Lexical reconstruction and early African history: Insights

Download Report

Transcript Lexical reconstruction and early African history: Insights

Lexical reconstruction and
early African history:
Insights from Bantu crafts
vocabulary and plant names
Koen Bostoen
Royal Museum for Central Africa, Tervuren
[email protected]
Bantu expansion
 Bantu : Africa’s largest language group, in terms
of number of speakers and languages as well as
in terms of geographical spread
 Number of speakers: ± 240 million (< ± 400m
NC < ± 750m Africans)
 Number of languages: 440 à 680
 Geographical spread: central + southern Africa
BENUE-CONGO
BANTU
Nurse &
Philippson
2003:2
‘Bantu Problem’ (Eggert 2005)
 Striking contrast between shallow time
depth and widespread distribution
 Lower node in the Niger-Congo tree: subbranch of a sub-branch of a sub-branch of
a sub-branch of Benue-Congo
 Niger-Congo expansion: at least 10.000 to
12.000 bp
 Bantu dispersal: only 4.000 to 5.000 bp
Williamson &
Blench 2000;
Schadeberg
2003
Driving forces behind Bantu
expansion
 Source of multidisciplinary speculation since
1950’s (linguistics, archaeology, evolutionary
genetics, …)
 Farming/language dispersal hypothesis:
Bantu as a textbook case of the concurrent
dispersal of early agriculture, human genes
and languages
Farming/language dispersal
hypothesis
Bantu expansion: one of the clearest examples of the
concordance of five independent types of evidence
for the replacement of local hunter-gatherers by
expanding farmers, all traceable back to the
farmers’ homeland of origin (Diamond & Bellwood
2003)
 Own archaeologically visible culture
 Domesticates
 Skeletal types
 Genes
 Languages
Bantu expansion ≈ spread
agriculture
Holden (2002: 793):
The Bantu language tree reflects the spread of
farming across this part of sub-Saharan Africa
between ca. 3000 BC and AD 500. Modern Bantu
subgroups, ..., mirror the earliest farming traditions
both geographically and temporally
Direct evidence for early ‘Bantu’
farming traditions
 Independent archaeological evidence for food
crop production in earliest Bantu speech
communities is very sparse
 Unambiguous data on crop remains or
agricultural tools are largely missing
 Central African rainforest: Banana phytoliths
from Nkang, Cameroon (840 and 370 BC,
Mbida et al. 2000)
Indirect evidence through lexical
reconstruction
 Comparative study of cultural vocabularies:
important source of circumstantial evidence
on the subsistence economy and sociocultural organisation of early Bantu speech
communities
 Especially in areas where other historical
sciences, such as archaeology or
archaeobotany, are inadequate or have not yet
been exploited to their full potential.
Proto-Bantu reconstructions for
domesticates
 Mainly restricted to ‘yam’ reconstructions: *kʊ̀á , *-bàdá, *-kódò,*-dɪ̀gà (Maniacky 2005;
Philippson & Bahuchet 1994/95)
 Two Vigna species: *-kʊ́ndè (cowpea, Vigna
unguiculata) + *-jʊ̀gʊ́ (Bambara groundnut,
Vigna subterranea) (Philippson & Bahuchet
1994/95)
 Most terms are reconstructable at different
levels of Benue-Congo
Proto-Bantu reconstructions for
domesticates
 Banana: irregular reflexes of *-kòndè, probably
not reconstructable into PB (De Langhe
1994/95; Philippson & Bahuchet 1994/95)
 No PB-reconstructions for ancient
domesticates, such as okra (Abelmoschus
esculentus), roselle (Hibiscus sabdariffa) or
amaranth (Amaranthus sp.) (Blench 2006)
Lexically-based ideas on the
subsistence economy earliest Bantu
speakers
 Root-crop centred agriculture based on
different yam species
 Probably supplemented with two Vigna
species
 Banana?
Ambiguity of crop name
reconstructions
 Crop name reconstructions not unproblematic as
evidence for early plant food production
 Domestication of an indigenous plant does not
necessarily lead to lexical change
 Yam family (Dioscoraceae): many species, both
wild and domesticated, difficult to associate specific
name to specific species  PB reconstructions =
reliable indication of food production?
Analogy with Bantu crafts
vocabulary
 Semi-specialized verbs often constitute the
conceptual center of the semantic field linked to
crafts as pottery, metallurgy,…
 Refer to the craft in its entirety, without being
genuinely generic terms
 More closely linked to the fashioning phase as
prototypic centre of the chaîne opératoire
 Several verbs/nouns linked to different steps of the
production process are derived from these verbs
Analogy with Bantu crafts
vocabulary
 Bantu verbs meaning “to make pottery”, “to
forge” originally had another, more general
basic meaning; underwent semantic shift at ≠
chronological stages
 Their dominant semantic core is quite general,
and thus fairly vague in terms of historical
information  technical changes do not
necessarily lead to lexical changes
 Peripheral, more technical meanings are more
scattered and often created through common
derivational procedures
Reliable lexical evidence for early
agriculture
 the reconstruction of vocabulary related to
farming tools and techniques
 PB reconstructions for utensils generated
from food plants, e.g. containers from
Lagenaria (Bulkens 1999a), and tools for
exploiting food plants, such as grindstones,
pestles, and mortars (Bulkens 1999b)
Reliable lexical evidence for early
agriculture
 PB reconstructions: *-dɪ̀m- ‘to cultivate (especially
with hoe)’, *-tém- ‘to cut; cut down; clear for
cultivation’, *-gʊ̀ndà, ‘garden’  original meaning?
 Absence of unquestionable PB reconstructions for
typical farming practices, such as planting, sowing,
making mounds, weeding and harvesting, or for
typical farming utensils, such as the hoe, the
digging stick and the bush-knife  allows debate
on the state of agricultural advancement at that
time (Jacquot 1991)
Crop-centred lexical approaches
 Biased focus on vocabulary for food crops
 Little attention for wild or semidomesticated food plants  their role in
the subsistence systems of early Bantu
speech communities often overlooked
 More attention needed for mixed nature of
these subsistence economies
Proto-Bantu lexical evidence for
the exploitation of wild trees
Bostoen (2005, forthcoming):
 Elaeis guineensis (oil palm): *-bídà +*-téndé (ProtoBC)
 Canarium schweinfurthii (safoutier): *-pátù (ProtoEast-BC) + *-bɩ́dɩ́ (PB)
 Dacryodes edulis (African plum): *-cákú (PB)
 Cola spp. (cola nut): *-bɩ̀dú (Proto-BC?)
 Parinari curatellifolia (mobola plum): *-bʊda (PB?)
Subsistence strategies of early
Bantu speech communities
 Necessity to get off the beaten tracks instead
of repeating weakly supported ideas on the
driving forces behind the Bantu expansion
 Use of lexical data to acquire a more refined
idea of early Bantu subsistence strategies
 Lexical ‘evidence’ should speak for itself
and not purely serve as confirmation of
(pre-established) historical theories
Proving the impossible: pearl millet
cultivated by early Bantu speakers?
 Common belief: pearl millet (grain
cultivation in general) not part of original
agricultural traditions of Bantu speakers
 Challenged by discovery of archaeobotanical
evidence of domesticated pearl millet
(Pennisetum glaucum) in the rain forest of
southern Cameroon, dated between 400 and
200 BC (Eggert et al. 2006)
Bantu pearl millet vocabulary as
evidence for early agriculture
 No wild pearl millet varieties attested in the
Bantu area
 Centre of early domestication: southern fringe of
the Sahara (Brunken et al. 1977, Harlan 1971);
southern Mauritania, Senegal and eastern Mali
(Tostain 1998)
 Bantu lexical reconstructions for ‘pearl millet’
certainly designate a domesticate having
undergone man-induced spread
Earlier lexical approaches to
‘Bantu’ history of pearl millet
 Unanimous on the fact that the first Bantu
speakers leaving the Nigerian-Cameroonian
borderland were not familiar with this cereal
 Bantu speakers first acquired it when the crop
started to spread in the eastern part of Central
Africa, only from about the beginning of the
first millennium AD (Ehret 1974, 1998; Vansina
1990, 1994/1995, 2004; Philippson & Bahuchet
1994/1995)
Earlier lexical approaches to
‘Bantu’ history of pearl millet
 Introduced among Bantu speakers from a single
eastern centre of origin
 Either more or less concomitantly with part of the
Bantu language dispersal after an ancestral Bantu
speech community (Proto-East-Bantu) had adopted
the cereal (Philippson & Bahuchet 1994/1995)
 Or independently, from one Bantu group to the
other, after Bantu languages had spread over large
parts of their current distribution area (Ehret 1974)
The lexical evidence
 Previous studies quote 2 very common ‘pearl
millet’ terms reconstructed as *-bèdé and *-cángʊ́
(Bastin & Schadeberg 2003; Ehret 1974; Guthrie
1967-1971; Homburger 1925; Philippson and
Bahuchet 1994/1995;Vansina 1990, 1994/1995,
2004)
 complementary geographic distribution
 *-bèdé : the eastern part of the Bantu domain
 *-cángʊ́ : almost exclusively western distribution
The lexical evidence
 Both terms may also refer to sorghum, finger
millet and maize, but ‘pearl millet’ is in both
cases the predominant meaning
 None of them thought to have a distribution
among Bantu subgroups that is representative
enough to be reconstructable into PB
 Both terms are very widespread
 Reflexes are overall phonologically regular
Historical implications
 Introduction pearl millet among Bantu
speakers must be relatively old (e.g. sorghum
and maize vocabulary much more diverse)
 Its spread must have been concomitant with
certain stages of the Bantu language dispersal
(no loanwords diffused across languages)
 Two terms: why assume a single eastern
point of origin?
New look at the lexical evidence:
*-bèdé
 Reflexes in Guthrie’s zones D, E, F, G, J, K, L, M,
N, P and S: East-Bantu
 Reconstructable to Proto-East-Bantu
 Spread of pearl millet coincided with spread of
East-Bantu languages from the East African Great
Lakes region (Philippson & Bahuchet 1994/1995)
 Ultimate origin: Nilo-Saharan loanword > Central
Sudanic (Ehret 1973, 1974), Sog (Eastern Sahelian)
(Ehret 1998)
New look at the lexical evidence:
*-bèdé
Language
Term
Meaning
Source
Nuer
bɛ́l
‘durra’
(Storch 2005: 203)
Anywa
bɛɛl
‘durra’
(Storch 2005: 227)
Belanda Bor
bɛ́l
‘sorghum, durra’
(Storch 2005: 304)
Thuri
bɛ̤́́ɛ̤́ ́ l
‘durra’
(Storch 2005: 299)
Alur
bɛ́l
‘sorgho’
(Ukoko et al. 1964: 28)
Dholuo
bél
‘sorgho’
(Tucker 1994: 495)
New look at the lexical evidence:
*-bèdé
 Widespread in West-Nilotic: also in Päri, Luwo,
Shilluk, Adhola, Chopi, Labwor, Kuman, Lango,
Acholi (Anne Storch pers. comm.)
 Meaning: SORGHUM !!!!
 Complicates the cultural historical interpretation
of the comparative lexical data: did Bantu
speakers adopted sorghum and/or pearl millet
from Nilotic speakers?
New look at the lexical evidence:
*-cángʊ́
 exclusively western Bantu distribution
 most reflexes occur among South-WestBantu languages
 only area in the western Bantu half that is
entirely savanna and where climatic and
ecological conditions are generally favorable
to pearl millet cultivation
New look at the lexical evidence:
*-cángʊ́
 Phonologically regular reflexes meaning
‘pearl millet’ occur in more West-Coastal and
Inner-Congo-Basin Bantu languages than has
usually been accepted
 Fewer in number: most often languages in
transition areas between the rain forest and
the savanna, such as the Bandundu and
Lower Kasai areas in DRC
New look at the lexical evidence:
*-cángʊ́
 South-West Bantu, West-Coastal Bantu
Inner-Congo-Basin Bantu often seen as
forming a larger unit, i.e. ‘Narrow West
Bantu’
 Absolute time depth?
 Relative time depth: possibility to reconstruct
*-cángʊ́ in their most recent common ancestor
New look at the lexical evidence:
*-cángʊ́
 (Phonologically regular?) reflex of *-cángʊ́ also
attested more northerly in the western half, i.e. in
the Lebonya/Boan Bantu subgroup
 The evidence is weak (1 attestation), but this is
rather unsurprising, since this subgroup is very
poorly documented and mainly occurring in a
rainforest environment
 Kumu language: rain forest-savanna transition
area in the Maniema and Orientale Provinces of
DRC
Language
Term
Meaning
Source
Pindi (B85)
masangu
‘millet’
(Obenga 1985)
Nsong (B85)
másàŋ
‘pearl millet’
(Koni Muluwa 2006)
Pende (L11)
disangu
‘millet’
(Gusimana 1972)
Kwezo (L13)
másàngù
‘millet’
(Forges 1983)
Bushong (C83)
maʃáaŋ
‘millet’
(Vansina 1959)
Lele (C84)
masangu
‘millet’
(Obenga 1985)
Wongo (C85)
masangu
‘millet’
(Obenga 1985)
Gesogo (C53)
-sangu
‘millet’
(Harries 1955)
‘millet’
(Harries 1955)
Kumu (D37)
West-Coastal
muyáŋgu
Inner Congo Basin
Lebonya/Boan
New look at the lexical evidence:
*-cángʊ́
 Lebonya/Boan Bantu subgroup: status
ambiguous, but possibly a primary offshoot of
the Bantu nucleus
 Relative time depth: *-cángʊ́ ‘pearl millet’ was
part of Bantu vocabulary long before the Bantu
languages started to spread over the southwestern savannas
 Reconstructable to Proto-Bantu?
 Can be answered affirmatively if one finds
cognates beyond Narrow Bantu
Non-Narrow Bantu reflexes of
*-cángʊ́
 Bagangu, Bafut, Nkwen, Bambui, Awing
languages of the Ngemba-cluster (Grassfields
Bantu, S-Bantoid): -sāŋ ‘maize’ (Leroy 1980)
 Proto-Eastern-Grassfields: -sáŋ ́ ‘maize’
(Hyman 1979)
 Ring (Western Grassfields):-sáŋ/-ʃáŋ ‘maize’
in several languages (Hyman & Jisa 1977)
 Reconstructable in Proto-Grassfields: which
meaning?
Cross-River reflexes of *-cángʊ́ ?
LeYigha (Upper-Cross) nsaŋe ‘maize’
Legbo (Upper-Cross) nzana ‘maize’
Usakade (Lower-Cross) úsân ‘maize’
(Blench et al. 1994): ~ -caam (possibly
Adamawa loanword)  link with
*cángʊ́ ?
New look at the lexical evidence:
*-cángʊ́
 Existence of possible cognates beyond Narrow
Bantu: Cross River attestations doubtful, but
Grassfields Bantu attestations more promising
 Problem: all signify ‘maize’  what was
original meaning? Possibly ‘pearl millet’, but
impossible to prove, as long as no ‘pearl millet’
attestations are found in the same groups
 Do we have proper ‘pearl millet’ terms for these
languages?
Historical implications
 The term *-cángʊ́ for ‘pearl millet’ is definitely
older than was supposed so far
 Relative time depth: certainly ‘early western
Bantu’, maybe Proto-Bantu
 Unlikely that the current-day Bantu distribution
of *-cángʊ́ reflects the introduction of pearl millet
from the East
 A independent early western introduction is
more likely
Historical implications
 Was pearl millet cultivation part of
agricultural traditions of earliest Bantu
speech communities?
 Lexical evidence: possibly
 Archaeobotanical evidence: possibly
 More unequivocal evidence, both lexical
and archaeological, is needed
Methodological lessons
 Consider as many languages as possible
instead of ‘representative’ sample
 Do not solely rely on earlier large-scale
lexical reconstruction databases as Guthrie
(1967-71), Bastin & Schadeberg (2003)
 Search data in all Bantu subgroups, also
‘Wide Bantu’ and if possible, wider
Benue-Congo and Niger-Congo
Methodological lessons
 Cultural vocabularies (plant names, crafts
vocabulary, …) first need to be carefully
studied in their own right, and only then be
used as a means to reconstruct broader and
encompassing human histories
 An integrated onomasiological and
semasiological approach of the diachronic
evolution of a global lexical field instead of
focussing on isolated individual words
Methodological lessons
 Lexical ‘evidence’ should speak for itself and
not purely serve as confirmation of (preestablished) historical theories
 Take into account limits of lexical
reconstruction and ‘W&T-method’: semantic
vagueness, semantic shifts, lexical
fragmentation
Methodological lessons
 Realize that “… language [can] be used only
as a diacritic, not as a primary source for
reconstruction of early culture…” (Lehmann
1970)