Africa over the last 12000 years: how we

can interpret the interface of archaeology

and linguistics?

This chapter has been prepared for a volume edited by François-Xavier Fauvelle on Africa has been
translated into French and is currently being copy-edited.

[NOT FOR CITATION WITHOUT REFERENCE TO THE AUTHOR]

Roger Blench
McDonald Institute for Archaeological Research
University of Cambridge
Correspondence to:
8, Guest Road
Cambridge CB1 2AL
United Kingdom
Voice/ Ans (00-44)-(0)1223-560687
Mobile worldwide (00-44)-(0)7847-495590
E-mail rogerblench@yahoo.co.uk
http://www.rogerblench.info/RBOP.htm

This version, Cambridge, 21 October 2017

Roger Blench Africa: the last 12,000 years. Draft

TABLE OF CONTENTS

1. Introduction ................................................................................................................................................. 1
2. Archaeology, genetics and their relevance ................................................................................................ 2
2.1 Overview 2
2.2 Reconstructing palaeo-climate 3
3. African language phyla............................................................................................................................... 3
3.1 Nilo-Saharan 3
3.1.1 The languages.................................................................................................................................. 3
3.1.2 Archaeological interpretation .......................................................................................................... 5
3.1.3 Central Sudanic ............................................................................................................................... 7
3.1.4 Songhay ........................................................................................................................................... 8
3.2 Niger-Congo 11
3.2.1 Overview ....................................................................................................................................... 11
3.2.2 Kordofanian................................................................................................................................... 13
3.2.3 Mande............................................................................................................................................ 14
3.2.4 Benue-Congo, Bantoid and Bantu ................................................................................................. 14
3.3 Afroasiatic 16
3.3.1 Overview ....................................................................................................................................... 16
3.3.2 Omotic and Cushitic ...................................................................................................................... 18
3.3.3 Berber and Guanche ...................................................................................................................... 19
3.4 Khoisan 20
3.5 Austronesian 23
3.6 Unclassified languages, marginal peoples and their significance 23
4. Synthesis..................................................................................................................................................... 24
References ...................................................................................................................................................... 25

TABLES

Table 1. Numbers of African languages by phylum 1

Table 2. Data categories, their attributes and values in reconstructing African prehistory 2
Table 3. Proto-Songhay terms associated with urbanism 10
Table 4. African language isolates 23
Table 5. Dates, homelands and causes of phylic expansion in Africa 25

FIGURES

Figure 1. Elements in reconstructing African history 2

Figure 2. Ounanian point, Northern Mali 3
Figure 3. A proposal for the structure of Nilo-Saharan 5
Figure 4. Classification of Central Sudanic languages 8
Figure 5. Internal structure of Songhay 10
Figure 6. Niger-Congo restructured 12
Figure 7. Revised subclassification of Benue-Congo languages 15
Figure 8. Proposal for the divergence of Bantoid languages 16
Figure 9. A proposed internal structure for Afroasiatic languages 17
Figure 10. Northern Khoesan (Kx’a) languages 22
Figure 11. Khoe languages 22
Figure 12. !Ui-Taa languages 22

i
Roger Blench Africa: the last 12,000 years. Draft
MAPS

Map 1. Nilo-Saharan languages 4

Map 2. Distribution of barbed bone harpoons 6
Map 3. Central Sudanic languages 7
Map 4. The Songhay languages 9
Map 5. Niger-Congo languages 11
Map 6. The Mande languages 14
Map 7. Synthesis of the Bantu expansion 16
Map 8. Current distribution of Afroasiatic languages 17
Map 9. The Omotic languages today 18
Map 10. The Berber languages today 20
Map 11. The Khoisan languages 21
Map 12. Language isolates and residual forager populations in Africa 24

ABSTRACT

Africa is extremely linguistic diverse, but most of its languages fall into four major phyla, plus isolates. The
paper describes the current classification of these and maps their distribution. It is argued that the only way
these can be effectively interpreted in terms of prehistory, is the use of an integrated approach which
incorporates archaeology and palaeoclimate. Potential interpretations of the four major phyla are proposed,
including sections which focus on specific families.

Keywords; Africa; language families; archaeological; palaeoclimate

ii
Roger Blench Africa: the last 12,000 years. Draft
1. Introduction
African languages represent nearly a third of all the languages spoken in the world today; apart from their
incalculable value as intangible human heritage, they represent a rich source of information on the history of
the continent. Their intricate, interlocking distributions must also reflect both the archaeological and cultural
map of the continent. Although the history of humanity in Africa goes much further back, most of the
languages today evolved during the last 12,000 years and this is the time-frame used in this synthesis. This
chapter presents an overview of current perspectives on individual language phyla and possible correlations
with archaeology and palaeoclimate; however, it must be underlined that there is much controversy and we
are still far from a consensus position. Both archaeology and genetics are rapidly revising our understanding
of the prehistory of the continent and if we wish to generate consilience between these very different
disciplines it is crucial to keep up to date with the current literature. The chapter begins with a general
overview of the language families of Africa and their approximate size and distribution, and then considers
each one in turn. It is not possible present a consensus view, nor even in the space available, to indicate in
any detail the disagreements between scholars. These can be followed up in the references. Important
subfamilies such as Berber, Songhay and Central Sudanic are given a more extended discussion. It argues
that an integrated methodology combining linguistics, archaeological results and palaeo-climate can produce
a richer picture than individual disciplines. It concludes with a brief agenda for ongoing research.

African languages are conventionally divided into four continental phyla, Niger-Congo, Nilo-Saharan,
Afroasiatic, Khoisan as well as Austronesian on Madagascar (Blench 2006). Two of these phyla have
significant numbers of speakers outside Africa; Afroasiatic, because of the expansion of Arabic northwards
and eastwards and Austronesian, which is mainly centred on SE Asia and Oceania. Using the estimates from
Ethnologue (Lewis et al. 2017), there are around 2139 African languages spoken today. Language numbers
are distributed very unevenly across the phyla (Table 1);

Table 1. Numbers of African languages by phylum

Phylum Number Source
Niger-Congo 1537 Lewis et al. (2017)
Nilo-Saharan 206 Lewis et al. (2017)
Afroasiatic 318* (i.e. 376-58) Lewis et al. (2017)*
Khoisan (70) Vossen (2013)
Kx’a 4 Lewis et al. (2017)
Khoe 13 Lewis et al. (2017)
Tuu 2 Lewis et al. (2017)
Austronesian 1 (in Africa) Lewis et al. (2017)
Isolates 4~5 Blench (in press)
*Arrived at by deducting non-African lects

In the case of Khoisan, many languages have become extinct in historic times and only inadequately
transcribed data remain. The table gives the figure from Vossen (2013) which includes languages for which
records exist but also the Ethnologue figures for living languages. New Niger-Congo languages continue to
be reported every year, although none of these have radically challenged existing classifications.

This division into phyla owes much to the work of Joseph Greenberg (1963), although there have been many
changes and additions since his proposals were first set out. The explosion of data in recent decades has
stimulated a constant process of reclassification and revision. The coherence of the first three phyla is
generally accepted among scholars although single, authoritative sources to provide the type of proof usual
in Indo-Europeanist or Austronesianist circles are lacking. Debate on the unity of Khoesan swings back and
forth, partly because of the inadequate documentation of so many languages and partly because of the
wayward transcription of clicks (e.g. Westphal 1962, 1963). In the 1980s, most specialists considered that
Southern African Khoesan languages did form a group (Traill 1986; Voßen 1997). However, more recent
syntheses revert back to the prior model of three independent phyla (Honken 2013).

1
Roger Blench Africa: the last 12,000 years. Draft
2. Archaeology, genetics and their relevance

2.1 Overview
The reconstruction of the remote African past is at present pursued via three major disciplines, archaeology,
comparative linguistics and more recently, genetics. However, other ancillary disciplines also play an
important role in reconstruction, in particular palaeoclimate. Figure 1 shows an ideal configuration of these
disciplines;

Figure 1. Elements in reconstructing African history

History

Documentary records
Genetics

Palaeoclimatology
Oral traditions

Comparative ethnography Comparative and Archaeology,

historical linguistics Ethnoarchaeology,
Archaeoscience

Paleobiogeography

Table 2 shows the disciplines used for the reconstruction the African past and assesses various features
associated with both their collection and availability. It gives impressionistic estimates both of the type and
amount of data available in specific disciplines and also the extent to which such data has been exploited.

Table 2. Data categories, their attributes and values in reconstructing African prehistory
Features Geographic Linguistics Archaeology Iconography Textual Oral Genetics Ethnogr
zone traditions aphy/
ethnosci
ence
Samples Very large Small Highly Very small Extensive Large Very
number number of variable sample, number small
point sample chronologically number
samples limited
Precision Low High High Medium Low High Very low
Dating Low High Low High Medium Medium None
Degree of North Medium High High High Low Low Medium
exploitation Africa
in:
E. and S. Medium Medium Medium Medium Medium Low Medium
Africa
Central Medium Low Low Low Medium Low Medium
Africa
West Africa Medium Medium Low Medium Medium Low Medium

Technical advances are likely to further change our interpretations of the ethnolinguistic map in the next
decade. To date, DNA has been analysed based entirely on modern samples. However, Skoglund et al.
(2017) have shown that it is possible to extract DNA from tropical skeletal material, and that this can
illuminate one of the key questions of Afican history, the transition to farming and relations with hunter-

2
Roger Blench Africa: the last 12,000 years. Draft
gatherers. In another development, Bonneau et al. (2017) provide direct dates for rock paintings in Southern
Africa, adapting techniques now used in other parts of the world. Once rock art can be dated in this way,
speculations based on style can give way more accurate chronologies.
2.2 Reconstructing palaeo-climate
The palaeoenvironment of Africa remains somewhat patchily understood, due to the rather approximate
nature of sampling. The most important tool available is lake cores, can be analysed for the fossil pollens
found in their stratified sediment and satellite imagery, which through innovative Digital Elevation Models
(DEMs), can reveal the presence of past water bodies and other key topographical features relevant to
human dispersal. Blome et al. (2012) have synthesised the climate record 150-30 KYA and support the
‘pulse’ model of alternating wet and dry periods, which created a series of potential periods for trans-
Saharan movement of modern humans. An extremely arid period after 80,000 BP may well be correlated
with ‘out of Africa’ movements, when the Horn of Africa developed a very challenging arid environment.
Whether these patterns can be correlated with the earliest dispersals of Khoisan (or at least click speakers)
remains controversial. Apart from the isolates, the language phyla of today originate in a more recent period.

The somewhat fragmentary record suggests that between 18,000 and 13,000 BP, Africa went through an
extremely arid phase, which saw the Sahara extend down towards the West African coast in places. After
this, the climate began to improve, the rivers filled, and an extensive humid period lasted until around 5500
BP (Shanahan et al. 2015). For around 1500 years until 4000 BP there was a hiatus when the weather
became drier after which humidity returned. Current opinion is that the first evidence for agriculture can be
discerned during this period, with finds of domestic millet in Mali (Manning et al. 2011). It is likely that
pastoralism, which is attested in the Acacus by 7000 BP, only reaches West Africa at around 4000 BP.

The extreme aridity of the Late Pleistocene Figure 2. Ounanian point, Northern Mali
was such that there is hardly any evidence
for subsistence systems during this period.
The advent of the Holocene seems to have
stimulated two interlocking strategies, bow
and arrow hunting and the exploitation of
aquatic resources. Far back into the history
of modern humans, seashores, lakes and
rivers would have had abundant supplies of
shellfish and inshore species. The worldwide
distribution of specific types of fish-trap
(Leth & Lindblom 1933) suggests that these
may have been known when modern humans Source: Katzman (2012)
left Africa. During the Holocene, rivers also played an important role in animal and human dispersals. Drake
& Blench (2017) have argued that the green corridors which traversed the Sahara would have allowed Sub-
Saharan animal species, especially large mammals, to disperse to North Africa. This in turn had the
consequence of attracting specialised bow-and-arrow hunters south, in turn initiating the linguistic
revolutions which were responsible for the pattern of languages seen in the present. Cancellieri & di Lernia
(2014) have shown that the Acacus Mountains in SW Libya were re-occupied by foragers from 9800 BP
onwards, during this period a wide variety of sites is characterised by a diverse toolkit, including Ounanian
points (Figure 2), such as are reported from Mauretania to Kharga Oasis (Vernet 2007; Gifford-Gonzalez
2008; Caton-Thompson 1952).
3. African language phyla

3.1 Nilo-Saharan

3.1.1 The languages

The Nilo-Saharan languages stretch from Tanzania to Mauritania and isolated pockets of speakers are found
in Upper Egypt (Map 1). Nilo-Saharan has the distinction of being the ‘youngest’ of the world’s language
phyla to be identified; prior to Greenberg (1963, 1971) there was no literature suggesting that a disparate
3
Roger Blench Africa: the last 12,000 years. Draft
group of languages stretching across Africa constituted a single phylum. The main monographs on the
phylum are Bender (1997) and Ehret (2001). The internal diversity of Nilo-Saharan is such that doubts about
its coherence remain, particularly among non-specialists, but to all intents and purposes it can be treated as
established. Nilo-Saharan has undergone considerable expansion and renaming since Greenberg’s first
proposal. Greenberg (1971) included Meroitic within Nilo-Saharan although his evidence for this
assignation was very tenuous. Despite this it now seems that this intuition was correct (Rilly & de Voogt
2012).

Map 1. Nilo-Saharan languages

The discovery or reclassification of some languages has added further to the complexity of the picture of
Nilo-Saharan. Bender (1997) assumes the members of Nilo-Saharan were as follows;

Songhay Fur Berta Kuliak

Saharan East Sudanic Kunama Kadu
Maban Central Sudanic Gumuz Koman

The Nilo-Saharan classification in Ehret (2001) is strongly at variance with the views of Bender. Blench
(2002) is a critical comparison of the methods and results of these two authors. This chapter follows
Bender’s more conservative view of both the structure of Nilo-Saharan and the potential for reconstruction
of proto-forms. Figure 3 presents a new proposal for the internal structure of Nilo-Saharan, based on the
massive body of fresh evidence that has become available in the last two decades.

Map 1 shows that the Nilo-Saharan languages are marked by extreme fragmentation and dispersal of even
major subgroups such as Eastern Sudanic. Part of this may be attributed to the disruptive effects of the slave
trade across Central Africa in recent centuries, but more probably it reflects a period when group sizes were
very small, practised either foraging, pastoralism or low-input agriculture, and moved freely when climatic
or security conditions were unfavourable. The branches of Nilo-Saharan show no ‘nesting’, in other words
they seem to be scattered across Africa, representing a series of individual migrations. Each subgroup shows
considerable differences, making lexical cognates and thus reconstructions relevant to early subsistence hard
4
Roger Blench Africa: the last 12,000 years. Draft
to detect. As the map also shows, the diversity of Nilo-Saharan is in the southeast, with a number of
divergent groups clustered along the Ethio-Sudan borderland. It is therefore assumed that the spread of the
phylum was from east to west.

Figure 3. A proposal for the structure of Nilo-Saharan

Proto-Nilo-Saharan

Bertha
Shabo [?]
Kunama Koman Gumuz

*Central
African
Songhay Saharan

Kuliak

Maban Fur Central Kadu Eastern

Sudanic Sudanic

3.1.2 Archaeological interpretation

One of the earliest proposals for an archaeological correlate for Photo 1. Wavy-line pottery
Nilo-Saharan is the ‘Aqualithic’ (Sutton 1974, 1977). Sutton
noticed an approximate correlation between the language
distribution, and a cultural complex associated with humid
environments, bone harpoons (presumed to be used for hunting
hippos and crocodiles) (Photo 2) and ceramics (Photo 1), possibly
with wavy-line decoration (Mohammed-Ali & Khabir 2003).
Archaeologists have raised objections to the coherence of this
proposed ceramic, but recent research has indeed shown that
ceramics appear in the Central Sahara, on the Nile and in West
Africa (Mali) in a very short ‘window’ around 10,500 to 10,000
BP (see review in Huysecom 2009). To complete the picture,
ceramics have also been uncovered from the same period around
Lake Turkana (Lahr pers. comm.). Drake et al. (2010) relate this to the
‘Green Sahara’, the expansion of aquatic resources in the desert some Photo 2. Bone harpoon
eleven thousand years ago, which was responsible for images of species
such as hippos in arid regions. Their supplementary tables show that
linguistic data also supports this, with reconstructed forms for ‘hippo’ and
‘crocodile’.

5
Roger Blench Africa: the last 12,000 years. Draft
Broadly speaking, therefore, Sutton’s prescient hypothesis can be reconciled with this new data. Map 2
shows the distribution of barbed bone harpoons, which should be compared with the distribution of Nilo-
Saharan languages (Map 1), corresponding neatly to the hypothesis that Nilo-Saharan first developed in the
southeast, where its diversity is still maintained. Datapoints for early ceramics are much less abundant, but
the distribution is also similar. As the ‘green Sahara’ developed, speakers travelled westward, attracted by
abundant aquatic resources, including hippos, crocodiles and the Nile perch, Lates niloticus.

Map 2. Distribution of barbed bone harpoons

Source: Courtesy Nick Drake

The Gobero site in Northern Niger (ca. 10,000 BP) is the largest cemetery uncovered in Sub-Saharan Africa
and displays a remarkable range of physical types (Sereno 2008; Garcea 2013; Stojanowski 2013). The
associated artefacts and ornaments indicate a culture with time to manufacture luxury goods and presumably
reflect complex social patterns. The exact subsistence strategies of the peoples who buried their dead at
Gobero remain under discussion1, but the current view is that the phenotypic affinities of the first occupants
(named ‘Kiffians’ after a proposal by Abbé Breuil) were with North Africa, and seem to have been hunters
and fishers. After 8200 BP, a millennium of high aridity led to the lakeside settlements being abandoned and
then re-occupied by more gracile physical types, with a greater focus on aquatic resources. A few cattle
bones among the later assemblages probably imply they were trading meat with pastoralists further north in
the Air, not that they were actually producing cattle.

1 My thanks to Paul Sereno for summarising the debates over the interpretation of Gobero, to general stimulating
discussions particular concerning the Ounanian, to Konstantin Podzniakov for lehoping update me on Atlantic
languages, to Nick Drak for continuing debate over the interpretation of paleo-climate data.
6
Roger Blench Africa: the last 12,000 years. Draft
3.1.3 Central Sudanic
One of the most widely dispersed families of Nilo-Saharan is Central Sudanic, which is scattered across
some of the most inaccessible regions of Central Africa (Map 3). In recent times these have been further
troubled by vicious civil wars. Ironically, this is not the first episode of major insecurity to rock the region;
from the eighteenth century, the slave trade had an important role in further dispersing the western branches.
The consequence of this has been fragmentation and dispersal, with Central Sudanic intertwined with
regional Arabic, Fur, Ubangian and Eastern Sudanic languages. Sara-Bongo-Bagirmi (SBB) languages form
a relatively coherent subgroup and the centre of diversity is all in the Southeast. This suggests that Central
Sudanic began to diversify in the region north and west of Lake Albert. There is no evidence that agriculture
can be reconstructed in Central Sudanic which points to a dispersal of hunters east to west. A credible
scenario is that this occurred in the period after 4000 BP, when the climate began to improve and large
mammal populations increased again.

Map 3. Central Sudanic languages

Source: updated and redrawn from Boyeldieu (2004)

Pascale Boyeldieu (2000) recalls a science fiction story, published before the First World War, which
envisaged Bagirmi, the Central Sudanic language of a small state in Central Africa, becoming the lingua
franca of Africa in the year 9040, when Europe has sunk beneath the waves (Van Gennep 1911). This is still
in the future, but it is pleasant to see an African language being given such high status in fiction.

Central Sudanic is usually divided into two major branches, East and West (Figure 4). The documentation of
Birri is too weak to be sure of its position and it is provisionally given a branch of its own, pending further
research. Formona-Sinyar is also sparsely documented and its position as a part of the Western branch must
remain a best guess.

7
Roger Blench Africa: the last 12,000 years. Draft
Figure 4. Classification of Central Sudanic languages
Proto-Central Sudanic
West East

Sara-Bongo Sinyar- Kresh- Mangbetu-Asua

-Bagirmi Formona Aja
Mangbutu-Efe Lendu- Moru-Madi
Birri
Ngiti

Historically speaking, the puzzle is to explain how Central Sudanic has become so territorially dispersed.
The core of its diversity appears to be west of Lake Victoria on the Uganda-Zaire border, although most
languages are far to the northwest, in the Sara area. Fragmented population islands connect them, including
some relatives of Sara-Bagirmi now spoken in the extreme southwest of Sudan. Separating them today is the
great eastward salient of Ubangian languages which extend from North Cameroun to Southern Sudan.
Bouquiaux & Thomas (1980) point out that the Ubangian expansion must have taken place north of the
forest prior to the Bantu expansion and they assign it a tentative date of 35-4000 BP. The Ubangian
expansion must have broken up a continuous chain of Central Sudanic languages from NE DRC to the
borders of Cameroun. The Sara languages subsequently underwent a secondary expansion north of
Ubangian.
3.1.4 Songhay
The Songhay languages are spoken along the Niger between Timbuktu and Gao, stretching into the Sahara
of Niger and South and East into Benin Republic and Nigeria (Map 4). There are also Songhay speakers in
Sudan, remnants of the pilgrimage to Mecca (Abu Manga 1995). An isolated population, the Kwarandzyey
of Tabelbala (‫ﺔ‬ ‫)ﺗﺒﻠﺒﺎﻟ‬, live in a small community on the Moroccan-Algerian borderland (Souag 2010).
Songhay is often treated in earlier literature as if it was a single language, but it is now recognised as a
complex cluster.

Songhay is undoubtedly Nilo-Saharan, as it shares a significant number of basic lexemes with remote
geographical languages which are neither Afroasiatic nor Niger-Congo in origin. But it is distant from other
Nilo-Saharan branches and suggests that either pre-proto-Songhay was spoken in an isolated community
with little differentiation or that the relatives of proto-Songhay were subsequently assimilated by other
languages. Songhay appears to have come under strong Mande lexical and grammatical influence (perhaps
specifically from an ancestor of Soninke) at an early stage in its evolution (Creissels 1981). At the same time
proto-Songhay was diverging, it must also have been in contact with Berbers, to judge by a small number of
Berber borrowings in early or proto-Songhay (Souag 2010).

Songhay is traditionally divided into two primary subgroups, Northern and Southern; a revised classification
is proposed by Souag (2012). Figure 5 shows the internal structure of the Songhay languages using this new
model. According to this hypothesis, the first split within the family was between Eastern Songhay, probably
spoken around Gao, and Northwestern Songhay, somewhere further north; it was followed by a more
prominent split between Western and Northern Songhay. Eastern Songhay is close to being a dialect
continuum, although a handful of extra-riverine varieties at Hombori and Kikara in Mali and Djougou and
Kandi in Benin show greater divergence.

8
Roger Blench Africa: the last 12,000 years. Draft
Map 4. The Songhay languages

Western Songhay, spoken around Timbuktu and at Djenné, remained in fairly close contact with Eastern, a
situation intensified not just by the ease of riverine trade but also by the Songhay Empire's conquest of
Timbuktu; as a result, words attested only in Eastern and Western Songhay can be securely reconstructed
only for proto-Eastern. The divergence of the Northern Songhay split with the rest has been far more
complete, thanks to its speakers’ dispersal in oases and desert areas dominated by speakers of Berber and
Arabic (Wolff & Alidou 2001).

Two varieties (Tadaksahak and Tagdal) are spoken by nomadic groups in the desert; since agricultural
vocabulary is reconstructible for proto-Northern Songhay, these are likely to result from a later change of
lifestyle or language shift. The dispersal of Kwarandzyey as far as Tabelbala (‫ﺔ‬ ‫)ﺗﺒﻠﺒﺎﻟ‬, a thousand
kilometres north of the rest of the family, is linked to the trans-Saharan trade. Tabelbala was an important
halt on a trade route linking Morocco and the Sahel. Souag (2012) points out that Songhay terms related to
agriculture, such as ‘hoe’ (kumu), ‘sow seed’ (dzʊm) and ‘irrigated garden (lәmbu) are retained in
Kwarandzyey. This suggests the their ancestor were brought there as horticulturalists, whether as slaves or
through some other type of contract relationship.

As with Berber, the puzzle of Songhay is that it is remote from other branches of its parent phylum, Nilo-
Saharan, yet all its lects are very close to one another, implying a relatively recent dispersal. Souag (2012)
observes that lexemes concerned with both livestock and agriculture reconstruct in proto-Songhay, as well as
words reflecting urbanism, such as ‘kitchen’, ‘key’ etc (Table 3).

9
Roger Blench Africa: the last 12,000 years. Draft
Table 3. Proto-Songhay terms associated with urbanism
Gloss Proto-Songhay
village, town *ko(y)ra
house *hugu
room *gar
kitchen *fuutay
key *karkabu
clay *laabu
Source: adapted from Souag (2012:201)

Figure 5. Internal structure of Songhay

Proto--
Songhay

Eastern: Koyraboro Northwestern

Senni, Kaado, Zarma,
Dendi, Hombori Senni,
Tonday Songway,
Kiini

Northern: Kwarandzyey, Western: Koyra Chiini,

Tadaksahak, Tagdal, Djenne Chiini
Tasawaq, Emghedesie
Source: Souag (2012)

This implies that it was the development of urbanism which kick-started the expansion of Songhay as we see
it today and it is reasonable to associate it with the incipient urbanism occurring in the archaeological record
from ca. 200 BC. Park (2010) describes the ceramic phases which mark a transition to urbanism in the
Timbuktu area. There is considerable linguistic evidence for a shift from aquatic subsistence to cereal
agriculture and livestock. Importantly, none of the significant terms reconstructed for either architecture or
livestock in proto-Songhay are Berber or Arabic loans.

In subsistence terms the Songhay apparently split into those who develop the urban motif and spread east
and south as traders and those who adopt cattle pastoralism from their Berber neighbours and move into the
arid zones around Agadez. They become the ‘Wangarawa’ who traded gold and other products with the
Hausa in the early Middle Ages, marked by a significant number of borrowings from Songhay into Hausa
(Skinner 1996). The surveys reported in Haour et al. (2011) in the Niger River Valley on the border of Benin
and Niger indicate proto-urban settlement formation from around 400 AD. Given this whole region is still
Dendi-speaking, these sites probably mark the first expansion of the Songhay along the Niger River.

10
Roger Blench Africa: the last 12,000 years. Draft
3.2 Niger-Congo

3.2.1 Overview
Niger-Congo has more languages than any other in the world and it occupies a greater land area than any
other African phylum (Williamson & Blench 2000). It includes the well-known Bantu languages, which are
spread over nearly all of Eastern and Southern Africa, although they are only a sub-branch of Niger-Congo.
Map 5 shows the present-day distribution of Niger-Congo languages. If compared with Nilo-Saharan (Map
1), one difference is immediately apparent. Niger-Congo languages form large territorial blocks with much
less of the fragmentation and geographical isolation characteristic of Nilo-Saharan. Typically, this suggests
both more recent expansions and the gradual spread of more sedentary populations colonising areally.
Nonetheless, Niger-Congo has its own perplexing problems; why should Bantu have covered such a large
area and how is it that the Kordofanian languages are isolated from the remainder by a large zone of Nilo-
Saharan speakers?

Map 5. Niger-Congo languages

Making a tree of Niger-Congo is no simple task; Niger-Congo has the largest number of languages of any
phylum in the world and is far more internally diverse than Austronesian, its nearest counterpart. Formerly
accepted groups such as Kordofanian, Kwa and Atlantic have been split into individual branches, which may
be independent. There are also isolates, such as Sua, Limba and Mbre, which have no place with established

11
Roger Blench Africa: the last 12,000 years. Draft
families. Nonetheless, it is essential to understanding early Niger-Congo that we model its internal structure,
and with suitable caveats, this is presented in Figure 6.

Figure 6. Niger-Congo restructured

Proto-Niger-Congo

Dogon

Ijoid
Mande

Ijo Defaka

Rashad Kwaalak-
Domurik
Tegem-Amira
‘Kordofanian’
Senufic
?
Kru
Talodi Heiban

Atlantic
Mel

*Gur-Adamawa continuum
Limba

Gola

Sua
‘Periphera Central Adamawa Adamawa Adamawa Adamawa
l Gur’ Gur 1 2,4,5,12 6,13,14,+ 7,9,10
Fali Day

Bere Gbaya
Kwa linkage Adamawa
8

Ubangian
Ega

Nyo West Central Ga-

Bank Togo Dangme

Volta-Niger linkage
Ikaan

Benue-Congo linkage

Gbe Yoruboid Akpes Ayere- Nupoid

Akokoid Ahan Okoid
Edoid, Idomoid
Igboid

12
Roger Blench Africa: the last 12,000 years. Draft
The reasons for the dispersal of Niger-Congo remain controversial. It is evident that its early phases were
pre-agricultural; ‘bow’, ‘arrow’ and ‘spear’ can be reconstructed, but no names for crops or domestic
animals (Blench 2006). One proposal is that the introduction of the bow is marked by the spread of
microlithic points to the southwestern Sahara by 12,000 BP. The archaeological culture known as the
Ounanian, first recognised by Breuil in 1930 at Ounan, south of Taodeni in northern Mali, consists of tanged
arrowheads (Figure 2). These occur in Mauretania (Vernet 2007), modern-day Mali by 9-10,000 BP
(Raimbault 1990) and in the Libyan Acacus (Cancellieri & di Lernia 2014). A small number of similar
points were found at Kharga Oasis (Caton-Thompson 1952:162). Ounanian points are arrowheads, or
‘projectile points’, in the cautious language of lithics. The appearance of these across the West-Central
Sahara can be connected with northward movement of large animals along trans-desert corridors with the
opening up of the ‘green Sahara’. An enhanced capacity to bring down large mammals would provide a
significant competitive advantage, thereby stimulating early language expansions.

The linguistic geography of Niger-Congo suggests an early split between the bow and arrow hunters and the
evolution of groups focusing on aquatic resources. The coast is marked by a series of branches, Atlantic
Bak, Kru, Kwa, Ịjọ situated at the terminus of river systems. We know that canoes were widely in use in
West Africa by 8000 BP following the discovery of the Dufuna canoe in 1994 (Photo 3). If early speakers of
these groups began to explore the river systems by canoe then they would in due course arrive at the sea.
Other populations, notably the Kordofanian, Gur, Adamawa and Ubangian peoples have a land-based
distribution which points to a spread across the savannas, presumably as foragers.

Photo 3. Excavating the Dufuna canoe

Source: Naijaland blog

3.2.2 Kordofanian
An unusual feature of the geography of Niger-Congo is the location of the Kordofanian languages. The
hypothesis that a set of languages in the Nuba mountains formed a coherent branch of Niger-Congo is due to
Greenberg (1963). Greenberg made one significant error, including the Nilo-Saharan Kadu languages (see
Schadeberg & Blench 2013 for fresh data and a history of these ideas). However, the other groups he
identified are generally accepted as Niger-Congo. Abundant new data has now questioned the coherence of
Kordofanian as a single branch. Despite their isolation, Kordofanian languages may constitute as many as
four independent branches (Blench 2013). This points to the existence of a ‘corridor’ joining the Nuba Hills
to West Africa, along which various foraging groups may have travelled. No reconstructions for the names
of crops have ever linked Kordofanian with West Africa, so these would have been hunters, trekking
eastward, prior to the arid period after 5500 BP. Blench (1995) argued that the riverine corridor represented
by the Wadi Hawar (which formerly ran from the Nile Confluence to Eastern Chad) attracted pastoralists
westwards. The huntable mammals along its banks may have functioned in a way similar to the green
corridors across the Sahara.

13
Roger Blench Africa: the last 12,000 years. Draft
3.2.3 Mande
The Mande languages are spoken in the western half of West Africa, but have far-flung outliers as far as
Nigeria (Map 6). A large proportion of the population of Mali, Côte d’Ivoire, Guinea, Sierra Leone, and
Liberia speak Mande; they are also found in substantial numbers in Burkina Faso, Senegal, Gambia, and
Guinea Bissau, with outlying groups in Mauretania, Benin, Ghana, Togo, and Nigeria (Dwyer 1989;
Kastenholz 1991/2). The Mande languages have long been recognised as a unity.

Map 6. The Mande languages

Source: Vydrine et al. (2000)

The distribution of Mande languages clearly indicates a core area of widespread and closely related
languages, the west/southwest languages, which include Mandinka, Malinke, Bambara, and the outliers,
such as Busa in Nigeria, which is extremely different from its relatives. This points to two-stage dispersal;
an early period when the remote Mande languages scattered, and the later period associated with the early
kingdoms on the Niger, perhaps as late as the twelfth century when the village mounds of the Inland Delta
were being built up. Vydrine (2009) has argued that the reconstruction of livestock terms in early Mande
points to a pastoral origin, probably on the southern edge of the Sahara. This would potentially date it to
around 4000 BP, which would make broadly correspond to the internal diversity of Mande.
3.2.4 Benue-Congo, Bantoid and Bantu
One of the largest and most significant subgroups of Niger-Congo is Benue-Congo. Originating with
Westermann’s (1927) Benue-Cross-Fluss, it took shape in Greenberg (1963), Williamson (1971) and De
Wolf (1971). The name Benue-Congo was introduced by Greenberg (1963) who argued that Bantoid and
Bantu represent nested subsets. This was a radical interpretation which swept away the prevailing view,
derived from Meinhof (1910) and Guthrie (1967-1971) that Bantu was a separate family. Benue-Congo itself
is scattered across present-day Nigeria, and parts of Western Cameroun. Figure 7 shows a revised
subclassification of Benue-Congo languages;

14
Roger Blench Africa: the last 12,000 years. Draft
Figure 7. Revised subclassification of Benue-Congo languages
Proto-Benue-Congo

Central Nigerian Ukaan ?

Bantoid-Cross
Kainji Plateau Jukunoid

Cross River
Bantoid
Central Tarokoid
Ndunic Plateau Eggonic-
Alumic Jilic Upper Lower Ogoni Delta
Pl Cross Cross Cross
Beromic SE Plateau
Northwest
Plateau South
North
Intervening
Tikar groups

Dakoid Mambiloid Bantu

Benue-Congo must have expanded outwards from the confluence of the Niger-Benue, several thousand
years before the beginning of the Bantu expansion, although exactly what drove this expansion is presently
unclear (Blench 2010). There is no evidence these were fishing peoples, so perhaps improved projectile
technology was responsible for this striking dispersal. The second wave of Benue-Congo expansion led to
the evolution of Bantoid, the languages that ‘stand between’ Eastern Benue-Congo and Narrow Bantu
(Blench 2015). Bantoid languages do not fit together is a nested pattern, and in the absence of more
extensive historical linguistics it is assumed individual groups split away from a common stem, and
developed their own characteristics. The order in which this took place remains controversial, and will take
considerable further work to resolve in a satisfying manner. A proposal is presented in Figure 8;

As Map 5 shows, the Bantu languages cover most of central and southern Africa. The Bantu expansion is
one of the most striking phenomena in Africa linguistic history. Bantu, far more than other groups of Niger-
Congo, has been the subject of conflicting attempts at historical interpretation. Guthrie (1962, 1967-71)
considered it to have originated somewhere in present-day Zambia. Greenberg (1955) situated its homeland
in NW Cameroun, a view previously espoused by Johnston (1919-1922). It is now generally thought that
around four thousand years ago, populations in southern Cameroun began and striking and rapid expansion,
down the west coast, eastwards along the northern edge of the equatorial forest and across the forest itself,
making use of the abundant river systems. Linguistic, palynological, vegetation and climatic evidence
concur with archaeology in tracing this movement but its initial stimulus is still obscure (Bostoen et al.
2015). The archaeological and the palynological data and point to two major peaks, roughly 3800-3500 and
2500-2000 BP, where there is forest disturbance and expansion of archaeological sites (Oslisly et al. 2013).
The earlier peak of vegetation change is prior to the introduction of iron tools, which suggests humans must
have taken advantage of newly opened natural savannah corridors, which may have initially attracted
hunters, something also inferred from anthropic tree species distributions (Blench in press a).

15
Roger Blench Africa: the last 12,000 years. Draft
Figure 8. Proposal for the divergence of Bantoid languages
South Bantoid

Bendi ?
Tivoid
Buru
Furu cluster
Yemne-Kimbi
East Beboid
Nyang
Ekoid
-Mbe
Grassfields
Part of Bantu A group
including Jarawan
Narrow Bantu
Source: Blench (2015)

Map 7 presents a synthesis of current ideas about the dates and routes of the Bantu expansion. The
movement across the forest is represent as dotted lines, since it is likely it was less a coherent progress more
a slow network-like movement.

Map 7. Synthesis of the Bantu expansion

3.3 Afroasiatic

3.3.1 Overview
The Afroasiatic (Hamito-Semitic, Afrasian) language phylum consists of some 376 languages spoken mainly
in Africa but with extensions of Arabic and Aramaic (its most well-known members) through the Middle
16
Roger Blench Africa: the last 12,000 years. Draft
East into Russia. Map 8 shows the present-day distribution of Afroasiatic languages in Africa, with extinct
languages such as Ancient Egyptian marked by bracketed names. Their distribution is extremely skewed,
since one language, Arabic, has more than 100 million speakers, i.e. as much as all the others combined.
Hausa, with up to 25 million speakers, is numerically the next most important. Two competing
reconstructions of Afroasiatic have been proposed, ironically published in the same year (Ehret 1995; Orel
& Stolbova 1995).

Map 8. Current distribution of Afroasiatic languages

The internal structure of Afroasiatic is controversial, with almost every configuration possible being
proposed (see detailed review in Blench 2006). In the light of this, Figure 9 presents a compromise proposal.

Figure 9. A proposed internal structure for Afroasiatic languages

Proto-Afroasiatic

Elamitic? North Central

Afroasiatic Afroasiatic

Chadic Cushitic

Berber Egyptian Semitic West Central Masa East Beja Agaw East South Dahalo? Omotic Ongota?

17
Roger Blench Africa: the last 12,000 years. Draft
An interesting question is whether Elamitic, a language isolate in Ancient Iran might have been an early
branch of Afroasiatic (Blažek 1999). If so, it argues for an early spread of Afroasiatic prior to Semitic.

Two quite different models of the expansion of Afroasiatic are current in the literature. The existence of
ancient written sources for Semitic languages and the archaeology of the Near East has suggested to
researchers unfamiliar with the African evidence that the Near East was the origin of its dispersal (e.g.
Bellwood 2013). For example, the Afroasiatic dispersal has been identified with the Natufian culture (Bar-
Yosef 1998). However, this is highly problematic, because all the diversity in Afroasiatic is in Sub-Saharan
Africa, particularly Ethiopia. A more credible alternative is that Afroasiatic originated in southwest Ethiopia,
where its most internally diverse branches are found today. The primary split would have been between
Omotic and Cushitic, with Chadic as Cushitic speakers who migrated westwards along the former Wadi
Hawar, the corridor which may also have carried the Kordofanian hunters. North Afroasiatic, i.e. Berber,
Egyptian and Semitic would have followed the Nile Corridor northwards. The Berber then moved into the
Western Desert, the Egyptians remained along the Nile and the proto-Semites entered the Near East as
pastoralists. This scenario is strongly contrary to researchers focused on Near Eastern origins, but it does
explain the linguistic data in a far more satisfactory way.
3.3.2 Omotic and Cushitic
The Cushitic language family is largely in Ethiopia and Omotic is almost entirely in the southwest of the
country. Omotic is the most divergent branch of Afroasiatic and for this reason is considered a primary split.
Map 9 shows the distribution of Omotic languages today. The linguistic evidence suggests strongly that the
split between Omotic and Cushitic was between foragers and vegeculturalists and livestock producers.
Words for ‘honey’, ‘enset’ can be reconstructed in proto-Omotic (Bender 1988; Blench 2007a), whereas
words for livestock cannot (Blench 2008). The other striking aspect of the Afroasiatic lexicon is the
reconstruction of terms for ‘cattle’. All branches except Omotic attest a form similar to *ɬa for ‘cow’ and a
possible interpretation is that early Afroasiatic speakers began the management of wild cattle in the
Egyptian-Sudan borderland some 10,000 years ago.

Map 9. The Omotic languages today

18
Roger Blench Africa: the last 12,000 years. Draft

Chadic, the most diverse branch of Afroasiatic, is spread across the region from Nigeria to Chad. Blench
(2014) has argued that the salience of ‘fish’ in Chadic argues for a migration westwards along the now-arid
waterways joining the Nile Basin to Lake Chad.
3.3.3 Berber and Guanche
The Berber languages constitute a major branch of the Afroasiatic language phylum and are spoken today
both by settled and nomadic populations along the North African coast and far down into the Sahara,
presently reaching the borders of Nigeria (Blench in press, c). Today, Berber languages are confined to a
series of islands surrounded by Arabic except where they touch Sub-Saharan African languages (Map 10).
This is a considerable geographical range, but it has been regularly argued that Berber culture and by
implication, people, reached as far as the Nile Confluence (e.g. Behrens 1989). Bechhaus-Gerst (1989)
claimed to detect loans from Berber into Nubian. Such a stretch is not inconceivable geographically, but the
evidence for this remains weak, both linguistically and archaeologically (also a negative evaluation in
Kossmann 2013). Nonetheless, Berber must once have been the dominant population throughout much of
North Africa and the Sahara in the past (Basset 1952; Brett & Fentress 1996; Blench 2001). Although the
Tuareg are presently the most widespread group, found across much of Algeria, Niger and southern Libya
(Bernus 1981), their expansion is probably relatively recent as they may have entered the south-central
Sahara as late as the 6th century AD (Camps 1980).

The Berber remain highly mobile, the Tuareg in particular forming new communities in the coastal cities of
West Africa and they are adept at maintaining a strong media presence. The Zenaga in SW Mauretania were
a significant group when first described, but are now down to some 300 speakers (Faidherbe 1877; Taine-
Cheikh 2008). North of Agades in Niger live the Tetserret, who language shows correspondences with
Zenaga and who are now encapsulated by the Tuareg (Lux 2013). Other islands of Berber speakers occur
with the Arabic-speaking zone further east, most notably at Awjila (‫( )ﺃﻭﺟﻠﺔ‬Paradisi 1960; Putten 2013),
formerly at El-Fogaha (Paradisi 1963) and Siwa (Laoust 1932). Furthermore it is often claimed that Berbers
reached the Canaries at an unspecified date in the past, leading to the formation of the Guanche, the now-
vanished aboriginal population (Wölfel 1965).

Despite an abundance of information, there are a series of major unanswered questions about the affiliations,
origins and date of diversification of the Berber languages (Galand 1970-1). Berber is Afroasiatic, and its
nearest relative is Semitic. Yet when deep-level Arabic borrowings are weeded out it retains only a very
small corpus of established Afroasiatic roots, pointing to a ‘long tail’, a split from Afroasiatic at a
considerable time-depth. When and where this took place is uncertain, but presumably in the Nile Basin.
Similarly, the dates of the primary expansion of Berber are problematic; its extremely low internal diversity
points either to a recent epoch or to an episode of language levelling. Evidence from Neo-Punic and Latin
borrowings suggests a date for proto-Berber of 100-200 AD. Under no circumstances can Berber be
identified with the Capsian (ca. 12,000-8000 BP) or even the first stages of the Neolithic in North Africa (?
7000 BP onwards), both of which are far too early to be reconciled with its internal diversity. If this is
indeed so, what process is congruent with the archaeological record?

The only way to account for the distinctiveness of Berber is to suppose that the speakers of the proto-
language must have been resident somewhere for a long period, diverging from Afroasiatic but not
diversifying internally2. At a point in the more recent past, a social or economic change must have
transformed their society, stimulating a major expansion. Blench (2001) argued that this was pastoralism, on
the basis that a quite detailed lexicon of livestock-keeping can be reconstructed for proto-Berber. This ought
to correspond to the expansion of pastoralism across the Central Sahara, which is around 5-4000 BP (Brooks
et al. 2009).

2 Alternatively they could have diversified but the branches that developed at that period were then assimilated by other
languages.
19
Roger Blench Africa: the last 12,000 years. Draft
Map 10. The Berber languages today

When the Spanish first reached the Canary Islands in the sixteenth century, they found the inhabitants were a
people called the Guanche, with four dialects spread across seven main islands. So entrancing were their
dances that these were carried back to Europe and entered the repertoire of classical music, hence the
canaries in the harpsichord suites of J.S. Bach. Unfortunately, the fate of the language was less iconic, as by
the end of the eighteenth century it had disappeared, with the speakers killed, dying from disease or being
assimilated. Archaeology currently points to an initial settlement from the North African mainland around
1000 BC (González & Tejera Gaspar 1990). There are many perplexing aspects of the culture of the
Canaries. The most notable is the presence of mummified bodies, which use techniques specific to Ancient
Egypt. Similarly, there are small earth constructions that look very like attempts to replicate pyramids.
Whether this implies the ancestors of the Guanche were carried to the Canaries by Egyptian ships remains a
point for speculation.

The records of Guanche are only those recorded by

travellers and amateur enthusiasts. Unfortunately, little of Photo 4. Guanche inscription, Gran Canaria
the vocabulary is basic (there are, for example, almost no
records of body parts), and it is impossible to identify the
affiliation of Guanche unambiguously. The classic
synthesis, Wölfel (1965), noted many similarities with
Berber and much smaller number with Basque. A
persistent history of theories relating Basque to African
languages, particularly Berber, goes as far back as
Gabelenz (1894) and Wölfel (1955). Most researchers
who have looked at the fragmentary records of Guanche
have identified it as a branch of Berber (e.g. Galand
1987/88). Moreover, the very short inscriptions on rock
in the Canaries are in the old North African Numidian
script (Photo 4). We cannot say for certain that Guanche
was a Berber language, but that there was a major influx of Berbers, which introduced the pastoral culture of
North Africa and transformed the languages. Roman contact is now also demonstrated and it has been
argued that the later inscriptions show familiarity with the principles of Latin writing.
3.4 Khoisan
The Khoisan languages are easily the most problematic phylum in Africa, primarily because the extent to
which they constitute a genetic grouping is uncertain. The languages themselves are spoken by small
scattered foraging populations in south-western Africa and they are under threat from their dominant
20
Roger Blench Africa: the last 12,000 years. Draft
neighbours. Records suggest the existence of some 100 Khoisan languages, yet only 23 are spoken today.
Map 11 is a composite, showing the distribution of all Khoisan languages spoken either in the present or
recent past and their approximate extent. Many of these languages now have very few speakers, scattered
among the migrant Bantu and white populations of Southern Africa.

Map 11. The Khoisan languages

The Khoisan or ‘click’ languages in Eastern and Southern Africa are similar to the languages of Australia, in
that they are defined by shared phonological and morphological features rather than by an evident common
lexicon. Khoisan as presently understood was probably first outlined by Bleek (1956) in her ‘Bushman
Dictionary’ but Greenberg (1963) is given credit for the ‘click’ language grouping, later renamed macro-
Khoisan. This was the grouping in a single phylum of all the African languages with ‘click’ sounds and not
otherwise classified, i.e. excluding Southern African Bantu and Dahalo. This joined Hadza and Sandawe in
Tanzania with the Khoisan languages of Southern Africa. The link with Sandawe and Hadza is supported by
Ehret (1986) but questioned by other Khoisanists (e.g. Elderkin 1983 for Sandawe; Sands 1998 for Hadza).
Sands’ (1998) study of Khoisan relationships has shown that especially in the case of Hadza most of the
lexical arguments advanced to support of its affiliation rest on very doubtful correspondences or erroneous
lexical citations.

The language of the Kwadi in Southwestern Angola is known only from field notes left by Westphal, and it
is doubtful whether any Kwadi speakers exist today. The Kwadi are pictured in Estermann (1976) and they
are visibly not of Khoesan physical type. Güldemann (2004, 2008) argues forcefully that Kwadi was indeed
related to the Khoe (Central) Khoisan. Certainly it shares a number of lexical items, although its morphology
is very distinctive. Figure 10, Figure 11 and Figure 12 show composite trees adapted from Honken (2013).

21
Roger Blench Africa: the last 12,000 years. Draft
Figure 10. Northern Khoesan (Kx’a) languages
Northern

‡Hõã Ju

Northern Southern

Kwando |Akhoe !Xuun Ju|Hoan ‡X’auǁen

Figure 11. Khoe languages

Proto-Central

Kwadi ?

Khoekhoe Naro-ǁGana Shua Tshwa Kxoe

Figure 12. !Ui-Taa languages

!Ui-Taa

!Xóõ |Xegwi N|uu |Xam Others

Archaeological evidence suggests that people of Khoesan type and culture may have been present in Eastern
and Southern Africa for >100,000 years and that those remaining are inheritors of this culture. The click
languages of Eastern Africa, such as Hadza, point to the earlier foraging layer which preceded Bantu.
However, as Vossen (1997) points out, the reconstruction of food production terms, especially livestock,
suggests that Central Khoesan is a relatively recent expansion, perhaps only 2000 years, in contact with
Bantu and Cushitic livestock producers. The !Ui-Taa and Kx’a branches of Khoesan are thus older layers
and Khoe an innovative wave reflecting contact with new subsistence strategies. Guldemann (2008) has
indeed argued that the Central Khoisan were originally much further north and irrupted into the Southern
Africa area as pastoralists, but this does not seem to be supported by archaeological evidence. A more
plausible interpretation of the archaeological evidence is that early Khoe speakers were in situ foragers in
contact with livestock producers, primarily Southern Cushites (accounting for the transfer of cattle) and later
the Iron Age Bantu (Blench 2009a).

22
Roger Blench Africa: the last 12,000 years. Draft
3.5 Austronesian
Austronesian is not usually regarded as an African language phylum, but it is included here, since it is
spoken throughout Madagascar and on the Comoros. Austronesian is a very large language phylum, spoken
from Taiwan to New Zealand, comparable in size to Niger-Congo. The outlines of the Austronesian family
were first recognised in the early eighteenth century by the Dutch scholar Adriaan van Reeland, who
compared Malay, Malagasy and Polynesian (Relandus 1708). Malagasy is Austronesian, but its precise
genesis has been much debated. It is generally considered to belong genetically to the Barito languages,
today spoken in southeast Kalimantan (Simon 2006). However, it has clearly undergone considerable
influence from Malay, whence it draws many nautical and other technical terms (Adelaar 1994). In addition,
there are numerous loans from the Bantu languages of the East African coast, some from Swahili, but others
from different languages, particularly those of Mozambique which probably post-date the Swahili
borrowings (Blench 2009b). It is likely the ancestors of the Malagasy came in multiple waves from Island
SE Asia from the 5th century onwards and settled both the East African Coast and the island of Madagascar.
However, the rise of Arab shipping in the Gulf overwhelmed the coastal settlements and left Austronesian
only spoke on the island.
3.6 Unclassified languages, marginal peoples and their significance
Apart from the well-known and largely established phyla, a small number of African languages defy easy
classification. Actually, it is very surprising that they should be so few. There are many isolates in the New
World, as in Papua, Australia and Siberia. On the assumption that the origin of modern humans lies in
Africa, there should be many more. As this chapter has suggested, this pattern of African language phyla
must reflect large-scale population movements, change and assimilation in a relatively recent period. Papua,
by contrast, has been largely isolated from the major pulses of population change and many isolated
languages have been able to continue unaffected for tens of millennia. Even the status and classification of
language isolates in Africa remains controversial. Table 4 lists the languages that have remained unclassified
and Map 12 shows their locations;

Table 4. African language isolates

Language Name Location Source Comments
Jalaa (=Cuŋ Tuum) Nigeria Kleinwillinghöfer (2001) Isolate
Bangi Me Mali Blench (2007b) Isolate
Laal Chad Boyeldieu (1977) Isolate
Kujarge Sudan Lovestrand (2012) East Chadic
Ongota Ethiopia Fleming (2006) Perhaps Afroasiatic
Oropom Uganda Wilson (1970) Probably spurious
Hadza Tanzania Sands (1998) Isolate
Sandawe Tanzania Elderkin (1983) Probably Khoisan
Kwadi Angola Güldemann (2004) Perhaps Khoe

With the exception of Bangi Me all of these are either foragers or were so until recently, which suggests that
they were marginalised communities, relics of a once more widespread interlocking network of hunter-
gatherers.

23
Roger Blench Africa: the last 12,000 years. Draft
Map 12. Language isolates and residual forager populations in Africa

Source: Blench (in press)

4. Synthesis and the way forward
The above descriptions provide a background to archaeological interpretations of African prehistory. Blench
(2006) proposed a series of correlations with technological innovation and climate which would provide
very approximate date for the early expansion of these phyla. Table 5 is a significantly updated synthesis of
this approach based on the scenarios in individual sections above;

24
Roger Blench Africa: the last 12,000 years. Draft
Table 5. Dates, homelands and causes of phylic expansion in Africa
Phylum Date BP Original Homeland Stimulus of dispersal
Kx’a >? South-Eastern Africa Unknown (? microlithic technology)
Khoe 2500 South-Central Africa Contact with livestock producers
!Ui-Taa >? Southern Africa Unknown
Nilo-Saharan 11,000 SW Ethiopia, Lake Turkana Expansion of aquatic resources
region
Afroasiatic 9,000 SW Ethiopia Management of livestock/enset
cultivation
Niger-Congo 10,000 South-Central Sahara Bow and arrow introduced and
improving climate
Malagasy 1500 Migration from ISEA Trade and quest for natural resources

African languages encompass a record of human interaction with the environment as well as a palimpsest of
social and economic change. The rapid expansion of archaeological and palaeoclimatic data has improved
the potential to interpret language dispersals in a more precise way. Nonetheless, the map still outlines large
language blocks for which there is still no convincing explanation. The next decades can allow research to
explore this is much greater detail.

References
Adelaar, K.A. 1994. Malagasy culture-history: some linguistic evidence. In: The Indian Ocean in Antiquity.
J. Reade (ed.) 487-500. London and New York: Kegan Paul/British Museum.
Bar-Yosef, Ofer 1998. The Natufian Culture in the Levant, Threshold to the Origins of Agriculture.
Evolutionary Anthropology, 6 (5): 159–177. doi:10.1002/(SICI)1520-6505(1998)
Basset, Andre 1952. La langue berbère. London : Oxford University Press.
Bechhaus-Gerst, Marianne 1989. Nubier und Kuschiten im Niltal. Sprach- und Kulturkontakte im ‘no man’s
land’. Afrikanische Arbeitspapiere, Köln, Sondernummer 1989.
Behrens, P. 1989. Langues et migrations des premiers pasteurs du Sahara: la formation de la branche
berbère. Libya Antiqua, 31-53. Paris: UNESCO.
Bellwood, Peter 2013. First migrants. Chichester: John Wiley & Sons.
Bender, M.L. 1988. Proto-Omotic: Phonology and Lexicon. In: Cushitic-Omotic: Papers from the
International Symposium on Cushitic and Omotic Languages. M. Bechhaus-Gerst, F. Serzisko. eds. 121-
162. Hamburg: Helmut Buske Verlag.
Bender, M.L. 1997. 2nd ed. The Nilo-Saharan languages: a comparative essay. Munich: Lincom Europa.
Bernus, E. 1981. Touaregs Nigériens. Unité culturelle et diversité régionale d'un peuple pasteur. Mémoires
ORSTOM no. 94. Paris: Éditions de l'Office de la Recherche Scientifique et Techniques Outre-Mer.
Blažek, Václav 1999. Elam: a bridge between Ancient Near East and Dravidian India? In: Archaeology and
Language, IV. eds. R.M. Blench and M. Spriggs. 48-78. London: Routledge.
Bleek, Dorothea F. 1956. A Bushman dictionary. New Haven: American Oriental Society.
Blench, Roger M. 1995. A History of Domestic Animals in Northeastern Nigeria. Cahiers de Science
Humaine, 31, 1:181-238. ORSTOM, Paris.
Blench, Roger M. 2001. Types of language spread and their archaeological correlates: the example of
Berber. Origini, XXIII: 169-190.
Blench, Roger M. 2002. Besprechungsartikel. The classification of Nilo-Saharan. Afrika und Übersee,
83:293-307.
Blench, Roger M. 2006. Archaeology, Language and the African Past. Lanham: Altamira Press.
Blench, Roger M. 2007a. Enset culture and its history in highland Ethiopia. In: Omotic and Cushitic
language studies. Papers from the IVth Cushitic-Omotic Conference, Leiden April 10-12, 2003. A.
Amha, M. Mous & G. Savà eds. 99-112. Köln: Rüdiger Köppe.
Blench, Roger M. 2007b. Bangi Me: a language of unknown affiliation in Northern Mali. Mother Tongue
XII: 147-178.

25
Roger Blench Africa: the last 12,000 years. Draft
Blench, Roger M. 2008. Omotic livestock terminology and its implications for the history of Afroasiatic. In:
Semito-Hamitic Festschrift for A.B. Dolgopolsky and H. Jungraithmayr. G. Takacs ed. pp. 63-78. Berlin:
Dietrich Reimer. Download.
Blench, Roger M. 2009a. Was there an interchange between Cushitic pastoralists and Khoesan speakers in
the prehistory of Southern Africa and how can this be detected? Special Volume Sprache und Geschichte
in Afrika: Language Contact, Language change and History, based on language sources in Africa.
Wilhelm Möhlig, Frank Seidel & Axel Fleisch eds. Sprache und Geschichte in Afrika, 20: 31-49.
Blench, Roger M. 2009b. The Austronesians in Madagascar and their interaction with the Bantu of East
African coast: surveying the linguistic evidence for domestic and translocated animals. Philippines
Journal of Linguistics, 182: 18-43. S. Brainard ed. Manila: SIL.
Blench, Roger M. 2010. The linguistic geography of Nigeria and its implications for prehistory. In: West
African Archaeology New developments, new perspectives. BAR International Series 2164. Philip
Allsworth-Jones ed. 161-170. Oxford: Archaeopress.
Blench, Roger M. 2013. Does Kordofanian constitute a group and if not, where do its languages fit into
Niger-Congo? Nuba mountains language studies. Thilo Schadeberg & Roger Blench eds. 571-586. Köln:
Rudiger Köppe.
Blench, Roger M. 2014. Linguistic evidence for the chronological stratification of populations South of Lake
Chad. In: Les échanges et la communication dans le basin du Lac Tchad. S. Baldi and G. Magrin eds.
395-420. Naples: Università degli Studi di Napoli L’Orientale.
Blench, Roger M. 2015. The Bantoid languages. Oxford Linguistics Online.
Blench, Roger M. in press, a. The translocation of useful trees in African prehistory. In: Proceedings of the
8th International Workshop for African Archaeobotany, Modena, Italy, 23 – 26 June 2015. Alexa Hoehn,
Katharina Neumann eds.
Blench, Roger M. in press, b. African language isolates. In: Language Isolates. Lyle Campbell, Alex Smith,
and Thomas Dougherty eds. Routledge Language Family Series. London: Routledge.
Blench, Roger M. in press, c. The linguistic prehistory of the Sahara. In: State Formation, Migration and
Trade in the Central Sahara (1000 BC - AD 1500). D. Mattingley, M. Gatto eds. Cambridge: Cambridge
University Press.
Blome, M. W., Cohen, A.S., Tryon, C. A., Brooks, A. S., & Russell, J. 2012. The environmental context for
the origins of modern human diversity: a synthesis of regional variability in African climate 150,000–
30,000 years ago. Journal of Human Evolution, 62(5), 563-592.
Bonneau, A. et al. 2017. The earliest directly dated rock paintings from southern Africa: new AMS
radiocarbon dates Antiquity 91, 322–333;
Bostoen, K., Clist, B., Doumenge, C., Grollemund, R., Hombert, J.M., Muluwa, J.K. and Maley, J., 2015.
Middle to Late Holocene Paleoclimatic Change and the Early Bantu Expansion in the Rain Forests of
Western Central Africa. Current Anthropology, 56(3): 354-384.
Boyeldieu, Pascal 2000. Identité tonale et filiation des langues sara-bongo-baguirmiennes (Afrique
centrale), Köln: Rüdiger Köppe Verlag, (SUGIA-Beiheft 10).
Boyeldieu, Pascale 1977. Éléments pour une phonologie de Laal de Gori. In : Études phonologiques
tchadiennes. J-P Caprile ed. 186-198. Paris : SELAF.
Brett, M. & E. Fentress 1996. The Berbers. Oxford: Blackwell.
Camps, Gabriel 1980. Berbères. Aux marges de l’histoire. Toulouse: Hespérides.
Cancellieri, E. and di Lernia, S., 2014. Re-entering the central Sahara at the onset of the Holocene: A
territorial approach to Early Acacus hunter–gatherers (SW Libya). Quaternary International, 320: 43-62.
Caton-Thompson, G., 1952. Kharga Oasis in prehistory. University of London, Athlone Press.
Creissels, Denis 1981. Songhay et Niger-Congo (mandé). In: Nilo-Saharan. Schadeberg, Thilo C. and M L.
Bender eds. 307-27. Holland: Foris Publications.
De Wolf, Paul Polydoor 1971. The Noun-Class System of Proto-Benue-Congo. The Hague: Mouton.
Drake, Nick A. & Roger M. Blench (in press) Human and faunal dispersal across the Sahara: when and
under what circumstances? In: Human dispersals. M. Petraglia and N. Boivin eds. Springer.
Drake, Nick A., Roger M. Blench, Simon J. Armitage, Charlie S. Bristow, and Kevin H. White 2011.
Ancient watercourses and biogeography of the Sahara explain the peopling of the desert. Proceedings of
the National Academy of Sciences of the United States of America, 108(2):458-62.
Dwyer, David J. 1989. Mande. In: The Niger-Congo Languages. J. Bendor-Samuel (ed.) 47-65. Lanham:
University Press of America.
26
Roger Blench Africa: the last 12,000 years. Draft
Ehret, Christopher 1986. Proposals on Khoesan reconstruction. Sprache und Geschichte in Afrika, 7(2), 105-
130.
Ehret, Christopher 1995. Reconstructing proto-Afroasiatic (Proto-Afrasian): consonants, vowels, tone and
vocabulary. Berkeley: University of California Press.
Ehret, Christopher 2001. A historical-comparative reconstruction of Nilo-Saharan. Köln: Rudiger Köppe.
Elderkin, E.D. 1983. Tanzanian and Ugandan isolates. In: Nilotic Studies: proceedings of the International
Symposium on Languages and History of the Nilotic peoples, Cologne, January 4-6, 1982. R. Vossen &
M. Bechhauus-Gerst eds. 499-521. Berlin: Dietrich Reimer.
Estermann, Carlos [ed. and trans. Gordon G. Gibson] 1976. The ethnography of southwestern Angola,
Volume I. New York: Africana Publishing Company. [original dated 1958]
Faidherbe, Louis L. 1877. Le Zenaga des tribus Sénégalaises. Paris : Leroux.
Fleming, Harold C. 2006. Ongota: a decisive language in African prehistory. Mainz: Harassowitz.
Gabelenz, G. 1894. Die Verwandschaft des Baskischen mit den Berbersprachen Nordafrikas. Braunschweig:
A.C. Graf von der Schulenburg.
Galand, Lionel 1970-1. Unité et diversité du vocabulaire berbère. In A. Giuffre (ed.), Atti della settimana
magbribina (Cagliari 22-25 maggio 1969). 5-16. Milan, 17-20.
Galand, Lionel 1987/88. Berberisch : der Schlüssel zum Altkanarischen? Almogaren, 18-19: 7-16.
Garcea, E.A. ed., 2013. Gobero: The no-return frontier: Archaeology and landscape at the Saharo-Sahelian
borderland (Vol. 9). Frankfurt: Africa Magna Verlag.
Gifford-Gonzalez, Diane ed. 2008. Adrar Bous : archaeology of a Central Saharan granite complex in
Niger. Tervuren: Musee Royale de l’Afrique Centrale.
González Antón, R. & A. Tejera Gaspar 1990. Los aborígenes canarios: Gran Canaria y Tenerife. Oviedo:
Ediciones Istmo.
Greenberg, Joseph H. 1963. Languages of Africa (Indiana University Research Centre in Anthropology,
Folklore, and Linguistics, Publication 25). The Hague: Mouton & Co.
Greenberg, Joseph H. 1971. Nilo-Saharan and Meroitic. In: J. Berry & J.H. Greenberg (eds.) Current Trends
in Linguistics Vol. 7, Sub- Saharan Africa. Mouton, The Hague.
Güldemann, Tom 2004. Reconstruction through ‘de-construction’: the marking of person, gender and
number in the Khoe family and Kwadi. Diachronica, 21(2):251-306.
Güldemann, Tom 2008. A linguist’s view: Khoe-Kwadi speakers as the earliest food-producers of southern
Africa. Special Issue of Southern African Humanities, Karim Sadr and François-Xavier Fauvelle-Aymar
eds. 93-132. Pietermaritzburg: Natal Museum.
Guthrie, Malcolm 1962. Some developments in the prehistory of the Bantu Languages. Journal of African
History, 3 (2):273-282.
Guthrie, Malcolm 1967-71. Comparative Bantu: An Introduction to Comparative Linguistics and Prehistory
of the Bantu Languages. 4 vols. Farnborough: Gregg International Publishers.
Haour, A., Banni Guene, O., Gosselain, O., Livingstone Smith, A., N'Dah, D. 2011. Survey Along the Niger
River Valley at the Benin-Niger Border. Nyame Akuma, 76.
Honken, Henry 2013. Genetic Relationships: an Overview of the Evidence. In: The Khoesan Languages
Huysecom, E. et al. 2009. The Emergence of Pottery in Africa during the 10th millennium cal BC: new
evidence from Ounjougou (Mali), Antiquity 83: 905-17.
Johnston, Harry H. 1919/1922. A Comparative Study of the Bantu and Semi-Bantu Languages. (2 Volumes).
Oxford: Clarendon Press.
Kastenholz, Raimund 1991-2. Comparative Mande studies: state of the art. SUGIA 12/13.107-58.
Kleinewillinghöfer, Ulrich 2001. Jalaa –an almost forgotten language of Northeastern Nigeria: a language
isolate? Sprache und Geschichte in Afrika, 16/17:239-271.
Kossmann, Maarten 2013. The Arabic Influence on Northern Berber. Leiden: Brill.
Leth, T. and Lindblom, G. 1933. Two kinds of fishing implements: the plunge-basket (Stülpkorb) in Africa
and elsewhere, the circular cast-net in Africa. Stockholm: Riksmuseets etnografiska avdelning.
Lewis, M. Paul, Gary F. Simons, and Charles D. Fennig (eds.). 2016. Ethnologue: Languages of the World,
Nineteenth edition. Dallas, Texas: SIL International. Online version: http://www.ethnologue.com.
Lovestrand, J. 2012. Classification and description of the Chadic languages of the Guéra (East Chadic B).
SIL Electronic Working Papers 2012-004. Dallas, Texas: SIL International.
Lux, Cecile 2013. Le tetserrét, langue berbère du Niger : Description phonétique, phonologique et
morphologique, dans une perspective comparative. (Berber Studies 38). Koln: Rudiger Koppe.
27
Roger Blench Africa: the last 12,000 years. Draft
Manning, Katie, Ruth Pelling, Tom Higham, Jean-Luc Schwenniger and Dorian Q. Fuller 2011. 4500-year
old domesticated pearl millet (Pennisetum glaucum) from the Tilemsi Valley, Mali: new insights into an
alternative cereal domestication pathway. Journal of Archaeological Science, 38 (2): 312-322.
Meinhof, Carl 1910. Grundriss einer Lautlehre der Bantusprachen. Berlin: Dietrich Reimer.
Mohammed-Ali, A.S. & A.-R.M. Khabir 2003. The wavy line and the dotted wavy line pottery in the
prehistory of the Central Nile and the Sahara-Sahel Belt, African Archaeological Review, 20(1): 25-58.
Orel, V. & O. Stolbova 1995. Hamito-Semitic Etymological Dictionary. Leiden: Brill.
Oslisly, R., White, L., Bentaleb, I., Favier, C., Fontugne, M., Gillet, J.F. and Sebag, D., 2013. Climatic and
cultural changes in the west Congo Basin forests over the past 5000 years. Philosophical Transactions of
the Royal Society of London B: Biological Sciences, 368(1625):1-10.
http://dx.doi.org/10.1098/rstb.2012.0304
Paradisi, U. 1960. Il berbero du Augila (materiale lessicale). Rivisita degli Studi Orientalia, 35:157-177.
Paradisi, U. 1963. Il linguaggio berbero di El-Fógǎha (Fezzân) (Testi e materiale lessicale). Annali, NS,
13:93-126.
Park, Douglas Post 2010. Prehistoric Timbuktu and its hinterland. Antiquity 84: 1076–1088.
Putten, M.V. 2013. A grammar of Awjila Berber (Libya): based on Umberto Paradisi’s material. PhD.
Leiden University.
Raimbault, M. 1990. Pour une approche du néolithique du Sahara Malien. In: Travaux de LAPMO 1990: 67-
81.
Relandus, Hadrianus [Adriaan van Reeland] 1708. Dissertationum Miscellanearum, Pars Tertia et Ultima.
55-139. Trajecti ad Rhenum: Guilielmus Broedelet.
Rilly, Claude and Alex de Voogt 2012. The Meroitic Language and Writing System. Cambridge: Cambridge
University Press.
Sands, B. 1998. Eastern and Southern African Khoesan: evaluating claims of a distant linguistic
relationships. Quellen zur Khoesan-Forschung 14. Köln: Rüdiger Köppe.
Schadeberg, Thilo & Roger Blench eds. 2013. Nuba mountains language studies. Thilo Schadeberg & Roger
Blench eds. Köln: Rudiger Köppe.
Sereno, P.C., Garcea, E.A., Jousse, H., Stojanowski, C.M., Saliège, J.F., Maga, A., Ide, O.A., Knudson, K.J.,
Mercuri, A.M., Stafford Jr, T.W. and Kaye, T.G., 2008. Lakeside cemeteries in the Sahara: 5000 years of
Holocene population and environmental change. PloS one, 3(8): p.e2995.
Shanahan, T.M., McKay, N.P., Hughen, K.A., Overpeck, J.T., Otto-Bliesner, B., Heil, C.W., King, J.,
Scholz, C.A. and Peck, J., 2015. The time-transgressive termination of the African Humid Period. Nature
Geoscience, 8(2): 140-144.
Simon, Pierre R. 2006. La langue des ancêtres : Ny fiten-dRazana: Reconstitution et périodisation du
malgache ancien jusqu'au XVè siècle. Paris: l’Harmattan.
Skinner, Neil 1996. Hausa Comparative Dictionary. Köln: Rüdiger Köppe.
Skoglund, Pontus et al. 2017. Reconstructing Prehistoric African Population Structure. Cell,
http://dx.doi.org/10.1016/j.cell.2017.08.049
Souag, M. Lameen 2010. Grammatical contact in the Sahara: Arabic, Berber, and Songhay in Tabelbala
and Siwa. PhD. University of London.
Souag, M. Lameen 2012. The subclassification of Songhay and its historical implications. Journal of African
Languages and Linguistics, 33(2): 181–213.
Stojanowski, CM. 2013. An archaeological perspective on the burial record at Gobero. In: Garcea EAA, ed.
Gobero: The No Return Frontier. Archaeology and Landsape at the Saharo-Sahelian Borderland. 44-64.
The Journal of African Archaeology Monograph series number 9. Frankfurt: Africa Magna Verlag.
Sutton, John E.G. 1974. The Aquatic Civilization of Middle Africa. Journal of African History, 15,4:527-
546.
Sutton, John E.G. 1977. The African aqualithic. Antiquity, 51:25-34.
Taine-Cheikh, Catherine 2008. Dictionnaire Zénaga-français (berbère de Maurétanie). Köln: Rudiger
Köppe.
Traill, A. 1986. Do the Khoi have a place in the San? New data on Khoesan linguistic relationships. Sprache
und Geschichte in Afrika, 7,1:407-430.
Van Gennep, Arnold 1911. Les demi-savants. Paris: Mercure de France.

28
Roger Blench Africa: the last 12,000 years. Draft
Vernet, R. et al. 2007. Fouille de la Butte de FA 10 (Banc d’Arguin) et son apport à la connaissance de la
culture épipaléolithique de Foum Arguin, nord-ouest du Sahara. Journal of African Archaeology, 5: 17-
46.
Vossen, Rainer 1997. Die Khoe-Sprachen. Ein Beitrag zur Erforschung der Sprachgeschichte Afrikas. Köln:
Rüdiger Köppe.
Vossen, Rainer ed. 2013. The Khoesan Languages. Abingdon: Routledge.
Vydrin, V. 2009. On the problem of the Proto-Mande homeland. Вопросы языкового родства–Journal of
Language Relationship, 1:107-142.
Vydrine, V., Bergman, T.G. and Benjamin, M. 2000. Mandé language family of West Africa: Location and
genetic classification. SIL Electronic Survey Reports 2000, 3, pp.2000-003.
Westermann, Diedrich 1927. Die Westlichen Sudansprachen und ihre Beziehungen zum Bantu. Berlin: De
Gruyter.
Westphal, E.O.J. 1962. A re-classification of Southern African Non-Bantu Languages. Journal of African
Languages, 1 (1): 1-8.
Westphal, E.O.J. 1963. The linguistic prehistory of Southern Africa: Bush, Kwadi, Hottentot and Bantu
linguistic relationships. Africa, 33:237-265.
Williamson, Kay & R.M. Blench 2000. Niger-Congo. In: African languages: an introduction. B. Heine & D.
Nurse eds. 11-42. Cambridge: Cambridge University Press.
Williamson, Kay 1971. The Benue-Congo languages and Ịjọ. In: Current Trends in Linguistics, Vol. 7. T.A.
Sebeok series ed. 245-306. Berlin: De Gruyter.
Wilson, John G. 1970. Preliminary observations on the Oropom people of Karamoja, their ethnic status,
culture and postulated relation to the peoples of the late stone age. Uganda Journal, 34(2):125-145.
Wölfel, Dominik Josef 1965. Monumenta linguae Canariae. Austria: Akademische Druck.
Wolff, Ekkehardt & Husseina Alidou 2001. On the non-linear ancestry of Tasawaq (Niger), or: how
“mixed” can a language be? In: Historical Language Contact in Africa. Derek Nurse, ed. 523-574. Köln:
Rüdiger Köppe.

29