The Worcester School for the Deaf and Blind was established in 1881, by the
Dutch Reformed Church, for the children of the Dutch settlers. The 1904 census
report states that combined oral and manual methods were used in the school.
The folklore is that Jan de la Bat, a Dutch Reformed Church missionary, taught
his Deaf brother by means of signs, and that this heralded the beginning of the
signed language used in Worcester, which is claimed by this community to be
indigenous. Only ‘European’ children were permitted to attend this school.
In 1884, German Dominican nuns established a school at Kingwilliamstown
in the eastern Cape. This too was a school for the ‘European Deaf’ and fol-
lowed a policy of strict oralism, presumably because of the overwhelming
influence of oralism in Germany. The German Dominican School later moved
to Johannesburg, where it became St Vincent’s School for the Deaf, which took
in only white Deaf children.
In 1933, the Dutch Reformed Church set up another school, for the coloured
Deaf, known as Nuwe Hoop. The language policy was the same as that at the
Worcester school for the white Deaf: spoken Afrikaans, and some manualism.
The Grimley Institute for the Deaf in Cape Town remained racially integrated,
and in the 1920s segregated the children on the basis of whether they were to use
manualism or oralism. This occurred after one of the sisters visited the German
Dominican School in Kingwilliamstown, and instituted a policy that all but
the most ‘backward’ children would be taught using the oral method. In 1937,
the Irish Dominicans opened a separate school for the ‘non-European’ Deaf
in Cape Town at Wittebome. Both coloured and African Deaf children were
admitted to the school. However, by 1953, once the Nationalist government
refined the policy of apartheid even further, the Dominican Grimley School at
Wittebome was declared a school for coloured Deaf only.5
In the 1960s, the white Dominican Grimley School for the Deaf moved to
Hout Bay in Cape Town and adopted a policy of strict oralism which it has
continued to this day. Pupils are expected to maintain strict separation from any
signers, and absolutely no signing is permitted on school premises.
In 1962, apparently because there were still African students trying to attend
the Wittebome school, a separate school for African Deaf children was set
up in Hammanskraal (then in the Transvaal Province, some 1,600 kilometres
away from Wittebome), also by Irish Dominican nuns, from the Wittebome
school. There was no school for the African Deaf in the western Cape and
no attempt to set one up until 1986. This was in accord with the Nationalist
government’s policy of influx control (in terms of which no African children
actually officially belonged in the western Cape). Only after influx control had
South African Sign Language: one language or many? 133
been officially scrapped in 1986 did the Dutch Reformed Church set up a school
for the African Deaf in Khayelitsha, Cape Town.
The first school for black Deaf children, Khutlwanong, was opened in 1941,
near Roodepoort in the Transvaal. Started originally by the Johannesburg Deaf
and Dumb Society, it was taken over by Dutch Reformed Church trustees in
1954. At this school, a system of signs, invented in Britain, known as the
Paget–Gorman system, was introduced, and teachers and pupils were to speak
and simultaneously use the Paget–Gorman signs. This was a policy that was
to spread to other schools for black Deaf pupils. The Paget–Gorman system
was not a language but a set of invented signs, based on unnatural handshape
permutations, lacking a grammar at any level.
As a result of the homelands policy,6 a number of additional schools for
the African Deaf were established in the rest of the country, divided according
to the spoken language of each ethnic group, and in line with the Bantustan
separate development policy. Thus, from the mid-1950s, the following schools
for the African Deaf were established: The Khutlwanong School moved to
Rustenberg and served the so-called Tswana, South Sotho and North Sotho
‘speakers’; in 1959, the Efata School in the Transkei, for Xhosa ‘speakers’ was
established, also under the auspices of the Dutch Reformed Church; in 1962,
the Dutch Reformed Church set up Bartimea School at Thaba’nchu for Tswana
and South Sotho ‘speakers’, and in 1965 the Vuleka school at Nkandla for Zulu
‘speakers’. The Catholic Church established St Thomas’s at Stutterheim for
Xhosa ‘speakers’ in 1962. Kwa Vulindlebe was set up by the Catholic Church
in 1979, Kwa Thintwa was established by the Catholic Church in 1983; Indaleni
was set up by the Methodist Church of South Africa in 1986, St Martin’s by
the Catholic Church in 1991, all for Zulu ‘speakers’. The Tsilidzini school
at Shayadima was established to serve Venda and Tsonga ‘speakers’ and the
Thiboloha School at Witsieshoek, for South Sotho ‘speakers’. Yingisani was
established by the Department of Works in 1989 to serve Tsonga ‘speakers’. The
school set up in 1962 by the Dominicans at Hammanskraal officially catered
for Sotho ‘speakers’. In 1978 and 1981, two day schools were set up for urban
black Deaf children, one in Soweto and one in Katlehong.
Until the 1980s the official medium of instruction in all these schools was the
mother tongue, although in the case of Deaf children, it was not clear what this
was. Additionally, the schools were instructed to integrate the Paget–Gorman
signing system with mother-tongue speech. As is the case generally with educa-
tion for black people in South Africa in these years, the whole idea of dividing
schools up on the basis of the mother tongue of their pupils was fraught with
inconsistencies, and based on partial, and often incorrect, information. In the
case of the Deaf children, this was even more confused. Further, the use of
rudimentary signs to accompany the spoken language made the official lan-
guage practices even less communicatively accessible to the Deaf children.
134 D. Aarons and P. Akach
African Signs (Penn 1992a). However, there are a number of facts that cast
doubt on its veracity.
Deaf people moved around the country. As a result of the apartheid system of
schooling, Deaf children often had to leave their home districts to go to school.
After leaving school, they either returned home or went to work and live in
other parts of the country. Deaf people socialise with other Deaf people. More
recently, there has been signing on television in programmes for the Deaf,
and interpreting of national news, and thus, Deaf people are exposed to the
signing of different sectors of the Deaf community. There are frequent local and
national Deaf events of a sporting, cultural and educational nature. Initiatives
have been launched for the Deaf people within provinces to hold regular forums;
in the last few years, national Deaf indabas (conferences) have been held.
Deaf people are beginning to train other Deaf people to teach Sign Language
irrespective of whether they are from the same community. Anecdotally, the
most convincing piece of evidence is that Deaf South Africans seem perfectly
able to communicate easily with one another, although it is revealing that many
Deaf people believe that there are different sign-language varieties in South
Africa.
There seem to be reasons to claim that if there are different varieties they are
converging,7 as is the case in South Africa with different Englishes.8 There is a
strong possibility that convergence is taking place, owing to the far greater mix-
ing of different communities with one another, and the (somewhat minuscule)
integration of Deaf schools.
The linguistic decision as to whether there is one South African sign language
or whether there are many can only be made on the basis of linguistic research.
To this end there is a project under way to investigate the structural properties
of the signed language used by different communities in South Africa.9 How-
ever, the decision is also a social one, as people’s perception of whether they
use the same language as another person, or a different one, is frequently based
on considerations other than the structural properties of the language. In the
remainder of this chapter we examine other considerations, some of which are
pertinent to languages in general, and some of which have particular bearing
on South African Sign Language.
The question that seems to beset the official development of South African Sign
Language in South Africa is one that might appear to be a non-question: how
many different signed languages are there in South Africa? There are many
different ways of going about answering this question, the first of which is to
ask why it is being asked. Generally, the official response has been that until we
136 D. Aarons and P. Akach
know the answer to this question, we cannot choose a standard variety. Only
then can we begin with interpreter training, and sign-language training for pre-
and in-service teachers, and with the introduction of television interpreting,
school curricula and syllabi for South African Sign Language, and so on. The
next question we might ask is: ‘Who is asking?’ And we may find that it is not
Deaf people who are asking this question, but educators of the Deaf, would-be
interpreters, bureaucrats and financial managers. For it is costly in terms of
time, effort and money to have to take responsibility for the promotion and
development of yet another language group in South Africa.
We propose to examine a number of the arguments, claims and beliefs that
underlie the commonly heard statement that there is more than one sign language
in South Africa. Not all the claims are compatible with one another, as they
are merely culled from received wisdom, and set down here as a list. We show
that in all these cases, the factors that are brought to bear on the discussion of
the signed language are non-linguistic ones. They have nothing to do with the
structure of the language itself. We will list these below as baldly as possible
in order to explicate them:
4.1 Claims
(1) For every different spoken language in South Africa, there is an equivalent
signed language, i.e. there is an English Sign Language, an Afrikaans Sign
Language, a Sotho Sign Language, a Zulu Sign Language, etc.
(2) For every different racial and ethnic community in South Africa, there
is a different sign language. Thus, for example, ethnically Indian South
Africans have their own signed language, English-speaking coloured South
Africans have their own signed language, Afrikaans-speaking coloured
South African have their own signed language, and these are different from
the white English or Afrikaans South African signed languages.
(3) For every different geographical or ethnically separated Deaf community
there is a different signed language. Thus, Deaf people from an English
school in Natal use a different signed language from that used by Deaf
people from an English school in Johannesburg.
(4) There is a word for every sign and a sign for every word (for argument’s
sake, in English).
(5) Signed languages mirror the morphological and syntactic structure of the
spoken languages from which they derive.
(6) Signs or signed-language utterances do not vary in their context of use.
(7) A standard language does not allow for regional, ethnic, gender, situational
or contextual variation.
(8) If people are born into a certain community, or culture, their primary loyalty
and identification must be to the language used in that particular community.
South African Sign Language: one language or many? 137
In order to make our case quite clear we will make some counter-claims about
South African Sign Language and then substantiate them. We claim first that the
reason some people say that there are different signed languages in South Africa
is based on a fundamental misunderstanding of the signed language itself: they
assume that signed language is a manual version of spoken language (claims 1,
2 and 3). Second, we claim that the Dictionary of Southern African Signs is
based on a false hypothesis about the effects of apartheid on the signed language
used in South Africa. It also seems to presuppose a close relationship between
words and signs, and fails to recognise variation within different contexts of use
(claims 4, 5 and 6). Although there are certainly different varieties of the signed
language used in South Africa, most Deaf people in the country control many of
these varieties, as is the case for speakers of any other language (claim 7). Third,
we claim that Deaf people themselves frequently confuse language identity
with other kinds of identity and thus sometimes reject the signing of other Deaf
people as ‘other’ (claim 8).
this explains why Deaf black South Africans from ten different mother-tongue
backgrounds communicate easily with one another in signed language, nor why
there are conflicting reports from Deaf South Africans of different racial groups
about whether or not they use the same signed language.
What we do know is that Deaf people seem to manage very well to com-
municate with one another across racial boundaries, until there are hearing
people (teachers, social workers, ‘interpreters’) involved. Apparently, many
hearing people use manual codes that are associated with a particular spoken
language. Then only the Deaf people who understand the particular spoken
language understand them. Similarly, some hearing people may understand a
signed form of a particular spoken language, but not the natural signed language
used by the Deaf themselves. Invariably, it is the hearing people who raise the
complaint that they do not understand ‘Zulu sign language’ or ‘Afrikaans sign
language’.
Combined with the complication of accommodation with hearing people’s
signing is the issue of colour. It is our observation that signers tend to decide
whether someone else’s sign language is the same as or different from theirs
on the basis of their skin colour. A skilled signed-language interpreter in the
Western Cape (totally bilingual in English and Afrikaans as well) who hap-
pens to be a coloured South African was informed that the white Afrikaans
Deaf did not understand him. Conversely, one of the authors, who happens to
be Kenyan, and knows no local Bantu languages, is frequently complimented
by black South African Deaf people on how well he uses the local signed
language.
Clearly, as shown above, the various Deaf communities did not mix much
over the years preceding the dictionary, and as one would expect, the signed
language used by different groups would have shown some lexical variation, a
variation perpetuated by apartheid divisions. The dictionary focused on these
lexical differences, attempting to correlate the different lexical items with the
spoken-language communities into which the Deaf users were born. To this
end, the project team documented signs from eleven different racially and re-
gionally based areas in South Africa. Researchers used English words and
phrases to elicit particular signs from the representatives of each community.
These signs were video-recorded, and then a still frame was made from each,
and presented as the sign used by the different communities for the particular
English word or phrase. Thus, each page of the dictionary listed an English
item, then showed eleven or so different signs that informants claimed were
the ways in which the sign for this English word was used in their language
variety.
It is difficult to see what purpose the dictionary would have served in standard-
ising the signed language used in South Africa into a single signed language. It
seems more likely that the dictionary’s purpose was to standardise each of the
different varieties. This idea is quite in accord with the practice under apartheid,
whereby language boards for each Bantu language were set up, usually compris-
ing non-native speakers of that language. The standard for a particular language,
for instance Xhosa, was then decided upon, and then this standard variety was
prescribed for use in and teaching in schools. Native speakers of the language
would find that their own variety was then deemed to be faulty as a consequence
of the decrees of the language board.10
It should also be noted that the dictionary had a stated pedagogical aim (Penn
1992a; Penn and Reagan 1994). Thus, its purpose was not only to describe the
varieties used by the different communities, but to use the items for teaching one
or other signed language. The issue of signed-language syntax is not addressed
directly in the dictionary itself (although there is some discussion of the syntax
of signed languages in general, in the introduction). The pedagogical aim, then,
seems geared more to teaching some sign vocabulary within the context of
an English sentence structure. The pedagogical outcome of such an approach
is unlikely to be the acquisition of a natural sign language. In any event, no
dictionary of any language could be said to actually teach a language.
The first serious misunderstanding upon which the Dictionary of Southern
African Signs is based, then, is that the structure of a signed language is depen-
dent on the structure of a related spoken language. The second misunderstanding
is that there is a one-to-one relationship between a lexical item in one language
and a lexical item in another, in other words, that there is a simple word–sign
relationship. In any event, the base items for elicitation in the dictionary were
English sentences. This seems to underplay the relationship among the different
142 D. Aarons and P. Akach
As Deaf people in South Africa become more committed to Deaf rights, Deaf
consciousness, Deaf pride, Deaf unity and Deaf power, these language differ-
ences seem to become smaller. Deaf people start to see themselves as bound by a
common language and a common struggle. The debate about how many signed
languages there are in the country becomes a non-question. This divisiveness
serves the needs of communities other than the Deaf and must be recognised
as arising out of important social forces that have a bearing on the social and
political, but not the linguistic, status of the natural language of the Deaf people
in South Africa. The real issue is how the rights of Deaf people as a linguis-
tic minority can be achieved, including the right to have signed language as a
medium of instruction in schools for the Deaf, state funding for the training of
skilled signed-language interpreters and signed-language teacher trainers, and
the provision of interpreters and services to ensure equal access for Deaf people
to the life of the community.
notes
1 In accordance with convention in the field of Deaf Studies, we use upper-case D (Deaf )
when we refer to people who identify with the Deaf community and who use signed
language, and lower case d (deaf ) to refer merely to the audiological condition.
2 See, for further argument and discussion, Aarons (1996).
146 D. Aarons and P. Akach
3 For an interesting and full discussion of, for example, American Deaf culture, see
Padden and Humphries (1990).
4 Note that this was before South Africa existed as a single national state, some forty-
seven years before Union.
5 We use apartheid terminology in order to show the distinctions that were maintained.
6 This was the apartheid policy of separate development, in which the idea was to
separate white South from black South Africans, and then further divide black South
Africans into a number of ethnic groups, each with its own ‘homeland’. Black people
were then considered ‘citizens’ of their designated homeland, and not South Africans.
7 For a discussion of convergence, see Thomason and Kaufmann (1988), and for a
discussion of the convergence of signed language varieties see Okombo and Akach
(1997).
8 See, for example, Lanham 1996.
9 ‘An investigation into the linguistic structure of the signed language/s used in South
Africa’. CSD Grant number 15/1/3/16/0125 to Debra Aarons and Ruth Morgan.
10 See, for example, Nyamende (1994).
11 To establish this uniformity, in conjunction with the well-documented ubiquitous
process by which classifier morphemes are used in signed languages of the world,
it would seem to be sufficient to make the claim that the same language is being
investigated, irrespective of lexical variation.
12 See Aarons (1994, 1996).
bibliography
Aarons, D. 1994. ‘Aspects of the Syntax of American Sign Language’. Ph.D. dissertation,
Boston University.
1996. ‘Signed languages and professional responsibility’. In Stellenbosch Papers in
Linguistics, 30: 285–311.
Aarons, D. and R. Morgan 1998. ‘The Structure of South African Sign Language after
Apartheid’. Paper presented at the Sixth International Conference on Theoreti-
cal Issues in Sign Language Research, Gallaudet University, Washington, D.C.,
November 1998.
Constitutional Assembly of the Republic of South Africa. 1996. Constitution of the
Republic of South Africa.
Heap, M. (in progress) ‘An anthropological perspective of the Deaf people in Cape
Town’. University of Stellenbosch, Department of Anthropology.
Lanham, L. W. 1996. ‘A history of English in South Africa’. In V. De Klerk (ed.), Focus
on South Africa. Amsterdam: John Benjamins, pp. 18–34.
Nieder-Heitmann, N. 1980. Talking to the Deaf. South African Department of Education
and Training and the South African National Council for the Deaf.
Nyamende, A. 1994. ‘Regional variation in Xhosa’. Stellenbosch Papers in Linguistics
PLUS, 26: 202–17.
Ogilvy-Foreman, D., C. Penn and T. Reagan 1994. ‘Selected syntactic features of
South African Sign Language: a preliminary analysis’. South African Journal of
Linguistics, 12, 4: 118–23.
Okombo, O. and P. Akach 1997. ‘Language convergence and wave phenomena in
the growth of a national Sign Language in Kenya’. International Journal of the
Sociology of Language, 125: 131–44.
South African Sign Language: one language or many? 147
Padden, C. and T. Humphries 1990. Voices from a Culture. Cambridge, Mass.: Harvard
University Press.
Penn, C. 1992a. Dictionary of Southern African Signs, 5 vols. Pretoria: Human Sciences
Research Council.
1992b. ‘The sociolinguistics of South African Sign Language’. In R. K. Herbert (ed.),
Language and Society in Africa. Johannesburg: Witwatersrand University Press,
pp. 277–84.
Penn, C. and T. Reagan 1994. ‘The properties of South African Sign Language: lexical
diversity and syntactic unity’. Sign Language Studies, 84: 319–28.
Thomason, S. and T. Kaufmann 1988. Language Contact, Creolization and Genetic
Linguistics. Berkeley and London: University of California Press.
7 German speakers in South Africa
Elizabeth de Kadt
148
German speakers in South Africa 149
situated in the urban Transvaal, and substantial numbers (7,500) in the urban
Cape. KwaZulu-Natal has relatively small numbers of German speakers: the
1980 figures put the German speakers in the Durban–Pinetown–Inanda area at
somewhat under 2,000. To these, however, must be added possibly twice as
many rural German speakers in KwaZulu-Natal; the census does not provide
details as to these.
The above already gives some indication of the two main groups of German
speakers in KwaZulu-Natal, the urban and the rural. In the rural areas small
communities are scattered throughout KwaZulu-Natal, but especially in north-
ern KwaZulu-Natal and the midlands. The smallest of these (e.g. Harburg,
Hermannsburg (275)1 in the midlands, Braunschweig (219) and Luneburg (316)
in northern KwaZulu-Natal) consist of little more than a church, school, post
office and shop, which serve the surrounding farming communities. Among
the white population these tend to have a majority of German speakers. How-
ever, the larger the settlement the lower the proportion of German speakers
150 E. de Kadt
until, in Dundee (427) and Vryheid (420), for example, it is less than 5 per
cent of whites. This, however, has so far been sufficient to sustain a church and
school. Typically, these German speakers are descendants of the settlers and
missionaries who came to South Africa during the second half of the nineteenth
century.
In the urban areas, on the other hand, German speakers are more scattered,
although here too they tend to be more concentrated in certain (often wealthier)
suburbs (in Durban and surroundings, for example, in the suburbs of Westville,
Kloof, New Germany and Gillitts). Again, these may be descendants of the older
settler families, but they also include a considerable number of more recent
immigrants (pre- and post-Second World War). To generalise again, the ma-
jority of urban German speakers in South Africa tend to be professionals in
technical and managerial fields. In the major urban centres there are generally
German-medium churches and schools available, but considerable numbers
choose English- or Afrikaans-medium facilities.
KwaZulu-Natal German speakers, with the exception of very recent immi-
grants, are typically multilingual: they speak German, English and/or Afrikaans
and, in rural areas, Zulu. On the whole German tends to be used in very restricted
domains: family, church and, to a certain extent, school. Even in the relatively
homogeneous rural communities family life is open to the influence of the media
(newspapers, magazines, radio and television), but here social life is mainly
based on German. In urban areas, in spite of the existence of German clubs,
social life is more open to English and Afrikaans (wider circles of friends, cine-
mas, etc.); however, here there exist some possibilities for the use of German
in a professional capacity, in industry, import–export businesses, shipyards,
travel agencies, etc. One informant spoke of relatively large numbers of German
speakers in middle management in Durban, but no hard data is available on this.
The present trend both in urban and rural communities is increasingly towards
language shift; there is an awareness among German speakers that the next two
generations may well see an irrevocable decrease in numbers. The clearest
indication lies in the rapidly increasing number of so-called Mischehen, ‘mixed
marriages’, meaning marriages between German and English or Afrikaans
speakers, which nowadays, as opposed to twenty years ago, tend to result in the
children speaking English or Afrikaans as L1. This has had important conse-
quences for schools and churches, which have previously played a crucial role
in maintaining German.
The schools that had been founded by the early settlers were, in the course of
time, integrated into the Natal Education Department (NED) schooling system
as so-called German primary schools. This means that they have departmental
German speakers in South Africa 151
permission to teach the first four years through the medium of German; in
Grades 5–7 English or Afrikaans is used as a medium, and German is taught as
a subject. The number of such schools has eroded to a certain extent over the last
thirty years: at present there are KZNED primary schools in Luneburg, Uelzen,
Wartburg, Izotsha, Harburg and Moorleigh; state-aided schools in Vryheid and
New Hanover; and private primary schools in Durban and Hermannsburg. Two
‘German’ high schools exist, Wartburg-Kirchdorf, a government school (which,
however, only offers German as a higher-grade subject), and, most importantly,
the private high school at Hermannsburg, which includes a hostel. The much
lower numbers of urban German speakers in KwaZulu-Natal mean that Durban
cannot support a combined primary and high school (the Durban primary school
for example has only approximately eighty pupils), as opposed to the ‘German’
schools in Johannesburg with one thousand one hundred pupils, Pretoria with
approximately seven hundred and Cape Town with over four hundred. Hence
the children from ‘German’ primary schools in KwaZulu-Natal continue their
schooling either at Hermannsburg or at the local high schools, where they
form a tiny minority and are only catered for by the subject ‘German as a
foreign language’. For native speakers of German, this is probably worse than
useless. The Hermannsburg school, on the other hand, uses English as medium
after the first four years and leads to the KZNED matric; but it also offers the
subject ‘German as a mother tongue’ through the Independent Examinations
Board (formerly the Joint Matriculation Board). This should be compared to
the ‘German’ schools in Johannesburg, Pretoria and Cape Town, which use
German as a medium to a much greater extent. They offer a local matric and
hence switch to English as a medium in part in Standard 6 and completely
in Standard 8, but it is then possible to continue with a thirteenth school year
taught completely in German, leading to the German Abitur.
Most of the rural ‘German’ primary schools are now threatened with closure
due to low numbers; they have been able to justify their continued existence
only by opening to (white) non-German speakers. For example, in Uelzen (near
Dundee), half the pupils (forty out of eighty) are now English speaking; even
in the very homogeneous area around Luneburg, ten out of seventy children are
Afrikaans speaking. It is only the state-aided Michaelisschule in Vryheid that
has so far been able to remain closed to non-German speakers; in 1989 there
were sixty-five pupils.
The private schools, on the other hand, while also faced with similar prob-
lems, show a somewhat different pattern: they too have introduced a stream
of ‘German as a foreign language’, but this is at least in part to accommodate
black pupils. This is a new development over the past few years, and is to a
certain extent the result of pressure applied by the official funding sources in
Germany. In Cape Town, for example, 120 out of just over 400 children are
now non-German speakers; in Johannesburg, the figure is approximately 100
152 E. de Kadt
out of 1,100. The Hermannsburg primary school remains solely German, but
forty non-German pupils have now been admitted into the high school, where
their curriculum includes ‘German as a foreign language’.
Let us now consider the changes taking place in the churches. Most of the
German-speaking churches originated in connection with the Lutheran mis-
sion in KwaZulu-Natal. Today there are a number of different branches of
Lutheranism in KwaZulu-Natal, some of which cater largely or solely for non-
white communities and no longer use German. German-speaking parishes still
exist in the following centres: Izotsha, Durban (Renshaw Road with a second
church in Hillcrest, Westville and New Germany), Pietermaritzburg, Wartburg,
Harburg, New Hanover, Hermannsburg, Moorleigh, Winterton, Elandskraal,
Dundee, Vryheid and Braunschweig. These are all small parishes: for example
in Durban all three parishes together have only 650 members. Increasing in-
termarriage with mainly English or Afrikaans speakers has led most of these
parishes to cater for non-German speakers too. Hence the last ten years or so
has seen the introduction of church services in English, at first once a month,
now generally more frequently; Uelzen (near Dundee) has both a German and
an English service each Sunday. It is only a few parishes in northern KwaZulu-
Natal, e.g. Luneburg and Braunschweig, that even today offer solely German
services, as ‘mixed marriages’ are still the exception in these communities. In
the larger urban areas, on the other hand, services in English also enable parishes
to minister to those non-white members for whom new Lutheran parishes have
not been established; some of these parishes lay great stress on non-racialism.
Of the Durban parishes, New Germany has an English service each week, the
Hillcrest church every second week, Renshaw Road every fourth week and then
together with its English-language sister parish of the Union Lutheran Church.
It is clear that policies of non-racialism will have linguistic effects. For example,
the lingua franca at the small Lutheran residence that accommodates students
training as Lutheran priests in the department of theology at the University of
Natal in Pietermaritzburg has changed during the past few years from German
to English. This is due to the fact that black Lutherans are now also being
admitted.
The tendencies discussed above also hold for the other provinces, with
the difference that German in the rural areas has been eroded to a much
greater extent. The German influence in the Free State was limited from the
start; today small groups totalling perhaps 250 in all are to be found in the
area from Kimberley and Kroonstad to Bethlehem, including Bloemfontein,
which has a German-Afrikaans parish and a German club. In Gauteng only
three ‘German’ primary schools remain: Kroondal, Wittenberg, Gerdau, with
Kroondal (fifty-five to sixty pupils, plus twenty in the pre-primary phase) being
the largest school; in the Eastern Cape schools have been closed and parishes
still exist only in the urban centres, such as Port Elizabeth and East London.
German speakers in South Africa 153
3 LANGUAGE MAINTENANCE
the central role of the petty bourgeoise worldview which has been perpetuated
in the ethos of these communities, and furthermore of an ethnicity based on
German Romanticism (Pakendorf 1997). Forsythe has investigated mainland
perceptions of the term ‘German’ in the late twentieth century and has shown
how racial, genetic and linguistic elements are intertwined in the central concept
of Deutschstämmigkeit, ‘being of German stock’ (Forsythe 1989). It is self-
perceptions such as these that would seem to underlie the determination to
maintain the German language.
One of the most far-reaching mechanisms of language retention was the
establishment of ‘own’ schools and churches, which was always one of the
priorities of a new German-speaking settler community. For these some financial
support may have been forthcoming from missionary societies in Germany, but
on the whole the settlers were willing to bear the expenses themselves, in spite
of the often enormous struggle to establish themselves in the new country.
It was of considerable significance for language-maintenance efforts that the
‘German’ schools were allowed to retain something of their own identity when
finally taken over by the Natal Education Department in 1925 and funded by
state resources. While in part due to the number of well-functioning schools
in existence, it is also a reflection of the economic and social power of the
German community in KwaZulu-Natal.2 Clearly, state resources available to
such initiatives are likely to dwindle in the future: the maintenance of German
then becomes a matter of the resources its speakers can muster, either from
inside the community or from abroad.
The German-speaking communities have from the start been willing to con-
tribute substantially to the maintenance of churches and schools, and the com-
munities appear to be aware that state funding can no longer be relied on.
In KwaZulu-Natal, the three primary schools which at present receive little
or no state funding survive on the basis of trust funds and extensive fund-
raising. (It is unclear to what extent German firms in South Africa contribute.)
In Vryheid, private fund-raising recently enabled the building of a boarding
establishment solely for German-speaking children at Vryheid’s primary and
secondary schools: this is intended to cater for the whole of northern KwaZulu-
Natal and Gauteng. In northern KwaZulu-Natal in particular, German speakers
seem aware of the financial implications of maintaining schools, and they seem
willing to make substantial sacrifices to achieve this aim.
German cultural foreign policy has over the last decade stressed the necessity
of promoting the German language abroad, in view of the world-wide decrease
of interest in the language. In this context South Africa receives substantial
financial aid which is directed primarily at the four main German schools in
the country: Johannesburg, Pretoria, Cape Town and Hermannsburg; indeed,
these four schools have a top rating among the German schools supported
world-wide. The aid given takes the form of funding forty-one teachers sent out
German speakers in South Africa 155
from Germany, and a subsidy for each pupil, which, in 1990 with 2,465 pupils,
amounted to a total of approximately R4 million. Some contributions are also
made for essential building projects, textbooks and so on. A co-ordinating
subject adviser oversees the whole aid programme. There are also an exchange
programme and scholarships for pupils and teachers, which in 1990 amounted
to close to R300,000. The four schools concerned would clearly be hard put to
continue, were this aid from Germany withdrawn.
‘jam’, ‘lift’, etc. A considerable number, however, have been integrated: ‘fence’
as Fenz; ‘krans’ as Kranz; ‘hooter’ as Huter, etc. New words have been formed
according to English and Afrikaans patterns: Armstuhl, ‘armchair’; Dornbaum,
‘thorn tree’; Fruchtkuchen, ‘fruit cake’; Grosskinder, ‘grandchildren’; Kohl-
mine, ‘coal mine’; Seekuh, ‘hippopotamus’ (from Afrikaans seekoei); Werkwort,
‘verb’ (from Afrikaans werkwoord ).
More subtly, the meanings of a number of already existing German words
have changed: for example, Hochschule from ‘university’ to ‘high school’.
Garage has come to include the English ‘garage’ which sells petrol and repairs
cars (HG Autowerkstatt). Erbe, HG ‘inheritance’, has gained the meaning of
Afrikaans erf, ‘plot of land’.
The few borrowed nouns of Zulu origin are, on the whole, words that have
also been adopted into South African English, such as donga, ‘dry, eroded
water-course’; masi, ‘thick soured milk’; muti, ‘(African) medicine’. Borrowed
adjectives are, as to be expected, fewer in number: busy (as in Ich bin busy, ‘I am
busy’); mal, ‘crazy’; pap, ‘exhausted, soft, deflated’; and sorry, ‘sorry’ (very
frequent). Borrowed verbs have generally been integrated: abswitchen, ‘switch
off’; booken, ‘book’; huten, ‘hoot’; kloppen, ‘do better than, beat’ (e.g. Karl hat
mich (im Test) gekloppt, ‘Karl has beaten me in the test’); posten, ‘post’; swotten,
‘swot’. There have been some substantial changes in meaning: (ver)missen, HG
‘miss a person’, now also used in the sense of ‘miss a bus’; ringen, HG ‘wrestle’,
now ‘to ring a doorbell’.
Morphology and syntax show changes in a number of central features of
German grammar: the marking for gender of non-personal nouns; the obligatory
link between preposition and specific case; the governing of cases by verbs.
Indeed, there seems to be a considerable amount of uncertainty as to the need
for case in the language, which might well reflect the influence of English and
Afrikaans, neither of which is structured by case to the same extent. In the
following, some of the more frequent changes are listed. Personal pronouns
(third person) are frequently used in the dative case, in the place of direct
objects (examples 1–3 below), or with prepositions that standardly govern the
accusative (examples 4 and 5 below).
(1) Ich hoffe, sie wird ihm heiraten. (HG ihn) ‘I hope she will marry him.’
(2) Die Katze beisst ihr. (HG sie) ‘The cat bites her.’
(3) Frag ihr doch! (HG sie) ‘Do ask her!’
(4) . . . meinen Dank an Ihnen richten (HG Sie) ‘to express my gratitude to you’
(5) Er hat es für ihr getan. (HG sie) ‘He did it for her.’
This contradicts the increasing spread of the accusative in High German, at the
cost of the dative and the genitive. Stielau (1980: 218–19; see also Russ 1990:
17, 47) notes the lack of differentiation between dative and accusative of these
pronouns both in certain North German dialects and in English and Afrikaans.
German speakers in South Africa 157
In the latter two languages the form of the third-person pronoun in the object
position – hom, ‘him’ and haar, ‘her’ – is closer to the German dative ihm and
ihr than to the accusative ihn and sie.
On the other hand, there is frequent use of the accusative instead of the dative
with prepositions that govern the dative or dative/accusative (examples 6–8
below), and with many verbs that govern the dative (examples 9–12 below).
This change seems to be particularly common with feminine nouns requiring
the article die, which may suggest Afrikaans influence.
(6) bei die Kirche (HG der) ‘at the church’
(7) mit viele Firmen (HG vielen) ‘with a lot of firms’
(8) Er war auf die Stelle tot. (HG der) ‘He was dead immediately.’
(9) Ich werde es Sie erklären. (HG Ihnen) ‘I will explain it to you.’
(10) Ich gratuliere dich! (HG dir) ‘I congratulate you!’
(11) Sie hilft die Studenten. (HG den) ‘She helps the students.’
(12) Ich werde dich nicht sagen. (HG dir) ‘I will not say (it) to you’
Similarly, these intransitive verbs are used to form a non-standard personal
passive, as opposed to the HG impersonal passive:
(13) Wir werden nie gesagt, wann . . . (HG Uns wird nie gesagt . . . ) ‘We are
never told when . . .’
(14) Sie werden geholfen . . . (HG Ihnen wird geholfen) ‘They are helped . . .’
(15) Er wurde erzählt . . . (HG Ihm wurde erzählt) ‘He was told . . .’
Also very frequent is a structure replacing the genitive case which has a close
parallel in Afrikaans – but also in some Low German dialects (Russ 1990: 13
for Frisian, 1990: 43 for North Saxon):
(16) in Kaiser seinem Drama (HG in Kaisers Drama) ‘in Kaiser’s play’; com-
pare Afr., in Kaiser se drama.
(17) Die Mutter ihre zweite schwere Sünde (HG Die zweite schwere Sünde der
Mutter) ‘The mother’s second great sin’; compare Afr., Die moeder se
tweede groot skuld.
It will have become clear that the Low German dialects originally spoken by
the settlers continue to exercise a perhaps somewhat unperceived influence,
in spite of the shift in articulation to High German. This influence has been
facilitated and doubtless reinforced by the new linguistic context, dominated as
it is by two other closely related Germanic languages. A detailed investigation
and comparison with Low German dialects would be of great interest. However,
the impression created by SAG is rather one of attrition and uncertainty than
of an emerging new linguistic system, and this can surely be ascribed in the
main to the often overwhelming influence of English and Afrikaans. In this
regard much research remains to be done, especially into the phonology and
158 E. de Kadt
5 FUTURE PROSPECTS
What are the prospects for the future of German in KwaZulu-Natal (see also
de Kadt 1998a)? In the urban areas, assimilation seems likely fairly soon. Here
the community backing essential to the maintenance of the language is largely
lacking. Links with the church tend to be more tenuous, and intermarriage
with the wider community is very frequent. There are substantial numbers of
post-Second World War immigrants, who, in the aftermath of Nazism, have
been less eager to cling to their German identity, and who similarly have found
it difficult to identify with the earlier settlers who have had no direct experi-
ence of developments in Germany in the late nineteenth and twentieth centuries.
The present generation of young German-speaking adults is therefore challeng-
ing the rigid value system and conservative upbringing associated with local
German to a far greater extent than before, and seems to be more accepting
of the prospect of language shift. Within the rural communities too it seems
likely that, in spite of determined resistance, assimilation will take place in
the not-too-distant future. The previous success in language maintenance has
been largely a function of the economic isolation of these communities. First,
isolation forced the settlements to be economically self-sufficient, and placed
no limits on economic growth, which meant that they could expand to include
the following generations. Second, the only challenge to the German culture
was that of Zulu culture, which was not in a dominant position. Third, this fur-
ther underpinned the dominance of the church: the pastor, as the only educated
person in the community, was regarded as the source not only of learning (and
correct German) but also of moral principles.
Although the present rural German speakers still take considerable pride in
being ‘different’, economic necessity is forcing changes on these communi-
ties. Economic self-sufficiency and further expansion are no longer possible,
with the result that the traditional way of life is increasingly being challenged:
although many of the sons stay on the land and ‘uphold the tradition of their
fathers’, increasingly the daughters are training for professions, marrying out of
the community and moving to the towns. Such contact with the urban areas
and their dominant culture(s) inevitably poses a challenge to rural German
culture. This, in turn, cannot fail to affect the position of the church, which,
although still powerful, is perhaps beginning to adopt something more of a
social role than a purely religious one. Such changes are clearly reflected in the
Vryheid community, for example, where the (traditionalist) decision to build a
German hostel was by no means a unanimous one. Some community members
argued that German speakers should not be shutting themselves off from other
German speakers in South Africa 159
South Africans in this way, lest the children find it difficult to adjust to their
larger community in adult life. One cannot escape the conclusion that even the
most determined proponents of German maintenance are swimming against the
tide and it is doubtful whether they will be able to hold out in the face of the
wide-ranging structural changes now facing our country. In short: the further
maintenance of German is directly linked to the degree of closure in the com-
munity; and it is the more progressive groupings that are most likely to lose
German first.
notes
I would like to thank representatives of the German-speaking communities who have
provided me with information, the embassy of the Federal Republic of Germany in
Pretoria, and especially Prof. J. Fedderke of the department of economics, University
of Natal in Pietermaritzburg, and Mrs. G. Strauss of Durban.
1 Figures in brackets denote the number of parish members in these communities.
2 The perceived significance of these settlers is indicated by a recent publication, a
special issue of Lantern, a journal under the patronage of the Directorate of Cultural
Affairs of the Department of National Education. In the context of the German Festival
Year 1992, commemorating the German settlers, the issue is devoted to the ‘German
contribution to the development of South Africa’, and includes a message from the
state president. Details as to many of the individual settlers and communities can be
found in this volume.
3 There has also been considerable research undertaken on German in Namibia: while
Schlengemann (1928–9), Nöckler (1963) and Gretschel (1984) concentrate on vocab-
ulary, Kleinz (1981) investigates the various functions of the three Germanic languages
in what was then South West Africa.
bibliography
de Kadt, E. 1998a. ‘Die deutsche Muttersprache in Südafrika – gegenwärtiger Bestand
und Zukunftsperspektiven’. Muttersprache, 108, 1: 1–14.
2000. ‘“In with heart and soul”: the German-speakers of Wartburg’. International
Journal of the Sociology of Language, 144: 69–93.
Forsythe, D. 1989. ‘German identity and the problems of history’. In E. Tonkin,
M. McDonald and M. Chapman (eds.), History and Ethnicity. London and
New York: Routledge, pp. 137–56.
Gretschel, H.-V. 1984. ‘Südwester Deutsch – eine kritische Bilanz’. Logos, 4, 2: 38–44.
Grüner, R. 1979. ‘Brauchtum und Schulunterricht in deutschen Siedlungen mit
besonderer Berücksichtigung der Verhältnisse in Kroondal bei Rustenburg’. In
L. Auburger and H. Kloss (eds.), Deutsche Sprachkontakte im Übersee. Tübingen:
Günter Narr, pp. 15–40.
Kleinz, N. 1981. Die drei germanischen Sprachen Südwestafrikas – Politische und sozi-
ologische Gesichtspunkte ihrer Lage und Entwicklung. Wiesbaden: Steiner.
Lantern 1992. Special issue: The German Contribution to the Development of
South Africa. February 1992. Pretoria: Foundation for Education, Science and
Technology.
160 E. de Kadt
R. Mesthrie
1 HISTORICAL BACKGROUND
Indian languages have existed in large numbers in South Africa, chiefly in the
province of Natal, since 1860. Their existence in this country is ultimately a con-
sequence of the abolition of slavery in the European colonies. Colonial planters
in many parts of the world looked to migrant labour from Asian countries to
fill the gap caused by the understandable reluctance of slaves to remain on
the plantations once they were legally free. The British-administered Indian
government permitted the recruiting of labourers to a variety of colonial terri-
tories. This resulted in a great movement of hundreds of thousands of Indian
labourers, first to Mauritius (1834), then British Guyana (1838), Jamaica and
Trinidad (1844), and subsequently to various other West Indian islands, Natal,
Suriname and Fiji. Although Natal was a new colony that had not employed
slave labour, the policy of consigning the indigenous, mainly Zulu-speaking
population to ‘reserves’ created a demand for Indian labour on the sugar, tea
and coffee plantations (see further Bhana and Brain 1990: 23–4). Just over
150,000 workers came to Natal on indentured contracts between 1860 and
1911. A large majority chose to stay on in South Africa on expiry of their
five- or ten-year contracts.
The languages spoken by the indentured workers were as follows:
(1) From the south of India chiefly Tamil and Telugu and, in small numbers,
Malayalam and Kannada. The latter two languages did not have sufficiently
large numbers of speakers to survive beyond a generation in South Africa.
(2) From the north of India a variety of Indo-European languages including
Bhojpuri, Awadhi, Magahi, Kanauji, Bengali, Rajasthani, Braj, etc. These
are related in varying degrees to Hindi, the main official language of India
since independence in 1947. These dialects coalesced to form one South
African vernacular, usually termed ‘Hindi’.
(3) A small number of Muslims among the indentured labourers (about 10 per
cent among North Indians and slightly fewer among South Indians) would
have spoken the village language of their area as well as varieties of Urdu.
161
162 R. Mesthrie
Burushaski
KASHMIRI
Do
gr
TO
i
SH
Pa
Majhi
PU
ha
ri
PANJABI
Hi
nd
us Lepcha
Brahui tan E
i Aka ES
BALUCHI ari Bhota Dafia M
Naw A
WE
R
Awadhi Nepali SS
A
A
H r a j Naga
STE
E AS
J A
Maithili
(BHO
i Garo Khasi
BIH PURI)
ar
RN
SINDHI
TE R
arw H I N D I
S
ARI
M
J
Meithei
HIN
T H
ra
NH
pa
BENGALI Ti
DI
IN
A N
Ku Mundari
DI
ru
ki
GUJARATI
I
Korku
Santali
ili
Bh
ORIYA
Kolami Parji Kui
Gondi
MARATHI
Ko
ya
TELUGU
KO
KANNADA
NK
AN
I
Tu
lu
Kodagu
MA
TAMIL
LAY
ALA
Indo-European Family
Tibeto-Burmese Family
Dravidian Family
Munda Family
2. Languages
Languages are shown thus: H I N D I
3. Dialects
Dialect groups are shown thus: WESTERN HINDI (at a slant)
Literary dialects are shown thus: A w a d h i
The Calcutta man told me 1/- would be deducted from my wages for the sheet being
torn – and I said ‘Sooga wina manga’ [=‘Get away, you’re lying’] and went away to my
work – this was about four o’clock in the afternoon.
I did not use the words ‘Sooga wina manga’ to the mistress, but she mistook me, and
she gave me ten cuts with a riding whip.
Two uses of the pidgin can be inferred from this. First, the plaintiff (a South
Indian) claimed to have used Fanakalo in communicating with a man from
Calcutta (i.e. a North Indian); second, the English mistress must have been
accustomed to being addressed in Fanakalo since she took the sentence to be
aimed directly at her.
However, Fanakalo was not the only lingua franca in use. In the plantation
barracks some bilingualism developed between Hindi/Bhojpuri and Tamil
speakers. We would expect this to have been more common among second-
generation Indian South Africans. Mahatma Gandhi (who played a central role
in South African Indian politics between 1893 and 1913) argued that ‘almost all
164 R. Mesthrie
more viable than Tamil or Hindi for careers in teaching, the civil service and so
on. Almost overnight in the 1960s Indian teachers of subjects such as history
and geography were pulled out to be retrained as teachers of a language they
did not understand. My informal interviews with hundreds of Indian pupils and
parents leave little doubt that Afrikaans has been the least popular and most
inaccessible subject in Natal Indian schools in the last three decades.
Another difficulty was that no one Indian language could serve as a language
of integration within the evolving community. The dominance of Hindi, the chief
official language of post-independence India, would not have been acceptable
to the large South Indian community, any more than Tamil would have been
to the North Indians, or Urdu to the Hindus and Christians. That is to say,
although there was a fair amount of multilingualism in Indian languages it
was at a functional level. No one language could come to symbolise unity or
integration in the way that Hindi has among Indian Fijians. English was in the
end able to fulfil this role of ‘horizontal’ communication as well as of ‘vertical’
communication with the ruling class of colonial Natal. Even today pride in
one’s ancestral language can very easily be mistaken for overzealous allegiance
to one sub-group within the larger Indian community.
Despite these difficulties Indian languages were well maintained up to the
1960s. The census figures for 1960 for these languages record the highest ever
returns in South Africa:1
The suggestion of a dramatic decline between 1970 and 1980 is not quite
accurate: the process was much more gradual than the figures suggest, with the
real turning point being the (early) 1960s, rather than the 1970s. The question
posed by the census – ‘What is your home language’ – is not a clear (or useful)
one in a community whose linguistic norms are changing. The figures for 1970
are probably too high for L1 usage, or the figures for 1980 onwards should be
at least doubled if we wish to include those who still have an Indian language
as second language. The picture in the 1990s is also not entirely as hopeless as
the figures suggest, since we must again include people with second-language
competence or the ability to understand an Indian language at least. Symbolic
attachments to the Indian languages as well as passive interaction in terms of
watching films, listening to songs, performing prayers and so on are aspects
that the census figures do not reflect. Another positive consideration is the
166 R. Mesthrie
The rest of this chapter will survey some salient characteristics of one of the lan-
guages, Bhojpuri, stressing South African sociolinguistic developments. Infor-
mation on Gujarati, Telugu and Urdu may be found in Desai (1998), Prabhakaran
(1991) and Aziz (1988) respectively. Konkani and Meman have yet to be
studied. In order to appreciate the development of a distinctly South African
variety of this language we must picture shiploads of people coming from a
vast geographical territory, stretching from the Bengal coast on the east to well
into north-central and even north-west India. In this area a number of languages
exist, the best-known being Bengali, Bhojpuri, Awadhi, Braj, Hindi, Panjabi,
Rajasthani and Kashmiri; many other languages and dialects of these languages
can also be added to the list. We are in the fortunate position of having reasonably
detailed records of all indentured workers, concerning their castes and places
of origin. We are also fortunate that at the period of indentured immigration
Sir George Grierson was undertaking his eleven-volume Linguistic Survey of
India, with notes, skeleton grammars and detailed speech samples of village
speech throughout North India. One consistent failing of commentators on
South African Hindi (Bhojpuri) was to compare it with standard Hindi of Delhi
and other prestige centres. The historical records show that a more accurate
procedure would be to compare South African Bhojpuri with its antecedents
in village speech in north-east India, the crucial districts being Basti, Gonda,
Azamgarh, Gazipur, Sultanpur, Fyzabad, Patna, Gaya, Allahabad and Rae Bareli
and Lucknow. These districts are part of today’s provinces of Uttar Pradesh
and Bihar, and have Awadhi and Bhojpuri as their main vernacular languages,
together with Hindi as supra-regional language. Map 8.2, with its display of
the languages involved and the percentages of immigrants per district, clearly
indicates the diffuseness (heterogeneity) of the linguistic situation as people
mingled together at the port depots, on board ship and in the plantations of Natal.
It is not surprising that a ‘common denominator’ speech form arose in Natal
among North Indian immigrants. This process might be termed ‘koineisation’ –
the development of a new dialect from existing dialects of a language and/or
other closely related languages. Some features of present-day South African
Indian languages in South Africa 167
PAHARI PERCENTAGE
TEHRI More than 5 BASTI, GONDA, AZAMGARH
GARHWAL
4−5 GAZIPUR, SULTANPUR, FYZABAD
KHARIBOLI GARHWAL
SHARANPUR 3−4 PATNA, GAYA, ALLAHABAH, RAI BARELI, LUCKNOW
UR
DA
MP BALLIA, HARDOI
T
LY HI
A
RA
Less than 1
MOR
IB
BULAND-
RE
PIL
SHAHR
AWADHI
BA
UR
NP KHERI UP − Bihar border
BRAJ HA
BA
JA
M AH
HR
A FA SH
AIC
TH
R
BHOJPURI
UR
N
RA
H
HARDOI
A
KH
BARA-
AD
CK
ET BASTI
AM
AW BANKI
NO
PA
A FYZ GORAKHPUR
W
AB
R
UNAO
H
AD
AN
SU
LTA
KANAUJI KANPUR RAE NP
UR
G
A
N
BARELI
HA
SARAN
JALAUN FA AZAMGARH B MAITHILI
RB
TE ALL
HP PARTABGARH IA
DA
HAMIRPUR UR
AL JAUNPUR GHAZIPUR
JHANSI LA MONGHYR
BANDA H D PATNA
BUNDELI
BANARAS BA
AB
A
AH
AD
MIRZAPUR SH GAYA
0 100 200
Kilometres
BAGHELI HAZARIBAGH
PALAMAU
NOT INDICATED RAIPUR, ARRAH, JAIPUR
MAGAHI
Based on SIEGEL 1987: Pg. 144 CHATISGARHI
8.2 Areas of origin of North Indian immigrants to Natal, and principal dialects
Bhojpuri which are from originally different source languages are outlined
below.
(i) Features from Bihari dialects alone
(a) The past tense endings in -l, e.g. ham laut.ailī, ‘I returned’ (see further
2.1 below).
(b) (Optional) plural marker -jā, as in ham log dekhli-jā, ‘we saw’.
(c) Obligation construction with dative particle ke after the subject, the
main verb in stem + -e (i.e. infinitive form) followed by ke again, plus
an auxiliary verb expressing obligation:
(1) chokrı̄ ke cāı̄ banāwe ke par.ı̄.
girl dat tea make-caus-inf dat fall.3sg.fut
‘The girl will have to make tea.’
(d) Emphatic construction with verbal noun in -be, plus verb kar: This
use of the oblique form of the verbal noun, coupled with the verb kar,
‘to do’, places emphasis on the agent’s intentions or actions (in contrast
to the usual indicative form of the verb).
(2) tab ham bollī nei – ham jai-be karab.
then I say.1sg.past no I go-vn do.1sg.fut
‘Then I said, “No, I will go.”’
(ii) Features from eastern Hindi dialects alone
(a) Third person singular past tense ending -is (e.g. dekhis ‘she saw’).
(b) Third person plural past tense ending -in (e.g. dekhin, ‘they saw’).
168 R. Mesthrie
(the easterly varieties having -li instead). On the other hand, the third per-
son form seems to derive from the easterly varieties of Bhojpuri, (the westerly
varieties having -lai here). The other Bihari varieties, while sharing -l endings
with Bhojpuri and SABh, have different vowels following the -l.
learnt the form from a parent born in India. Among these idiolectal forms are
present participles in -it, rather than the usual -at; second person future endings
in -bā, rather than the usual -be; third person transitive past endings in -le or -lis,
rather than the usual -las or -lak, and the use of the endings -wā for the third
person singular of past intransitive verbs. All of these forms show the marginal-
isation of non-Bhojpurian features in the coastal SABh dialect.
One pair of variants form a notable exception in that they occur equally
frequently in coastal SABh in apparent free variation. These are the third person
past transitive marker -las (from Bhojpuri) and -lak (its equivalent in Magahi
and Maithili and some Bhojpuri dialects bordering upon them). Speakers are
not sensitive to the difference in the phonological form of these items; that is,
they are not indexical of social meanings.
does not occur in SABh, being replaced by a lesser used (but less irregular)
alternate of Indian Bhojpuri and Awadhi, ho-.
In this section I will examine borrowings and other neologisms in South African
Bhojpuri, stressing how the linguistic practices of a group of people can serve
as antennae to aspects of their social history. The word for ‘indenture’ still used
among older speakers is girmit., based on the English word agreement, with
an agentive noun girmit.yā, ‘one who signed a girmit., an indentured worker’.
Other loanwords referring to events occurring prior to departure include depo
(from English depot) for the building in which immigrants were housed while
awaiting the next ship to Natal. Terms for ‘recruiters’ speak volumes for the
unethical practices of these Indians in the employ of the British: thagwā and
luterā. These are equivalent to ‘thug’ and ‘looter’ respectively. It is ironic that
these two terms (originally from Hindi), which the British picked up in India,
should be used in connection with the practices of those serving their interests.
Many historians (e.g. Tinker 1974: 122) confirm the unscrupulous practices of
the recruiters, their false tales and occasional kidnapping of reluctant village
folk.
Another neologism of this period is Kalkatyā, which signifies ‘one who
embarked ship at Calcutta’, (not ‘a native of Calcutta’). This term stresses the
importance of the port of Calcutta in re-shaping the lives of the original migrants
and their very identity. They referred to the new form of speech (the koine) as
Kalkatyā bāt (‘Calcutta language’). In the same vein there arose new kinship
terms such as jahajı̄ bhāı̄, ‘ship brother’ and jahājı̄ bahin, ‘ship sister’ denoting
the special bonds that arose between those who travelled on the same ship. I
have oral evidence that this relationship was treated for a while as a true blood
relationship, with marriage between immediate descendants of ship brothers
and/or sisters being discouraged.
Early loanwords from languages of Natal are extremely interesting in that
they capture something of the mental struggle to become familiar with the
new environment, its peoples, languages and customs. Thus āfkaran became
the word for ‘twenty-five pence’ (from half-a-crown); d.amolā the word for
‘sugar mill’ (from Mauritian Bhojpuri spoken by some indentured workers
and plantation owners in Natal, ultimately based on Creole dã mulẽ, from
French dans le moulin). South African place names proved tongue-twisters
to the first generation, who modified them to suit the phonological and some-
times semantic structure of their own language. Thus the Afrikaans suffix -burg,
‘town’, seems to have been identified with the Bhojpuri word for garden, bāg
(which is also a place-name suffix in India). Hence ‘Johannesburg’ became
Jobāg (‘Joe’s garden’?) while ‘Pietermaritzburg’ became Mirichbāg, literally
172 R. Mesthrie
‘garden of chillies’, which one might want to link with the persistent myth
presented to immigrants that Natal was a fabled land in which money grew
on chilli-trees. A few words from Fanakalo have passed into Bhojpuri, notably
bagāsha, ‘to visit’ (ultimately from Zulu ukuvakashela). Loanwords from other
Indian languages encountered for the first time in Natal are not very com-
mon, apart from some food terms from Tamil (e.g. polī, a savoury pie stuffed
with coconut and fried in oil). There are very few loanwords from Gujarati,
perhaps reflecting the class distinction between trading class and indentured
workers.
The language that has influenced South African Bhojpuri the most is English.
From the earliest times English words were incorporated into the language,
often out of necessity as is the case with girmit.. However, they were not nu-
merous, and were adapted to the phonological structure of Bhojpuri: e.g. pilā˜ k,
‘wood, plank’, shows the breaking up of the pl cluster of English, changing the
[æ] vowel (non-existent in Bhojpuri) to a nasalised [ã:] with deletion of the
nasal consonant. Since the 1950s, however, with increasing English–Bhojpuri
bilingualism the prestige of English was the cause of a flood – almost a tor-
rent – of loanwords, often ousting native words and phrases, and considerably
affecting the phonological system of Bhojpuri.
However, one should not be too dismissive of English loans in Bhojpuri or
any other local language. As the article by Branford and Claughton (chap. 10,
this volume) shows, borrowing is an essential ingredient in lexical growth and
adaptation. In addition to being overtly influenced by English, Bhojpuri has
undergone other internal changes. Of particular interest is the linguistic change
contingent upon social change. With the early collapse of the highly stratified
caste system among indentured workers, many words denoting caste occu-
pations have become archaisms or been lost altogether. Thus terms such as
dusadh, ‘corpse bearer’, d.om, ‘a type of out-caste’ and kamangar, ‘bow maker’,
are unknown to South African Bhojpuri speakers. In addition, some terms that
in India still denote a low caste or an out-caste have different semantic import
in Natal, no longer denoting a particular social group but rather stereotypic or
derogatory characteristics associated by some with those who used to belong to
those groups. One example (among five) is the term can.d.āl, which in India still
denotes a particular out-caste group. In South Africa it has become a swear-
word, an epithet for a ‘good-for-nothing’, ‘an upstart’, etc. Even the word for
‘caste’ itself, jāt, seems to me to have undergone subtle change of meaning in
actual usage to denote ‘one’s nature’ (especially in a derogatory sense). The
common phrase Okar jāt oise he, which historically and literally means ‘That’s
characteristic of his/her caste’ in effect usually conveys ‘S/he’s like that, that’s
his/her way.’
As an example of a semantic field that has been particularly susceptible to
vocabulary loss I shall illustrate the sphere of ploughing. In Bihar and Uttar
Indian languages in South Africa 173
Pradesh the word for ‘to plough’ is har jot, with many dialect variants. There
are different words for the first ploughing (pahil cas), the second ploughing
(dokhar), the third ploughing (tekhar) and so on. The ploughing of millet when
it is a foot high is known as bidah, while the ploughing of a rice field after it has
been flooded is called leo. There are separate phrases for ‘to plough with a new
plough’ (nawthā ke jot) and for ‘to lightly replough in order to clear weeds and
cover the seed’ (unah). There are special terms for ‘cross ploughing’, ‘ploughing
in diminishing circles’, ‘ploughing in progressively larger circles’, ‘ploughing
diagonally’, ‘ploughing breadthwise’, and special terms for concepts such as
‘the centre plot in the middle round which the bullocks have no room to turn’,
‘small pieces of a field which a plough has not touched’, etc. In South Africa,
with the rapid shift away from a village-based agricultural economy such spe-
cialised terms do not seem to have lasted beyond the first generation of immi-
grants. Only the general term for ‘to plough’ was known to informants that I
questioned. The language has, instead, had to adapt to a different technology,
with not a little help from English. Today you hear even home-bound, elderly
persons saying in connection with automobile travel: on kar or swı̄c on kar,
‘start’; of kar, ‘switch off’; pāk kar, ‘park’, mot.ar jek karat he, ‘the car is
jerking’, etc.3
The term ‘language shift’ denotes the gradual replacement of one language by
another as the common means of communication within a community. This is
undoubtedly happening within the Tamil, Telugu and Bhojpuri (Hindi) commu-
nities of South Africa. Initially English was used in formal domains (education
and public speaking) but gradually entered into informal domains such as the
neighbourhood and home. The shrinkage of domains in the course of shift
is paralleled by receding generational competence in the outgoing language.
In her pioneering study of shift from Scots Gaelic to English, Dorian (1981)
characterised four levels of competence, ranging from full command of the
outgoing language to zero command. In between these are the competences
of young fluent speakers, semi-speakers and passive bilinguals. In Dorian’s
scheme ‘young fluent speakers’ are those who have native command of the an-
cestral language, but who show subtle deviations from the fluent older speakers’
norms. ‘Passive bilinguals’ have full understanding of the ancestral language,
but are unable to use it in productive speech. ‘Semi-speakers’ are those who
have had insufficient exposure to the ancestral language, but continue using it
in an imperfect way some of the time, out of a high degree of language loyalty.
Dorian characterises the semi-speakers of Gaelic in East Sutherland in Scotland
as having relatively halting delivery, speaking in short bursts and exhibiting
linguistic deviations, of which older speakers are mostly aware. On the other
174 R. Mesthrie
hand, they are able to build sentences and alter them productively, a trait which
distinguishes them from the passive bilinguals.
In my fieldwork on Bhojpuri in Natal in the early-to-mid-1980s, all four
types of speakers were found. There was the difference that semi-speakers of
Bhojpuri did not converse with each other (except in jest); they usually used
the language out of necessity in communicating with those elders who lacked
a command of English. In Mesthrie (1991: 202–39) I characterise the unstable
competence of such semi-speakers. I will confine myself to two lexical examples
here to illustrate the effects of the narrowing of the range of contexts in which the
language is used. The phrasal verb lapet. kar- in older fluent-speaker speech has
the general sense of ‘to wrap, to roll (transitive), to entangle’. The only meaning
I could extract from semi-speakers was ‘to make a sandwich out of roti [round,
flat unleavened bread] and curry’. They did not think that the word could be
used in any other sense, as in ‘to get entangled in a fight’. This restriction of
meaning is clearly due to the domestication of the language. Likewise the word
naksān, which in older fluent speech means ‘wastage’ (of energy, life, food, etc.),
has been restricted in semi-speaker competence to refer solely to the wastage
of food.
5 CONCLUSION
Although the outlook for Indian languages as spoken idioms seems bleak, I
believe that they should continue to be fostered by schools, temples and private
organisations with state funding where possible. Individual spoken competences
might vary, but many people hold the Indian languages in great esteem for
cultural and religious purposes. This was clear not just for Indian but also for
other Asian heritage languages generally (Chinese and Malay) in the report
of the sub-committee on heritage languages to the government (LANGTAG
1996). With the end of the apartheid era there are now new ties with India
(the first country to have imposed sanctions against South Africa – in 1947),
which opened up possibilities of cultural and linguistic renewal in small ways.
Prabhakaran (1998) has undertaken an interesting initial study of the interaction
between people belonging to the South African Telugu community and more
recent Telugu-speaking immigrants from India. In his book Reversing Language
Shift Fishman (1991: 35) puts the case for community languages succinctly:
RLS [reversing language shift] appeals to many because it is part of the process of
re-establishing local options, local control, local hope and local meaning to life. It
basically reveals a humanistic and positive outlook vis-à-vis intragroup life, rather than
a mechanistic and fatalistic one. It espouses the right and ability of small cultures to
live and inform life for their own members as well as to contribute to the enrichment of
humankind as a whole.
Indian languages in South Africa 175
Regretfully for Indian languages in South Africa, the situation is much more
complex (see Mesthrie 1995). There is an ongoing renegotiation and redefi-
nition of the notion of community from within (i.e. a sense of Indian South
Africanness, rather than a narrow sub-group thereof ) as well as from without
(a sense of growing beyond apartheid as part of the larger society). Further-
more, in the new non-racial education system pupils of Indian origin are spread
more widely – but also more thinly – than before. This makes it increasingly
difficult for individual Indian languages to meet minimum required numbers
in the state schools. At tertiary level there is similar cause for concern: at the
time of going to press the Indian languages department at the University of
Durban-Westville has been shut down and staff redeployed to other tasks or
allowed to offer only basic courses in Indian languages. It is indeed tragic that
Indian languages continue to be neglected in the country.
notes
1 The 1996 census does not give figures for Indian languages.
2 Further details on koineisation can be found in Mesthrie (1991: 55–76).
3 Note that in these examples kar is not the English ‘car’, but the verb ‘to do’ which
converts other parts of speech and loans into verbs.
bibliography
Aziz, A. K. 1988. ‘An Investigation into the Factors Governing the Persistence of Urdu
as a Minority Language in South Africa’. MA thesis, University of South Africa.
Bhana, S. and J. B. Brain 1990. Setting Down Roots: Indian Migrants in South Africa,
1860–1911. Johannesburg: Witwatersrand University Press.
Cole, D. T. 1953. ‘Fanagalo and the Bantu languages of South Africa’. African Studies,
12: 1–9.
Damsteegt, T. 1988. ‘Sarnami: a living language’. In R. K. Barz and J. Siegel (eds.),
Language Transplanted: The Development of Overseas Hindi. Wiesbaden: Otto
Harrassowitz, pp. 95–120.
Desai, U. K. 1992. ‘The Gujarati Language amongst Gujarati-speaking Hindus in Natal’.
MA thesis, University of Durban-Westville.
1998. ‘Investigation of the Factors Influencing Maintenance and Shift of the Gujarati
Language in South Africa’. Ph.D. thesis, University of Durban-Westville.
Dorian, N. 1981. Language Death: The Life Cycle of a Scottish Gaelic Dialect.
Philadelphia: University of Pennsylvania Press.
Fishman, J. 1991. Reversing Language Shift. Clevedon: Multilingual Matters.
Gambhir, S. K. 1981. ‘The East Indian Speech Community in Guyana: A Sociolinguistic
Study with Special Reference to Koine-formation’. Ph.D. thesis, University of
Pennsylvania.
Gandhi, M. K. 1958–84. Collected Works of Mahatma Gandhi, 90 vols. Delhi: Govern-
ment of India.
Grierson, A. G. 1903–28. Linguistic Survey of India, 11 vols. Calcutta: Government of
India; repr. Delhi: Motilal Banarsidass, 1967.
176 R. Mesthrie
Language contact
Pidginisation, borrowing, switching and intercultural contact
9 Fanakalo: a pidgin in South Africa
Ralph Adendorff
1 INTRODUCTION
Fanakalo (also spelled ‘Fanagalo’) is an intriguing South African pidgin lan-
guage, for at least four reasons. First, its origins are uncertain, even though a
number of explanations have been proposed to account for them. Second, from
a structural point of view, the Fanakalo variety spoken on the mines in South
Africa is atypical: for instance, it exhibits a number of features that pidgins do
not typically possess. A third reason is the assumption by many that it is used
only in the mining industry. Closer examination shows that it is an interactional
resource which is employed for a range of purposes and in a range of settings.
Finally, Fanakalo conveys at least two social meanings, one pejorative, the other
positive in its associations. Because of its pejorative connotations Fanakalo is
being replaced on certain gold mines because of what it connotes, yet it is relied
on in other settings because it enables some people to express solidarity with
one another and reinforce their interpersonal relationships. These features are
sufficient reason to explore Fanakalo in some detail.
Elsewhere (Adendorff 1993) I have summarised what I see as the salient contex-
tual features, i.e. the domains, role and power relationships, racial identities and
attitudes of mind that characterise the unmarked and marked use of Fanakalo. By
‘unmarked’ I mean the conventional or predictable contexts in which Fanakalo
is used. By ‘marked’ contexts I mean those in which the use of Fanakalo is
unexpected or unconventional. As an unmarked choice we find that:
(a) Fanakalo is usually restricted to work, i.e. to non-affective domains;
(b) it is used in interactions where there is an asymmetric role and power
relationship between the participants, usually that of master–servant;
(c) the less powerful participant is black;
(d) Fanakalo is negatively evaluated by blacks – others who use it in interaction
with them are either positively disposed towards it, or else are indifferent
towards what it symbolises.
The use of Fanakalo in marked settings, by contrast, defines and reflects a rather
different dispensation as regards the balance of rights and obligations between
the interacting parties, because its use often calls into question the existing rights
and obligations. It can be used to play down asymmetry in the relationship;
indeed, rather than signalling disparities in power, it is always instrumental in
signalling solidarity.
An ethnography of Fanakalo is needed if we are to have a more detailed under-
standing of the range of settings and domains in which it is used, the functions it
fulfils and the participants who use it. In the absence of comprehensive accounts
of this kind, I refer readers to Chamber of Mines (1982), Wessels (1986), Brown
(1988) and Radise et al. (1979) for insights into the teaching and learning and
the underlying ideological agenda, as well as the use and evaluation, of Fanakalo
on the mines; Mesthrie (1989) for insights into the functions and possibly di-
minishing role of Fanakalo in the Indian community; and Ribbens and Reagan
(1991) who contextualise the role of Fanakalo in industry more generally.
4.1 Introduction
I collected the data on which the following grammatical description is based in
1978 from interviews in Fanakalo between a white training-school supervisor
at a Gauteng gold mine and three black instructors at the training school. The
instructors (mother-tongue speakers of Tsonga, Xhosa and Zulu) had worked
on the mines in various capacities for an average of twenty-five years each and
had first learnt Fanakalo on the mines. I see them therefore as good exponents
of Mine Fanakalo.
Fanakalo: a pidgin in South Africa 181
Until richer linguistic descriptions are available, I believe Fanakalo is best un-
derstood as describing a continuum of varieties which, in their typical linguistic
features, range from Zulu at one pole to South African English at the other. The
varieties draw their linguistic resources largely from Zulu and South African
English, and vary in overall lexico-semantic and morpho-syntactic complexity.
The Mine Fanakalo data, I believe, belong nearly as close to the Zulu end
of the continuum as the variety illustrated, for example, by Trapp (1908) and
partly described by Mesthrie (1989: 219). In contrast, the data which I call
Garden Fanakalo, as well as that employed in the poem ‘A Kafir Lament’ (see
section 7.1) are very different, and can be located close to the English end of
the continuum.
The grammatical description that follows is very selective, for reasons of
space. As far as syntactic structure is concerned, I shall be outlining the sim-
plest sentence structures in Fanakalo and so deal with the noun phrase (NP),
functioning as subject and direct (but not indirect) object, and the verb phrase.
I shall ignore construction markers within the NP, such as ka and na and the
genitive/possessive and associative constructions in which they function. I shall
also do no more than refer to one kind of complex sentence, the type that includes
one or more relative clauses in addition to a main clause. As regards morphology
and lexico-semantics, I shall simply list the most prominent types of morpho-
logical processes of inflection and derivation in the data, and briefly summarise
key features of the lexical data.1
In essence, the canonical order of constituents in Mine Fanakalo is subject–
verb–object. The phrase-structure patterns conform to those of English rather
than Zulu, and at the morphological level also Fanakalo is closer to English than
to Zulu. This is because what were affixes in Zulu, a language rich in affixes,
are often free forms in Fanakalo. The lexicon of Mine Fanakalo, in contrast, is
strongly Zulu based, most notably in semantic domains not linked to mining-
industrial activity. Cole (1953) estimated that 70 per cent of the Fanakalo lexicon
derives from Zulu, 24 per cent from English and 6 per cent from Afrikaans. I have
expressed misgivings about these figures (in Adendorff 1993: 24, n.1), and the
statistics provided in section 4.3 of this chapter offer further grounds for caution.
(6) Yena fundisa tina zonke lo into aikona funeka tina enza
He teaches us all the thing not wanted/desirable we do
lo yena mubi.
which they bad
‘He teaches us all the things which are undesirable that we do which are
bad.’
It is plausible for a mod2 (e.g. munye, ‘other’) to precede the head noun, into,
in (6), but such a structure is not evident in my data.
The VPs in the following sentences demonstrate the five kinds of pred-
icative types: simple verbs (7); serial verbs (8); khona (existence) (9); khona
(attribution) (10); and enza (copula) (11). ‘Predicative’ is chosen as a superor-
dinate term to subsume formally distinct constituents.
The statistics indicate that Mine Fanakalo relies heavily on Zulu for its lexical
stock, which can perhaps be taken to suggest ready access to Zulu at the time of
Fanakalo’s genesis and subsequently. The type–token ratio suggests uncommon
richness for a pidgin, and the frequencies of lo, ka, yena and lapa underscore
their grammatical significance in the language. The percentage figures also
suggest comparative norms in terms of which one might place this and other
Fanakalo data on the Zulu–South African English continuum referred to in
section 4.1.
Moreover, each has a very narrowly defined referent, for example, muzi denotes
not simply a dwelling, but specific additional surrounding features (kraal, huts,
etc.); bazal denotes both mother and father in one word; and slalo an abstract
concept: the office someone holds.
(16a) muzi homestead
bazal parents
slalo office
Verbs are listed in (16b) and, like the nouns in (16a), are also opaque. Those
in (16b) constitute a lexical set having to do with nurturing. Semantic richness
shows itself in the way that different types of nurturance are lexicalised. Thus,
lima relates to crops, and zala and kulusa to animate objects. What distinguishes
zala and kulusa semantically is that zala refers to the initial stage of rearing,
to ‘bringing into the world’, whereas kulusa refers to later stages, to ‘bringing
up’:
(16b) lima cultivate
zala father
kulusa rear
In most pidgins such semantic precision is impossible in a single lexical item.
Pidgins, typically, are analytic and their simplification takes the form, further-
more, of a restriction in the number of function words available to signal syn-
tactic relations and other grammatical information. Zulu, the principal source
of the Mine Fanakalo lexicon, is a clear example of a synthetic language being
rich, for example, in affixes. What is interesting about Fanakalo is that, while
it evinces considerable simplification and avoidance of the types of morpho-
logical processes employed in Zulu (see Cole 1953 for details), the Mine data
nevertheless indicate the use of certain inflectional and derivational processes
(see (a)–(d) below) which are not usually found in pidgins. As I indicate in
section 5, this is another feature that should oblige us to enquire into the origins
and history of Fanakalo if we are to explain the presence of these processes.
Fanakalo: a pidgin in South Africa 187
In earlier varieties of Fanakalo the prefix zo- signalled future tense. In Trapp’s
(1908) data, zo is a free form. In my data, zo- is replaced by azi as the only
marker of future tense. It, too, is a free form.
Morphological affixes in the Mine data that are less evident than those listed
so far are the following:
(e) -ini: locative adverbial (mgodini, ‘in the mine’)
(f) -ana: diminutive (nouns) (mtwana, ‘little child’)
ana: reflexive/reciprocal (verbs) (izwana, ‘hear mutually’,’understand’;
fanana, ‘resemble one another’)
(g) -ela: applied suffix (layitela, ‘light up’ in a particular place)
The following exchange took place between a white employer, J, and her black
gardener, V, neither of whom can speak the other’s language (English and Zulu
respectively). M and B were bystanders:
J: now V (8 secs) wena funa faga lo . . . rocks . . . lapa . . . okay . . . V wena buga . . .
lapa lo top lapa . . . all these . . . buga (10 secs) lapa . . . round there . . . okay . . .
yah . . . manje noka wen ai faga lo end . . . and lapa . . . alright . . . just get them up
here for me . . . wen’ can you carry them up here . . . hey?
V: (Inaudible response)
J: yah . . . you’re strong (5 secs) okay . . . let me get out the way
M: better be careful these don’t . . . [tip
J:B: [break . . . yah
J: . . . right just put those up there . . . thanks (at this point J addresses B) sand in that
little hole there . . . he could just throw them up . . . ah . . . that’s
V: . . . (whistles)
J: (increased volume) ai no (J is now addressing V) . . . no good aikona . . . wena
hamba lo side . . . tata lo side . . . round
M: what’s no . . . good? (M is speaking to J)
J: well lo’ look already this has broken off (J, here, responds to M, then addresses
V) manje now we must put more daga there . . . okay . . . come down . . . yah . . .
faga bitjane more daga . . . lo no good
M: (M is speaking to J) bit dry
J: (J responds to M) yah it is too dry . . . dad mixed it . . . I dunno who mixed it
188 R. Adendorff
J: (J addresses V). . . yah put some more there . . . you’ll have to go round . . . no
good walking up there . . . put some more in there too (shovelling sound)
key:
... noticeable pause (+0.5 seconds)
( secs) pauses exceeding 0.5 seconds – exact duration specified
[ overlapping speech
(italicised ) contextual information
Fanakalo clearly plays a part in this exchange, but it is obvious that contextual
support, in particular by way of referents (objects, locations) in the setting, is
crucial to V’s decoding of J’s message and to J’s formulation of it in the first
place. Such strong reliance on contextual support is not evident in the Mine
Fanakalo data, where speakers have richer linguistic resources with which to
verbalise their communicative intentions. Lapa (deictic ‘there’) and lo are the
highest frequency forms in the Garden Fanakalo extract, each occurring five
times. Lapa, we note, often occurs with both paralinguistic and additional ver-
bal support: ‘lapa lo top lapa’, ‘lapa . . . round there’, ‘lapa . . . just get them
up here for me’. Lo, throughout, is deictic. Wena, the second person pronoun,
labels J’s addressee (four times) and manje, used to stage instructions, occurs
twice. Okay also has a similar discourse function and is used more often than
manje. For the rest, J uses five verbs: funa, ‘want’; faga, ‘put’; buga, ‘look’;
hamba, ‘go’; and tata, ‘take’; two negators: ai and aikona, an adjective, bitjane,
‘a little’, and noka, ‘if’. Semantic richness is not a feature of the lexical items.
Most content words are superordinate terms or hyperonyms. They are seman-
tically shallow, because, unlike the evidence from Mine Fanakalo in (16), their
meaning is not specified as finely. Syntactic complexity is also not apparent.
J uses phrasal fragments for the most part and she does not inflect or modify
the basic form of the Fanakalo words in any way. Marking of time is restricted
to the present in keeping with the general contextualising of activity at the time
of speaking. It should be clear that the variety of Fanakalo under considera-
tion is very different, linguistically speaking, from that discussed previously,
and the circumstances under which it was acquired by J and in which it is
used by her in her dealings with V are very different from those of the Mine
variety.
(a) Fanakalo originated in the eastern Cape and Natal somewhere between 1820
and 1850 as a result of interaction between English-speaking settlers and
speakers of Nguni languages.
(b) Fanakalo originated in Natal in the 1860s from interaction between
indentured and trader Indians and users of Zulu and English.
(c) Fanakalo arose in Kimberley and the Witwatersrand after 1870 from
interaction between those drawn to the diamond and gold fields.
Cole himself favoured the second hypothesis, though not without misgivings.
In particular, he was concerned that his Fanakalo data showed no perceptible
influence from the Indian languages, a fact which seemed to deny the central
contribution of Indians to the creation of Fanakalo. Notwithstanding this reser-
vation, Cole’s support has been a major reason for the widespread acceptance
of the ‘Indian hypothesis’ as an explanation for the origins of Fanakalo.
was to meet the Zulu mind to mind. He was not willing to accept the ‘miserable
gibberish’ that he was offered (and which others, such as traders, accepted). As
a means to understanding Zulu culture and eliciting, recording and interpreting
Zulu nursery tales, Callaway took very active steps to learn Zulu. Such commit-
ment was not characteristic of the colonists in general. He spent a considerable
amount of time with his informants or teachers, and required that they provide
him with Zulu as they would use it among themselves.
Complementing the information Callaway provides is Alone Among the
Zulus (especially chapter 6), an account of life in Natal in the early 1850s,
written by Mrs Charles Barter, a missionary’s wife, under the nom de plume
‘A Plain Woman’. Its value lies in the fact that it offers a richer picture than
Callaway provides of social life at that time, and of the distinctive role of the
missionaries.
The writer again emphasises the missionaries’ fluency in Zulu. She herself
‘can speak Zulu correctly’ (Barter 1855: 70) and it is in Zulu that a Norwegian
missionary in the area conducted a service. What is also interesting about her
account of the service (1855: 66–7) is the composition of the congregation –
‘some of the natives from the neighbouring kraals’ as well as English traders –
and the information that the European missionaries in the area had had to
negotiate with the Zulu king the right to preach to his people. Thus, she writes:
‘On first settling in the country, the chief pastor of the mission had requested
leave from the King to teach the Truth to his subjects. He had graciously issued
a mandate to all the great men in the neighbourhood of the several stations,
ordering the people to assemble when called by the missionary.’ I quote this
by way of confirmation of the negotiated, rather than imposed, presence of
missionaries in the area, and of their facility in Zulu.
Like Callaway, Mrs Barter contrasts the linguistic abilities of the missionaries
with those of the traders and, more generally, ‘the uneducated persons, who form
the majority of colonists’ (1855: 71). She notes, for example, with regard to
the church service referred to above: ‘Our traders, being incapable from their
ignorance of the language of being themselves edified by the service, thought
it desirable to command the attendance of their native servants, which they did
in these words: “Zonke umuntu oza.”’ A further anecdote is revealing:
There are many stories current which exemplify this ignorance; but I think the best that
I have heard is that of a colonist, who meeting with a newly arrived emigrant, offered
to be his interpreter with his native workmen. They were building or fencing, and the
stranger begged his friend to desire one of the men to hand him a small axe.
‘With the greatest pleasure in life,’ said the professor: ‘Here, you! shaya me lo piccanini
bill!’ It is difficult to construe this sentence literally; but I think he meant, ‘Shy me that
infant bill-hook’ which was done immediately. The demonstrative particle lo was the
only word at all resembling anything in the language he intended to speak.
192 R. Adendorff
For Mrs Barter, as for Callaway, pidgin symbolises condescension. She differs
from him, though, on the matter of the direction of that condescension. It is the
whites, according to her, who are condescending.
Lastly, Mrs Barter makes the following comments on the Zulus’ abilities and
interest in English (1855: 70):
There is at this moment among civilized natives of Natal a great desire to obtain English
teaching for their children, and I have seen several who could read well in the Bible. I
have myself had pupils who could read and write an English diction in words of two
and three syllables with very few mistakes, and one of them writes a tolerable English
letter, and is able to read what is written in answer.
technical domains. Moreover, many of the items are notable for their
semantic richness.
(b) Fanakalo employs more productive inflectional devices (all but one are
suffixal) than appears to be usual in Eurocentric pidgins.
(c) Lo as a nominal marker (deriving its form from Zulu, but its function from
English and Afrikaans) appears both in early data and, more consistently,
in later data, e.g. (3) illustrated in section 4.2.1. In one of its functions,
illustrated in (17) below, lo marks relative clauses:
(17) Lo kuba yena lo into lo tina azi lima ka yena.
‘A hoe it is a thing rel.marker we will cultivate with it.’
(d) A frequent construction within the NP is the genitive/possessive, marked
by the free-form ka, which derives from the Zulu prefix -kwa. This is illus-
trated in (18). The same form (ka) marks instrumental adverbials. In this
second function, illustrated in (19), it derives from the bound Zulu (prefixal)
morpheme -nga:
(18) lo baba ka mina
the father of me
‘my father’
(19) Mina washa ka lo manzi
‘I wash by means of the water.’
(e) In addition, yena often marks the boundary between subject and predicate
in Fanakalo, as it does in:
(20) Lo pomp yena donsa lo manzi.
‘The pump releases the water.’
It is probably premature to argue that the presence in Fanakalo of these five
features is a consequence of missionaries’ involvement with Zulus, in other
words that the missionaries were instrumental in ‘casting the die’ of Fanakalo.
To do so would be to ignore earlier patterns of interaction in Natal (possibly
not even involving Europeans), which might for instance have influenced Zulu
foreigner-talk to the missionaries, or patterns that might have spread from else-
where, e.g. the eastern Cape. What is evident is that it is not necessary to identify
the diamond and gold fields as contexts for the initial origins of Fanakalo, since
it was clearly in use well before the diggings began in the 1870s. The hypothesis
favoured by Cole, as we saw, has also been firmly refuted by Mesthrie. That
said, evidence in support of the missionary hypothesis is scanty, but, I believe,
suggestive and worthy of further research.
of ‘A Kafir Lament’, a poem first published in 1890 and, though dated, heavily
suggestive of a persistent set of social connotations; second, on video-recorded,
spoken data; and third, on self-reported data dealing with when and why a Zulu
student’s elderly father employs Fanakalo in his dealings with his children.
I argue that Fanakalo is a widely used interactional resource, associated with
which are many shades of social meaning which are exploited in multiple ways.
I understand ‘social meaning’ in the sense of Downes (1984: 51), as ‘the set
of values which a language itself encodes or symbolises, and which its use
communicates’.
‘A Kafir Lament’
I lofe Umlungu very much, I love the white man very much,
Him much my fren’ we’ az. He’s quite a friend you know.
M’ningi promise eb’ry night, Each night he promises me a lot,
Ikona give kusas’. But next day – not a thing.
It’s always ‘Wacht een beetje, – It’s always ‘Wait a jiffie,
Hlan’ ncozana,’ then Sit down young man,’ and then
‘Footsack, suka, spuk-a-spuk! ‘Beat it, buzz off, moron!
Ngi bulala wen.’ Or I’ll murder you.’
M’ningi much sebenza, We have a lot of work,
Pesula Baas lo mine, This Baas of mine is tops,
Baas biza mina ‘Gashli!’ He just says ‘Be careful!
‘Gashli!’ eb’ry time. Careful!’ every time.
It’s always ‘Wacht een beetje, – It’s always ‘Wait a jiffie,
Hlan’ ncozana,’ then Sit down young man,’ and then
‘Footsack, suka, spuk-a-spuk! ‘Beat it, buzz off, moron!
Ngi bulala wen.’ Or I’ll murder you.’
(Butler and Mann 1979: 33–4)
Garden Fanakalo data, I suggest one would want to locate it close to the English
pole of the English–Zulu continuum. In addition, there is clear evidence that
the Fanakalo used in the poem has been crafted in order to satisfy the demands
of writing. Hence, it is somewhat of an artefact. Nevertheless, it makes use of
common Fanakalo words such as skoff, ‘food’ (and picannin, ‘child’, later in
stanza 6), lo as a nominal marker, reduction of morphological complexity (e.g.
ngi, ‘I’ and bulala, ‘kill’ as two free forms rather than being bound), and so
forth. In terms of the context in which Fanakalo is used, we notice, over and
above the detail provided earlier, that the speaker (the black person) endures
extensive humiliation. We also note the brutality and ruthlessness of the baas
(‘boss’) towards his black assistant. It is Fanakalo, largely, that encodes the
general subjugation which is described in the poem – not English or some
other language. Fanakalo, too, is clearly the language of invective and, from
the speaker’s point of view, of his distrust of the white boss. These are key
components of the social meaning of this variety of Fanakalo in the setting and,
as I show in Adendorff (1993), of its use more generally as an unmarked choice.
Where it is unrepresentative is that it suggests a degree of spitefulness that is
less apparent these days.
I shall contextualise the next piece of evidence by presenting it as a scenario.
Unlike the poem, it is evidence of Fanakalo as a marked choice.
This scenario shows Fanakalo being used by status-equal whites, for which rea-
son it is a marked choice. The significance of its use lies in the fact that Fanakalo
is quintessentially a South African phenomenon. By using it, therefore, the
speaker is signalling his common South African background and identity and, in
196 R. Adendorff
drawing attention to what they share, is conveying solidarity with his audience.
The interrogative mode of Hey wena? Ini wena buka? recalls the stereotypi-
cal asymmetry often associated with the use of Fanakalo in unmarked settings.
However, in the context in which it is uttered, such choices and connotations are
obviously inappropriate, because they presume an asymmetry unlikely to exist
between the immigrant South African in New Zealand and his white viewers
in South Africa. It is this very inappropriateness of Fanakalo, in fact, that is
the source of amusement. At face value, the sender’s questions are insults and
yet, as Labov (1972) has shown, some insults function as markers of solidar-
ity in male culture. Inasmuch as the utterances in the New Zealand scenario
constitute a form of teasing and banter, they again qualify as solidarity-seeking
behaviour.
8 CONCLUSION
Fanakalo is a code that linguists have long neglected. This is possibly because
pidgins, generally, were disregarded as topics worthy of serious consideration
until fairly recently. The unfavourable connotations of Fanakalo (as an un-
marked choice) probably also deflected academic attention from it. Nevertheless,
it holds a wealth of research opportunities for students both of South African
sociolinguistics and of linguistic theory in general.
Fanakalo: a pidgin in South Africa 197
notes
I acknowledge with gratitude the assistance of Marina Savini-Beck and Jane
Boustred, who read and critically commented on drafts of this chapter; Rodrik Wade,
who provided critical comments and gave me permission to use the Garden Fanakalo
data; Professor L. W. Lanham, who first guided my understanding of Fanakalo some
years back; and Raj Mesthrie, for his support and editorial suggestions.
1 A fuller account particularly of the syntactic characteristics of Mine Fanakalo is
provided in Adendorff (1995).
2 Recent evidence is Mesthrie (1998) which explores the sociohistorical context in
which language contact and learning took place in the eastern Cape in the first half
of the nineteenth century. In this account Mesthrie includes ‘the earliest sentence of
Fanakalo recorded in the English sources’, dated 1816, and uses it to test the hypothesis
advanced in section 6.3.
bibliography
Adendorff, R. D. 1987. ‘The origin of Fanakalo’. Unpublished paper, Bloomington:
Indiana University.
1993. ‘Ethnographic evidence of the social meaning of Fanakalo in South Africa’.
Journal of Pidgin and Creole Languages, 8, 1: 1–27.
1995. ‘A description of selected grammatical characteristics of Mine Fanakalo’. South
African Journal of Linguistics, Supplement 27: 3–18.
Barter, Mrs Charles 1855. Alone Among the Zulus. The Narrative of a Journey Through
the Zulu Country. London: Society for Promoting Christian Knowledge.
Brown, D. 1988. ‘The basements of Babylon: English literacy – and the division of
labour on the South African gold mines’. Social Dynamics, 14: 46–56.
Butler, F. G. and C. Mann 1979 (eds.). A New Book of South African Verse in English.
Oxford: Oxford University Press.
Callaway, H. 1868. Izinganekwane, Nensumansumane, Nezindaba Zabantu: Nursery
Tales, Traditions, and Histories of the Zulus. London: Trübner.
Chamber of Mines, Mine Safety Division 1982. Fest-Eyid ka lo Mayini. Johannesburg:
Chamber of Mines.
Cole, D. T. 1953. ‘Fanagalo and the Bantu languages in South Africa’. African Studies,
12: 1–9.
Downes, W. 1984. Language and Society. London: Fontana.
Gal, S. 1989. ‘Language and political economy’. Annual Review of Anthropology, 18:
345–67.
Hancock, I. F. 1977. ‘Lexical expansion within a closed system’. In B. G. Blount and
M. Sanches (eds.), Sociocultural Dimensions of Language Change. New York:
Academic Press, pp. 161–71.
Labov, W. 1972. ‘Rules for ritual insults’. In D. Sudnow (ed.), Studies in Social Inter-
action. New York: The Free Press, pp. 120–68.
Mesthrie, R. 1989. ‘The origins of Fanagalo’. Journal of Pidgin and Creole Languages,
4, 2: 211–40.
1998. ‘Words across worlds: aspects of language contact and language learning in the
eastern Cape: 1800–1850’. African Studies, 57, 1: 5–27.
198 R. Adendorff
1 INTRODUCTION
This chapter is an exploratory piece, which will focus where possible on the
motives for lexical borrowing and on the lexical fields in which borrowings take
place. It will not discuss the effects of borrowing on linguistic form. Xhosa,
English and Afrikaans have been chosen because the authors, between them,
have some working acquaintance with all three, and because they are major
languages of southern Africa with contrasting social histories.
Our point of departure will be the classical paper of Haugen (1972 [1950]) on
borrowing. Although Haugen is pleasantly clear on what borrowing is, we shall
take the liberty of offering a definition based on his, but somewhat different
in wording: ‘the adoption into one language of items, patterns and meanings
from another’. Here the term ‘adoption’, which we owe to Desmond Cole, is
important to distinguish between nonce-words, borrowed ad hoc, as in ‘ons moet
daardie fridge nou laai’ (we must now load that fridge), and words established
in a language, as commandeer (from French via Afrikaans) is now established
in English. Whether a word is really established or not can be decided only on
the basis of a respectable body of evidence of use, as in the collections of actual
contexts that form the data base for every entry in the Oxford English Dictionary
or the Dictionary of South African English on Historical Principles (DSAE
Hist). Most of us, however, make this judgement impressionistically and without
data, other than those of daily experience. But the point is basic: adoption
implies adoption into the ‘public’ language system, as opposed to an individual’s
perhaps momentary extension of a personal repertoire in code-switching.
A borrowed item typically undergoes some change of form. Bus, reconsti-
tuted as Xhosa ibhasi, has taken on a Xhosa phonetic form and acquired a
noun-class marker. However, a signification may be borrowed on its own and
‘attached’ to a word in the ‘borrowing’ language. Thus Dutch snoeck, originally
meaning ‘fresh-water pike’, acquired at the Cape an additional signification,
that of a marine fish of different species but equally impressive teeth, our own
South African snoek. And in South Africa the original Dutch signification has
199
200 W. Branford and J. S. Claughton
fallen away. Haugen offers a useful three-way distinction, which again we have
ventured to adjust:
(1) Loanwords, borrowed with sound changes only, though their grammatical
markers are likely to be those of the borrowing language. An example is
Afrikaans gogga (insect), from Khoe xoxon, with obvious sound-changes
and a normal Afrikaans plural form.
(2) Loanblends, of native origin with borrowed lexical morphemes, as in South
African English (SAE) muti-man (‘herbalist’, ‘doctor’) in which an Angli-
cised Zulu umuthi (medicine) combines with English ‘man’.
(3) Loan translations, also called ‘calques’, e.g. Afrikaans swarthout from
English blackwood, where only a meaning is borrowed but ‘the forms rep-
resenting that meaning are native’ (Appel and Muysken 1987: 15).
It might be convenient to group loanwords and loanblends together as ‘adop-
tives’ (Cole’s term), since both involve new or altered forms, as opposed to loan
translations which involve new meanings only. A question of some interest is
what determines the choice between borrowing meanings only and borrowing
forms as well.
There seems at present to be no comprehensive theory for the social pro-
cesses involved in lexical borrowing. Borrowing typically involves bilingual-
ism (Haugen 1972 [1950]: 79), but the bilingualism may be of a very slight and
feeble kind. For the processes by which a borrowed item becomes established,
some pointers may be found in Labov’s detailed hypothesis for the ‘social moti-
vation of a sound-change’ (Labov 1963; Wardhaugh 1986: 202 n.) We shall not
linger on these. Some languages appear to borrow more readily than do others.
Among speakers of the same language, some will borrow more frequently than
others and be more tolerant of borrowings.
Accounts of the motives for borrowing will probably vary according to the
writer’s general theory of language functions. Common motives for borrowing
include:
(a) extension of range of reference (Appel and Muysken 1987: 171): Xhosa
ialfabet (‘alphabet’), naming an addition to the Xhosa cultural repertoire;
(b) structural convenience, as with English numerals borrowed into Southern
Bantu languages;
(c) directive capability as in Voetsek! and hamba in SAE contexts, and in the
former Anglo-Indian repertoire of curt ‘vernacular’ commands;
(d) expressive force as in borrowed swearwords, e.g. Afrikaans donnerse (‘con-
founded’) in SAE: expressive language presents ‘its originator’s feelings
and attitudes’ (Leech 1974: 47);
(e) social solidarity, as in the SAE repertoire of greetings and other phatic items
from African languages, e.g. molweni, ‘greetings’, itself originating in an
Nguni borrowing from Dutch, and hamba kahle, ‘go well’;
Mutual lexical borrowings 201
(f) stylistic effect, e.g. the ‘local colour’ created by the scraps of Afrikaans and
other African languages in SAE fiction or the sketches of Casey Motsisi
and his successors.
The same borrowing may, of course, serve more functions than one, and it may
be difficult at times to distinguish clearly between motives. But some tentative
framework of classification is preferable to none. Many borrowings in language-
contact situations may function at least in part as signals of social solidarity, as
in Giles’ accommodation theory as mentioned by Trudgill (1986: 2), which for
our purposes can be restated thus: ‘A sender who wishes to gain a receiver’s
approval may choose vocabulary items thought acceptable to that person.’
A related remark by Trudgill (1986: 61) is also relevant: ‘Whole new lan-
guage varieties, many of them eventually spoken by millions of people, grow
and develop out of small-scale contacts between individual human beings’ –
although there must be convergences in these if parole is to become langue.
2 XHOSA
Xhosa history and social experience are clearly reflected in the various strata
of established loanwords in the Xhosa vocabulary. Most of the indigenous
languages of southern Africa, including Xhosa, have borrowed extensively from
Afrikaans and English. Xhosa, together with Zulu to a somewhat lesser extent, is
unusual in its extensive borrowings at an early stage from Khoesan languages,
principally Khoe. This reflects an early Xhosa–Khoesan symbiosis of which
tantalisingly little is at present known. There is evidence, however, from the
late eighteenth century of Xhosa marrying Khoekhoe women and of Xhosa
communities incorporating tribal units of Khoekhoe (Harinck 1969). Walker
(1928: 102) remarks upon ‘the Gunukwebes, half Xhosa half Hottentot’.
Xhosa in its earliest forms was without clicks. Most Khoe words contain
clicks, and when Xhosa borrowed words from Khoe, it took over the clicks in
them, thereby extending not only its vocabulary but its phonemic inventory too
(Lanham 1964). We can thus assume that most words containing clicks have
been borrowed from Khoe (the counter-examples are very few – see chap. 15,
this volume). Furthermore, except in obvious loanwords from Afrikaans, the
voiceless velar fricative [x], spelled rh in the current orthography as in irhamba,
‘puffadder’, also indicates a borrowing from Khoe.
Words of Khoesan origin constitute a remarkably large proportion of the
Xhosa vocabulary. In a present-day dictionary of Xhosa, approximately
one-sixth of the words begin with a click, so that it appears that at least
one-sixth of the Xhosa vocabulary is derived from Khoe. In Zulu the corre-
sponding proportion is about one-seventh (Bourquin 1935). Swati and Southern
Sotho, on the other hand, contain far fewer click words. Nearly all the click
202 W. Branford and J. S. Claughton
colonists and missionaries, and fought long and hard for their lands and inde-
pendence (Peires 1981). Another facet of this contact was the establishment of
missionary institutions, including schools, in Xhosa territory from the 1820s
onwards (Walker 1928: 185).
Thus when we turn to borrowings from Afrikaans and English we turn to a
process that began in the eighteenth century and continues to the present day.
Most of these borrowings reflect the need for words to express new concepts.
A few do not, as is shown by Afrikaans tog, borrowed as torho, ‘yet’.
Owen Lloyd (1955: 12), in a short study of 300 Afrikaans-derived words
he had collected from printed Xhosa sources, showed that ‘it is in the spheres
of church life, the law, the army, labour, trade, dress, building, farming, do-
mestic service and fight conversation that Afrikaans has influenced the Xhosa
language’. In the following examples the original Afrikaans terms occur in
brackets:
It is interesting that many of the words that Owen Lloyd cites, although widely
current in early times, are rare in the present-day language. In many cases
they have been replaced by words of native Xhosa origin, and sometimes by
English words. Thus umaneli, ‘minister’, from meneer, ‘sir’, has been replaced
by umfundisi (literally ‘teacher’; possibly a loan translation from Afrikaans
leraar). Bedesha, ‘pray’, although surviving in the Methodist Prayer Book, has
generally been replaced by thandaza. For noyisha, ‘invite’, from Afrikaans nooi,
a present-day speaker would use mema or possibly even invayita. Isoldati has
been replaced by ijoni, ‘soldier’ (pl. amajoni). English soldiers in the nineteenth
century are reported to have addressed any Xhosa men as ‘Johnny’. The Xhosa
reportedly returned the compliment by using amajoni to mean English soldiers
(R. Mesthrie, personal communication).
Some borrowings from English are definitely early, such as ititshala,
‘teacher’. The absence of aspiration suggests that this was borrowed either
via Afrikaans or via dialectal and especially Scottish missionary pronuncia-
tion. Early borrowings show the /l/ replacing post-vocalic /r/, as Xhosa orig-
inally did not have an /r/ phoneme. Compare Zulu uthishela and utisha, both
‘teacher.’
In many cases the word used in the spoken Xhosa depends on the dominant
‘Western’ language in the area. Professor Peter Mtuze tells us that when, having
grown up in Middelburg, an Afrikaans-speaking area, he visited Cradock, which
204 W. Branford and J. S. Claughton
3 AFRIKAANS
What is now ‘Afrikaans’ was in the 1860s an unstandardised language of hearth
and home, with various designations, e.g. Cape Dutch, Kaaps, the Taal, and
so on. It was the ‘L’ partner in an uncomfortable diglossic relationship with
Nederduitsch, or standard Dutch, the official language of the Reformed Church
and the Afrikaner Republics. By the mid-1920s a recreated Afrikaans had be-
come a fully standardised national language, fulfilling – up to a point – the
promise of the nineteenth-century battlecry ‘de taal is gansch het volk’: roughly,
‘the language and the people are one’ (Zietsman 1992: 196). The social forces
behind this transformation included the political drive for the establishment of
ons eie, ‘our own’. Afrikaans acted both as a symbol of the political hopes of
the Afrikaner people and as one of the instruments for their realisation (van der
Merwe 1966: 24; Zietsman 1992: 197).
Hudson (1980: 34) points out four key processes in the establishment of a
standard language:
r selection of a variety for standardisation;
r codification;
r elaboration of functions;
r acceptance.
These provide a helpful framework for this sketch of the vocabulary of Afrikaans
in its relations with other vocabularies of southern Africa. As point of entry to a
time sequence stretching from 1652 to the present day, it will be convenient to
take 1902, the year of publication of the Patriot-woordeboek, the first Afrikaans–
English bilingual dictionary.
3.1 Selection
The variety selected for standardisation was ‘Afrikaans’, rather than ‘Hollands’,
which had, of course, an already standardised variety. This is somewhat re-
markable since it is estimated that about 90 to 95 per cent of the present-day
Afrikaans vocabulary is ultimately of Dutch origin (Raidt 1976: 177; Carstens
l989: 144; P. Harteveld, personal communication). Afrikaans, however, did
not originate from contemporary or even from nineteenth-century Dutch, but
‘from the colloquial Dutch of the 17th century’ (van der Merwe 1951: 23),
much affected by its use by people of non-Dutch descent. ‘In the 18th century
there were more non-White speakers of Afrikaans than White’ (Donaldson
1991: 30). These included slaves and political prisoners from Indonesia, subju-
gated Khoekhoe, people of mixed descent and French- and German-speaking
settlers.
206 W. Branford and J. S. Claughton
Cape Dutch borrowed on a small scale from the languages of these people and
others. From Malay there came, for instance, amper, ‘almost’ (Malay hampir);
baadjie, ‘jacket’ (Malay badju); baie, ‘much, many’ (Malay banyak); and pier-
ing, ‘saucer’ (Malay piring). From Portuguese came bredie, ‘a stew’ (bredo);
kombers, ‘blanket’ (cobertas); and kraal, ‘cattle enclosure’ (curral). From
Khoesan languages came names of plants (boegoe, dagga), animals (koedoe,
kwagga), artefacts (kaross, kierie), topographical features (Karoo) and numer-
ous place names.
More commonly, however, Dutch had adjusted itself to the ‘new world’ of the
Cape by simply using words and formatives of Dutch for new referents. Thus
a species of large evergreen or semi-evergreen tree was designated essenhout,
‘ash’; the rock hyrax was called a dasje, ‘little badger’ (now dassie), Khoesan
groups living at or near the coast were designated Strandlopers, ‘beach-walkers’
and a large ox-like antelope goes to this day as wildebeest, ‘wild ox’. For other
items see J. Branford (1988).
Vocabulary, however, was a minor differentiating factor between Afrikaans
and Dutch as compared with pronunciation and grammar (du Toit 1897, cit.
Donaldson 1991: 33). President Paul Kruger, speaking in Holland in 1902, had
to warn his audience that they might not be able to understand him because
‘Ik spreek niet Hollandsch, maar Hollandsch-Afrikaansch’ (‘I am speaking not
Dutch, but Dutch-Afrikaans’) (cit. Zietsman 1992: 125).
Afrikaner scholarship at this time defined Afrikaans as a ‘white man’s lan-
guage’: as Langenhoven put it in 1914, ‘die één enigste witmans-taal wat in SA
gemaak is’ (‘the one and only white man’s language that was made in South
Africa’) (cit. Zietsman 1992: 197). Witmanstaal marginalises by implication the
Afrikaans of coloured speakers. Van Rensburg (1993: 146) distinguishes among
modern varieties between Kaapse Afrikaans, ‘Cape Afrikaans’ and Oosgren-
safrikaans, ‘Eastern Border Afrikaans’, claiming that the latter, regarded as
a ‘white’ language, is the principal basis for Standaard-Afrikaans, which is
impoverished as a result.
3.2 Codification
An important step in the codification of Afrikaans was the development of
a simplified spelling system which achieved a better fit between spelling and
pronunciation and a transformation of the appearance ‘on the page’ of thousands
of words of Dutch origin. Thus, for example, paard, ‘horse’, became perd,
schuit, ‘boat’, became skuit and vrouw, ‘woman’, became vrou.
The spelling rules as revised up to 1991 are published by the Suid-Afrikaanse
Akademie vir Wetenskap en Kuns with an official word list (hereafter SAAWK
1991), of which the first version appeared in 1917. A full-scale Woordeboek
van die Afrikaanse Taal (‘Dictionary of the Afrikaans language’), on which
Mutual lexical borrowings 207
3.4 Acceptance
Evidence for the widespread acceptance of Standaard-Afrikaans as a model,
especially as regards vocabulary, is the success of such handbooks as H. J. J. M.
van der Merwe’s Die korrekte woord (seven impressions, 1966–82) and of
Combrink and Spies’s Sakboek van regte Afrikaans (1986). Anglicisms are
frequently targeted in both of these. The market for such guides may indicate a
measure of linguistic insecurity on the part of some of their buyers.
Such insecurity is not surprising in view of the wide gap between Standaard-
Afrikaans and the language of informal exchanges in shops, offices and homes.
In these – perhaps particularly among less-educated people (Donaldson 1991) –
both established and ad hoc borrowings from English abound. Consider the
following:
(1) Hoe kan ons hulle support as hulle ons nie kan supply nie? (‘How can we
support them if they can’t supply us?’)
(2) Hulle pay mos Vrydae. (‘Friday is their pay-day.’)
(3) Jy’s gechuff met jouself met daardie. (‘You’re chuffed about that.’)
(4) Maar [Dr X] was so boring. (‘But [Dr X] was so boring.’)
In (1), a remark by a bilingual East Cape shop assistant, support and supply
are from the specialised vocabulary of retail trading, used here probably to
extend the speaker’s range of reference and, like gechuff, probably an ad hoc
borrowing. Pay in (2) is a well-established lexical item in informal ‘coloured’
Afrikaans, used possibly for structural convenience. Gechuff was overheard in
Woolworths from a well-dressed white customer, and was chosen possibly for
its expressive force. Expressive force may also explain the choice of boring, the
verdict of a distinguished bilingual journalist in a post-mortem on an upmarket
Cape Town dinner-party. Goosen (1990) captures hundreds of fictional code-
switchings from Parow railway workers in the 1950s, e.g. ‘ons gaan Aunt Mavis
se hare perm’ (‘we’re going to perm Aunt Mavis’s hair’) and ‘hulle is usherettes’
Mutual lexical borrowings 209
(‘they are usherettes’) in which the borrowings are again ‘specialised’ words,
like support and supply in (1).
Donaldson (1991: 281–4) notes the dependence of Afrikaans on English
for several swear-words, notably blerrie, ‘bloody’, and fokken, ‘fucking’. He
also notes the Afrikaans use of English farewells, among which he seems to
have missed koebaai, ‘goodbye’. Both swear-words and farewells look like
borrowings for expressive force.
Recent proposals for ‘alternative Afrikaans’ (Gerwel 1988), ‘Afrikaans and
liberation’ (Brink 1988) and ‘the democratisation of Afrikaans’ (van Rensburg
1993) are critical, directly or by implication, of Standaard-Afrikaans in its
present form. The focus of Brink and Gerwel, however, is less on linguistic
specifica than on the unhappy associations of Afrikaans with apartheid. Brink,
in particular, is concerned to point out that Afrikaans, now to some a symbol
of the oppressor, has early connotations of ‘freedom fighter’ echoing back to
Hendrik Bibault’s cry of defiance in 1707: ‘Ik ben een Afrikaander.’
In a new South Africa, Standaard-Afrikaans is likely to draw more widely for
its word-stock upon other languages of Africa. There are indicators in Woordelys
1991 (cf. SAAWK 1991), such as amalaita, that this process is already begin-
ning. But some critics of Standaard-Afrikaans seem to miss an important point.
This is that a standard variety, closely associated as it is likely to be with written
texts and formal occasions, is specialised for a particular set of contexts and
functions. Two long-established words in the colloquial ‘coloured’ Afrikaans
of George are klôgoed, ‘kids’, and klimmeid, ‘young girl’. Useful as these two
are, they are hardly items of standard Afrikaans. (Compare juleit, ‘to work’,
as in waar juleit julle? though this is perhaps Flaaitaal rather than Afrikaans
proper.) Hudson (1980: 34) points out that standard varieties are unusual be-
cause they are almost ‘pathological in their lack of diversity’. Hudson seems
to regret this, but greater diversity – such as the adoption, say, of klôgoed and
juleit – would deprive Standaard-Afrikaans of the relative homogeneity that
is the price of its necessarily public role. The key point is that any language
serving as many different communities as does Afrikaans will need not only a
standard variety but others too, some with specialised repertoires of vocabulary
beyond the standard’s domain.
4 ENGLISH
and springbok (in its ‘animal’ sense) in 1775. Lord Chesterfield had already
referred to Dr Samuel Johnson as ‘a respectable Hottentot’.
Dutch was already well established at the Cape. The British moved into a
long-established colonial society with its own language and its own powerful
traditions and dynamics. Hence early borrowings include not only the expected
words for topographical features, living creatures and artefacts, e.g. drift (1795),
springer (1797), knobkerrie (1832) and veld (1835), all from or via Dutch, but
a substantial vocabulary of social institutions and the people who ran them, e.g.
commando (1790), field-cornet, drostdy (1796) and Volksraad (1836). Nouns
with concrete senses predominate in these early borrowings, and indeed in most
later ones, and loan translations are less common than in Dutch-Afrikaans.
This points to an ‘assimilative’ tendency in English generally despite the
efforts of some English purists, so that Lady Anne Barnard was able to refer
to her husband as Mynheer de Secretarius and sign herself in 1802 – probably
with at least an inner smile – ‘Your ever-faithful Vrouw’.
South African English in the early twenty-first century is not, of course, a
single variety. Though ‘racial’ labels are unfashionable and at times misleading,
the preliminary census data of 1996 indicate that English was spoken as ‘home
language’ by about 1.71m ‘whites’, 0.97m ‘Asians’, 0.58m ‘coloureds’, and
0.11m ‘blacks’. ‘Asian’ refers usually to Indians in census returns. In addition,
English is used as a ‘second first language’ by some thousands of Africans, many
in positions of influence and power. The figures and designations are here given
to underline the point that in a population where the barriers between one social
group and another are still as formidable as they are in present-day South Africa,
a language such as English will encode not one ‘world of experience’ but many,
so that the notion of ‘South African Englishes’ is at least complementary to that
of ‘South African English’.
Thus loanwords of African origin, such as mantshingilane, ‘night watchman’,
maphepha, ‘money’ and mashonisa, ‘money-lender’, are immediately intelligi-
ble to some users of English and opaque to others. An important range of vocab-
ulary, that of Indian English, has recently been opened up for study in Rajend
Mesthrie’s Lexicon of South African Indian English, with entries ranging from
bunny-chow, ‘a take-away meal comprising curry stuffed into the hollowed-
out part of a half or quarter loaf of bread’ (Mesthrie 1992: 8) to satyagraha,
‘non-violent struggle, soul-force, passive resistance as advocated by Gandhi’.
Gandhi coined this now-international word in Durban from Sanskrit.
Until the publication of Mesthrie’s data, it appeared that about half the vo-
cabulary of the ‘South African’ components of ‘South African English’ were of
Dutch-Afrikaans origin. W. Branford (1994) cites two estimates of languages of
origin, one based on 500 items chosen at random from Pettman’s Africanderisms
(1913) and the other on 2,549 drafts in the holdings in 1988 of the DSAE Hist.
at Rhodes University. The second listing, of all items for which the dictionary
unit had at that time drafted entries, is given in table 10.1.
Mutual lexical borrowings 211
Items of Khoesan origin were unfortunately counted only for the drafts of
1988, among which they numbered 1 per cent. There is a much higher input
from Dutch-Afrikaans than from Southern Bantu languages. This reflects both
the relative social distance of black people from white and the high degree of
Afrikaans–English bilingualism in the white population since the 1930s.
As already suggested, it seems that SAE has borrowed more extensively
from other languages of Africa than has Afrikaans. The apparent motives for
borrowings into SAE are, broadly speaking, those suggested in section 1:
(a) extension of range of reference in the majority of cases, but this often
influenced by (b);
(b) social solidarity (see below);
(c) in a few cases, directive needs as in hamba (Nguni; pleasantly glossed as ‘get
you gone’), voetsak (with similar meaning, a South African Dutch contrac-
tion of voort seg ik, ‘away, I say’) and pas op (‘watch out!’ from Afrikaans);
(d) convenience, as with bakkie, ‘light delivery van’ (Afrikaans) or tollie, ‘cas-
trated bull-calf’ from Xhosa via Afrikaans;
(e) in many cases, stylistic effect.
The clearest signals of social solidarity are, in encounters across languages, the
use of ‘other-language’ greetings, courtesies and farewells, such as the greeting
sakubona, ‘we see you’ (Zulu); enkosi (Xhosa) as a word of thanks; or hamba(ni)
kahle (Nguni) and its loan translation ‘go well’.
The commonest stylistic effect is in the use of ‘localising words’ as in Thomas
Hardy’s Drummer Hodge:
His landmark is a kopje-crest
That breaks the veldt around.
Here ek sê, ‘I say’, now a common interjection in English contexts, serves both
to ‘Afrikanerise’ the notion of ‘Doornfontein’ and as a phatic signal of solidarity
with the reader, identified as a person able to respond to ek sê.
The newspaper vocabulary of political abuse of the Afrikaner establish-
ment from 1948 onwards includes some key words of Afrikaans origin, such
as baasskap, ‘domination’ (1935); kragdadig, ‘forceful’ (1949); verkramp,
‘bigoted’ (1969); verlig, ‘enlightened’, often ironic (1968); and swart gevaar,
‘black danger’ as in swart gevaar politics (1939). How long these will last into
post-apartheid South Africa remains to be seen. The fate of Bantu/Bantoe was
sketched in section 3, but black liberation movements continue to use it as a
term of abuse, notably in ‘bantu education’, which in ANC usage has forfeited
the capitals that once distinguished it.
A few items of Afrikaans, notably the obscene moer as in a moerova klap, ‘a
helluva blow’, and the moer in, ‘fed up’, have also extended the SAE repertoire
of swear-words.
The popular stereotype of ‘Sow Theffricun Innglissh’ (Malan 1972: 5) is
still in some quarters something of a laughing matter – for details see Branford
(1976: 298). Serious study of the variety dates back at least to Charles Pettman’s
remarkable Africanderisms (1913). South Africa’s place in history and the
achievements of South African writers in English from Olive Schreiner onwards
have put a number of items of South African origin into world currency, among
them apartheid, Boer, commando, trek and veldskoen, all five of them with
immediate origins in Dutch-Afrikaans, though commando goes further back,
to Spanish. Jean Branford’s Dictionary of South African English (Branford and
Branford 1991) and Penny Silva et al.’s Dictionary of South African English
on Historical Principles (1996) have brought home to some of their readers the
semantic range and ‘resourcefulness’ of the English of their country, as well
as its ‘fun’ components, and the availability of these texts has probably had
some effect on the relative frequency of ‘South Africanisms’ in the press since
their first publication.
5 CONCLUSION
languages’ for their mother-tongue speakers ‘can be a refuge away from the
manipulative impersonality associated with corporate English language acqui-
sition’. The same may hold for Afrikaans in a ‘new’ South Africa.
One aim of this chapter has been to suggest how the diversification of English
vocabularies in South Africa reflects the diversification of English-speaking
subcultures. Ndebele (1987: 2) reminds us that ‘the development of English in
many parts of the world has taken forms that have gone beyond the control of the
native speakers’. Surviving English purists may take heart from the following,
from a recent advertisement in the Eastern Province Herald:
bibliography
Appel, R. and P. Muysken 1987. Language Contact and Bilingualism. London: Edward
Arnold.
Baugh, A. C. 1959. A History of the English Language, 2nd edn. London: Routledge &
Kegan Paul.
Bourquin, W. 1935. ‘Click words that Xhosa, Zulu and Sotho have in common.’ African
Studies, 10: 59–81.
Branford, J. 1978. A Dictionary of South African English, 1st edn. Cape Town: Oxford
University Press.
1988. ‘Adam’s dilemma: a note on the early naming of kinds at the Cape’. In
E. J. Stanley, and T. F. Hoad (eds.), Words: For Robert Burchfield’s Sixty-Fifth
Birthday. Cambridge: D. S. Brewer, pp. 69–80.
Branford, J. (with W. Branford) 1991. A Dictionary of South African English, 4th edn.
Cape Town: Oxford University Press.
Branford, W. 1994. ‘English in South Africa’. In The Cambridge History of the English
Language, vol. V: English in Britain and Overseas, ed. R. Burchfield. Cambridge:
Cambridge University Press, pp. 430–96.
Brink, A. P. 1988. ‘Afrikaans en bevryding’. In van den Heever (ed.), pp. 23–41.
Carstens, W. A. M. 1989. Norme vir Afrikaans. Pretoria: Academica.
Chapman, M. 1981 (ed.). A Century of South African Poetry. Johannesburg and London:
Ad Donker.
Combrink, J. and J. Spies 1986. SARA: Sakboek van regte Afrikaans. Cape Town:
Tafelberg.
Donaldson, B. C. 1991. The Influence of English on Afrikaans. Pretoria: Academica.
1995. ‘Language contact and linguistic change: the influence of English on Afrikaans’.
In R. Mesthrie (ed.), Language and Social History: Studies in South African
Sociolinguistics. Cape Town: David Philip, pp. 222–9.
Gerwel, J. 1988. ‘Alternatiewe Afrikaans op höerskool’. In R. van den Heever
(ed.), Afrikaans en bevryding. Kasselsvlei: Cape Professional Teachers’ Union,
pp. 7–19.
214 W. Branford and J. S. Claughton
K. McCormick
1 INTRODUCTION
The deft weaving of English and Afrikaans that characterises the above extracts
is a feature of those Cape Town speech communities in which code-switching
(CS) and mixing are common2 . At times it is conscious, and the listener is
aware that the speaker is enjoying playing with the languages, juxtaposing ele-
ments from each to create a particular effect. At other times language switching
appears to be quite unconscious, with none of the participants noticing where
switches occur (as with the child talking to herself). That, of course, can happen
only if frequent switching is part of the normal way of talking in the community.
Where it is, it can become a marker of the community’s sense of identity – it has
done so in District Six. CS takes many forms and serves a variety of purposes.
In this chapter we will focus on its forms and functions in one Cape Town com-
munity, but before doing that we will briefly consider how the phenomenon is
defined, and the kinds of contexts that are conducive to its occurrence.
CS can be used as a superordinate term, broadly defined as the juxtaposition
or alternation of material from two (or more) languages or dialects. (There are
some who argue that ‘style’ is also a code and that style shifting, even within
one dialect, should be regarded as CS.) When it comes to distinguishing among
different kinds of alternation, there is disagreement among scholars as to the
criteria for the definition of sub-categories of CS, particularly with regard to
length of switched elements (see Romaine 1989: 114). For example, some
scholars would use one category for alternation involving elements of any
length, from single words to long chunks of conversational turns, but would
216
Code-switching, mixing and convergence 217
distinguish this category from lexical borrowing. Others would agree that
lexical borrowing should be kept as a separate category but would argue that
a distinction should be made between intraclausal switches and interclausal
(or intersentential) switches because the first of these is likely to be syntactically
more complex.
The main kinds of criteria used for categorising types of language alternation
are formal ones, but other factors may also be considered. Two of these are
discernible stylistic or social function and apparent level of awareness of the
code change. As the words ‘discernible’ and ‘apparent’ suggest, judgements
involving these criteria are often subjective: sometimes in the situation there
are clear signals of function and speaker’s awareness, but often there are not.
I believe that the appropriateness of categories and their criteria cannot be
separated from what one wants to achieve in using the categories. Thus, in
this chapter I shall use different combinations of social, functional and formal
criteria for categorising material involving language alternation, depending on
what I want to illuminate in the related analysis. In the last section I shall
discuss the theoretical and metatheoretical problems that are raised by attempts
to categorise and analyse phenomena involving code alternation.
For the purposes of making some generalisations about social and discourse
patterns in the District Six speech community, I shall work initially with two
categories, CS and code-mixing (CM), and I shall argue that CM as a common
practice in a speech community can result in a fairly stable mixed code. I shall
define CS in formal terms as referring to the alternation of elements longer than
one word from two languages or dialects. In functional terms CS can often be
seen to serve specific purposes or have certain stylistic or social effects. CM, on
the other hand, will be taken to refer to speech in which the alternation is of
shorter elements, often just single words. There is thus intra-phrasal mixing
of vocabulary from two (or more) languages, and there may also be evidence
of distinctive structures from one language being ‘realised’ partly or wholly in
vocabulary from the other language. Clearly learners of a new language often
go through a phase that could be characterised in this way, but there are also
speech communities whose vernacular bears these characteristics, and it is with
that kind of situation that this chapter is concerned. In such cases, code-mixing
is not simply an individual speaker’s strategy, but is a commonly used process
which can result, as it has in District Six, in a fairly stable, widely used mixed
code. Of course, if one goes back far enough in time, almost every language can
be shown to be a mixed code. But in the type of mixed code we are dealing with
here, synchronic comparisons can be made between the mixed code and the
standard (or other regional) dialects of the languages whose elements comprise
the mixed code. The level of density of mixing in mixed vernaculars such as
that of District Six is striking to contemporary speakers of one or other of the
contributing languages. It may, however, be much less apparent to speakers who
218 K. McCormick
do not also speak the standard dialects of the languages involved: they may be
unaware of which language particular words belong to. That is particularly likely
when lexical items have been phonologically and morphologically adapted to
fit the receiving language. In terms of the contrasts I wish to work with initially,
the first extract below is an example of CS, and the second of CM:
(1) My ma het nie gewerk nie, my ouma het (My mother didn’t work, my grandmother
nie gewerk nie – she was a housewife. didn’t work . . . )
(CTOHP G Hend p. 4)
(2) Ou stock is lekkers wat gechip is af. (Old stock is sweets that have been
(CTOHP P Mil p. 27) chipped off.)
The two often occur together – in other words, a speaker will shift between a
dialect of one language and a mixed code, as in the following extract:
(3) Nou daai dae daar was nie carriers (Now in those days there weren’t
gewies nie, soos nou nie. Ons het carriers like [there are] now. We went
gegaan groceries haal en they made a to fetch groceries and . . . )
lovely parcel of brown paper. (CTOHP
G Hend. p. 2)
Switching of this kind, between a mixed code and one of its constituent lan-
guages, poses interesting practical, theoretical and metatheoretical challenges.
They will be touched on in the final section of the chapter.
The extracts given above are examples of one type of CS, commonly called
‘conversational code-switching’. The alternation of languages or dialects hap-
pens within a conversation on one topic, often within one speaker’s turn, and,
as we see above, even within one sentence. In another type, called ‘situational
code-switching’, the choice and changing of codes depend on situation, topic
and interlocutors. Bilingual or bidialectal speakers choose the code that is re-
garded by the community as appropriate for particular situations (e.g. a casual
conversation about children’s behaviour) or they are influenced by their sense
of who the person is whom they are addressing (e.g. a school principal, or a
stranger who is obviously a foreigner). A speaker may, of course, choose a code
that is unexpected in the circumstances, in order to make a point about how he or
she feels about the situation or the person being addressed. For example, he or
she may make a speech at a ceremonial occasion in informal vernacular instead
of in the formal register of the standard dialect in order to indicate disappproval
of the solemnity of the occasion.
Situational CS occurs to some extent in most bilingual and bidialectal speech
communities, but members of such communities do not necessarily engage in
conversational switching. Neither does CM automatically occur as a result of a
community’s bilingualism. The pattern of social relationships that is conducive
to conversational CS is one that involves frequent contact between speakers
Code-switching, mixing and convergence 219
The earliest map showing habitation of the area that later came to be called
District Six seems to be one dated 1820 (Cape Archives Depot M5/16), but
it was in the 1830s that rapid settlement began (Warren 1985). This was the
period of slave emancipation, and many former slaves made their homes there.
So did traders, merchants, artisans, specialists in various crafts and unskilled
labourers. Street directories show that it was not only a residential area but also
one in which many different economic activities had their base. This meant that
residents would have been able to interact not only as neighbours but also as
fellow workers and as providers and recipients of various goods and services.
A socio-economic framework of this kind provides the opportunity for the de-
velopment of what Lesley Milroy describes as ‘multiplex networks’ (1980: 21).
(These are networks whose members habitually interact in a number of differ-
ent domains.) The nature of such networks is of sociolinguistic significance
because it entails the need for network members to be able to talk to one an-
other on a range of topics – thus necessitating a wide shared vocabulary, and in
different domains – which means that different registers will be used. If two or
more languages are spoken in the community, then community members either
have to be very proficient in one another’s languages, or they all have to develop
the required proficiency in one or more lingua francas. From the evidence that
has come to light so far, it seems that it was the second of these alternatives that
facilitated communication in District Six. The languages that served as lingua
francas were Dutch and, later, English.
By the time people started to settle in the area that was later to be named Dis-
trict Six, a non-standard dialect of Dutch had been in use for some time among
many of the slaves born in the Cape, their descendants and the descendants
of those blacks who had been classed as ‘free blacks’ before emancipation.
Another dialect of Dutch was spoken in the homes of free burghers – whether
220 K. McCormick
Jewish community retained only their businesses in the area but lived elsewhere.
But even before they left, Yiddish was on the wane because most of the immi-
grants’ children became English speaking. The history of other immigrant com-
munities in the area has not been as extensively reported as has that of the Jewish
community, but it seems that it was common for children to become English-
dominant even if their parents’ language continued to be used in the home.
Although English was not the home language of the majority of District Six
families, all but six of the twenty schools in the area taught through the medium
of English only. Their classrooms and playgrounds provided the environment
in which many children – including Afrikaans speakers – gained their skills
in English. By the middle of the twentieth century, few languages other than
Afrikaans and English were heard on the streets and in the homes of District Six.
Our sense of what the day-to-day social relationships were like comes mainly
from oral history interviews recorded in and after the 1970s. Old people’s
memories, and their accounts of what their parents had experienced, give us
images of the neighbourhood going back as far as the last two decades of the
nineteenth century. It seems that, in spite of the variety of their origins, reli-
gions, languages and cultures, the area’s inhabitants did constitute a community
(Nasson 1988: 13). It was not without the tensions and divisions characteristic
of most communities, but there were also strong bonds forged by living in close
proximity and sharing facilities and also by some of the strategies devised for
coping with financial difficulties. Living conditions provided fertile ground for
language contact. Were attitudes towards the languages such that they would
facilitate or inhibit linguistic convergence? To answer that question, we need to
see whether people felt that their identity was strongly tied to the distinctiveness
of their home language.
In other parts of South Africa language has been a key factor in constructing
people’s sense of group identity, their own and that of others – take, for example,
constructs such as ‘the Zulu nation’, ‘the Afrikaners’ (meaning white Afrikaans
speakers, often nationalists) and ‘the English’ (meaning English speakers of
British descent). But this kind of strong link between one language and a group’s
sense of their identity did not mark the District Six community. By the 1950s
the vast majority of English and Afrikaans speakers living in District Six did
not fall into the categories usually denoted by the terms ‘the Afrikaners’ and
‘the English’, as they did not share the classification ‘white’. They had been
marginalised, largely through social and legal processes generated by racial pre-
judice. Their racially mixed ancestry was turned into a salient factor in excluding
them from rights and privileges enjoyed by other English and Afrikaans speak-
ers who were, or claimed to be, ‘pure’ white. It is well known that a concern for
racial or ethnic purity is often accompanied by a concern for linguistic purity –
miscegenation and linguistic borrowing both being regarded as unacceptably
222 K. McCormick
contaminating (see Ross 1979). In District Six, a concern for linguistic purity
came to be seen as the province of those whites who had declared them ‘other’
and rejected and often humiliated them. This was particularly the case regard-
ing Afrikaans after the nationalist ideal of racial segregation had led to the
implementation of a series of laws which destroyed the community, forcibly
removing 30,000 people and scattering them in far-flung areas.
In this context, the local dialect of Afrikaans, the lexicon of which contained
many items originally English, grew in symbolic value. It was the language of
neighbourhood solidarity, its form clearly a product of easy contact between
different ethnic groups and thus a reminder of a valued social order that had
prevailed in District Six for more than the first century of its existence. This
order was very different from the one that was the goal of successive apartheid
governments, namely the rigid separation of ethnic or racial groups.
Only one small area of District Six, an island created by arterial roads, was
left intact. Among its factories and warehouses there were 212 homes, 2 schools,
a church, 2 community centres and 3 shops. The rest was demolished.
3 DATA GATHERING
It was this neighbourhood that provided the data on which the following analysis
of CS is based. Access to the community came through a chance meeting with
someone who served on the management committee of the Marion Institute,
one of the local community centres. The Marion Institute became the base
from which I made many other contacts. Data include fifty-two hours of tape-
recordings of interviews, meetings, families at home and pre-school children
playing and talking to one another and to me. (There are recordings of approx-
imately one hundred and forty-three speakers who come from 35 per cent of
the homes in the neighbourhood.) I did some interviewing, but most of it was
done by people who had either been to school or lived in the area and shared
the neighbourhood’s linguistic repertoire. (For a detailed account of the use of
observation and interviews, see McCormick 1989b: 31–57.)
As no sociolinguistic history of the area had been written, I examined other
kinds of historical analyses in order to get an idea of some of the factors that
influence language use: socio-economic conditions, social relationships, reli-
gious affiliations, educational institutions, cultural groups and leisure activities
(see McCormick 1989b: 61–95). I also examined primary sources such as street
rolls, church records and transcripts of oral history interviews. The range of sur-
names on street rolls gave some indication of home languages that might have
been spoken in that street. Some street rolls and church records provided infor-
mation about the employment of people whose names they recorded. Church
records such as marriage registers, which required signatures, gave a sense of
the level of basic literacy among congregants. The two oral history projects on
Code-switching, mixing and convergence 223
which I drew had been designed to give broad coverage of social history. (These
are both based at the University of Cape Town: the Cape Town Oral History
Project in the history department and the Jewish Oral Histroy Project in the
Isaac and Jessie Kaplan Centre for Jewish Studies and Research.) Neither had
a strong focus on language, but the transcripts were rich in sociolinguistically
relevant information, particularly about social networks of various kinds.
Before going on to look at CS, let us establish briefly what the codes in the
repertoire are. (Of course not everyone has equal command of all of its codes.)
The names in inverted commas are those most commonly used by members
of the speech community. Contrasting modes of referring to local English and
Afrikaans indicate that the English is not seen as a dialect but as imperfectly
learned English, whereas speakers are quite clear that their Afrikaans is a dialect
in its own right, albeit one with low prestige. The term kombuistaal, as members
of the speech community use it, covers both non-standard Afrikaans (nsA)
and switching between English and Afrikaans strings. They do not distinguish
between mixing and switching, whereas I have found that for certain purposes
I need to be able to do so. The term that I use to cover all non-standard local
usages (of English, Afrikaans, switching and mixing) is ‘the vernacular’.
Kombuistaal should, I think, be regarded as a mixed code in itself – mixing
is not just a speaker strategy. English loanwords form a high proportion of its
speakers’ vocabulary and there seems to be English influence in verb-placement
rules. But for the rest the syntax is clearly Afrikaans. Kombuistaal has been used
for a long time; thus, while the gates of its lexicon are open, its grammatical
structure is fairly stable. The case of local English is almost the reverse: as
the higher prestige language its lexicon has absorbed little from Afrikaans,
but because it is in the process of becoming a first language, its grammatical
structure is not as stable. It is commonly introduced to children by parents whose
first language is the local dialect of Afrikaans. Because of the nature of that
dialect, speakers have quite a large English vocabulary but because they have
V E R N A C U L A R
However, if people from the area are gathered together for a formal occasion,
such as a meeting, then the code that is felt to be appropriate is sE. An analysis
of tape-recordings of meetings shows that in committee meetings, at points
where heated debate arises, English may well be abandoned in favour of the
vernacular without any official sanctioning of this change of code, and very
possibly without the participants’ being aware that there has been a change. In
the extract below, which is taken from a rugby club’s committee meeting, we
see the chairperson (whose turns are the first and the last) using fairly formal
English. The members, who are heatedly debating the penalties for dagga-
smoking, break with the formality of English discussion and use the vernacular.
In attempting to establish or restore order to the discussion, the chairperson
(A) uses English, which is the accepted language of the discourse of meetings.
(The recording as a whole showed that when debate was not heated, it was
conducted in English.)
Code-switching, mixing and convergence 225
(7) Ek praat all the way net Engels met (I talk all the way only English to them.
hulle. Nou ek dink as jy vir hulle in Now I think that if you put them in
Engelse klasse sit hulle kry eerste English classes they get first privilege,
privilege, man. Because why? Jy man. Because why? You go far in
gaan ver deur die lewe nou met life now with English. You get first
Engels. Jy kry first privilege as jy privilege if you now – Look, even if you
nou – Kyk, even as jy nou ’n bruin are now a brown girl but if you speak
meisie is nou but as jy praat Engels, English then you get a job really easily.)
dan kry jy sommer gou ’n job.
(8) M: Maar nou sy was sy was ook so (But now she was she was also like
gewees but now she can’t that . . . )
anymore.
Now it makes her feel
depr essed
C: Depress, ja
M: You see
Because
C: ja
M: she like also –
like you, she likes to visit in the
afternoons
C: ja lekk er (Yes, nice
M: as sy nou if she had now
klaar
werk gedoen het in die finished
working in the
ogg ende mor nings
C: ja man yes, man
M: dan lyk sy – then she liked . . .)
O: You must
have fresh air. You must have
fresh air.
M: Then she goes
and visit this one and that one.
O: Yes. Even sy – gaat jy nie – as (Even she – if you don’t go – if I don’t
ek nie uitgaan uit my huis uit nie go out of my house then I don’t feel right.)
dan voel ek ook nie reg nie. I
must have fresh air.
A language switch may signal a shift in focus, as in the third and fourth switches
in the example given above. It can also enable the speaker to incorporate useful
set phrases or idioms from the other language:
Code-switching, mixing and convergence 227
(10) Die Slaams sê regtig, straight,seg hy, (The Muslim says straight. . . he says
hy kan nie gaat baklei vir king and he can’t go fight for king and country.
country nie. Hy moet baklei vir sy He must fight for his faith.)
geloof.
The idiomatic phrase ‘king and country’ has no Afrikaans equivalent, and it
has strong emotive associations of particular kinds of loyalties and bonds from
which he wants to dissociate himself, as the rest of the interview makes clear. In
terms of the definition I gave in the opening section of the article, the first English
element in this extract, ‘straight’, would not normally qualify for classification
as a switch because it is too short. But, as is often found in code-switches, it
has a clearly discernable rhetorical function: it is repeating the meaning of the
previous word and therefore seems to have been included for the purpose of
emphasis. It is not filling a lexical gap. I would thus regard it as an example of
the phenomena that challenge the validity of categorical classification on one
criterion alone. Creating emphasis by means of repetition in another language
is a common strategy in this community (and many others). The repeated ele-
ments can vary in length from a word to a whole sentence. Quotations can be
foregrounded by a language switch:
(11) Now when I get home I tell them (Bring now your two rand, the two
‘Bring nou julle twee randjies, julle rand [that] you pay.)
twee randjies julle pay.’
But, as is often the case when a language switch coincides with quotation,
the code chosen for the direct speech should not be assumed to be the one
that was actually used. In this case there is clearly room for uncertainty as to
which code was used in the exchange she reports because, two turns later, when
completing her account of this particular conversation, the speaker says: ‘Now I
tell them, “Don’t bring nothing to eat.”’ The function of the switch in such cases
is often simply to signal that material is being quoted and to foreground it – it
may be doing little more than serving the function that inverted commas do in
writing.
Semantic contrast and balance in syntactic structure can be highlighted by
having the two focal sections in different languages:
(12) Somige van die members praat (Some of the members speak Afrikaans . . . )
Afrikaans but the predominant
language is English.
5 CONVERGENCE
Given the nature of the social context in which English and Afrikaans have been
in contact in District Six, and given the fact that many Afrikaans-dominant
parents are speaking mainly L2 English to their children, it is not surpris-
ing that local varieties of both languages display signs of linguistic conver-
gence. The influence of English on Afrikaans is clearly present in the lexicon
of local Afrikaans, which has absorbed an enormous number of English words.
It may also have influenced some syntactic forms. The extent of the influence
of Afrikaans on English is not easy to establish. The lexicon has absorbed
relatively little, but the syntax may well have been affected by contact with
Afrikaans. Several non-standard structures parallel those of Afrikaans but, as
they are also found in several regional dialects of British English, it is not clear
what their origin was. Another possibility is that some of the features of local
English arose as an effect of the processes of second-language acquistion. We
know that, in the nineteenth century, there was a variety of home languages in
District Six, while in the twentieth century Afrikaans would have been the most
common first language. Should this lead us to ask when the distinctive features
of local English were crystallised, and thus to try to establish which features
would have come from which of the community’s home languages? Probably
not. The view that transfer of features from the learner’s home language is
what accounts for deviations from the target variety has been challenged as
being only partly true: some features will be directly traceable to the learner’s
first language but others will not. Recent research on the acquisition of both
first language and additional languages provides grounds for believing that
psycholinguistic processes are at work which lead to similarities in learners’
production of the target language, whatever their home language. Williams
(1987) discusses several features of non-native varieties of English that occur
in places with very different linguistic contexts. In South Africa some of these
features can be seen in District Six and in the English spoken in the Natal
Indian community (see Mesthrie 1992). The language backgrounds of the two
communities are very different. Williams also argues that there is a category
of similar features in L2 varieties of English that probably stems from pecu-
liarities of English itself, which learners try to regularise. Thus we see that
it is not a simple matter to account for the characteristic features of a dialect
that has arisen in a language-contact situation. Whatever their origin, however,
syntactic convergence is noticeable in District Six dialects in structures that are
still distinctive in the standard dialects. Let us look in more detail at the signs
Code-switching, mixing and convergence 229
in Afrikaans, but in most cases are also found in other dialects of English and
also in L2 varieties.
5.2.1 Afrikaans
The most striking syntactic feature of nsA distinguishing it from sA is the
violation of verb-placement rules. In sA, when the verb consists of a modal
or a tense/aspect auxiliary plus main verb, these verb components are split in
certain contexts and other elements are placed between them. In such contexts,
the first auxiliary is in second position (or third, if the clause is introduced by
some conjunctions), and the rest of the verb is in clause-final position. What
happens in nsA is that there are various kinds of rightward movement of material
normally found in between auxiliary and main verb in sA, with the consequent
concentration of verb components in second position in nsA. This means that
the word order in nsA is closer to its English counterpart than in the sA word
order. In the examples that follow, the sA version is given first, and the nsA
second. Verb and auxiliary are in bold print to highlight their positions relative
to each other and to other material in the clause or sentence.
(14a) Ons moet altwee bestudeer.
(14b) Ons moet study altwee. (We must study both.)
(15a) toe het hy na die prokureur gegaan
(15b) toe hy het gegaan na die lawyer (then he went to the lawyer)
5.2.2 English
Local English syntax has far more non-standard features than does local
Afrikaans. In the following list of distinctively non-standard morpho-syntactic
features of English, those constructions that have direct parallels in sA are
marked with (+A), those that occur in other first-language dialects of English
are marked with (+OE) and those that are very frequently found in the second-
language varieties spoken in different parts of the world are marked with
(+EL2). This marking serves to remind the reader that convergence is not
always clearly traceable to its source(s).
(a) Verb-related features
(i) Number concord
The verb ‘to be’ as both auxiliary and main verb usually has the same form
for third person singular and plural. It is the standard English third person
singular form. This is more predictable where the subject is not a pronoun
and where the verb is the copula or one of the auxiliaries ‘has’ or ‘have’.
(+A +OE +EL2)
(16) The neighbours is bringing me up to school.
Code-switching, mixing and convergence 231
Standard English rules for number concord are found to apply more often to
other verbs, but where they do not, some speakers reveal an interesting re-
versal that may, perhaps, be based on the extension of the rule for pluralising
nouns: ‘add an -s to indicate plural’. Thus we find that verbs may take a word-
final -s if the subject is plural, and have no word-final -s where the subject
is singular. (In other words they place what is the sE plural form after a singu-
lar subject and vice versa.) But this near reversal is not found in the majority
of speakers in my data: there are more instances of non-standard concord
rules applying in singular subject–verb constructions than in plural ones. (As
the second example below suggests, the plural concord pattern is unstable.)
(17) If somebody chop it then it fall down.
(18) They drink and they makes a lot of noise.
Perhaps what we see here is tension between two tendencies, the one to-
wards simplification (moving towards having one verb form for both singu-
lar and plural subjects), and the other towards regularising ways of indicat-
ing singular and plural in nouns and verbs (word-final -s for plural but not
for singular forms). It will be interesting to examine concord data from this
community twenty to thirty years hence, when today’s pre-schoolers are
raising children, having themselves had twelve years of English-medium
schooling – concord rules receive a great deal of attention in the classroom!
(ii) Tense, aspect and modal marking
Past tense is frequently indicated by using did unemphatically. (+A +OE
+EL2) This is particularly common among children but it is not exclusive
to them.
(19) He did eat his food.
It is possible that the form is created by analogy with Afrikaans, which does
not use a dummy verb in the past tense but which almost always has two
elements to mark the past tense: the participle het and the main verb pre-
fixed by ge-. Mesthrie (1999) argues that Afrikaans influence is but one of
many possible convergent influences here. He attempts to trace the origins
of this unstressed dummy do to the standard English of the late eighteenth
century, but finds only a few cases in early settler English in South Africa
that resemble present-day District Six and Cape Flats usage. He concludes
that the archaic English preaching style of the missionaries is a more likely
influence. Another contribution of Mesthrie is to observe that rather than
functioning as a grammatical marker of the past tense, unstressed do is
today used with pragmatic effect: to mark off a verb phrase as ‘salient’.
(b) Placing of adverbials
The adverbial may immediately precede the object where it cannot do so in sE.
(+A +EL2)
232 K. McCormick
Adverbs of time precede those of place instead of the reverse, which is the sE
order. (+A +EL2)
(21) I’ll go now on the bed.
Given that the two codes share lexical items and syntactic structures which
their standard counterparts do not, and given also that there is evidence of some
phonological convergence – a matter not touched on here – it is easy to see that
the boundaries between the local dialects of English and Afrikaans are far less
clear than those separating the standard dialects of the two languages. Where
there is CS between the two local dialects it is often impossible to pinpoint
the site of the switch because a word that may seem English to an outsider
is, to an insider, firmly established as an item in the local Afrikaans lexicon.
The speaker may not even know that it was originally English and that there
is another word for it in Afrikaans. Whose perspective should be used in cate-
gorising these lexical items? Different perspectives would give rise to different
analyses.
If one seeks a theoretical framework in an attempt to find reliable reference
points, one finds that there are no clear answers to the questions ‘When does
a loanword cease to be regarded as a foreign item?’ and ‘On what basis are
converging language structures accepted (by theorists) as rules in the codes that
are being forged through intensive language contact?’ (See Clyne 1987 and
Code-switching, mixing and convergence 233
McCormick 1989a). Without answers to these and related questions, one cannot
proceed with a linguistic analysis of CS that attempts to discover whether there
are syntactic constraints on intra-sentential CS. Metatheoretical issues about
perspectives on and criteria for classification need to be identified and clarified,
otherwise ad hoc decisions will continue to be made about linguistic aspects of
convergence and CS in situations of prolonged and intensive contact, such as
that in District Six.
notes
1 The initials CTOHP indicate that the tape is in the possession of the Cape Town Oral
History Project, and the accompanying initials and page numbers identify the section
of the relevant transcript. The spelling conversions used in the CTOHP transcripts have
been retained in quotations. The other transcripts, which are all from my data, do not
attempt to capture pronunciation, except in cases where it is sufficiently well known
to have been represented in literary texts dealing with the area. Here and throughout
the chapter, italics are used for Afrikaans words and utterances.
2 The research on which this chapter is based was made possible by the generous
financial grants received from the University of Cape Town and the Human Sciences
Research Council. The research would not have been possible without the friendly
co-operation of members of the speech community and without the skills of the
interviewers. My thanks to all the people and institutions involved.
bibliography
Bynon, T. 1977. Historical Linguistics. Cambridge: Cambridge University Press.
Chana, U. and S. Romaine 1984. ‘Evaluative reactions to Panjabi/English code-
switching’. Journal of Multilingual and Multicultural Development, 5, 6: 447–73.
Clyne, M. 1987. ‘Constraints on code switching: how universal are they?’ Linguistics,
2, 5: 739–64.
Gumperz, J. J. and E. Hernandez-Chavez 1972. ‘Bilingualism, bidialectalism, and class-
room interaction’. In C. Cazden, V. P. John and D. Hymes (eds.), Functions
of Language in the Classroom. New York: Columbia Teachers’ College Press,
pp. 84–108.
Gumperz, J. J. and R. Wilson 1971. ‘Convergence and creolization: a case from the
Indo-Aryan/Dravidian border in India’. In D. Hymes (ed.), Pidginization and
Creolization of Languages. Cambridge: Cambridge University Press, pp. 151–67.
McCormick, K. 1989a. ‘Unfiltered talk: a challenge to categories’. York Papers in
Linguistics, 13: 203–14.
1989b. ‘English and Afrikaans in District Six: A Sociolinguistic Study’. Ph.D. thesis,
University of Cape Town.
Mesthrie, R. 1992. English in Language Shift – the History, Structure and Socio-
linguistics of South African Indian English. Cambridge: Cambridge University
Press.
1999. ‘Fifty ways to say I do: tracing the origins of unstressed do in Cape Flats
English’. South African Journal of Linguistics, 17, 1: 58–71.
Milroy, L. 1980. Language and Social Networks. Oxford: Basil Blackwell.
234 K. McCormick
Nasson, B. 1988. ‘Oral history and the reconstruction of District Six’. Upstream
Magazine, 6, 4: 12–19.
Raidt, E. 1984. ‘Vrouetaal en taalverandering’. Tydskrif vir Geesteswetenskappe, 4, 4:
256–86.
Romaine, S. 1989. Bilingualism. Oxford: Basil Blackwell.
Ross, J. A. 1979. ‘Language and the mobilization of ethnic identity’. In H. Giles and
B. Saint-Jacques (eds.), Language and Ethnic Relations. Oxford: Pergamon Press.
Warren, D. 1985. ‘The early years of “District Six”: District Twelve in the
eighteen-forties’. Cabo 3, 4: 13–18.
Williams, J. 1987. ‘Non-native varieties of English: a special case of language acquisi-
tion’. English World-Wide, 8, 2: 161.
12 Code-switching in South African townships
1 INTRODUCTION
Language contact, including contact within the Bantu language family, is not a
new phenomenon. However, switching languages or linguistic varieties within
the same conversation, or code-switching (CS) as it is termed, is a dynamic
and growing phenomenon as contact between speakers of various languages
throughout the world continues to increase. This chapter offers an introduction
to the relationship between social history and linguistic studies on CS in South
African townships, as well as some contexts within which CS is practised in
South African townships. Thereafter, a sociohistorical overview of a variety of
perspectives on CS research will be given.
The term ‘code-switching’ will be used in this chapter to include full consti-
tuents as well as single lexemes. No distinction will be made between ‘noun
switching’ (Poplack 1981: 171), ‘nonce borrowings’ (Sankoff and Vanniarajan
1990), code-mixing (CM) (Kachru 1978) and code-switching. Myers-Scotton
(1993b: 24) notes in this regard that many CS researchers do not make clear
how they classify single lexemes. In particular they are not willing to label
as borrowed forms all singly occurring forms drawn from another lan-
guage, so ‘code-mixing’ becomes a compromise designation. The distinction
between CS and borrowing is also still unresolved. In this chapter the term
‘code-switching’ will include references to singly occurring forms as well as to
intrasentential and intersentential CS within, as well as across, conversational
turns.
235
236 S. Slabbert and R. Finlayson
Brown (1992: 71) refers to the problematic nature of ‘the relationship between
individual and social behaviour on the one hand, and language on the other’. He
argues that it is necessary to consider language, in all its modes and forms, as
a social product (Dittmar 1976). For him a critical analysis of linguistic theory
in South Africa and its practices is needed to reveal the social perspectives
and ideologies that have underpinned it. This remark is apt in the context of
African language research. Initially, interest in the African languages stemmed
from a strictly structural perspective which included phonetics, phonology,
morphology and historical comparative linguistics with a strong emphasis on
what was regarded as ‘the exotic’ and the unusual. More recently, however,
interest has shifted to the social and functional use of the African languages in
South Africa. This shift is indicative of parallel developments in social history.
For example, in the past there was a strict division of communities into
racially divided ethnic groups with a concomitant pressure for language purity.
This ideology spawned studies which frequently focused on the phenomenon of
‘borrowing’ or the ‘adopting’ of speech sounds and lexical items from the colo-
nial languages by the indigenous languages. The emphasis on the interaction
between the colonial languages and the indigenous languages in CS patterns
is furthermore a typical feature of CS studies in Africa and other post-colonial
societies.
To a large extent interest in the social and functional use of the African
languages goes hand in hand with the increase in contact between different
communities. The more intense the interlingual contact between the peoples of
South Africa has become, the more complex the exchange of linguistic items at
all levels has become. Whereas in the past language contact would have resulted
primarily in phonetic and lexical interference, today in the urban areas all
linguistic levels show the effects of contact: phonology, morphology, syntax and
discourse. Extensive CS raises on its part theoretical questions about language
change, shift and convergence, and the pragmatic issues that are associated with
each of them.
The aim of this chapter therefore is to collate, contextualise and discuss,
from a sociohistorical point of view, different perspectives evident in the var-
ious studies of CS in South African townships undertaken during the last ten
years.
Harrison et al. (1997: 43) describe ‘townships’ as ‘all those areas previously
reserved for African settlement under apartheid laws, including formal town-
ships, site-and-service areas and informal settlements’. Because of their ethnic
diversity, the townships spawned an increasing and urgent desire for people to
demonstrate both their independence and interdependence. At the same time
Code-switching in South African townships 237
and ‘urban’ varieties. ‘Deep’ refers to the older, rural and relatively ‘pure’
varieties of African languages, in contrast to their urbanised forms which have
greatly departed from those norms. Inter-ethnic communication is enhanced
by the urban varieties and CS, while use of a single ‘deep’ variety can lead
to miscomprehension. A related development in the Gauteng area is the use
of Southern Sotho as a bridge between the Nguni and Sotho language groups.
There is often a switch to Southern Sotho when an interaction between a Nguni
and a Sotho speaker occurs.
All these factors have added a further dimension to the already complex nature
of the language milieu. The following example from a conversation in Soweto,
a township in the Johannesburg area, is typical of the type of switching that
occurs (English, caps; Tswana, italics; Afrikaans, bold caps; Southern Sotho,
bold; Tsotsitaal, italic caps):
it depends gore o na le bomang. for instance, if ke na le mathoka go ya ka gore ba
KRYile re bua eng – if ba fitlha ke bua Sezulu, ba tla joina – if slang sa mothaka ele
ONS SAL ALMAL WITHI. the situation, gore o na le bo mang.
[It depends on whom you are with. For instance, if I am with my friends and they find
us speaking something – if they arrive and I am speaking Zulu, they will join in – if slang
of the friends, then we’ll all join in [the slang]. The situation is dependent upon whom
you are with.] (Joe with shebeen friends, cited in Finlayson and Slabbert 1997a: 399)
5 PERSPECTIVES ON CODE-SWITCHING
influences, such as the presence of clicks in Xhosa and Zulu, are old compared to
influences relating to contact between Southern Sintu [= Bantu] languages and the
Germanic languages of the first white settlers.
In this regard, Koopman (1994: 13, 14) quotes Delegorgue in connection with
early records of language contact dating back to 1847. He also refers to a
fascinating extract from Brownlee (1975: 65) of an event recorded some time
in the mid-1880s in which the narrator, telling a story of a soldier and a Xhosa
tracker, indicates a switch in conversation from Xhosa to English. Another
noteworthy reference to early comments on language contact made by Koopman
(1994: 18) concerns van Warmelo’s work of 1927 in which ‘he [van Warmelo]
found the number of loan-words “staggering”. I [Koopman] doubt that a scholar
of the stature of Van Warmelo has used this word naively: I am sure that he really
was surprised at the rate at which foreign words were entering Sotho.’ Koopman
(1994: 19) adds further that Cole (1990) saw ‘language mixing’ in a positive
light and as part of an enrichment and strengthening process. Cole himself
(personal communication) was averse to the term ‘loan’ word and ‘borrowing’
since, as van Warmelo (1927) also noted, it was hardly possible for a lexeme
to be on loan with the prospect of being returned! Cole accordingly preferred
the term ‘adoptive’ (cf. Khumalo 1984 and Madiba 1994). On the other hand,
‘adoptive’ has some irrelevant connotations too.
Further to the early influences of the colonial languages, Finlayson (1993:
178) suggests, in diagrammatic form, a number of contributory factors which
have led to ‘the changing face of Xhosa’. She describes specific changes in the
phonetic, phonological, morphological, semantic and literary categories that
have occurred at different stages in the evolution of spoken and written Xhosa.
She argues that these changes have further enhanced the language rather than
contaminated it.
In tracing ‘foreign’ elements, Koopman (1994) divides African-language
lexical adoptives into various semantic categories such as utensils, government,
persons, food, oxen, industry and the church. These involve semantic processes
such as broadening, narrowing and onomastic shifts. Similar arguments are
found in studies of the ‘adaptation’ of phrases and sentences into the African
languages. Thipa (1992: 88) argues that Xhosa–English CS often arises from the
result of the native speaker’s unfamiliarity with, or ignorance of, an appropriate
word. He continues as follows: ‘That then forces the native speaker, especially
a bilingual one, to resort to the language with which he seems to be most
familiar, namely English in most cases amongst the Xhosa speakers. . . . [C]ode-
switching . . . serves to express ideas with which the vocabulary of Xhosa cannot
cope or ideas which are alien to indigenous Xhosa culture.’
Khati (1992), in his study of CS between English and Southern Sotho, has
also commented on the stigma that was originally attached to the transferring of
Code-switching in South African townships 241
Two studies were commissioned to address these issues: the study ‘Critical
mass’ undertaken in 1990 and 1993 and the ‘Language study’ (van Vuuren
and Maree 1994). ‘Critical mass’ crudely measured the percentages of adults
who could understand a particular language at three competency levels by
asking respondents a question in each competency level. The ‘Language study’
developed an index of multilingualism for the black population. The forty-two
focus-group discussions that acted as the basis for the subsequent quantitative
study revealed for the first time to the advertising industry the massive extent of
CS in the urban/township communities, the functions of CS in urban/township
communication and the language attitudes of the speakers with regard to this
language use. It confirmed the fundamental differences between the standard
African languages and the urban varieties. Contrary to the STANON report,
the emphasis here was not on the deviations from the standards, but on the fact
that urban values and urban affluence are associated with these varieties. This
observation was not in accordance with the language varieties on which the
SABC had been insisting.
These studies also questioned the relevance of the assumption of a single
home language as well as the concept of a lingua franca in the urban/township
context.
group came from the Sotho group of languages and four from the Nguni group.
The different neighbourhoods of Tembisa were also equally represented. The
moderator, who was fluent in both Sotho and Zulu, told subjects that their con-
versations were part of a study about how people communicate in Tembisa. Thus
this micro-study aimed at reflecting the language preference in a restricted area,
but taking into account prevailing sociological factors. Her study is discussed
further in section 5.6 below.
A slightly different angle to the influence of Indo-European languages in
urban/township CS patterns is found in the studies on Tsotsitaal and Iscamtho,
the CS slang varieties of the urban areas/townships (Mfenyana 1977; Schuring
1983; Janson 1984; Msimang 1987; Mfusi 1990; Ngwenya 1992; Slabbert
1994; Makhudu 1995; Ntshangase 1995; Childs 1996). A sociological per-
spective is central to these studies, although they also investigate aspects of
language structure. The conflicting sociopolitical perspectives on the origins
of Tsotsitaal and Iscamtho, and particularly the role of Afrikaans in Tsotsitaal,
illustrate very aptly the ideological and sociohistorical nature of linguistic study.
All studies acknowledge that Tsotsitaal is Afrikaans based; they differ, how-
ever, on its origin, its distribution and its future. According to Ntshangase (1995)
Afrikaans was brought to the freehold townships by Tswana-speaking groups
from the western Transvaal who learnt it from the people who dispossessed
them of their land. In contrast to this strong claim of a single origin, both
Makhudu (1995) and Schuring (1983) make reference to the contribution of
coloured speakers of Afrikaans. Ntshangase (1995) and Ngwenya (1992) say
that the 1976 Soweto uprising led to a decline in the use of Tsotsitaal because of
its association with Afrikaans. Slabbert (1994: 38) found in a national sample
of black adult males that almost 25 per cent claimed to be users of Tsotsitaal.
She states, however, that Tsotsitaal ‘is a language of the townships; which [for
its speakers] has nothing to do with the Afrikaans of the whites’. The uprising
has led to a decline in the knowledge of Afrikaans and as a consequence in the
ability of people to conduct fully-fledged conversations in Tsotsitaal (Slabbert
and Myers-Scotton 1996). However, according to Makhudu (1995: 304), the
use of Tsotsitaal/Flaaitaal is ‘widespread and increasing’.
Both the SABC’s macro-studies and Calteaux’s micro-study provided data
in the form of transcripts of hours of naturally occurring conversations. These
data subsequently formed the basis of more focused and in-depth studies into
CS, the major focus of these studies being structural, functional and pragmatic.
(a) having an awareness of what the addressee prefers and switching accord-
ingly;
(b) establishing common ground, i.e. meeting the addressee halfway with lan-
guage;
(c) a willingness to learn and experiment with other languages in the commu-
nication situation even to the point of moving out of one’s comfort zone;
(d) employing measures to make oneself understood;
(e) making adaptations on the variety continuum of ‘deep’ to urban.
This function of accommodation has been reiterated by Finlayson, Calteaux
and Myers-Scotton (1998: 401) in commenting on their data. They note that
in a multilingual setting, such as Tembisa, speakers are aware that communication prob-
lems will arise and that different accommodation strategies may be necessary. While CS
is the main strategy of accommodation, it may take many forms. The form it takes reflects
the norms as well as the demographics of the community. There are many inter-related
forces which are at play in the use of varieties in the speech event as the participants
consciously, but more often unconsciously, switch from one variety to the other. From
the speaker’s own point of view, CS offers a middle path with regard to the costs and
rewards which accrue from using any one language on its own. In this respect, CS is
a ‘safe choice’ [Myers-Scotton 1993b: 147]. Also, by using more than one variety in
a conversation, a speaker can evoke the multiple identities associated with each code
[Myers-Scotton 1993b: 122]. Thus, CS is both a reactive choice, as accommodation,
and a proactive choice, as a presentation of one’s multiple selves.
The authors propose three strategies in the process of accommodation. First,
one participant may speak his/her first language while the other speaks his/her
first language. A second strategy would be the use of the dominant language
of the community as the structurally dominant language in CS. A third strategy
would be for a speaker to repeat what he/she has just said in the language of
the addressee so as to ensure that the message has been understood. Since the
fundamental concern in such events is communication, any strategy will be
used in order to enhance it.
Central to an interactional perspective on CS is the ‘markedness model’
of Myers-Scotton (1993a). Based on extensive fieldwork data in Kenya, she
proposed a model that involved several dimensions:
(1) Bilingual people may use switching as both a tool and a symbol of social
relationships.
(2) Utterances can be used with intentional meaning (e.g. co-operation or
lack of it), and not just referential meaning. In particular, Myers-Scotton
(1992: 166) proposed that choice of one code rather than another is dri-
ven by a negotiation principle: ‘choose the form of your conversation
contribution such that it symbolises the set of rights and obligations which
you wish to be in force between Speaker and Addressee for the current
exchange’.
246 S. Slabbert and R. Finlayson
(3) Code choices reflect the fact that speakers are rational actors, whose social
and personal relationships influence code choice.
(4) The social meaning of a talk exchange is accomplished by the exchange
itself.
(5) A feature of the communicative competence of speakers is a ‘markedness
metric’ which provides speakers with the predisposition to perceive all
code choices as more or less unmarked or marked for a specific talk ex-
change.
Particularly important in the markedness model is the idea of the rights and
obligations of interlocutors, based on the norms of their community. When
a speaker engages in CS, it involves some negotiation over the balance of
rights and obligations between speaker and addressee. Arising from the model,
Myers-Scotton (1992: 169) posits four functions served by CS. Rather than
repeat her examples, we illustrate these four functions using South African
examples from Herbert (1997). Herbert found that all four functions indicated
by Myers-Scotton were prominent in his data set from the University of the
Witwatersrand.
(Another postgraduate student (C) from the same programme joins them. C is Venda
speaking.)
interaction between two strangers outside a city library. ‘A’ uses Sotho to iden-
tify himself and to initiate interaction:
A: Tsela e yang CINEMA ke efe? [Where is the street leading to the cinema?]
B: Uthini ndoda? [(Zulu) What are you saying, man?]
A: Ikuphi indela eya le e-CINEMA? [Where is the road to the cinema?]
B: Uhambe STRAIGHT. Uzobona abantu bayimi e-LINE. [You go straight. You will
see people in a queue.]
It is clear from the example that the initial choice did not work, and by his reply
(which switches the conversation to Zulu) B has successfully renegotiated a
common code.
status by the use of English. For a detailed example and a discussion of reverse
patterning in another setting see Herbert (1997: 412).
Finlayson and Slabbert (1997c) give an alternative interpretation to social
motivations for CS. According to them the markedness model is speaker ori-
ented. It defines markedness according to the speaker’s unwillingness to
conform to societal norms. The fact that Myers-Scotton (1993a: 82) regards
‘marked’ versus ‘unmarked’ as a continuum implies that these norms will not
always be clear to speaker and addressee nor always shared by them in inter-
actions. It implies further that certain interactions will be indeterminate with
regard to markedness. In this case the question can be asked: if speakers con-
sistently foreground the index of linguistic identity, as Zulu speakers are said to
do, are they acting against societal norms or merely affirming their own norms?
It is problematic to categorise this type of CS as either marked or unmarked.
The continuum of ‘unmarked’ to ‘marked’ is furthermore linked to the dy-
namics of the larger social context and to a historical process. What is unex-
pected today can tomorrow start to emerge as expected, due to a change in the
dynamics of the social context. For example, as we have already mentioned
(Finlayson and Slabbert 1997a: 416), it has been said by respondents that Zulu
speakers in Soweto have become more accommodating, since foregrounding
their linguistic identity in certain contexts has become politically sensitive.
As has been pointed out previously (Finlayson and Slabbert 1997a: 416), the
markedness model does not answer the question as to why the speaker would
not conform to the societal norms nor why the speaker would wish to increase or
decrease the social distance with regard to the addressee. Why are Zulu speakers
unwilling to be accommodating? Marked code choices do not take place in a
vacuum. They are the result of salient situational features in as broad a sense
as possible. This view is in line with Bourdieu’s theory of practice, which is
interpreted by Thompson (1991: 16) as follows:
While agents present themselves towards specific interests or goals, their action is only
rarely the outcome of a conscious deliberation or calculation in which the pros and
cons of different strategies are carefully weighed up, their costs and benefits assessed,
etc. . . . Since individuals are the product of particular histories which endure in the
habitus, their actions can never be analysed adequately as the outcome of conscious
calculation.
not only supplies the terms by which this identity is expressed, but a particular
pattern of language use also marks and constitutes this urban/township identity.
Various aspects of language use are singled out, although the authors caution that
they are inextricably entwined. These aspects include the ability, in a delicately
balanced manner, to:
r function in many languages;
r use complex CS patterns;
r adapt within the variety continuum of ‘deep’ versus ‘light’;
r use English and Afrikaans in very specific functions.
The above sentence contains six CPs, indicated by square brackets and sub-
scripts. Two of these are monolingual CPs from Zulu. It is the remaining four
CPs that are subject to the constraints suggested by the MLF model. The details
are rather complex for a detailed exposé here, but the following broader points
can be verified from the Finlayson, Calteaux and Myers-Scotton (1998) paper:
(1) The matrix language is Zulu, and the embedded language English.
(2) System morphemes in mixed constituents come from Zulu. Note, for exam-
ple, CP2 in which the system morphemes are from Zulu, resulting in a Zulu
syntactic pattern. (Gangs is clearly a content unit in itself; the plurality is
marked by the system morpheme ama-.)
(3) An example of an embedded language island is the clause it differs from
one gang to another.
Code-switching in South African townships 253
6 CONCLUSION
The perspectives described above have developed parallel to the political
democratisation process in South Africa. As interlingual and interracial contact
has grown and polarisations have become blurred, studies increasingly have
recognised that CS in the urban/township context is extensive, complex, irrevo-
cable and as such part of the fibre of South African society. The unique features
of South African urban/township CS generate data whose richness has already
contributed to linguistic theory on CS and language change and will continue
to do so.
The accommodation function of CS that has been described above further-
more symbolises values of democratisation: equality, coming together, mutual
understanding and respect. It follows therefore that CS studies have an important
contribution to make to the challenge of implementing a policy of multilingual-
ism in South Africa at all levels. With eleven official languages, duplication and
translation are impractical and costly options. Code-switching, however, offers
the possibility of creating multilingual programmes, advertisements, brochures,
political speeches, etc., which would enable communicators to accommodate
different sectors of our multilingual society not only in terms of understanding
but also in terms of solidarity in a cost-effective way.
note
The authors acknowledge with appreciation the editorial suggestions and changes made
by Rajend Mesthrie.
bibliography
Adendorff, R. 1993. ‘Teacher education: code-switching amongst Zulu-speaking pupils
and their teachers’. South African Journal of Applied Language Studies, 2, 1: 3–25.
Brown, D. 1992. ‘Language and social history in South Africa: a task still to be un-
dertaken’. In R. K. Herbert (ed.), Language and Society in Africa. Johannesburg:
University of the Witwatersrand Press, pp. 71–92.
Code-switching in South African townships 255
Sankoff, D. and S. Vanniarajan 1990. ‘The case of the nonce loan in Tamil’. Language
Variation and Change, 2: 71–101.
Schuring, G. K. 1985. Kosmopolitiese omgangstale. Die aard, oorsprong en funksies
van Pretoria-Sotho en ander koine-tale. Pretoria: HSRC.
1983. ‘Flaaitaal’. In G. Claassen and C. van Rensburg (eds.), Taalverskeidenheid.
Pretoria: Academica, pp. 116–33.
Slabbert, S. 1994. ‘A re-evaluation of the sociology of Tsotsitaal’. South African Journal
for Linguistics, 12, 1: 31–41.
Slabbert, S. and R. Finlayson 2000. ‘“I’m a cleva!”: the linguistic make-up of identity
in a South African urban environment’. International Journal of the Sociology of
Language, 144: 119–35.
Slabbert, S. and C. Myers-Scotton 1996. ‘The structure of Tsotsitaal and Iscamtho:
code-switching and in-group identity in South African townships’. Linguistics, 34:
317–42.
Slabbert, S. and I. van den Berg 1994. ‘Language study’. Management report, prepared
for the South African Broadcasting Corporation.
Thipa, H. 1992. ‘The difference between rural and urban Xhosa varieties’. South African
Journal for African Languages, 12, Supplement 1: 77–90.
Thompson, J. B. 1991. ‘Introduction’. In P. Bourdieu, Language and Symbolic Power,
ed. J. B. Thompson, trans. G. Raymond and M. Adamson. Cambridge, Mass.:
Harvard University Press, pp. 1–31.
van Vuuren, D. and A. Maree 1994. ‘Language and broadcasting in South Africa: a
research perspective’. SABC Broadcasting Research Unit.
van Warmelo, N. J. 1927. ‘European and other influences in Sotho’. Bantu Studies, 3:
405–21.
13 Intercultural miscommunication in South Africa
J. Keith Chick
1 INTRODUCTION
In this chapter I will be reviewing a selection of intercultural and cross-cultural
studies of communication in apartheid South Africa. My purpose in doing so
is, first, to distinguish between cross-cultural and intercultural communication
studies in terms of the theories that inform and the research methods that are
used in them. Second, it is to explore what each type of study has contributed to
an understanding of the sources and consequences of intercultural miscommu-
nication in South Africa and, more generally, of how dominant ideologies and
power relations associated with them affect, and are affected by, the quality of
such communication.
In sociolinguistics, as elsewhere, the terms ‘intercultural’ and ‘cross-cultural’
tend to be used interchangeably. However, following Carbaugh (1990: 292),
I distinguish between them, reserving the term ‘cross-cultural’ for studies
that explore particular features of communication (e.g. compliments, refusals,
apologies, turn-taking) across two or more cultures. I use the term ‘intercultural
communication’ to refer to studies that, by contrast, focus on particular inter-
cultural encounters, and attend to whatever communication features are salient
in them.
258
Intercultural miscommunication in South Africa 259
it’ is that it has the illocutionary force of an apology. It also showed that, though
certain forms and semantic formulas are conventionally used to perform partic-
ular speech acts (e.g. in English, imperative mood conventionally – in unmarked
cases – realises commands), form and function frequently do not coincide. When
that happens interlocutors usually infer the pragmatic force of the utterance by
drawing on relevant contextual information. For example, even though in un-
marked cases imperatives are used to perform directives, a pupil, by noting that
his teacher can see that he is not using a pen, and by recognising that his teacher
has the authority to issue directives, can infer that, when a teacher says ‘Are you
using a pen?’, she may be not asking for information, but directing him to use a
pen. Ethnography of speaking has supplied the understanding that the rules of
speaking for speech acts differ across cultures and languages. Ethnography of
speaking has also shown that interlocutors belonging to different language and
cultural groups often use different linguistic forms and semantic formulas to
realise particular speech acts. Thomas (1983) supplies the example of Russian
konesno, a formula used frequently by Russians in responding to requests. This
word translates literally as ‘of course’, in English. This frequently leads to
miscommunication, because whereas konesno in Russian conveys enthusiastic
affirmative, ‘of course’ in English implies that the speaker has asked something
that is self-evident. Ethnography of speaking has also shown that interlocu-
tors with different cultural backgrounds often ‘assess’ the situational context
of their talk differently, thus often having different views about what forms
and formulas are appropriate. For example, Zulu speakers tend to accord old
people more status than South African English (SAE) speakers do. This means
that they are shocked when SAE speakers, in issuing directives, address elderly
employees by their first names. Zulu speakers would, instead, use a respectful
address form such as baba (‘father’), no matter how humble the employee’s po-
sition. Conversational analysts have supplied the understanding that when form
and function do not coincide the interpretation of such ‘indirect speech acts’ is
accomplished not only inferentially but also interactionally. For example, the
force of the utterance ‘Are you doing anything at the moment?’ is contingent
when uttered and only definite in retrospect (Dore and McDermott 1982: 386).
It is contingent, for example, on the listener’s response. If he/she responds: ‘No,
why? Do you need a hand?’ it retrospectively takes on the force of a directive
(or pre-directive). On the other hand, if he/she replies: ‘Yes, I’m tied up with
my homework’ it retrospectively takes on the force of a request for information.
acts have used a wide range of data-collection procedures from, at the one
extreme, tape-recordings of spontaneous, naturally occurring exchanges to, at
the other extreme, obtaining written responses to discourse completion tasks.
These include contextual descriptions and dialogues in which the utterances that
realise the particular speech acts focused on are left blank so that respondents
can fill in the words they think they would use to complete dialogue. With
very few exceptions (see, however, Clyne 1995) the data is limited to single
utterances or short exchanges.
The researchers also use a wide range of data-analysis procedures, the
choice being determined partly by the nature of the data collected and partly by
the researcher’s purposes. Since it is beyond the scope of this chapter to illus-
trate even a representative sample of the methods employed, I have based my
choice not on the research methods employed, but on whether the researchers
have addressed the second of my objectives, namely how speech-act perfor-
mance relates to dominant ideologies.
The first of these studies is Herbert’s comparative study of patterns of compli-
ment responses among white, middle-class Americans and white, middle-class
South Africans (see Herbert 1985, 1989; Herbert and Straight 1989).
Herbert asked students in his linguistics classes at the State University of
New York at Binghamton (1980–1 and 1982–3) and at the University of the
Witwatersrand (1981–2) to record compliment-giving and responding se-
quences as they occurred spontaneously in public places on campus. He sub-
sequently coded the responses in terms of a typology of twelve response types
devised by Pomerantz (1978), which he subsequently refined. Figure 13.1
lists the twelve response types or strategies, giving examples of each.
Pomerantz (1978) explains that acceptance and rejection of compliments are
both problematic, as they violate one or other of two putative universal conver-
sational principles: agree with the speaker and avoid self-praise. She explains,
further, that many of the response types in her typology (3–12) are strategies
for resolving this conflict by exhibiting features of both acceptance and
rejection.
Finally, Herbert counted and aggregated tokens of each response type and
represented his findings in the form of a comparative table.
Table 13.1 reveals most Americans in the New York corpus tend not to accept
most compliments, while most South Africans in the Witwatersrand corpus tend
to accept them.
Of particular relevance to my second purpose, Herbert and Straight (1989)
attempt to explain the difference between the patterns of response types in
table 13.1 in terms of the dominant ideologies and social relations in the USA
and South Africa. They suggest that Americans reject compliments frequently
in order to avoid the implication that they are superior to their interlocutors, i.e.
by giving priority to the injunction avoid self-praise. This behaviour they
see as consistent with the ideology of egalitarian democracy most Americans
Intercultural miscommunication in South Africa 261
Accepting
R: Thank you.
R: I got it at Boscov's.
# % # %
Accepting
1. Appreciation token 312 29.4 162 32.9
2. Comment acceptance 70 6.6 213 43.2
36.0 76.1
Deflating, deflecting, rejecting
3. Reassignment 32 3.0 23 4.7
4. Return 77 7.3 12 2.4
5. Qualification (agreeing) 70 6.6 12 2.4
6. Praise downgrade (disagreeing) 106 4.5 0 6.3
7. Disagreement 106 10.0 0 0.0
31.4 15.8
Questioning, ignoring, reinterpreting
8. Question (query or challenge) 53 5.0 9 1.8
9. Praise upgrade (often sarcastic) 4 0.4 2 0.2
10. Comment history 205 19.3 24 4.9
11. No acknowledgement 54 5.1 1 0.2
12. Request interpretation 31 2.9 4 0.8
32.7 7.9
Totals 1,062 100.1 492 99.8
publicly espouse, and with the structure of a society in which social status is
open to negotiation. They suggest that, by contrast, white, middle-class South
Africans frequently accept compliments to keep non-equals at a distance, by
allowing the compliment to imply that they are superior to their interlocutor,
by giving priority to the injunction agree with the speaker. This tendency
they see as consistent with the ideology of ‘institutionalised social inequality
publicly enunciated in South Africa’ (1989: 43), and a social structure in which
social status is to a large extent predetermined.
Herbert and Straight, moreover, argue that there is a reflexive relationship be-
tween micro- and macro-phenomena. Not only do structural macro-phenomena
such as ideologies determine patterns of sociolinguistic behaviour, these pat-
terns are constitutive of structural macro-phenomena. They suggest that
Americans engage in compliment rejecting ‘not because they feel confident
that they and their interlocutors share feelings of mutual worth and equality,
but rather because they are trying to establish this mutual worth and equality’
(1989: 43). Similarly, they suggest that South Africans’ patterns of compliment
responding serve to ‘affirm’ solidarity with white peers and confirm their elite
Intercultural miscommunication in South Africa 263
# % # % # %
Accepting
1. Appreciation token 312 29.4 162 32.9 48 33.0
2. Comment acceptance 70 6.6 213 43.2 10 6.9
36.0 76.1 39.9
Deflating, deflecting, rejecting
3. Reassignment 32 3.0 23 4.7 3 2.0
4. Return 77 7.3 12 3.4 0 0.0
5. Qualification (agreeing) 70 6.6 12 2.4 8 5.5
6. Praise downgrade (disagreeing) 106 10.0 0 0.0 14 9.7
7. Disagreement 106 10.0 0 0.0 6 4.1
31.4 15.8 21.3
Questioning, ignoring, reinterpreting
8. Question (query or challenge) 53 5.0 9 1.8 19 13.1
9. Praise upgrade (often sarcastic) 4 0.4 2 0.2 5 3.4
10. Comment history 205 19.3 24 4.9 13 9.0
11. No acknowledgement 54 5.1 1 0.2 17 11.7
12. Request interpretation 31 2.9 4 0.8 2 1.4
32.7 7.9 38.6
Totals 1,062 100.1 492 99.8 145 100.0
did with his New York and Witwatersrand data, I gathered the additional infor-
mation I needed to produce a comparative table.1
Table 13.2 reveals that whereas only 23.7 per cent of the responses in the
Witwatersrand corpus involve saying something that can be interpreted as a
rejection or partial rejection (i.e. 15.8 plus 7.9), as many as 59.9 per cent of
responses in the Natal corpus fall into this category (i.e. 21.3 per cent plus 38.6
per cent).
Since no data are available for the Durban campus in 1981–2 (when Herbert
collected his Witwatersrand corpus) it is not possible to rule out the possibility
that the difference between the patterns of frequencies of compliment-response
types on the two South African campuses shown in table 13.2 reflects regional
variation in the relevant rules of use rather than the effect of historical change.
However, on the assumption that the patterns of response on these two campuses
were similar in the early 1980s, the table suggests that, indeed, the pattern of
responses of white, middle-class South Africans has changed markedly.
In attempting to interpret these findings, I suggested that the presence of
significant numbers of black students in formerly exclusively white institutions
Intercultural miscommunication in South Africa 265
might of itself have been a spur to white students to question traditional social
relations of power. I also suggested that political instability and the decline in
the economy, which characterised the 1980s, might have served to undermine
the unquestioning assumption of many whites that their high status would be an
enduring feature of South African society. I concluded that it is plausible that
increasing uncertainty about social relations of power led whites increasingly
to avoid the implication associated with acceptance of compliments, namely
that they are superior to their interlocutors.
To address the second of my objectives (establish the extent and nature of
cultural diversity in compliment-response behaviour on this campus), I counted
tokens of each response type in the entire corpus (with the exception of those
produced in intercultural encounters),2 and on the basis of the information
about the ethnic background of the respondents supplied by research assistants
produced a table showing totals and aggregates for each response type for the
three groups: whites, Indians and blacks.
The extent of the differences in the frequency of use of different response
types by the different groups reflected in table 13.3 suggests that on the Durban
campus there is considerable potential for intercultural miscommunication. For
reasons of scope I shall refer to just two differences.
There are marked differences in the frequencies of choice of response cate-
gory 7 (‘disagreement’). Whereas, in my corpus, as many as 10.4 per cent of
the total responses of Indian students fall into this category, only 3.6 per cent
of the total white responses and 3.1 per cent of the total black responses do so.
Moreover, what is distinctive about the Indian disagreements is that many are
very direct, such as in the following example:
A: Your hair looks nice today.
B: It’s a mess.
A: No, it’s not.
What this suggests is that, for whites, disagreements are particularly face-
threatening, and that they use devices such as hedges as a means of redress/of
resolving the conflict between the two principles. It follows that this group
would probably interpret the overt disagreements of Indian students as rude,
even where no offence was intended.
There are also noticeable differences in the frequency of choice of the
compliment-response strategy of ‘no acknowledgement’. Whereas as few as
10.7 per cent of ‘white’ and 11.5 per cent of Indian responses fall into this
266 J. K. Chick
Accepting
1. Appreciation token 8 12.5 29 33.3 62 36.9
2. Comment acceptance 9 14.1 7 8.1 10 6.0
26.6 41.4 42.9
Deflating, deflecting, rejecting
3. Reassignment 3 4.7 3 3.5 4 2.4
4. Return 1 1.6 1 1.2 0 0.0
5. Qualification (agreeing) 2 3.1 4 4.6 8 4.8
6. Praise downgrade (disagreeing) 3 4.7 5 5.8 15 8.9
7. Disagreement 2 3.1 9 10.4 6 3.6
17.2 25.5 19.7
Questioning, ignoring, reinterpreting
8. Question (query or challenge) 7 10.9 9 10.4 23 13.7
9. Praise upgrade (often sarcastic) 5 7.8 6 6.9 5 4.2
10. Comment history 2 3.1 2 2.4 13 7.7
11. No acknowledgement 21 32.8 10 11.5 18 10.7
12. Request interpretation 1 1.6 1 1.2 2 1.2
56.2 32.4 37.5
Totals 64 100.0 87 99.3 168 100.1
category, as many as 32.8 per cent of ‘black’ responses in the corpus do so.
Such conspicuous absence might easily be interpreted, by someone expecting a
response, as an unwillingness to engage and, therefore, as face-threatening. A
case in point is the following example from my corpus (followed by translation
in English) which is part of a conversation between two male Zulu students in
B’s university residence room:
A: (Knocks)
B: Come in.
A: Heita Bheki. [Hi Bheki.]
B: Eit kunjani mfowethu? [Hi. How are you brother?]
A: Ei grand man. [I’m fine thanks.] (moves towards the table) Hawis mfowethu, yaze
yayinhle le radio eyakho? [Hey brother I like your radio, it’s so beautiful. Is it your
radio?]
B: Yebo. [Yes.]
A: Yaze yayinhle futhi inkulu. Wayithenga kuphi? Ngamalini? [It is so beautiful and big.
How much did you pay for it? Where did you buy it?]
Intercultural miscommunication in South Africa 267
What is interesting about this example is that it suggests why the choice of ‘no
acknowledgement’ is not interpreted by Zulu interlocutors as unwillingness to
engage, and, therefore, as face-threatening. What the complimenter frequently
does is to make a response to the compliment less conspicuously absent by
adding another speech act immediately after the compliment. Thus, for exam-
ple, A, after recycling and embellishing his compliment (It is so beautiful and
big, line 7), asks two questions (How much did you pay for it? Where did you
buy it?). B is thus able to avoid responding to the compliments, by answering
the questions (In town – it was R399, line 8). It is possible, however, that mem-
bers of other groups who are unfamiliar with this strategy might not see B as
having been released from his obligation to provide a response. (For a study
of cross-cultural directives with specific reference to Zulu see de Kadt 1995.)
De Kadt (1998) deals with the concept of ‘face’ in relation to politeness in
Zulu.
Student: |I. . .think one and two are which was equally difficult
Professor: | equally difficult
Student: |yah
Professor: |and
Student: |and not actually difficult but I think er not prepared
270 J. K. Chick
The professor, by treating ‘equally difficult’ as a single tone group with nucleus
placement (a rise-fall pitch movement) on ‘equally’, signalled that this is the
part of the message that he would like the student to build on. However, as is
apparent from the student’s reply, which addresses whether or not the questions
were difficult rather than which of the two questions was the more difficult, the
ZE student did not perceive the accentuation cue on ‘equally’ as salient. There
are also differences in rates of speech and pause lengths, with Zulus, accord-
ing to one informant, valuing behaviour that proceeds at a steady, measured
pace. Ironically, attempts by the interlocutors to repair were frustrated by the
progressively increasing asynchrony. They failed to attend to one another’s re-
pairs, either because they were talking at the same time, or because the repairs
came at the ‘wrong’ times, i.e. when they were not expected.
I found that a further source of interactional synchrony in intra- as well as
intercultural encounters was the mismatch of ‘readings’ by the professor and
some students of their relations, and of how these are affected by the loss of
face experienced by the students who performed poorly in the examination.
Comparing interactions involving students (one a ZE speaker and the other
an SAE speaker) who fared relatively poorly in the examination, I found that
they both did considerable face-repair work but used very different politeness
strategies to do so (see Scollon and Scollon 1981). The ZE student in the
interaction above tended to use deference politeness strategies, i.e. strategies
that offset possible loss of face or redress of face by assuring the hearer that the
speaker respects his/her independence. For example, as the encounter became
more stressful, he used the address term sir, which contrasts with the absence
of any address term earlier in the interaction, and implies that he did not wish
to challenge the professor:
By contrast, the SAE student who perfomed poorly tended to use what Scollon
and Scollon (1981) view as a type of solidarity politeness, namely, ‘bald-on-
record’ without redressive action. The implicit assumption with bald-on-record
strategies is that relations between the interlocutors are so close that redressive
action is unnecessary. In this case, the student resisted the professor’s attempts
to get the floor, and put words in his mouth:
Intercultural miscommunication in South Africa 271
Student: |now I don’t think I did this in this essay um answered entirely in that frame
of reference
Professor: |ya
Student: |I think that is what
you’re going to say
Professor: |well well I’m I’m wanting to see
Student: |You’re you’re going
to say I didn’t actually um answer the essay in relation
See Chick (1990) for the full transcript.
I also discovered that the interactional consequences of the choices of strate-
gies for repairing face were different. By using ‘bald-on-record’ solidarity strate-
gies the SAE student contested the professor’s assumption that their relations
were friendly, but not his assumption that their relations were symmetrical, and
that, therefore, reciprocal solidarity strategies were appropriate. The ZE student,
by using non-reciprocal deference politeness ‘up’, more severely challenged
the professor’s assumptions about symmetrical power relations and, thus was
more negatively evaluated. The irony, as Carbaugh (1990: 156) points out, is that
the ‘use of one’s best cultural manners’ can lead ‘unknowingly and innocently’
to negative evaluation.
In attempting to sum up what this study contributes to an understanding of
the sources and consequences of intercultural communication, I argued that
racial segregation associated with apartheid ideology kept groups ignorant of
one another’s culturally specific discourse conventions. I argued further that
miscommunication and misevaluation in countless gatekeeping encounters (see
Erickson and Shultz 1981) such as these served to maintain the culture of racism.
It did so partly by ensuring that members of historically disadvantaged groups
were often negatively evaluated by gatekeepers (more often than not members
of dominant groups) and that, therefore, they did not get their fair share of
resources and opportunities. It did so also because repeated miscommunication
generated negative cultural stereotypes. Such stereotypes contributed to further
miscommunication by predisposing gatekeepers to perceive only behaviours
that matched the stereotypes, and apparently provided a justification for the
maintenance of discrimination and segregation that had been the source of
miscommunication in the first place.
Further insights into the sources and consequences of intercultural miscom-
munication in South Africa are provided by Kaschula’s (1989 and 1995) studies
of communication through the medium of Xhosa on farms and in courts in the
Eastern Cape. It is to these studies that I finally turn.
interactional styles differed from their own along a number of dimensions, but
they had to interact using a language not their own, and in which they often had
limited proficiency.
What makes Kaschula’s studies particularly significant is that he examines
intercultural communication through Xhosa. Most studies of intercultural com-
munication have involved communication through a dominant language, and
have been carried out by researchers who are themselves members of a dominant
group and native speakers of that dominant language (see Singh et al. 1988 for
a critique of these trends).
In the first of these studies Kaschula (1989) examines intercultural com-
munication between white farmers and Xhosa-speaking labourers. He notes
that while Xhosa is used as a medium, the very asymmetrical social relations of
power that conventionally obtain ensure that the farmers control topic choice and
turn-taking, and do most of the talking. He explains that, over time, this has led
to the development of a limited farming register characterised by considerable
code-mixing. Because farmers find it necessary to speak only the farming reg-
ister, and because the labourers’ share of the floor is restricted, opportunities to
resolve communication difficulties arising from the farmers’ lack of proficiency
in Xhosa and mismatch of culturally specific discourse conventions are limited.
In the second of these studies, Kaschula (1995) examines intercultural com-
munication in the law courts of the Eastern Cape. Here, too, Xhosa serves as
a medium, but only through the offices of an interpreter. Accordingly, Xhosa
speakers’ opportunities to resolve communication difficulties are even more re-
stricted than in the farming context. As Kaschula notes, courtroom procedure,
as elsewhere in the country, including choice of language medium, ‘belongs
to the dominant minority culture’ (1995: 9). Witnesses may use indigenous
languages but, significantly, it is the translations into English or Afrikaans, and
not the actual words of the Xhosa speakers, that are recorded, i.e. incorporated
into official courtroom discourse, and used as the basis for assessing character,
honesty, and so on. Kaschula points to many reasons for inaccurate translating
from Xhosa into English and vice versa. These include the absence of equiv-
alent one-to-one terms in the two languages; the often limited proficiency in
English of many translators; translators’ unfamiliarity with the dialect of the
Xhosa speakers; and the fact that pervasive asymmetrical social relations of
power do not encourage lawyers to become sufficiently proficient in Xhosa for
them to be able to check the accuracy of translations.
These studies add to our understanding of the sources and consequences
of miscommunication in South Africa. In other words, they add to our under-
standing of how structural arrangements of institutions constrain what occurs in
interactions, and how the outcomes of interactions serve to maintain structural
inequities in them. The special interest of these studies is that they help us to
understand the plight of some of the most disadvantaged people in the society.
Intercultural miscommunication in South Africa 273
4 CONCLUSION
In this chapter I have surveyed a selection of studies that attempt to trace the
sources of miscommunication in intercultural communication in South Africa,
and to indicate what insights they provide into how the structural circum-
stances of apartheid society affected the quality of communication and how
pervasive miscommunication impacted on that society. The apartheid regime
has fallen and, while some of the structural circumstances that characterised
it have changed radically, others are still very much intact, and intercultural
miscommunication is still pervasive. It follows that sociolinguists have at this
particular period in South Africa’s history a unique opportunity of further ex-
ploring the relationship between macro-structural phenomena and what occurs
in the micro-contexts of conversational interactions. What I suggest is particu-
larly urgent is comparative studies of intercultural communication in domains
in which there has been dramatic structural change and those in which there
has been minimal structural change.
notes
1 One departure I made from the procedures used by Herbert was in the coding of what
he terms compound responses, such as:
A: Nice coat.
B: Thanks. Katherine gave it to me.
Herbert (1985: 80) reports that he coded such responses on the basis of ‘perceived
intention’. Thus, for example, in the above exchange, he would have coded B’s re-
sponses, type 5 (qualification) even though the first part of the response, if it had oc-
curred on its own, would have been coded type 1 (appreciation token). My misgiving
about this way of proceeding is that it increased, to what I considered an unacceptable
degree, the subjectivity involved in coding responses. Accordingly, with compound
responses I adopted the policy of coding all the types involved. For example, I coded
the above response 1 + 5.
2 I chose not to include compliment responses produced in the intercultural encounters
because counting revealed that the patterns of choice differed considerably from those
in intracultural encounters.
274 J. K. Chick
bibliography
Beebe, L. M. and M. C. Cummings 1996. ‘Natural speech act data versus written ques-
tionnaire data’. In S. M. Gass and J. Neu (eds.), Speech Acts across Cultures. Berlin:
Mouton de Gruyter, pp. 65–86.
Carbaugh, D. 1990. Cultural Communication and Intercultural Contact. Hillsdale, N.J.:
Lawrence Erlbaum.
Chick, J. K. 1987. ‘Linguistics, language and power’. In D. Young (ed.), Festschrift
in Honour of Len Lanham. Cape Town: Maskew Miller Longman, pp. 115–
31.
1990. ‘The interactional accomplishment of discrimination in South Africa’. In
Carbaugh (ed.), pp. 225–52.
1996. ‘English in interpersonal interaction in South Africa’. In de Klerk (ed.),
pp. 269–84.
Clyne, M. 1995. Inter-cultural Communication at Work: Cultural Values in Discourse.
Cambridge and New York: Cambridge University Press.
Cohen, A. 1996. ‘Speech acts’. In S. Mackay and N. Hornberger (eds.), Sociolinguis-
tics and Language Teaching. Cambridge: Cambridge University Press, pp. 383–
420.
de Kadt, E. 1985. ‘The cross-cultural study of directives: Zulu as a non-typical language’.
South African Journal of Linguistics, Supplement 27: 45–72.
1998. ‘The concept of face and its applicability to the Zulu language’. Journal of
Pragmatics, 29: 173–91.
de Klerk, V. 1996. Focus on: South Africa. Varieties of English around the world GS.
Amsterdam and Philadelphia: John Benjamins.
Dore, J. and R. P. McDermott 1982. ‘Linguistic indeterminacy and social context in
utterance interpretation’. Language, 58, 2: 374–98.
Erickson, F. 1978. ‘Timing and context in everyday discourse: implications for the study
of referential and social meaning’. Paper presented at a conference on children’s
oral communication skills, University of Wisconsin.
Erickson, F. and J. Shultz 1981. ‘When is a context? Some issues and methods in the
analysis of social competence’. In J. L. Green and C. Wallet (eds.), Ethnography and
Language in Educational Settings. Advances in discourse processes 5. Norwood,
N.J.: Ablex, pp. 147–60.
Gumperz, J. 1982. Discourse strategies. Cambridge: Cambridge University Press.
Herbert, R. K. 1985. ‘Say “Thank you” – or something’. American Speech, 61: 76–88.
1989. ‘The ethnography of English compliments and compliment responses: a
contrastive sketch’. In W. Olesky (ed.), Contrastive Pragmatics. Amsterdam and
Philadelphia: John Benjamins.
Herbert, R. K. and H. S. Straight 1989. ‘Compliment-rejection versus compliment-
avoidance: listener-based versus speaker-based pragmatic strategies’. Language
and Communication, 9, 1: 35–47.
Kaschula, R. H. 1994. ‘Cross-cultural communication in a north eastern Cape farming
community’. South African Journal of African Languages, 9, 3: 100–4.
1995. ‘Cross-cultural communication in Eastern Cape with particular reference to
law courts’. South African Journal of African Languages, 15, 2: 9–15.
Mehan, H. 1979. Learning Lessons. Cambridge, Mass.: Harvard University Press.
Intercultural Miscommunication in South Africa 275
Language contact
Gender, language change and shift
14 Women’s language of respect:
isihlonipho sabafazi
R. Finlayson
1 INTRODUCTION
The 1990s witnessed an ever-quickening pace in southern Africa towards urban-
isation and modernisation at the expense of custom and tradition. The socially
accepted form of behaviour in a community and the customs and beliefs be-
ing handed down from one generation to the next have become blurred as the
many forces being exerted upon the traditional family lifestyle take effect. The
object of this chapter is to trace the development of a specific sociolinguistic
phenomenon which has been particularly affected by modernisation. This con-
cerns a language variety associated with respect practised by certain southern
Bantu-speaking people, more specifically Nguni and Southern Sotho-speaking
women. This chapter considers this phenomenon as related in particular to the
Xhosa-speaking women.
First, it would assist those unfamiliar with this interesting linguistic phe-
nomenon to put it in context by a hypothetical example from English. One could
consider the following situation: Robert and Grace Green have three children –
William, Joan and Margaret. William marries Mary and takes her home to
his family. Here she is taught a new vocabulary by Joan, her sister-in-law and
where necessary advised by Grace, her mother-in-law. This is because from now
on she may never use the syllables occurring in the names of her husband’s
family, i.e. simplistically rob, ert, green, will, may and grace. Thus for the sen-
tence ‘Grace will not eat green yoghurt’, Mary would have to say something
like: ‘The older daughter of Smith refuses to eat grass-coloured yomix.’ This
(in simplified form) demonstrates the linguistic constraints to which one would
be subjected in conforming with this linguistic custom.
This custom, known among the Nguni as ukuhlonipha, (literally ‘to respect’)
and among the Southern Sotho as ho hlonepha, has been defined in a number
of ways. Kropf and Godfrey (1915: 161) describe it as to ‘be bashful, respect,
keep at a distance through reverence and to shun approach’. They go on to add:
279
280 R. Finlayson
the names of their chief’s or their husband’s relations, especially their father-in-law;
they must keep at a distance from the latter. Hence they have the habit of inventing new
names for those persons.
being eroded and replaced by other forms of social behaviour. With regard to
the Xhosa, Hoernlé (1946: 67) noted that there was an ‘ordered group-life,
with reciprocal rights and duties, privileges and obligations, of members, and
moulding the feelings, thoughts, and conduct of members according to these
patterns, so that it is only in and through them that the individual can achieve
his personal self-realization and participate in the satisfaction offered by the
life of his community’. To this effect a baby (usana, class 11/10) is accorded
no gender until it is able to play a role in society. This is evidenced by the
prefixation of class 11 u(lu)-sana which is not a personal class (cf. um-, class 1,
personal). In such a patriarchal society, it is the mother who has the task of
bringing up the child, who teaches it to speak and learn the rules according
to which it must live in order to be an accepted member of the community. A
young boy (inkwenkwe, 9/6) at the age of about five or six, among other things,
is taught to herd the calves or lambs. He and his peers (intanga, 9/10) would set
off at an early hour in the day, returning in all probability after the morning’s
meal, thereby having to make do with a cold rather than a freshly prepared
meal. Thereafter he would set out once more to spend hours in the fields taking
note of nature, learning the names of edible plants and birds, setting traps and
hunting small animals and playing games, such as mock fights with his friends.
Meanwhile, the girl (intombazana, 9/6), who is the sole concern of her mother,
is kept at home to do the necessary household chores such as collecting water
and wood and acting as a nurse for a younger brother or sister. Previously her
main recreation would have included playing with clay or mealie-cob dolls.
From the age of approximately ten to twelve, the responsibility of the child
generally changes. The young boy will now begin to herd cattle and traditionally
will be instructed in such things as spear throwing and stick fighting. His father
now becomes involved in the education of his son, but the son’s ultimate loyalty
is to the chief. The young girl, on the other hand, will now learn how to perform
more responsible work in the home and its surroundings, and will be taught by
the women in the home how to cook, make clothes and work in the garden. At
each stage of development, the ultimate code is one of respect for one’s seniors
and a particular naming procedure is adhered to (Finlayson 1986).
The progress of a traditional Xhosa child continues to the initiation schools,
where, as custom prescribes, a girl would become an intonjane (9/10) and
the boy would go through the circumcision rites as an umkhwetha (1/2). Both
of these ceremonies have linguistic rules attached to them. However, as the
intonjane ceremony is rarely practised now, its linguistic connotations have
fallen away (Jonas 1972). Isikhwetha, the language of the Xhosa initiates, is,
however, still practised in certain situations (Finlayson 1998). The value of the
ceremonies has been accentuated as they incorporate an oath of allegiance to the
chief and have an effect on the moral behaviour of the individuals (Hunter 1961).
The initiates are subjected to rituals and come through these rituals into a new
life of responsibility attached to adulthood.
282 R. Finlayson
This period of life is associated with courtship and games related to this
courtship, but strict rules are imposed upon any relationship existing between
young girls and boys. During this time the young girls would begin to learn the
rules of avoiding the syllables occurring in the family names of their boyfriends,
and thus would have the chance to practise complying with these rules in
preparation for married life. There were threats of severe punishment, such
as baldness, barrenness and other possible consequences, for those who did not
adhere to these rules regarding their relationship with the opposite sex.
1.2 Marriage
The process of courtship leads ultimately to the marriage contract and negotia-
tions between the parents-in-law. Stewart (1940) gives the purpose of ukulobola
(‘bridewealth’) as to ‘secure father’s consent to the marriage of his daughter,
translating her from his guardianship to that of her husband and [a transition] by
which the father would lose the benefit of his daughter’s services, the intending
husband was obliged by payment or by services rendered to the guardian to
prove his fitness to undertake the duties of husband and future guardian’.
This was a means of binding families and sub-clans together. It also implied
that the descent and, in turn the inheritance, was patrilineal. The Nguni are
exogamous and therefore marriage between related clans is forbidden. After
the marriage the young woman moves ceremonially from her home to that of
her in-laws. Here she is taught to respect all the senior relatives of her husband,
especially the male relatives and her mother-in-law. She has to avoid certain
areas of the homestead which are frequented by men and also the cattle kraal
(Hunter 1961: 36–47). As the daughter-in-law she is expected to be even more
responsible to her mother-in-law than to her husband and, in turn, the mother-in-
law is expected to protect her. Any misconduct on the part of the mother-in-law
leads to the invocation of the theleka custom, i.e. the daughter-in-law would
return home to her own parents until a fine (uswazi, 11) has been paid for any
misbehaviour towards her.
From the time that the woman enters her in-laws’ home she may not pronounce
words containing any syllable that is part of the names occurring among her
husband’s relatives. Various reasons for this linguistic form of respect may be
postulated, such as the intention that the daughter-in-law should be aware that
she has not been born into this particular family and thus should be distinguished
from the natural daughters. Further, she should also be conscious of her new
state and, by respecting her in-laws, some of whom may be deceased, she may
also be seen to be respecting the ancestors of her new home. In turn, she should
be respected and protected herself. As Herbert (1990: 463) notes in connection
with her dual life:
Such a transfer does not, however, terminate the woman’s membership in her birth group;
rather, marriage confers membership (or potential membership) in a second group. She
maintains links with the birth group, particularly with the ancestral shades of that group,
and she may return to her father’s homestead if she is expelled or seeks to escape from
her husband’s homestead. Members of the latter are acutely aware of the wife’s dual
membership, which is the basis for the view of wives as ‘strangers’ and ‘outsiders’
throughout their lives.
Herbert (1990: 471) also postulates that ‘it may be possible that avoidance prac-
tices such as hlonipha occur only in societies with a high incidence of unique
names and where names are derived from ordinary words of language’. This
is indeed the case among the Xhosa-speaking people. The traditional family
was generally a patrilineal extended family and the young daughter-in-law was
expected not only to respect the senior members linguistically but also to avoid
them physically. In fact, her movements in and around the home, her form of
dress and her eating habits would be severely restricted. Most of her instruction
would come from her mother-in-law, but her sisters-in-law (indodakazi, 9/6),
especially the eldest, would play an active role in instructing her. Once her
children are born the whole cycle begins again. The children are aware of their
mother’s speech in the home and are made conscious of the procedures involved
in the act of respect. The woman is expected to hlonipha throughout her life.
However, Herbert (1990: 461) notes that ‘the hlonipha customs affecting a
Nguni bride do not continue to operate through her life. Most of these prohibi-
tions are gradually relaxed by a special release ritual or by a verbal order from the
mother-in-law.’
Other authors on the subject as quoted by Herbert (1990: 461), however,
support the theory that women should hlonipha throughout their lives (Mzamane
284 R. Finlayson
1962; Mqotsi 1957), but in practice it has been found that recently, once the
woman’s sons are married, she becomes the senior partner in the homestead
and the rules do not apply as strictly. Dowling (1988: 30) has the following to
say in this regard:
The newly married woman is not allowed to treat this custom lightly, and is
subjected to severe public shame should she ignore the rules laid down for
her. The forces exerted by public opinion are a very important deterrent in
upholding these rules, as one may be ostracised from one’s community. In
general the communities were isolated and monolingual and in most cases the
members of the community were illiterate. Should the daughter-in-law disobey
the rules of the community, she might be sent home and have to return with a
gift of some sort in penitence. Herbert (1990: 459) notes in this regard: ‘The
idea of cleansing is central to the rite of repairing hlonipha infractions . . . In all
cases, it is most particularly the ancestors who must be appeased.’
2.1 Exemplification
In order to give some idea of how the women’s speech is affected by the rules of
hlonipha, an example has been extracted from a conversation recorded in a rural
part of the eastern Cape in 1978. Buziwe Diko is in conversation with Nogogose,
her daughter-in-law, who has her own mother, Zondiwe Qebeyi, present and
therefore feels confident, and verges on being rude to her mother-in-law, who,
in turn, responds angrily.
The italicised words in the conversation show the hlonipha words and their
Xhosa equivalents. While the conversation is heated, the hlonipha rules are
nevertheless adhered to throughout and the syllables occurring in the in-laws’
family are strictly avoided. Nogogose’s in-laws’ names are Mbombo, father-in-
law (utatazala, 1a/2a), and Buziwe, mother-in-law, whose in-laws in turn are:
Ncaphayi, father-in-law; Mthetho, brother-in-law (ubhuti, 1a/2a); Khethiwe,
sister-in-law (indodakazi, 9/6); Xoseka, grandfather-in-law (utatomkhulu, 1a/
2a); Ntobeko, uncle-in-law (umalume, 1a/2a); Msongelwa, uncle-in-law; and
Diko, great-grandfather-in-law (ukhokho, 1a/2a). Figure 14.1 illustrates the
family tree, which has been empirically derived only from those members of
the family whose names were avoided in the above conversation.
Diko
Xoseka
Ngogose Thami
3 EFFECTS OF MODERNISATION
The evolution of hlonipha should be seen within the urbanisation framework
as it exists in the various regions of southern Africa (Dewar et al. 1982). Prior
to the discovery of minerals in South Africa in 1860 the scale and rate of
urbanisation were relatively low, with traditional subsistence farming providing
a more beneficial alternative to wage labour in the cities. Between 1870 and
1913, however, a rapid increase took place in the rate of urbanisation. This
trend continued, and today different regions experience varying degrees of
rapid urbanisation. Much has been written regarding the effects of urbanisation
and modernisation on the Xhosa people (e.g. Hunter 1961; Jonas 1972). Pauw
(1976: 159) notes:
Generally speaking, relations between husband and wife are no doubt closer in the
urban household than they used to be in the traditional Xhosa umzi, where avoidance
customs and ritual emphasized the position of the wife as an outsider in her husband’s
homestead . . . Where marriage takes place in town, a young couple usually choose
each other in the first place, . . . and in married life they tend to be less involved with
their parents and in-laws and more aware of their exclusive responsibility for an own
household from an early stage.
Personal research in Cape Town, Pretoria and Soweto has reaffirmed this with
the general trend being away from the extended family and the hardships in-
volved for the daughter-in-law. In the urban areas hlonipha is not taken as
seriously as it used to be, and in most cases is not adhered to at all. Dowling
(1988) similarly has found in her research conducted in three areas, Tsolo and
Mqanduli in the Transkei and Cape Town, that attitudes towards this custom
have changed. She notes (1988: 3) that ‘the fact that all three communities in-
dicated some doubt as to the continued existence of the custom demonstrates
the likelihood of such a situation developing’. However, Dowling points out
that the most important variables in analysing the different attitudes towards
hlonipha are age, social mobility and education. She also comments (1988: 68)
that poverty and its associated problems retarded economic growth by inhibiting
social development and encouraged a certain linguistic conservatism.
people
ityhagi (5/6) inkwenkwe (9/6) boy
inikazi (9/10) intombi (9/10) girl
incentsa (9/6) indoda (9/6) man
umnyepha (1/2) umlungu (1/2) white man
ityubuka (9/10) usana (11/10) baby
ikhitha (5/6) ixhego (5/6) old man
body parts
iphoba (9/10) intloko (9/10) head
amagabuko (6) amehlo (6) eyes
umnakazo (3/4) ingalo (9/10) arm
isinyamba (7/8) isifuba (7/8) chest
ikruqelo (5/6) idolo (5/6) knee
umnabo (3/4) umlenze (3/4) leg
miscellaneous
inkumba (9/10) indlu (9/10) house
umbaso (3) umlilo (3) fire
isilozelo (7/8) isipili (7/8) mirror
umgaqo (3/4) indlela (9/10) road
inkwezi (9/10) inyanga (9/10) moon
isotha (9/10) ilanga (9/10) sun
ihloma (5/6) izulu (5/6) heaven
isichopho (7/8) isitulo (7/8) chair
ubuyiso (11/10) ucango (11/10) door
amanyiso (6) amabele (6) udder
ethameni (descrip.) phandle (descrip.) outside
-weke (qual.) -mhlophe (qual.) white
ukunyambela (15) ukufaka (15) to put on
ukunoboka (15) ukufa (15) to die
ukunawuka (15) ukuhamba (15) to walk
ukuhuka (15) ukusenga (15) to milk
ukukhuluma (15) ukuthetha (15) to speak
ukumathela (15) ukubaleka (15) to run
The second is that consonant deletion found in traditional hlonipha does not
apply to the core vocabulary, for example:
Traditional hlonipha Xhosa English
andi-uni (pred.) andifuni (pred.) I do not want
uku-ina (15) ukuqina (15) to tighten
uku-ondela (15) ukusondela (15) to come nearer
The rest of this section describes features of traditional hlonipha that do occur
in the core sample. The core vocabulary contains retentions of Common Bantu
forms as reconstructed by Guthrie (1970):
Here the hlonipha word has come from the qualificative -luhlaza meaning
‘green’ or ‘fresh’, and thus associated with fresh milk. Hence also such words as:
In Xhosa iphoba means ‘that part of the head with hair’ and intlumayo ‘a very
small bean’.
Many verbal derivatives occur in the core vocabulary, such as:
Traditional and core
hlonipha Xhosa English verb Meaning
umbaso (3/4) umlilo (3/4) fire ukubasa to kindle
impungo (9) ikofu (9) coffee ukuphunga to sip
amagabuko (6) amehlo (6) eyes ukugabuka to clear up
umnabo (3/4) umlenze (3/4) leg ukunaba to stretch a leg
Interestingly, the Xhosa form isonka (7/8), ‘bread’, has its hlonipha equivalent
isiqhusheko (7/8), which could possibly have come from the verb ukuqhusheka,
‘to put under something’, which probably indicated the way of baking the
bread.
290 R. Finlayson
One of the most interesting facets of the lexical core of hlonipha is the coining
of new words and these abound in the core vocabulary, for example:
Core
hlonipha Xhosa English
ukumunda (15) ukutya (15) food/to eat
ukunawuka (15) ukuhamba (15) to walk
umolulo (3) utywala (15) beer
iwaku (5/6) icephe (5/6) spoon
educated and had grown up in Soweto, but was prepared to return to her in-laws
and take up the traditional customs. Her father-in-law blamed the influence of
Western culture for the fact that daughters-in-law no longer cared to hlonipha.
He commented: Yile mpucuko le ibangela kumke amasiko. Abantwana
abakhoyo abasihloniphi kuba bayakhumsha (‘It is this civilization which has
caused customs to go. The children here do not hlonipha because they speak a
foreign language’). However, Dowling (1988: 58) states:
As a custom however, hlonipha will persist because of its historical authority and legit-
imacy. Research that has been conducted in both the Ciskei and the Transkei indicates
that for many people its survival is desirable and important. Apart from what people
desire and consider important, however, there are other considerations involving factors
such as political change, imported values and syncretism, the implications and effects
of which being as yet not entirely predictable.
A strange dual life sometimes occurs where the daughter-in-law may be edu-
cated and working as a nurse in an urban area, returning home in the evening
and adopting once more her traditional attire and customs. This is accepted
by the family but proves very difficult for the daughter-in-law who cannot be
understood at work should she hlonipha, but who has to revert to hlonipha once
she is at home. This dual life cannot be expected to persist so will eventually
lead to the falling away of hlonipha – a sad event, one informant’s father-in-law
commented, as he felt that when tradition dies, the nation dies.
In many of the rural areas researched it appears that there are three distinct
categories of hlonipha users – the older group, who still hlonipha and strictly
uphold all the customs; the middle group, who have a partial retention of the
hlonipha vocabulary; and the younger set, who hardly hlonipha at all, and when
they do, include many words of English and Afrikaans origin. It has been found
that in some nuclear families, it is often the husband who will teach his wife
how to hlonipha. So while it appears that there is a distinct movement away
from hlonipha (see also Levin 1946), there is also some pressure to retain this
custom.
Pauw (1976: 198) notes that ‘it is probably a general feature of the urban-
isation of African peoples that customs, values and beliefs relating to certain
principles of social structure change more slowly than the structure itself’.
Those features of an institution demonstrating resistance to change will in-
evitably be seen to be the most resilient. However, although still retained by
many Xhosa women, the hlonipha custom is changing. According to Herbert
(1995: 61) many anecdotal reports exist ‘of situations in which individuals are
forced to violate a taboo’. He cites, for example, Kunene (1958: 165), who
described the frustration and difficulties experienced by post office workers
when they attempted to determine the name of someone whom the individual
was obliged to hlonipha. Herbert states that a further indication of the decrease
292 R. Finlayson
in the use of this custom was initially identified in Kunene’s research where
some of his informants reported that a woman could ‘whisper a taboo word to a
child’ if she fails to use the standard hlonipha form after which the child would
whisper the taboo word out loud. Finlayson in her research also notes that when
women were gathered together and a discussion was taking place, guesswork
would ensue as what a certain woman was trying to convey when there was
any confusion. All the women would participate in the guesswork which would
cause much mirth, sometimes ending up with the interlocutor having to utter
the tabooed key word quickly after which she would spit over her right shoulder
in order to appease the ancestors. Herbert (1995: 61) reports that the practice of
the avoidance of reference to names and food terms continues to exist, whereas
reference to spatial characteristics, articles of clothing and personal property
tends not to be found as frequently.
Many younger Xhosa speakers in rural situations do not practise the custom
of hlonipha at all, and, when they do, they use Afrikaans or English words.
In fact, many present-day students have been quoted as referring to hlonipha
as ‘like reading about a foreign culture’ (Herbert 1995: 62). With the changes
in hlonipha through the use of English and Afrikaans lexical items hlonipha
may become a language variety whose application generates another form of
identity for its users. Implications for research into code-switching can also not
be discounted.
It has been found over a period of some twenty years that more and more
English and Afrikaans words are being used in order to avoid the tabooed
syllables. From the original list of hlonipha terms (Finlayson 1984c), the
numbers of words of Afrikaans and English extraction were thirty-three and
twenty-seven out of sixty, respectively; more recent research has shown that
more words from English are being used. The rural communities have be-
come increasingly exposed to English and Afrikaans through the media, mobile
shops and, most importantly, through education. This influence of English and
Afrikaans on hlonipha is best illustrated in the following conversation which
took place in the Trappes valley area of the Eastern Cape province.
Nominiti Velani and Nomisile Myali are conversing about the problems of
the day.
The words in italics indicate borrowings from English and Afrikaans. In fact
a total of five words were used in two sentences from one woman in conversa-
tion, i.e.
The Xhosa word for ‘bottle’ is ibhotile, so here, in the hlonipha word, consonant
replacement has occurred, i.e. the breathy-voiced bilabial plosive has been
replaced by the breathy-voiced alveolar affricate in initial consonant position.
Until recently the coining of new words has been found to be mostly through
invention. English and Afrikaans now offer the user of hlonipha a rich source
of new words which may be used to avoid the tabooed syllables. Generally the
words borrowed from English and Afrikaans fit into the semantic categories of
clothing, household utensils and food. The majority of the women interviewed
could not speak either English or Afrikaans but must have been exposed to these
languages. Even some of their names are of English or Afrikaans origin, e.g.,
294 R. Finlayson
4 CONCLUSION
Lifestyles today in the urban areas make it virtually impossible for the women
to retain the custom of ukuhlonipha umzi wabo, ‘the custom of respecting the
homestead’. The custom involves the initial marriage negotiations when the
suitor has his bride negotiated for him while he remains at home. They also
involve the young girl’s avoidance of her suitor’s home during all the marriage
transactions, until the woman begins life in her new home. Such customs have
undergone drastic changes. A young girl used to accompany the bride to enable
the latter to communicate through the girl in her new home, but this no longer
happens as, in most cases, the bride will live in a home away from her in-laws
and will not have to worry about linguistic as well as physical avoidance of
things pertaining to them.
There is no doubt that hlonipha is still practised in many rural communities
today, but there is a change in the nature of this form of hlonipha. Previously
women who did not uphold this tradition were ridiculed and ostracised, but
today this does not generally happen. In the urban areas the converse occurs –
women are ridiculed for upholding the tradition. Such women would be con-
sidered uneducated. Many schoolchildren consider the whole concept a joke.
As Dowling (1988: 145) notes with regard to its survival: ‘This is a language
that will survive only as long as certain other institutions survive. This will
require social stability, a world view that is firmly based in oral culture and
a patriarchal ordering of society uninfluenced by any feminist perspectives or
demands.’
In this modern world of ours, there appears to be no time for the finer details of
customs of respect. People must answer for themselves and answer quickly. No
bureaucrat has time to decipher the intricacies of linguistic avoidance patterns
and thus, as the basis for a tradition changes, so the tradition itself falls away
and in some areas it begins to die. A strong case therefore exists for the accurate
documentation of this unique and changing tradition.
note
The original paper ‘The changing nature of isihlonipho sabafazi’ appeared in African
Studies, 43, 2, 84. This chapter is an updated revised version.
Women’s language of respect: isihlonipho sabafazi 295
bibliography
Dewar, D., A. Todes and V. Watson 1982. ‘Theories of urbanization and national settle-
ment strategy in South Africa’. University of Cape Town, Urban Problems Research
Unit, May, Working Paper No. 21.
Dowling, T. 1988. ‘Isihlonipho Sabafazi – the Xhosa Women’s Language of Respect. A
Sociolinguistic Exploration’. MA thesis, University of Cape Town.
Finlayson, R. 1978. ‘A preliminary survey of hlonipha among the Xhosa’. Taalfasette,
24, 2: 48–63.
1982. ‘Hlonipha – the women’s language of avoidance among the Xhosa’. South
African Journal of African Languages, 1, 1, Supplement: 35–60.
1984a. ‘The changing nature of isihlonipho sabafazi’. African Studies, 43,
2: 137–46.
1984c. ‘English and Afrikaans in hlonipha’. Unpublished paper presented at the
African Language Association of Southern Africa Seminar, Windhoek.
1986. ‘Linguistic terms of respect among the Xhosa’. In P. E. Raper (ed.), Names 1983.
Proceedings of the Second Southern African Names Congress: Pretoria: Human
Sciences Research Council, pp. 128–38.
1993. ‘The changing face of Xhosa’. In R. Hill, M. Muller and M. Trump (eds.),
African Studies Forum, vol. II. Pretoria: Human Sciences Research Council,
pp. 175–194.
1998. ‘The linguistic implications of the Xhosa Initiation Schools’. Language Matters,
29: 101–16.
Guthrie, M. 1970. Comparative Bantu. London: Gregg Press.
Herbert, R. K. 1990. ‘Hlonipha and the ambiguous woman’. Anthropos, 85: 455–73.
1995. ‘The sociohistory of clicks in Southern Bantu’. In R. Mesthrie (ed.), Language
and Social History. Cape Town: David Philip, pp. 51– 67.
Hoernlé, A. W. 1946. ‘Social organisation’. In I. Schapera (ed.), The Bantu-speaking
Tribes of South Africa. Cape Town: Maskew Miller, pp 67–94.
Hunter, M. 1961. Reaction to Conquest. London: Oxford University Press.
Jonas, P. J. 1972. ‘Die veranderende posisie van die vrou in die huwelik en gesin
by die stedelike Xhosa van Oos-Londen, met besondere verwysing na die dorp
Mdantsane’. MA thesis, University of South Africa.
Kropf, A. and R. Godfrey 1915. A Kafir–English Dictionary. Lovedale: Mission
Press.
Kunene, D. P. 1958. ‘Notes on hlonipha among the Southern Sotho’. African Studies,
17: 159–82.
Kuper, H. 1982. Wives for Cattle: Bridewealth and Marriage in Southern Africa. London:
Routledge & Kegan Paul.
Levin, R. 1946. ‘Marriage in Langa Native Location’. MA thesis, University of Cape
Town.
Mayaba, P. M. 1972. ‘A discussion of the hlonipha language among the Xhosa’. BA
Hons. article, University of South Africa.
Mqotsi, L. 1957. ‘A Study of Ukuthwasa’. MA thesis, University of the Witwaters-
rand.
Myburgh, A. C. 1942. ‘EzakwaZulu, ‘n Volkekundige beskrywing van die Zoeloe in die
volkstaal’. Unpublished paper.
296 R. Finlayson
Robert K. Herbert
The peoples and languages of southern Africa soon attracted the attention of
linguistic and cultural evolutionary theorists, who saw southern African hunters
and their languages as representing ‘primitive types’. The view, first expressed
by van Ginneken (1911: 346–7), that clicks were the phonetic material from
which human language first arose was developed and ardently championed by
the Polish linguist Roman Stopa (1935, 1979). Clicks were seen as arising from
‘the condensed expression of the gesticulatory part of speech’ (Stopa 1979:
28). Both van Ginneken (1938) and Stopa proposed comprehensive theories
whereby clicks have evolved into the diversity of human speech sounds. This
approach mirrors Bleek’s much earlier (1869) view that ‘Those languages . . .
in which the sounds are easiest of utterance are the farthest removed from the
primitive phonetic systems [i.e. San languages] of human speech’ (cited in
Theal 1910: 27). Attempts to discover the ‘origin’ of click sounds in Khoesan
and to discern developmental links between click and non-click consonants
have been largely abandoned.
With regard to the Bantu languages of southern Africa, it has long been recog-
nised that the click consonants are not reflexes of inherited elements; rather,
the clicks were ‘borrowed’ from Khoesan contact languages and incorporated
within Bantu phonological systems at some point during the prehistory of
297
298 R. K. Herbert
southern Africa. Within the Bantu languages, clicks have been most widely in-
corporated within the Nguni subgroup, which includes several major languages.
The relations among the various Southern Bantu languages are represented and
their relative geographic distribution are indicated on map 15.1.
Both Xhosa and Zulu exhibit a three-way opposition: dental [/], (pre)palatal
[!] and lateral [//]. Other Bantu languages display either only the dental and
palatal or only the palatal click. That is, possible Bantu language click inven-
tories include [/, !, //], [/, !] and [!]; such distributional considerations dictate a
relative markedness of clicks within Bantu: lateral → dental → palatal.1 In ad-
dition to place of articulation, various manners of articulation are distinct within
the class of clicks, including the oppositions plain:breathy:aspirated:nasalised.
It is frequently estimated that about 15 per cent of Xhosa and Zulu words ex-
hibit clicks; the vast majority of these are words of demonstrable or presumed
Khoesan origin, but there are examples where a click inexplicably substitutes for
an inherited Bantu consonant. According to Lanham (1964), between twenty-
one and twenty-five of the fifty-five Xhosa consonants are non-inherited and
confined almost exclusively to the borrowed vocabulary.
sociolinguistic avoidance custom provided the major impetus for click incor-
poration in Bantu languages; this argument is developed in section 3 below.
There is wide agreement now that the myth of the ‘invading Bantu male’
has been seriously overplayed in the literature (e.g. Harinck 1969; Marks 1969;
Ownby 1981, 1985; Wilson and Thompson 1969). Traditional ethnography of
the area has come under attack for a number of reasons. First, there has been a
tendency to treat the Bantu-speaking ‘tribes’ as monolithic units migrating and
displacing other peoples with abandon. Such mass migrations generally occur
less often than ‘the sporadic progressions of a set of segmentary interrelated
parties’ (Nurse et al. 1985: 64). Second, the ‘angry man’ theory (Ownby:1985:
32ff.) of a hostile relationship between the Nguni and Khoesan populations
is simply untenable.2 In place of this view, the Khoesan–Bantu relationship is
seen as a symbiotic one, characterised by frequent and intimate interaction over
several centuries in several domains, such as trade and intermarriage. As has
been noted in several publications, the Khoe and the Xhosa, for example, were
culturally compatible, not only in such areas as social and political organisation
but also in a high value orientation towards pastoralism expressed in an elabo-
rate cattle cult associated with the veneration of ancestors (Harinck 1969: 147).
Additionally, the nature of loanwords in Nguni languages (cf. Louw 1977; also
Werner 1902), in the socio-economic and ritual spheres, is incompatible with
the traditional ethnographic vision of the Khoesan–Bantu relationship. In place
of ‘invading Bantu males’, one needs to consider ‘Khoesan–Bantu composite
groups which existed well into the nineteenth century, if not the twentieth, all
over southern Africa’ (Marks 1969: 134). Though the usual pattern seems to
have been for Khoesan peoples to be assimilated within Bantu-speaking groups,
the reverse pattern also occurs. Harinck (1969: 157–9) discusses the assimi-
lation of leaderless Xhosa refugees into Khoe chiefdoms, where subsequent
generations spoke a ‘mixture of Khoi and Xhosa, with Khoi predominating’. It
seems most likely that the relationship between Khoesan and Bantu speakers
only became hostile some time in the nineteenth century when regular raids on
Nguni cattle began.
Thus, Bryant’s explanation of the incorporation of clicks within Nguni, of-
fered in the context of a discussion of Nguni migrations, is deficient in a number
of respects:
Here a new and difficult problem confronted them – tiny yellow men, more wily than
themselves, more treacherous and aggressive than the beasts, contested their very rights
to cattle, land and life. They must now perforce either fight, be pauperized, or die; and
so this endless warfare with the pygmy foe, while causing a marked recession in all arts
and industries of peace, trained them into a warrior race. Captured Bushwomen became
common in their homes as concubines and slaves, and sometimes, it is plain, as mothers.
And the children, ignorant of the consequences, adopted as their own, but in a Bantuized
form, much of the slave-girl’s speech and grew up with it on their tongues. Hence the
clicks in Nguni speech. (1929: 5)
The sociohistory of clicks in Southern Bantu 301
In addition to the inaccuracy and racism of the sketched relationship, this lin-
guistic explanation fails to offer any coherent reason why the nature of Khoesan
influence on Bantu is so restricted. It is clear, for example, that modern Zulu
is not a Khoesan language ‘in a Bantuized form’ (cf. section 3 below). Further,
the social asymmetry of the contact relationship would argue against Bantu lan-
guages borrowing heavily from Khoesan since, as noted by Moravcsik (1978:
109), ‘nothing can be borrowed from a language which is not regarded to be
prestigious by speakers of the borrowing language’.
A contributing factor in the incorporation of Khoesan sounds into Bantu
phonological systems must have been the very distinctive acoustic quality of
clicks. Clicks are perceptually sharp and distinct as a class, although to the
untrained ear there is much confusion within the class. The particular brand of
bilingualism present in the contact situation, the distinct quality of the clicks,
and the absence of any inherited Bantu sound types with which they might be
matched are all factors that contributed to their borrowing.3
the following derivational morphemes to Khoe sources: -se, -she (< Khoe -s, a
feminine suffix) used for a variety of functions in Xhosa; -sholo (<soro, ‘bad,
ugly, coarse’) used for derogation. Most of these incorporated morphological
formatives are non-productive. Thus, if one assumes some very extensive and
intense brand of bilingualism in order to explain the borrowing of such a large
number of consonants, one is hard pressed to explain why that bilingualism had
so little effect elsewhere in the recipient languages, for example in the vowel
inventory.
A third surprising aspect of the Khoesan–Bantu contact is that the borrowed
consonants occasionally appear in inherited Bantu lexical items. Consider in-
ternal correspondences such as Zulu kh:xh as exhibited in -xhopha, ‘to hurt
the eye’, vs. ukhophe, ‘eyelash’, and ukhopho, ‘a person with deep-set eyes’;
c:th as in -consa, ‘fall, drip, leak’ vs. ilithonsi, ‘a drop of liquid’. Comparative
forms occasionally show the same bizarre correspondences: -cima (Proto-Bantu
*-lima), ‘extinguish’ (cf. Sotho – tima). More commonly, both the inherited
Bantu form and a modified form with click coexist with differentiated mean-
ings, such as -cwazimula/-nyazimula, ‘to shine brightly/to flash, shine’; -chela/-
thela, ‘to sprinkle (ceremonially?)/to pour, pour out’; -qhuma/-duma, ‘to burst,
explode, pop/to thunder, rumble, reverberate’.5
Nguni languages thus exhibit a type of contact borrowing that appears most
unusual in terms of the intensity (yet very restricted nature) of the linguistic
effects. The usual explanation advanced for this extraordinary situation refers
simply to bilingualism and to the type of social contact between the Khoesan
and Bantu-speaking populations. Scholars have pointed to the primary influence
of the mother (Faye 1923–5; Lanham 1964) and suggested that this explains
consonant incorporation. Were this an adequate explanation, we should expect
to find Khoesan influence elsewhere in the phonological system and in the
grammar. Hagège and Haudricourt (1978: 112) suggest that phonetic invento-
ries are increased through language contact whenever a borrowed word fails to
undergo loan phonology: ‘Les phonèmes, étrangers à la langue emprunteuse,
qui peuvent faire partie du signifiant des mots empruntés, s’y introduisent tout
naturellement en même temps que ces mots’ (sounds that are foreign to the
borrowing language but part of the phonetic form of borrowed words are in-
troduced naturally and at the same time as the words themselves (emphasis
added)). Such a view is hard pressed to explain why universally highly marked
sound types are incorporated when less marked phonetic elements are not.
Further, this approach neglects the often cited observations (e.g. Weinreich,
Jakobson) that languages borrow linguistic elements only when these elements
correspond to internal tendencies of development, e.g. filling phonetic gaps, and
that lexical borrowings are neither sufficient nor necessary to produce phono-
logical borrowing.6 It is precisely the peculiar distributional fact of Southern
Bantu borrowing effects that requires explanation.
The sociohistory of clicks in Southern Bantu 303
I have argued elsewhere that the extent of Khoesan linguistic influence was
greater still and that contact accounts for the Southern Bantu opposition of as-
piration and ejection in the absence of a phonetically unmarked series of plain
voiceless consonants (Herbert 1987; cf. Louw 1986). In this analysis, the effects
of contact included not only the enormous influence on the consonantal inven-
tory and the lexicon, but also the development of a Southern Bantu ‘articulatory
mode’, which involves a predilection for glottalic consonants, clicks, aspirates,
and so forth. The concept of articulatory mode, as it has been developed by
British phoneticians, requires some further elucidation; in the present case, it
is suggested that Southern Bantu languages (apart from Shona) operate with a
phonetic mode in which ejective quality is characteristic of otherwise unmarked
voiceless consonants. Whatever the theoretical status of such a concept, there
is good reason to believe that the Southern Bantu articulatory mode is indeed
a Khoesan influence, especially since the geographic extent of this phonetic
mode corresponds to the range of heavy lexical borrowing from Khoesan. For
example, all of the Southern Bantu languages borrowed from Khoesan (directly
or indirectly) words for ‘cow’, ‘sheep’ and ‘milk’ – except for Shona, which
shows Bantu reflexes for these items and lacks the borrowed articulatory mode
as well (Westphal 1963: 253 ff.; Wilson and Thompson 1969: 104).
Hlonipha (discussed more fully in Finlayson: chap. 14, this volume) is the name
given to a range of social avoidance customs practised by Nguni speakers. The
dictionary definition of the term is something like ‘respect through avoidance’,
covering a wide range of behaviours, especially those expected of married and
engaged women. The general use of the term in the sociolinguistic literature
is restricted to a linguistic taboo process whereby women are barred from
pronouncing the names of their fathers-in-law and other senior male affines.9
Among traditional Xhosa and Zulu speakers, it is not only the name itself that
must be avoided, but also any of its composite syllables. Thus, a woman whose
father-in-law is named Bongani must avoid the name itself and the syllables
306 R. K. Herbert
bo and nga – wherever they occur in speech.10 Since it is not only the name
of the father-in-law but those of all senior male affines and the mother-in-law
that must be avoided, the effect on each individual woman’s speech may be
dramatic.
A variety of linguistic mechanisms are used to achieve avoidance, including
consonant deformation (substitution) (e.g. ulunya, ‘cruelty’ → ulucha), ellipsis
(e.g. umkhono, ‘foreleg’ → um’ono), synonymy (e.g. kufa, ‘to die’ → kushona,
‘to set; to die’), derivation (inkhuleko, ‘thing for tethering’ for imbuti, ‘goat’ <
kukhuleka, ‘to tether’), as well as neologism, archaicism and borrowing. What
should be noted is that the majority of practices involve lexical substitution, i.e.
the replacement of one word with another; in some geographic areas, lexical
strategies have entirely replaced non-lexical ones, particularly among younger
speakers. There is good reason to believe, however, that phonetic strategies of
syllable deformation (including consonant substitution and elipsis) represent
original hlonipha practices (Herbert 1990a: 460ff.).
It is reasonable to conclude that the process of hlonipha itself is the essen-
tial part of the explanation for click incorporation in Southern Bantu.11 There
is no way to understand the intensity and restrictedness of Khoesan influence
without recourse to some very peculiar aspect of the social contact situation.
Specifically, it is argued that the native (i.e. Khoesan) phonological invento-
ries provided Khoe, San and Nguni women with a ready-made and ‘natural’
source for consonant substitutions as required by hlonipha. That is, it is in
some sense natural that a woman who enjoys a prohibition against uttering
the syllables bo, nga, ni, di, ke, sa, etc. would look to this alternative phonetic
inventory in order to replace Nguni consonants. Bear in mind here that the pre-
contact Nguni consonant inventory was relatively small. The substitution of a
foreign element such as a click is perceptually salient and deforms the offending
syllable acceptably. Furthermore, the use of non-Bantu consonants for this pur-
pose precludes the possibility of the deformed word being homophonous with
some other pre-existing word in the lexicon. The pre-contact phonologies of the
relevant Bantu languages were quite simple in terms of consonant and vowel
inventories and syllable-structure constraints. The existence of an extraordinary
phonological inventory that could be invoked in hlonipha therefore served an
important sociolinguistic function.
A number of advantages derive from the preceding explanation. First, the
presence of click consonants in inherited Bantu words is explained. The seem-
ingly random substitution of a click for an inherited consonant represents the
‘fixing’ of a hlonipha form. This idea is far from novel. Faye (1923–5) lists a
number of examples of such fixed forms. Faye also notes that the replacement
of an inherited Bantu word with a hlonipha alternative is rare, whereas the co-
existence of both forms – with a semantic differentiation – is more common.
The sociohistory of clicks in Southern Bantu 307
practised in those Southern Bantu languages that show limited click incorpo-
ration, such as Tsonga, which shows clicks only in Zulu borrowings and in
ideophones.
The connection between hlonipha and consonant incorporation in Bantu lan-
guages is further supported by the non-click consonants that act as favoured
substitutes in hlonipha. For example, Finlayson (1982: 49) notes that two of
the most common consonant substitutes in Xhosa are ty [c’] and dy [j], neither
of which is a reflex of Proto-Bantu consonants. Thus, their preferred status in
hlonipha is like the status of clicks – that is, they became established as preferred
substitutes precisely because they did not occur in native Bantu words. Also,
in earliest times (i.e. before they were incorporated into the Bantu languages),
these foreign consonants did not themselves require avoidance: they did not
occur in Bantu names. Lanham (1964: 389) noted that some borrowed words
with palatals correspond to the Khoesan dental click. Such correspondence may
reflect something about the shared extraordinary status of these borrowed sound
types. Lanham vacillates between listing twenty-one or twenty-five Xhosa con-
sonants as Khoe borrowings; it is the status of the palatals that is in doubt.
Two obvious questions that require asking in the context of the above propo-
sals have to do with the ‘fixing’ or standardisation of hlonipha terms. Given that
the names to be avoided varied from one woman to another, how does an inher-
ited Bantu form come to be displaced by a hlonipha form? Residence among
the Nguni is patrilocal, and wives in the affinal homestead would therefore
share a significant number of male in-laws whose names required avoidance.
How though would a hlonipha term become established enough on a wide scale
to acquire a separate semantic identity and diverge from its original role as a
simple avoidance form? There are no good answers to either of the above ques-
tions. The simplest case to understand would be that in which a particular name
was taboo for an entire large community, but such cases are relatively few in
number. For example, among Zulu groups the name of the great chief Shaka
was universally taboo; thus, Zulu speakers would not utter the word -shaya,
‘to hit’, or -shanela, ‘to sweep’ (J. de N.R. 1899/1900: 446).13 Similarly, Soga
cites the example of one Xhosa clan, the AmaBamba, for whom the name of
a distinguished ancestor named Tangana, ‘Little Pumpkin’, is taboo: ‘Hence
every Bamba woman, whether she be by blood of the clan or has married into
it from some other clan, observes the hlonipa custom in connection with the
name “tanga”– “pumpkin”. So that no woman of the clan ever speaks of the
pumpkin vegetable as “itanga” but as igabade’ (1932: 209).
Mzamane (1962: 256) noted that hlonipha words are often standardised
within a large area and there are numerous reports of young wives being in-
structed in the appropriate hlonipha of the homestead. In a discussion of the
general question of standardisation, Kunene (1958: 163) observes that ‘in actual
fact, however, there is so often a sameness or similarity of family names’. He
The sociohistory of clicks in Southern Bantu 309
also notes that Southern Sotho informants will argue about the ‘correctness’ of
a hlonipha substitute; this too points to an unconscious effort to regularise the
alternate vocabulary.14
5 CONCLUSION
This chapter has considered the very unusual language contact influence of Khoe
and San languages on a subset of the Southern Bantu languages, with particular
attention to the issue of click incorporation. What makes the Southern Bantu
case unique in the historical sociophonology literature is the essential role of
a sociolinguistic taboo in an extensive restructuring of the sound system of a
language.
There are several remaining points that deserve mention in this context.
First, I do not claim that all of the Khoesan words appearing in any Southern
Bantu language are hlonipha forms for taboo Bantu words. Rather, the claim
is that the practice of hlonipha ‘primed’ the language to be receptive to click
incorporation. That is, young children – even in earliest contact times – were
exposed to varieties of language in which clicks were regularly employed. All
scholars agree that the young child’s primary linguistic influence would be
the mother, and children would therefore acquire from their mothers words
that included clicks, though these words varied somewhat from one area to
another. Consider, too, the various reports of young girls ‘practising’ syllable
avoidance so that they could comply with the custom after marriage, especially
in the light of the real or imagined consequences of hlonipha violation, such
as insanity (Mncube 1949: 47), baldness (Soga 1932: 209), stillborn children
(Raum 1973: 12) and infertility (Finlayson 1984: 143), as well as the more
general ‘risk of death, madness, maladjustment to life or some kind of tragedy
or malady’ (Mzamane 1962: 231). Clicks may originally have been restricted
to a supplementary vocabulary – a vocabulary recognised as being set outside
‘normal language’. However, over the course of time, the special phonological
status of Khoesan consonants disappeared or was blurred, and these consonants
310 R. K. Herbert
were absorbed into the native inventory, leading the way for borrowings from
Khoesan to be taken over with these consonants intact, although other aspects of
the lexical shape, such as vowel sequencing, vowel nasality, syllable structure,
etc. were subject to loan phonology.
A puzzle for comparative Southern Bantu linguistics has been posed by the
fact that there is so little overlap in the actual click words appearing in the
various individual languages. Bourquin (1951) examined 2,395 click words
in Xhosa, but only 376 (16 per cent) were shared with Zulu and/or Southern
Sotho. Apart from click words, approximately 80 per cent of the vocabular-
ies of Xhosa and Zulu is cognate. Whether the 376 common words represent
words borrowed in very early contact times (e.g. by speakers of Proto-Nguni) or
whether they represent borrowing from one Bantu language to another or par-
allel borrowings from Khoesan is an open question. Most probably, the terms
common to Xhosa and Zulu reflect early Nguni–Khoesan contact, whereas
those shared with Southern Sotho represent inter-Bantu borrowing or parallel
borrowing. This problem cannot be addressed in the present context, but it is
interesting to note that many of the terms common to Zulu and Xhosa are rather
unexpected borrowings, for instance, words for ‘to urinate’, ‘man’, ‘to stab’,
‘egg’, ‘full’, ‘bark’, ‘knee’, ‘navel’, ‘lake’, ‘name’, ‘swell’, ‘sing’, and so on
(cf. Ownby 1981). Such borrowings also point to the rather unusual nature of
Khoesan–Bantu contact.
A final question concerns the actual distribution of languages during the
earlier periods of Khoesan–Bantu contact. Wide-scale bilingualism is generally
assumed, and Wilson (1969: 80) suggests that the acquisition of the Nguni-
speaking father’s language points to the status of mothers as concubines rather
than wives. There is no compelling reason to believe that this was the case.
It is impossible to assert anything about the contact situation with complete
certainty. Reconstructing sociolinguistic history for southern African groups
will continue to pose a challenge to linguists and anthropologists in the region.
notes
Following established usage, the term ‘Bantu’ is used as a shorthand reference for
‘Bantu languages’ and, occasionally, ‘Bantu-speaking people(s)’; within South Africa,
the term is strongly offensive, but it is so established in the scientific literature that no
readily acceptable and recognised substitute is available. Similarly, the older terms
‘Bushman’ and ‘Hottentot’ are occasionally used here alongside the alternate terms
‘San’ and ‘Khoe’; the former terms are of non-African origin and have strongly
derogatory connotations. Unfortunately, the term ‘San’, itself a Khoe word, is also
derogatory and not readily accepted by the people named, who often prefer, pace
linguists and anthropologists, to call themselves Bushmen. Further terminological
problems arise from the use of all these terms to refer to cultural groups, physical
types and language units, the distribution of these three variables not necessarily
The sociohistory of clicks in Southern Bantu 311
coinciding. Some of the problems of distinguishing (and relating) Khoe and San
populations are discussed in Wilson (1986). Compare also Nurse et al. (1985: 79),
who note that a ‘good case can be made out for supposing the Khoi and the San to be
simply economically differentiated segments of the same people . . . There is evidence
that Khoi who lost their cattle reverted to a hunting way of life, while some San who
became pastoralists would merge into the Khoi population.’ The differentiation of
Khoe and San populations does not bear crucially on the arguments to be advanced
in this chapter.
The transcriptions used in published sources have generally not been modified
here. The usual symbols used for writing clicks in Khoesan languages include /, =| ,
!, // for the dental, alveolar, palatal and lateral types, respectively; the IPA has voted
to adopt the ‘Africanist’ set of symbols above. Bantu-language orthographies most
frequently employ c, q and x for the dental, palatal and lateral varieties.
1 Such considerations of relative markedness within the class of click consonants are
discussed in Herbert (1990b).
2 According to this view, the name ‘Xhosa’ is itself derived from a Khoi verb meaning
‘to destroy’. Harinck (1969: 152) cites Maingard’s tracing of the word to Kora //kosa,
‘angry men, the men who do damage’. There are alternate etymologies in the literature;
see, for example, Louw (1977: 139, 1979: 9, 18).
3 The argument is often made in the literature, but open to serious question, that non-
‘click incorporating’ languages in the area have occasionally managed to nativise
clicks, usually by substituting velars. Such a substitution would not be surprising,
given the essential role of the velum in the production of clicks. This pattern is also
observed in children’s acquisition of clicks (cf. Herbert 1990b), in the substitutions
made by students in introductory phonetics courses, and in historical change (Traill
1986; Herbert 1990b).
4 The stability of the Bantu noun classes and, especially, the system of concordial agree-
ment in language contact situations and in Bantu languages used as lingue franche,
is remarkable. These features seem largely resistant to pressures of ‘simplification’.
Reduction in class numbers and ‘semanticisation’ of concordial agreement are seen
in some cases (Herbert 1985).
5 As noted above, the most common lexical context for clicks is in borrowed vocab-
ulary, e.g. Zulu -qiqinga, ‘tie in a bundle’ < Korana !ai, ‘bind’; Z. -qhosha, ‘button,
fasten’ < K. !goi-is, ‘button’; Z. iqhubu < K. !hubu-b, ‘swelling on the body’; Z.
incuke < K. /hu-khã-b, ‘hyena’. Bourquin (1951: 75) also relates Zulu and Xhosa
-nci, ‘small’, to Korana /a, ‘small’, although he notes the existence of a Bantu stem
-nı̂, -nyı̂. Meinhof (1932: 103) derives -nci from the Bantu form and attributes the
click to hlonipha influence. The latter seems a more plausible line of development
and is the one more commonly cited in the literature.
6 Hagège and Haudricourt (1978) do cite these considerations elsewhere. Their ex-
amples (pp. 112–13) of phonological borrowing, however, involve the incorporation
of marginal elements, e.g. the velar nasal in French words such as smoking, living,
parking, rather than the full integration of a phonetic element within a sound system.
7 ‘Gene flow between San and Khoi appears to have occurred mainly from the former
to the latter; hence it is possible that a certain proportion of the San contribution to
the Negroes has occurred via the Khoi’ (Nurse et al. 1985: 131).
8 The clicks in Ndebele are generally thought to have been lost through extensive contact
with Tswana and other non-Nguni groups.
312 R. K. Herbert
9 Linguistic avoidances are also practised by men towards the names of their mothers-
in-law and, occasionally, other persons. Hlonipha refers generally to ‘name avoid-
ance’: it is usually accomplished by some lexical substitution. What distinguishes
women’s language behaviour vis-à-vis their male in-laws is the far-ranging effects
of avoidance and the diverse linguistic practices that effect avoidance of the name,
including consonant substitution and deletion. A general review of name avoidance
practices in Southern Bantu is given in Herbert (1990a).
10 There is much variation from one locale to another as to whether prefixal and suffixal
elements within names must be avoided in hlonipha. In some areas, the final syllable
ni might require avoidance. See Raum (1973) for a detailed description of the range
of Zulu hlonipha practices and variants.
11 Such a possibility is mentioned by Louw (1962), but he does not ascribe the im-
portance to the role of hlonipha suggested here. Also, he puts equal stress on the
use of clicks in onomatopoeic words in explaining their incorporation. I suspect that
the latter is of marginal influence since there are several languages that show clicks
in such expressives without any indication of clicks being incorporated into normal
phonology.
12 These two facts may have independent historical explanations, however. The rela-
tively impoverished click inventory of Swati may be due to speakers of pre-Swati
not having incorporated very many borrowed words on account of less contact with
San populations. Note that the admixture of Gm1,13,17 , representing incorporation
of Khoesan peoples, is considerably less in Swati (25 per cent) than in Zulu- or
Xhosa-speaking populations. As noted earlier, the number of borrowed words in
Nguni languages that can be traced to contact during the Common Nguni period is
quite small; most of the Khoesan words in Zulu and Xhosa were acquired after dif-
ferentiation of the Nguni dialects/languages. A second possible, though less likely,
explanation is that Swati previously had more click distinctions and that they were
lost when the ancestors of the present Swati population migrated to the area around
the Usutu river (modern Swaziland) from the east and ‘mingled with the “Sotho”’
(Nurse et al. 1985: 143) who were resident there. One must also recognise certain
Tsonga influences on Swati.
The lesser role of hlonipha in Swati is more likely due to the second of the above
factors, i.e. contact and incorporation of a non-Nguni Bantu-speaking population:
hlonipha needs to be reconstructed for the Proto-Nguni people since all of the modern
languages show some trace of the practice. Numerous other sociocultural features
distinguishing Swati and Zulu are also due to Sotho ‘influence’, e.g. cross-cousin
marriage, which is anathema to the Zulu and Xhosa (Wilson and Thompson 1969:
97, 159).
13 Werner (1905: 352–3) reported that certain Zulu animal names also enjoyed near-
universal avoidance, for example, ingwe, ‘leopard’, was replaced by isilo, ‘wild
beast’. Imfene, ‘baboon’ and impaka, ‘wild cat’ were similarly avoided. Werner
uses the term hlonipha to describe these taboos and others such as avoidance of
the names of certain mythological creatures. Although they obviously share certain
features, it seems advisable to distinguish these patterns and the linguistic avoidances
of Nguni wives.
14 The question of whether Southern Sotho hlonepha was already in a weakened form
at the time of Kunene’s fieldwork is an open one. Some support for this notion
comes from the fact that there were no real punishments for non-observance of
The sociohistory of clicks in Southern Bantu 313
the taboo; rather, hlonepha was viewed simply as a sign of ‘good upbringing’
(1958: 165).
15 Editor’s note: The author has graciously permitted the excising of a large section here
in the interests of space, and since the section is discussed by Finlayson in chapter
14. The excised text can be found in earlier versions of this article (Herbert 1990c
and 1995).
bibliography
Beach, D. M. 1938. The Phonetics of the Hottentot Language. Cambridge: W. Heffers
& Sons.
Bourquin, W. 1951. ‘Click words which Xhosa, Zulu and Sotho have in common’.
African Studies, 10: 59–81.
Bryant, A. T. 1929. Olden Times in Zululand and Natal. London: Longmans.
Faye, C. U. 1923–5. ‘The influence of “hlonipa” on the Zulu clicks’. Bulletin of the
School of Oriental and African Studies, 3: 757–82.
Finlayson, R. 1982. ‘Hlonipha – the women’s language of avoidance among the Xhosa’.
South African Journal of African Languages, Supplement: 35–60.
1984. ‘The changing nature of isihlonipho sabafazi’. African Studies, 43: 137–46.
Hagège, C. and A. Haudricourt 1978. La Phonologie panchronique. Paris: Presses
Universitaires de France.
Harinck, G. 1969. ‘Interaction between Xhosa and Khoi: emphasis on the period 1620–
1750’. In L. Thompson (ed.), African Societies in Southern Africa. New York:
Praeger, pp. 145–69.
Herbert, R. K. 1985. ‘Gender systems and semanticity: two case histories from Bantu’.
In J. Fisiak (ed.), Historical Semantics/Historical Word-Formation. Berlin: Mouton
de Gruyter, pp. 171–97.
1987. ‘Articulatory modes and typological universals: the puzzle of Bantu ejectives
and aspirates’. In L. Shockey and R. Channon (eds.), Festschrift for Ilse Lehiste.
Dordrecht: Foris, pp. 401–13.
1990a. ‘Hlonipha and the ambiguous woman’. Anthropos, 85: 455–73.
1990b. ‘The relative markedness of click sounds’. Anthropological Linguistics, 32,
1–2: 295–315.
1990c. ‘The sociohistory of clicks in Southern Bantu’. Anthropological Linguistics,
32, 3–4: 120–38.
1995. ‘The sociohistory of clicks in Southern Bantu’. In R. Mesthrie (ed.), Language
and Social History: Studies in South African Sociolinguistics. Cape Town: David
Philip, pp. 51–67.
J. de N.R. 1899/1900. ‘Een week in Kafferland.’ Ons Tijdschrift (February): 441–8.
Jacottet, E. 1896. ‘Moeurs, coutumes et superstitions des Ba-Souto’. Bulletin de la
Société Neuchâteloise de Géographie, 9: 107–51.
Kunene, D. P. 1958. ‘Notes on hlonepha among the Southern Sotho’. African Studies,
17: 159–82.
Lanham, L. W. 1964. ‘The proliferation and extension of Bantu phonemic systems
influenced by Bushman and Hottentot’. Proceedings of the Ninth International
Congress of Linguists. The Hague: Mouton, pp. 382–91.
Louw, J. A. 1962. ‘The segmental phonemes of Zulu’. Afrika und Übersee, 46: 43–93.
314 R. K. Herbert
Robert K. Herbert
1 INTRODUCTION
316
The political economy of language shift 317
intelligible languages. Thus, while it is true to argue, as Harries has, that Swiss
missionaries ‘invented’ the Tsonga, the raw materials for this invention included
diverse peoples whose everyday speech was sufficiently similar to allow for a
single written standard language to be developed.
Within South Africa, there are two diverse groups who are generally in-
cluded within a broad scope of ‘the Tsonga-speaking peoples’: (a) the groups
in the north-eastern Transvaal (formerly the Gazankulu homeland, now in
Mpumalanga and Northern Provinces), usually described as eighteenth- and
nineteenth-century coastal immigrants; and (b) a small group resident in
northern KwaZulu-Natal. Beyond a shared linguistic heritage, these two groups
have little in common. The former group, sometimes classed as Gwamba (Doke
1954), is actively Tsonga in language and custom, and is found in several
Tsonga-dominant rural areas in the former Gazankulu (Grobler et al. 1990),
and in several major urban centres. The latter group, part of the Ronga division,
is represented among the peoples resident south of the Mozambique border,
the area once known as Maputaland, part of former Thongaland, which was
formally incorporated into the KwaZulu ‘homeland’ as part of the Ingwavuma
318 R. K. Herbert
district in 1982. This chapter concerns the latter group of people, often known as
the Tembe–Thonga, and focuses in particular on the changing role of
language(s) in attributions of identity.
2 THE TEMBE−THONGA
The current linguistic situation is this area is complex. For longstanding polit-
ical reasons, the region is generally classed as ‘Zulu speaking’ (e.g. Grobler et al.
1990). Ngubane (1992), for example, identifies the local language as isiZulu
sase Nyakatho, which he glosses as ‘northern Zululand Zulu’, or isiNyakatho,
‘northern language’. On the other hand, Kubheka (1979) recognised an ad-
mixture of Swati and Thonga, with generally stronger Thonga influences on the
eastern side of Ingwavuma, and Swati influences in the west. The label isiTembe
is sometimes used to name a variety of language with strong Thonga features.4
The language traditionally spoken by the Tembe–Thonga is known as Thonga
or Ronga, the latter term being used by Junod (1896, 1927) to name the
southernmost of the six Thonga groups.5 These six groups may conveniently
be reduced to three ‘tolerably well-defined sections’ (van Warmelo 1974: 69):
northern group: Hlengwe [Tswa (and others)]
central group: Nwalungu, Bila, Hlanganu, Djonga
southern group: Ronga
It is likely that neither of the labels ‘Thonga’ and ‘Ronga’ was originally endony-
mous for any group. The Tembe–Thonga claim to have originated in Zimbabwe,
but there is no doubt that they have been in Mozambique since at least the
sixteenth century. Zulu incursions in the nineteenth century resulted in a series
of southward migrations. Tradition holds that Thongaland was first occupied
by the Tembe during this period. The Tembe lineage is traced back more than
ten generations to the founding chief, Tembe.
The term ‘Thonga’/‘Tsonga’ has a disputed etymology, often given as a Zulu
word for ‘slave’ or as relating to ‘east/dawn’ (Junod 1905: 223; Felgate 1982: 9).
Felgate reported that the term was resented and not used locally at the time of
his fieldwork in 1964–5; my own experience twenty-five years later is that the
same people do occasionally self-identify as Thonga, most especially when
they choose to deny their links to Zulu identity and hegemony.6
The Tembe clan predominates numerically and politically in the area. Indeed,
they sometimes claim to be ‘the original owners’ of the land, although this
claim is disputed by the next-strongest clan, which calls itself Ngubane. The
Ngubane reserve the ‘original-owner’ distinction for themselves, and claim to
have been dominated by in-migrating Tembe people, who were assisted by ‘the
Europeans’. Whatever the historical facts of early occupancy, both groups prob-
ably represent Thonga groups that migrated southwards from Mozambique.7
The linguistic situation among Tembe-Thonga is very interesting. Despite the varying
ethnic origins of the people, the languages spoken are exclusively Zulu and Thonga, with
Zulu being predominantly the language of men, and Thonga the language of women.
On the South African side of the border men never speak Thonga . . . On both sides of
the border women speak Thonga almost exclusively. It is not at all uncommon to find
men addressing women in Zulu and the women answering them in Thonga.
These same facts are echoed by Webster (1989) and are consistent with ear-
lier descriptions by van Warmelo (1935) and Allison (1951); cf. also Junod
(1896: 6), who noted that the Ronga spoken in this area was ‘un language
intermédiaire entre le ronga et le zoulou’.
Certainly, a historical explanation for the presence of Zulu within the group
is not hard to find. As noted above, there were steady Zulu incursions into
Mozambique during the Shakan period and there was a Zulu influence of long
duration. Zulu formed the prestige group, and their language was a prestige
language. There is also good evidence of trading between Zulu and Thonga
groups for a period long before Shaka’s ascendancy; Thonga men probably
first learned Zulu in this context. Later, in the nineteenth century, there may
have been further pressure for men to speak Zulu as a result of what Junod
(1927: 33) called Zulu ‘despotic domination’ of the Thonga. Junod himself
noted the fact that women were not learning Zulu and that women ‘are the best
safeguards of the purity of the language’, an idea promoted by Jespersen (1922)
and several pre-feminist generations of linguists. Women’s non-shift to Zulu
and their historic maintenance of Thonga has gone largely unexamined in the
literature, which has focused on language shift among Thonga men. Certainly
the facts as they have been presented here are not extraordinary. What makes
the Tembe–Thonga case interesting from a linguistic perspective is not the
association of one sex or the other in taking the lead in language shift. Rather,
the case warrants closer investigation on account of the more than one hundred
and fifty years of sex-determined bilingualism in the area. (Cf. also Bryant 1929:
292.) Perhaps the most intriguing aspect of the linguistic situation is, however,
an ongoing reinterpretation of the sex-differentiated behaviour, in which cultural
patterns are reproduced while the outward vehicles of expression, the languages
of Thonga men and women, are shifting.
4 LANGUAGE SHIFT
An initial question here, however, must concern the accuracy of historical re-
ports of language use among the Tembe–Thonga. The earliest reliable report
seems to be van Warmelo’s A Preliminary Survey of the Bantu Tribes of South
Africa (1935), in which he noted that the VaTonga, known to the Zulu as
abakwaTembe or abakwaMabhudu (= Maputu), ‘have adopted Zulu custom
and language to a far extent and must be mentioned among the “Natal Nguni”
for that reason’ (1935: 81; emphasis added). Later, van Warmelo noted that
‘I have met members of the tribe who understand practically nothing of Tonga.
The customs observed by such are also more likely to be Zulu’ (1935: 91). At the
same time, older descriptions of the area are complicated precisely because it
was a linguistic border zone. The Language Map of South Africa (van Warmelo
1952) clearly indicates the northernmost portions of the district as ‘Tsonga-
speaking’, with a mixed Zulu–Tsonga zone to the south. On the Maputaland
322 R. K. Herbert
16.3b Domain of the Thonga language (van Warmelo 1935); note the spread
of Zulu
doctors in the area, many of whom were undoubtedly Swiss, learned and used
Thonga in everyday communication. These residents reported that the foreign
men’s use of Thonga, rather than Zulu, was a source of amusement to young boys
at the time. Assuming this period to have been in the 1920s provides indirect evi-
dence for sex-based distribution of language already being well established at
that time. The striking sex-based distribution of languages was evident even to
324 R. K. Herbert
outsiders: ‘Most of the people living in this area use two languages, Thonga and
Zulu. Thonga is the language used in the home by the women and children, but
Zulu is the “official” language, and the language of the men’ (Allison 1951: 7).9
Both Felgate (1982) and Webster (1989) note that the South African migrant
labour situation would reinforce the association of men with Zuluness. Zulu
serves, to a certain extent, as the lingua franca among much of South Africa’s
ethnically diverse population and it is the African language of prestige through-
out much of the country. Thonga, on the other hand, is a distinctly non-prestige
language, reflecting the non-prestige status of Thonga ethnicity in the wider
national context. Men who seek work therefore may feign Zuluness in order to
improve their status in the employment setting. The prestige of Zulu explains
the tendency for men to change their surnames, e.g. many Tembe men have
taken the Zulu Mtembu as their name, especially when at work. Similarly, the
Gubande, who are now often known as Ngubane, often attribute the presence of
individuals named Gubande, the original form of the name, in their genealogies
to European mishearings and faulty transcriptions. There are some individu-
als whose own recounting of their genealogies will include both original and
re-formed varieties of the names.
In addition to language shift, Thonga men have abandoned their participation
in agriculture, which is seen by the Zulu as a female activity. In Mozambique,
Tembe men participate in agriculture (Junod 1927; Felgate 1982) and often tend
their own gardens. Yet it would be wrong to assume that Zulu influence has
been pervasive throughout all areas of sociocultural organization. The system of
land tenure, homestead arrangements, marriage laws, ritual life and taboos, etc.
are still distinctly Thonga, or at least non-Zulu. As Felgate noted, ‘the women
remain Thonga in activity and outlook . . . and the Thonga way of life persists’
(1982: 27). One reason for the limited influence of Zulu, as Webster noted, may
be that the low-lying, marshy, mosquito- and tsetse-infested area of Thongaland
was not compatible with Zulu lifestyle and the Zulu were therefore tradition-
ally content with a raiding and tribute-paying relationship. However, the limited
influence of Zulu is also explained by the non-participation of women in the
linguistic and cultural shift.
6 POLITICAL ADMINISTRATION
Territorial control has been disputed since European presence in the region.
The British navy persuaded two local groups to place themselves under ‘British
protection’ in the early nineteenth century; one of these groups was the
Tembe–Thonga. In 1875, the present Mozambique–South African border was
held to be the dividing line between the Portuguese and British spheres of influ-
ence. This line cut the Tembe–Thonga area, although there was little immediate
effect on traditional life since the population comprised small, independent
The political economy of language shift 325
groups, which did not come under central political and judicial control of
Ngwanase until the 1890s. The British proclaimed British Amatongaland a
Protectorate in 1895. Two years later it was incorporated into Zululand, with a
special status marked in part by a succession of Thonga chiefs, including the
present incumbent, who is a direct successor in the Tembe line. Both Thonga-
land and Zululand were incorporated into Natal in 1897.
Zululand control of Thongaland was more forcefully exerted after 1982, fol-
lowing the sordid ‘Ingwavuma land deal’, in which the South African Republic
offered the district to Swaziland as a Swazi corridor to the sea in return for
Swaziland’s complicity in accepting the so-called KaNgwane homeland as
part of the kingdom. The land deal would have provided a buffer between
Mozambique and Natal, and it was formally endorsed by the chief. Under pres-
sure from Thonga headmen, who claimed that they had not been consulted and,
more importantly, from the KwaZulu government, which claimed that the dis-
trict has been an integral part of Zulu territory since time immemorial, the deal
was aborted. While the latter assertion is patently untrue, the practical conse-
quence of the incorporation into KwaZulu was a forced recruitment into Inkatha,
the national cultural liberation movement associated with Chief Buthelezi, and
a community sense of a Zulu occupational force in the area. A Thonga Indepen-
dence Party existed more or less clandestinely, but residents were unwilling to
discuss anything to do with the organisation. A few individuals indicated that
they would like Thonga to be taught in the schools and that their language had
been ‘stolen’ from them by the Zulu. On the other hand, most residents believed,
rightly or wrongly, that any expression of Thonganess, including the use of the
language, would serve as a pretext for their being denounced by Inkatha as
refugees and then repatriated to Mozambique. Indeed, one of the only two men
resident in one fieldwork area in 1990 was described as ‘the repatriation officer’.
For these reasons, the label ‘abantu basenyakatho’ (Zulu, ‘people of the north’)
was sometimes used rather than (a)maThonga in self-identifications.10 As ex-
pected, the Language Atlas of South Africa (Grobler et al. 1990) describes the
area as Zulu speaking, though it should be noted that the atlas is based on the
1980 population census which asked for self-reports of home-language use. The
older Language Map of South Africa (van Warmelo 1952) describes most of the
Ingwavuma district as mixed Zulu and Thonga, with the northern and eastern
regions exclusively Thonga speaking. Junod (1896) gives most of the district
as Thonga speaking (Ronga), although he also described the local linguistic
variety as ‘un langage intermédiaire entre le ronga et le zoulou’ (1896: 6).
more readily in public arenas’ (1989: 255) needs to be read with caution. In
KwaNgwanase, younger women no longer speak Thonga, and the language of
both public and private discourse is Zulu. It is no longer true, for example, that
mothers address their young children in Thonga – except perhaps in the most
remote villages along the Mozambique border.
This is not to claim that women, following one hundred and fifty years of
resistance, have now followed the men’s shift to Zulu. Indeed, the above descrip-
tion should not be understood as suggesting that men and women now speak the
same language. Men’s Zulu is hardly ‘pure Zulu’, although its speakers insist
that it is. Visitors from deep Zululand and urban Zulu workers in Johannesburg
often remark on the ‘bad Zulu’ of these men. The reaction of Thonga men is
typically one of strong offence: they insist that they are true Zulu. They quickly
show their identity documents, which records their identity as ‘Zulu’, and they
make conscious efforts to ‘pass themselves’ as Zulu.
Women’s Zulu, on the other hand, is described by all as ‘very bad Zulu’.
Their language is so replete with Thonga lexical residue as well as phonological
and morphosyntactic influences that some outside Zulu speakers claim it to be
unintelligible.13 The linguistic accommodation that has been made by women
is thus limited, and was described to me by one (male) informant thus: ‘The
328 R. K. Herbert
women think in Thonga still, but they have Zulu in their mouths.’ It is surely
not surprising that some accommodation to Zulu has been made by women
in the light of the political intimidation described earlier and the fact that the
prescribed syllabus in all schools was that of KwaZulu. This ‘makes sense’
to local residents as they are publicly Zulu, but the cultural component of
the language syllabus, such as those parts relating to hlonipha language, is
completely foreign to pupils, since local women do not hlonipha. Girls’ poor
performance in Zulu is popularly attributed to a general devaluation of female
abilities. Men and women simply agree that girls cannot learn Zulu properly.
This attitude channels reports of language use: since women cannot learn Zulu,
it follows that they must be speaking something else. The ‘very bad Zulu’
of women is sometimes called Thonga, by both women and men. The label
has thus been redefined in at least some contexts. However, much of what
passes for ‘Thonga’ is clearly Zulu. The notion that women speak Thonga is
a convenient fiction, maintained for a variety of reasons by both women and
men.
There are several notable differences in male–female Zulu performance lo-
cally. Women, particularly middle-aged women, tend to tekela whereas men
prefer to zunda, i.e. women use t in place of men’s z. In addition, the nominal
morphology of women is more ‘Thonga-like’ than that of men. For example,
women use more CV- prefixes than men, who employ the canonical VCV-
Zulu form of prefixes. Women’s speech occasionally has a fully class 5 li- pre-
fix, which Zulu does not. These are, however, variable features of women’s
performance. There are also some striking differences in vocabulary used by
women and men, with the general trend being that women’s vocabulary is
more ‘conservative’,14 i.e. Thonga.15 However, these differences are not reli-
ably present in everyday speech.
On account of the highly variable nature of women’s linguistic performance,
it is not possible to provide a list of ‘characteristic features’. Interestingly,
their speech ranges on a continuum from Gonde, the strongest form of Thonga
spoken in the area, to Zulu spoken with a Thonga underlay. In the latter case,
the most notable features may be phonetic and a few prefixal marks; in the case
of Gonde, there are large differences in vocabulary between Zulu and women’s
speech. Some representative points of differences between standard Zulu and
local speech patterns, here termed Nyakatho, are given below.
7.1 Vocabulary
7.2 Phonetic–phonological
(a) Tekela (women) Zunda (men/Standard Zulu) English
-enta -enza do something
-buta -buza ask a question
timbuti izimbuzi goats
(b) elision of initial /l/ in class 3 nouns
Nyakatho Zulu English
unilo umlilo fire
unomo umlomo mouth
unente umlenze leg
(c) nk-w alternation
Nyakatho Zulu English
iwuku inkukhu fowl
iwomo inkomo beast
7.3 Morphological
(a) Characteristic palatalisation of labials and some other consonants, e.g.
w/diminutive and passive suffixes, does not ocur in Nyakatho.
Nyakatho Standard Zulu English
isintombana (intombi) intonjana small girl
isimotwana (imoto) imotshwana small car
-lume (-luma) -lunywa be bitten
(b) absence of class 1a marker in Nyakatho
Nyakatho Standard Zulu English
Mame uMama Mother
Baba uBaba Father
(c) Nyakatho has a full CV class 5 marker /li-/ as opposed to Zulu /i-/
Nyakatho Standard Zulu English
litinyo izinyo tooth
lilanga ilanga day
litimba izimba ear of corn
330 R. K. Herbert
(d) Class 8 /swi-/ as plural of class 7 rather than Zulu use of class 10 as the
plural of both class 7 and class 9
Nyakatho Zulu English
swiwoni izoni sinners
(f) Nyakatho often exhibits a CV class 11 prefix /lu-/ rather than Zulu /u-/
Nyakatho Zulu English
lukunyi ukhuni firewood
luvemvane uvemvane butterfly
(g) Nyakatho occasionally shows a class 15 prefix hu- with vowel stems in
place of Zulu /uku-/
Nyakatho Zulu English
huwenta ukwenza to do
huwaha ukwakha to build
The social meanings of the two shifts from Thonga to Zulu, first by men and
recently by women, are different. In part, this difference relates to the variable
link between language and ethnicity discussed above. In the case of Thonga
men, the linguistic shift correlates with a strategic shift in identity. In the
case of Thonga women, however, any attempt to negotiate identity is sharply
circumscribed.
The intriguing question in this scenario, as mentioned earlier, concerns female
behaviour. Webster summarised Thonga men’s own view of the situation in the
title of his 1989 article ‘Abafazi bathonga bafihlakala’ (Thonga women are a
mystery). Why do women not follow men’s lead and more thoroughly embrace
Zulu performance identity? Having given up, albeit under some force, Thonga
The political economy of language shift 331
language for all intents and purposes, what are the rewards associated with
speaking ‘very bad Zulu’?
The answer may be profitably viewed from an ethnography of speaking ap-
proach of the sort championed by Hymes (1962, 1982) in which the variable of
language is seen as inextricably linked to other variables of sociocultural organ-
isation such as religion, kinship and other social relations, economy, political
organisation, and so on. Using this approach, the different cultural value sets as-
sociated with Zulu and Thonga qua languages are readily apparent. Of notable
prominence are those relating to the role and status of women, and these are key
pieces in the explanation of the dynamics whereby men actively espouse Zulu
identity while women do so reluctantly, half-heartedly and (deliberately) badly.
The position of Nguni, including Zulu, women has been amply documented in
the ethnographic literature (e.g. Ngubane 1981; Herbert 1990). In broad outline,
a Zulu women is a ‘perpetual minor’, who moves at marriage from the control
of her father to that of her husband and his male kin, especially the father-
in-law. It is not possible in the present context to review sex-based relations
in Zulu society in any detail. Suffice it to say that Zulu women, particularly
wives, operate with few rights of respect and privilege: they are ‘strangers’ in
the husband’s homestead, sources of potential contamination, and causes of all
manner of misfortune. It is not coincidental that the majority of accusations of
witchcraft in Nguni society are made against wives.
On the other hand, Thonga women traditionally enjoyed a number of rights
and showed a great deal of independence. Several positions within the kinship
system, most notably father’s sister and mother-in-law, carry privilege. The
former traditionally played an important role in political, economic and social
decisions, for example, at family councils. There is no equivalent in Zulu culture.
Similarly, sisterhood brings certain respect in Thonga society; elder sisters may
be called manana, ‘little mother’. Such usage is absent in Zulu. As Webster
(1989: 256ff.) points out, men and women use different kinship systems, Zulu
and Thonga respectively, to negotiate relations in a sort of ongoing contest to
define the situation.
As noted above, hlonipha, an elaborate system of respect through avoidance –
linguistic and otherwise – practised by Zulu women, is absent in Thonga society.
Thonga women are proud that they do not hlonipha, despite male demands that
they ‘show respect’. Thonga women traditionally enjoyed more liberal divorce
customs than their Zulu sisters (Clerc 1938). The overall impression is that the
position of women in Thonga society is, in some sense, ‘better’ than its Zulu
counterpart.
For women to embrace Zulu identity, ethnicity and custom as their men do
would involve a marked diminution of their status and power. Women’s recogni-
tion of this relationship underlies their reluctance to follow men’s shift to Zulu.
The private domain, which is to a large extent still controlled by women, remains
332 R. K. Herbert
9 CONCLUSION
Creolised cultures are characterised by rapid change, the result of individual and
collective agency. In this regard, creolisation is the normal state of human affairs.
notes
1 There is much onomastic confusion here. The terms ‘Tsonga’, ‘Thonga’, ‘Tonga’,
‘Shangaan’ and ‘Gwamba’ are often used interchangeably. Tsonga has emerged as
the label used for all of the various linguistic subvarieties as well as the name of
the standardised variety used in South Africa (based on the Nkuna and Gwamba
varieties). The term ‘Thonga’ is reserved in this chapter for the linguistically related
group residing in northern KwaZulu-Natal; this form is generally taken to be Zulu,
but is occasionally used by this group to name itself. The South African Thonga are
a subgroup of Ronga, itself a subdivision within the larger Tsonga group (S.50).
2 ‘The clans comprising the Thonga people have nothing in common beyond a few
disappearing customs. The only thing that they hold in common is a rich, ancient and
distinct language. The unity of this tribe is indeed more linguistic than national.’
3 Field research for this chapter was supported by a grant-in-aid from the research
foundation of the University of the Witwatersrand and Title F support from the State
University of New York at Binghamton. Fieldwork was conducted during 1989, 1990
and 1992, mainly around KwaNgwanase, but also in subdistricts to the north and east,
including KwaMshudu and the Kosi Bay area. The chapter describes the language
situation prior to the South African transition to democracy. Reports from the area
since that time indicate some slight rise in moves to reassert Thonga identity, though
not expressed in any language revitalisation movement.
4 Whether this variety is viewed as Zulu with Thonga features, or as Thonga with Zulu
admixture, is usually decided on political grounds. Other analysts skirt the issue by
labelling it, for example, ‘a Thonga–Zulu dialect’ (Ngubane 1992: 11).
5 It should be noted that ‘Ronga’, like the appellation ‘Thonga’, is a label of convenience
based on linguistic, rather than social or cultural, facts. The Ronga never constituted
a social or political unit. The term was not used by the peoples so named as a self-
appellation. It is derived from the local word for ‘east’.
6 The meaning glossed as ‘slave’ was probably added after Zulu incursions into Thonga
territory in the nineteenth century. A more likely etymology is suggested by the
geographical distribution of the distinct peoples known as Thonga, Tonga and Tsonga.
All of these groups surround the ancient Zimbabwe empire. An etymology more likely
to mean ‘foreigner, outsider’ is suggested by this distribution. See Herbert (1996:
1345–6). As noted above, the Tembe–Thonga claim an ancient origin in Zimbabwe.
That several outsider groups adopted this name for themselves reveals something
about power and status and about the interactions between groups.
7 Felgate (1982: 13) suggests that the Ngubane were a Thonga group who migrated
southwards before the Tembe.
8 This summary ignores the important theoretical debates concerning the nature of the
variable labelled ‘ethnicity’. In South Africa, some of the relevant issues are partic-
ularly complex on account of the ways in which early administrative classifications
were used and transformed into politicised ethnicity after the advent of the Nationalist
government.
9 That children learn their mothers’ language, Thonga, is not surprising. This is the
language that children use when speaking to their siblings and playmates: ‘It is only
334 R. K. Herbert
when they [boys] become conscious of themselves as males that the switch to Zulu
as the dominant language is made’ (Felgate 1982: 24).
10 There are numerous historical reports that the label ‘(a)maThonga’ carried derogatory
overtones and ‘was never used by the people to whom it was applied’ (Harries
1983: 19). While it is certainly true that European perception/promotion of Tsonga/
Thonga/Tonga linguistic unity underlay a promoted sense of shared identity and
cultural unity, an argument Harries has made elsewhere (1988), one should not think
the Tsonga unique, or even unusual, in the southern African context. The derogatory
overtones associated with the label were largely purged, and the reluctance to use
the term today derives from other, more narrowly political, forces.
11 What is most striking here is that the typical Zulu central cattle pattern is missing;
huts are arranged in a line facing east rather than a circle, and the cattle kraal is not
within the homestead.
12 Belatedly, I need to record my debt to David Webster, who first suggested to me that
an anthropological linguist would find the study of language issues in the area inter-
esting. Several of my informants came to me through Webster’s recommendation. I
am also grateful to Sihawukele Ngubane and his father, Mr E. S. T. Ngubane, whose
advice, insights and assistance were invaluable during the fieldwork period.
13 Obviously, these outsider reports of unintelligibility need to be assessed with caution.
14 The basis for comparison is the Thonga/Tsonga spoken in southern Mozambique,
rather than the South African standard Tsonga of the Eastern Transvaal.
15 The linguistic situation is complicated by the presence of variety known as Gonde/
Gondzze/Konde, classed by Baumbach (1987) as ‘a Tsonga dialect’. Gonde is spoken
exclusively by old women along the coast from Lake Sibhayi to Bhanga Nek. It seems
to represent the ‘purest form of Thonga’ in the area, i.e. it is least influenced by Zulu,
though Ngubane (1992: 12) classifies it as a Zulu dialect that has been ‘externally
influenced by Swati and Thonga’.
bibliography
Allison, A. A. 1951. ‘A Maputaland school: the Star of the Sea School’. Native Teachers’
Journal 31: 7.
Baumbach, E. J. M. 1987. ‘Klasprefikse van Gondzze’. South African Journal of African
Languages, 7: 1–6.
Bryant, A. T. 1929. Olden Times in Zululand and Natal. London: Longmans.
Clerc, André 1938. ‘The marriage laws of the Ronga tribe’. Bantu Studies, 12: 75–104.
Doke, C. M. 1954. The Southern Bantu Languages. London: International Africa
Institute.
Felgate, Walter 1982. ‘The Tembe Tonga of Natal and Mozambique: an ecological
approach’, ed. Eileen Jensen Krige. Department of African Studies, University of
Natal, Durban, Occasional Papers No. 1.
Grobler, E., K. P. Prinsloo and I. J. van der Merwe 1990. Language Atlas of South Africa.
Pretoria: Human Sciences Research Council.
Harries, Patrick 1983. ‘History, ethnicity and the Ingwavuma land deal: the Zulu northern
frontier in the nineteenth century’. Journal of Natal and Zulu, 6: 1–27.
1988. ‘The roots of ethnicity: discourse and the politics of language construction in
south-east Africa’. African Affairs, 87: 25–52.
Herbert, Robert K. 1990. ‘ “Hlonipha” and the ambiguous woman’. Anthropos, 85:
455–73.
The political economy of language shift 335
Language contact
New varieties of English
17 From second language to first language:
Indian South African English
R. Mesthrie
1 INTRODUCTION
Indian South African English (henceforth ISAE) is worthy of the attention
of sociolinguists and scholars concerned with new Englishes, for a variety of
reasons. It offers the opportunity of examining in a relatively fossilised form
(on account of former rigid segregative tendencies in South Africa) the evolution
of a dialect of English under less than perfect conditions concerning educational
and social contact with target-language speakers. It also provides, again in a
relatively fossilised form, the opportunity of studying the changes a language
undergoes as it shifts from L2 to L1.
This chapter has two aims: (a) to complete the sociohistorical background
to language maintenance and shift among Indian South Africans begun in the
article on Indian languages (chap. 8, this volume); and (b) to examine the con-
sequences of social history on linguistic and sociolinguistic structure, manifest
in the dialect of English spoken by Indians in KwaZulu-Natal.
As a prelude to the history of ISAE the reader is referred to the back-
ground of indenture and immigration set out in chapter 8. Historical records
suggest that the vast majority of Indian immigrants (perhaps 98 per cent – see
Mesthrie 1992b: 12) had no knowledge of English. The language of the new
colony that Indians learnt quickest was the pidgin, Fanakalo. For communica-
tion among themselves Indians used an Indian language (usually Bhojpuri or
Tamil) and sometimes Fanakalo. Generally, the use of English among Indians
in nineteenth-century Natal was the exception rather than the rule.
Bughwan (1970: 503) states that English was first transmitted to Indians
by native speakers of the language – English missionaries, British teachers
and English-speaking sugar-estate owners. This is far too optimistic a view of
the social conditions prevailing in the colony. We can instead posit four main
possible sources of input to the learner:
(a) schooling, with teachers being native speakers of English;
(b) schooling, with teachers being non-native speakers of English;
(c) contact with native speakers of English in Natal;
(d) contact with non-native speakers of English (chiefly Indians).
339
340 R. Mesthrie
Written records suggest that all four sources were significant in shaping ISAE.
As far as education in the nineteenth century was concerned, it would appear
that the number of non-native English-speaking teachers was at least as great as
that of mother-tongue English teachers. This includes the missionaries, many
of whom were of Continental European origin (see further Mesthrie 1992b:
19–22).
By the 1930s the pattern of language learning had not changed much from
nineteenth-century trends. English was learnt as a second or third language out-
side the home (in classrooms or, in the case of large numbers who had no school-
ing, either at work or not at all). By the late 1950s, when education facilities
had improved, English began to be introduced in the home and neighbourhood
by children. In some homes a rapid inversion of roles took place. Whereas the
first- and second-born child might have arrived at school with no knowledge
of English, their subsequent influence in the home was in some instances so
significant as to cause the last or second-last child in a large family to arrive
at school with English as dominant language. In the 1960s and 1970s English
became the first language of a majority of Indian schoolchildren. A process of
shift is under way, with the Indian languages surviving with some difficulty.
The period of language shift can be thought of as gradual or rapid, depending
on one’s defining criteria. As 1960 was exactly one hundred years since the first
immigrations, the shift might seem a gradual one; but as 1960 was also less
than fifty years since the last shipload, the shift is perhaps not all that gradual.
In some rural homes, parents (especially mothers) learnt English from the
youngest children, rather than vice versa. This process, which I call a ‘closed
cycle of reinforcement’ in language shift, continues until today, though it is now
manifest in the interactions between grandparents (especially grandmothers)
and grandchildren. That is, those grandparents with little or no schooling who
spoke an Indian language to their children a generation ago are now forced
to learn English in order to be intelligible to monolingual grandchildren.
(In some homes grandchildren are lucky enough to receive input from grandpar-
ents in an ancestral language, but this is increasingly rare.) The closed cycle of
reinforcement is remarkable for its potential two-way influence: the grandpar-
ent learns from and with the grandchild, and in turn reinforces the grandchild’s
child-language. As one of my interviewees discussing his wife’s knowledge
of English put it, ‘Now with her purposes too, her grandchildren all growing
y’see, now she must communicate with them in the language they understand
[English]. So she goes along with that language. They teach the grandparents
how to speak the language’ (emphasis mine).
3.2 Vocabulary
At its most formal, the ISAE lexis differs only slightly from general South
African English; at its least formal, it is exceedingly different. The differences
found in informal speech are catalogued in my Lexicon of South African Indian
English (Mesthrie 1992a), a work comprising about 1,400 items character-
istic of this dialect. Entries describe specific lexical items from a variety of
sources, points of grammatical usage (e.g. y’all as plural second-person pro-
noun), specific pronunciation traits (e.g. bagit for ‘bucket’ among many older
speakers), adolescent slang and proverbs (e.g. to want mutton curry and rice
everyday, which means ‘to expect a good time/the best always’). The majority
of lexical items are drawn from Indian languages in the sphere of kinship, re-
ligion and culinary practices. Interestingly, some of these still vary from home
to home, depending on the ancestral language. A good example is the word
for ‘spicy food’, for which the adjective hot is ambiguous. Speakers wishing to
describe spicy food rather than food that is hot in terms of the temperature use
one of the following, depending on the ancestral language: karo (Tamil); karum
(Telugu); thikku (Gujarati); thitta (Hindi/Bhojpuri); thikka (Urdu). Since these
are used in primarily domestic settings, the terms are not widely known. In
public discourse a term such as pungent or chilli-hot may be used. Three areas
of vocabulary are listed below: the lexis of love (which hint at a time when love
was not spoken about directly); some salient semantic shifts; and common lexis
drawn from Indian languages.
The lexis of love
future (n.) husband or wife-to-be
interested in in love with
get in touch have a romantic involvement
proposed affianced, engaged
disappointed jilted in love
spoilt carrying a child out of wedlock
marry out to marry outside one’s traditional sub-ethnic group
Some semantic shifts in ISAE
lazy unintelligent
interfere to molest
hint to speak ill of (not necessarily obliquely)
independent stand-offish, haughty
From second language to first language 343
Basilect
Question: How often (do) you go to Durban?
Response: Where we go! Hardly we go, visit Durban too. Sometime ’olidays, my
’usband take his brother’s house an’ his sistern-law there an’ all of his
344 R. Mesthrie
(Loose sE equivalent: We don’t go. We hardly ever go to visit people in Durban. Some-
times during holidays my husband takes us to his brother’s and sister-in-law’s house or
to other relatives. My relatives live in Merebank. We sometimes go on holidays, but this
year we had some problems; even if we want to visit, we have to consider that to stay
for two or three weeks is an imposition, since they live a hard life, with a little money
only. We must be considerate; we can’t just pitch up and remain there for long. Even if
it’s our own brother or sister, or anyone close we have to realise that in Durban people
have to pay for services like lights, water, etc.)
Mesolect
Question: Tell me about the time you had a heart attack.
Response: I went an’ bought one soda water. So I had a soda water in the cafe, I took
my coat out, took my jersey an’ all out, I chucked it on the table. I sat, sat, sat – I
said no’, I felt I must reach home. I didn’t trust anybody to drive that van because
it was lent to me from somebody else. So somehow or other I managed, I jumped
into the van, an’ I drove the van an’ came, I just came an’ parked here an’ lied
down. My son was here, this second, third fellow of mine. Phoned by Dr T. G.
Singh, while I’m lying on the bed, I donno what happened, the wife gave me little
bit of sugar-water. I just drank that sugar-water, and eh, just when I finished
drinking the sugar-water I became normal. (sixty-year-old, urban, male, working
class)
who understands all the nuances of the basilect. In Mesthrie (1990) I discuss the
similarities and possible links between the pre-basilect and Butler English, a
rudimentary pidgin of South India. This is the earliest inter-language form dis-
cernible within ISAE and forms an important foil for the proper understanding
of the more focused and stable basilect.
New developments of the 1990s include: (a) the rise of a small group of well-
educated people who acquired newly created jobs as television announcers;
and (b) students graduating from private schools in which the norms of white
SAE prevail. Such ‘post-acrolectal’ speakers, who use an essentially non-ISAE
system, fall outside the ambit of this study.
Putting the main lects in boldface, the ISAE continuum may be presented as
follows:
Another speaker (a college lecturer), whose interview style can best be de-
scribed as upper-mesolectal to acrolectal, showed similar patterns of style shift-
ing in a generally relaxed interview. In the middle of the session she turned to
her husband, who had just returned from shopping, and asked, You bought
cheese, Farouk? Once again, an intimate style required a switch away from the
acrolect constructions (do-support, in this instance). The reply of the husband,
a high-school English teacher, was even more revealing. Not realising that the
tape was running he said, in an ultra-casual style, No, but lot butter I bought.
This single utterance contains a number of basilectal features that he himself
might harangue against in the classroom: a predilection for topicalisation; lot
for ‘a lot of’ (or ‘much’ in classroom English); and a basilectal pronunciation,
[nɔ:] for acrolectal no (= [noυ]).
Another argument for not considering the ISAE system to comprise a bipolar
‘dialect plus standard’ mechanism comes from a study of the acrolectal end of
the continuum. The acrolect is not the same as sE or the (white) SAE variety
used in Natal. In terms of both accent and syntax there are subtle boundaries
which few speakers traverse. Only a few speakers are genuinely bidialectal in
ISAE and SAE. These tend to be young professionals employed in prestigious
commercial houses, where they come into contact with SAE employers and
clients. The group also includes a few radio and television announcers as well
as students in private schools. Those who carry the SAE dialect home run the
risk of being gently ridiculed (Your mummy’s using her Standard Bank accent)
and, in my observation, switch to the mesolect in intimate styles.
What are the syntactic features that the acrolect shares with the basilect and
mesolect, and that mark off the acrolect from general SAE?
(a) y’all as plural pronoun form. This form, which is below the level of social
consciousness for most ISAE speakers, occurs in informal letters (where
it is usually spelt you’ll) and formal speeches. It has a genitive form,
y’all’s.
(b) Copula attraction to wh- in indirect questions, which results in sentences
like (1)–(2):
(1) Do you know what’s/what is roti?
(2) I don’t know when’s/when is the plane going to land.
In sharp contrast is the (standard) SAE equivalent with the copula occurring
after the subject NP of the embedded clause. The equivalent of (1) would have
stress on sentence final ‘is’: Do you know what roti is?
(c) The use of of in partitive genitive constructions beyond standard English
contexts:
(3) The trouble with him is he’s got too much of money.
From second language to first language 347
5.1.1 Complementation
Among the more striking differences between basilectal ISAE and sE comple-
mentation are the following:
(a) Sentence-external placement of modal-like modifiers.
(4) Lucky, they never come. (= ‘We were lucky that they didn’t come.’)
(5) Must be, they coming now. (= ‘It must be that they’re coming now/they
must be on their way.’)
(b) This pattern is extended to constructions that would require raising and the
use of infinitives in standard English.
(6) They told I must come an’ stay that side. (= ‘They asked me to come
and live there.’)
(7) I like children must learn our mother tongue. (= ‘I’d like our children
to learn our mother tongue.’)
(8) Then Ram told Devi’s mother must tell I must come. (= ‘Then Ram
asked Devi’s mother to ask me to come.’)
Like (4)–(5) these show the pattern modal-like element +S, with the structure
of S unchanged. Although the usual English pattern with to infinitives and
raising does occur in the basilect, they are not as frequent as patterns exhibited
in sentences such as (6)–(8).
(c) The use of clause-final too as hypothetical marker in conditional clauses.
348 R. Mesthrie
(9) It can be a terrible house too, you have to stay in a terrible house.
(= ‘Even if it’s a terrible house, you have to live in it.’)
(10) Very sick an’ all too, they take them to R. K. Khan’s. (= ‘If they’re very
sick, they are taken to R. K. Khan Hospital.’)
5.1.2 Co-ordination
Both the basilect and pre-basilect often favour the paratactic stringing of clauses
instead of overt co-ordination markers.
(11) She was calling, she was telling . . . (= ‘She called and said . . .’)
(12) Born over there, I’m brought up over there. (= ‘I was born and brought
up over there.’)
When co-ordination is marked, a variety of strategies arise. The ones that are
‘created’ rather than ‘inherited’ (from Natal English) are exemplified below.
(a) The use of too clause-finally: there were a few instances of these in the
corpus.
(13) I made rice too, I made roti too. (= ‘I made both rice and roti.’)
(14) You walk into town too its difficult, you wanna do shopping too, its
difficult. (= ‘If you want to walk in town and do your shopping, it’s
difficult.’)
(There are parallels in such clause final marking in Indic and Dravidian lan-
guages – and rigid OV languages generally.)
(b) The use of salient phrase-final quantifiers:
(15) I speak English, Tamil, both. (= ‘both X and Y’)
(16) . . . rose-water, vicks, coconut oil, nothing. (= ‘neither X nor Y nor
Z . . .’)
(17) We had to take out our shirt, tie, vest, everything. (=‘all of X, Y, Z . . .’)
(18) They must have one cup porridge, water, anything. (= ‘one of X,
Y . . .’)
(a) Aspect marking The verbs stay and leave are used in non-target
language ways to convey aspectual distinctions. An’ stay after a verb signals a
habitual sense; an’ leave him/her/it is a completive marker.
(25) We’ll fright an’ stay. (= ‘We used to be afraid (for a long while).’)
(26) When mother-all here, we’ll talk mother, and laugh an’ stay. (= ‘When
my mother and others were alive we used to talk merrily (at length).’)
(27) She filled the bottle an’ left it. (= ‘She filled the bottle completely.’)
(28) We whacked him an’ left him. (= ‘We beat him up thoroughly.’)
350 R. Mesthrie
This construction, which is not a very common one, might be part of a larger
tendency to replace adverbs by verbs in sentences denoting habitual action:
(29) He’ll run an’ come. (= ‘He’ll come running.’)
(30) They only laugh and talk. (= ‘They always speak lightly/in a laughing
manner.’)
Sentences (25) and (26) show another aspectual difference from sE: the use of
the reduced form of will to denote past habitual action (equivalent to sE would).
The most striking innovation in aspect marking in ISAE is however, should for
standard English used to (see Mesthrie 1992b: 130–3).
speech they are easily able to produce a standard form, but in spontaneous dis-
course, under either very relaxed or very tense situations, they display several
‘near-misses’. By contrast, forms typical of the basilect often involve not-so-
near misses. The first of our two sets of illustrations concerns items that are
recognisably part of the dialect (chiefly the mesolect).
(a) (i) Prepositions:
(49) He’s got no worries of anyone. (‘about’)
(50) I’m not fluent with Afrikaans. (‘in’)
(51) I was good in arithmetic. (‘at’)
(52) You should be appreciated with that thing. (‘appreciative of’)
(ii) Adverbials, adjectives and quantifiers
for really really, truly
farest furthest
worst worse (in addition to its usual meaning)
more worse worse
the both both
(iii) Lexis and Idioms
sincing since
catch up catch on
scratch itch
can’t stick the heat can’t stand the heat
play fools play the fool
long-cut long route
to run a mock to run amuck, to revel
to pick somebody out to pick on someone
to take out one’s clothes to take off one’s clothes
Some of the ‘near misses’ involve a conflation of two target-language items,
or the influence of one over another. Thus sincing seems to be based on
both since and seeing; long-cut is formed by analogy with short-cut, etc. Such
neologisms, overgeneralisations and recategorisations are very common in the
new Englishes generally. Sey’s grammar (1973) of Ghanaian English and
Nihalani, Tongue and Hosali’s (1978) lexicon of Indian English give examples
which suggest that these processes occur to a much greater extent than in ISAE.
Indeed, these are the most salient feature of those varieties of English. In ISAE
they are one of a widely varying set of processes, and not the most divergent of
these from sE.
(b) Another interesting set involves items that are ‘one-off’ errors, used by
mesolectal speakers in the interviews. Although the individual items exempli-
fied in (53)–(55) are not characteristic of the dialect, the process is widespread
enough in the mesolect to warrant our attention.
(53) I accompany all the vegetables with spices. (= ‘I mix the vegetables with
spices.’)
354 R. Mesthrie
7 CONCLUSION
The differences that ISAE exhibits from sE are, I believe, much greater than
those exhibited by other new Englishes. The circumstances of its origins and
development have more than passing similarities with those that engendered
pidgin and creole languages. However, the reader should not be misled into
anticipating all kinds of problems in the classroom because of the differences
stressed in this chapter. For one thing, many young speakers are increasingly
able to shift to more standard ways of speaking in public and formal discourse.
However, the characteristics of ISAE usually surface in intimate and informal
conversations of even the most educated speakers. This closely follows Labov’s
(1972: 134) characterisations of covert prestige attached to the vernacular. In
moving from L2 to L1 within the special case of apartheid-dominated society
ISAE has not abandoned its former L2 features.3 These features have gone
underground, so to speak.
notes
1 The zero should not be interpreted literally; it signifies that no post-acrolectal speakers
turned up in the sample. The actual proportion is close to zero.
2 Valdman (1977: 155) discusses these processes in relation to creolisation.
3 This phenomenon still holds in the new post-apartheid society, but may gradually
change.
bibliography
Andersen, R. 1983 (ed.). Pidginization and Creolization as Language Acquisition.
Rowley, Mass.: Newbury House.
Bickerton, D. 1975. Dynamics of a Creole System. Cambridge: Cambridge University
Press.
From second language to first language 355
Bughwan, D. 1970. ‘An Investigation into the Use of English by the Indians in South
Africa, with Special Reference to Natal’. Ph.D. thesis, University of South Africa.
Kachru, B. B. 1983. The Indianization of English. Delhi: Oxford University Press.
Labov, W. 1972. Sociolinguistic Patterns. Philadelphia: University of Pennsylvania
Press.
Mesthrie, R. 1990. ‘Did the Butler do it?: on an analogue of Butler English in Natal,
South Africa’. World Englishes, 9, 3: 281–8.
1992a. A Lexicon of South African Indian English. Leeds: Peepal Tree Press.
1992b. English in Language Shift: The History, Structure and Sociolinguistics of
South African Indian English. Johannesburg: Witwatersrand University Press;
Cambridge: Cambridge University Press.
Nihalani, P., R. K. Tongue and P. Hosali 1978. Indian and British English: A Handbook
of Usage and Pronunciation. Delhi: Oxford University Press.
Sey, K. A. 1973. Ghanaian English. London: Macmillan.
Slobin, D. 1973. ‘Cognitive prerequisites for the development of grammar’. In C. A.
Ferguson and D. I. Slobin (eds.), Studies of Child Language Development.
New York: Holt, Rinehart & Wilson, pp. 175–208.
Valdman, A. 1977. ‘Creolization: elaboration in the development of Creole French
dialects’. In Valdman (ed.), Pidgin and Creole Linguistics. Bloomington: Indiana
University Press, pp. 155–89.
Washabaugh, W. 1977. ‘Constraining variation in decreolization’. Language, 53:
329–52.
Williams, J. 1987. ‘Non-native varieties of English: a special case of language acquisi-
tion’. English World-wide, 8, 2: 161–99.
18 Black South African English
356
Black South African English 357
tuition. The recent flood of heated and critical complaints from purists following
its increasing use in public media underline its erstwhile status as second-rate
‘non-standard’ variety. But increasingly it is being viewed (and described) as
a variety in its own right (Buthelezi 1995; Gough 1996;2 Wade 1997), worthy
of recognition, but more than that: an unavoidable fact of life. At the same
time, its status and prestige appear to be undergoing a rapid change for the better
(see section 3).
2.1.2 Consonants
Consonantal systems in the local African languages are fairly complex, and
the only English phonemes lacking generally are /ð/ and /θ/. These are typ-
ically pronounced as dental or alveolar stops [d] and [t]. Other features can
be attributed to specific native-language influences. /tʃ/ is a marginal phoneme
in Zulu and may be replaced with /ʃ/ by Zulu speakers (Jacobs 1994), while
Sotho speakers may pronounce the consonant cluster /kl/ as an ejective lateral
affricate /tl’/, a phoneme which occurs in the Sotho languages. There is lit-
tle evidence in the literature that such differential pronunciation features are
generally evaluated as markers of ethnicity or regional origin as they may do
in Nigeria (Schmied 1991: 57). However, anecdotal evidence from casual in-
terviews suggests that African speakers interacting in English can very often
identify a person as having a different first language from their own.
Other more widespread consonantal features are a trilled /r/ sound (as opposed
to an approximant). In addition, stops in the Bantu languages also appear to
have a later voice onset time in comparison to white South African English
(WSAE), and may also tend to be devoiced in word-final position. This may
result in voiced stops being perceived as voiceless. Jacobs (1994: 23) claims
that the cumulative effect of such consonantal and vowel features in what she
Black South African English 361
In BSAE phonology, prominence may not have the discourse functions of sig-
nalling contrast and the difference between given and new information (as is
typically the case in native varieties), because in the Bantu languages gram-
matical and syntactic means are used to indicate contrast and given and new
information. Lanham (1984) finds that in BSAE such prominence appears to
be assigned to content words generally in pre-coded speech (i.e. written text
read out loud), while Genrich de Lyle (1985: 98) finds that it may be assigned
more or less arbitrarily in conversation. Casual observation of speakers with an
otherwise high degree of fluency in English also suggests that if there is an aux-
iliary within a sentence, this tends to attract sentence stress (without necessarily
indicating contrast or emphasis) as in ‘the results WILL be announced later’.
Very little study has been done on phonetic variation along the basilect–
acrolect continuum. Hundleby (1964) does, however, discuss some variation in
this respect and seems to indicate that differences between standard WSAE and
BSAE tend to be more phonetic than phonological. This area of study is still a
wide open one.
362 V. de Klerk and D. Gough
the features listed here (including the lack of tense and concord marking) are
common to pidginisation processes universally.
It is important not to treat all of these grammatical structures monolithically
and as all equally representative of a uniform BSAE. There is, in this respect,
some evidence of considerable variability with regard to the relative accept-
ability and utilisation of such structures. Gough (1996), for instance, shows
that grammaticality judgements of twenty Xhosa-speaking teachers indicate
that some grammatical features may to be more acceptable than others among
educated speakers of BSAE. Thus, for example, structures such as (24) and (25)
below were described as ungrammatical by about 90 per cent of the sample.
However, structures such as (26) and (27) were typically regarded as gram-
matical, with only around 20 per cent of the sample indicating that they were
ungrammatical.
(24) I tried that I might see her.
(25) He was carrying a luggage.
(26) She was refusing with my book.
(27) He explained about the situation.
Such figures suggest that, at least in more acrolectal varieties, certain features
are more ‘entrenched’ or ‘fossilised’ than others. The study also suggests that
among educated speakers a fairly traditional norm of ‘correctness’ continues to
act as model. Somewhat contrary results, however, are suggested in preliminary
research involving first-year students at the University of the Western Cape
(a university with predominantly African and coloured students). Given the
task of correcting sentences such as:
(28) After chairperson have being chosen, she will leave for Cape Town.
around 90 per cent of the 50 (‘coloured’) Afrikaans-speaking students produced
standard versions of the sentence, while only around 30 per cent of the 250
speakers of African languages did so. Surprisingly, the feature most commonly
changed in the above sentence by the students speaking African languages
was to change being to been (around 50 per cent of students), while over
80 per cent of the students failed to correct either the missing article or to correct
have to has. This may be due to the stigmatising of particular constructions in
formal education, which may raise them more to the level of awareness than
others. More generally we may note that there is increasing research that re-
lates similar grammatical variation to pragmatic functions and social variables
(Mesthrie 1997; Wade 1995).
2.3 Vocabulary
A range of words from African languages reflecting African experience are
commonly used in BSAE and have indeed become part of SAE usage more
Black South African English 365
(33) He proposed love to her (‘He told her that he loved her’).
(34) Pass my regards to the family.
(35) I must quickly touch the beauty salon (i.e. ‘drop into’).
(36) You are scarce (i.e. ‘I haven’t seen you for a while’).
The idioms in (34) and (36) may well be spreading in SAE. Other examples
that have commonly been noted are the predicative use of late (as a euphemism
for ‘die’) as in My father is late, and the use of somebody for ‘person’, as in
He is a very important somebody (see Buthelezi (1995) and Adey (1977) for
more South African examples). Another common feature is the redundant use
of each and every, used synonymously with each. There are also differences
and restrictions in stylistic range, so that word-pairs such as abode/house and
mommy/mother may not necessarily be differentiated in terms of relative de-
grees of formality or informality. There may also be connotational differences:
it appears, for instance, that the public use of terms for sexual organs is even
more strongly tabooed than in Anglo-Saxon communities – a fact that has made
AIDS education problematic (see Crawhall 1993). A number of these particu-
lar lexical characteristics have also been noted in African varieties of English
outside South Africa (Jowatt and Nnamonou 1985).
educational context, where lack of student participation may indicate respect for
the teacher as the repository of knowledge. Peires (1992: 11) also notes very little
participation from females in mixed-sex conversations, a carry-over of the tra-
ditional subservience of African women in African society (cf. de Klerk 1997).
(i) in fact
The phrase in fact in native English conversation (as in In fact, he’s very rich,
for instance) is used for emphasis or to underscore a point. In black English
discourse this marker is particularly common and does not appear to have this
meaning, as is evident in the following conversation:
A. Hello, doctor.
B. Hi, Vuyisile. (pause)
A. In fact, I want to talk to you about my essay.
The origin of this in fact appears to be from an African-language equivalent
which is used in conversation as a topic-changing or topic-initiating marker.
(ii) I can say that
The use of the italicised expressions below is another feature of BSAE conver-
sation:
(37) In my opinion I can say he is not correct.
(38) I can say that this is an interesting book.
(39) On my side I do not think this is relevant.
These expressions are essentially equivalent to a common expression in African
language discourse, such as ndingathi, ‘I can say’ or ngoluvo lwam, ‘in my
opinion’. They are routinely and formulaically used in conversation in a way
somewhat equivalent to the English ‘I think’. The phrase I can say does not
really mean what it does in native-speaker conversation (cf. also Peires 1992:
6; Wissing 1987: 84).
(iii) Again
Again may be used as a marker of additional information (something like ‘in
addition’). This also appears to be due to the influence of a marker common in
African languages which covers the sense of both ‘again’ and ‘in addition’:
(40) Smoking is bad for health. Again it affects people around the smokers.
(iv) By all means
The phrase by all means, rather than being a discourse marker of assurance, as it
is in native-speaker English, typically functions as an intensifier, as exemplified
in (41), which could be glossed as ‘You should try to do the best you can’:
(41) You should try by all means to do your best.
themselves) word order and morphological devices appear to be used far more
often for signalling informational relatedness. As we have seen, in fact, stress
within the sentence may not indicate salience at all, and very often seems
somewhat randomly assigned, as the following relevant examples from BSAE
show (Gough 1996: 67; Wissing 1987: 151):
(42) The best education, I need to get it.
(43) A student, if he cheats, he will be expelled.
(44) She informed her lecturer that the extension, she wanted it for another
reason.
Mesthrie (1997) notes that, as opposed to the general explanation of simple
transfer from African languages, such instances, although occurring more fre-
quently in BSAE, are very similar in function to topicalisation, contrast and
focusing in general English usage. Wade (1995) holds, on the other hand, that
in many instances such topicalisation constructions, unlike the case of native
varieties, may be used to indicate change of topic in extended discourse.
It is important not to forget that there is a struggle at an advanced stage going on in South
Africa, Transkei and logically also at the Faculty of Education at Unitra. Our struggle,
with an important dimension of empowering the indigenous black victims of apartheid,
is to be pursued in every corner and sphere of life in our country including in the Faculty
of Education in Unitra. It must be waged here and everywhere . . . Indigenous people in
the faculty must be given way to take the bull by the horns and advance our struggle.
We cannot sit idly by and read about contributions from [names of various universities]
while we sit like frightened frogs in the face of a dying snake . . . Indigenous people are
knocking on the door to take what is their birthright.
A range of explanations for the occurrence of this style has been offered both
for new Englishes in general (Platt et al. 1984: 148–50) and BSAE in particular
(Wissing 1987: 179; Scheffler 1978: 26). Such explanations include rhetorical
transfer from native language, the attitude that the only good English is more
formal English, or simply restrictions in the knowledge of English register.
Black South African English 369
Problem I experienced with second language was that I was not able to perform and also
not good in writing as well. I only speak second language when I was going to the shop
by asking prices of the goods. The only thing that motivated me was because I wanted
to pass and my teacher used to tell us whoever wants a job one will be forced to speak
English.
I was having problem speaking English. I was having an attitude towards second
language but there wasn’t a choice. I was forced. I was experiencing some problem
with learning second language because in my family English was not a spoken language
even by mistake. I was speaking English in the classroom after that I was not speaking
because there wasn’t a person to speak with. The fact that I was coming from a working
class family made me to suffer at school because every subject was taught in English
during my Higher Primary [Standard 3–5]. There were no people through which I can
speak the second language with except with my teacher at school
The worst of all because I was from working class there were no enough source like
libraries within the location and even at home there were no English books to read not
television unlike middle class and upper class homes whereby one was enjoying the
privileges of many be having some magazines and books
Besides poignantly revealing many of the grammatical features discussed above,
this essay also demonstrates the ‘waffle phenomenon’ characteristic of certain
types of non-native referential discourse more generally. This phenomenon
appears to result from writers feeling that they are not expressing their ideas
clearly in the linguistic forms to which they are restricted and needing to repeat
themselves in order to get their message across.
2.5 Code-switching
The mixing of English and vernacular languages in the same conversation
is a common feature of black South African discourse, as is the case more
generally in the new English-speaking world (Myers-Scotton 1989; D’Souza
1992), where it forms part of its users’ total stylistic repertoire. It may be the
norm or, in Myers-Scotton’s terms, the unmarked choice among certain social
groups (typically the educated elite) whose membership is symbolised by using
both languages. As Myers-Scotton (1989: 343) notes, while English symbolises
membership of the elite, educated and powerful, because the participants’ other
370 V. de Klerk and D. Gough
3.1 Attitudes
Given the dramatic shifts in sociopolitical power in South Africa recently,
and because of the growing demographic status of its speakers, attitudes to-
wards BSAE are rapidly changing and it is enjoying increased vitality in a very
favourable ideological milieu (Wade 1997). The stigma associated with the use
of non-standard varieties, so strong in the past, has been replaced by a growing
assertiveness and confidence in the value of SAE varieties, including BSAE.
One may go so far as to say that it is no longer simply a case of covert prestige
being attached to BSAE (see Smit 1996); instead, the prestige is becoming
more overt. Contributing factors are its use as a major language of government
combined with the rising socio-economic status of its speakers, who are rapidly
forming a black middle class: the Financial Mail (13 June 1997) shows that
almost as many blacks (3.5 million) as whites (4 million) comprise the top
socio-economic bracket in the country (cited in Wade 1997).
Given the current focus in South Africa on democracy, non-racialism and
egalitarianism, there has been increasing emphasis on democratic language
rights as well as an awareness of the linguistic difficulties prevailing in the
country. As a result, prescriptive concern for correctness has declined, and
tolerance and mutual respect have led to more emphasis on getting the message
across, rather than on elitist requirements regarding concord, tense and other
Black South African English 371
grammatical niceties. The prospects are very good for greater acceptance of
variability in educational contexts and in business. We are certainly witnessing
this in the media, where serious announcements and up-market advertisements
are increasingly in BSAE accents,6 reflecting changing perceptions of its status,
authority and persuasive appeal. This growth in the prestige of BSAE is likely
to lead to increasing confidence among BSAE speakers and learners, who,
because they can identify strongly and positively with the variety, are likely to
master it more easily. In addition, such a variety may act as a powerful national
unifier, bridging the gap between speakers of often very different indigenous
languages.
While acknowledging the growth of more positive attitudes to BSAE in South
Africa, it would be inaccurate to claim that support for English (of any variety) is
unequivocal: trends in the growth of the appeal of BSAE will probably be slowed
down by three different factors. First, the traditional voices of prescriptivists,
both academic and non-academic, continuing to make themselves heard, fight-
ing against changes in standards and refusing to recognise the validity of BSAE.
Second, it is very likely that the steadily increasing numbers of young black
South Africans emerging from former whites-only English-medium schools,
who typically acquire something closer to standard SAE by the time they leave
school, will counteract the appeal of BSAE. With their privileged educational
backgrounds, these young people undoubtedly will form the elite class of the
future, and are very likely to work towards maintaining the normative value
of SAE, or even conservative (exonormative) English as opposed to ethnically
marked BSAE (‘elite closure’ in Myers-Scotton’s terms (1993); see also Wade
1997). Third, many speakers of African languages experience a love–hate rela-
tionship with English, and find themselves forced to make an effort to master it
only for instrumental reasons. Many learners of English experience deep am-
bivalences in their relationship to it, and Peirce (1995: 19) uses the notion of
‘investment’ instead of instrumental or integrative motivation to explain the
apparent contradiction between learners’ motivation to learn English and their
sometimes ambivalent desire to speak it. An investment in English will pay off
only in certain aspects of the lives of non-English South Africans, and it needs
to be worth their effort.
In addition it will be increasingly important to monitor ongoing changes in
the role and status of BSAE. Patterns of language choice and use are related to
socio-economic and political processes, and to the distribution of knowledge
and power, and for these reasons the role, status and development of BSAE in
South Africa are likely to change dramatically in the next decade, significantly
altering patterns of communication.
Peirce (1990: 108) argued that ‘People’s English . . . is a struggle to appro-
priate English in the interests of democracy in South Africa’: nevertheless one
must ask to what uses such a variety (if it is one) is put, and what meanings
it may carry. Several educationists support efforts to maintain some sort of a
372 V. de Klerk and D. Gough
standard (Wright 1996; Titlestad 1996), while others argue persuasively for
the need to resist these pressures (Webb 1996) and advocate the recognition of
naturally evolved local forms of English suited to the needs of their speakers.
Wright (1996: 154) points out the irony inherent in the support of BSAE by
elite black writers and academics who express their views in sE.
The problem seems to lie in interpretations of the concept of ‘democratic
language rights’. For some, this implies the need to recognise and promote
the localised forms that language takes as equally viable and effective as the
standard. For others, it implies an obligation (on the part of the state and edu-
cators) to deliver to learners a model of English that has international currency
and will afford them the advantages that others the world over enjoy once they
have mastered sE. Proponents of this point of view recognise that black vari-
eties of English are legitimate and meet the immediate communicative needs
of local speakers. However, they also see such varieties as stigmatising their
speakers in wider linguistic contexts, and limiting their comprehensibility and
their opportunities to participate in the global village on an equal intellectual and
economic footing with speakers of other varieties of English which have greater
currency worldwide. The attitudes towards BSAE of both South Africans and
other speakers of English will need careful investigation before any serious
claims can be made about changes in the status of BSAE.
English, has on the other prevented them from acquiring sufficient competence
in it to further their personal ambitions, by failing to provide adequate support
for its acquisition and by severely limiting access to mother-tongue English
speakers. Most speakers of African languages encounter very little English
of any kind, and it could be argued in some cases that they do not speak a
recognisable variety of BSAE, but that each has arrived at a different stage
on a learner–language continuum (de Kadt 1993: 314). The very rudimentary
(pidgin-like) forms of English used by many black South Africans cannot be
classified as representative of BSAE, and in such cases speakers could not
be termed ‘bilingual’ in the stricter sense of the word. Their level of compe-
tence in English reflects an incomplete educational process and this explains
why descriptions of black varieties of English often tend to highlight negative
aspects and to compare them with better-recognised varieties of English, rather
than acknowledge the development of a unique BSAE.
Indeed, to view BSAE uncritically as a means of access to power and self-
improvement which will automatically be accompanied by a range of social and
educational benefits is grossly misguided. At present BSAE offers no guarantee
to any of these, and ‘to lead students to believe that there is a one-way relation-
ship between particular genres taught in school and those positions [of power] is
to set them up for disappointment and disillusion’ (Street 1993: 122). Literacy
is a set of social practices that function to empower or disempower people, and
the real literacies of true power, while being understood implicitly by those who
use them, in commerce and government, are not taught in educational institu-
tions. They certainly are not taught through the medium of BSAE, written or
spoken. It would probably be true to say that the vast majority of black South
Africans achieve, at best, a functional command of English, enabling them to
understand signs, read newspaper headlines, fill out applications, etc. They lack
the more empowering cultural and critical literacies (see Williams and Snipper
1990) which usually operate through more prestigious forms of English.
devising L2 syllabuses for the indigenous languages is slow, and support among
mother-tongue speakers for their own languages is worryingly low. This atti-
tude is partly attributable to apartheid policies of the past: while the Nationalist
government ensured the development of the nine indigenous languages (via
separate language boards and enforced mother-tongue instruction), this rein-
forced a view (among mother-tongue speakers) that these languages must be
inferior if they were reserved for black people. This attitude shows now in
rising registration figures of black children at English-medium schools, and in
increasing evidence of lack of support from mother-tongue speakers for their
indigenous languages. In addition, reports of some African children speaking
English at home and African languages at school are indicative of an identity
crisis among these children.
and entrenched, while the masses will find themselves unable to improve their
own English because of a massive national decline in competence. The ques-
tion that therefore remains to be answered is the degree to which variants of
BSAE will drift away from L1 English before a backlash arises, either from
educators or from the learners themselves. The success of current efforts to
resist value judgements and recognise the worth of BSAE will depend not only
on the goodwill of South Africans, and on the co-operation of all speakers of
English, worldwide, but on the rate at which the variety drifts away from recog-
nised standard forms of English. It remains to be seen whether a recognisable
variety of BSAE will make its mark proudly and globally as a distinctive and
recognisable variety of English, equal in all respects to British, Australian or
South African standard varieties.
notes
1 Platt et al., however, seem uncertain about the status of BSAE, since to their thinking
it developed in a territory where there was a sizeable presence of L1 speakers.
2 The interested reader is referred to Gough (1996) for a fuller, more comprehensive
literature survey of relevant works in this area. Reference to all of these works has not
been repeated in this chapter.
3 The big differences in estimates probably relate to the question of what constitutes
‘knowledge of English’ in qualitative terms – a highly problematic concept (see Gough
1996: 53).
4 Statistics from the Development Bank of South Africa cited in Democratic Party
discussion document (1995: 2).
5 Referees and injury time are better considered loanwords rather than examples of
code-switching.
6 One needs to be careful to distinguish between accent and other features of BSAE.
While the phonological effect on English of indigenous black languages is distinctive
and almost unavoidable, the syntactic and lexical aspects of BSAE are not much in
evidence in formal contexts of use at this stage.
bibliography
Adendorff, R. and M. Savinni-Beck 1993. ‘The teaching of English vowels and conso-
nants in the new South Africa’. Journal of Language Teaching, 27, 3: 232–48.
Adey, A. D. 1977. ‘South African Black English’. English Usage in South Africa, 8, 1:
35–9.
Bamgbose, A. 1992. ‘Standard Nigerian English’. In B. Kachru (ed.), The Other Tongue:
English across Cultures. Urbana: University of Illinois Press, pp. 148–61.
Bokamba, E. G. 1992. ‘The Africanisation of English’. In B. Kachru (ed.), The Other
Tongue: English across Cultures, 2nd edn. Urbana: University of Illinois Press,
pp. 125–47.
Branford, W. 1987 (ed.). The South African Pocket Dictionary. Cape Town: Oxford
University Press.
Bua 1993. Facts and Figures: laat die feite self praat. 8(3) September.
376 V. de Klerk and D. Gough
Buthelezi, Q. 1995. ‘South African black English’. In R. Mesthrie (ed.), Language and
Social History: Studies in South African Sociolinguistics. Cape Town: David Philip,
pp. 242–50.
Central Statistical Services 1994. STATS, monthly statistical and marketing digest, 30,
4. Pretoria: Central Statistical Service.
Chick, J. K. 1985. ‘The interactional accomplishment of discrimination in South Africa’.
Language in Society, 14: 299–326.
1989. ‘Intercultural miscommunication as a source of friction in the workplace and in
educational settings in South Africa’. In O. Garcia and R. Otheguy (eds.), English
across Cultures, Cultures across English. Berlin: Mouton de Gruyter, pp. 130–60.
Crawhall, N. 1993. ‘Consenting adults: talk about sex and AIDS’. Bua, 8, 2: 23–5.
de Kadt, E. 1993. ‘Attitudes towards English in South Africa’. World Englishes, 12, 3:
311–24.
de Klerk, V. 1996a (ed.). Focus on South Africa. John Benjamins: Amsterdam.
1996b. ‘Use of and attitudes to English in a multilingual university’. English
World-wide, 17, 1: 111–27.
1997. ‘Encounters with English over three generations in a Xhosa family: for better
or for worse?’ In E. W. Schneider (ed.), Englishes around the World: Studies in
Honour of Manfred Görlach. Amsterdam: John Benjamins, pp. 97–118.
de Klerk, V. and G. Barkhuizen 1998. ‘Language attitudes in the South African National
Defence Force: views from the Sixth SA Infantry’. Multilingua, 17, 2–3: 155–80.
Democratic Party 1995. Discussion and information document on language. Unpub-
lished manuscript.
D’Souza, J. 1992. ‘The relationship between code-mixing and the new varieties of
English’. World Englishes, 11, 2/3: 217–23.
Genrich de Lyle, D. 1985. ‘Theme in Conversational Discourse: Problems Experienced
by Speakers of Black South African English, with Particular Reference to the Role
of Prosody in Conversational Synchrony’. MA thesis, Rhodes University.
Gough, D. 1994. ‘English people and people’s English’. In E. Reckwitz, L. Vennarini
and C. Wegener (eds.), The African Past and Contemporary Culture. Essen: Die
Blaue Eule, pp. 221–31.
1996. ‘Black English in South Africa’. In de Klerk (ed.), pp. 53–78.
2000. ‘Discourse of students’ experience of higher education’. In B. Liebowitz and
Y. Mohamed (eds.), Routes to Writing in Southern Africa. Cape Town: Silk Road
International, pp. 43–58.
Herbert, R. K. 1996. ‘The meaning of language choice(s): social and pragmatic factors
reconsidered’. Unpublished conference handout, First International Conference of
African Linguistics, Swaziland.
Hundleby, C. E. 1964. ‘Xhosa–English Pronunciation in the South-East Cape’. Ph.D.
thesis, Rhodes University.
Jacobs, M. 1994. ‘Consonantal variation in Zulu English mesolect’. South African
Journal of Linguistics, 12, 1: 16–25.
Jefferay, C. R. 1993. ‘The Design, Implementation and Evaluation of an English
Language Development Component within Grahamstown Community Project’.
MA thesis, Rhodes University.
Jowatt, D. and S. Nnamonou 1985. Common Errors in English. London: Longman.
Kachru, B. B. 1986. The Alchemy of English. Oxford: Pergamon Press.
Black South African English 377
Language contact
New urban codes
19 The lexicon and sociolinguistic codes of
the working-class Afrikaans-speaking
Cape Peninsula coloured community
Gerald L. Stone
1 INTRODUCTION
Ethnographic research between 1963 and 1991 (Stone 1991) has confirmed that
members of the working-class Afrikaans-speaking Cape Peninsula coloured
community speak a distinctive dialect (mother tongue of a region or commu-
nity), and not merely slang. The dialect is a marker of the community’s identity,
which is reflected in endogamy, ties of descent, kinship and preferential asso-
ciation, and shared residential areas, both voluntary and enforced. Irrespective
of ‘racial’ appearance, members of this community have tended throughout the
period of research to identify themselves informally as bryn1 ‘brown’; the for-
mal English translation, ‘coloured’, has to some extent different connotations,
although the denotation is identical.
Very broadly speaking, coloured identity is part of a national system of
communal identity formation whose two poles are ‘black’ and ‘white’. Coloured
identity is regarded as intermediate, paradoxical, anomalous, deracinated and
liminal in South African society (Turner 1969). This opens it to ambivalence:
one the one hand, to sacralisation as humble, egalitarian and creative of identity,
and on the other to stigmatisation as bastardised, outcast and destructive of
identity. To the extent that the national system is consensually maintained,
coloured communal identity may well continue as long as the system does
(despite the collapse of legalised apartheid).
White domination promoted, exploited and rigidified this system, with vir-
tually total success from 1948 until it bred the first mass civil rebellion in black
and coloured communities in 1976. The term ‘coloured’, and even more its
formal Afrikaans version, Kleurling, were thus doubly stigmatised, and the
very constructs of communal identity and ethnicity themselves were anathe-
matised as ‘racist’ by the politically helpless and humiliated coloured middle
class. The traditional working class, on the other hand, have contended that this
stigmatisation and disavowal of coloured identity were fraudulent, and that the
confirmed middle class were snobs (sturfies, ‘stiffies’, ‘starched’), who had be-
trayed the working class and hou hulle wit (‘act white’, ‘pretend to be white’),
381
382 G. L. Stone
and have sought favour with the politically dominant of the time, white or
black, against the interests of the working class and against their own nature.
For the traditional working class, class has been colour, and colour, authentic,
unchangeable nature. For the middle class, colour has been class, and class,
changeable culture.
Between 1963 and 1990 the size of the Peninsula community classified
‘coloured’ increased by about 317 per cent (Stone 1991: 146). On demographic
evidence I have suggested that the size of the Peninsula coloured speech com-
munity dominant or bilingual in the working-class Afrikaans dialect increased
by 365 per cent to about 863,000 speakers during this period (Stone 1991: 147).
This growth cannot be attributed wholly to natural increase, but also reflects
large-scale immigration from rural and other urban areas. It is impossible to fix
precise geographical boundaries to the dialect, which begins to fade among the
working class into more middle-class Afrikaans beyond greater Cape Town and
Atlantis, but it can be heard to varying degrees – especially among speakers
who present themselves as ‘disreputable’ – in the larger Peninsula and Boland
towns: Stellenbosch, Paarl and even Worcester.
Marked changes in inter-dialectal and intra-dialectal codes have been
wrought by many factors, including the imposition of the Group Areas Act and
the establishment of townships on the Cape Flats, rapid growth of population
and of the national economy, upward mobility, the introduction of television
and the decline of white domination since 1976. Notably there has been an
increasing shift to bilingualism or dominance in middle-class English and to
a far lesser extent middle-class Afrikaans. There have also been increases in
lexical innovation following civil rebellions in 1976, 1980 and 1985; a consid-
erable increase in the size of the speech community’s lexicon and versatility
in code-switching; and the sudden emergence in 1980 of the lexis of prison
gangs into the lexis of delinquent gangs outside prison. Since the repeal of
all the major legislation of white domination in 1991, a trend towards indi-
vidualism has intensified, and there has been a greater degree of upward and
downward mobility. The working-class coloured trend towards the domestic,
or at least public or formal, use of middle-class English or working-class vari-
ants of English has also grown rapidly. Since the mid-1960s it has become
increasingly common to encounter parents who converse with each other in
the dialect, rear their children in a variant of English, and prefer English mass
media.
Since the international resurgence of Islam in 1973 the Peninsula Muslim
community of Indonesian and Indian origins has increasingly switched from
Malay to Arabic lexis in its variant of the working-class dialect, and espe-
cially in customary religious discourse, which was previously Arabic in refer-
ence to religious ritual and Malay in reference to communal custom. Finally,
The lexicon and sociolinguistic codes 383
since 1985, urban working-class and ex-peasant Xhosas in the Peninsula have
increasingly adopted the dialect as their variant of Afrikaans in discourse
with coloureds and whites. (Urban working-class Zulus, from Natal or the
Witwatersrand, have tended to continue to prefer English.)
2 METHOD
gangs, which contains a useful if partly unreliable and misspelt list of over four
hundred items of what they refer to as ‘jargon’ (1984: 187, my translation),
including many items which are dialectal but not prison lexis.
Items unique to a friendship group or an individual were excluded from the
verified lexicon. I have often found such lexical innovations, but have never
encountered an individual who claims to have introduced an item that has
entered the communal dialect. On the basis of daily usage, habitual alertness
and frequent sociolinguistic discussion with speakers since 1975 (except for
a short break in recording) I conclude that I have thus far recorded 90 per
cent or more of all the dialectal items in present or past use by speakers who
have become adolescent between 1963 and 1991. The current verified lexicon
numbers nearly five thousand items (counting multiple meanings in some cases),
and several hundred more are in the process of verification.
A far fuller account of the ethnographic and lexicographic issues and meth-
ods is in Stone (1991: 148–250). To my knowledge this chapter (originally
appearing in 1995) is the first publication on lexicography of the dialect as
a whole. There is some unpublished work, on segments only: Kotzé’s (1983)
study of lexical variation of Afrikaans among Java Muslims (of claimed
Indonesian descent), Heilbuth’s (1984) lexicon of the small Cloragail argot of
moffies ‘effeminate male homosexuals’, Fagan’s (1984) translation of a letter
ordering prison-gang murders in prison-gang lexis on toilet paper and Stone’s
work (1991) setting out the Disreputable Lexicon of adolescent and young
adult males. These and Lötter and Schurink (1984) apart, no formal or disci-
plined lexicography appears to have been undertaken. I have also made use
of Makhudu (1980) and Mfenyana (1981), which contain lexis in Flaaitaal
or Tsotsitaal on the Witwatersrand (including some items found also in the
Peninsula dialect).2
No other community in South Africa has had to contend with such universal and
extreme stigmatisation – including self-stigmatisation – as coloured commu-
nities and their subordinated antecedents. For three centuries prior to the elec-
tions of April 1994, every daily act of political subordination, indifference or
resistance has been informed by consciousness of outcastness and non-entity
unique in the country. This has been too formative to be ignored or readily
transcended, and remains too harrowing and mortifying even to be verbalised.
Moreover, coloured communities have been the receptacle of universal con-
tempt by white, black and themselves, as uniquely lacking in ‘racial’ and cultural
integrity.
Halliday (1978: 181) defines a (sociolinguistic) code ‘as a systematic pattern
of tendencies in the selection of meanings to be exchanged under specified
The lexicon and sociolinguistic codes 385
English translation of it, with highly idiosyncratic spelling. In the past few
years, commercial television and radio advertisements have utilised stereotypic
folk characters speaking the dialect.
The stigma attached to the dialect has waned since the 1960s. Yet it is still
regarded as the marker of static communal identity, and currently all speakers
(apparently the large majority) who aspire to upward mobility periodically code-
switch into middle-class English or Afrikaans, signifying their sociolinguistic
competence in upward mobility, especially to middle-class strangers.
Apart from the peripheral lexicons of Java Muslims, ‘moffies’ and Rastafar-
ians, and the specialised jargons of speakers in the construction and fishing
industries, the lexicon consists of a hierarchy (in terms of socio-economic
status and psychosocial development) of four lexicogrammatical codes, sig-
nifying the enaction of four corresponding working-class identities. Speakers
implicitly assign all the dialectal lexis to one or more of these codes. They
are respectively the ‘respectable’, ‘disreputable’, ‘delinquent’ and ‘outcast’.
Codes are commonly switched in address to different respondents and during
discourse with the same respondents. For instance, it is utterly inappropriate to
use the delinquent code to one’s respectable mother, and the respectable code
when the enaction of delinquency is expected by one’s delinquent peers, unless
one intends to signify disorientation or defiance.
Codes are implicitly graded internally: respectable/middle class, respectable/
disreputable, disreputable/respectable, disreputable/delinquent, delinquent/
disreputable, delinquent/outcast, outcast/delinquent and outcast/silence. The
terms of the corresponding intracommunal identities are set out in detail in
Stone (1991: 245 and 250–359). However, it is impossible to understand the
codes – and thus begin to understand the consciousness of the speech commu-
nity – without some brief outline of the corresponding identities.
As is apparently universal, all identity enaction is constructed in terms of
the religious antinomy of ‘Nature’ versus ‘Culture’ (Lévi-Strauss 1964). In this
speech community, Natural identity is regarded as consisting of one’s body and
all one’s innate urges, impulses, emotions, limitations and abilities. Unless con-
gruent with Culture, Natural identity is regarded ambivalently, on the one hand
stigmatised as potentially egocentric and antisocial, on the other hand sacralised
as a source of individual survival, vitality and identity in the face of anti-Natural
Culture. Natural identity enaction regarded as unmannerly, uncouth, obscene
or uncivilised is termed rou (‘raw’, ‘crude’). Culture is regarded as consisting
of all honourable, prosocial self-regulation and self-creation that tend to psy-
chosocial and communal development and oppose anti-Cultural Nature. One
must bear in mind that these are folk constructions. In scientific terms, all iden-
tity enaction is psychosocially acquired and thus cultural, whether or not it is
construed in folk terms as Natural.
The lexicon and sociolinguistic codes 387
3.1 Respectability
Speakers enacting respectability describe themselves as decent mense (decent
people), but give the Respectable Lexicon no particular name. Respectability
is regarded as sacramental Cultural omnipotence over stigmatised Nature in all
one’s identity enactions, and thus as the pursuit of perfect order, stability, privacy
and development in upward socio-economic mobility or at least avoidance of
downward mobility. It is a struggle for control over the imputation or enaction
of stigmatising disreputability. Decent mense define themselves in opposition
to skollies ‘delinquent ruffians, riff-raff’ – the stigmatised Natural – who are
said to act ignorantly, vulgarly and malignantly in public and indeed to have no
standards of decency at all.
The Disreputable, Delinquent and Outcast Lexicons are collectively termed
skollie tale ‘riff-raff terms’ by the respectable (compare with the Witwatersrand
term ‘Tsotsitaal’), and ou roeker tale by the disreputable, delinquent and out-
cast, who scrupulously avoid the term skollie as sanctimonious and dishonouring
(except as a humorous verb meaning ‘to scavenge’ or as the proper name for a
domestic dog). Ou roeker (delinquent male adolescent or adult, unless stated as
a child or female) literally means ‘old smoker’, one who has rebelliously been
smoking cigarettes and perhaps dagga ‘cannabis’ even prior to adolescence.
3.2 Disreputability
The boundaries of disreputability may be focused or extended. Focused disrep-
utability (referred to as ‘disreputability’ except where otherwise stated) rejects
confirmation in delinquency, let alone outcastness. Extended disreputability
includes confirmed delinquency and outcastness. In disreputability, Culture –
mainly the rules of respectability – is ambivalently violated and intermittently
yielded to Nature. Disreputability is the pursuit of instability, ambivalence,
optionality, lability, aggrandisement and creativity in communal identity and
in the conflict between Culture and Nature. It may be frank or ambiguous. If
ambiguous, it may engage in masquerade – flirtation with, or passing as, an-
other communal identity. And it proposes itself as ‘I-don’t-care’ in flaunted
imperviousness to all consequences.
From 1975 to 1990 disreputable/respectable and respectable/disreputable
adolescent and young adult males were termed cats: playful, upwardly mo-
bile, English-dominant or bilingual individuals sharing a highly gregarious
and public subculture that idealises modernity, pop music, material display
and adult sophistication. Cats were sharply differentiated by outies (English
lexis, from outlaw, exotically Ngunicised pronunciation as in awuti) from
themselves. An outie is a disreputable/respectable or disreputable/delinquent: a
388 G. L. Stone
more impoverished, socially static, serious, tough, rough, plainly spoken male
regarding himself as more authentically masculine and working-class coloured.
Whereas a cat was non-delinquent and non-violent, an outie is ready for vio-
lence and competent in it, although he might not initiate it.
3.3 Delinquency
Delinquency is the violent pursuit of disorder: the publicly flaunted, malignant
triumph of stigmatised Nature over persecutory Culture. The delinquent identity
is the disreputable writ large, stripped of its ambiguity, optionality and ambiva-
lence, and taken to its sadistically rebellious and violent conclusion within the
social network of the community (beyond which the actor becomes outcast).
In oral myth, gang name, graffito, tattoo and enaction, actors of the identity
dramatise and romanticise themselves as heroic criminal warriors, bearers of a
tradition of vengeful, triumphant defiance of the violently sanctimonious, depri-
vatory and exclusive Cultural authority of parents, the respectable, the coloured
middle class, the dominant white stratum of society, and the identity style of
the Christian West. They are contemptuous of the merely disreputable, who
are ineffectual, timid and phony ‘inauthentic’, whereas vollende (‘authentic’,
‘confirmed’, ‘full’) ou roekers kyk their ding (‘pursue their thing to the hilt’) and
dala (‘do the deed’, ‘act boldly, consequentially, ruthlessly and remorselessly’;
from standard Zulu dala, ‘create’, ‘form’, ‘conceive’, ‘cause’).
3.4 Outcastness
Outcastness is the nameless spectre that has haunted Peninsula coloured iden-
tity since its formation from the ruins of Khoekhoe and San society and slavery.
Outcast identity is disavowed and alexicalised: discourse on it is always avoided
except in utter despair, and it has no name. For total outcastness signifies dis-
integration, annihilation and death, and any identity – however stigmatised – is
to be embraced as preferable.
Outcast identity is the pursuit of chaos and disintegration in violence
(rebellion against fate) or silence (submission). The identity is located below the
hierarchy of working-class communal identities, and is regarded as outside soci-
ety and Culture, in the community but not socially of it. Outcastness is victimisa-
tion by both stigmatised Nature and persecutory Culture, precluding or disqual-
ifying the actor from normative participation in communal life. The stigma may
be external (racial appearance, gross physical deformity, crippling or profound
injury) or internal (Nature: severe or recidivistic criminality, especially mur-
der, addiction, compulsive promiscuity in females and homosexuals, histrionic
effeminate homosexuality, paedophilia, parent–child incest, sadomasochistic
The lexicon and sociolinguistic codes 389
Western society since the time of Aristotle, and now academically termed ‘com-
munication’. Architecture, religion, sport, dress, violence, rituals, propaganda
and advertising are of course all forms of rhetoric.
I suggest that the dialect constitutes linguistic bricolage. The ‘ends’, the ‘stan-
dard’ dialects from which it is composed, are appropriated and adeptly made
to constitute a new ‘means’, the working-class dialect, under the noses (so to
392 G. L. Stone
Language actively symbolizes the social system, representing metaphorically in its pat-
terns of variation the variation that characterizes human cultures. This is what enables
people to play with variation in language, using it to create meanings of a social kind, to
The lexicon and sociolinguistic codes 393
participate in all forms of verbal context and verbal display, and in the elaborate rhetoric
of ordinary daily conversation. It is this same twofold function of the linguistic system,
its function both as expression of and as metaphor for social processes, that lies behind
the dynamics of the interrelation of language and social context. (Halliday 1978: 3)
Equally, language
not only serves to facilitate and support other modes of social action that constitute its
environment, but also actively creates an environment of its own, so making possible all
the imaginative modes of meaning, from backyard gossip to narrative fiction and epic
poetry. The context plays a part in determining what we say; and what we say plays a
part in determining the context. (Halliday 1978: 3)
With this theoretical framework, positing that language and reality generate
each other, Halliday provides the context for the social semiotic presentation
of a phenomenon that he terms ‘antilanguage’, i.e. the ambivalent reversal of
the rules of a referent language.
Detailed exegesis of this working-class dialect and possibly others as anti-
language is precluded here (cf. Stone 1991: 68–77, 517–35). I can only intro-
duce such an exploration by pointing in this dialect to the intracommunal
and sociolinguistic hierarchy of the ‘standard’ referent codes (of middle-class,
‘standard’ Afrikaans and English and standard Zulu) and the four working-class
codes of respectability, disreputability, delinquency and outcastness in the fol-
lowing terms: on the one hand, of corresponding intracommunal identity and
degree of stigma, dishonour, disorganisation, antagonism towards and linguistic
distance from referent language(s); and on the other hand, reversal of these by
counter-reality, creative relexicalisation, overlexicalisation, metaphoric con-
notation and rhetorical foregrounding of interpersonal meaning and social
hierarchy.
The selections below, constrained by space, are chosen for their inclusion of
items especially significant in terms of the corresponding communal identity’s
frame of consciousness, myth and ritual. Undated items were in circulation
before, during and after the research period for this chapter, 1963 to 1991.
law and ways of freedom of the world outside as opposed to the hlahluka
(or tambuku) van die Point: the totalistic brutality and fatefulness of the prison
world, mythified as ‘another world’ (translation).
7 CONCLUSION
Freud (1900: 608) observed that ‘the interpretation of dreams is the royal road
to a knowledge of the unconscious activities of the mind’. Similarly lexis, in the
ethnographic context of discourse, social situation, ritual, myth and community
or society, may be described as the royal road to the consciousness, reality,
ideology and identity of the speech community.
notes
1 adjective: literally, ‘brown’; noun in plural only, brynes. See lexicon entries below.
2 Flaai is not found in Afrikaans. I suggest the spelling ‘Flytaal’, i.e. ‘sly language’, from
the British English slang fly, ‘artful’, ‘wide awake’, ‘knowing’ and fly-boy, ‘artful,
knowing man’, usually working class. See Stone (1991: 405) for fly-boy, ‘habitually
cunning working-class youth or man’, sometimes a nickname, usually for male who
has traded illicitly in drugs or liquor since adolescence; fly-fly, ‘casually’, ‘briefly’,
‘in passing’, i.e. circumspectly; and fly padda, ‘sly chap’, in the Peninsula dialect
(frogs are usually hidden, appear watchful, and fall silent when exposed).
3 Section 6 contains selected quoted material from Stone (1991).
bibliography
Brillouin, L. 1968 [1949]. ‘Life, thermodynamics and cybernetics’. In W. Buckley
(ed.), Sociology and Modern Systems Theory. Englewood Cliffs: Prentice Hall,
pp. 147–56.
Fagan, E. 1984. ‘An Examination of Prison Gang Language: Analysis of a Letter’. LLB.
research project. Institute of Criminology, University of Cape Town.
Freud, S. 1900. The Interpretation of Dreams. London: Hogarth Press.
Halliday, M. A. K. 1976. ‘Antilanguages’. UEA Papers in Linguistics, 1: 15–45; reprinted
in Halliday 1978, pp. 164–81.
1978. Language as Social Semiotic: The Social Interpretation of Language and
Meaning. London: Edward Arnold.
Heilbuth, D. 1984. ‘Cloragail: The Code the Queens Use on the Cape Flats’. BA research
report, Department of English, University of Cape Town.
Kotzé, E. F. 1983. ‘Variasiepatrone in Maleier-Afrikaans’. Ph.D. thesis, University of
the Witwatersrand.
Lévi-Strauss, C. 1962. The Savage Mind. Chicago: University of Chicago Press.
1964. The Raw and the Cooked. London: Cape.
Lötter, J. M. and W. J. Schurink 1984. Gevangenisbendes: ‘n Ondersoek met Spesiale
Verwysing na Nommerbendes onder Kleurlinggevangenes. Pretoria: Human
Sciences Research Council.
The lexicon and sociolinguistic codes 397
K. D. P. Makhudu
398
An introduction to Flaaitaal 399
3 SPEAKERS OF FLAAITAAL
20.1 Townships in the PWV (now Gauteng) area during the apartheid era
Tuka-toun as their sources, respectively. The Flaaitaal speakers of the West Rand
townships of Kagiso and Mohlakeng cite Munsieville and Madubulaville, and
those from the East Rand cite Dukathole, Dindela and Etwatwa as their places
of origin. The political significance of keeping alive the memory of the old
townships should not be overlooked; few of the townships’ property holders
were adequately compensated for the deprivation of their property rights. One
use of Flaaitaal was to denote resistance or defiance; the common cry in the
Defiance Campaign against the infamous Group Areas Act was Ons dak nie,
ons phola hier! (literally ‘We won’t move, we’re staying put!’). Interestingly,
this sentence shows the interweaving of the grammar and lexis of the dispos-
sessor – dak from Afrikaans slang nak, ‘to leave’ or from English duck (v.) and
phola from Zulu ukuphola, ‘to be cool, to sit down and reflect’.
There is one other strand in the origins of Flaaitaal: the contribution of
coloured speakers. In Kimberley and suburbs adjacent to Johannesburg such as
Eldorado Park, Eersterus, Rust-Ter-Val, Bosmont and Riverlea coloured male
speakers employ a variety of Flaaitaal and have, over the years, contributed to the
association of Flaaitaal with Afrikaans (D. Mattera, personal communication).
The relationship between Flaaitaal and non-standard lects of the Cape such as
Kaaps, Skollietaal and Gamtaal needs to be thoroughly researched.
4 CONTEXTS OF USE
Some studies of Flaaitaal or Tsotsitaal have uncovered links with prison varieties
in gaols such as Leeuwkop in Bryanston and the Fort in Johannesburg. These
connections with hardened gangs such as the Big Five and the Jerries (Germans
or Majeremane), whose members are serving or have served long-term sen-
tences, have given Flaaitaal speakers a notoriety that has overshadowed the
reality of its communicative function. Several writers, such as Schuring (1983),
Mfusi (1992) and Ntshangase (chap. 21, this volume) accept such criminal links
but point out that Flaaitaal is used in a range of contexts that go beyond the
underworld.
Flaaitaal can count poets and authors among its users, notably Sipho Sepamla,
Achmat Dangor, Essop Patel and Gibson Kente (Mutloatse 1981). The inclu-
sion in the list of Dangor and Patel, who are not mother-tongue speakers of
an African language, suggests that Flaaitaal is spreading as a general urban
phenomenon. The use of Flaaitaal in literary texts should not create the im-
pression, however, that Flaaitaal is stable and unchanging. It has other eminent
users, if only for symbolic purposes such as black solidarity. These include
the well-known jazz musician Hugh Masekela (originally from Sophiatown)
and Archbishop Desmond Tutu (originally from Munsieville) who both claim
to use Flaaitaal when speaking casually with other Flaaitaal-using friends
(Sunday Times, 25 April 1993; 28 November 1993). Around the time of a
402 K. D. P. Makhudu
new post-apartheid order, ANC leaders tried to establish a rapport with their
supporters by making the occasional remark in Flaaitaal, notably at the begin-
ning of rallies. Few observers will deny the effective use of Flaaitaal by Cyril
Ramaphosa (a prominent ANC politician and former trade unionist) to praise
ANC leader Nelson Mandela at the April 1993 funeral of Chris Hani: Heitha,
Comrade Pres. Mandela, Heitha ‘Hail to you President Mandela, Hail’.1
Flaaitaal can best be understood in terms of Halliday’s concept of ‘anti-
language’.2 In his terms, an anti-society is a society that is set up within another
society as a conscious alternative to it, as a mode of resistance. ‘An anti-language
stands in relation to an anti-society in much the same relation as does a language
to society. The simplest form taken by an anti-language is that of the creation of
new words for old; it is a language relexicalised’ (Halliday 1978: 165). I shall
illustrate two characteristics of anti-language as outlined by Halliday (1978) and
by Stone (1991) for Cape Coloured varieties. The first is ‘overlexicalisation’: the
anti-language is not merely relexicalised in certain areas, it is overlexicalized;
for example, the word fly/flaai itself denotes ‘city-wise, urbane, slick’. This
dichotomy fly/not fly is represented by words for authentic Flaaitaal persons die
main ou, ‘the main man’, or group terms such as die autis, die ouens and majitas.
This contrasts with the foolish, dim-witted or ‘slow’ person characterised as
mugu [muxu], bari, mumish, pop, gashu [xaʃu:] etc.
The volume of terms for the notion of ‘friend’ or ‘in-group member’ is
overwhelming. These may be adopted from other South African languages or
be new coinages.
FT word Source language
bra [bra] brother (English)
bab [ba:b] baba (‘father’ – Zulu)
bri [bri] brigade (English, French, Italian)
budi [budi:] boetie (‘little brother’ – Afrikaans)
brikhado [bri:kado] obrigardo (‘thank you’ – Portuguese)
mri [mri:] mratho (‘younger brother’ – Pedi)
Each of these examples denotes ‘a friend’ in Flaaitaal, yet in each case some
sort of semantic shift has taken place through either calquing or specific phono-
logical processes. These internal processes in the lexicon of the argot render
the language highly changeable, so that anyone who does not keep abreast is
soon left behind by the rapid turnover of vocabulary. For example, the concept
‘friend’ has at any one time up to thirty synonyms. This raises the issue of
whether Flaaitaal is an in-group or ‘secret’ means of communication. This is
certainly the case with the ‘slang’ employed by prisoners but is less likely to be
the case in the generally common township ‘lingo’. However, it is true that each
variety of Flaaitaal used in a specific locality is slightly different. The residents
of Kagiso would differ from those of Atteridgeville in the way they refer to a
‘despised person’, for example:
An introduction to Flaaitaal 403
FT word Locality
bari [bari] generic
barzen [barzən] Kagiso
mumish [mɒmi:ʃ] Atteridgeville
jankrap[daŋkrap] Meadowlands
dat [da:t ] Eldorado Park
mqhaka [m!aka] Sharpeville
hamish [hami:ʃ] Rockville/Soweto
The second feature of the anti-language that Halliday (1978: 175) noted as
a defining characteristic is its metaphorical character: ‘An anti-language is a
metaphor for an everyday language; and this metaphorical reality appears all
the way up and down the system. There are phonological metaphors, gram-
matical metaphors and semantic metaphors.’ Those structures and collocations
are usually self-consciously opposed to the norms of the established language
or languages. Examples from Flaaitaal of processes that fit Halliday’s scheme
abound, and I shall describe only a few here (for further details consult Makhudu
1980):
5.1 Nasalisation
Sound change English/Afrikaans FT Gloss
/b/ > /m/ beer /miya/ beer
baikie /maikie/ jacket
/d/ > /m/ dom /mom/ foolish
/f/ > /m/ vang /maŋ/ arrest
/v/ > /m/ vest /mesten/ clothing
/p/ > /m/ papier /mamir/ papers
/t/ > /n/ timing /naimiŋ/ wise
met or moet /mun/ with
/d/ > /n/ dak or duck /nak/ dodge
5.4 Reduplication
(a) Complete reduplication with change of meaning:
Stem Language Meaning FT FT meaning
nice English nice /naiza-naiza/ party
vang Afrikaans catch /fang-fang/ hit repeatedly
-thenga Zulu to buy /thenga-thenga/ cheap woman
-tama FT to eat /tama-tama/ delicious food
(b) Partial reduplication
Stem Language Meaning FT FT meaning
ndama FT money /ndadama/ money
blackjack English policeman /jekeja/ policeman
dikoto Sesotho clubs /makotokoto/ a firearm
snaaks Afrikaans funny /snakanaka/ a foolish person
An introduction to Flaaitaal 405
6 CONCLUSION
The widespread and increasing use of Flaaitaal raises questions about its status
in the future. It is certain to continue flourishing in urban multilingual centres,
and to continue influencing standard forms of the African languages. In the past
communicative needs in the urban centres led to the formation of this informal
and lively means of expression among initial strangers who then became close
associates through it. This makes Flaaitaal similar in many ways to pidgin and
creole speech forms. Whether, like creole languages, Flaaitaal will stabilise
into a first language is uncertain. The matter is certainly deserving of future
research.
notes
1 As Ntshangase notes (chap. 21, this volume), this particular form of greeting is com-
mon to Flaaitaal and Iscamtho.
2 I am indebted to Stone (1991) in this section for drawing my attention to the applica-
bility of Halliday’s work in the South African context.
bibliography
Halliday, M. A. K. 1978. Language as a Social Semiotic: The Social Interpretation of
Language and Meaning. London: Edward Arnold.
406 K. D. P. Makhudu
1 INTRODUCTION
Language practice, by nature a complex phenomenon, is yet more complex
in Soweto where together with English and Afrikaans many African languages
are spoken in almost every resident’s immediate experience. Apart from the
standard African languages used in Soweto, there is also another form of lan-
guage that seems to cut across all linguistic, political and ethnic barriers created
by the apartheid state but which also reflects other barriers. This language is
commonly called by its speakers Iscamtho [is/amtho]. This name is probably
derived from the Zulu word ukuqamunda [uk’u!amunda], which means to talk
volubly.1 Iscamtho has been confused with Flaaitaal or Tsotsitaal, with which
it has many parallels.
The purpose of this chapter is to show that Iscamtho is a different variety
from Flaaitaal. Iscamtho has very strong leanings towards Zulu and Sotho:
both of these influence the lexical base of Iscamtho even though there are
social and linguistic differences between them. Iscamtho also forms a very
important marker of urban identity, particularly a Soweto identity which re-
flects a number of social phenomena. Languages are not abstract entities but
important social and historical phenomena which bind, and sometimes re-
flect cleavages within, communities. Thus, Iscamtho reflects an urban iden-
tity and, at the same time, the social barriers between its users and
non-users.
Iscamtho is a language that is used ‘through’ another language – a type
of basilect, yet it retains its own defining features, i.e. it has no structure of
its own since it relies heavily on the language structures of the languages
from which it ‘operates’. This means that it has not yet developed its own
syntactic base which will make it linguistically independent of the base
languages.
In Soweto, Iscamtho is used mainly ‘through’ Zulu and Sotho. Below, I offer
renditions of the English sentence ‘I am going’ in both the Sotho-based and
the Zulu-based Iscamtho, Standard Zulu and Standard Sotho to illustrate the
changes the language undergoes and how it retains its features.
407
408 D. K. Ntshangase
A Zulu speaker will use (1) and an Iscamtho speaker from a Zulu background
will use (1a), a Sotho speaker (2) and an Iscamtho speaker from a Sotho back-
ground (2a). It is worth mentioning here that the etymology of the Iscamtho
word vaya can be traced to the Afrikaans word waai (‘blow’, as in ‘the wind
blows’). Below, I offer renditions of the same English sentence in Flaaitaal and
Afrikaans.
(3) Afrikaans Ek loop.
1sg.-go
(3a) Flaaitaal Ek thler.
1sg.-go
What these sentences illustrate is that Iscamtho draws its lexical base from Zulu
and Sotho while Flaaitaal draws its lexical base from Afrikaans. This linguistic
difference shows that Iscamtho cannot be considered a variety of Flaaitaal, as
Mfenyane (1977; 1981) asserts.
The speech communities of Flaaitaal and Iscamtho derive from totally different
social and historical backgrounds. Flaaitaal emerged, and draws its speech
community from, the freehold townships of the Western Areas of Johannesburg
(Sophiatown, Martindale and Alexandra). Iscamtho, on the other hand, emerged
from an argot called Shalambombo and draws its speech community from the
squatter communities of Orlando, Pimville, the Eastern Native Township and
the Moroka Emergency Camp.
Freehold townships and the squatter communities mark different processes of
African land dispossession and urbanisation; and these two languages not only
reflect different social transformations but also mark permanence in black urban
settlement. Freehold townships were largely occupied by Tswana-speaking
groups from the Western Transvaal, while the squatter communities were largely
Nguni, particularly Zulu and Xhosa. This can be explained by understanding the
different ways in which African communities were dispossessed of their land.
Language and language practices in Soweto 409
There are regional and historical variations in the process, which are important
in understanding the processes and periodisation of urbanisation (see Bundy
1987; Delius 1983).
Iscamtho and Flaaitaal developed as argots or criminal languages. Iscamtho
developed from an argot called Shalambombo used by a criminal gang network
called Amalaita, which operated in and around Johannesburg between 1890
and 1930. The gang, which used a mine dump in Crown Mines as their head-
quarters, lived mainly in Orlando and Pimville and were composed mainly of
Zulu migrants. Their counterparts, AmaRussia (sometimes spelt AmaRashea),
who were mainly Sotho migrants from Lesotho and the Orange Free State,
lived in Newclare and later the Moroka Emergency Camp (see Bonner 1987;
1990). Flaaitaal, on the other hand, developed among the criminal gangs of the
Western Areas who were composed mainly of urban male youths. As Makhudu
notes (chap. 20, this volume), Flaaitaal is also known as Tsotsitaal. Flaai meant
‘citywise’ and tsotsi meant ‘urban citywise and slick’.2
Crime and criminal gangs became very popular in black urban settlements
around Johannesburg between the 1930s and the 1950s and, as Glaser (1990)
says, young children growing up in Johannesburg identified more with criminals
than with professionals. This resulted in the increased use of the languages as-
sociated with criminality. Thereafter Flaaitaal and Iscamtho no longer reflected
the life of the underworld but that of the young and urban-wise, and assumed
an urban identity, which distinguished itself from the rural identity of migrant
workers.
The same is true of language practices in South African prisons. Keswa (1975)
has documented the history and practices of prison gangs in South Africa. Prison
gangs can be divided into two sorts; those who use Flaaitaal (the 26, 25 and
Air Force) and those who use Shalambombo. In interviews I undertook with a
number of former prison inmates and from my experiences in a South African
prison, members of the 28 and 27 prison gangs said they use Shalambombo;
for them it was not exactly the same as Iscamtho, though very similar. They
could understand both Shalambombo and Iscamtho but could also exclude an
Iscamtho speaker if they opted to use Shalambombo.
The linguistic structure of Iscamtho suggests that it has some linguistic pro-
cesses that are different from those of Zulu. It is worth noting that the word
‘Iscamtho’ itself reflects a peculiar linguistic process. As mentioned before, it is
probably derived from the Zulu word ukuqamunda. There is a use of the dental
click (spelt <c>) instead of the palatal click (spelt <q>). This can also be noted
in words such as icanda for egg; the Zulu word is iqanda.
There is also a high degree of vowel elision in Iscamtho. Zulu nouns have a
V (vowel), VCV (vowel–consonant–vowel) or VN (vowel–nasal) structure in
their prefixes, as in nouns like u(V prefix)-limi (tongue); isi(VCV prefix)-lwane
(animal); and um(VN prefix)-fana (boy). The word Iscamtho itself reflects this
410 D. K. Ntshangase
vowel elision process in the noun prefixes, for in Zulu the word would be
isicamtho. This is also evident in a number of other words, e.g. iskole instead of
isikole (school). Vowel elision is an important morphological and phonological
difference between standard Zulu and Iscamtho. We can also note variation in
the syntax of the language. The question form in Zulu ends with the question
formative na, as in sentence 4:
(4) U-ya-hamba na?
2nd sg. prefix-long form tense-go interrog.
‘Are you going?’
However, in Iscamtho there is a different process operative.
(5) Why u-zunda ama-jents?
Interrog. 2nd sg.-hate noun prefix-young men
Why do you hate gents? (gents is derived from gentlemen and means ‘young
men’).
If Iscamtho strictly followed Zulu syntactic patterns, (5) would have been
(6) Yini u-zonda izi-nsizwa na?
Interrog. 2nd sg. -hate noun prefix - young men interrog
This shows that Iscamtho does not use the na Zulu question formative. There
are numerous examples where the na formative is not used, as in (7) below
(7) U-zo-vaya?
2nd sg.-future aux-go
‘Will you go?’
This sentence has a higher intonation contour than an ordinary declarative
sentence. Apart from the differences in the lexical items (i.e. where Iscamtho
uses different words from Zulu with semantic shift), our examples show that
there is a systematic difference in the language structure between Zulu and
Iscamtho. Afrikaans not only influences the linguistic base of Flaaitaal, it also
shapes a majority of its lexical items. Iscamtho, on the other hand, has very little
Afrikaans and a heavy English influence apart from its Zulu and Sotho base.
This has significant political undertones. The anti-Afrikaans events of 1976 had
more to do with this than is commonly recognised. The differences in the base
(or matrix) of Flaaitaal and Iscamtho as well as in the code-switching practices
characteristic of the two varieties are analysed by Slabbert and Myers-Scotton
(1997: 338):
Those versions that we call Tsotsitaal always have a variety of Afrikaans – generally a
nonstandard variety – as their matrix language (ML). In contrast, those versions that we
call Iscamtho always have a South African Bantu language as their ML; most often it is
Zulu, but Sotho-based Iscamtho is also attested, and there may be other versions with
other Bantu languages as their ML.
Language and language practices in Soweto 411
patronise shebeens3 and stokvels4 and speak it to their peers and boyfriends who
are usually shebeen and stokvel patrons themselves. The following are words
associated with female users of this language:
concepts. If this happens in a Zulu lesson, it shows that students do not neces-
sarily speak standard Zulu or even understand all of it.
An increasing number of families use Iscamtho as a first language. Moreover,
the advertising and entertainment industries have also begun to accept Iscamtho.
Today, there are many electronic and print adverts which use Iscamtho as images
of urban culture and communication. These range from adverts on Radio Metro
to designer-label clothing adverts in the press. Many theatre plays also use
Iscamtho. Brenda Fassie, Stimela and Senyaka10 use Iscamtho in their songs.
3 CONCLUSION
An important aspect of both Iscamtho and Flaaitaal is that they embody salient
features of black urban culture. These languages and the urban culture they
support, ironically, involve an acceptance of the townships as ‘home’, even
though they were created by the apartheid state to serve its own interests. Current
evidence shows that these languages are growing in numbers of speakers and
functions. Whereas previously Flaaitaal and Iscamtho were not used within the
family, they are increasingly used in this domain, even by women and children.
notes
1 Doke et al. (1982).
2 Bothma (1952) suggests that the word tsotsi comes from the word tsetsefly, which
is an insect. I would suggest that the word is derived from the Sotho verb go tsotsa,
which means ‘to rob’ and is directly used with the original meaning.
3 Shebeens are family houses where people who want to purchase liquor are allowed
to come and drink inside. They can be seen as a black alternative to pubs in white
suburbs.
4 Stokvels are savings schemes where members invest an agreed sum of money over an
agreed period. Once it is the turn of a member to receive money, he or she is encouraged
to invite many people to come and purchase liquor and meals at an inflated price. This
usually happens over the weekends and is done from Friday until Monday. There are
a number of such schemes which include burial societies, syndicates and ‘kitchen
parties’ (a modern version of the old tea parties). It is estimated that stokvels in South
Africa turn over 30 million rand a month.
5 Glaser (1990) asserts that this word is coined from the 1940s film The Magic Garden.
It is nonetheless used to refer to a young man who is citywise, probably with criminal
inclinations.
6 Interviews with Sizakele Tshabalala, Nandi Tshabalala, Anonymous (preferred
name), Thembeka Galeta, Zinhle Galeta (December 1992); Kholeka Mange, Gabisile
Mange, Lawukazi Mange (January 1993); Maureen Kaunda (February 1993).
7 I am indebted to Steve Lebelo for raising this issue. More investigation has to be done
as to whether apart from Iscamtho there is an urban form of Zulu. Presently I am
414 D. K. Ntshangase
unable to offer any solutions to this but from anecdotal evidence would suspect that
there is an urban form of Zulu; I am not yet sure what distinguishes it from Iscamtho.
8 I am indebted to Dr Adam Ashforth for this realisation that hostel dwellers become
suspicious of users of non-standard Zulu and that the use of standard Zulu serves as a
means of converging with hostel dwellers. It will be interesting to note how language
use between hostel dwellers and Soweto residents became a means of identification
during the political violence of the early 1990s.
9 For more findings on attitudes towards Iscamtho see Ntshangase (1993, chapter 6).
10 Brenda Fassie’s song is ‘I-straight le ndaba’; Stimela’s song is ‘Iscamtho asikho’;
Senyaka’s song is ‘Why uzonda amajents’.
bibliography
Bonner, P. 1987. ‘Desirable or undesirable Sotho women? Liquor, prostitution and mi-
gration of Sotho women to the Rand, 1920–1985’. In C. Walker (ed.), Women
and the Organisation of Gender in South African History. Johannesburg: Ravan
Press.
1990. ‘An evil empire? The Russians on the Reef, 1947–1957’. Paper presented at
the History Workshop Conference, University of the Witwatersrand.
Bothma, C. V. 1952. ‘’n Volkekundige ondersoek na die aard en ontstaans van tsotsi-
groepe en hulle aktiewiteite soos gevind in die stedelike gebied van Pretoria’. MA
thesis, University of Pretoria.
Bundy, C. 1987. The Rise and Fall of the South African Peasantry. London: Heinemann.
Delius, P. 1983. The Land Belongs to us. Johannesburg: Ravan.
Doke, C. M., D. M. Malcolm and J. M. A. Sikakana 1982. English – Zulu Dictionary.
Johannesburg: Witwatersrand University Press.
Glaser, C. 1990. ‘Anti-Social Bandits: Juvenile Delinquency and the Tsotsi Youth
Sub-Culture on the Witwatersrand, 1935–1960’. MA thesis, University of the
Witwatersrand.
Keswa, E. R. G. 1975. Outlawed Communities: A Study of Contra-acculturation among
Black Criminals in South Africa. Pretoria: n. p.
Lacey, M. 1982. Working for Boroko. Johannesburg: Ravan Press.
Lebelo, S. M. 1988. ‘Sophiatown removals and political acquiescence’. BA (Hons)
dissertation, University of the Witwatersrand.
1990. ‘Apartheid’s Chosen Few: Urban African Middle Classes from the Slums of
Sophiatown to the Northern Suburbs of Johannesburg, 1935–1985’. Paper presented
at the History Workshop Conference, University of the Witwatersrand.
Lukhele, A. K. 1990. Stokvels in South Africa: Informal Savings Schemes by Blacks for
the Black Community. Johannesburg: Amagi Books.
McRobbie, A. and Garber, J. 1976. ‘Girls and sub-cultures’. In T. Jefferson and
S. Hall (eds.), Resistance through Rituals: Youth Sub-Cultures in Post-war Britain.
London: Hutchinson.
Mfenyane, B. 1977. ‘Iskhumsha nesiTsotsi: The Sociolinguistics of School and Town
Sintu’. MA thesis, Boston Graduate School.
1981 ‘Scamto – Isjita: the black language arts of SasAfrika’. In M. Motloatse (ed.),
Reconstruction. Johannesburg: Ravan Press., pp. 294–302.
Language and language practices in Soweto 415
T. G. Reagan
That language planning should serve so many covert goals is not surprising.
Language is the fundamental institution of society, not only because it is the
first human institution experienced by the individual, but also because all other
institutions are built upon its regulatory patterns . . . To plan language is to plan
society. (Cooper 1989: 182)
1 INTRODUCTION
In 1971, Rubin and Jernudd edited a book entitled Can Language be Planned?
That was, and remains, an important question, and one that linguists and policy
makers are increasingly confident in answering in the affirmative. As Robert
Cooper noted in the quotation above, ‘to plan language is to plan society’ – and
the planning of society is, if anything, an increasingly common phenomenon in
both the ‘developed’ and ‘developing’ worlds. In fact, the significant question
is not whether language can be planned, but rather how and by whom. In this
chapter, the nature of language planning as an applied sociolinguistic activity
will be explored, with a particular focus on the challenge of linguistic diversity in
South Africa, as well as on the ways in which that diversity has been addressed in
the past and is likely to be addressed in the years ahead. Further, policy issues
in language planning in the South African context, the challenge of ethnicity
for language planning, and finally the future of language policy and language
planning in South Africa will be discussed.
419
420 T. G. Reagan
(2) The justness test. Is the policy just and fair? That is, does it treat all people
in an equitable and appropriate manner?
(3) The effectiveness test. Is the policy effective? Does it achieve its objectives?
(4) The tolerability test. Is the policy resource-sensitive? Is it viable in the
context in which it is to be effected?
These four ‘tests’ are quite useful in evaluating language policies, and may
serve as a working model for analysing different language-planning processes,
providing us with a series of questions that can be used in evaluating different
language-policy options. At this point, though, it is useful to look briefly at the
current state of linguistic diversity in South Africa, and to consider the changes
in that diversity that are likely to occur in the short- and intermediate-term
future.
As a general rule, it is safe to say that the more developed a nation is (primarily,
though certainly not exclusively, in economic terms), the greater the degree of
linguistic uniformity that will characterise it. If one keeps this correlation in
mind (and the relationship is correlative rather than causal), South Africa falls
just where one might expect – somewhere between the developed nations and
the countries of the so-called Third World. The language situation in South
Africa is characterised not only by the number and variety of African, Asian
and European languages that coexist, but also by alternative varieties of these
languages. Specifically, there are the koine languages of the townships (see
Schuring 1985), the Afrikaans of the coloured population (see van den Heever
1987, 1988) and a number of distinct native and non-native varieties of South
African English (see, for instance, Mesthrie 1992). There are also three lan-
guages – Arabic, Sanskrit and Hebrew – used almost exclusively for religious
purposes. Finally, there are the various natural sign languages used by the dif-
ferent deaf communities in South Africa (see Penn 1993; Penn and Reagan
1990, 1994, 1995; Reagan and Penn 1997).
However, despite the high degree of linguistic diversity in the country, South
Africa also shares a number of linguistic characteristics with the world’s
‘developed’ nations. The country’s linguistic diversity includes a language of
wider communication, English, which is widely spoken throughout the country,
and by members of virtually all of the different ethnolinguistic groups. There
is a high level and degree of bilingualism and even multilingualism, reflect-
ing the educational level of the population as well as the extensive inter-group
contact that continues, in spite of the legacy of apartheid, to characterise South
African society (see Kaschula and Anthonissen 1995). And, although still far too
low to be acceptable, and certainly skewed disproportionately towards certain
groups at the expense of others, the literacy rate in South Africa is impressive
422 T. G. Reagan
by ‘Third World’ standards, if not by Western ones (see, e.g., French 1982;
National Education Co-ordinating Committee 1993: 69–70). In short, the no-
tion of South Africa as a ‘Fourth World’ society (i.e. one in which elements of
both the ‘First’ and ‘Third’ worlds coexist) clearly makes a great deal of sense
from the perspective of the country’s linguistic situation.
And what of the future? At least in the short term, the language situation is
likely to remain basically unchanged. Those changes that do occur will fall into
four well-documented linguistic processes: language spread, language change,
language emergence and language death. Further, it is important to emphasise
that regardless of the nature of recent political change in South Africa, it is
virtually assured that linguistic diversity will remain a feature of social life
for generations to come, and that bilingualism and multilingualism will remain
commonplace for many, perhaps even most, South Africans well into the next
century.
for blacks (and, in fact, for almost all children in the country), but for arguably
quite different reasons from those used to defend mother-tongue instruction for
white children. It is clear that mother-tongue programmes for blacks were not
only consistent with the ‘ideology of apartheid’, but that they functioned as
one of the pillars of apartheid in perpetuating both racial and ethnolinguistic
divisions in South African society (see Reagan 1987b). Mother-tongue school-
ing for blacks was employed from the passage of the Bantu Education Act of
1953 to the end of the apartheid era to support the social and educational goals
of Verwoerdian-style apartheid. The apartheid regime used such programmes
to reinforce ethnic and tribal identity among black schoolchildren, seeking to
‘divide and conquer’ by encouraging ethnolinguistic divisions within the black
community (see Hartshorne 1992, 1987; Heugh 1985). As Barnard perceptively
noted,
Moedertaalonderwys . . . is not the Afrikaans term for mother-tongue instruction. It is
a political concept which has its roots in the dogma of Christian National Education.
According to this dogma, each ‘race’ or ‘volk’ has its own identity which sets it apart
from all others . . . Surely one has to wonder and become suspicious when there is
this insistence on the part of the authorities to force upon all children, against the
wishes of their parents, a particular language . . . What is being attempted is certainly
not mother-tongue education in the interests of the children but the enforcement of
‘moedertaalonderwys’ as an instrument of social control and subjugation, as a means to
an end. (Quoted in Heugh 1987b: 143–4)
Further, the process by which these policies were determined, developed and
implemented was fundamentally undemocratic. In a society as highly politicised
Language planning and language policy 425
as that of South Africa, such policies were doomed almost from the start. The
end result was that the polices – regardless of any objective merit – were either
accepted (as in the case of the first policy) or rejected (as in the other four cases)
largely on political and ideological grounds alone.
Notice that we have now, in essence, returned to our starting point: the im-
portant issue for language planning and language planners in South Africa
is ultimately how the planning is to be done, and by whom it is to be done.
Language planning is unlikely to be successful without the active support and
participation of the community towards which it is directed. This is why, in part,
the Afrikaans language movement (both in terms of status and corpus planning)
was so successful; it is also why efforts to plan and develop the African lan-
guages are likely to fail if (as in the past under National Party rule) they are met
by resistance from their own speakers.
As new language policies are developed in the South African context, such
policies will inevitably have to address a number of interrelated status- and
corpus-planning issues, among which are the role and place of English (status
planning), the role and place of Afrikaans (status planning), the role and place of
the various African languages (status planning), the need for lexical develop-
ment in specific languages (corpus planning) and the place of and limits on
‘mother-tongue’ programmes (status planning). Further, any policy adopted
should (at least ideally) be able to pass all four of the policy tests (desirability,
justness, effectiveness and tolerability) discussed above.
Finally, we come to current government language policy. The Government of
National Unity, as well as the new constitution, recognised eleven official lan-
guages, rejecting the historical bilingual policy which reflected only the linguis-
tic diversity of white South Africa with a multilingual policy more accurately
reflecting the reality of South African society. Further, the Reconstruction and
Development Programme of the ANC called for the development of ‘all South
African languages and particularly the historically neglected indigenous lan-
guages’ (African National Congress 1994: 71). This commitment to multilin-
gualism is commendable on a number of grounds, and meets all of Kerr’s policy
tests, with the possible exception of the tolerability test in so far as maintaining
all public and private sector services in all eleven official languages would be
almost certain to prove cost-prohibitive. However, this assumes that past models
of bilingualism are superimposed on current realities – that is, that the absolute
equality of English and Afrikaans sought by the apartheid regime (primarily
as a component of Afrikaner political ideology) is the same kind of equality
to be pursued by the democratic government of South Africa with respect to
all eleven official languages. This, of course, need not be the case, and in fact
is almost certainly not the case. The recognition of eleven official languages
in South Africa does not by any means necessarily imply that all public- and
private-sector services will inevitably be provided in all eleven languages; other,
more cost-sensitive, options and outcomes are possible.
In short, with the end of the apartheid era and the election of a democratic
government in South Africa, language policy in general, and in education
in particular, has received considerable attention as the institutions of South
African society are transformed. One powerful example of this concern with
language policy, especially in the educational sphere, is A Policy Framework
for Education and Training, which is a discussion document issued by the
Education Department of the African National Congress, which sets out pro-
posals related to issues of education and training (African National Congress
1995). Included in this document are four lessons that are identified as be-
ing of ‘the utmost importance’ in order that the ‘cycle of language oppression
be broken’ in South African society in general, and in education in particular
(African National Congress 1995: 62). These four lessons are:
Language planning and language policy 427
In order to ensure that these lessons are reflected in any language policy to be
developed in South Africa, the African National Congress discussion document
goes on to identify three general principles upon which educational language
policy should be based. These principles are:
(1) The right of the individual to choose which language or languages to study and to
use as a language of learning (medium of instruction).
(2) The right of the individual to develop the linguistic skills, in the language or lan-
guages of his or her choice, which are necessary for full participation in national,
provincial, and local life.
(3) The necessity to promote and develop South African languages that were previously
disadvantaged and neglected. (African National Congress 1995: 63)
It seems clear that both the lessons to be learned from past experience and the
general principles upon which educational language policies are to be based
are reflective, in large part, of concerns about past practices in South Africa,
and are intended to be consistent with the goal of a democratic and non-racial
language policy – as well as with the constitutional recognition of the equality
of the eleven official languages of South Africa.
An excellent example of the sort of approach to language policy formula-
tion envisioned by the ANC is the National Education Policy Investigation’s
work on language (NEPI 1992, 1993). The National Education Policy Investiga-
tion (NEPI) was a project undertaken by the National Education Co-ordinating
Committee between 1990 and 1992, the purpose of which was to explore policy
options in the educational sphere ‘within a value framework derived from the
ideals of the broad democratic movement’ (NEPI 1992: vi). The NEPI was
intended, in short, to achieve three principal functions:
(1) the provision of information and a lens to focus on the values which underpin specific
policies;
(2) the stimulation of public debate on educational policy in all spheres of society;
(3) the development of capacity for policy analysis. (NEPI 1992: vi)
In other words, what the NEPI sought to accomplish was to set the stage for
ongoing, and indeed protracted and extensive, debates about educational policy
issues. The language component of the investigation was typical in this regard,
428 T. G. Reagan
and its conclusions provide at most very broad, general guidelines for future
policy development. This can clearly be seen in the final concluding paragraph
of the language report, which argues that
any [language policy] option that is chosen can have an empowering or a disempowering
effect on learners, depending on its suitability for the particular school’s context, on how
it is implemented, and on how it relates to the national language policy of the country.
There is no one policy that is ideal for all schools. Language policy for education needs,
therefore, to be flexible without being so laissez faire as to allow the perpetuation of
present discriminatory policies or ill-informed choices of alternatives to them. (NEPI
1992: 93)
Current efforts now under way in South Africa, whether knowingly or un-
knowingly, are in fact moving in accord with this advice, and as a result the
educational language policies that are in the process of being developed are far
more likely to receive broad popular support than have past policies. Perhaps
the most outstanding example of this has been the reception of the final report of
the Language Plan Task Group (LANGTAG). This group was created in 1995
by Dr B. S. Ngubane, the Minister of Arts, Culture, Science and Technology,
with the explicit task of devising a national language plan for South Africa. The
final LANGTAG report, issued in August 1996, clearly attempted to achieve
the following objectives, which had been identified by Dr Ngubane:
(1) All South Africans should have access to all spheres of South African society by
developing and maintaining a level of spoken and written language which is appro-
priate for a range of contexts in the official language(s) of their choice.
(2) All South Africans should have access to the learning of languages other than their
mother tongue.
(3) The African languages, which have been disadvantaged by the linguicist policies of
the past, should be developed and maintained.
(4) Equitable and widespread language services should be established. (Language Plan
Task Group 1996: 7)
The central problem that South Africa faces in responding to the very real
fact of language diversity is the understandably close linkage in the minds of
Language planning and language policy 429
many South Africans between ethnicity and apartheid. ‘Ethnicity’ in the South
African context is not merely a descriptive term, it is rather a normative term,
and to defend ethnicity as a legitimate manifestation of human experience and
awareness has for many become synonymous with defending apartheid. As
Heugh (1987: 208) noted before the 1994 election: ‘Ethnicity is regarded by
the government’s extra-parliamentary critics as a euphemism for racism and a
policy not only inimical to black unity but also part of the government’s grand
apartheid scheme of divide and rule.’
This need not be the case, though, and it is important that appropriate mani-
festations of ethnicity be recognised, as the new government has already shown
signs of doing. As Adam and Moodley argued in South Africa without Apartheid
(1986: 220):
Liberalism has for the most part failed to recognise the legitimate aspects of mobilized
ethnicity, by associating ethnicity solely with unfair advantage or the height of irrational-
ity. But insofar as ethnicity expresses cultural distinctiveness and the quest for individual
identity through group membership, it may fulfill desires that liberalism ignores. People
do not necessarily want to be the same . . . Cultural ethnicity only becomes problematic
if is transformed into economic and political ethnicity for the advantage of its members
at the expense of outsiders.
The South African government has interpreted the concept of group rights over a long
period of time to the advantage of whites. Yet such distortions of the concept of group
rights in favour of group privilege should not invalidate the concept. One should rather
introduce the principle of justice, to help evaluate the applications of the concept of
group rights.
7 CONCLUSION
bibliography
Adam, H. and K. Moodley 1986. South Africa without Apartheid: Dismantling Racial
Domination. Berkeley: University of California Press.
African National Congress 1994. The Reconstruction and Development Programme:
A Policy Framework. Johannesburg: African National Congress.
1995. A Policy Framework for Education and Training (Discussion document).
Braamfontein: Education Department, African National Congress.
Alexander, N. 1989. Language Policy and National Unity in South Africa/Azania. Cape
Town: Buchu Books.
Language planning and language policy 431
Bullivant, B. 1981. The Pluralist Dilemma in Education: Six Case Studies. Sydney:
George Allen & Unwin.
Bunting, B. 1986. The Rise of the South African Reich. London: International Defence
and Aid Fund for Southern Africa.
Cluver, A. 1992. ‘Language planning models for a post-apartheid South Africa’.
Language Problems and Language Planning, 16: 105–36.
Comrie, B. 1981. The Languages of the Soviet Union. Cambridge: Cambridge University
Press.
Cooper, R. L. 1989. Language Planning and Social Change. Cambridge: Cambridge
University Press.
Coulombe, P. 1993. ‘Language rights, individual and communal’. Language Problems
and Language Planning, 17: 140–52.
Degenaar, J. 1987. ‘Nationalism, liberalism and pluralism’. In J. Butler, R. Elphick and
D. Welsh (eds.), Liberalism in South Africa: Its History and Prospect. Cape Town:
David Philip, pp. 236–49.
Dogancay-Aktuna, S. 1995. ‘An evaluation of the Turkish language reform after
60 years’. Language Problems and Language Planning, 19: 221–49.
Dunja-Blajberg, J. 1980. Sprach und Politik in Südafrika. Bonn: Informationsstelle
Süidliches Afrika.
Eastman, C. M. 1983. Language Planning: An Introduction. San Francisco: Chandler &
Sharp.
French, E. 1982. The Promotion of Literacy in South Africa. Pretoria: Human Sciences
Research Council.
Hartshorne, K. 1987. ‘Language policy in African education in South Africa, 1910–1985,
with particular reference to the issue of medium of instruction’. In D. Young (ed.),
Bridging the Gap between Theory and Practice in English Second Language
Teaching: Essays in Honour of L. W. Lanham. Cape Town: Maskew Miller
Longman, pp. 62–81.
1992. Crisis and Challenge: Black Education, 1910–1990. Cape Town: Oxford
University Press.
Heugh, K. 1985. ‘The relationship between nationalism and language in education in
the South African context’. In D. Young (ed.), UCT papers in language education.
Cape Town: University of Cape Town, Language Education Unit, Department of
Education, pp. 35–70.
1987a. ‘Trends in language medium policy for a post-apartheid South Africa’. In
D. Young (ed.), Language: Planning and Medium in Education. Rondebosch:
Language Education Unit (UCT) and SAALA, pp. 206–20.
1987b. ‘Underlying Ideologies of Language Medium Policies in Multilingual
Societies’. M. Phil. thesis, University of Cape Town.
Heyd, U. 1954. Language Reform in Modern Turkey. Jerusalem: Israel Oriental Society.
Hirson, B. 1981. ‘Language in control and resistance in South Africa’. African Affairs,
80: 219–37.
Kaschula, R. and C. Anthonissen 1995. Communicating across Cultures in South Africa:
Toward a Critical Language Awareness. Johannesburg: Witwatersrand University
Press.
Kerr, D. H. 1976. Educational Policy: Analysis, Structure, and Justification. New York:
David McKay.
432 T. G. Reagan
Sarah Murray
1 INTRODUCTION
South Africa is undergoing great change, not least in the areas of educa-
tional and language policy. We are moving away from policies that emphasised
strong boundaries between languages and people, towards those that encourage
people to learn and use many languages to communicate with each other. South
Africans – a descriptor which only now includes everyone in South Africa –
are urged to discard their old singular identities, rooted in an intimate bonding
of race, language and culture, in order to embrace a more complex sense of
self. This new South Africanism is dynamic: it is rooted in a previously unac-
knowledged commonality, forged through historic and economic processes, but
it also acknowledges and gives expression to different languages and cultures
(Alexander 1996).
Alexander – one of the main proponents of this view – argues that multilin-
gualism, which challenges the inseparability of language, culture and identity,
will play a large part in achieving this new identity. He sees education as an
important means through which South Africa’s multilingualism can be both
validated and developed (1996: 11).
In this dynamic view of language and culture, multilingualism is seen as a
resource to be drawn upon, in much the same way that Thornton (1988) views
culture. This more fluid notion of languages, which does not insist on their
separation or containment, is ever present in the educational language issues I
have chosen to discuss in this overview.
They are: the choice of languages to be used in school and for what purposes;
the issue of code-switching in the classroom; the question of literacy – a point
at which variety and standard often come into conflict; and the preparation of
teachers, both in terms of their own language proficiencies and their capacity
to teach linguistically diverse classes equitably.
To set the scene for this discussion, it is necessary first to look back briefly
in order to understand some of the perceptions and realities that remain with us
in the current education system.
434
Language issues in South African education 435
After the elections, the old separate departments of education fell away and all
government schools became the responsibility of provincial departments. The
separate legislative acts that had governed schooling were replaced, in 1996,
by the South African Schools Act, which provided a ‘uniform system for the
organisation, governance and funding of schools’ (South Africa 1996: 5).
The interim constitution – accepted in its final form by the Constitutional
Assembly in 1996 – declared eleven languages official: Pedi, Sotho, Tswana,
Swati, Venda, Tsonga, Afrikaans, English, Ndebele, Xhosa and Zulu. It has
given every person the right to basic education in the language of his or her
choice where this is reasonably practicable. The various language boards, re-
sponsible in the past for the development and regulation of individual languages,
have been replaced by the Pan South African Language Board (PANSALB).
This move was called for as early as 1992 by Neville Alexander (1992: 160–1),
who has been a driving force behind the board, and – until recently – its deputy
chairperson.
The specific form that educational language policy should take had been
vigorously debated since the early 1990s. Kathy Luckett (1992, 1993), a re-
searcher for the National Education Policy Investigation (NEPI), had put the
notion of ‘national additive bilingualism’ firmly on the agenda, although her
ideas had not been unequivocally accepted by NEPI itself (Young 1993a). In
essence her position was that any policy involving a transition from home- to
second-language medium of instruction was ‘subtractive’; she argued for some
form of dual-medium policy. The ANC’s own Language Policy in Education
Working Group had supported Luckett’s position, arguing strongly for an
‘additive bilingual/multilingual approach’, and also for the recognition of multi-
lingualism as a ‘national resource’ and a move away from terms such as ‘medium
of instruction’ to ‘language(s) of learning’ (Centre for Education Policy Devel-
opment (CEPD) 1993). This position had been promoted and popularised by the
National Language Project (NLP) in its publication, Language Projects Review
(later renamed Bua!), and by the Project for the Study of Alternative Education
in South Africa – PRAESA – of which Alexander is director (see Heugh
et al. 1995).
Language issues in South African education 437
4.2 Code-switching
Code-switching (CS) is a typical feature of multilingual societies such as South
Africa (see McCormick and Slabbert and Finlayson, chaps. 11 and 12, this
volume). However, previous educational language policies, which were
premised on strong boundaries between languages, viewed the practice un-
favourably. Nevertheless, it was widespread; in the NEPI survey a third of all
teachers interviewed said they used more than one language in the classroom
(Bot 1993). This was the case in primary schools where an African language
was the medium of instruction and in high schools where English was the
medium. Nomvete (1994: 13) quotes an experienced Grade 1 teacher from a
farm school in Bronkhorstspruit talking about her use of two closely related
Nguni languages under the old political dispensation: ‘Most of my learners are
Ndebele and the inspector insists that I teach them in isiZulu because that is the
official DET policy for that area, but I often use isiNdebele and only insist on
isiZulu for written work and tests and when I know the inspector is coming.’ In
the same study, a high school English teacher describes the difficulty of teaching
in English and the problematic consequences of CS when written assessment
is in that language:
My learners are mainly Zulu, Ndebele and North Sotho speaking. At school the policy
is that they learn all the subjects in English with the exception of isiZulu and Northern
Sotho as subjects. These children speak the three languages most of the time and English
only in class. These children are not exposed to English most of their time and yet they
are expected to learn in English in order to be successful in their lives. There are those
440 S. Murray
teachers who sometimes use isiZulu . . . even if a teacher can teach something in isiZulu
and those children understand him very well, he is going to test them in English since it
is not allowed to test in isiZulu. So those poor children are now going to find it difficult
to answer those questions even if they have understood their work. (Nomvete 1994: 13)
over the distinction between first and second language, particularly in relation
to English (Mawasha 1984). As Janks (1990: 243) put it, ‘subtle undertones of
“second-language, second-class” exist in the society’. This was not an issue in
relation to African languages, which were in any case not taught as second but
as third or foreign languages.
There was much debate, too, about the use of different varieties of languages
in the classroom, with calls for the Africanisation of English (Ndebele 1987). In
the case of African languages the debate centred around the issue of standardi-
sation and the work of the language boards. The standard languages prescribed
in the school syllabuses bore little relation to the varieties of those languages
spoken in students’ homes, particularly for those living in metropolitan areas
where varieties are characterised by high levels of CS (Botha 1994). Ntshangase
(chap. 21, this volume) describes a teacher who finds it necessary to switch to
Iscamtho to explain difficult things in a Zulu lesson. In the classroom, this was
exacerbated by the dry and formal way in which African languages were taught
(Nokaneng 1986), resulting in the claim by many African students that they
find it easier to study English as a subject than their home languages.
In recent debates the whole notion of a single home language or mother
tongue has been challenged, and it has been suggested that for many urban
children English is very much part of the home repertoire (Winkler 1997). This
raises questions regarding the choice of languages as subjects (and as languages
of learning) and the levels at which they should be taught and learned. In
considering the issue it is important to bear in mind that both parents and students
seem to be against the use of non-standard varieties of African languages for
educational purposes (Pather 1994; Ntshangase, chap. 21, this volume).
The new curriculum takes account of these concerns. There are no longer
separate syllabuses or, in fact, syllabuses at all; there is a ‘learning area’ called
‘Language, literacy and communication’ in which are described the common
‘outcomes’ for all languages. While learners are to be given access to standard
forms of a language where appropriate, acceptance of different varieties is en-
couraged. A distinction is made between ‘main’ and ‘additional’ languages, but
all learners will be expected to achieve the same outcomes, albeit at different
levels. Some questions have been raised about the appropriacy of a core cur-
riculum, which has been conceptualised in English and now applies to other
languages. Maclean (1997), for example, believes that common outcomes need
to be written at a sufficient level of generality to be inclusive of discursive dif-
ferences between languages; he believes that some in the new curriculum are
generic in this way, but others are not.
It has yet to be seen what all this will mean in practice. At the time of writing,
we are five months into the Grade 1 implementation. In the Eastern Cape,
with which I am familiar, things have got off to a rather slow start, especially
in rural schools, and there is a shortage of learning materials. The learning
442 S. Murray
programme for Grade 1 is holistic and integrated, and the language used for
different aspects of the curriculum seems to be determined by the language of
the learning materials delivered to the school. These materials, in any event,
have been conceptualised in English and translated into other languages.
In previously white schools, many of which now have large enrolments of
African children, English or Afrikaans are still the languages of learning and
teaching. However, in some schools Xhosa has been introduced as a subject,
and in multilingual classes – alongside their English- and Afrikaans-speaking
classmates – Xhosa-speaking children are being taught to say Molo, titshalakazi
(‘Hello, teacher’)! This, incidentally, has become a phenomenon higher up in
such schools, where Xhosa-speaking children study Xhosa as a third language.
There are exceptional schools. One such is Collegiate Junior School in Port
Elizabeth where all the children learn three languages – English, Xhosa and
Afrikaans. The school has employed Xhosa-speaking teachers and developed a
communicative methodology for teaching the language with the assistance of an
NGO called L-MAP (Language Methods and Programmes).1 This is no mean
feat in a time of financial stringency and teacher retrenchments. However, it is
worth noting that in a parent survey, less than half of the Xhosa-speaking parents
regarded first-language support for their children as important (Plüddemann
1995).
4.4 Literacy
Levels of literacy are a matter of concern in all schools where enliteration is
complicated by the language issue (Winkler 1997: 35), but especially in African
schools where this is compounded by a lack of resources (Ntete 1998).
The Molteno Project has researched and developed a programme – Break-
through – for enliterating children in African languages, over a period of more
than twenty years. Breakthrough is available in all eleven official languages;
it is a national programme in Botswana and has been developed in a number
of the languages of Namibia. Molteno has also developed Bridge to English,
a programme which systematically builds on the foundation of literacy in an
African language and transfers this to English. Evaluations of the project’s work
have been overwhelming positive, though it could be criticised on the grounds
that it is ‘subtractive’ – something which it is now remedying; it has also been
criticised for its instrumentality. (See also Walters 1996.)
The lack of reading materials in African schools is a serious problem; cur-
rently 80 per cent of schools do not have libraries (Garson 1998). An NGO
called READ has done much to remedy this by creating class libraries in the
form of book boxes, developing local reading materials and educating teachers
in their use. READ has been criticised for its overemphasis on developing a
reading culture in all contexts, both rural and urban, through books in English
Language issues in South African education 443
(Garwen 1995). Alexander (1996: 11) has argued that ‘the lack of a reading
culture in the African languages in South Africa is the single most important
sociocultural factor that explains the continued low status of these languages’.
The Teacher, a monthly newspaper, has consistently advocated the use of
public libraries. However, the current crisis in funding places their very exis-
tence in jeopardy, and alongside the lack of money for school textbooks and
the consequent downsizing of educational publishing, this does not bode well
for the availability of books in future. This is particularly sad when one was
beginning to see reading material in African languages coming onto the market.
The issue of how children learn to read and write should be at the top of our
research agenda, but recently it seems to have been eclipsed by concerns about
multilingualism. And although literacy is foregrounded in the new curriculum,
reading and writing is subsumed under ‘literacies’ – language literacy, cultural
literacy, critical literacy, visual literacy, media literacy and computer literacy –
a move, incidentally, that carries the potential to create new boundaries (see
Street 1997). While it is important that a more sophisticated understanding of
the nature of literacy and literacy practices should inform our curriculum, it is
equally important that we should not lose sight of the important task of schools
to teach children to read and write. Researchers should, in particular, be con-
sidering the relationship between spoken and written varieties of languages in
multilingual communities and the implications of this for enliterating children.
There has been a wealth of research in South Africa in higher education,
as this is the point at which students’ poor levels of literacy become apparent.
This research focuses almost exclusively on academic literacy in English and
is widely reported in conference proceedings of the South African Association
for Academic Development (SAAAD) and the Southern African Applied
Linguistics Association (SAALA), and in journals such as the South African
Journal of Higher Education. Recently, too, there has been research into literacy
practices in the field of adult education (Prinsloo and Breier 1996).
5 CONCLUSION
The idealistic goals of South Africa’s multilingual language policy in education
are hard to take issue with, but difficult to achieve in practice. Outside the
classroom, people use their linguistic resources in flexible ways to achieve their
communicative purposes. Inside the classroom, however, the teacher is expected
to develop students’ linguistic abilities in particular languages in demonstrable
ways. To achieve this teachers must make endless decisions: how to ensure, for
example, in a multilingual class, that home-language speakers of, say, Xhosa are
sufficiently stretched in their Xhosa lessons. What to do about an essay written
in non-standard Zulu, and so on. In such situations, ‘standards’ in at least
two senses of the word become an issue. And for the teacher multilingualism
becomes not just a resource, but also a problem. (See Barkhuizen in Young et al
1995: 91.)
Language issues in South African education 445
notes
1 This organisation used to be known as ELMAP (English Language Methods and
Programmes).
bibliography
Adendorff, R. 1992. ‘Code-switching among Zulu-speaking teachers and their pupils:
its functions and implications for teacher education’. In J. Boughey (ed.), Visions
and Realities: Proceedings of the 11th Annual SAALA Conference. Cape Town:
University of Cape Town, pp. 23–42.
Alexander, N. 1992. ‘A language policy for a future South Africa’. Proceedings of the
English Academy of Southern Africa Conference. 1–3 July. Cape Town: University
of Cape Town, pp. 154–62.
1996. ‘Mainstreaming by confluence: the multilingual context of literature in South
Africa’. World Literature Today, 70, 1: 9–11.
Bengu, S. M. E. 1997. Statement by the Minister of Education, 14 July (on the Internet:
http://www.polty.org.za/govdocs/misc/langpol/html).
Bosch, B. and V. de Klerk 1996. ‘Language attitudes and their implications for the
teaching of English in the Eastern Cape’. In V. de Klerk (ed.), Focus on South
Africa. Amsterdam: John Benjamins, pp. 231–250.
Bot, M. 1993. ‘Language use in schools: a survey of teachers’ opinions’. Edusource
Data News, 3, August: 1–9.
Botha, E. M. 1994. ‘Manifestation of language varieties in the classroom in the PWV’.
ELTIC Reporter, 18, 1/ 2: 27–31.
Centre for Education Policy Development (CEPD) 1993. ‘Language Policy in Education:
Report of the Language Policy in Education Working Group’. June.
Cummins, J. and M. Swain 1986. Bilingualism in Education: Aspects of Theory Research
and Practice. London: Longman.
de Kadt, E. 1993. ‘The dangerous power of English’. In Proceedings of the English
Academy of Southern Africa Conference. Cape Town, July: 6–16.
446 S. Murray
Department of Education 1997. Curriculum 2005: Lifelong Learning for the 21st
Century. Pretoria: Department of Education.
Donaldson, A. 1992. ‘Financing education’. In R. and A. McGregor (eds.), McGregor’s
Education Alternatives. Cape Town: Juta, pp. 231–50.
Dowling, T. and P. Maseko 1995. African language teaching at universities. In Heugh
et al. (eds.), pp. 100–6.
English Language Teaching Information Centre (ELTIC) 1997. Multilingual Learning:
Working in Multilingual Classrooms. Cape Town: Maskew Miller Longman.
Freer, D. 1992 (ed.). Towards Open Schools: Possibilities and Realities for Non-Racial
Education in South Africa. Johannesburg: Witwatersrand University Press.
Garson, P. 1998. ‘Reading’s vital statistics’. The Teacher, 3, 1: 5.
Garwen, M. 1995. ‘Interim Evaluation Report’. Johannesburg: READ/DBSA.
Gough, D. 1993. ‘Code-switching in the classroom – thetha or khetha/choose or
loose/kies of vries’. Unpublished article.
Hartshorne, K. 1992. Crisis and Challenge: Black Education 1910–1990. Cape Town:
Oxford University Press.
Herbert, R. K. 1992. ‘Language in a divided society’. In R. K. Herbert (ed.), Language
and Society in Africa: The Theory and Practice of Sociolinguistics. Johannesburg:
Witwatersrand University Press, pp. 1–19.
Heugh, K., A. Siegruhn and P. Plüddemann 1995 (eds.). Multilingual Education for
South Africa. Johannesburg: Heinemann.
Janks, H. 1990. ‘Contested terrain: English education in South Africa 1948–1987’. In
I. Goodson and P. Medway (eds.), Bringing English to Order. London: Falmer
Press, pp. 242–61.
Johanson, E. 1996. ‘“TALK”-ing at Princess High’. ELTIC Reporter, 20, 1: 77–81.
Kgomoeswana, V. N. 1993. ‘The case against codeswitching in multilingual class-
rooms’. ELTIC Reporter, 17, 2: 13–16.
Langhan, D. 1996. ‘The roles of languages in learning across the curriculum’. ELTIC
Reporter, 20, 2: 4–23.
Luckett, K. 1992. ‘National Additive Bilingualism. Report from the Medium of Instruc-
tion Sub-group to the Language Policy Research Group’. NEPI Working Paper.
1993. ‘“National additive bilingualism”: towards the formulation of a language plan
for South African schools’. Southern African Journal of Applied Language Studies,
2, 1: 38–60.
Macdonald, C. 1990. Crossing the Threshold into Standard Three. Main Report of the
Threshold Project. Report SOLING-16. Pretoria: HSRC.
Maclean, D. 1997. ‘Conditions under which outcomes can be common’. SAALA
Communiqué, June: 4–5.
Makoni, S. B. 1994. ‘Mother-tongue education: a literature review’. ELTIC Reporter,
18, 1/2: 18–26.
Mawasha, A. 1984. ‘Your English, our English: English in black South Africa’. English
Usage in Southern Africa, 15, 2: 12–18.
Mhlanga, S. I. 1995. ‘Parental Preferences regarding Medium of Instruction in Primary
Schools in the Nongoma District of KwaZulu-Natal’. M.Ed. half-thesis, Rhodes
University.
Murray, S. 1998. ‘Talk Schools Programmes: Interim Evaluation’. Unpublished draft
discussion document.
Language issues in South African education 447
Murray, S. and H. van der Mescht 1996. Preparing students to teach English first and
second language. In V. de Klerk (ed.), Focus on South Africa. Amsterdam: John
Benjamins, pp. 251–68.
Murray, S. and M. Hendricks 1997. ‘Multilingual Education: From Policy to Practice’.
Paper presented at the Kenton Education Conference, Hermanus, 31 October–2
November.
National Commission on Higher Education (NCHE) 1996. Report: A Framework for
Transformation. Cape Town: NCHE.
Ndebele, N. 1987. ‘The English language and social change in South Africa’. English
Academy Review, 4: 1–16.
Nokaneng, M. 1986. ‘The Methodology of African Languages’. Paper presented
at CSIR Conference on the Role of Language in Black Education, Pretoria,
February.
Nomvete, S. 1994. ‘From oppression to opportunity: multilingual policies for schools’.
ELTIC Reporter, 18, 1/2: 11–17.
Ntete, S. 1998. ‘Case Studies of Second Language Learners who Excel in English’.
M.Ed. thesis, Rhodes University.
Pather, E. 1994. ‘Language attitudes and multilingualism: classroom-based action
research’. ELTIC Reporter, 18 1/2: 32–8.
Peires, M.-L. 1994. ‘Code-switching as an aid to L2 learning’. Southern African Journal
of Applied Language Studies, 3, 1: 14–22.
Plüddemann, P. 1995. ‘Girls go for Xhosa’. Bua! 10, 1: 15.
Prinsloo, M. and M. Breier 1996. The Social Uses of Literacy: Theory and Practice in
Contemporary South Africa. Amsterdam: John Benjamins.
Roberts, J. E. 1997. ‘A Study of the Development of Institutional Language Policies in
Two Gauteng Primary Schools’. M.Ed. thesis, University of the Witwatersrand.
Shoai, C. 1997. ‘New policy on language’. The Teacher, August 1997: 3.
Smit, U. 1994. ‘Investigating language attitudes as a basis for formulating language
policies – a case study’. Southern African Journal of Applied Language Studies, 3,
1: 23–35.
South Africa (Republic) 1996. South African Schools Act No. 84. Pretoria: Government
Printer.
Street, B. 1997. ‘The implications of the “New Literacy Studies” for literacy education’.
English in Education, 31, 3: 45–59.
Technical Committee for the Revision of Norms and Standards in Teacher Education
1997. Norms and Standards for Teacher Education, Training and Development:
Discussion Document. Pretoria: Department of Education.
Thornton, R. 1988. ‘Culture: a contemporary definition’. In E. Boonzaier and J. Sharp
(eds.), South African Key Words: The Uses and Abuses of Political Concepts. Cape
Town: David Philip, pp. 17–28.
Walters, P. 1996. ‘Issues in English language teaching in primary schools’. In V. de
Klerk (ed.), Focus on South Africa. Amsterdam: John Benjamins, pp. 211–30.
Winkler, G. 1997. ‘The myth of the mother tongue’. Southern African Journal of Applied
Language Studies, 5, 1: 29–39.
Young, D. 1993a. ‘The NEPI language report: product or process?’ Proceedings of the
12th Annual SAALA Conference, Part 2. University of Port Elizabeth, 28–30 June,
pp. 267–78.
448 S. Murray
Kathleen Heugh
1 BACKGROUND
Language-policy developments in South Africa have undergone dramatic
changes over the last decade. Explicit statements of policy have shifted away
from the segregationist mould of the previous apartheid government with the
widely divergent roles and functions it ascribed to the various languages of
the country. There is now a move towards principles that espouse the equal
status and functions of eleven of the country’s languages in addition to the
promotion of respect for, and use of, other languages. The extraordinary cir-
cumstances surrounding the political negotiations that led to a sharing of power
after the country’s first democratic elections of 1994 created the opportunity
for ‘proposals from below’ (from civil society), to take root in a manner which
has never before been possible in South Africa. Many of the proposals for new
language policy have been accepted on an official level and an encouraging,
optimistic environment seemed, in the early years of the new government of
national unity, to promise a vibrant future for language development and mul-
tilingualism. In the era of globalisation, however, there are larger structural
forces at play, which influence international and domestic economic and devel-
opment policies. These forces are generally antithetical to multilingualism. It
should therefore not be surprising that tensions in language-policy development
are beginning to manifest themselves. In part, these tensions are discernible in
other multilingual societies, particularly in Africa; in part they are peculiar to
South Africa. In this chapter, the emergence of a language policy based on the
multilingual reality of the country will be discussed against a background of
developments in other African countries as well as the impact of globalisation
and the Western political economy.
A conservative estimate of about five thousand languages used in about two
hundred countries indicates that multilingualism is a global reality. However, as
David Crystal (1987: 360) points out: ‘The widespread impression that multilin-
gualism is uncommon is promoted by government policies: less than a quarter
of the world’s nations give official recognition to two languages . . . and only
six recognise three or more.’ South Africa is thus in the pioneering position of
449
450 K. Heugh
now recognising eleven official languages, more than any other country; and its
new constitution (1996) together with the Pan South African Language Board
Act (1995) and a later amendment to this Act impel, in principle, the promotion
of respect for other languages as well as the promotion of multilingualism and
the development of languages. These impulses place the country at the cutting
edge of international language-policy development.
Ingrid Gogolin (1993) refers to the ‘monolingual habitus’1 in which the
general, Western perception about language resides. The South African govern-
ment, however, nurtured a public and bilingual ‘habitus’ during the half-century
of Afrikaner nationalist rule, and it has not, since Milner’s Anglicisation pol-
icy at the turn of the last century, projected a monolingual view of the world
through its language policy. Language policy under apartheid was driven by
a two-pronged logic: to counteract the hegemony of English foisted upon the
country under Milner and to pursue the principle of separate development.
This involved extensive modernisation of Afrikaans, in addition to limited and
separate language development in each of nine African languages/varieties of
languages. In this way, the social, educational and political segregation of the
users of the different languages was encouraged. Social stratification was fur-
thered through the application of this policy since the development of Afrikaans
was to ensure unrestricted functional use of the language, whereas the develop-
ment of African languages was only intended to occur for restricted purposes.
They were never intended for use in the upper levels of education, the economy
or political activity.
Despite conscious language-planning activities designed to limit the hege-
mony of English, the functional status of English increased, particularly in
the economically powerful, English-speaking private sector. The monolingual
habitus of English speakers seemed to be in conflict with the apartheid language
policy’s promotion of Afrikaans, giving the illusion that there was an overlap
of interests between the major opponents of apartheid and the proponents of
English in South Africa. This permitted, for a time, a faulty public percep-
tion that it was primarily the Afrikaans-speaking white community that was
responsible for the segregationist political and economic system in the country.
It hid the structural and economic support of the powerful corporations largely
owned by the English-speaking community. The emphasis given Afrikaans
in the secondary schools for African-language-speaking students until 1977
gave rise to a justifiable belief that the government was attempting to limit
access to English. Just as Milner’s Anglicisation policy engendered resistance
from those to whom it was primarily directed, so too did the National Party’s
attempt to foist Afrikaans upon African-language speakers spark resistance.
This culminated in the 1976 revolt of students led by the Black Consciousness
Movement in Soweto, which has been well documented elsewhere (for example,
Hartshorne 1992: 195–205). Thus, the majority of African-language speakers
Recovering multilingualism 451
became neither willing champions of their own languages nor willing users of
Afrikaans. Instead they became committed to English for high-status commu-
nicative functions. In so doing they have, in effect, collaborated with the larger
political and economic interests of the West in the mistaken belief that access
to English will deliver power, both economic and political, to the majority.2
Developments during the 1990s have catapulted official statements on lan-
guage policy out of the domain of an official Afrikaans–English bilingualism,
with privileged status for two languages, to that of functional multilingualism.
Both the interim and final constitutions of 1993 and 1996 task government and
civil society with addressing a multilingual reality. This, however, needs to be
understood against the gathering momentum of an overwhelming drive toward
the monolingual habitus, and the dynamics of linguicism (linguistic racism).3
Language policies tend to reflect the interests of the ruling elite. But the
argument being made here is that the application of language policy is very
often a reflection of a more complicated set of relationships between overt
political ideology and the more covert aspects of the political economy. To
compound matters, it is not just the political economy of a particular country
that would affect that country’s language policy. The hegemony of the Western
free-market economy is such that it influences the economies of developing
countries. A Western economy is also very often accompanied by linguicism
which places high status on English, for example, and low status on other
languages. Western aid packages to the developing world have impacted, and
continue to impact upon, the implementation of language policy. It is important
to look at international trends in language policy, and their relation to political
ideology and free-enterprise economy, in order to assess the implications of
implementing new language-policy options for South Africa. Implicit in the
arguments of this chapter is the recognition that language policies are usually
arrived at by a top-down process which rarely accommodates the perspective
or needs of people from below.
2 LANGUAGE-POLICY PARADIGMS
For the purpose of identifying where the points of tension are to be found
in relation to language-policy developments in South Africa, the following is
presented as a frame of reference. Ruiz (1984) has articulated a way of viewing
language from three different theoretical positions: language as a problem;
language as a right; and language as a resource. A number of sociolinguists have
explored and found useful this avenue for understanding the origin of language
policies as well as their implementation strategies. (See, for example, Lo Bianco
1990, with regard to Australia and Akinnaso 1991, with regard to Nigeria.) An
analysis here of language policies and the language-planning mechanisms that
ensue, when viewed against the Ruiz typology, reveals the following:
452 K. Heugh
education effectively advantages the children of the ruling class and disadvan-
tages or marginalises the remainder of society.
The World Bank and the IMF have become the principal organisations through which the
capitalist West seeks to control the destiny of the rest of the world. In this respect,
the establishment and reconstitution of structural inequalities . . . and cultural inequali-
ties . . . between the imperial European languages and other languages becomes indis-
pensable strategies towards that attempted control. (Mazrui 1997: 43)
In the Zambian situation, English had been selected in the belief that it could
function as a neutral language to further the interests of national unity. The strat-
egy for implementing this policy was to select English as the primary language
of education, the results of which are reflected above. Thus the implementation
strategy failed to achieve what was intended by policy.
In other situations, even when African languages are used in the education
system, tensions arise between the language in education policy and national
policy. Akinnaso (1991) argues that the national (economic) plan subverts the
language-in-education plan in countries such as Nigeria and Tanzania where
the promotion of the use of African languages in education is neutralised by
an English-proficiency criterion to positions of national political and economic
power. Tripathi, Akinnaso and Siachitema all argue, therefore, that national
language-policy formulation and implementation need to be knitted into the
overall plan for national development, of which education is one domain. The
structural tensions discussed earlier contribute to the lack of synchrony between
language policy and effective implementation; or between a national develop-
ment plan and a language plan.
At another level, a number of sociolinguists extend this argument by exam-
ining the relationship between the failure of development programmes and the
failure of education on the continent to embrace the reality of multilingualism.
Paulin Djité, for example, argues that African development agencies need to
recognise the role of the larger indigenous languages ‘in a process of global and
integrated development’: ‘Reliance and dependency on superimposed interna-
tional languages to achieve development in Africa over the last three decades
have proven to be a failure. Instead of leading to national unity, this attitude
has significantly contributed to the socioeconomic and political instability of
most African countries’ (1993: 149). Djité’s argument is based on an analysis
of the use of lingua francas in West Africa particularly, which ease channels of
communication among people across national boundaries on a significant and
useful level. The use of the standard varieties of English, French and Portuguese
on the continent, however, are limited to the upper levels of government and
administration.
The masses have managed and developed these networks of communication over the
years . . . The linguae francae, because of these communicative and socioeconomic real-
ities, are progressively being perceived as neutral languages and are increasingly being
learned as second languages. They are, in the true sense, the de facto national and inter-
national mediums of communication, for they satisfy the criteria of efficiency, adequacy,
and acceptability (Haugen 1966: 61–3). (Djité 1993: 159)
In other words, despite what agencies such as the World Bank and other Western
interests would have Africans believe, the continent does not suffer from the
linguistic confusion of Babel. Furthermore, the role of the superimposed
456 K. Heugh
Towards the end of the period of multiparty negotiations that resulted in the
interim constitution of 1993, the various possibilities for a new language policy
for South Africa had become clearly distilled. The strong association between
language and cultural identity for white Afrikaans-speaking people sown during
the terminal stages of British colonial control of South Africa grew, during
National Party rule, into a defining characteristic of Afrikaner nationalism. The
Afrikaans language assumed such importance in the identity of this community
that it became one of a few major issues upon which the success or failure of the
constitutional negotiations hinged towards the end of 1993. At this point, the
multiparty negotiations were really being conducted between the two strongest
political forces, the National Party (NP) and the African National Congress
(ANC). The bottom line for the NP negotiators was that Afrikaans should not
lose its privileged status in the new dispensation; in other words, it had to retain
its official status. In contrast, the ANC did not attach a similar importance
to language issues. Much greater significance was placed on neutralising or
removing a wide range of symbols of apartheid, of which language policy
was only one. The official policy of the ANC was that all languages would be
regarded as equal, but that none should be accorded official status. The unofficial
conviction, however, was that English, for apparently pragmatic reasons, would
function as the official language of government. This view of English had
its origins in the early history of the ANC, when English had been regarded
as a language of liberation and a language through which opposition to the
Afrikaans-speaking government would be mediated. Additionally, many senior
members of the ANC had been exiled in English-speaking countries for many
years prior to 1990 and had therefore come to believe in the importance of
English as the lingua franca in the country.
The ANC supports the deliberate fostering of multilingualism in schools, adult education
programmes, in the workplace and in all sectors of public life . . . Though language ex-
perts argue that initial education is best conducted through the ‘mother tongue’ . . . large
sections of black urban communities have already pressurised primary schools into
beginning with English as the medium of instruction from day one . . .
Any language policy must reflect the voice of the people and this voice is more
important that any model which emerges.
The ambiguities present in this document reflect precisely those current in the
broader context of South African society with regard to the weight given the
role of English vis-à-vis African languages. A significant contribution, from
within the ANC, but independently of the language commission, was made via
a submission to the constitutional committee by Zubeida Desai and Robin Trew.
They distinguished between passive and positive rights where, by adopting a
strong position on effecting rights (positive rights), citizens could be protected
from ‘exclusion from effective participation in public debate and the inequitable
enjoyment of public services, justice, education, power and economic advance-
ment’ (Desai and Trew 1992). Crawhall in analysing this contribution draws
attention to the importance of a comment included in Desai and Trew’s docu-
ment: ‘Language rights need to deal both with what Chinua Achebe has called
the unassailable position of English, and with the fact that African languages
are the primary linguistic resource of most South Africans’ (quoted in Crawhall
1993: 21). This contribution signalled a shift from an entirely rights-based po-
sition within the ANC to a stronger, more vigorous approach to rights that
acknowledges the view of language as a resource. Nevertheless, the position of
powerful forces within the ANC was more accurately captured by the ANC’s
‘Language Policy Considerations’.
vertical use. Apartheid language policy, which had partly been fuelled by a fear
of the international hegemony of English, had failed. The privileged position
of Afrikaans could certainly not continue if African languages remained weak
while English gained ground. English was being advanced by African-language
speakers who had lost confidence in the wider functionality of their languages.
Ironically, the Afrikaans-language lobbyists began to seek succour from the
proposals emanating from the NLP and PRAESA. The future of the Afrikaans
language would lie within a multilingual paradigm and a strengthening of the
functional use of African languages. The idea that languages could be viewed
as resources rather than as problems became increasingly attractive, and the
Afrikaans language lobby shifted from the segregationist position to a divided
commitment to language as a right, in order to protect its inevitable minority
situation in the future, and a tentative commitment to language as a resource.
A vigorous part of the Afrikaans lobby was spearheaded by the Stigting vir
Afrikaans, which saw the strategic value that resources, built up in the Afrikaans
language, might contribute to development in other languages. More conser-
vative elements within the white Afrikaans-speaking lobby threw their support
behind the notion of language as a right.
By early December 1993, the multiparty constitutional negotiations were
reaching closure. An eleventh-hour compromise was that there would be eleven
official languages. Afrikaans would not lose its official status, and the equal
status of eleven languages would be enshrined in the constitution. Additional
protection for Afrikaans was built into the interim constitution with what has
come to be known as the non-diminution clause:
3.(2) Rights relating to language and the status of languages existing at the commence-
ment of this Constitution shall not be diminished, and provision shall be made by an Act
of Parliament for rights relating to language and the status of languages existing only at
regional level, to be extended nationally. (Constitution of the Republic of South Africa
Bill 1993)
further the development of languages and protect the rights of each linguistic
community to use its language.
a final constitution which was eventually adopted in May 1996, and amended
on 11 October of that year. Again the language clauses could not be agreed
on until the final days of negotiations in October. The 1996 constitution sub-
stantively altered the language clauses and scaled down many of the earlier
provisions.
The expansive commitment to achieve the equal status of eleven languages
in the 1993 constitution has been de-emphasised, but the language clauses have
been tightened and should make it easier to develop an explicit national language
policy. It also makes clear the division of responsibility between government
and PANSALB (Constitution of the Republic of South Africa 1996: clauses
6.(1) – (5)). These are summarised as follows:
Principles that are the responsibility of government
r There are now eleven official languages in the country.
r The state has a responsibility to elevate the status and practical usage of those
official languages that did not previously enjoy official status.
r National and provincial governments must use at least two official languages.
r National and provincial governments must regulate and monitor their equi-
table use of official languages.
Principles that are the responsibility of PANSALB
r PANSALB must promote, and create conditions for, the development and use
of all official languages, the Khoe and San languages, and South African Sign
Language.
r PANSALB must promote and ensure respect for all other languages used in
the country.
The interpretation here of the clauses contained in the 1996 constitution are
that the frequency of terms such as ‘status’, ‘use’ and ‘usage’ point clearly
towards a paradigm based on functional multilingualism. The state is charged
with the responsibility of giving effect to the official status of eleven languages.
PANSALB’s role is that of strengthening and initiating the establishment of civil
society structures which support the development of interlinguistic/multilingual
skills not only in the official languages but also other languages used in the
country.
While the second round of constitutional negotiations were in progress a
number of language-policy and planning developments in line with the 1993
constitution were begun. These are as follows:
r Responsibility for language was brought under the newly structured Ministry
and Department of Arts, Culture, Science and Technology (DACST).
r The Senate as custodian of the PANSALB initiated legislation for its estab-
lishment. (The Senate was replaced by the National Council of Provinces in
the 1996 constitution.)
Recovering multilingualism 463
r The new Department of Education began to interpret the constitutional
implications for language in education.
r The Department of Defence began a lengthy process of defining a language
policy for itself.
(The first three of these will be discussed in more detail below. For a study of
language in the Defence Force see de Klerk and Barkhuizen (1998).)
During the LANGTAG process it became clear that the private and public
sectors continued to view languages other than English as a problem and, at best,
from the perspective of language as a right. It is not surprising therefore that the
LANGTAG report recommended various strategies for raising public awareness
about the value of multilingualism. The report specifically drew attention to the
need to identify goals and timeframes for implementation programmes.
Other early initiatives specific to the work of the DACST in relation to ex-
tending language services, notably in the form of a proposed trial telephone
interpreting service for South Africa (TISSA), signalled important changes.
The telephone interpreting service, designed to facilitate access to emergency
services for persons who speak languages other than English and Afrikaans,
would have been a major advance in language planning for South Africa. The
DACST, on behalf of government, could have added other services to TISSA.
Difficulties under Mtshali’s term of office effectively placed TISSA on hold
for the next four years. The DACST also took the initiative, through its State
Language Services, now the National Language Services, to focus on the role of
language in the economy and has encouraged exploratory research in this area.
Particularly useful have been their explorations into the role of languages and
trade for South Africa with the rest of the continent (DACST 1997). After the
second general election in mid-1999, Ben Ngubane was returned as Minister
of Arts, Culture, Science and Technology. Immediately the follow-up work
of LANGTAG was resumed vigorously by the department. Neville Alexander
was asked to convene the ministerial advisory body on language policy and
planning for the country. This advisory body was to oversee the drafting of
language legislation and define the regulations for language policy and plan-
ning. By 2001 the Cabinet of Ministers was presented with a final draft of
the South African Languages Bill and the Language Policy and Plan for South
Africa. Both of these documents concretise the clauses that give official status to
eleven languages and support multilingualism in the country at large. Whether
or not the national government will eventually have the Bill enacted is not yet
clear.
The structural conditions, however, under which its legislation (and subsequent
amendments to its legislation in 1999) placed it, as well as political pressures
which threaten the independence of the board, have rendered the body instru-
mentally weak. It took the board two years to be given the go-ahead to establish
its full-time staff and gain access to funds allocated to it from central govern-
ment via the DACST. Under Mtshali, the DACST saw its role as overseeing
and issuing directives towards the work of the board, whereas the board’s view
was that it was established to advise and monitor government activities rather
than take on tasks that should be performed by government departments.
PANSALB was legislatively bound to establish subsidiary structures, some
of which are consistent with promoting multilingualism co-operatively, such
as the establishment of advisory Provincial Language Committees comprising
representatives from each of the languages/clusters of languages in the respec-
tive provinces. Other structures, namely, the national language bodies for four-
teen languages or categories of languages, could be undermined by lingering
separatist interests. These bodies may not synchronise with an interdependent
approach to language development.
The 1996 constitution substantively altered the language clauses, and this
necessitated amendments to the legislation under which PANSALB exists. The
DACST drove the amendment process without negotiating with the successor to
the Senate in terms of the 1996 constitution, the National Council of Provinces
(NCOP). The DACST also compromised heavily with regard to consultation
with PANSALB itself. The board objected to the tabling in parliament of the
Pan South African Language Board Amendment Bill in 1998 on the grounds
that it was not given an opportunity to make known its reservations concerning
the amendments. Essentially, the process that unfolded seriously undermined
the autonomy of the board. The first deputy chairperson of PANSALB, Neville
Alexander, resigned from the board in March 1998 as soon as it became clear that
its autonomy was under threat. Structural arrangements in terms of the amended
legislation in effect make the board a sub-department of the DACST. In partic-
ular, the department succeeded in relocating lexicography units for each of the
official languages under PANSALB. This involves the board in hands-on lan-
guage development, and thus compromises its monitoring function, as it could
not both undertake and monitor its own work. The board had not necessarily
prioritised the making of dictionaries as the most immediately urgent activity
in the area of language development. Comprehensive dictionaries take decades
to compile and edit, whereas there were a number of language-acquisition and
other corpus-planning activities which the board had prioritised. Such units, in
fact, fall directly under the state’s responsibility for effecting the equal status
of the official languages.
One of the most successful interventions from the board was the commission-
ing of a national sociolinguistic survey of the country (PANSALB 2000). This
466 K. Heugh
survey has provided a wealth of current data which is necessary for accurate and
appropriate language-planning ventures. The prognosis for PANSALB, how-
ever, is not at this stage promising. Whereas encouraging developments have
emerged from within the DACST’s own language-policy and planning processes
since the return of Minister Ngubane in late 1999, the same cannot be said for
PANSALB. Damage caused during Minister Mtshali’s three-year period (late
1996–mid-1999) has not been repaired. PANSALB has not been given suffi-
cient funds to successfully establish the national lexicography units, and by late
2000 Minister Ngubane was insisting that PANSALB sacrifice funds already
committed to specific PANSALB programmes, in order to subsidise language
projects under the control of the DACST. Although at the final joint meeting of
the board’s first membership in November 2000 a decision was taken to refuse
the minister’s request, the officials of the board subsequently acceded to politi-
cal pressure. The capacity of PANSALB to act as a watchdog on government’s
implementation of the language clauses of the constitution has thus been suc-
cessfully eroded. The first five-year term of office of the board members came
to an end in March 2001 and appointment to the second board had not by early
2002 been completed by the DACST.
7 FURTHER OBSTACLES
of any government of Africa to implement the OAU’s 1986 Language Plan for
Africa is one of many examples of educational policies of good intent which
have never been effected. Since only a handful of sociolinguists or language
planners are presently committed to a resource-based approach, the more pow-
erful structural forces will undermine even the rights-based approach in South
Africa. There have to be instrumental or functional reasons why there should
be a paradigm shift towards harnessing the resources that African languages
offer. Unless the economic advantages of harnessing the languages of a country
in its economic development can be demonstrated unequivocally, the drive to-
wards monolingualism and the closure of access to power for the majority will
proceed unchecked. The economic benefits that local languages bring to small,
medium and micro-enterprises, as well as the degree to which use of local
languages might save time and costs in the activities of large corporations need
to be demonstrated.
Francois Grin, the most prolific of the writers on the relationship between
language and the economy or the economics of language has this to offer:
‘Economics can provide some of the essential ingredients to build a con-
vincing case to the effect that minority language promotion could deserve
state support – not for moral, political or cultural reasons, but for economic
reasons’ (1996: 16). Grin goes on to argue that there has to be an underlying
demand for language maintenance if language promotion is to have success:
‘The strategic implication is that demand must be strengthened, supported or
created prior to any other form of action. I consider this to be one of the very
few general results to hold in all minority language situations’ (1996: 16).
The key issue is thus, in his view, to establish the tangible value in linguistic
diversity/plurality. Grin’s argument is equally valid here where our concern in
Africa is in fact with indigenous languages, many of which are languages of
the majority.
A related matter which needs further cost–benefit analysis is the financial im-
plication of delivering educational resources and materials in local languages.
Surprisingly, World Bank researchers have provided evidence that the increased
cost of such materials has been exaggerated in the past. The viability of edu-
cational materials production in local languages would increase, however, under
the following conditions which take into account languages shared across
national boundaries: ‘A strategy for cost-efficient production of learning and
teaching materials would be to develop a unified curriculum and manufacture
books for target groups that encompass language groups beyond the boundaries
of one country’ (Vawda and Patrinos 1998: 24).
Ironically, of course, under apartheid, even though government provided
meagre financial support for the education of African children, schoolbooks in
several African languages for the first eight years of schooling were published,
and cost was clearly not an impediment then.
470 K. Heugh
Language policy and planning activities cannot be left divorced from the full
spectrum of economic activities and these, in turn, are related to areas of knowl-
edge and expertise. Thus if governments do not have the capacity to link the
domains and activities, smaller projects initiated from civil society need to do
this. If they were able to demonstrate how small systems which tap into local
expertise and knowledge are able to contribute to local and regional economies,
then eventually the prevalent consciousness might be altered. Maurice Goula
(1998) refers to this as a ‘rediscovery of Africa’s (Bio)diversity and endoge-
nous knowledge as a basis for social advancement’. It is a matter of reclaiming
a vast store of knowledge, indigenous to Africa. The new political space cre-
ated in South Africa during the 1990s provided an opportunity during which,
if it were possible for government to resist the structural pressures alluded to
by Pool, Tollefson, Skutnabb-Kangas and Phillipson, as discussed earlier, a
radical reconceptualisation of language policy and planning would have taken
root. That the South African government, despite arguably the most enabling
constitution in the world, has been floundering with regard to an unambiguous
definition of and commitment to a new language policy is sufficient warning to
language planners and service providers elsewhere on this continent not to wait
for transformation. The way forward now in South Africa is via the setting up,
and strengthening, of existing networks of experimental developmental projects
which, for the time being, have to be independent of the national system.
notes
1 Gogolin (1993: 3) defines this term as follows: ‘It is the basic and deep-seated ob-
session that monolingualism in a society, and particularly in schools, is the one and
only, overall, forever and always valid normality . . . The “monolingual habitus” is an
intrinsic characteristic of the classical national state.’
Recovering multilingualism 471
2 Sarah Murray (chap. 23, this volume) delineates a number of the tensions, paradoxes
and complexities of attitudes toward languages in education which bedevil the South
African school system.
3 See for example Skutnabb-Kangas 1988: 10: ‘Is monolingualism in fact a reflection
of an ideology, akin to racism, namely “linguicism”, the domination of one language
at the expense of others . . . ?’ (See also Phillipson 1992: 50–7.)
4 Michael MacMillan (1986) gives a detailed analysis of the tensions that which arise
between individual, group and collective language rights. Ultimately, in the absence
of highly regulatory legislation, the guarantee of language rights is difficult to meet.
5 Josef Schmied (1991) reminds us that in multilingual societies speakers of one lan-
guage incorporate elements of other languages into their language and vice versa.
6 This includes cultural and religious attachments.
7 ‘A sense of national identity is more likely to develop out of functional relationships
within a society than out of deliberate attempts to promote it’ (Kelman quoted in
Alexander 1989: 52).
8 Many of the influences on the South African education system reflect not only forces
peculiar to the country but ones that have had, and continue to have, their effect on the
rest of the continent. The history of the relationship between language in education, in
Africa, during the twentieth century, is one of repeated commissions of enquiry that
result in recommendations based on the centrality of indigenous languages as initial
languages of literacy and languages of learning. There are literally dozens of major
reports attesting to this particular issue; the following have been selected purely
for illustrative purposes: Phelps-Stokes Commissions of the 1920s; UNESCO’s
Report on the Use of the Vernacular Languages in Education 1953; the Lagos
Conference of Education Ministers of African Member States 1976 – which, ac-
cording to Bamgbose, recommended that ‘democratization, national character,
authenticity and modernization of education’ could only be achieved if national
languages are restored as national languages of instruction (Bamgbose 1996: 9)
(the Lagos conference also advocated the use of indigenous languages as the vehi-
cles for scientific and technical progress); the Harare Declaration by Ministers of
Education of African Member States linked education in African languages to socio-
economic development (UNESCO 1982); OAU (1986); the Pan African Colloquium
on Educational Innovation in Post Colonial Africa, Cape Town 1994.
9 There is a vast body of evidence, seldom made available in accessible form to parents,
which demonstrates that in the majority of situations, young learners in subtractive
bilingual programmes (i.e. those where the home language is not maintained and
extended for the duration of schooling) achieve poor academic results. Furthermore,
they fail to become proficient in English, and are only able to use the home language
for a limited range of communicative functions. (See, for example, Ramirez et al.
1991; Thomas and Collier 1997; Christian et al. 1997.)
10 The conference was held by the NLP at the University of Cape Town in 1991. It
was the first occasion when scholars from across the continent were able to attend
a sociolinguistic conference in South Africa after a lengthy period of academic and
other boycotts directed against the government. An important result of this event was
that valuable contacts in relation to language-policy developments in other African
countries were established.
11 This colloquium was co-ordinated by Professor Kwesi Prah of the University of the
Western Cape, but held at the University of Cape Town, 11–14 July.
472 K. Heugh
bibliography
Agnihotri, R. 1992. ‘India: multilingual perspectives’. In N. Crawhall (ed.),
Democratically Speaking: International Perspectives on Language Planning. Cape
Town: National Language Project, pp. 46–55.
Akinnaso, F. N. 1991. ‘Toward the development of a multilingual language policy in
Nigeria’. Applied Linguistics, 12.1: 29–61.
Alexander, N. 1989. Language Policy and National Unity in South Africa/Azania. Cape
Town: Buchu Books.
1992a. ‘Language Planning from Below’. In R. K. Herbert (ed.), Language and Society
in Africa. Johannesburg: Witwatersrand University Press, pp. 56–68.
1992b. ‘Harmonising Nguni and Sotho’. In N. Crawhall (ed.), Democratically
Speaking: International Perspectives on Language Planning. Cape Town: National
Language Project, pp. 143–9.
1998. ‘The political economy of the harmonisation of the Nguni and the Sotho
languages’. Lexicos, 8 (AFRILEX – reeks/series 8: 1998): 1–7.
Alexandre, P. 1972. An Introduction to Languages and Language in Africa. London:
Heinemann.
ANC 1992. ‘ANC language policy considerations’. Johannesburg: ANC, Department of
Arts and Culture.
Baker, C. 1988. Key Issues in Bilingualism and Bilingual Education. Clevedon:
Multilingual Matters Ltd.
Bamgbose, A. 1987. ‘When is language planning not planning?’ Journal of West African
Languages, 18: 6–14.
1996. ‘African Language Situation and its Implications for Language Planning and
Education’. Paper presented at the Pan African Seminar on the Problems and
Prospects of the Use of African National Languages in Education, Accra, 26–30
August 1996.
Christian, D., C. L. Montone, K. J. Lindholm and I. Carranza 1997. Profiles in Two-
Way Immersion Education. McHenry, Ill.: Center for Applied Linguistics and Delta
Systems Co., Inc.
Chumbow, B. 1987. ‘Towards a language planning model for Africa.’ Journal of West
African languages, 18: 15–22.
Recovering multilingualism 473
NLP and PRAESA 1995. ‘Urgent Submission to the Constitutional Assembly: Language
Clauses in the Final Draft of the Constitution’. Cape Town: Project for the Study
of Alternative Education in South Africa and the National Language Project.
Organisation of African Unity (OAU) 1986. ‘The Language Plan of Action for Africa’.
Twenty-Second Ordinary Session of the Heads of State and Government of the
Organisation of African Unity, Addis Ababa, 28–30 July.
Pan South African Language Board (PANSALB) 2000. Language Use and Language
Interaction in South Africa: A National Sociolinguistic Survey. Pretoria:
PANSALB.
Phillipson, R. 1988. ‘Linguicism: structures and ideologies in linguistic imperialism’.
In Skutnabb-Kangas and Cummins (eds.), pp. 339–58.
1992. Linguistic Imperialism. Oxford: Oxford University Press.
Pool, J. 1993. ‘Linguistic exploitation’. International Journal of the Sociology of
Language, 103: 31–55.
Rado, M. 1991. ‘Bilingual education’. In Liddicoat (ed.).
Ramirez, D., S. Yuen, D. Ramey, and D. Pasta 1991. Final Report: Longitudinal Study
of Structured English Immersion Strategy, Early-exit and Late-exit Transitional
Bilingual Education Programs for Language Minority Children. San Mateo, Calif.:
Aguirre International.
Ruiz, R. 1984. ‘Orientations in language planning’. Journal of the National Association
for Bilingual Education, 8: 15–34.
Schmied, J. 1991. English in Africa: An Introduction. London: Longman.
Siatchitema, A. K. 1992. ‘When nationism conflicts with nationalist goals: Zambia’. In
N. T. Crawhall (ed.), Democratically Speaking. Cape Town: National Language
Project, pp. 17–21.
Skutnabb-Kangas, T. 1988. ‘Multilingualism and the education of minority children’.
In T. Skutnabb-Kangas and J. Cummins (eds.), pp. 9–44.
2000. Linguistic Genocide in Education. Or Worldwide Diversity and Human Rights.
Mahwah, N.J.: Lawrence Erlbaum Associates.
Skutnabb-Kangas, T. and J. Cummins 1988a (eds.). Minority Education: From Shame
to Struggle. Clevedon: Multilingual Matters.
1988b. ‘Concluding remarks: language for empowerment’. In Skutnabb-Kangas and
Cummins (ed.), pp. 390–4.
Smolicz, J. J. 1990. ‘Language and economy in their cultural envelope’. Vox, 4: 65–79.
Thomas, W. and V. Collier 1997. School Effectiveness for Language Minority Students.
Washington: National Clearinghouse for Bilingual Education.
Tollefson, J. W. 1991. Planning Language, Planning Inequality. London: Longman.
Tripathi, P. D. 1990. ‘English in Zambia: the nature and prospects of one of Africa’s
“new Englishes”’. English Today, 6.3: 34–38.
UNESCO 1953. The Use of Vernacular Languages in Education. Paris: UNESCO.
1977. Education in Africa in the Light of the Lagos Conference 1976. Paris: UNESCO.
1982. Conference of Ministers of Education and those Responsible for Economic
Planning in African Member States: Final Report. Paris: UNESCO.
Vawda, A. and A. Patrinos, 1998. ‘Producing Educational Materials in Local Languages:
Costs from Guatemala and Senegal’. Paper presented at the Tenth World Congress
of Comparative Education Societies (WCCES), 12–17 July, University of Cape
Town.
Index
names
Aarons, D. 143 Buccini, A. F. 79, 85, 86
Achebe, C. 458 Bullivant, B. 423
Adendorff, R. 163, 399, 440 Bundy, C. 409
Alexander, N. 22f., 67, 425, 434, 436, Buthelezi, Chief M. G. 325
443, 456, 457, 463, 465, 467, 468 Buthelezi, Q. 357
Alexandre, P. 454 Bynon, T. 229
Allison, A. A. 324
Anders, H. 37, 41 Callaway, H. 190
Anderson, R. 349 Calteaux, K. 6, 244, 245, 251
Anthonissen, C. 421 Carbaugh, D. 258
Appel, R. 200 Carstens, P. 33, 205
Arbousset, T. 40, 69 Chick, K. 365
Armstrong, J. C. 89 Cluver, A. D. de V. 35, 36, 430
Ashforth, A. 414 Clyne, M. 232
Cobbing, J. 16
Bailey, R. 70, 71 Cole, D. 163, 188, 189, 199, 240
Baker, P. 89 Combrink, J. G. H. 86, 98, 208
Bamgbose, A. 454, 457 Comrie, B. 420
Barkhuizen, G. 24, 359, 463 Conradie, C. J. 85
Baumbach, E. J. M. 64, 71 Constable, P. 459
Beach, D. M. 34, 35, 313 Cooper, R. 419, 420
Bengu, S. 437 Corne, C. 89
Bennett, P. 56 Crawhall, N. 12, 22, 28, 365, 457–9
Bennie, J. 16, 31 Crystal, D. 449
Bhana, S. 161 Cummins, J. 438, 467
Bleek, D. F. 27, 36, 37, 39, 41, 43, 44 Cupido 40
Bleek, W. H. I. 27, 36, 38–40, 43, 50, 51,
297 Davids, A. 17, 83
Blench, R. 55–7 Deacon, J. 39
Bokamba, E. G. 365 de Kadt, E. 153, 373, 440
Bonner, P. 409 de Klerk, V. 6, 24, 263, 359, 365, 374,
Boonzaaier, E. 34 438, 463
Bot, M. 438 de la Bat, J. 131
Boyce, W. 16, 31 Delbridge, A. 110
Brain, J. B. 161 Delius, P. 409
Branford, W. 172 den Besten, H. 14, 82, 85, 88–90, 94, 95,
Brink, A. P. 208, 209 97, 98
Brown, D. 180 Desai, Z. 164, 457, 458
Bruyn, A. 87 Deumert, A. 83, 98
Bryant, A. T. 62, 300, 320 de Villiers Pienaar, P. 34
476
Index 477
languages
/Xam 27, 36–40 Afrikaans 1, 5, 15, 17–19, 29, 31, 32, 34,
/’Auni 27, 28, 36, 42, 44 41–4, 79–99, 106–9, 113, 120, 122, 134,
/’Auo 43, 44 136, 144, 150, 155, 157, 164, 181,
//Ku//e 37, 40 199–213, 216, 221, 223, 224, 228–31,
//Ng !k’e 37, 40, 41 237–9, 292, 358, 381–96, 398, 401, 402,
//Xegwi 37, 41, 42 404, 408, 410, 411, 420–4, 435–7, 440,
=| Khomani 27, 28, 36, 42–4 442, 444, 450, 456, 459, 460
!Gã !ne 37, 41 Cape Afrikaans 87
!Xóõ 44 Dutch-Afrikaans 209, 220
!Kwi 28, 36, 39–44 Eastern Cape Afrikaans 83
Hollands 80
Aarschot dialect 85 Khoekhoe Afrikaans 32, 33, 35
480 Index
English 1, 2, 15, 17–19, 29, 31, 32, 34, 41–4, Japanese 137
79–99, 106–9, 113, 122, 134, 136, 144, Ju/wasi 6
150, 155, 157, 164, 181, 199–213, 216,
221, 223, 237–9, 241, 292, 358, 381–96, Kalanga 11
398, 401–4, 410, 412, 421, 422, 424, Kanauji 161, 167
435–44, 450, 452, 454–6, 458–61 Kannada 161
Afrikaans/Afrikaner 122f. Kashmiri 166
American 106–7 Kgatla 52, 69
Australian 106, 112 Khoekhoe (also Khoe) 29–36, 63, 79,
British 106 84, 93–5, 97, 98, 306
Canadian 106 Khoesan 1–3, 11, 27–49, 53, 54, 297,
Cape 18 301, 303, 306, 308, 398
Cape Flats 123 Konkani 11, 162
Index 481
subjects
academic literacy 443 capitalism 18, 454
acrolect 91f., 96, 343ff. Centre for Education Policy Development
additive bilingualism 253, 439, 458 (CEPD) 436
Africanisation of English 440 chain shifts (in SAE) 113f.
African National Congress (ANC) 22, 402, China 12
423, 426, 427, 436, 456–9 Ciskei 13, 291
Afrikaans, codification of 206, 207 click consonants 29–31, 34, 62, 297–313
Afrikaners 15–18 cluster reduction (in Dutch) 92
Amsterdam 79 code-mixing 3, 216–33, 235
Anglicisation policy 450 code-switching 3, 16, 86, 216–33, 235–54,
Angola 11, 57 369f., 411, 434, 439, 440
anti-language 393, 402 conversational 225–8
apartheid 3–5, 18, 19, 141, 144, 340, 357, 374, conversational vs. situational 218f.
402, 412, 421–3, 426, 429, 435, 449, 453, markedness model of 245ff.
459, 468 matrix language frame model 251ff.
assimilation policies 452, 461 situational 224–5
attitudes codification 206–7
to BSAE 370ff. compliment-response studies 260–7
gender 327ff. compliment-response types 261
to Iscamtho 412f. constitution of South Africa 23–5, 450,
Australasia 106 460–7
Australia 1, 107, 458 contextualisation cues 268
convergence 96, 216–33
Bantu Education 18, 19, 357 corpus planning 420, 424, 465
basilect 343ff., 347ff., 351, 407 creolisation 85ff., 343, 351
basilectisation 98, 99 creolist hypothesis 87–92
Basters 33, 35, 37, 42, 81, 95 cross-cultural communication 258ff.
Basuto 17 Curriculum 2005 467
Basutoland 41
Bechuana 17 Delhi 166
Bhaca people 41 Democratic Party 111, 376
Bihar 166, 172 Department of Arts, Culture, Science and
bilingualism 13, 37, 43, 44, 87, 120, 200, 239, Technology 25, 72, 462–6
326, 372–3, 421, 422 Department of Bantu Education 19
Black Consciousness Movement 450 Department of Education 437, 463, 466, 467
Bophuthatswana 13 Department of Education and Training 358,
borrowing 3, 16, 84, 199–213, 217, 239–41, 435
293, 308, 310 dialect 105
Botswana 14, 44, 69, 307, 442 Dictionary of South African Signs 135, 140ff.
Britain 1, 17, 108 District Six 216, 217, 219–22, 228
Dutch East India Company 79, 81, 107
calques 200, 232
Cameroon 53, 55, 57 East Africa 57
Canada 1, 7 Eastern Cape 358, 441
Cape 11, 12, 14, 15, 18, 27, 29, 31–3, education 18, 22, 24, 28, 104, 129, 164, 253,
36, 37, 50, 88, 89, 91, 108, 149, 254, 340, 356–9, 373, 374, 421, 423, 427,
210 434–45, 453–5, 457, 459, 466, 467
eastern 11, 15, 16, 18, 24, 108, 189 elite closure 371
northern 27 embedded language 411
western 31–3 England 104
Cape Colony 79, 80, 86, 87, 89 extraterritorial varieties of English 106f.
Cape Flats 382
Cape Khoekhoe 27–9 Fingo 17
Cape Peninsula 381 France 128
Index 483
Ulundi 326 Xhosa people 16, 38, 108, 201, 279–94, 299,
‘unassailable position of English’ 458 300, 383
Union of South Africa 18, 108
United States of America 1, 106–8 Zambia 454f.
urbanisation 16, 279 Zimbabwe 15, 63, 71, 72, 319, 326–8, 330–2
Uttar Pradesh 165, 172 zones of convergence (SAE) 114
Zulu people 15, 17, 190f., 320, 324–6, 409,
variationist / interlectalist hypothesis 85–7 412
Venda 1, 2, 11, 50, 62 Zululand 15, 17