Southern African Humanities Vol. 20 Pages 7–35 Pietermaritzburg December, 2008
The early livestock-raisers of southern Africa
Department of History, University of California, 6265 Bunche Hall, Box 951473,
Los Angeles, CA 90095-1473, firstname.lastname@example.org
The Khoekhoe have long been recognized as historically distinctive livestock-raising people, whose ancestors
were responsible for the establishment of cattle-raising across the western half of southern Africa. A further,
no longer extant Limpopo Khoekhoe people have been identified as having had a major impact on the
establishment of cattle and sheep-raising in the eastern side of southern Africa as well. What has been less
clearly depicted is where the linguistically very closely related peoples of the Kwadi-Khoe branch of southern
African Khoesan stand in these developments, and what the impact of these changes might have been on
other, non-Khoe peoples. A third element, of particular relevance in the potentially correlative archaeology,
is the place of ceramic technology in this story. Together, these themes are key in proposing wider linguistic,
historical and archaeologically informative perspectives on the early history of livestock and livestock-
raising peoples in southern African history.
KEY WORDS: Loanwords, Khoe, Zhu, Tuu, Kwadi, Nguni, Sotho, livestock, ceramics.
Recent advances in the historical reconstruction of southern African Khoesan allow us
to revise and refine our understanding of early Khoekhoe history. The expanded detail
in particular gives fresh scope for correlation with the evidence from archaeology.
Two works contribute new and essential primary evidence, Rainer Vossen’s Die Khoe-
Sprachen (1997), which lays out the basic features of Khoe reconstruction, and this
writer’s Modeling historical linguistic reconstruction: proto-southern African Khoesan,
which is in preparation and will add further detail to Vossen’s findings. A third line of
investigation, undertaken by Tom Güldemann (e.g. 2004, 2006), sheds entirely new
light on the issue of how to relate the Kwadi language to the rest of the Khoe branch.
Together, these various findings require the discarding of several mistaken lexical
identifications (in Ehret 1982 and rightly criticized by Haacke this volume), and allow
a more extensive cultural and economic historical reconstruction based on new, well
Vossen (1997) constructs the detailed linguistic stratigraphy of the Khoe branch of
southern African Khoesan (SAK), of which Khoekhoe forms an offshoot. The Khoe
branch has two coordinate sub-branches, Khoekhoe and Kalahari Khoe (Vossen’s ‘Nicht-
Khoekhoe’ sub-branch). A multiplicity of evidence validates this binary division. To
Vossen’s evidence can be added two divergent sound change histories, not noted by
him, that reconfirm his conclusions as to the primary division of Khoe into Khoekhoe
and Kalahari Khoe sub-branches. One diverging pair of sound outcomes affected
aspirated clicks. Two rather than just one kind of aspirated click characterized proto-
Khoe, and these two clicks had alternative outcomes in the two sub-branches.1 The
sign K in these rules denotes a click:
1See link to http://www.history.ucla.edu/people/faculty?lid=472 for a tabling of cognate sets attesting rules
1–4. Vossen 1997 merges the cases of *tsx’ with those of a distinct PKhoe consonant *ts’. We can distinguish
PKhoe *tsx’ and *ts’ because *tsx’ had *|x’ as its PKK reflex, while PKhoe *ts’ became PKK *|’. It is
easy to miss this distinction, though, because PKK *|x’ and *|’ merged as |’ in the Namibian Khoekhoen
language, from which the bulk of our evidence comes, and because too little evidence is available for the
!Ora data, the one language to retain the distinction.
2SOUTHERN AFRICAN HUMANITIES, VOL. 20, 2008
1. Proto-Khoe (PKhoe) *Kh > proto-Kalahari Khoe (PKalK) *Kh
PKhoe *Kh > proto-Khoekhoe (PKK) *Kh
2. PKhoe *Kkh > proto-Kalahari Khoe (PKalK) *Kh
PKhoe *Kkh > proto-Khoekhoe (PKK) *Kh
The second pair of sound changes differently affected the proto-Khoe alveolar affricate
with ejective affricated release, *tsx’:
3. PKhoe *tsx’ > PKalK *ts’
4. PKhoe *tsx’ > PKK *|x’ (with a further change to Namibian Khoekhoen
As Vossen (1997) also establishes, the Kalahari Khoe branch divides into two
primary subgroups, West and East Kalahari Khoe. What Vossen does not situate in
this picture is the Kwadi language. Kwadi is clearly a close relative of the extant
Khoe languages. Of the 23 Kwadi core vocabulary (Swadesh 100-meaning-list)
items published by Westphal (1966), at least 11 (for man, woman, eye, nose, heart,
tongue, breast, meat, skin, night, and moon) and possibly 13 or 14 are common
Khoe roots. Arguing from that published wordlist, Ehret (1982) concluded that the
closest relationship of Kwadi was to the East Kalahari Khoe subgroup. Specific
indicators suggestive of an East Kalahari Khoe affiliation were the Kwadi attestation
of a systemic shift of reconstructed proto-Khoe dental clicks to non-click affricates
(5 instances among the 23 basic words) and the deletion of alveolar clicks, both of
which kinds of sound changes also characterized the East Kalahari Khoe subgroup
(e.g., Kwadi xwa- ‘zebra’ versus East Kalahari Khoe *kwaxa, from proto-Khoe
But this view is no longer tenable. Güldemann (2004) builds a strong case, based
on the reconstruction of pronoun histories, that Kwadi stands not only outside of
East Kalahari Khoe, but that Khoe and Kwadi should be considered sister sub-
branches of a deeper-level Kwadi-Khoe group. His conclusions have the added
virtue of accounting for certain previously unexplained lexical evidence from
Kwadi. Two Kwadi core vocabulary words, for woman and for moon, have
occurrence patterns that exclude Kwadi from membership in East Kalahari Khoe
as well in Kalahari Khoe as a whole. In both instances Kwadi uses a root word
present also in Khoekhoe, whereas the proto-Kalahari Khoe language innovated a
different word for each meaning: Khoekhoe // khãã and Kwadi khã- ‘moon’ versus
PKalK *|| noe; Khoekhoe tara- and Kwadi tala- ‘woman’ versus PKalK *|| gae-
khoe. A third Kwadi core vocabulary item directly supports Güldemann’s conclusion
that Kwadi stands as a sister to the Khoe group as a whole. The Kwadi word for
tree, thi-, appears to preserve a much older Khoesan root word also seen in Sandawe
thee ‘tree’, whereas all of the Khoe group, including Khoekhoe, share a common
innovated root *hai for tree.2 The preservation of the older root word is most
economically explained by Kwadi’s having diverged from the Khoe line of descent
before the new word *hai came into use in proto-Khoe.
2Vossen (1997) did not take into account diphthong simplification and a regular proto-Khoe h to y change
in front vowel environments in many Kalahari Khoe languages, and so he reconstructs this root only for
the Kalahari group and as *yi.
EHRET: EARLY LIVESTOCK-RAISERS 3
Together these findings depict the Kwadi-Khoe stratigraphic and historical tree as
shown in Figure 1. This tree adds to Vossen’s findings a further Khoe subgroup, Limpopo
Khoekhoe (identified in Ehret 1982). The diagnostic evidence for the classification of
Limpopo Khoekhoe is as yet sparse. One key datum is the proto-Southeast Bantu (PSEB)
borrowing of the word *-komo ‘cow’. The borrowed shape reveals an apparently
distinctive element of Limpopo Khoekhoe phonology: the retention of a pre-proto-
Cape-Gariep Khoekhoe word-final *o after the consonant /m/. As Beach (1938) showed
long ago, !Ora and Namibian Khoekhoe share a phonological structuring rule that,
among other things, disallows stem-final /o/ after the penultimate consonant /m/. Any
root word of the shape *CVmo (C = consonant, V = vowel) would have changed by this
rule to *CVma in proto-Gariep Khoekhoe. The same shape, with final *a, was recorded
from eastern Cape Khoekhoe dialects as well (Nienaber 1963: 414). Limpopo Khoekhoe
*komo thus must preserve the word as it was pronounced before the proto-Cape-Gariep
node on the Khoe family tree, requiring that proto-Limpopo Khoekhoe diverged from
the Cape-Gariep line of descent before that time.
Another sound change visible in *komo, PKK *g > Limpopo Khoekhoe *k, is also
found in the Nama-Damara branch of Gariep Khoekhoe, but not in !Ora. On the basis
of this shared devoicing shift, we might be tempted to classify Limpopo Khoekhoe
with Namibian Khoekhoen. But the PSEB language clearly dates to the early first
millennium AD, when Bantu peoples were first establishing themselves more widely in
Fig. 1. Kwadi-Khoe Stratigraphy.
4SOUTHERN AFRICAN HUMANITIES, VOL. 20, 2008
southern Africa (Ehret et al. 1972). The Southeast-Bantu borrowing of *-komo from
the Limpopo Khoekhoe thus had to have taken place centuries before the divergence of
proto-Gariep Khoekhoe into !Ora and Namibian branches, a development dating
probably to no more than about a thousand years ago (Ehret 1982). The Limpopo
Khoekhoe devoicing rule is thus best understood as a parallel but independent sound
change from that found in Namibian Khoekhoen.
The PSEB borrowing -gu ‘sheep’, in sharp contrast, does not show the Limpopo
Khoekhoe devoicing of PKhoe *g. The maintenance of *g in this PSEB borrowing has
two possible explanations: that the word came from Khoe languages other than Limpopo
Khoekhoe, or that it was an earlier borrowing from the Limpopo Khoekhoe language,
dating before the sound change. In either case its phonological shape supports the Khoe
branch evidence, to be presented below, that the adoption of sheep-keeping predated
the establishment of cattle-raising in southern Africa.
Names for an additional animal, the goat, also have a place in Figure 2. Their
cumulative implication is that goats spread to Khoe peoples later than either sheep or
cattle, a topic we will return to later.
SERIATING THE LEXICAL EVIDENCE FOR LIVESTOCK
Using the linguistic stratigraphy of the Khoe group as our framework, we discover that
the lexicon of livestock-raising evolved in two and possibly three stages (Fig. 2). In the
current state of our knowledge, two livestock terms, *goe ‘cow’ and *gu ‘sheep’ trace
Fig. 2. Khoe stratigraphy with seriation of food-production terminology.
EHRET: EARLY LIVESTOCK-RAISERS 5
to the proto-Kwadi-Khoe era. To the subsequent proto-Khoe stage, three additional
food-production-related words can be reconstructed: *||go, either ‘ram’ or ‘bull’ or both;
*tsx’ao ‘to milk’; and *||nubu ‘to churn (milk)’ (Vossen 1997). A fourth proto-Khoe
term, originally *||hoe ‘wooden vessel’ (Haacke & Eiseb 2002; but ‘water pot’ in Kilian-
Hatz 2003), may also be diagnostic of livestock-raising if its original primary use, as it
was in later times among the Khoekhoe, was as a milking container. The reconstructions
in Figure 2 apply rules established in and findings from Vossen (1997), and lexical data
from Barnard (1985), Haacke and Eiseb (2002) Kilian-Hatz (2003), Nienaber (1963),
Tanaka (1978), Visser (2001), and Vossen (1997) (for Western Cape Khoekhoe reflex
of *goe, see Valentyn (1726), Kolbe (1727); and in 1691 letter from Nicolaas Witsen,
cited in Nienaber 1963: 117, 345).
The early generic term *goe for cow is recorded, interestingly, to have two meanings
in the Khwe language of the West Kalahari Khoe subgroup of the Kalahari Khoe—
‘cow’ and ‘game [animal]’ (Kilian-Hatz 2003). The second meaning ‘game’ looks very
much like a retention in Khwe of the original meaning of the root, as a generic term for
large grazing animals that were important sources of meat. If this implication is correct,
*goe would have been an existing Kwadi-Khoe term specialized in meaning to apply to
a new source of meat, the cow. This kind of meaning shift would make particular sense
if it took place in the context of the spread of the first knowledge of cattle to people who
were still hunter-gatherers in economy and in their outlooks and attitudes.
There is a strong possibility that the meaning cow for *goe may not actually go back
to the proto-Kwadi-Khoe period, but may have come into use only subsequently. If
indeed the original meaning of this root word was ‘game animal’, then a plausible
candidate for a cognate is the Zhu|’hoansi word g!ò’é ‘oryx’ (Dickens 1994). If these
two words are cognate—and at this point it is only a provisional proposition—the loss
of the alveolar click *! would mean that this term spread among the Kwadi-Khoe
languages by borrowing from a language in which the loss of the click was regular.
This sound change is in fact regular in the Kwadi language, in the proto-language of the
East Kalahari Khoe subgroup of Kalahari Khoe, and isolated in the Khwe dialects of
the West Kalahari Khoe subgroup (Vossen 1997). The term does not appear as a word
for cow, however, in the East Kalahari Khoe branch languages at all, and the deletion of
the click in Khwe took place in relatively recent centuries, far too late in time to account
for ancientness of the original adoption of *goe. That would leave the Kwadi language
as the remaining possible source for the word, providing of course that *goe and the
Zhu terms are indeed old cognates. Such a history would place the spread of cows after
the initial divergences within Kwadi-Khoe had already taken place and after an early
loss of the alveolar click in the ancestral Kwadi language.
There is another datum favouring the conclusion that cows became known to Kwadi-
Khoe peoples later than sheep. In contrast to the universal use throughout the Kwadi-
Khoe languages of the same term *gu for sheep, two further generic terms for cattle,
additional to *goe, were innovated in different Khoe subgroups after the initial divergence
of proto-Khoe into daughter languages: *gomo in proto-Khoekhoe (PKK) and *be in
the proto-East Kalahari Khoe (PEK) subgroup of Kalahari Khoe. Whether or not the
*goe root and the Zhu word for oryx are cognate, this second stage of name adoption,
with new generic terms for the cow adopted in different subgroups, makes better sense
if it reflects a second stage in the knowledge of cows—that these animals became a
6SOUTHERN AFRICAN HUMANITIES, VOL. 20, 2008
familiar, established part of economic practices only after the divergence of proto-Khoe
into daughter languages was already underway. Viewed from this perspective, *goe
because of its wide occurrence would have been the earliest cattle term, presumably
drafted into use by the ancestral Kwadi people at a time when cows were viewed as a
new kind of game animal. They then diffused to the closely related proto-Khoe. In the
subsequent period, as proto-Khoe diverged into proto-Kalahari Khoe and proto-
Khoekhoe, the proto-Khoekhoe began to shift to full cattle pastoralism, acquiring a
second generic term for the cow from an as yet unidentified non-Khoe language (see
below for more on this point). The proto-East Kalahari Khoe descendants of the proto-
Kalahari Khoe later took up another distinct term, *be, for cow, the source of which is
as yet uncertain.
Still another datum, this one ethnographic, favours sheep as the earlier adoption. As
Monica Wilson has pointed out (Wilson & Thompson 1983), in Khoekhoe culture, in
direct contrast to the practices of Bantu-speaking cultures in southern Africa, the primary
animal of ritual was the sheep rather than the cow. A long-established axiom of
comparative ethnographic reconstruction is that the primacy of a particular animal or
crop in ritual observances is an almost certain marker of the greater antiquity of that
animal or crop (Sapir 1916).
The most notable indication of the overall lexical evidence is that it was the proto-
Khoe, and not their Khoekhoe descendants, who fully brought into existence the new
economy that blended livestock-raising, presumably initially just sheep-raising, with
hunting and gathering. Perhaps most interesting of all, the lexical reconstructions imply
that the proto-Khoe initiated the practices of milking their animals and of churning the
milk to produce a kind of butter.
Did this process of adding food production to foraging pursuits actually begin even
earlier, at the preceding proto-Kwadi-Khoe period? A distinct possibility is that the
adoption of herding might have proceeded in three stages, with the raising of sheep for
meat at the proto-Kwadi-Khoe period, and for milk as well as meat at the subsequent
proto-Khoe stage. Alternatively, the raising of sheep for both meat and milk might go
back to the proto-Kwadi-Khoe stage, but the available Kwadi lexical evidence is
inadequate to address this possibility. The addition of cattle-raising, which can be argued
to have constituted a third stage in this process, appears still somewhat later. As a primary
economic pursuit, it may have been a specifically Khoekhoe elaboration.
LOCATING AND DATING THE EARLY KHOEKHOE SOCIETIES
From the language evidence we must envision the original proto-Kwadi-Khoe as forming
a grouping of related hunting-gathering bands speaking one language, although possibly
with small dialect differences. In inferring the early home territories from the more
recent distributions of Kwadi-Khoe speakers, two questions must be considered. Where
did the centre of linguistic diversity lie at different periods of language and societal
divergence? What is the most parsimonious history of population movement that would
have accounted for the subsequent expansions from that centre of diversity? In the case
of the Khoe branch of Kwadi-Khoe, the linguistic geography of its latter-day subgroups
is most simply explained by a location for the proto-Khoe society in the far northeastern
Kalahari regions. For the Kwadi branch, a series of language contact evidence supports
placing its origins, surprisingly, in or close to the same broad region.
EHRET: EARLY LIVESTOCK-RAISERS 7
The Khoekhoe sub-branch of Khoe, because of the complexities of its history,
provides the most fruitful starting point for applying these principles. Two primary
branches of Khoekhoe once existed, Limpopo Khoekhoe, now entirely extinct, and
Cape-Gariep Khoekhoe, which comprised all the remainder of the Khoekhoe languages
and dialects (Ehret 1982). The initial centre of diversity within the Khoekhoe sub-
branch would have lain where the territories of these two sub-branches conjoined.
Where would that region have been?
The earliest set of Limpopo Khoekhoe loanwords date to the proto-Southeast Bantu
(PSEB) language. The most economical explanation of the geographical distribution of
the latter-day descendant languages of the proto-Southeast Bantu is to trace them back
to an original settlement somewhere between the southern Mozambique coast and
northern South Africa during the earlier first millennium AD (Ehret et al. 1972). GiTonga
of southeastern coastal Mozambique appears to constitute a sixth primary branch of
Southeast-Bantu (Laumann 1995), distinct from the neighboring Chopi branch, as well
as from the other four branches, Venda, Tsonga, Sotho (i.e. Sotho-Tswana), and Nguni.
Five of the six branches of Southeast-Bantu cluster in a compact stretch of country,
running from the modern-day Venda and northern Sotho lands in northern South Africa
across the lower Limpopo to the coast at Inyambane, where two of the branches, Chopi
and GiTonga, concentrate today in one small coastal region of higher rainfall. The
adoption of Khoekhoe loanwords into PSEB in the early first millennium AD must
therefore have taken place somewhere in that stretch of country, hence the name
‘Limpopo’ Khoekhoe for the source language of those loans.
After the divergence of proto-Southeast Bantu, two of its daughter languages entered
into a closer and more pervasive kind of cultural interaction with Khoekhoe speakers,
very different from the narrowly focused economic encounter betokened in the loanwords
of the PSEB era. A very different kind of word borrowing accompanied these interactions,
indicative in each case of close intertwining of the different communities and the eventual
merging of the Khoekhoe into the Bantu-speaking societies. Limpopo Khoekhoe
loanwords in the proto-language of the Sotho subgroup of Southeast-Bantu, along with
word borrowings from a less certainly identifiable Khoekhoe language in proto-Nguni,
reveal a widespread Khoekhoe presence in the northern provinces of modern-day South
Africa, lasting into the second half of the first millennium.
The loanword set in proto-Nguni includes words of the kind that fit a category of word
borrowing that has been called ‘heavy general’ (Ehret 2005). This category is characterized
by the penetration of the borrowings into relatively basic and not just cultural vocabulary,
as a sample of the proto-Nguni evidence identified by Carolan Ownby (1985) shows (Table
1). A notable additional indicator of the depth of the Khoekhoe impact in proto-Nguni was
the adoption of the click sounds along with the words in which they occurred.
Proto-Nguni English Khoekhoe / proto-Khoekhoe
*-n|| ele ‘left (side)’ proto-Khoekhoe (PKK) *|| are
*-!hwa ‘ice’ Khoekhoe !khoa-b ‘ice’; !khoa “to freeze”
*-tsheli ‘yellow’ known from Khwe tcere- (West Kalahari Khoe)
*-|ham- ‘to urinate’ PKhoe and PKK *|xam
Sample Khoekhoe loanwords in proto-Nguni.
8SOUTHERN AFRICAN HUMANITIES, VOL. 20, 2008
The history of language shift that accompanies heavy general borrowing typically
has two characteristics. First of all, it happens as part of the gradual absorbing, over an
extended period, of the people from whose language the loanwords come into the
society that borrows the words. Secondly, the donor society of the heavy general
loanword set, at the initial stages of interaction, is at least equally, and often more,
populous than the society into which they are eventually absorbed. When significant
new additions to the phonology accompany heavy general borrowing—in this case, a
wholly new category of consonants, the clicks—the demographic implications are
doubly strong. The source Khoekhoe community of the loanwords in proto-Nguni
could well have comprised a majority component in the evolving proto-Nguni society.
This characterization should not be taken as implying that great numbers of people
participated in this history. The categories of word borrowing tell us something
about the relative proportions involved in a history of language shift; the absolute
sizes of the communities involved might have been no more than a few thousand people
As to the historical geography of these contacts, scholars have often read recent
history back into the deeper past. They have assumed that the proto-Nguni society
originated below the Great Escarpment and have thus attributed the Nguni-speaking
populations of the Highveld—the ‘Southern Ndebele’, ‘Northern Ndebele’, and Phuthi—
to early pre-Mfecane movements inland across the escarpment. Carolan Ownby (1985)
proposes, however, that the Highveld Nguni languages actually represent the first three
divergences within Nguni. In her analysis, all the Nguni below the escarpment, from
Swati to Zulu to Xhosa, comprise a fourth branch, which she calls Kahlamba. In Ownby’s
classification, in other words, three of the four deep branches of Nguni are restricted to
the Highveld. Only one of the four is situated primarily in Kwazulu-Natal. If she is
right, the center of diversity, and thus the most probable origin land of the proto-Nguni,
lay somewhere in the eastern Highveld, between Ngwane, Lesotho, and the areas east
of the Rand.
Ownby’s findings have fundamental implications for interpreting the archaeology of
KwaZulu-Natal. In particular, she shows—and this is the incontrovertible part of her
study—that two successive Bantu populations, of which the Nguni were the second,
have occupied KwaZulu-Natal and areas as far south as the Kei over the past two
millennia. The very large numbers of loanwords from an extinct set of Bantu languages
related apparently closest to Shona, which occur in all the Nguni tongues spoken from
Swaziland in the north to IsiXhosa in the south, put this conclusion beyond doubt.
These loanwords penetrated sometimes into the most basic vocabulary—they fit the
categories of either ‘heavy general’ or ‘intensive’ borrowing (Ehret 2005)—
demonstrating that the Kahlamba Nguni spread across these regions, initially as an
incoming minority, by successively absorbing into their communities the previously
established Bantu societies of KwaZulu-Natal and the Transkei. Ownby concludes that
this cultural and linguistic changeover began in the eleventh century, accompanying
the one major period, attested in the archaeology, of a sweeping shift in the economy
and settlement patterns across those regions.
The germane point for Khoekhoe history is that, if the proto-Nguni of the late first
millennium AD were inhabitants of the eastern Highveld, then the Khoekhoe with whom
they interacted so extensively as to borrow even peripheral basic words (e.g. ‘ice’, ‘to
EHRET: EARLY LIVESTOCK-RAISERS 9
urinate’) must have lived somewhere in the same broad region. This Khoekhoe
community cannot be identified with the separate and distinct Eastern Cape Khoekhoe
populations who so strongly influenced the Southern Nguni societies living between
the Mzimvubu and Kei Rivers much later on, in the middle centuries of the second
millennium (Ownby 1985).
A second group of Southeast-Bantu communities which had major early interactions
with Khoekhoe communities were the proto-Sotho, whose language was ancestral to
the all various Sesotho and Setswana dialects spoken today. The proto-Sotho society,
like the proto-Nguni, probably dates to the latter parts of the first millennium.
One striking indicator of the Khoekhoe impact on proto-Sotho is the phonological structure
of the Sotho consonant system. Even though proto-Sotho did not preserve the clicks in
borrowed Khoekhoe words, it did change radically the overall structure of the consonant
system in ways characteristic of Khoe languages. In Khoekhoe, for example, /t/ does not
exist intervocalically (between vowels) in stems. The only alveolar/dental sound possible
in that position is /r/ or, alternatively in some Khoe group languages, /d/. Proto-Sotho did
away with proto-Bantu *t, changing it to *r between vowels. A second major sound change
in proto-Sotho shifted the consonant system further in a Khoekhoe-like direction. Voiced
prenasal stops, a characteristic and defining feature of Bantu languages, do not exist in
Khoekhoe. They no longer exist in Sotho either. A combined de-nasal/devoicing shift
changed proto-Bantu *mb to proto-Sotho *p, *nd to *t, and *ng to *k. (A side effect of this
sound change was the creation of new intervocalic *t in proto-Sotho, replacing the original
Bantu *t which had previously become *r in early proto-Sotho.)
The history of word borrowing in the Sotho branch of Southeast Bantu has yet to be
given the kind of study that Ownby applied to Nguni. It is an urgent and important task
to pursue. Even without that work yet undertaken, it is relatively easy to identify
considerable numbers of possible and probable Khoekhoe loanwords in proto-Sotho. A
selective listing of such items follows (Table 2). The proto-Sotho roots are reconstructed
on the basis of their occurrence, with regular sound correspondences, in both North and
South Sotho languages (Kriel 1983, Mabile 1976). In these data the Bantu suffixes and
verb extensions that the Sotho languages subsequently added to the borrowed roots are
marked off from the roots by hyphens.
Several patterns emerge in these data. Unlike proto-Nguni, proto-Sotho, rather than
adopting the original click consonants converted them to non-clicks. The lateral and
alveolar clicks (|| and !) became velar affricates at the beginning of a word or verb stem
in Sotho, but velar fricatives in the middle of a word. The Khoekhoe palatal click ( ), in
contrast, apparently yielded the affricate /ts/ in Sotho. The alveolar, palatal, and velar
affricates of Khoekhoe and the Khoe branch remained affricates (ts and kx) in the
borrowings, while the Khoe languages’ aspirated stops also preserved that articulation
(th and kh) in the loanwords. The preservation of PKK *r as *r in the proto-Sotho word
borrowings shows that most of the Khoekhoe words entered the proto-Sotho lexicon
after the proto-Sotho sound change rule, PB *t > proto-Sotho *r between vowels, took
effect. If the words had been borrowed before the sound change, the cases of borrowed
Khoekhoe *r would have been changed to *l in proto-Sotho, as did happen in at least
one case, the word for ‘mead’. This particular item thus appears to have been borrowed
early in the contacts between Khoekhoe and proto-Sotho speakers or, alternatively,
from a different Khoesan language, in which PKhoe *d became *l.
10 SOUTHERN AFRICAN HUMANITIES, VOL. 20, 2008
The kinds of borrowings provisionally identified for proto-Sotho appear to belong to
the category of heavy general borrowing. As for proto-Nguni, this category implies that
the Khoekhoe who interacted with the proto-Sotho were possibly even a majority demic
component in the formation of that society.
Where do we locate the proto-Sotho? Lacking as yet the sort of study Ownby
has carried out for Nguni, we have to fall back for now on the general recognition
that the dialectally most diverse region within the wider Sotho-speaking territories
lies in the northeastern parts of South Africa, where such divergent branches of
Sotho as Pedi, Lovedu, and Palaborwa are spoken. Provisionally, we can propose
Proto-Sotho English Khoekhoe / proto-Khoekhoe
*kxama hartebeest Khoekhoe and PKK *|| xama (Southern Nguni *-|| hama is a
separate, historically recent borrowing from Eastern Cape
*-kxob-e- to run about in PKK *|| gubu ‘to run with difficulty’ (PSAK *g|| xobu).
*-kxab- to cut up in pieces; PKK *|| ’abe ‘to chop, cut up’ (also PSAK).
*-kxar-its- to scratch Khoekhoe || gara (PSAK *g|| x’adV).
*se-xwaxwa frog Khoekhoe || goã-‘large frog’.
*-kxao-l- to cut through; Khoekhoe !gao ‘to cut, cut up, cut apart, slice’.
*-kxao-x- to be cut through
*kxai piece of cloth Khoekhoe !khaib ‘head-cloth, scarf’ (areal Southern Africa
Khoisan (SAK) *!xai).
*kxadi mead Khoekhoe !kharib ‘mead’ (implies borrowing in which PKK
*r > pre-proto-Sotho *l > proto-Sotho *d /_i by regular Sotho
rule;hence borrowed before *t > proto-Sotho *r; rest of cases of
PKK *r in this table took place after rule creating Sotho *r
*tsiritsiri cricket Khoekhoe giri giri.
*-tsi-ets- to deceive Khoekhoe tsiitsii-sen ‘to disguise oneself’ (root *tsi ‘to
deceive’ reduplicated plus recipr.-sen).
*-tsar-an-y- to spit out with Khoe: Khwe tc’aro ‘to spit’.
*-tsarar-a very bitter, sour PKhoe *tsadu ‘to be sour’ (*d > PKK *r).
*thari skin for PKhoe *thadi ‘leather garment worn around shoulders’
carrying child (PSAK *thadi; !Xoõ thari ‘skin for carrying child’).
*-kham-ol- to separate, divide Khoekhoe kham ‘to skim off foam’ (PSAK *kham ‘to separate
[one thing from another]’).
*-kxom-ol- to pull or pluck Khoe: Khwe kx’om ‘to pull out, tear out’.
*-kxom-ox- to break off (intr.)
*-kxom- to rub, smear PSAK *Gq’om ‘to rub’.
*-par-ax-an-y- to race, as a horse PKhoe (and PSAK) *badV ‘to run (away)’.
1(S.Sotho -par-atl- (PKhoe *d > PKK *r).
1‘to make a noise
*-par-al-al- to become stiff, Zhu bara ‘to rise (hair on back of neck, quills
erect of porcupine)’ (not yet recorded for Khoe branch).
Sample Khoekhoe loanwords in proto-Sotho.
EHRET: EARLY LIVESTOCK-RAISERS 11
this region as encompassing the most probable locations of the emerging proto-
Sotho society in the later first millennium AD. For that reason we must also place
in or near these areas the Khoekhoe populations with whom the proto-Sotho so
Were the Khoekhoe communities who contributed to the proto-Nguni and the proto-
Sotho societies the descendants of the Limpopo Khoekhoe from whom the PSEB had
previously adopted cattle-raising and milking? The available evidence in the proto-
Nguni case does not yet provide diagnostic indicators of such a link. In contrast, the
loanwords in proto-Sotho do provide such indicators. Specifically, in the last two root
words in the sample, for ‘to race’ and ‘to become stiff’, PKhoe voiced *b became
voiceless /p/. Because proto-Sotho did retain the voiced stop consonant *b in its inherited
old Bantu lexicon, its speakers would have kept *b if it had been present in the loanwords.
The only viable explanation for the occurrence of *p in the borrowings is that *b became
*p in the Khoekhoe source language before these roots were borrowed. The Limpopo
Khoekhoe language, as we saw in the PSEB borrowing, *-komo ‘cow’, from original
PKK *gomo, apparently did have just such a rule devoicing voiced stops. The most
parsimonious explanation is that the two loanwords with *p in proto-Sotho came from
a later descendant dialect of the same language, Limpopo Khoekhoe, as the PSEB word
The linguistic evidence from Sotho thus supports the proposal that the Limpopo
Khoekhoe cattle pastoralists, after influencing the proto-Southeast-Bantu in the first
half of the first millennium AD, continued to be a significant presence in northern
South Africa on into the second half of the millennium. If the particular Khoekhoe who
interacted with the proto-Nguni were also of Limpopo Khoekhoe ancestry, the lands of
the Limpopo people would have extended as far south as the eastern Free State.
Alternatively, if the proto-Nguni borrowings derive from an extinct Cape-Gariep dialect,
the evidence would imply a former, otherwise unsuspected eastward extension of Cape-
Gariep speakers across the southern Highveld.
The known Cape-Gariep languages belong to two sister groupings, the Cape Khoekhoe
and the Gariep Khoekhoe (Ehret 1982). The most economical explanation of the later
Cape-Gariep Khoekhoe distributions is that the proto-Cape-Gariep society lived
somewhere in the broad sweep of lands around and north of the Vaal-Gariep confluence.
The early Cape Khoekhoe diverged out of this grouping of communities by moving
southward, initially to the Eastern Cape, before spreading out eastward from there to
beyond the Kei River and westward as far as the Cape of Good Hope (Ehret 1982).
The communities that remained around and northward from the Vaal-Gariep
confluence became the proto-Gariep Khoekhoe. According to previous proposals,
centuries later an offshoot of the proto-Gariep people expanded westward (Ehret 1982;
Elphick 1977). Spreading through the areas between and around the Gariep River and
the Malopo, their descendants brought the Namibian Khoekhoen language to Namibia.
Two newly available lines of linguistic evidence support these proposals. One is the
presence in the Gariep Khoekhoe languages of Tuu Khoesan loanwords, most clearly
attested in Namibian Khoekhoen; this will be described in detail in my forthcoming
book mentioned earlier. These borrowings require a history of direct contact between
these peoples. Güldemann (2006) independently establishes the same case, but from
complementary evidence, namely, the presence of notable Tuu grammatical influences
12 SOUTHERN AFRICAN HUMANITIES, VOL. 20, 2008
on the Namibian Khoekhoen and !Ora languages. The old territories of the northerly
Tuu peoples, from whom these language influences and this new demographic element
would have come, straddled the regions extending from the Vaal and upper Gariep
Rivers westward through the lands around and between the courses of the Malopo and
the middle Gariep. Thus the early Gariep Khoekhoe encounters with Tuu hunter-gatherers
must have taken place in these regions as well.
As previously argued, the evidence of loanword histories in proto-Southeast Bantu
and its daughter language, proto-Sotho, locate the Limpopo Khoekhoe, the sister branch
of Cape-Gariep Khoekhoe, north and northeast of these areas in more northerly parts of
South Africa. The region where the early lands of the early Limpopo Khoekhoe would
have adjoined those of the proto-Cape-Gariep people would have thus have been to the
north of the lower and middle Vaal River.
LOCATING THE PROTO-KHOE AND THE PROTO-KWADI-KHOE
The most parsimonious explanation for the distributions of the two Limpopo and Cape-
Gariep branches of Khoekhoe, is that their common cultural and linguistic ancestors,
the proto-Khoekhoe people, came from the areas immediately to the northwest, where
Botswana today borders on northern South Africa and southwestern Zimbabwe. The
reason for placing the proto-Khoekhoe in these areas is that it most economically accounts
for the earliest locations of the two coordinate branches of the Khoe group. A location
of the proto-Khoekhoe in the borderlands of Botswana with southwest Zimbabwe and
the Northern and Limpopo provinces of South Africa would place them directly adjacent
to and along the eastern edges of the long-time lands of the second branch, Kalahari
Khoe. Just one initial expansion of each of the daughter societies of the proto-Khoe is
sufficient to explain the subsequent early population distributions—a spread of the proto-
Kalahari Khoe from this region westward across northern Botswana, in the process
diverging into proto-West and proto-East Kalahari Khoe societies, and a spread of the
proto-Khoekhoe southeastward into northern South Africa, subsequently diverging into
the proto-Cape-Gariep and the early Limpopo Khoekhoe societies. This history of
expansion places the most probable location of the ancestral proto-Khoe also in
northeastern Botswana and the adjacent western fringes of Zimbabwe. Placing the proto-
Khoe location farther west in northwestern Botswana (e.g. Heine & König this volume)
is possible, but less probable.
Locating the still earlier proto-Kwadi-Khoe stage rests on a different kind of reasoning.
The last speakers of the Kwadi language lived in the coastlands just north of the lower
Kunene River in Angola, far to the west of the early stages of Khoe history. Three
pieces of evidence, however, suggest that the Kwadi originated close to the proposed
lands of the proto-Khoe in northeastern Botswana.
One datum is a word, kx’a, ‘soil, earth’, uniquely shared between Kwadi and the
distinct Zhu group of languages (Westphal 1966). The evidence is insufficient to rule
out the possibility of this word’s having been an old shared cognate going back to a
much more ancient common southern African Khoesan ancestry for the Zhu and Kwadi-
Khoe language groups. The probability is high, however, that it is a borrowing either
from Zhu into Kwadi or from Kwadi into the Zhu group. Whichever the direction, the
borrowing of a core vocabulary item is strong evidence of earlier close interactions
between the borrowing and the adopting language, indicating that the Kwadi at an
EHRET: EARLY LIVESTOCK-RAISERS 13
earlier stage in their history inhabited areas in northwestern Botswana or southeastern
Angola where the languages of the Zhu group are, and have long been, spoken.
A second datum is the *goe term for cow. If as proposed, this was a Kwadi loanword
in the West Kalahari Khoe and Khoekhoe languages, then the ancestral Kwadi in the
era of the first knowledge of cattle must also have resided close to the regions of northern
Botswana in which the first divergence of proto-Khoe took place. Because cattle first
reached this region around or before 2000 years ago, the contacts would date to that era
Still more notable are the shared phonological features of Kwadi and the East Kalahari
Khoe subgroup, namely (1) the shift of the palatal clicks to fricative and affricate dentals/
alveolars and (2) the entire deletion of the alveolar click. The loss of the alveolar click
spread in very recent centuries to dialects of the northern subgroup of West Kalahari
Khoe, but was a far older change in East Kalahari Khoe. The additional sound changes
affecting the palatal clicks are strictly Kwadi and East Kalahari Khoe phenomena.
Güldemann’s finding that Kwadi is a sister language to the Khoe group means that
these parallel sound changes (contra Ehret 1982) cannot be attributed to a closer genetic
relationship between the two.
One possible explanation is that separate histories of long-term contact of proto-East
Kalahari Khoe and Kwadi with non-click languages brought about the parallel changes
in phonology (Tom Güldemann pers. comm.). But that explanation seems improbable
in this instance on at least two grounds. In the first place, it is unlikely that separate
contact histories would entirely independently engender such precisely parallel sound
changes, even if contacts with non-click languages do tend to lead to click consonant
loss, and even if certain kinds of clicks tend to be more susceptible to loss than others.
As an example, the now extinct ||Xegwi language, embedded probably for centuries in
Sotho-speaking lands, lost some but not all of its palatal clicks. It did so differently
from proto-East Kalahari Khoe and Kwadi, by changing them to lateral obstruents (tl’
and : e.g. Westphal 1966: || Xegwi tl’oo ‘man’ versus N|huki (|| -) o; || Xegwi tl’e
‘persons’ versus N|huki (|| -) e; || Xegwi wi ‘ear’ versus N|huki || uí- and ||Xegwi
tl’óló ‘moon’ versus N|huki ’óló).
Secondly, contacts with non-click languages do not necessarily lead to click loss. Sandawe
of Tanzania, despite having borrowed more than sixty per cent of its vocabulary from
neighbouring Southern Cushitic, Southern Nilotic (Datoga), and Bantu languages, shows
no signs of click loss, and in its non-borrowed lexicon click words occur in proportions
rivalling those of South African click languages (initial assessment by Christopher and
Patricia Ehret of the Sandawe dictionary materials recorded on around 10 000 note cards in
the 1960s and 1970s by the anthropologist, Erik ten Raa, archived in the Special Collections,
Young Research Library, University of California at Los Angeles).
The remaining alternative explanation is that the shared phonological changes owed
principally to spatial developments, spread from either ancestral Kwadi to proto-East
Kalahari Khoe or from proto-East Kalahari Khoe to Kwadi, at a period when the ancestral
Kwadi and proto-East Kalahari Khoe speakers were closely interacting neighbours.
The entire restriction of the East Kalahari Khoe group to northeastern Botswana and
the adjoining southwestern fringe of Zimbabwe therefore supports the proposition that
the ancestral Kwadi once inhabited lands in or adjacent to these anciently East Kalahari
14 SOUTHERN AFRICAN HUMANITIES, VOL. 20, 2008
Viewed as a whole, the contact evidence in Kwadi is most economically accounted
for by a three-stage westward expansion of the ancestral Kwadi. In the period following
the divergence of proto-Kalahari-Khoe into proto-West and proto-East Kalahari Khoe,
the Kwadi phonological sharings with early East Kalahari Khoe would place them in
northeastern Botswana, where proto-East Kalahari Khoe undoubtedly would have been
spoken. The occurrence of the *goe in both proto-Khoekhoe and Kwadi, if it indeed
spread by borrowing from early Kwadi, would also place the ancestral Kwadi people
somewhere in and close to northern and northeastern Botswana. At a second stage,
dating to or not long after the era of first acquaintance with cattle, and thus probably to
2000 years ago (e.g. Robbins et al. 2005, 2008) or somewhat earlier, the Kwadi would
have advanced westward, possibly along the Caprivi Strip or adjacent areas of northern
Botswana, encountering the Zhu peoples. The Kwadi sharing of at least one core-
vocabulary word uniquely with the Zhu group may indicate that significant numbers of
former Zhu speakers were absorbed into the ancestral Kwadi society in this era or,
alternatively, that the ancestral Kwadi had a strong impact on the Zhu; but more evidence
would be needed to explore these possibilities.
This sequence of contacts draws the origins of the Kwadi branch of Kwadi-Khoe
back eastward, to the same broad region in which the arguments from linguistic
geography and parsimony would place the proto-Khoe society. Such an origin area for
the ancestral Kwadi would place the still earlier divergence of proto-Kwadi-Khoe into
the ancestral Kwadi and proto-Khoe communities also somewhere in or close to
northeastern Botswana, perhaps in the Chobe area or in southwestern Zimbabwe, or
possibly even across the Zambezi river in far southwestern Zambia. Locating the proto-
Kwadi-Khoe in this region, with sheep, makes good sense in view of the probability
that both sheep and cattle reached these areas from northeasterly directions, ultimately
from East Africa. The fact that sheep-keeping spread a further 1500 km south from
northern Botswana to the coastlands of the Eastern Cape already by the first century
AD requires the original proto-Kwadi-Khoe acquaintance with sheep to have begun, at
a minimum, several centuries earlier than that, well back in the first millennium BC.
HERDING ECONOMIES AND POPULATION EXPANSION
The successive elaborations of the lexicon of livestock in the early periods of Kwadi-
Khoe history (Fig. 2) coincide with widening expansions of the speakers of these
languages. This parallel between lexicon and migration history points up the effects
that livestock-raising would have had in increasing subsistence productivity and the
pressure to find additional grazing lands for growing herds. With sheep the need for
expanded grazing might be minimal, depending on the extent to which sheep-keepers
might keep animal population growth in check by relying on their animals for meat. In
the case of cattle, important for milk and for reckoning wealth, the growth of herd size
has high positive value, the butchering of the animals tends to be infrequent, and the
pressure for expansion into new areas tends to be much greater.
The low frequency of small-stock-raising, or the lack of livestock altogether, among
most West and East Kalahari Khoe speakers of more recent centuries might reflect a
loss of livestock brought about by the competitive inroads of more numerous Bantu-
speaking livestock-raisers in the past five centuries. But the tendency to strongly forager
worldviews among many of the West and East Kalahari Khoe; the drying of climate in
EHRET: EARLY LIVESTOCK-RAISERS 15
the northern Kalahari after a wetter period c. 2000–1500 b.p. (Robbins et al. 2008); and
the historical example of the spread of the Khoekhoen language to hunter-gatherer
populations in Namibia fit an alternative historical scenario. The original proto-Kalahari
Khoe people can be proposed, in this view, to have possessed a fully established mixed
sheep-raising and foraging economy. Their expansion in the late first millennium BC
and early first millennium, and their divergence into proto-West and proto-East Kalahari
Khoe societies would have come about because of their possession of this more
productive combination of subsistence practices.
Their dominant economic position in turn would have led to the adoption of the East and
West Kalahari Khoe languages by many of the existing hunter-gatherer communities of
those regions. Foraging communities in northern Namibia, such as the Hei|| kom, similarly
can be proposed to have adopted Namibian Khoekhoen as their home language during the
eras of Nama expansion into northern Namibia—before the recent expansions of the Herero
and other Southwest Bantu groups pushed the Nama pastoralist boundaries back southward.
In the Kalahari the distribution of the West Kalahari Khoe languages suggests that sheep-
raising proto-West Kalahari Khoe groups pressed south toward the centre of the Kalahari,
spreading their language to local hunter-gatherer communities, before shifts to drier climate
by or before mid-first millennium AD curtailed the viability of this economy.
The acquisition of cattle, because of the greater tendency of cattle-keeping to lead to
pressures for expansion, likely coincided with the inception of a different and much
more extensive period of language expansions. Consistent with the lexical evidence for
the adoption of cattle, the set of expansions in which cattle ownership was key were
those of the proto-Khoekhoe and their descendant communities southward, first into
the watershed areas of the middle Limpopo River; then farther south in northern South
Africa and in the Gariep-Vaal confluence region; and finally into the Eastern Cape and
from there eastward to beyond the Kei River and westward to the Cape of Good Hope.
In this interpretation, the adoption of cattle had a transformative economic impact on
the Khoekhoe, shifting their productive activities to full-scale pastoralism, with hunting
and gathering very much supplementary. The adoption of Limpopo Khoekhoe pastoral
loanwords, specifically *-komo ‘cow’ and *-pi ‘milk’, already in proto-Southeastern-
Bantu (PSEB), a language spoken in the early or middle first millennium AD (Ehret et
al. 1972), places the inception of these developments among the proto-Khoekhoe
ancestors of the Limpopo Khoekhoe no later than the early first millennium AD and
possibly in the late first millennium BC.
The proto-Khoe term *gu ‘sheep’, besides diffusing early to the PSEB communities
(Ehret et al. 1972; Westphal 1963), was separately borrowed into proto-Southwest Bantu
(ancestral to Herero, Nyaneka, Kwanyama, etc.), proto-Kavango (ancestral to Kwangali,
Sambiyu, and Gciriku), and Yei (Pfouts 2003). In those regions the knowledge of sheep
would have come from the ancestral Kwadi and possibly early West Kalahari Khoe
groups, after the initial divergence and expansion of the early Khoe communities was
well underway. The distribution of these additional borrowed forms of *gu, as far west
as the lower Kunene, fits in with the proposal that sheep-keeping spread initially
westward with the ancestral Kwadi, in advance of the full establishment of Bantu-
speaking communities in each of these areas, again at a period dating to 2 000 or more
years ago (Pfouts 2003). It also suggests that the Kwadi may have been sheep- rather
than cattle-raisers themselves.
16 SOUTHERN AFRICAN HUMANITIES, VOL. 20, 2008
Notably missing from Figure 2 are old root words diagnostic of the crop-
cultivating side of food production. It is difficult, even with concerted searching,
to identify any such terms. With one notable exception, the words connoting
cultivating activities in Khoe languages tend to be subgroup- or even language-
specific. Most derive from earlier root words not specifically relating to agriculture
or are borrowings from Bantu languages. The single term occurring in more than
one Khoesan subgroup is a proto-Kalahari Khoe root *||hada ‘cultivated field,
garden’ (Vossen 1997). This term appears in a Tuu language, !Xoõ, probably via a
borrowing in recent centuries from a nearby Kalahari Khoe language, possibly
Naro. These citations (see Fig. 2) suggest that, by the proto-Kalahari Khoe and
proto-Gariep Khoekhoe periods, cultivation had probably become part of the
knowledge of Khoe peoples. The inability to reconstruct supplementary terminology
to any of the early periods favours the conclusion, however, that cultivation
remained something known to, but not practiced by the early Kalahari Khoe and
the Gariep Khoekhoe communities.
Farther south and much later in time, the Eastern Cape Khoekhoe did indeed engage
to some extent in cultivation, as literate observers of the later eighteenth century noted
(e.g. Sparrman 1786). Whether these practices had spread to the Eastern Cape
communities not long before, from their Nguni neighbours to the east, or went back
significantly earlier in Cape Khoekhoe history cannot be resolved from the very little
evidence available to us as yet. The complete lack of cultivation amongst the Western
Cape Khoekhoe does not help in answering this question, because its absence there
would have been simply a function of environment. The staples of southern African
agriculture of earlier centuries, sorghum and pearl millet, germinate in hot rainy seasons.
Westward from the Great Fish River, the summer rain regime of the eastern side of
southern Africa gradually yields to a winter rain climate, in which the cultivation of the
tropical African crops was not a viable option. The Western Cape Khoekhoe would
perforce have been solely pastoralist, whether or not their proto-Cape Khoekhoe
ancestors had been so.
Far to the north, some West Kalahari Khoe speakers, living around and north of the
Okavango areas of northern Botswana, may separately have taken up cultivation of a
few crops in the early eras of Kalahari Khoe divergence. The Kxoe language, in particular,
generally borrowed its words for those crops introduced to Africa in the last few centuries
from nearby Bantu languages. In contrast, for the old indigenous African crops, such as
sorghum and Voandzeia subterranea; also called the earthpea or the Bambara groundnut),
Kxoe tends to possess its own unique terms, the origins of which remain to be discovered
(see Köhler 1986 for the detailed lexical evidence and analysis).
A variety of related issues remain to be considered. One key problem brought to the
fore in the meeting at which the original version of this paper was delivered can be
epitomized in the question, ‘did the keeping of sheep spread to Khoesan hunter-gatherers
of southern South Africa, or to Namibia, ahead of the spread of Khoekhoe or Kwadi
populations’ (e.g., Sadr this volume; Fauvelle-Aymar this volume)? The lexical
distributions give no reason to think so, but do not entirely rule out that possibility
EHRET: EARLY LIVESTOCK-RAISERS 17
In the regions of the northern Kalahari and southern Angola, most Zhu dialects have
adopted Khoe *gu for ‘sheep’. Virtual identity, tonally and segmentally, obtains between
the Zhu and Khoe pronunciations, and this is what one expects if Zhu speakers acquired
the word and the knowledge of sheep from Khoe communities since the end of the
A different version of the *gu root appears in the northerly Tuu language. !Xoõ which
has kûu as its term. The !Xoõ devoicing of the voiced stop *g suggests either Namibian
Khoekhoen (as in Traill 1994: 172) or Limpopo Khoekhoe, both of which attest this
particular devoicing, as a potential source language. There is no other evidence of a
Limpopo Khoekhoe presence in Botswana, but the northern fringes of the Gariep
Khoekhoe expansion that brought the early Namibian Khoekhoen speakers to Namibia
might have brushed this area. Another possible solution is that the !Xoõ word is an
adoption from a Kgaladi Sotho dialect of the phonologically regular Sotho reflex, *nku,
of PSEB *-gu ‘sheep’. The phonological constraints of !Xoõ do not allow initial *nk,
which thus regularly to !Xoõ /k/ in Setswana loanwords. The !Xoõ adoption of the
word thus cannot be placed with certainty any earlier than the first Kgalagadi settlement
in the region, a development unlikely to have taken place much before AD 1000.
In the southern regions of the Kalahari and farther south in the western and eastern
Cape, to the extent that we have evidence, a different pattern emerges. The languages
of those areas have coined a number of different unique words for sheep, the derivations
of which remain as yet unknown (e.g. Bleek 1956, Westphal 1966). A pattern of this
kind, in which almost every language has a different term, has two alternative
explanations. One is that numerous names have arisen because the item has been around
for thousands of years, and so different peoples have gradually replaced an original
common name with various later innovated terms. Sheep are too recent an arrival on
the southern African scene for the variability of Tuu names to be explained in this
fashion. The other way this kind of distribution of terms can arise is when the item in
question is something people have some acquaintance with but is not part of their own
cultural practices. In the case of the more southerly Tuu languages, we are left with this
latter solution—that sheep were not raised, at least by the ancestors of any of the Tuu
peoples for which there is direct linguistic evidence. If the animal had been raised
by their ancestors at an earlier era in the past 2000 years , we would expect patterns
similar to those of cow and sheep terms farther north, with the same root words occurring
over contiguous areas or in most of the member dialects/languages of the particular
On the other hand, the data do not entirely rule out the possibility of the spread of
sheep ahead of the Khoekhoe in the southernmost regions of the Western Cape, for the
simple reason that we lack any livestock terms from that region other than those brought
in by the speakers of the Cape Khoekhoe dialects (compiled in Nienaber 1963).
If there was such a prior spread of sheep, the arrival of the early Cape Khoekhoe
cattle pastoralists had to have followed within a very few centuries. The proposed dating
of the Limpopo Khoekhoe interactions with the proto-Southeastern Bantu, to the first
half of the first millennium AD, places the prior divergence of proto-Khoekhoe into its
Limpopo and Cape-Gariep branches at least a few centuries earlier. The first spreads of
Khoekhoe populations east into northern South Africa and south to the Gariep-Vaal
confluence, the events that initiated this era of linguistic divergence, thus may date to
18 SOUTHERN AFRICAN HUMANITIES, VOL. 20, 2008
2000 years ago or perhaps a bit earlier. The dating arguments previously published
(Ehret 1982), based on the degrees of lexical divergence among the Cape-Gariep
languages and dialects, would allow for the further spread of the proto-Cape Khoekhoe
south from the Gariep-Vaal confluence into the Eastern Cape to have taken place as
early as the early first millennium. The evidence used in these arguments, though, is
scant and uneven, and a dating of the spread of the Cape Khoekhoe east and west along
the southern coasts to as late as the early second half of the first millennium, later than
the first sheep-keeping in those regions, would be equally possible. The proposal that
Khoekhoe communities might have arrived in the Western Cape only slightly before
their early encounters with Europeans (e.g. Fauvelle-Aymar this volume) does not seem
probable, however, from these data.
On the archaeological side, the presence of sheep but no cattle bones in the earliest
Eastern and Western Cape sites with domestic animal remains also does not resolve the
debate in favour of the priority of sheep. The reason is straightforward. The Khoekhoe
were clearly strong cattle pastoralists, and the normative pattern in such societies is a
greater willingness to butcher small stock and to part with them in trade. The early
Cape sites with sheep were not pastoral habitation sites of the types we know from
Khoekhoe ethnography. They may have been the sites of pastro-foraging sheep-raisers.
But the more plausible explanation in my view is that these were the sites of hunter-
gatherers who obtained sheep, an animal cattle pastoralists tend to be more willing to
trade, from fully pastoralist neighbouring peoples (see also A.B. Smith this volume).
The problem of archaeological invisibility is a real one in the case of mobile
pastoralists. As David Anthony (2007) has pointed out in the case of the highly mobile,
highly pastoral Yamnaya culture of Ukraine, had these people not fortuitously buried
their chiefs in kurgans along with the material accoutrements of elite status,
archaeologists would have virtually no evidence that anyone lived in the Pontic Steppes
for much of the second half of the fourth millennium BC. The Khoekhoe, as far as we
know, lacked elaborate chiefly burial customs and structures, and their ephemeral and
shifting habitation sites are likely to be no more visible in the archaeological record
than the wagon encampments of the Yamnaya peoples.
GOATS IN KALAHARI KHOE AND KHOEKHOE PASTORAL PURSUITS
The terminology of goats reflects a still different history. The lexical evidence shows
that goats reached Khoe and other Khoesan peoples significantly later in time than
sheep and later also than cattle, and via at least three separate adoptions from Bantu-
speaking societies (Fig. 2). To the west, the Kwadi term khobo-de (Westphal 1966) is a
borrowing of the neighbouring Southwest-Bantu *-kombo. A second Bantu root *m-
pene (Bantu class 9 noun prefix plus stem *-pene) appears in the Khwe subgroup of the
West branch of Kalahari Khoe, in the shape míní. This borrowing probably came
specifically from nearby Yeyi impèné, the only Bantu language in southern Africa to
have this root with this particular meaning. This root word also applies to the goat in
several East African languages; it occurs in Shona (Hannan 1981) and Tsonga (Cuenod
1967), but there it has a different meaning, steenbok.
In most of the Khoe and the other Khoesan languages of the Kalahari as far south as
the Gariep River, still a third Bantu root, *m-buli, gave rise to the words for goat. The
original borrowed shape, found in some Zhu dialects and in the East branch dialects of
EHRET: EARLY LIVESTOCK-RAISERS 19
Kalahari Khoe, was *puli or *puri, depending of the language or dialect involved. The
devoicing of initial proto-Bantu *b requires that the word reached these languages via
a language with a stop devoicing rule. Two possible sources exist, either proto-Sotho,
in which proto-Bantu *m-buli became *pudi, or Limpopo Khoekhoe, whose stop
devoicing rule and lack of prenasal stops would have converted *m-buli to *puri. A
proto-Sotho source for goats is improbable because it would require a delayed spread
of the knowledge of goats to Khoe peoples, only after proto-Sotho’s own devoicing
shift, which likely dates no earlier than the later first millennium AD. In addition, the
proto-Sotho lands are best placed in northeastern modern-day South Africa, several
hundred kilometres east of the probable early Khoe lands.
The more probable answer is that the acquaintance with goats reached the northern
Botswana peoples indirectly from the proto-Southeastern Bantu peoples in the first 500
years AD, via the Limpopo Khoekhoe, who fit both the geographical and linguistic
requirements. The habitation areas of this society in the early first millennium, as already
argued, would have included the areas between Botswana and the proposed early
Southeast-Bantu lands, and the regular devoicing rule and the presence of intervocalic
/r/ but not /l/ in Limpopo Khoekhoe would have converted PSEB *m-buli into the
attested borrowed shape, *puri.
A succession of irregular deviations from the original pronunciation of *puri appear
to track the successive stages of its spread. The first change was of original *puri, found
in the East subgroup of Kalahari Khoe, to *piri in the Cape and Gariep Khoekhoe
languages. This latter pronunciation occurred as far south as the Cape Khoekhoe dialects
of the Eastern Cape, although a distinctive term for the animal was innovated by the
Western Cape Khoekhoe (Nienaber 1963: 224; see Fig. 2). It also travelled to Namibia
with the expansion of the ancestral Namibia Khoekhoen speakers to that region. The
same pronunciation diffused west, too, to the ancestors of the Naro and the G|wi and
G|| ana of the Central Kalahari.
A further change of *piri to *pari, then took place in an undetermined language of
that region. This tertiary shape spread north and south, persevering today in two languages
located at the ends of a north-south axis running through the Central Kalahari—as páli
in the northerly Tuu language !Xoõ (Traill 1994) and as párí in Ju|’hoansi (Dickens
1994). An alternative course of sound change, from *puri to *pari can be ruled out
because the vowel-consonant combination *uri is frequent in both languages and
therefore would have been preserved. The sequence *iri, in contrast, is not normally
present in Ju|’hoansi and very rare in !Xoõ and so would be susceptible to replacement
by the much more familiar Khoesan sequence *ari.
The most parsimonious accounting of these distributions implies the following history
1. The initial spread of the term in the shape *puri or *puli, and presumably
of the animal, from the proto-Southeast Bantu via the Limpopo Khoekhoe
westward to the proto-East Kalahari Khoe people in eastern Botswana;
2. The subsequent spread of the term and animal from the early East Kalahari
Khoe to early Cape-Gariep Khoekhoe speakers, with a change in shape of
the word to *piri, and from them to Central Kalahari hunters;
3. The still later spread of the term north and south in the Central Kalahari,
with a further distortion of *piri to *pari or *pali.
20 SOUTHERN AFRICAN HUMANITIES, VOL. 20, 2008
The adoption of goats may have taken place early enough for the animal to have
accompanied the proto-Cape-Gariep Khoekhoe in their initial southward expansion
toward the southern Cape regions. More probably, goats diffused later via the proto-
Gariep Khoe to the Eastern Cape Khoekhoe and then from them westward to the Western
Cape communities. Badenhorst’s (2006) proposal that the Khoekhoen used goats to
herd sheep makes good sense of the low frequency of goats in the archaeological sites
and among many Khoekhoe pastoralists as late as the era of written documentation.
What the linguistic evidence implies in addition is that this practice would have had a
different source than the raising of sheep and cattle among the Khoekhoe and would
have reached them after sheep and cattle herding were already established economic
THE WIDER DIFFUSION OF CATTLE KNOWLEDGE
The knowledge of cattle outside the Khoe group followed still a different historical
course from that of either sheep or goats. The words for cow both in the Zhu group and
in the Tuu languages of the southern Kalahari derive from a single root of a distinct
original shape *gumi, derivable from the Bantu root word *ngombe. The distribution
of the term implies the diffusion of the knowledge of the cow southward through the
Central Kalahari without any Khoekhoe agency in the process at all. Zhu and Tuu
phonological constraints would require the conversion of a Bantu shape *ngombe to a
Khoesan shape *gome (since *ng and *mb were originally lacking in Khoesan
languages). The actually attested *gumi, with raising of expected *o to *u and expected
*e to *i, thus presumes the former existence of a still unidentified intermediate source
language in which the vowel raising took place. The overall import of this evidence is
that the knowledge of cows became established in the Zhu regions from a different
source area north of the Kalahari than did cattle amongst the proto-Khoekoe.
From the early Zhu the knowledge of cattle spread at a later period apparently directly
southward to the Tuu peoples of the southern Kalahari Khoesan. The fact that the
settlement areas of the Kalahari Khoe peoples of the north-central Kalahari breaks the
continuity of the distribution of *gumi indicates that this term and the knowledge it
represents diffused southward ahead of the spread of Kalahari Khoe from the east into
these regions, although that conclusion leaves unanswered the question of just when
the Kalahari Khoe spread into these areas took place.
CERAMIC TECHNOLOGY IN KHOE HISTORY
One further issue of early Khoekhoe and Khoe history to take up, important especially
because of its usually greater visibility in the archaeology than sheep, goat, or cattle
bones, is the technology of pottery making. The available lexical evidence for the history
of ceramics consists primarily of nouns for pot—the collectors of dictionaries commonly,
although not always, having neglected to collect verbs for the processes of pot-making
or words for related features, such as temper.
Pottery among Khoe peoples goes back to at least the proto-Khoe period, evidenced
by the proto-Khoe root word, *su, for such vessels (Fig. 2). Its introduction (or its
independent invention) therefore took place as early in time as, or only shortly after, the
introduction of sheep to southern Africa, and it developed among the same peoples
who early took up sheep-raising. This term spread wherever the Khoekhoe descendants
EHRET: EARLY LIVESTOCK-RAISERS 21
of the proto-Khoe spread, revealing the continuous association of Khoe people in general,
and of Khoekhoe in particular, from the proto-Khoe period onward, with the fashioning
of pots. Ceramics should be a recurrent feature of the archaeology of their sites.
Still fragmentary language evidence suggests, however, that the early pot makers of
southern Africa included others than just the Khoe peoples. We can provisionally identify
a lexical distribution in the central Kalahari, paralleling the distributions of *gumi for
cow and the *pari term for goat. Again the link is from the Zhu group to !Xoõ, with the
terms for a ceramic pot in the two languages being, respectively, n ànì and ’ nàna.
This pairing of terms points to the central Kalahari as an additional area of the early
spread of pot making. We will need further lexical evidence, however, if we are more
fully to explore what the linguistic evidence might imply about the directions of spread
of potting technology in this region, the sources of such a development, and its links to
potting among the early Khoe groups. Sadr and Sampson’s (2006) examination of early
Southern Africa pottery is an intriguing start from the archaeological side in tackling
this issue. If Andrew Smith’s (this volume) generalizations about forager pottery hold,
we might expect this spread of pottery technology to be attested by vegetal-tempered
vessels with no sign of elaborate burnishing or sophisticated firing controls.
On a different note, the parallels between the distributions of this pair of pot terms
and those of *pari and *gumi for goats and cows suggests a further line of investigation
that archaeologists might wish to pursue—the possibility of an early, but ultimately
ephemeral, extension of cattle- and goat-raising by Zhu-related people into the central
Kalahari during a period of wetter climate. If such a population movement took place,
did it precede, follow, or parallel in time the spread of West Kalahari Khoe speakers,
proposed above to have been sheep-raisers, into some of the same areas?
LINGUISTIC HISTORICAL PROPOSALS
The recasting of Khoe group linguistic history, allowed by the recent, solidly grounded
new work in the field, offers a variety of old and new implications for the archaeology
of southern Africa.
1. The historical inferences allowed by the far more detailed and systematic
linguistic information we now have for the Khoe group accord in many
respects with the earlier views (Ehret 1982; Elphick 1977) on the history
of Khoekhoe and Kwadi pastoralist expansions, although they require major
modifications and add numerous new elements helpful in understanding
the earliest stages.
2. It can now be suggested from the linguistic evidence, as has been argued
from the archaeology, that the raising of sheep as an adjunct to gathering
and hunting was probably established among the proto-Kwadi-Khoe and/
or the proto-Khoe prior to the development of full cattle and sheep
3. Sheep-keeping spread prior to the arrival of Bantu speakers all across the
northern Kalahari zone and diffused to the proto-Southeastern Bantu from
a different source from, and probably ahead of, the Limpopo Khoekhoe
4. The available language evidence does not offer any explicit indications
that sheep-raising preceded the expansion of cattle-and-sheep pastoralist
22 SOUTHERN AFRICAN HUMANITIES, VOL. 20, 2008
Cape Khoekhoe communities into the far southern coastlands of southern
Africa. The chronological indications of the evidence are consistent with
dating the initial Khoekhoe settlement period in the southern Cape areas
from as early as the first century or two AD (Ehret 1982), contemporary
with the first appearance of sheep and ceramics in the archaeology.
5. The degree of linguistic differentiation among the Cape Khoekhoe dialects
is not consistent, however, with dates for an arrival of the Cape Khoekhoe
in the Eastern Cape significantly more recently (see A.B. Smith, Sadr and
Fauvelle-Aymar, all this volume, for contrasting arguments from the
archaeology). The depth of the dialect differences in the late seventeenth
century across the 800–900 km east-west chain of Cape Khoekhoe
communities, noted by early literate observers—with speakers of the
dialects at the far east and west ends of the chain hard put to understand
each other—commonly betokens a period of dialect divergence on the
rough order of 1000 years; and longer, if little grammatical or phonological
change has taken place, as is evident from the recorded Cape Khoekhoe
lexicons. These features, in conjunction with the indications of the uneven
lexical evidence from western and eastern Cape Khoekhoe dialects, allow
a large range of possible times during which the full extension of Khoekhoe
populations between the Eastern and the far Western Cape could have
taken place—from the early first millennium AD, up to the early second
half of the first millennium (Ehret 1982 presents arguments for these dating
6. The proto-Khoekhoe generic term for cow, *gomo, very probably did
derive, at least indirectly, from the same root seen in the old Eastern Bantu
root *-gombe. If it were a direct borrowing from Bantu, however, this
word should have had the proto-Khoekhoe shape *gomi or *gome, rather
than *gomo. It probably entered the proto-Khoekhoe lexicon indirectly
via a non-Bantu language.
7. A separate spread of a different term for cow, *gumi, passed through the
central Kalahari. Its usage in those regions may reflect a very early
southward spread, ultimately ephemeral, of cattle-keeping into those areas,
possibly by Zhu-related people. Like proto-Khoekhoe *gomo, the term
*gumi seems ultimately to derive from the same root as Bantu *-gombe,
but via a different intermediate source. This conclusion reminds us that
there is still much to learn about the history and archaeology of, and the
peoples involved in, the spread of cattle-raising between East Africa and
the Zambezi River.
88. The Limpopo Khoekhoe cattle pastoralists had a long-lasting presence in
the first millennium AD in northern and northeastern South Africa. Their
influences during the first millennium AD directly spread cattle-raising
and milking to the proto-Southeastern Bantu (ancestral in language to the
Chopi, GiTonga, Tsonga, Sotho, Venda, and Nguni peoples).
89. The Limpopo Khoekhoe were also the probable intermediaries in the spread
of goats to the Cape-Gariep Khoekhoe to the south and to other peoples in
EHRET: EARLY LIVESTOCK-RAISERS 23
10. Khoekhoe communities continued to be a historical factor into the period
AD 500–1000 in the northern provinces of modern-day South Africa. The
Limpopo Khoekhoe contributed strongly to the evolution of the proto-
Sotho society, which resided, probably, in the farther northeastern parts of
South Africa during that era, and the descendants of the Limpopo Khoekhoe
likely constituted a major element in the demographic ancestry of the proto-
Sotho. Either the Limpopo Khoekhoe or a related Khoekhoe people made
an equally strong contribution to the emergence of the proto-Nguni as a
distinct people, probably in the eastern Highveld, also before AD 1000
(for this dating see Ownby 1985).
11. Ceramic technology began to be practiced by Khoesan peoples before the
arrival of Bantu speakers. The proto-Khoe had pots and may have been
the first practitioners of this technology; and their descendant groups, the
Khoekhoe in particular, continued to carry on this kind of artisanship into
later eras. The language evidence also supports the archaeology in
identifying cases where non-Khoe may very early have taken up potting,
in particular in the central Kalahari. The non-traceability of any of the
early Khoesan terms for pot, at least as yet, to outside sources is consistent
with proposals for a separate invention of ceramic technology in southern
Africa in the first millennium BC (Sadr & Sampson 2006), but it does not
at all rule out an introduction of the technology from outside the region
ARCHAEOLOGY AND LANGUAGE
These linguistics-based proposals about the early locations and expansions of the Kwadi
and Khoe peoples raise a number of suggestive possibilities for the archaeological
correlations of Khoe and Khoekhoe populations. From the linguistic side, the areas of
highest probability as the original country of the proto-Khoe and their immediate
descendant communities, the proto-Kalahari Khoe and the proto-Khoekhoe, include
northeastern Botswana and western Zimbabwe. Less direct and less compelling evidence
suggests that the sister ancestral Kwadi society lived in or near parts of this same overall
The most parsimonious history of proto-Khoekhoe divergence from the proto-Kalahari
Khoe postulates a single southward or southeastward extension into areas west and
south of the Shashi-Limpopo confluence. From there the proto-Khoekhoe quickly
expanded more widely southward, giving rise in northern South Africa to the Limpopo
Khoekhoe and, in the areas around and north of the Gariep-Vaal confluence, to the
proto-Cape-Gariep Khoekhoe communities, who spoke a proto-Khoekhoe dialect
ancestral to the widespread Cape Khoekhoe and !Ora and Namibian Khoekhoe languages
of later times.
The proto-Khoe descendant groups who remained behind in northeastern Botswana
evolved into the proto-Kalahari Khoe community. A subsequent single westward
expansion of proto-Kalahari Khoe speakers into central and western parts of northern
Botswana equally parsimoniously accounts for the divergence of the proto-Kalahari
Khoe into the proto-West and proto-East Kalahari Khoe communities. The proto-West
Kalahari Khoe emerged out of those groups who moved west, while the proto-East
24 SOUTHERN AFRICAN HUMANITIES, VOL. 20, 2008
Kalahari Khoe derived from the proto-Kalahari Khoe speakers who remained behind in
the original proto-Khoe lands in northeastern Botwana and adjacent far western
Three features of the archaeology of those regions have close bearing geographically
and in time with this linguistically postulated history. One feature, well represented in
recent literature, has to do with rock art. The other two relate to proposed ethnic links of
Benjamin Smith and his colleagues have identified a non-entoptic geometric rock art
tradition of southern Africa with the Khoekhoe herding societies (e.g. Eastwood 2003;
Hall & Smith 2000; Smith & Ouzman 2004). This art tradition occurs in a wide band
through the Limpopo Basin and northern South Africa southward into the interior Eastern
and Western Cape provinces and also westward through areas around the Vaal and
Gariep Rivers, with Limpopo Basin occurrences dating to before 1500 years ago (Smith
& Ouzman 2004). These distributions match up well with the proposed early Khoekhoe
areas, with the subsequent spread of Khoekhoe speakers into the Cape regions, and
with the posited westward spread of the ancestral Namibian Khoekhoen speakers along
the Gariep River toward southern Namibia. The fact that similar geometric designs in
central Namibia appear to be historically recent (Smith & Ouzman 2004: 512) is in
keeping with the linguistic arguments that Namibian Khoekhoen speakers arrived in
these regions from the south only some time after AD 1000. The contrasting older
occurrence of this artistic tradition in northern Namibia and northern Botswana suggests,
however, that its southern Africa cultural linkage was wider than just the Khoekhoe
herders—that it was made by early Kwadi as well, and that it may therefore go back to
the origin period of the Kwadi-Khoe group as a whole.
In the ceramic record, most immediately striking are the detailed distributional parallels
between Bambata pottery and the proposed early lands of the Khoe branch of Kwadi-
Khoe. Bambata pottery finds extend from western Zimbabwe southward across the
Limpopo Basin and westward from Zimbabwe through the Makgadikgadi Pans to Toteng,
covering the same overall span of territories proposed to form the areas of the earliest
expansions of Kalahari Khoe and Khoekhoe speakers. A further match of Bambata
with the linguistic mappings is its suggested subdivision into western and eastern
varieties, with potsherds in the Limpopo Basin including more incised decoration than
in the Kalahari Basin, but lacking the charcoal temper common in the latter area (Sadr
2008). The distribution of the eastern variety closely fits the proposed early geographical
spreads of the proto-Khoekhoe to the southeast. The western variety occurs in those
areas where the early spread and divergence of proto-Kalahari into proto-East and proto-
West Kalahari Khoe is proposed to have taken place. The sites with Bambata ware
often show the presence of cattle as well as sheep, unlike the ripple-rim and southern
Cape sites, the early occurrences of which reveal only sheep.
The earliest dates so far for Bambata ware, to the last three or four centuries BC, are
from eastern sites. If these finds (with not the most reliable dates, as Sadr 2008 puts it)
are not simply artefacts of site selection, their locations and dates are in keeping with an
adoption of livestock-raising by Khoe speakers in northeastern Botswana and western
Zimbabwe in the earlier second half of the first millennium BC and the spread of the
earliest emerging proto-Khoekhoe, with cattle, southeastward into the Limpopo Basin
before the end of the millennium. It is perhaps time to look beyond a broad connection
EHRET: EARLY LIVESTOCK-RAISERS 25
of Bambata with LSA herders (Beaumont et al. 1995; Reid et al. 1998; Robbins et al.
2005, 2008; Sadr & Sampson 2006) to consider whether Bambata pottery may more
specifically have been the wares of the proto-Khoe or their early proto-Kalahari and
A second proposed set of ceramic correlations for LSA sheep-herding peoples is
with ripple-rim ware, found in northern Namibia and in the Limpopo Basin. On the
basis of a suite of shared technical characteristics, Andrew Smith (this volume) sees a
wider connection of ripple-rim ware with the ceramics, also associated with sheep bones,
of early first millennium sites in the western and southern Cape regions. Like ripple-
rim ware, the Cape pottery is thin-walled, mineral-tempered, and high-fired. The earliest
dated Cape pottery, of around 2000 years ago at Die Kelders, has a surface treatment
not dissimilar, as Andrew Smith (this volume) puts it, to that used for the ripple-rim
wares of northern Namibia and the Limpopo Basin. Other early Cape pottery, from
slightly later sites, did not apply, or no longer applied, this technique, although it retained
other features of the Die Kelders pots.
If ripple-rim ware is indeed ethnically linked, its distribution in the Limpopo Basin
and in northern Namibia has a broader geographical fit than does Bambata with Kwadi-
Khoe speakers: it appears both in the probable early Khoekhoe-speaking areas and in
the proposed areas of the spread of the ancestral Kwadi west to the Atlantic. The early
Cape ceramics, if they do belong to the same tradition as the ripple-rim pottery, could
be the reflections of a further early, rapid spread of Khoe-related herding people to the
southern coastal regions.
Within this wider geographical context, Bambata pottery manifests as a regional
phenomenon. It is thin-walled like other LSA herder ceramics and thus, in this respect,
is unlike the thicker EIA pottery of Bantu speakers; but it has more decoration and also
different decorative elements from those of other LSA herder pottery. Its distribution
fits closely with the proposed early proto-Khoe and proto-Khoekhoe locations. If it was
a marker of Khoekhoe presence, however, its absence in the areas farther south with
Khoekhoe-speaking populations needs explaining. Did the same people fashion less-
decorated or undecorated and sometimes ripple-rim, thin-wall pottery along with the
thin-wall Bambata ware, often found at the same sites, and did those descendant groups
who spread southward simply favour the less decorated style? Or might Bambata have
been a regional development in fashion that spread in the northern areas after Khoekhoe
expansion toward the southern coastlands was already underway?
The alternative attribution of Bambata in the literature is as the ceramic tradition of a
distinct people living on the peripheries of the proto-Kalahari and proto-Khoekhoe
territories. If so, who were these people and what happened to their society? One possible
candidate would be the early ancestral society of the Bantu-speaking Yeyi, who live
today at the western edge of the Bambata distribution. Their language forms a subgroup
of its own, rather distinct from the rest of the southern African Bantu languages. On the
other hand, Yeyi does share an odd item or two with Shona, notably the basic vocabulary
word for ‘head’. If Yeyi did originate as a very early offshoot of the Shona group, this
might accord with Huffman’s (2005 and elsewhere) strong avowal of Bambata as a
facies of the Kalundu tradition, to which Shona-speakers’ pottery belongs.
One problem with attributing Bambata to a Bantu-speaking society, however, is the
lack of any indication whatsoever in the extant Khoekhoe linguistic evidence, either in
26 SOUTHERN AFRICAN HUMANITIES, VOL. 20, 2008
the form of loanwords or in phonology or grammar, for any direct contacts of the proto-
Khoe or the proto-Khoekhoe with Bantu speakers and, except for the word *gomo
‘cow’, of any indirect contacts either. This is in fact a more general problem. There are
indeed Bantu loanwords indicative of direct interactions with particular Khoe languages
of the northern Kalahari belt. But all of them reflect contacts of later eras. None of this
evidence traces back even to the proto-West or proto-East Kalahari Khoe periods, let
alone to proto-Khoekhoe or to the earlier proto-Khoe era. If Bambata was indeed a
pottery made by Bantu-speakers, then the ancestors of the speakers of the Khoekhoe
languages of recent centuries must already have spread southward out of the areas in
which Bambata ware occurs, before 2000 years ago.
The evidence on all sides that Khoekhoe speakers had moved well southwards in
South Africa by the early first millennium AD brings to the fore a pervasive question of
recent research. Did the early southward movement of Khoekhoe speakers continue
rapidly onward into the southern Cape regions 2 000 years ago, or might there have
been two expansions of herders in the southern regions, an early spread of sheep-raisers
overlain by a later spread of Khoekhoe cattle-raisers? One archaeological proposal is
that Khoekhoe communities arrived in the Western Cape concurrent with the shift from
spouted to lugged ceramic wares around 1000 years ago (Sadr 2003; see also Fauvelle-
Aymar this volume). As noted above, the degree of linguistic divergence between the
western and eastern Cape Khoekhoe dialects (for which see Ehret 1982) in the
seventeenth and eighteenth centuries was already great, suggestive of an initial spreading
out of the Cape Khoekhoe dialect chain more than 1000 years before that time. That
evidence does not preclude the possibility of a secondary spread of cattle-raising
Khoekhoe overlaying an earlier spread of sheep-raisers, but if so that expansion should
date at the latest several centuries before AD 1000. The shift from spouted to lugged
wares seems too late in time to be its material manifestation. Andrew Smith’s (this
volume) argument that the continuities in the temper, manner of construction and finish
reveal a change in ceramic function or fashion, rather than different peoples on the
landscape, seems to this writer the more likely explanation.
PASTORALIST IDENTITIES AND LONG-TERM HISTORICAL CONSEQUENCES
In assessing the possible archaeological correlations, the cultural depth of the Khoekhoe
identification with pastoral pursuits, beginning in the proto-Khoekhoe period, should
not be underestimated. The idea that pastoralism amongst them was an on-and-off
thing, with particular Khoekhoe who lost their cattle becoming no different from the
many San hunter-gatherer groups, has always seemed to this scholar a misunderstanding
of the historical evidence and the anthropology (see Barnard this volume). To call the
members of one’s society who were poor in livestock by the same term, san, as hunter-
gatherer people who had no livestock in the first place, was an economic statement,
not a cultural one. Khoekhoe societies organized themselves around patrilineages and
patriclans, to which people belonged by birth and to which and from which
responsibilities and social identities flowed. The connections and obligations of this
relationship did not just dissolve because of economic misfortune. The entire lack
of such institutions among the hunter-gatherer peoples of southern Africa shows
we are dealing with two distinct socio-cultural worlds, with distinctive ideologies to
EHRET: EARLY LIVESTOCK-RAISERS 27
The proto-Khoe along with one branch of their descendants, the Kalahari Khoe, may
have practiced an economy transitional between these two distinct worlds. On the one
hand, the proto-Khoe lexicon of livestock-raising, with its specialized terminology of
milking and butter making, makes sense only if at least some of the members of the
proto-Khoe raised sufficient numbers of sheep to form viable herds (see A.B. Smith
this volume, on the issue of viability; also A.B. Smith 1990). A transmission of sheep
through forager reciprocity relations might account for some occurrences of sheep bones
in otherwise forager sites, but only if nearby there lived communities who raised sufficient
numbers of sheep to perpetuate breeding and herd growth. On the other hand, the proto-
Khoe likely preserved the older Khoesan pattern, still retained by modern-day Kalahari
Khoe speakers, of bilateral descent. Transitional cultural attitudes more accepting of a
shift back toward foraging may well also have persisted in this intermediate cultural
world. But the full adoption of cattle and sheep pastoralism by the proto-Khoekhoe as
the core economic value, along with their reorganization into segmentary kinship
systems, projects a fundamentally different worldview.
These divergences surface overtly in the manners in which the pastoral Cape Khoekhoe
and the hunter-gatherer Tuu societies responded to the dire consequences of Boer and
Baster expansion in the eighteenth century. In the wake of depopulating epidemics and
the loss of livestock and grazing lands, the Khoekhoe frequently attached themselves to
the expanding frontier groups. In doing so, they drew on an older customary expectation
deeply rooted in their history, seeking to rebuild their herds by acting as clients and
herdsmen for the new kind of chiefs. In stark contrast, the independent hunter-gatherer
communities of the interior, little affected by the epidemics and with their social systems
intact, envisioned no such option. They fought back against the encroachment of Boers
and their Khoekhoe clients on their lands and against the killing of their wild game
resources—and for their efforts suffered genocidal attacks and the capture and enserfment
(so-called ‘apprenticing’) of their children over the course of the second half of the
The contrastive early histories of expansion of the two branches of the Khoe group
further point up the differing expansive capacities of a pastro-foraging economy (to use
the term of Kuper & Kröpelin 2006) and a pastoral one. The early Kalahari Khoe,
consisting, I would propose, of pastro-foraging communities, did expand their territories
widely in the early first millennium AD, but mostly within the confines of the northern
half of Botswana. The pastoralist and cattle-raising Khoekhoe, however, spread their
languages and economy to the farthest corners of southern Africa. The Khoekhoe
economy had radically greater subsistence productivity per unit of land than pure hunting-
gathering and significantly greater productivity than pastro-foraging. Only this
productive advantage can adequately explain the relative facility with which Khoekhoe
communities expanded over so great a set of territories.
It is an advantage implied as well by the linguistic evidence for the Khoekhoe demic
contribution to the formative stages of the proto-Sotho and proto-Nguni societies. The
Khoekhoe economy, these data indicate, supported a population in northern South Africa
in the first millennium AD comparable in magnitude to that of the neighbouring proto-
Sotho and proto-Nguni food-producing peoples with which they so closely interacted
and into which their descendants eventually merged. The archaeological visibility of
the Khoekhoe pastoralists is likely to be extremely low (for the reasons A.B. Smith this
28 SOUTHERN AFRICAN HUMANITIES, VOL. 20, 2008
volume cites; cf. Fauvelle-Aymar et al. 2006). But the language evidence reveals that
they were once major components of the southern Africa human landscape and that
they contributed in large measure to the ancestry of the modern-day Bantu-speaking
Anthony, D. 2007. The horse, the wheel, and language: how Bronze-Age riders from the Eurasian steppes
shaped the modern world. Princeton: Princeton University Press.
Badenhorst, S. 2006. Goats (Capra hircus), the Khoekhoen and pastoralism: current evidence from southern
Africa. African Archaeological Review 23 (3/4): 45–53.
Barnard, A. 1985. A Nharo wordlist with notes on grammar. Durban: Department of African Studies,
University of Natal.
Barnard, A. 2008. Ethnographic analogy and the reconstruction of early Khoekhoe society. Southern African
Humanities 20: 61–75.
Beach, D.M. 1938. Phonetics of the Hottentot language. Cambridge: W. Heffer & Sons, Ltd.
Beaumont, P.B., Smith, A.B. & Vogel, J.C. 1995. Before the Einiqua: the archaeology of the frontier zone.
In: Smith, A.B., ed., Einiqualand: studies of the Orange River frontier. Cape Town: University
of Cape Town Press, pp. 236–64.
Bleek, D. 1956. A Bushman dictionary. New Haven: American Oriental Society.
Cuenod, R. 1967. Tsonga-English dictionary. Johannesburg: Swiss Mission in South Africa.
Dickens, P. 1994. English-Ju/’hoan; Ju/’hoan-English Dictionary. Cologne: Rüdiger Köppe Verlag.
Eastwood, E.B. 2003. A cross-cultural motif in San, Khoekhoe and Northern Sotho rock paintings of the
central Limpopo basin, southern Africa. South African Archaeological Bulletin 58: 14–26.
Ehret, C. 1982. The first spread of food production to southern Africa. In: Ehret, C. & Posnansky, M., eds,
The archaeological and linguistic reconstruction of African history. Berkeley, Los Angeles:
University of California Press, pp. 158–81.
Ehret, C. 2005. Writing African history from linguistic evidence. In: Philips, J., ed., Writing African history.
Rochester: University of Rochester Press, pp. 86–111.
Ehret, C., Bink, M., Ginindza, T., Gottschalk, E., Hall, B., Hlatshwayo, M., Johnson, D. & Pouwels, R.L.
1972. Outlining southern African history: a reconsideration, A.D. 100–1500. Ufahamu 3 (2):
Elphick, R. 1977. Kraal and castle: Khoikhoi and the founding of white South Africa. New Haven: Yale
Fauvelle-Aymar, F.-X. 2008. Against the ‘Khoisan paradigm’ in the interpretation of Khoekhoe origins and
history: A re-evaluation of Khoekhoe pastoral traditions. Southern African Humanities 20:
Fauvelle-Aymar, F.-X., Sadr, K., Bon, F. & Gronenborn, D. 2006. The visibility and invisibility of herders’
kraals in southern Africa, with reference to a possible early contact period Khoekhoe kraal at
KFS 5, Western Cape. Journal of African Archaeology 4 (2): 253–71.
Güldemann, T. 2004. Reconstruction through ‘de-construction’: the marking of person, gender, and number
in the Khoe family and Kwadi. Diachronica 21 (2): 251–306.
Güldemann, T. 2006. Structural isoglosses between Khoekhoe and Tuu: the Cape as a linguistic area. In:
Matras, Y., McMahon, A. & Vincent, N., eds, Linguistic areas: convergence in historical and
typological perspective. Basingstoke: Palgrave MacMillan, pp. 99–134.
Güldemann, T. In press. Person-gender-number marking from proto-Kwadi-Khoe to its descendents: a
rejoinder with particular reference to language contact. In: König, C. & Vossen, R., eds,
Festschrift for Bernd Heine. London: Routledge.
Haacke, W.H.G. & Eiseb, E. 2002. A Khoekhoegowab dictionary with an English-Khoekhoegowab index.
Windhoek: Gamsberg Macmillan.
Haacke, W.H.G. 2008. Linguistic hypotheses on the origin of Namibian Khoekhoe speakers. Southern African
Humanities 20: 00–00.
Hall, S. & Smith, B.W. 2000. Empowering places: rock shelters and ritual control in farmer-forager
interactions in the Northern Province. South African Archaeological Society Goodwin Series 8:
Hannan, M. 1981. A standard Shona dictionary. Salisbury: College Press.
Heine, B. & König, C. 2008. What can linguistics tell us about early Khoekhoe history? Southern African
Humanities 20: 00–00.
Huffman, T.N. 2005. The stylistic origin of Bambata and the spread of mixed farming in southern Africa.
Southern African Humanities 17: 57–79.
EHRET: EARLY LIVESTOCK-RAISERS 29
Kilian-Hatz, C. 2003. Khwe dictionary. Cologne: Rüdiger Köppe.
Köhler, O. 1986. Allgemeine und Sprachliche bemerkungen zum Feldbau nach Oraltexten der Kxoe-
Buschleute. Sprache und Geschichte in Afrika 7 (1): 205–72.
Kolbe, P. 1727. Naaukeurige en Uitvoerige Beschrijving van de Kaap de Goede Hoope. Amsterdam: B.
Kriel, T.J. 1983. Woordeboek: Noord-Sotho–Afrikaans, Afrikaans–Noord-Sotho. Pretoria: J. L. van Schaik.
Kuper, R. & Kröpelin, S. 2006. Climate-controlled Holocene occupation in the Sahara: motor of Africa’s
evolution. Science 313 (11 August): 803–7.
Laumann, D. 1995. The Chopi and GiTonga. Unpublished UCLA seminar paper.
Mabile, A. 1976. Sesotho-English Dictionary. Morija: Morija Sesotho Book Depot.
Nienaber, G. S. 1963. Hottentots. Pretoria: J. L. van Schaik Beperk.
Ownby, C. P. 1985. Early Nguni history: the linguistic evidence and its correlation with archaeology and
oral tradition. PhD dissertation, University of California, Los Angeles. Ann Arbor, MI: University
Pfouts, A.M. 2003. Economy and society in northern Namibia 500 BCE to 1800 CE: a linguistic approach.
Ph.D. dissertation, University of California, Los Angeles.
Reid, D.A.M., Sadr, K. & Hanson-James, N. 1998. Herding traditions. In: Lane, P., Reid, D.A.M. & Segobye,
A., eds, Ditswa mmung: the archaeology of Botswana. Gaborone: Pula Press and the Botswana
Society, pp. 81–100
Robbins, L. H., Campbell, A. C., Murphy, M. L., Brook, G. A., Srivastava, P. & Badenhorst, S. 2005. The
advent of herding in southern Africa: early AMS dates on domestic livestock from the Kalahari
Desert. Current Anthropology 46 (4): 671–7.
Robbins, L. H., Campbell, A. C., Murphy, M. L., Brook, G. A., Liang, F., Skaggs, S. A., Mabuse, A. A.,
Srivastava, P. & Badenhorst, S. 2008. Recent archaeological research at Toteng, Botswana:
early domesticated livestock in the Kalahari. Journal of African Archaeology 6 (1): 131–49.
Sadr, K. 2003. The Neolithic of southern Africa. Journal of African History 44:195–209.
Sadr, K. 2008. An ageless view of the first millennium AD southern African ceramics. Journal of African
Archaeology 6 (1): 235–52.
Sadr, K. 2008. Invisible herders? The archaeology of Khoekhoe pastoralists. Southern African Humanities
Sadr, K. & Sampson, C.G. 2006. Through thick and thin: early pottery in southern Africa. Journal of African
Archaeology 4 (2): 235–52.
Sapir, E. 1916. Time perspective in aboriginal American culture: a study in method. Ottawa: Government
Sparrman, A. 1786. A voyage to the Cape of Good Hope. London: Robinson.
Smith, A.B. 1990. On becoming herders: Khoikhoi and San ethnicity in southern Africa. African Studies 29
Smith, A.B. 2008. Pastoral origins at the Cape, South Africa: influences and arguments. Southern African
Humanities 20: 49–60.
Smith, B.W. & Ouzman, S. 2004. Taking stock: identifying Khoekhoen herder rock art in southern Africa.
Current Anthropology 45 (4): 499–515.
Tanaka, J. 1978. A San vocabulary of the central Kalahari: G?ana and G|wi dialects. Tokyo: ILCAA.
Traill, A. 1994. A !Xoõ Dictionary. Cologne: Rüdiger Köppe Verlag.
Valentyn, F. 1726. Beschrijvinge van de Kaap der Goede Hoope. Dordrecht: van Braam.
Visser, H. 2001. Naro dictionary. D’Kar: Naro Language Project.
Vossen, R. 1997. Die Khoe-Sprachen: ein Beitrag zur Erforschung der Sprachgeschichte Afrikas. Cologne:
Rüdiger Köppe Verlag.
Westphal, E.O.J. 1963. The linguistic prehistory of southern Africa: Bush, Kwadi, Hottentot and Bantu
linguistic relationships. Africa 33: 237–65.
Westphal, E.O.J.1966. Linguistic research in S.W.A. and Angola. In: Südwestafrikanische Wissenschaftliche
Gesellschaft, eds, Die Ethnischen Gruppen in Südwestafrika. Windhoek: South West Africa
Scientific Society, pp. 125–44.
Wilson, M., & Thompson, L. 1983. A history of South Africa to 1870. London: Croom Helm.