Roger Blench

IS NIGER-CONGO SIMPLY A BRANCH OF NILO-SAHARAN?

From an early period, scholars have noticed a series of onological, between the Afiican

e phyla today called Niger-Congo and Nilo-Saharan. The of an argument for the coherence of Niger-Congo may be

to Westermann's (1 927) 'Die westlichen Sudansprachen', while e the recognition of Nilo-Saharan to Joseph Greenberg (1966, However, as early as 19 1 1, Westermann had joined the two

exical data, 'Die Sudansprachen' rsen (1972) put forward both

of lexical isoglosses as evidence macro-phylum for which he proposed the name Kongo-Saharan', njoin Niger-Congo and Nilo-Saharan.

Creissels (1 98 1) listed the many morphological and lexical ies between Mande and Songhay to raise doubts about the

of Niger-Congo and Nilo-Saharan. These similarities are ot be explained by chance

ever, Creissels' examples seem to conflate ancient sented here is correct) shared 81:3 16) notes the similarity

n Songhay kuuru and Mandinka kulu for 'skin/hide7. However, in 1972 that similar forms were recorded widely

of the ongin of Songhay listed many more 'rapprochements' betwc~l~ Mande and Songhay as well as systematically exploring the links w11l1 w~in basic itemr was more than could be accounted fbr by bychaaec. Saharan. More recently, Nicolal( 199 1) has established the comparlsr 11 I Bwover, he made no proposals for a genetic or historical schema to with Tamazight and claims that the language may have begun , I mtlt far this situation. If Nilo-Saharan and Niger-Congo are to be Berber before being Mande-ised. The conclusion, that Songllr~\ evolved as a creole used for intercommunication, certainly explains 1 I I ~ deep penetration of loan-words, but the argument for dismissing I I I ~ Nilo-Saharan affiliations is still opaque.

Not all authors have agreed with the methods or evidence us(*ll This paper2 argues that Niger-Congo and Nilo-Saharan are to support the existence of such a large macrophylum. Boyd (199~1 used the numerous overlapping glosses in languages from differc~~~ phyla in northern Carneroun to question the methodology ( 1 1 classification. More recently, Cloarec-Heiss (1992), in a study of the

shared isoglosses of Banda (Ubangian) and Central Sudanic has raisctl similar queries. : , ,' .

In order to discuss the hypothesis, the text refer to it before any proof is offered. This seems an i p p r o w $Q propose a new name for the macro-phylum, Nieer-balguqq~, rj;rnersen'u 'Kongo-Saharan' has been occasionally used, but it jdqa ). L tw@her . the two second terms in the convmtiond m e 8 apd giy . I, *leadins impression to non-specialists of the loution fl t b Proto- Niger-Saharan would then be abbreviated PNGS to COW it with Proto-Nilo-Saharan (PNS).

There has been relatively little fbrther progress on the ,' : , development of the Niger-Saharan hypothesis, in part because the expansion of data has made the whole project more d i c u l t to Evidence for such a broad hypothesis comes in three forms, encompass. Scholars such as Bender (1981) projected such a cal, morphological and lexical. Oregersen discusses unification, although only hinting at the evidence. In part, also, simply demonstrating the unity of Nilo-Saharan has not proved easy; shared innovations to demonstrate the unity of the phylum are elusive', and

P tho particles and morphemes are extremely short; vowel

some scholars remain dubious over the inclusion of families such as FI 11 r cspondences are uncertain and the case becomes difficult to make.

Kado and Kuliak. I 111ld like to thank Kay Williamson f i r bar comments on thne dr@ of this

1 Or at least published tabulations of evidence are lacking I am gat&l to -Liontl Bclrdcr and CWs ahtet fbr mhments arbd

1 1ons on the text.

with Kordofanian as the 'Tumtum' group but noted at the time that '11

shows considerable divergence'. Schadeberg (1 98 1) questioned t h ~ \ classification and suggested a Nilo-Saharan affiliation. Matsushitl\ (1984, 1986) was content to repeat Greenberg, but Dimmendaal ONOLOGICAL EVIDENCE (1987) argued this case in more detail after the publication of Reh'., (1985) grammar of Krongo. Finally, Stevenson (1991) set out the cast. There are two main pieces of evidence from the phonological for Nilo-Saharan affiliation using unpublished lexical data. Thl,. argument is now generally accepted (i.e. in Bender 1991b) althoupll the closer affiliations of Kado are still uncertain. Ehret (1992) has agall~ raised the question of the relationship to Nilo-Saharan.

The internal structure of Niger-Congo is broadly agreetl although doubt remains on the historical priority of several branchv- Figure 2. shows the 'tree' of Niger-Congo, based on Bennett and Stel I. (1977) and the contributions to Bendor-Samuel (1989) adapted ant1 hrther developed using recent unpublished material. VlI IEINCE FROM VOWEL HARMONY SYSTEMS

Figure 2. The Principal Subdivisions of Niger-Congo

Proto-Niger-Congo* + Kordofan~an

Mande-Congo* Mande -1

Atlantic ongo*

I Atlantic

Ijoid Dogon [?I

Volta-Congo*

I North Volta-Congo* t ia not my purpose to discuss the phonetic interpretation of

++ Kw+co"g,,

t rimply to observe that it is present in both Nilo-Saharan and o languages. The + or - ATR vowels most commonly form

Km

Pre Gur-Adamawa RMB December 1992

The most significant feature of this is that Kordofanian i1gi1111 becomes the primary branching of Niger-('o~~go, a reprisc ( 1 1 € h i theat: N~stems arc ilnicluc in thc languages ofthe world.

Greenberg's original hypothesis Although 1l1c. I r l t L u w~tlr the rcsl I I I I'he li~st c~lteliil~t to ~ - O I I I O H I C Ill[: two phyla is Hall et al. (1974)

NIK~I-C'OIIHO crlr still scccptcd. 11ic wcllh~~c-99 1'1 111111 11 oI'01e C V I ~ ~ ~ . I I I pt U V I ~ I C N I ~ I I W I I I I 1nI)lr 01 c*an~rll~lr,.r ol' vowel Iia~ liiot~y in Nilo- I I I ' \ v 111c'11 ,l ,11,1 \I ll', 'llplllll' rllllly , l ( ' l (* t ( I v " Y I I I , ,' 11lc'y

represents a hypothetical

2. g

2 R

EW

\O

o

m-

=

(D

F g

7.

em

=

TE

(D

c

5

E-

5.""

-Y

oJY

m

=h

e

8 3

-

?? =

%

o s

s 0

z

--

V)

-w

P sy-

;;t 0

0

-- u

5 %

$

E (D

=

=i'?

=*

$ *

To

y 9.

o

=w

E

Fg

c(

D 1

=-'

g

P 3 q.

0 = s

g

?(

Dm

5. EVIDENCE FROM LEXICAL ISOGLOSSES

Establishing lexical isoglosses is essentially a matter of cross- comparing large compilations of data. There are three primary sources for suggestions, Westermann (191 l), Gregersen (1972) and Creissels (1981). These have been checked, new materials added and compiled into data tables showing comparative series.

The situation for basic data in the case of the two phyla is very disparate. In the case of Niger-Congo there are two major works setting out comparative data tables, Westermann's 'Die westlichen Sudansprachen' (Westermann 1927) and Mukarovsky's reconstructions of 'Western Nigritic' (Mukarovsky 1976-7). Greenberg refers thc reader to Westermann for the evidence of the unity of Niger-Congo.

Both of these works have problems. Westermann Bg ant1 Adamawa-Ubangian as well as Kordofanian, whereas Mukarovsky deliberately omits these languages and Mande (which he is alone ill considering as related to Afro-Asiatic). Nonetheless, they represen1 major compilations of data which make possible comparisons ol widespread roots.

Nilo-Saharan is much less well-served, despite a majo~ expansion of publication in recent years. Greenberg's comparativl. series are inevitably the starting point, although his 'Chari-Nile' is now generally discounted. Bender (198 1, 1989b) has begun the process 0 1 seeking out isoglosses, although only the case for pronouns is set O ~ I I

in full. There are, however, a number of individual studies that can IN- used to establish comparative series;

Songhay Kanuri Maba Fur Central Sudanic East Sudanic Bender (in&b) Kado Berta Kunama Kol~lrl/ SlllllMl I.. 1111,lh

Ducroz & Charles (1978), Creissels (1981), Nicolai ( I 9H.l I M e r & Hutchison (1990) Edgar (1991) Beaton (1 968), Jakobi (1990) Greenberg (1966), Bender (1992, ined,a) TheIwall(1981), Fleming (1983a). Dimmendaal(1988).

Matsushita ( 1 984. 1986). Rch ( 1085) Bender ( I 989c) Caslclnrlovo ( 1050) l3c11dcr ( I ON 1 ) I i l c ' ~ ~ ~ ~ ~ ~ y , ( I ') ') I ) I r1c11 I I 1 1 1 ~ 1 1 1

I I I I I I I ~ 1 I ' J H

The search for isoglosses shared between Nilo-Saharan and -Congo has yielded two major results: items common through of Nilo-Saharan also widespread in Niger-Congo, and a series of

llosses that link Eastern and Central Sudanic, Kado and Niger- ango. This supports the evidence from vowel harmony systems.

The lexical data is set out in Appendix 1. It is divided into two or sections; a series of general Niger-Saharan isoglosses and ngo-Sudanic' isoglosses. The evidence for Niger-Congo is generally

nted in a more summary form, making use of existing structions, where these are supported by data tables6.

There are certain words that have been previously suggested as ~rible isoglosses for Nilo-Saharan. However, these words are also

~d in parts of Afro-Asiatic. They are therefore probable ancient -words and their tendency to 'jump' phyla argues that they cannot uoed as evidence for genetic classification. The three words didered here are 'fire', 'dog' and 'pig' (Appendix 2).

A comparison with reconstructions of Proto-Cushitic (Ehret, also produced a few common forms. The case of 'flyljump' is

ully striking as PC *pur- corresponds almost exactly to many NS NC forms. There are two possible explanations, apart from

8) an early loan to or from PC b) a general tendency to form words for 'fly' in this way (cf IE Irrre) through sound-symbolism

he second explanation is more attractive in many ways, simply e of the extra-African parallels. Appendix 3 considers the cs for 'fly' and 'cough' in the perspective of external cognates. In me way, Cushitic *fu 'blow' parallels forms in Niger-Congo, but uld clearly be dangerous to regard this as evidence for a phylum or even an early loan.

iuctln~l, ICCOIIN~IIIL lrtl It11 111s QIIIIIIIII 11111 OF IIIP(I m IIIICCI OVI~CIIL'C 1011i9,\,\ IIIC

The PNGS roots for 'knee' and 'tortoise' are also attested in Cushitic (see data tables). Since these do not appear to be Afroasiatic roots, they are most economically explained as loans into Cushitic, assuming that the dispersion and diversity of Nilo-Saharan is a reliable indicator of its antiquity.

One of the principal reasons why Kado (Kadugli-Krongo) was considered part of the Kordofanian cluster is the presence 01' alternating CV prefixes on nouns. Although Schadeberg (198 1) has argued that these cannot be compared directly with Niger-Congo sincc they play a role in a gender system, it is also true that some of them look remarkably similar to Kordofanian prefixes (op. cit, 299) especially to Talodi. However, the relatively low level of lexical similarity between Kado and Kordofanian and the absence ol convincing parallels for the CV alternation elsewhere in Nilo-Sahara11 inevitably suggests the speculation that these arose from borrowing.

However, in Central Sudanic and possibly beyond there arc distinctive traces of an affix system apparently similar to Niger-Congo One of the most notable class affixes that surfaces almost through0111 Niger-Congo is the ma- single gender for liquids or mass nouns (scc Greenberg 1966:lO). In Kordofanian, the g- classes for mass nouns correspond to ma- classes in other branches of Niger-Congo (Black and Black 1971:11, Schadeberg 1989:72). For this reason, Williamsorl (1989a:42) proposed a labial velar as the orginal initial consonant ol the affix. In the case of Niger-Congo class 6A, this would be *gW ,

giving the g- and ma- by different routes.

Gregersen (1972:75) discusses this point but provides no evidence for the ma- affix, mentioning only the presence of the velar nasal in some forms for water. However. stevenson' notcd that ma- is

pesent as an affix in a number of Central Sudanic languages, as a wrker of mass or as a collective. However, if the ma- is accepted as @m present in Nilo-Saharan, then such a reconstruction of initial *gm CYguld seem to be problematic.

able 1 compiles these examples from East Sudanic and some other lches of Nilo-Saharan.

hive been checkcd wllcrc possiblc against published material, but some are ly quolcd f'rolt~ 111s ow11 Ilcltl tlrlttt

Examples of ma- outside Central and East Sudanic are mmewhat thin and may perhaps be excluded as coincidence. The analysis of the m- as a remnant of a productive affix is its absence in aognates in closely related languages. For example, the Bongo word br water, drh, appears as ini in Baka. The attestations outside East md Central Sudanic may be genuine retentions, loans or accidental wremblances. If they are indeed regular cognates then this affix may be present through most of Nilo-Saharan.

The evidence presented in the appendixes is intended to show

ent fiom each of the two current phylum names.

would then be the group closest to Proto-Niger-Congo.

re 3. Proposed Niger-Soharan 'tree': Minimal Hypothesis

This tree makes no hypothesis about the internal classification of the left-hand (Songhai to Shabo) grouping.

Two historical conclusions are implied by these hypotheses;

a) the component families of Nilo-Saharan are significantly older than Niger-Congo

b) the homeland of Niger-Congo is probably significantly east of its usual suggested site

Assigning dates to language phyla is notoriously problematic and certainly no mechanical method such as glottochronology is likely to yield significant results. Nonetheless, it is tempting to try to correlate major periods of language evolution with ecological change

In the past 20,000 years, the two principal events in the eco history of Africa have been the beginning and end of the Holocene, all epoch of prolonged aridity, usually dated 20-12,000 B.P (Street X Gasse, 1981). This probably implies a hunting economy based on highly dispersed populations and may well be reflected in the present day scatter of Nilo-Saharan subgroups and the deep divisions betwee11 thems. Intractable remnant languages such as Ongota, Laal, Ndeewc. Kujarke, Hadza and Sandawe probably are all relics of this period.

Evidence for cultural practices from the PNGS reconstructior~v is fairly limited. 'Turtle' and 'frog' seem to be present at a deep timc level, arguing the importance of riverine resources. Poor lexical di1111 means that the potential to reconstruct pan-African fish species such i1.l

tilapia is limited. So far, it has not proved possible to reconstruct t l ~ c names of any hunting implements into Niger-Saharan. However, thrir is the possibility that 'canoe' will reconstruct to Niger-Central Sudnt111 If this is correct, then this node may be identified with the grat1111rl improvement in the climate after 12,000 B P The bow and at lo\+

which appears in North Africa by I I ,000 I I , rcconstt ~ t c I -1

convincingly back to Proto-Mande Congo and no further. Interestingly, there is no comparable reconstruction possible for the more scattered Nilo-Saharan, suggesting major dispersal took place before the technology spread south of the Sahara. Previous writers, noting the concentration of families in West Mica, have tended to assume a location somewhere near the headwaters of 'the Niger and explained Kordofanian by the migration of a single

roup. If the present classification is accepted, it becomes far more Cely that the homeland was in the centre of present-day Sudan and thrt Kordofanian represents the Niger-Congo speakers who stayed at

The idea of a Niger-Saharan macro-phylum has been present in ackground of African classification studies, but few scholars have

use of it, generally staying with the phyla proposed by berg. However, the evidence for a Niger-Saharan now seems to

ore. The argument of this paper is Niger-Congo is the aharan most closely related to the Central Sudanic

ges. There is a parallel with Greenberg's placing of Bantu as a branch of Benue-Congo and the initial resistance this excited

Bantuists. Researchers in both Niger-Congo and Nilo-Saharan now consider the detailed implications of such a conjunction.

A wide-ranging hypothesis of this type should be viewed more rtimulus to further research and revision than as a completed

still major gaps in the descriptive literature as well absence of detailed lexicons of some of the key languages. ver, as noted above, it is extremely difficult at this great time- to distinguish between true genetic affiliation and extensive

win8 Nonetheless I hope this paper will be considered a

Anbessa, Teferra ined. - Shabo wordlist. Typescript.

Beaton, A.C. 1968 - A grammar of the Fur language. Linguistic Monograph Series, I . Khartoum University.

Bender, Marvin L. 1975 - Omotic: A New Afioasiatic Language Family. Universilj Museum, Southern Illinois University, Carbondale. 1976 - Nilo-Saharan Overview. In Bender, M. L. & Fleming. H.C. (eds.) The non-Semitic languages of Ethiopia. Michigall University Press. : 439-483. 1981 - Some Nilo-Saharan isoglosses. In Schadeberg, T.C. airtl Bender, M.L, (eds.) : 253-267. 1983a (ed) - Nilo-Saharan Language Studies. Michigan Stutc* University Press. 19833 - Proto-Koman Phonology and Lexicon. a k a unit Ubersee, LXVI: 259-29 7. 1989a (ed.) - Topics in Nilo-Saharan Linguistics. Helmut Bu.\~I, Verlag, Ham burg. 19893 - Nilo-Saharan pronouns/demonstratives. In Bendur M.L, 1989a : 1-34 1989c - Berta lexicon. In Bender, M.L, 1989a : 271-304. 1991a (ed) - Proceedings of the Fourth Nilo-Saha~i~~r Conference. Bayreuth, 1989. Helmut Buske Verlag, Hamburg 1991 b - Subclass~fication of Nilo-Saharan. In Bender, M I 1991a : 1-36. 1991c - Genetic Classification of Nilo-Saharan. Paper fir 111r

Table Ronde du rdseau 'Dij$$sion lexicale en zone Suhcli~ Saharienne '. Nice. 1992 - Central Sudanic Segmental and 1,exical Recon.struc./~i J ~ I

Afrikanisticsche Arbeitspapicre, , , 29.3 6 1

med, a - Central Sudt~nic ~uoylosscs / ~ / J ~ ~ A ~ I I I ~ ~

ltlcpt/, h - I.HN~ St~dtt t~ i~ ~ N O H ~ O Y Y ~ ' ~ 1 \ ' / l t t \ d l l /b l

Bender, Marvin L., and Harold C. Fleming (eds.) 1976 - The non-Semitic languages of Ethiopia. Michigan University Press.

Bendor-Samuel, John (ed.) 1989 - The Niger-Congo languages. University Press of America, Lanham.

Onnett, Patrick R., and Jan P. Sterk 1977 - South-Central Niger-Congo: a reclassrfication. Studies in African Linguistics, 8,3 : 241-273.

@fry, Jack, and Joseph H. Greenberg (eds.) 1971 - Current Trends in Linguistics Vol. 7, Sub-Saharan Africa. Mouton, The Hague.

I971 - The Moro Language: Grammar and Dictionary. Linguistic Monograph Series, No. 6. Sudan Research Unit, tlniversity of Khartoum, Khartoum.

In prep - A subclassrfication of the Benue-Congo Languages derivedfrom isoglosses. Typescript.

h, Roger M., and Kay Williamson In prep - Bantoid revisited. Typescript.

19617 - Les Langues Fer (Kara) et Yulu du Nord Centrafiicain.

/u 78 - A pro/)os ( A t * ~r~v~v:~~rrhlrrrrces lexicales entre lanpes N t p t s - ( 'O~IAVJ (*/ NIIo-~Y, //I' 11 /a1tl dYb.'/A/l; 65, /'(~r~,s.

--

Castelnuovo, G.F. 1950 - Vocabulario della lingua Cunama Curia Generalizia do1 Cappucini, Roma.

Cloarec-Heiss, France 1992 - Comparaison lexicale au sein des languages oubangiennes Paper for the Fifth Nilo-Saharan Conference, Nice, Augu.~~, 1992.

Comrie, Bernard 1981 - The languages of the Soviet Union. C'ambri& University Press.

Creissels, Denis 1981 - Songhay et Niger-Congo (mandk). In Schadeberg, 1; '

and M. L Bender, eds. : 307-327.

Cyffer, Norbert, and John Hutchison 1990 - Dictionary of the Kanuri Language. Foris, Dordrecht.

Dirnmendaal, Gerrit J. 198 7 - Krongo: Between Universal, Areal and Genetic Norttl \ Journal of Afncan Languages and Linguistics, 9: 161-1 77. 1988 - The lexical reconstruction of Proto-Nilotic: A , / ; I

reconnaissance. Afrikanistische Arbeitspapiere, 16:5-67.

Ducroz, Jean-Marie, and Marie-Claire Charles 19 78 - Lexique Sog ey (Songhay) Franqais. Harmattan, Paris.

Dwyer, David J. 1989 - Mande. In Bendor-Samuel, J. (ed.), 1989 : 46-65.

Edgar, John T. 1991 - Maba-group lexicon. Reimcr, Hcrlrrr.

he und Geschichte in Afi-ika, 8: 7-180. 1989 - Subclass~fication of Nilo-Saharan: a proposal. In Bender, M.L, 1989a : 35-50 1992 - Does Krongo fit into Nilo-Saharan? Paper for the Fifth Nilo-Saharan Conference, Nice, August, 1992.

I971 - The origin of the domestic animals of Africa. (2 vols) A fricana Publishing Corporation, New York.

M. (eds.) Nilotic Studies. : 523-555. l9R3b - Kuliak External Relations: Step One. In Vossen, R. & Bechhaus-Gerst, M. (eds.) Nilotic Studies. : 423-478. 1991 - Shabo: presentation of data and preliminary rluss~fication. In Bender, M.L. (ed), 1991a : 389-402.

I986 - The Papuan Languages of New Guinea Cambridge Language Series, CUP.

- The Languages of Atiica Published for Indiana

reenberg (eds.), Current Trends in Linguistics Vol. 7, Sub- rrkrran Africa. Mouton, The Hague.

I986 - A preliminary Sketch of Kadugli Vocabulary. I. Verb and

Halaoui, Nazam, Kalilou Tera, and Monique Trabi Verbal Sentences. In Sudan Sahel Studies 11. ed. M. Tomikawa.

I983 - Atlas des languages mande-sud de CGte d'lvoir~~ ILCAA, Tokyo : 1 11-1 38. A CCT/ILA, Abidjan.

Hall, Beatrice L., R.M.R. Hall, M.D. Pam, A. Myers, S.A. h e l l , and 1980 - Bantu lexical reconstructions. MRAC, Tewuren. G.K. Cherono

1974 - African vowel harmony systems from the vantage point 01 Kalenjin. Afrika und Ubersee, 5 7, 4:241-26 7.

Hutchison, John P. 1981 - The Kanuri language. African Studies Progrcrlll Unrversiv of Wisconsin, Madison. 1984 - Preliminaires a une etude sur l'origne du Songhay.

Jakobi, Angelika 1991 - Les apparentements genetique du Songhay: elements 1990 - A Fur grammar. Helmut Bush Verlag, Hamburg. d'analyse. In Bender, M. L. (ed.), 199 Ia : 3 7-56

Kleinewillinghofer, Ulrich I990 - Aspects of Vowel Harmony in Waja and Tangale- W ~ I I ~ I 1989 - La Langue dyes Aiki. Geuthner, Paris. common vocabulary. Frankhrter Afrikanistische Blatter, 2 .') 1

106. on, Andrew M., and R. Janet 1991 - Mod0 -English dictionary with Grammar. SIL, Sudan.

Le Mbaindo, D.N., and Jacques Fedry 19 79 - Lexique ngam b@-Fran~ais, Fran~ais-ngambe. Co1li~)re Charles Lwanga, Sarh.

1985 - Die Krongo-Sprache (niino modi). (KBA 12). Relmer,

Lindau, Mona, Leon Jacobson, and Peter Ladefoged I972 - m e Feature Advanced Tongue Root, UCLA Wor L 111

Papers in Phonetics, 22: 76-94. 1982 - Die siidnilotischen Sprachen: Beschreibung, Vergleichung

Maddieson, Ian und Rekonstruktion Reimer, Berlin.

1984 - Patterns of Sound. CUP.

Matsushita, Shuji 1984 - A preliminary Skrldr of h'trtl14gl1 I'oc~trhrrlc~ry. I Norrtro c r , 7 ' I I I / A / 201-3015.

r 1 1 I I 111 SII I I~I I Sallel St\~clicn I I \ I l UNV - A'or~tloft~trrtrrr Irr / I r , t ~ d h j r ,\irrrrrccll, . I , IVHY . 66-80.

1 i 1 k 1 1 1 I I I t ' I

--a --

Schadeberg, Thilo C. and Philip Elias 1979 - A Description of the Orig Language. MRAC, Tewuren.

Schadeberg, Thilo C., and Marvin L. Bender (eds.) 1981 - Nilo-Saharan. Foris Publications, Holland

Stevenson, Roland C . 1981 - Adjectives in Nyimang with special reference to k- and r- prelxes In Schadeberg and Bender, M.L. 1981, : 151-1 65. 1991 - Relationship of Kadugli-Krongo to Nilo-Saharan Morphology and lexis. In Bender, M. L. 1991a : 34 7-369

Stewart, John M. 1967 - Tongue root position in Akan vowel harmony. Phonetics. 16: 185-204.

Street, Alan, and Frangois Gasse 1981- Recent Developments into the Quaternary Clirnafr( History of the Sahara. In Allan, J. A.. (ed) The Sahar;~ Ecological change and early economic history. Metttr\ monographs, 1, Outwell. : 8-29

Sutton, John 1974 - m e Aquatic Civilization of Middle Africa. Journal 0 1 African History, 15,4:52 7-546. 19 7 7 - The African aqualithic. Antiquity, 51:25-34.

Thelwall, Robin 1981 - Subgrouping and Reconstruction of Daju. In Schadehloly T.C. and M.L. Bender: 167-184.

Vossen, Rainer 1982 m e Eastern Nilotes: i n i i and Hi.s/ot.~~ti~ Reconstructions. Reimer, Herlitr.

Vossen. Rainer, and Mnl.intit~r IIC~*IIIIHIIW ( i r ~ rt (1~1s) IOA' 4 - N~lol~i* SIII(IIV% / < ~ + I I I I , = I 11, I 11t1

107

'Westermann, Diedrich I911 - Die Sudansprachen Hamburg. 1927 - Die Westlichen Sudansprachen und ihre Beziehungen zum

1973 - Benue-Congo comparative wordlist: Vol. 2. Ibadon: West African Linguistic Society. 1989a - Niger-Congo Overview. In Bendor-Samuel, .I (ed),

19896 - Beme-Congo Overview. In Bendor-Samuel, J. (ed.), I989 : 247-276. 1989c - Linguistic evidence for the prehistory of the Niger Delta. In The early history of the Niger Delta. E.J. Alagoa, F. N. Anozie and N. Nzewunwa (edr.), Helmut Buske Verlag,

In prep - Comparative IJq.

bunson, Kay and Kiyoshi Shimizu 1968 - Benue-Congo comparative wordlist: Vol 1. Ibaaizn: West Afrcan Linguistic Society.

1971 - The noun class system of Proto-Benue-Congo. Mouton,

NIGER-SAHARAN: DISTRIBUTION OF NILO-SNURAN AND NIGER-CONGO

- - - - - - ~alor- !---, , lndlan

0 lmo m Ocwn

L- - -L-- - I lYarr \

Acronyms and Terminology

nyms for Main Sources

C Creissels (1 98 1) Dirnrnendaal(l988) G Greenberg (1966)

M Mukarovsky (1 976-7) S Schadeberg (198 1) W Westermann (1927)

rources by these authors a re h l l y referenced

Reconstruction established from complete analysis of sound-change 'Pseudo-reconstruction' established from quick inspection of cognates

Common Bantu Guthrie, 1967-7 1

Proto-Atlantic-Congo (tentative only) Meussen, 1980

Broto-Benue-Congo de Wolf, 197 l9 Ehret, 1987 Thelwall, 198 1 Ehret, 1987

Proto-Eastem Nilotic Vossen, 1982 Williamson, in prep. Bender, 1983b Dwyer, 1989

Prolo-Mandc-Congo Derived from Westermann, 1927 Dimmendaal, 1988

Prolo-N~pcr-('o~~gc, PP~~~-N I~C~-S I I I I J I I~ I I I

- -

110 111

PNS Proto-Nilo-Saharan lometimes these authors did not uncover cognates in the families [hey PSN Proto-Southern Nilotic Rotland, 1982 search and I have tried to add these. The references after each PVC Proto-Volta-Congo PWN Proto-Western Nigritic Mukarovsky, 1976177" o-reconstruction are to guide the reader to existing references -

PWS Proto-West Sudanic Westermann. 1927 very have often replaced the citations proposed by these authors

V Vowel th more recent Or more convincing examples, Greenberg did not propose any speculative proto-forms and

1 have adopted the convention for reconstructions used in the Nigcl cd it is sometimes difficult to imagine the shape of the item he Congo volume edited by Bendor-Samuel (1 989). distinguishing th('',i Poses to link the individual forms. For ease of reference, I have established by regular sound-correspondences from those derived ''1 posed pseudo-reconstructions in the text, marked with a I#'

quick inspection of cognates. By this criterion, most mallb1 reconstructions are 'pseudo-reconstructions' (inevitably) The effecl

this is to translate the starred forms of various writers to hack ' / I to-Niger-Saharan

Other Terminology VkV 'arm, hand, shoulder' (C. :3 15, Gr. :so, G: 133)

TO shorten the rather clumsy term 'Kadugli-Krongo', Roland Stevell\(~II Saharan Zaghawa ba originally proposed 'Kado' a term for people in many languages Ills Berta Berta Ba'ba group his has been adopted by Bender and is used in this Paper ES Didinga iba

CS Bagirmi bo ko upper arm

APPENDIX 2 *PWS -buak- Bantu *PB *- bo ko

The following examples are Set out to provide \ ( ) I 1

justification for the tree proposed above They are arranged in sl'lr l rent l~ attested in Kordofanian where another root something Nilo-Saharan and Niger-Congo forms. Where a gloss has l'fi "* identified by Greenberg (1 966: 1 53), is dominant, Creissels discussed previously, I have referred to the authors in acronyms i l lr" 15) compares a set of words referring to a root #kamba which the table. The table of acronyms preceding the introduction sh()llltl @ornrnonl~ means 'shoulder', though these are probably distinct. used with the glosses.

Where possible I have cited the reconstructions of Weslcltllrl (pws) and Mukarovsky (PWN) as these direct the readel '11

tabulation of numerous individual forms. I have tried to adtl 1 1 ' from language families omitted by these authors -SO for PWN I 11

tried to find Kordofanian, Mande and Adamawa-Ubangian atlc~llllll whereas I have tried to add Kordofanian for PWS 1 tlave a l ~ ~ ~ ~ ~ ~ ' l l r

thus Westcrmannls I'WS is assig~lctl 10 I'M( ' 1'1 O I ~ M i l ~ l ( l ~ ~ (

1 1 1 l'wf'd l h ~ ~ ~ ~ p l O ~ k l l l k l ~ ~ ~ ~ I!III\ h d k 111 11' I ' A '