Introduction to Study of Language
Introduction to Study of Language LING 1
Popular in Course
Popular in Linguistics
This 26 page Class Notes was uploaded by Deborah Greenholt on Friday September 4, 2015. The Class Notes belongs to LING 1 at University of California - Los Angeles taught by Staff in Fall. Since its upload, it has received 66 views. For similar materials see /class/177859/ling-1-university-of-california-los-angeles in Linguistics at University of California - Los Angeles.
Reviews for Introduction to Study of Language
Report this Material
What is Karma?
Karma is the currency of StudySoup.
Date Created: 09/04/15
Prepublication draft THE USE AND MISUSE OF LANGUAGE IN THE STUDY OF AFRICAN HISTORY Russell G Schuh UCLA Department of Linguistics 1 Language Classi cation and History It is staggering to contemplate the massive number of historical events implied by the ve words Wolof and Zulu are related The speakers of these languages are separated by thousands of miles of rain forest and desert not to mention mountain ranges and some of the world s mightiest rivers Yet comparative linguistic evidence makes it clear that Wolof Zulu and more than 1000 other languages in Africa belong to the Niger Kordofanian language family a linguistic group which can ultimately trace its ancestry back many eons to a single language community probably somewhere in West Africa In the scheme laid out in Joseph Greenberg s Languages ofAfrican Greenberg 1966 the classification of African languages now univesally accepted among serious scholars NigerKordofanian is but one of four families into which all the languages of Africa fall Since the scienti c study of language began in the mid19th century linguists have developed effective techniques for determining whether or not particular languages show evidence of genetic relationship ie whether or not we can hypothesize that certain languages have descended from a single ancestral language even though the modern languages may differ from each other at every levelipronunciation vocabulary and grammar In a very few cases we have data from the ancestral language itself as well as its descendants the best known and best documented case being Latin and its descendants the Romance languages such as Italian Spanish and French Unfortunately we do not have such information for most of the languages in the world including virtually all those of Africa but when we nd that languages resemble each other in ways which could not be a result of pure chance we can infer that such languages must have come from a single ancestor for which we have no historical documentation but which must have existed at a speci c location and then spread through conquest cultural dominance or the simple need for more land As time went on communities speaking this original language became isolated from each other changing their manner of speaking in ways idiosyncratic to each community until the accumulated changes resulted in such differences between the diverging communities that they no longer spoke the same language As in many world areas documented history in Africa dates back a few centuries at best and in most areas there is no reliable historical linguistic documentation at all But what Africa lacks in historical documentation it makes up for in languagesilots of themiin fact more than in any other world area Ruhlen 1991 gives a total of 1474 known African languages probably a considerable underestimate of the actual total By Ruhlen s count one of the four language families of Africa NigerKordofanian alone has 1064 languages more than any other single language family in the world except Austric the family covering southeast Asia and most of the Paci c However many of the 1175 languages of this family are spoken on islands and this isolation of people from each other surely accounts in part for the great linguistic diversity The only contigous land area which rivals Africa for complexity is New Guinea with around 800 languages concentrated on a relatively small island but these languages are all members of a single family Compare this to Nigeria a single nation with speakers of around 400 languages from three of the four African language families Knowing nothing more than the genetic class cation of languages and their locations allows us to draw strong inferences about the history of their speakers Consider the case of Fula also known as Fulfulde Pulaar Toucouleur Peul Ful or Fulani depending on where it is spoken and the language of the person referring to it Speakers of this African language relationships 2 language range across West Africa from the western tip in Senegal into Chad Republic and possibly beyond Fula belongs to the West Atlantic branch of NigerKordofanian All the other West Atlantic languages are spoken in Senegal Sierra Leone and Guinea Bissau Based on genetic relationships and geographic distribution we can make only one reasonable inference the homeland of the ancestral language of the West Atlantic branch was at the far western end of West Africa and Fula has spread eastward In the case of Fula we know what the reason for the spread was The Fula are traditionally nomadic herders who must seek pasturage for their animals Beginning from their original homeland in what is now Senegal they could move only one direction in search of pasturageieast To the west was the Atlantic Ocean to the north the Sahara Desert and to the south the endemic tsetse y infestation of the forested areas In some parts of West Africa the Fula remain nomadic but in other areas they have settled Thus in northern Cameroon and contiguous parts of Nigeria Fula is the native language of many people who are settled on the land and whose lifestyles are essentially the same as those of people who speak languages indigenous to the area Nonetheless because Fula s nearest linguistic relatives are thousands of miles away we know there must be some explanation for the presence of Fula speaking communities in this area other than their having been there since time immemorial There are two related aspects of the West AtlanticFula model which have implications that are applicable where the situation is less transparent l the area of greatest diversity within a genetic linguistic group is the homeland and 2 linguistic uniformity over a large area suggests rapid linguistic spread This is evident for example in the great dialectal diversity of English in Britain the homeland of English vs the relative dialectal uniformity of English in the North America where English spread across the continent in a short period Ruhlen 1994303304 lists 46 West Atlantic languages 45 of which are spoken only at the far western tip of Africa The one West Atlantic language not restricted to this area Fula extends across a vast area yet is linguistically highly uniform across its extentithe variety of Fula spoken in Senegal is mutually intelligible with the variety spoken in Cameroon perhaps with a little practice comparable to that needed by a Chicagoan trying to communicate with a rural Scot We can apply these principles to one of the more notable features of African language distribution the Bantu area The 380 languages usually referred to as Bantu called Narrow Bantu in Ruhlen 1994312 include languages as far ung as Ewondo Cameroon Luganda Uganda Kikuyu Kenya Chaga Tanzania ChiNyanja Malawi Shona Zimbabwe Bemba Zambia Zulu South Africa and of course Swahili spoken throughout East Africa While the large number of Bantu languages would seem to represent diversity from a linguistic point of view they represent uniformity These languages do not differ linguistically from each other much more than do the Germanic languages of Western Europe Greenberg 1966 gives only a single entry Bantu to cover the entire group rather than listing the individual languages However if one moves to southwestern Cameroon and continguous areas of Nigeria one finds dozens of Bantulike languages which share some linguistic traits with Bantu but which are quite different from Bantu and from each other This is the area of diversity that we are looking for Clearly the homeland of the ancestor language to the Bantu languages was somewhere in what is now the southern NigerianCameroon border area The expansion of Bantu to cover nearly all the African continent from the Niger Delta east to the Indian Ocean and south to the Cape has taken place in very recent times in terms of language evolution There is still evidence of the linguistic situation prior to this expansion There are pockets of nonBantu languages in East Africa surrounded by Bantu particularly in Tanzania These include among others the Southern Cushitic languages Iraqw Dahalo Hadza and Sandawe The Southern Cushitic languages are related to languages such as Somali and Oromo much further north Greenberg 1966 classifies Hadza and Sandawe together with the Khoisan languages of the Kalahari desert and South Africa African language relationships 3 The Khoisan languages themselves are almost entirely surrounded by Bantu languages This language distribution implies that the Cushitic subfamily probably at one time extended along the entire eastern part of Africa north of the equator and the Khoisan languages probably occupied most of the continent south of the equator Moreover across the rainforests of equatorial Africa are the Pygmy peoples A number of Pygmy groups speak Bantu languages for example Babinga of southern Cameroon which is a cover term for several distinct Bantu languages As we will see in 2 it is dangerous to relate language and race in any direct way but assuming that the speakers of the ancestral language of the modern Bantu languages comprised a fairly homogeneous ethnic community and assuming the people of that community resembled the people in modern southwestern Cameroon it seems safe to say that the ancestors of the racially distinctive Pygmy peoples did not speak a Bantu language1 Survivals of nonBantu languages can be found in the Bantu languages themselves in the form of borrowed words from nonBantu languages In East African Bantu languages can be found words for ironworking for example which cannot be reconstructed for protoBantu see Ehret 1982 who also summarizes the reconstructed history of the Bantu expansion In summary we can make many inferences about undocumented history by piecing together a puzzle composed of a genetic classification of languages the present distribution of those languages and details from those languages such as the origins of particular vocabulary items However not all historical work using linguistic data has dispassionately approached the task letting the linguistic data lead where it might despite external factors such as race culture political importance or preconceived notions of what the history should be In the remainder of this paper I will discuss two studies where examination of the linguistic data as it has been used does not support the historical proposals The first Meinhof s notion of a Hamitic family I will discuss only brie y Meinhof s Hamitic has long been discredited as a valid linguistic group but his methodology provides a particularly clear example of the invalidity of introducing factors such as race and culture as evidence for linguistic classification Most of the paper will be devoted to the PanAfrican theory of Cheikh Anta Diop in which Diop claims to show the genetic unity of all African languages 2 Carl Meinhoi s Theory of Hamitic Racial and Linguistic Unity Carl Meinhof was perhaps the most important figure in African linguistics in the early 20th century He published voluminously and the validity of some of his work endures today in particular his work on comparative Bantu e g Meinhof 1910 However his book Die Sprachen der Hamiten Meinhof 1912 was an unfortunate publication which misdirected the course of the study of the languages involved in ways that endured for decades The term Hamitic in the title of Meinhof s book refers to the descendants of Ham Noah s youngest son who laughed at his father as he lay drunk and naked The term Hamitic is in contrast with Semitic which refers to the descendants of Shem Noah s 1The term Pygmy is in some disrepute butI use if for lack of an alternative Terms such as Baka preferred in some quarters are linguistic designations but the peoples designated by the ethnicracial term Pygmy speak a number of unrelated languages as noted below It is hard to say what kind of language the ancestors of the Pygmies did speak or even if they ever comprised a homogeneous ethnolinguistic community Pygmies today speak languages from three vastly different genetic groups Bantu languages as just noted languages of the AdamawaEastern branch of NigerCongo and languages of the Nile Saharan family One proposal is that the ancestors of the Pygmies originally spoke a language or languages of the Khoisan family Ian Maddieson in personal communication has told me that there is some DNA evidence for a genetic connection between the Pygmies and the modern speakers of Khoisan languages African language relationships 4 oldest son who along with his brother Japheth covered their father s nakeness without looking at him The Semitic languages Arabic Hebrew Aramaic Akkadian Amharic etc long the subject of linguistic scholarship form an obvious genetic unit in which the languages have much shared vocabulary and numerous grammatical similarities The languages originally referred to as Hamitic include Ancient Egyptian the Berber languages Somali and a few others They have clear affinities with the Semitic languages in both vocabulary and grammar Meinhof was not the first to use the term Hamitic to refer to a genetic linguistic group but he was the first to attempt to relate certain languages of subSaharan Africa to the languages already recognized as Hamitic Meinhof 1912 looks at seven languages in detail Ful Fula Hausa Schilh Shilha or Tashilhet a Berber language of Morocco Bedauye Beja a Cushitic language of northeastern Sudan Somali the national language of Somalia also Cushitic M aasaz39 a member of the Eastern Sudanic branch of the NiloSaharan family in Greenberg s classification and Nama a member of the Khoisan family in South Africa We now know that some of these languages are relatediHausa Schilh Bedauye and Somali are all members of the Afroasiatic family Fula Masai and Nama however each belong to one of the three other African language families Why then did Meinhof group them all as Hamitic The answer is that he used a mixture of racial and linguistic criteria most of which are spurious The introduction to Meinhof 1912 is a confused discussion which wanders between linguistic and racial issues The basic premise concerning race emerges in the following statement p 2 Es is ja bei einem Blick auf die Sprachenkarte Afrikas evident daB die hamitschen Sprachen als Sprachen von Leuten kaukasischer Rasse zusammenngetroffen sind mit des Sprachen der Nigritier Wie es scheint hat sich der Vorgang im Lauf der Geschichte immer wiederholt daB ham itische Stamme als Herrenvolk unter dunkelfarbigen anderssprachigen Volker39n auftraten sie unterwarfen and beher rschten Dabei fand selbstverstandlich ein sprachlicher Austausch zwischen der herrschenden Minoritat und der beher rschten Maj oritat statt It is quite evident from a glance at a language map of Africa that the Hamitic languages as languages of people of Caucasian race have come together with the languages of the Negroes As it seems the course of history has ever repeated itself in that Hamitic tribes have shown up as a dominant people among the dark colored peoples speaking other languages subjugated them and controlled them Therea er obviously a language exchange has taken place between the dominating minority and the dominated majority One learns about the prototype of these dominant Hamites in an appendix to Meinhof1912241256 Hamitische Typen by one one Felix v Luschan Herr von Luschan says p 241 242 Jedwede Betrachtung hamitischer Typen muB irhen Ausgang von des alten Agypter n nehmen Man braucht heute nur eine einzige der vielen altagyptischen Darstellungen zu betrachten auf denen wirkliche dunkle Afrikaner neben Agypter jene Unterschiede einschatzten als der modern Anatom das getan hat Any examination of the Hamitic type must take its starting point from Ancient Egypt Today one need only consider a single one of the many ancient Egyptian pictures in which true dark Africans appear next to Egyptians in order to see immediately how much more properly the ancient Egyptians assess that difference than modern anatomy has done This appendix goes on at length discussing hair type nose shape and other physical characteristics of Egyptian mummies comparing these physical parameters to measurements taken from various people around Africa and to pictures of mummies and Africans displayed in plates following the index Not surprisingly the speakers of 2The traditional name for Greenberg s Afroasiatic family is HamitoSemitic or SemitoHamitic a term still used by many Europeans The implication of this name is that there are two coordinate subfamilies Hamitic and Semitic While Semitic does constitute a genetic unit Hamitic does not ie Egyptian Berber etc do not together form a subfamily coordinate with Semitic African language relationships 5 Hamitic languages who lack the prototypical Egyptian features have undergone race m1x1ng Freilich sind die Ful und die Hausa alle mehr Oder weniger angenegert und haben teilweise durch direkte Aufnahme von uberwiegenden Mengen Negerblut auch wirkliche Negereigenshcaften angenommen p 255 Admittedly the Fula and Hausa are all more or less Negri ed and have in part taken on true Negroid attributes as a result of the predominant quantity of Negro blood All this of course is bad physical anthropology and bad history which would be dismissed today as having validity of any kind But even if the racial and historical criteria for grouping the Hamites are bogus we must still ask whether Meinhof provides valid linguistic evidence for the grouping he proposes The answer is no and the basic reason emerges in the first sentence of his introduction Unter des Sprachen Afrikas finden sich eine groBe Anzahl von Idiomen die durch das grammatische Geschlecht und des Ablaut an die semitischen und indogermanischen Sprachen Among the languages of Africa are found a large number of tongues which because of grammatical gender and ablaut are reminiscent of the Semitic and Germanic languages Meinhof in this sentence mentions two linguistic traits grammatical gender and ablaut changes in vowels to signal grammatical differences as in English sing sang sung Modern linguistics calls traits such as these typological features ie general features of a grammatical system To say that a language has grammatical gender is to say that it is a certain type of language but this observation alone says nothing about genetic relationship Thus both Hebrew and French categorize all nouns as masculine or feminine but these languages are entirely different from each other with respect to the markings they use to show gender distinctions Consequently no one has proposed that Hebrew and French are genetically related to each other at least on the basis of the criterion that they both have grammatical gender Meinhof s 1912 method of comparing the seven languages is to list a variety of typological features then to show how each language exhibits those features Some of the features he uses are the following consonantal changes ablaut tone reduplication gender number case verb stem shape He shows how each language exhibits those features but in very few cases does he explicity show that the manifestations of those features resemble each other across languages Without such specific resemblances the presence of such typological traits is meaningless for genetic classification For example almost all African languages Hamitic and otherwise are tone languages as are most languages of east Asia and Mexico all African languages use reduplication to show iteration frequency etc but this is probably a nearly universal trait of human languages cf splish splash rinlgz dingz etc in English Consider the case of grammatical gender cited as a feature relating Hamitic to Semitic For Hausa Meinhof 191272 notes that feminine nouns usually end in a eg abo39kz39 friend m abzikia friend f and he mentions a genitive n for masculine r tl for feminine with no example phrases For Masai p 195 he mentions a masculine definite article 01 feminine 311 For Nama p 219 he cites forms like khoe b the man khoes the woman The only possible reaction to this data is So what It would have been equally meaningful to include French grand tall m grande tall f On the other hand he does not call attention to comparisons which could support a hypothesis of genetic relationship Thus in Schilh he gives a number of examples where feminine is marked by t eg p 98 asgggn black m tasgggnt black f but nowhere in the book does he point out the fact that the t feminine of Schilh might resemble the t feminine of Hausa seen for example in ordinals eg p 73 nafari first m tafari first f African language relationships 6 There is no point in pursing further examples The outline in the preceding paragraph typi es the methodology of the entire book and Greenberg 1966 has thoroughly refuted Meinhof s treatment of Fula pp 2427 Masai pp 9095 and Nama pp 6772 showing that these languages are not Hamitic but rather are obvious members of the West Atlantic NiloSaharan and Khoisan families respectively One can only conclude that Meinhof approached the Hamitic enterprise with a preconception about what kind of people would speak Hamitic languagesiin particular Caucasian looking warlike cattle herders Some languages spoken by people lacking these racial and cultural traits have features so similar to those in recognized Hamitic languages that Meinhof could not ignore them This includes Hausa whose speakers despite being black agriculturalists managed a place in the Hamitic ranks A likely aid to Hausa s achieving Hamitic status was the fact that most Hausa speakers live in large wellorganized states whose rulers are of Fula descent On the other hand the Fula the Masai and the Nama were obvious candidates for Hamititude because of their pastoral cultures and indeed Meinhof managed to discover Hamitic traits in their languages never noticed before or since 3 The Pan African Theory of Cheikh Anta Diop The Senegalese scholar Cheikh Anta Diop has written a number of articles and books most notably Parent g n tique de l gyptien pharaom39que et des langues n gro afrz39caines Diop 1977 in which he claims to demonstrate a relationship between the language of Egypt in the Pharaonic period and the modern langues n gro africaines languages of black Africa3 The only language of the latter group from which he supplies significant amounts of data is his native language Wolof Were he able to show unequivocal resemblances between Egyptian and Wolof this would of course show that Egyptian must also be related to linguistic relatives of Wolof However I will show that Diop does not provide any convincing evidence for an EgyptianWolof connection Moreover he fails to mention several groups of langues n gro a icames which share no apparent resemblances to Wolof but which do share features with Egyptian namely the Chadic languages of west and central Africa the Cushitic languages of northeast Africa and the Semitic languages of Ethiopia and Eritrea Diop l977xxiiixxv makes lavish claims regarding both the innovative nature of his work and the certainty of his hypotheses eg Pour la premiere fois dans l histoii39e de la linguistique africaine il a t possible de rendre compte scientifiquement de l tat actuel d une langue morphologie syntaxe lexique walaf a partir de l gyptien ancien C est le caractere syst matique de cette explication quasi totale qui n a pargn presque aucun aspect de la langue expliqu e qui est vraim ent nouveau pp xxiiixxiv For the rst time in the history of African linguistics it has been possible to scienti cally account for the present state of a language morphology syntax and lexicon of Wolof beginning from Ancient Egyptian It is the systematic nature of this near total explanation which has left almost no aspect of the language under discussion untouched which is truly new Peut tre pour toutes ces raisons cet ouvrage fondetiil r ellement la linguistique historique africaine en tout cas il confere a la linguistique africaine la dimension historique qui lui faisait d faut p xxiv It may be that for all these reasons this work actually establishes African historical linguistics in any case it confers upon African linguisitcs the historical dimension which it lacked We will examine the extent to which Diop really demonstrates the connection of Wolof to Egyptian below But what about the dimension historique qui faisait defaut a la linguistique africaine To take Diop s statements at face value one would think no one had ever done any historical work on African languages before 1977 Looking at his 3Diop is not explicit about what he means by langues n graaf caines Taken in its most literal sense it would mean all the languages in Africa spoken by black people African language relationships 7 bibliography one nds no mention of any work in African historical linguistics For xample there is no mention of Carl Meinhof who as noted in the previous section did fundamental work in Bantu historical linguistics as well as his less creditable work on Hamitic languages no mention of Diedrich Westermann who wrote voluminously on both synchronic and diachronic aspects of African languages no mention of Pierre Alexandre whose popular book Langue et language en Afrique noire Alexandre 1967 summarized the generally accepted relations between African languages and most particularly no mention of Joseph Greenberg whose work in African language classification dates from the 1940 s and is now the universally accepted classificatory scheme It may be that Diop did not know of these works or on the other hand that he chose to ignore them Either way the scholarship underlying his work is suspect None of these recognized scholars and a number of others who have dealt with African languages on a continentwide basis has ever even hinted that there might be some demonstrable connection between Wolof and Egyptian4 To take Diop s theories seriously we would like at the very least some guidance to understand why his predecessors missed something which he was able to discover Diop does not provide this guidance but we can still examine his work and compare it to that of others to see whether it really does set African historical linguistics on a new and more profitable course In order to put some limit on the discussion I will concentrate on just two aspects of Diop s comparisons 1 his claim that the noun class systems of the langues n gro africaines derive from elements found in the Egyptian language and 2 his claim to have identified a large number of lexical items common to those languages and Egyptian Diop himself considers the first aspect to be particularly significant Pour tendre cette tentative de syst matisation de l explication de l e39gyptien aux autres langues africaines il nous a paru int ressant de partir d un trait dominant de la morphologie de cellesci classes nominales qui commande meme la syntaxe In order to extend this attempt at a systematization of the explication of Egyptian to other African languages it has seemed interesting to us to begin with a dominant trait in the morphology of the latter nominal classes which controls even the syntax A convincing scenario linking features of Egyptian word structure to the noun classes of the langues n gro a icaines would indeed constitute powerful evidence for the claimed genetic relationship Noun classes are deeply embedded in many aspects of grammar in large numbers of African languages making it unimaginable that clear resemblances between Egyptian word structure and noun classes could have developed independently or been borrowed As for lexical comparison the fact that Diop devotes 223 pages of his 400 page book to lexical resemblances between Egyptian and Wolof demonstrates the importance he gives to this evidence From a linguistic point of view lexical resemblances are perhaps the most convincing type of evidence for genetic relationship because of the arbitrary link between form and meaningisince there is no relationship between what a word sounds like and what it means cf the large number of homonymous word pairs in English the only good explanation for recurrent soundmeaning resemblances between words of two languages is that they descend from a single ancestral language where those words were pronounced in a way similar to their pronunciation in the descendant languages 31 Lexical evidence for an Egyptian Wolof connection Let us consider the lexical evidence first It is beyond the scope of this brief study to look at every item in Diop 1977 162384 I will therefore look at this evidence from several perspectives citing a few salient examples In every case the number of examples could multiplied manyfold 4Ironically Meinhof 1912 might be interpreted as having indirectly proposed such a connection with his classification of Fula which is closely related to Wolof as preHamitic and hence related in some vague sense to the Hamitic language Egyptian African language xehnanshps 2 The canchlsmn wdlbe um Du presentsvmna yno 12x1 xesemhlmcesbelween Egypm q mg Walafafthzniypr st can be fanndbemen languges mm m gemn y ssss s s u snsnss x2 2 srsn snif ng an gill sr resemblances n s wank mung 1mm quanmy sr resemblances um naps 2515 ns m atzdslgm cm ymm Ways FusL unsn sn cases MW a mug alaf em 1s pnmd wnh smm E ypnsn mm In ssnns sss snsn s an xsl sns cmdbelaw an ngqsmanvmxds sm mselves dznvedfmm a slug mm andshmlld smcd s akmg he s nm we unns enmes m an 115 r emncampnnnve um Engnsn 4m cmldvm ch dhood wnh an 5mg German ward and In sunsn sssss E s E E s s s E n nnn alnfvm snsn n asnssnns lzgmmmly c nsnmudueeenmesm hzcam vevm 15 Ifaneiymnlagwal nlaonnslu dues exstncm n1 bebet snuns s fvmxdandomafthzli ypnsn xds snnsy a a m chambths 1 g s up yt lust meiymalngy pxhapsmgethzxwmul Egyp anvmxd in taunt m um39is Wuhfwnld hawhaw uidgm afchddrench dr means snma39 kn handgnu calf mhhmhhmht nadzdhndsmm nadm exs39 biq aah nadzdlmd39 nnnhw sfmngexcellznce39 man ss 39 w faamxeed39 manqmanh svcknp39 sllrp39 mrmrmxw Te jardzsen39 mm mm nmj lnmmnm bedlbedmaldmxt5sleep39 mmm wss 1m dawn39 pmyp w pnnm mmmx39 mama cmmmdlvdz39 xx nth 39 mos msm z39 snmsnmwsnm cansnmzpmvlsmnsglnmny nsnn faad39 The secandwa ans number srnsnns m ms 395 lust 1s ex ed wnhnm pmwdm masnss rsn addxugml nynsxsyss nns s Evyquot11m Egyptmnwnnng sysnnn 1 my W s nannsn mnn nnnnmnssnns on n unuysnsx fhsnsmnngymunspmsssssnnnsnsnmsfsg g P gs mm PM Wm W m ms mud chzsl39 anmzen mmp bnasl39 w5 s WU nuns Emmmmna yhkzmmmwmbamadmamhs mimemsmzmagedw mm mam mmxmmnmsssmsmmwnnunsmw unssz nzxmh mm s unswmwnnmm nun Mm agMFIud lw nmmmlnsastgmhm m 51 ml African mung xelananslups 9 cm Km wc mm i W V chm 5cm mm ch 5W cw ccycccmcc ccmcc an MW Hcammwm mm mm mm W c m cmch cm Egypnanand Wm qualityan enmzs 5 gaad Buns m acmch few iyplcalcampunsans Emu wm hxly weals humblz39 x1 OnX anand39 m spmlse39 eet an mwa pmlse Mm spmlse39 inbnw mum mum d39nnz mmmm mammm femmz39 gt mam h Ndz39 O cam tzmemm W W P cw mmm mfaxway bewet39 13 Tank y MW 13 bmad wAdz39 mam ch 11m cnp39 m km my m mm llmw hne39 km my n mm ym m cm dlscxplmz39 m nymy m M chck mm Vulture39 w myquot manger Aves mm manly mmquot mnmmc m um mm get sexual pleasm39 doc agxeeahle39 mmcmmma mummy mm dwm mom cm affspnng39 mywl Ja 1139 mm 1c 239 m anch dzs Ammanx s eq Sikh manz axmay u M d amidhmquot r mm m a hkzmss39 m famdymmz39 w car So su am my cmmbssm Indus andm yank mm an eiymalngus Dm cms umc 15 m magi culmnl ax xcal Wm canmcuan human 2 E m Waaf has c M wcc frccx pumblzquotquot hz meaning 1an 5mm causes mm m muquot 5 WM m back39 15 m quotbmadzslquot pun am bady cupc39 and mnmrs are hallawed mu 1x 5 hale a hue can be used a a a hale etc Nam um cnpmnmrhmhnlz39 present a funth dluskanan afmulnplymg m number afenmzs African language relationships 10 by listing the same word under more than one etymology The Egyptian words ikn cup and iknw hoe are surely unrelated except that they happen to share some sounds as do English cup and cap for example The Wolof word lam hoe therefore cannot be paired with both Egyptian ikn and iknw as separate etymologies On the other hand the unrelated Wolof words lam hole and g nn mortar cannot both be paired with the same Egyptian word In some cases knowledge of word structure makes even the phonetic resemblances less compelling Though neither hbsjt nor hbsw both defined epouse female spouse by Diop are in Faulkner 1976 Faulkner does give a word hbs clothe cover HZst and hbsw are probably compounds based on this wordifew if any underived Egyptian words have more than three base consonants A link between Egyptian iPt back and Wolof yaatu be broad questionable semantics aside is rendered even less likely by the fact that the final t in the Egyptian word is a feminine suffix not part of the root and tu in Wolof is a verb derivational suffix also not part of the root Diallo 198343 I should stress that I did not select especially questionable looking comparisons for the list of examples aboveisimilar items can be found on every page of Diop s comparative word list To be sure there are some items which look relatively convincing eg Egyptian hrwy testicles Wolof xuur hwr testicle possibly rd child Wolof xale child and a few others but this is no surprise One can find a few chance resemblances in words between any two randomly chosen languages With selected words and a fertile imagination one can furnish apparent proof that English and Wolof are related Wolof English bey goat buck dee to die die fan when when gemmi mouth gums goor man guard6 lekk to eat lick man 1 me me nag cow nag in the sense of decrepit horse am food yumyum ndox water oc n1t person nat1ve nopp ear lobe safara fire fire 361 a prefix in Wolof xale child child ch lt kicf German Kind yow you sing you When comparing languages which are genetically related the main types of words where one expects to find a fair number of clear soundmeaning pairings are items of basic vocabulary ie words found in every language which remain relatively stable in meaning and which are resistant to replacement whether by borrowing from other languages or from language internal changes These include small numbers terms for body parts universal environmental elements sun moon water f1re and verbs 6Cf personne courageuse given as one of the meanings for go o r in Fal et al 1990 Ian Maddieson has suggested that an even better WolofEnglish cognate would be Wolof go o r English mwalf which originally meant m wolf 7Pairing yumyum with Wolof am is not as farcical as it may appear It is not unlike Diop s frequent inclusion of Wolof ideophones such as Xepp completement mouill in his comparative sets African language relationships 11 referring to basic life functions die eat drink Depending on the geographicalcultural area certain animals plants occupations etc might also be considered basic eg goat dog elephant guinea fowl okra and farming could be viewed as basic to the leXicons of African languages But working through Diop s list one finds few onetoone soundmeaning pairings in basic vocabulary between Egyptian and Wolof Below is an exhaustive list of the words I was able to find in Diop s EgyptianWolof comparisons where items which could be considered to be from basic vocabulary had the same or uncontroversially related meanings between Egyptian and Wolof 4 a ht eye of god get eye probably fem form of This is the plural of bar Sh spirit t is a eye hence not a good feminine suffix not part gL1 matCh of the root b nt neCk baat bat neck I is a fem suffix dW give jox dioh doh give th Wing dunq unh feather dns heavy diis dis heavy h h hrd nhnw Child xale child if Steal sacc sata steal itn sun jant sun nhdt tooth be be tooth er Pulaar HWY Come ew iew come but base meaning is retuIn rkh re light lakk burn roast con agration t hOt tang tang hOt spit t fli tef spit not in Faulkner but cited in Greenberg 196662 tp head bopp bop head Sound resemblances between Egyptian and Wolof in most of these items are dubious at best A couple look reasonable eg nwy ew come and inkdunq wingfeather but as already noted it would be surprising if there were not at least some similar words in large comparative lists of any two randomly chosen languages In most cases however there are sounds present in Egyptian which are uneXplicably absent in Wolof dnsdiis heavy and viceversa ZIPSlice steal there are different sound matches in different words Egyptian g 7 in wing and d in heavy both matched with d in Wolof and there are cases with no obvious similarity at all e g Egyptian tp head with initial t and final p has as much in common with French t te as it does with Wolof bopp The Wolof word xale child shows up literally dozens of times in Diop s wordlist compared with African language relationships 12 various Egyptian words Most of the proposed Egyptian cognates with xale are problematic for the same reasons given for other words In short it is not an exaggeration to say that the fictional cognates in the WolofEnglish list above are at least s 39 39 for J quot 39 quot as the items from Diop s EgyptianWolof a 1 list In contrast if we compare Wolof and a language to which it really is genetically related we do nd convincing cognate pairs in basic vocabulary Compare the Egyptian Wolof list above with the following items from Wolof and Fula Fula data from Labouret 1955 The Fula nouns and verbs have suffixes set off by hyphens These are noun class markers about which more below The two Wolof numbers at the end of the list have apparent prefixes seen also in aar two Wolof Fula nOPP ear nof ru ear b t 5g g t p1 eye yite re sg git e p1 eye lammi Tongue dem ngal tongue naaC SUH naa nge sun weer moon lew1u moon ndOX Water ndiy am sg di e pl water g r man gor 0 man nit Person ned do person 113 COW nag ge cow b y 03V mbee wa sg be i goat ey elepham nyii wa elephant am fOOd am de to eat e tt three tati three e nt four nayi four 32 Noun classes as evidence for an Egyptian Wolof connection I now turn to Diop s claim of having located the source of the noun classes of the langues n gro a icames in Egyptian Before examining his scenario for this development it is necessary to give a broad picture of how noun classes work in the languages which have them Languages with noun class systems in the sense to be described below are found in all branches of Greenberg s NigerKordofanian family with the exception of Mande the branch comprising languages such a Bambara Soninke Mende and others in the West African savannah region It is important to reemphasize a point made above about noun class systems These systems involve an elaborate subcategorization of all the nouns of a language into many distinct classes This classification system is integrated into grammar at all levels such as marking on the nouns themselves nominal derivation agreement patterns between nouns and their modifiers systems of pronominal forms etc It is inconceivable that such systems would have been created independently from language to language or that a language without such a system would have borrowed it in toto from another language The implication therefore is that African languages which have such noun class systems must have inherited them from an ancestral language which already had them Those NigerKordofanian languages which no longer have noun classes such as the Mande languages or many languages of the Kwa group along the West African coast must have lost the noun class systems that their ancestral languages possessed This hypothesis is supported by the fact that we can find many languages today where the noun class systems are in various stages of being lost as an active part of the grammar but essentially no languages which are creating noun class systems Noun classes are somewhat like grammatical gender familiar from IndoEuropean languages ie in the same way that languages like Latin or German categorize all nouns African language relationships 13 as masculine feminine or neuter languages with noun classes categorize all their nouns as belonging to abstract classes which determine what adjectival agreements to use etc There are differences between European gender systems and NigerKordofanian noun class systems however First class languages have many more than the two or three genders of European languagesidepending on the language the number of noun classes may range from 8 or 9 to over 20 Second sex does not play a role in determining noun class though in many class languages there are rough correlations between a noun s meaning and its class e g nouns referring to humans regardless of sex typically belong to a particular class animals particularly domestic animals regardless of sex will typically belong to a particular class things that occur as non countable masses water milk grain will belong to particular class etc Third and most important number singular or plural is not a distinct grammatical category from class That is there are classes that refer to individuals the singular classes and classes that refer to more than one the plural classes but there is no marking for plurality that is grammatically distinct from class marking For example in Wolof nag wi the cow is shown to be singular because it is modified by the definite article wi ofthe w class a singular class typically used with animals Nag yi the cows on the other hand is shown to be plural because it is modified by the definite article yi of the y plural class There is no extra marking that can be called a specific mark of plural like the s of French or Spanish which may be added to either masculine or feminine nouns Class languages spoken today mark nouns for class in a remarkable variety of ways Some examples are seen in the table below Wolof shows no class marking on the noun at alliclass of a noun shows up only on noun modifiers pronouns etc Fula marks nouns by suffixes and by a set of alternations in the initial consonants of nouns which depend on which class a noun is in Avatime a language of Ghana marks class by prefixes on nouns8 The table gives examples of nouns in each class of the respective languages modified by a definite marker roughly equivalent to a definite article In Wolof and Fula the definite markers are separate particles in Avatime they are suffixes on the noun Generally in class languages there is a singularplural class pair ie if a noun in the singular belongs to class x in the plural it will belong to class y In Wolof and Fula there are fewer plural classes than there are singular classes such that some of the plural classes pair with more than one singular class In Avatime on the other hand each singular class pairs with a distinct plural class 9 The table is laid out with each noun shown in a singularplural pairing This table does not claim to be aligning corresponding classes across languages a task beyond the scope of this paper and my knowledge of comparative NigerKordofanian man g man woman the men the men the women orange the women the heads the the the bulls the cows the the the thorns 8For the varying positions of noun class affixes in NigerCongo languages see Greenberg 1978 9Avatime is a tone language Some classes are distiguished from each other only by tone eg the kri singular class exemplified by kzZtse39 death where the noun prefix bears mid or high tone vs the la plural class exemplified by k zi bowls where the noun prefix bears low tone Avatime data are from Schuh 1995 Fula data here and elsewhere are from Sylla 1982 and Labouret 1955 African language relationships 14 nag nag ge nge the cows the cows the deaths person ngo the sands the birds the doors the roads the canoe the canoes the reeds the wind Diop 1977XxviXxviii proposes the following scenario to explain the development of class marking in the langues n gro africames 1 Egyptian had five distinct consonantal morphemes to question who what viz m p s L and t Diop 19774510 plus two further consonants k and w seen in the constructions ky the one the other we we the one the other It is this set of seven consonants which correspond to the consonants characteristic of the class markers in the langues n gro africaines Note that Diop is not claiming that the class languages got their classmarking consonants from Egyptian only that Egyptian gives us a picture of what the nonwritten African languages may have looked like at the time when Egyptian was spokenicf point 2 just below 2 The langues n gro a icaines extended the set of consonants seen on interrogative words by a principle of euphony with the initial consonants of the nouns to which they refer In Diop s words p xxvi Dans les autres langues de la famille africaine qui a l poque des pyramides se trouvaient dans le m me tat d e39volution que l gyptien ancien et qui n etaient pas crites la conjugaison euphonique fut le proc d le plus conomique pour la specialisation fonctionnelle de ces morphemes pour des langues de ce type pr sentant cette structure particuliere l oralit a d jouer un role acc l rateur In the other languages of the African family which in the epoch of the pyramids were in the same state of evolution as Ancient Egyptian but which were not written euphonic conjugation was the most economical process for the functional specialization of these morphemes for the languages of this type manifesting this particular structure orality ie being spoken but not writteniRGS must have played a role in accelerating the trend toward euphonic agreement 10As far as I can tell only the first four are really parts of question words The last I is part of a particle ter or 131 which accompanies the question words p xxvii This may not be the fifth consonantal morpheme that Diop had in mind butI cannot see any other candidates in the list he gives African language relationships 15 3 With these variant forms of the interrogative words as a basis demonstratives and relative pronouns developed with the same set of euphonic consonants e g from an interrogative like Wolof ban which one b class came demonstratives like bii this one b class and relative pronouns as in xale M me39 gis child whom I see b class Let us examine this scenario With respect to the Egyptian interrogatives referred to in l Diop says L Egyptien avait cinq manieres diff rentes cinq morphemes consonantiques diff rents d interroger de dire ltltquoigtgt ltltquigtgt c estadire apparemment la m me chose ou bien il existait d ja une conjugaison euphonique embryonnaire masqu aujourd hui par le systeme d ecriture hi roglyphique ou bien il s agit d une profusion de pl onasmes ce qui est improbable Egyptian had ve different ways ve different consonantal morphemes for asking questions for saying What Who that is to say for expressing apparently the same thing either an embryonic conjugation based on euphony already existed masked today by the hieroglyphic writing system or it had to do with a profusion of pleonasms which seems improbable But the suggestionclaim that these Egyptian morphemes meant the same thing is entirely gratuitous Callender 19759698 gives different meanings and different syntactic functions for all the Egyptian words in question He defines m as who and calls it an AGENTIVE he defines p as what specific entity what who and calls it an ADJECTIVAL he defines s as which one and calls it a NONSPECIFIC and he defines L as fromatin what place where and calls it an ADJECTIVAL or ADVERBIAL depending on syntactic environment To say that these Egyptian question words meant the same thing would be like saying that English has a set of morphemes om at ich ere that mean the same thing when attached to the question morpheme wh Perhaps more important for the discussion at hand is that the variant forms of the question words in Egyptian share no functional similarity to the noun class markers It is agreement with the class of a noun which accounts for the differences between noun class markers regardless of function interrogatives demonstratives etc In the quotation above even Diop only speculatively hints at anything like this to account for the variants of Egyptian question words when he speaks of une conjugaison euphonique embryonnaire but he presents no evidence for this nor does he pursue it anywhere in his book In short the claim in 1 would require us to believe that a set of functionally and phonologically distinct question words would first have to lose their functional distinctions then these apparently meaningless variants would have to be reinterpreted as having an entirely different function viz agreement with nouns starting with different consonants This brings us to 2 the notion that the languages qui n etaient pas ecrites which were not written multiplied the number of consonants in interrogatives beyond the seven seen in Egyptian by a principle of euphony with the initial consonants of nouns to which they referred Here Diop must have in mind the fact that in Wolof the consonant of class markers for a particular noun will often be the same as the initial consonant of the noun eg weft wi the y garab gi the tree fanciin the mouse meew mi the milk suufsi the sand etc Examples such as these aside it is easy to show that consonantal euphony plays no role in the structure of NigerKordofanian noun class systems Even in Wolof the number of nouns with euphonic class markers is very small when looking at the full nominal lexicon Diallo 19834850 lists this as only one out of a large number of criteria for determining a noun s class marking Some of the other criteria he mentions are the following i un genre humain is marked by singular k plural Fi nit ki the person nit iii the people ii nouns derived from verbs by initial consonant alternation belong to the 1 class caafli the roasted peanuts lt saaf to roast iii manner nouns derived with the suffix in belong to the w class doxin wi the conduct lt dox to walk iv liquid and mass nouns belong to the m class soow mi the cultured milk v fruits belong to the b class mango bi the mango etc There are even African language relationships 16 homophonous nouns which belong to different classes depending on meaning eg weft W the y but weft gz39 the iron These examples and thousands of others show that consonantal euphony as a determinant for noun class far from being the source of multiple noun class markers is obviously a rather recent development in Wolof In the first place as far as I know no other class languages show any evidence whatsoever of consonantal euphony as a principle for determining noun classia glance through the table above reveals no correlation between class marking consonants and initial consonants of noun stems in Fula or Avatime nor would similar lists from hundreds of other languages show any such correlation Rather the typical correlations with noun class are like those illustrated for Wolof from Diallo 1983 ie meaning ofthe noun as in i iv v or derived features of the noun as in ii iii The rough correlation in Wolof of noun initial consonant with class marking consonant is straightforwardly accounted for by the historical linguistic principle of analogy whereby apparently nonfunctional variation levels out for example as in English where nearly all nouns are now pluralized in s as opposed to the more complex system of Old English still re ected in modern German Indeed in Wolof the analogical principle continues to level the noun class system today In Dakar many nouns are being shifted to the default b class even though they belong to other classes in more conservative varieties of the language eg loos W the neck vs Dakar loos bi Diop l977xxvii fn 4 ff links the specific class marking consonants of Wolof to the Egyptian consonants he singles out under 1 above viz m p which he links to the Wolof default class b s L seen as c in Wolof locatives such as 0139 in here ca in there k w and a couple of other forms which need not concern us here Here again the specific comparison between Wolof and Egyptian is misleading In the first place Wolof has a smaller number of noun classes than most class languagesithe average is probably something between the 10 of Wolof and the 21 of Fula This makes it easier to find putative relations with Egyptian than would be the case say with Fula More important are the specifics of the noun classes themselves Perhaps some of the consonants of the Wolof classes can be matched with the consonants of Egyptian interrogatives but how characteristic are the Wolof classes of class languages in general A complete answer to this question would require an indepth comparative study of NigerKordofanian but one can note a few of points of interest One is the m class where Diallo 198349 notes an association with liquids This association runs throughout the NigerKordofanian family being seen in the m of Fula dam see the word for water in the table above and occurring universally in the Bantu languages eg Swahili m ji water One wonders what possible connection this m marking liquid or mass in NigerKordofanian could have to the Egyptian m interrogative who what Another comparative point is the Wolof 1 class Diop l9773 relates Wolof l to Egyptian 71 though he does not relate the 1 class marker to any specific Egyptian morpheme as far as I can tell This 1 class is found throughout NigerKordofanian class systems e g as illustrated by the Avatime word for cow above Swahili where it is the concord marker for the 11 classjz39 no pz39 which tooth and possibly the Fula nalz39 class cf the Fula word bull in the table above Notably absent in Wolof is a b class referring to plural humans seen in Fula 5e Avatime ba and in all Bantu languages the name Bantu itself is a generalized Bantu word meaning people Presumably if this b were related to Egyptian it would be to Egyptian p which is specifically a masculine singular form It is hard to imagine how this could have become related to a general human plural Consider now the Wolof default b which is a singular class Diop l9774 explains the predominance of this class marker in Wolof as follows en effet nous venons de voiI que l gyptien ne poss dait que le d monstratifpw devenu bw en walaf celuici qui est a l origine le morpheme de classe le plus sp cialis et le plus courant r git naturellem ent le plus grand nombre de term es dans la langue African language relationships 17 in fact we have just seen that Egyptian had only the demonstrative pw which has become bw in Wolof the latter which is the origin of the morpheme marking the most specialized11 and most common class naturally controls the greatest number of terms in the language Here again using Wolof as the point of comparison presents a skewed picture In fact a b singular class much less a b default class seems to be nonexistent elsewhere in NigerKordofanianiin fact it may well be restricted to Wolof The origin of the b class in Wolof is at the moment unclear12 but it is inconceivable that Wolof alone would retain it as an archaic feature whereas huge numbers of NigerKordofanian languages independently would have developed a plural human b class and lost a default singular b class 0 summarize the thrust of this discussion there is absolutely no evidence that the relatively large number of noun class distinctions in langues negro a icaines could have come from consonantal euphony with the initial consonants of nouns The following statement by Diop pp 34 shows not only a fundamental misunderstanding of the workings of noun class systems in general but also of Wolof in particular The emphasis ofa key phrases by small caps is mine see below Les langues africaines qui possedent un nombre variable de consonnes pouvant se substituer chacune au p du demonstratif gyptien pw SANS MODIFICATION DE SENS sont appel s des langues a classes Ces consonnes ou semiconsonnes sont pour le walaf au nombre de huit b m s w k g d l cela veut dire que l on peut remplacer par un permutation circulaires le p de l e39gyptien par chacune de ces huit consonnes et le sens grammatical reste le meme SEUL LA VALEUR EUPHONIQUE CHANGE African languages which have a variable number of consonants which can each be substituted for the p of the Egyptian demonstrative pw WITHOUT MODIFICATION IN MEANING are called class languages These consonants or semiconsonants y or wiRGS are for Wolof eight in number b m s w k g i I this means that one can replace by circular permutation the p of Egyptian with each of these eight consonants and the grammatical meaning stays the same ONLY THE EUPHONIC VALUE CHANGES One gets the picture of the preliterate ancestors of the speakers of these langues 61 classes euphonically alliterating modifiers with the nouns they modify because this was easier than keeping track of abstract semantic and grammatical classes of nouns This picture cannot possibly have been the case The discussion above has shown that euphony is not now and has never been an important force in noun classification Contrary to Diop s claim that class markers existed sans modi cation de sens classes surely did arise in semantic andor grammatical categories a number of which can still be clearly identified across the NigerKordofanian languages including Wolof even after millenia of class realignments reductions augmentations etc in individual languages In the light of these facts the credibility of a hypothesis whereby class markers arose from something like the interrogatives of Egyptian evaporates1 11My translating ability from French to English may be at fault here but it seems to me that there is a contradiction in calling Wolof 12 both le plus specialise and le plus caurant the most specialized and the most common Normally specialization should refer to use in a restricted number of contexts 12Diallo s 198349 association of bi to words for fruits may give a hint as to the origin of this class marker Throughout Africa the generic term for fruit is child of tree In NigerCongo the nearly universal word for child or beget is bi cf Fula ngel Avatime 6 e A bi root in this meaning seems to be absent in Wolof It may therefore be that the 12 class in Wolof originated in the expression child of Another possibility is that the NigerCongo b human plural class was generalized In a number of languages notably some of the Grassfields Bantu languages of Cameroon this class is being generalized as the plural for all nouns regardless of singular class Such a generalization could have taken place in Wolof then extended to all nouns regardless of number Whatever the case Wolof is unusual within NigerCongo in l lacking a 12 human plural class and 2 having a 12 default singular class 13The origin of lexical noun classes in semantic distinctions has a parallel in most East Asian languages where nouns are accompanied by classifiers in many contexts Thus in a language like Vietnamese one cannot say two children two pencils etc Rather one must say hai du d nho two beings children hai m vi t chi two sticks pencil etc Interestingly Diop does not mention a better Egyptian parallel to African language relationships 18 I have little to say about part 3 of Diop s scenario for the development of noun classes ie that classes first developed in interrogatives then spread to demonstratives and nally to relative pronouns The last part of this claim probably is trueia parallel exists in IndoEuropean languages where relative pronouns have come from question words as in the Romance languages and English cf French qui English who and from demonstratives as in German cf der the or relative who for masculine singular nominative nouns However the only claim of importance in part 3 of Diop s scenario is that interrogatives served as a model for demonstratives and this in turn is important only because the noun class markers of the langues n gro a icaines putatively are related to the consonants seen in Egyptian interrogatives Since there is no credible evidence for this relationship a directionality of relationship between interrogatives and demonstratives ceases to be of interest 4 The Real Linguistic Relatives of Egyptian in Africa Consider the title of the book under review here Parent g n tique de l gyptien pharaom39que et des LANGUES NEGRO AFRICAINES my capsiRGS In Chapter VII pp 11715914 Diop brie y mentions the following languages Serer S rere Diola 261a Fula Peul Nuer Shilluk Shillouk Swahili Kinyarwanda and Bambara Not surprisingly there are specific and convincing connections between most of these languages and Wolof inasmuch as all but Nuer and Shilluk are members of the Niger Kordofanian family and the first three along with Wolof are members of the West Atlantic subfamily of NigerKordofanian Diop makes only sparse and schematic comparisons of these languages directly with Egyptian In most cases there is an assumption that a relationship between Wolof and Egyptian has been established and therefore if a connection between language x and Wolof can be established then language x is likewise connected to Egyptian This is in principle a good assumption but it depends crucially on the strength of the EgyptianWolof connection A case in point is initial consonant alternations found throughout the West Atlantic languages Diop 1977117 says Les altemances consonantiques rencontres en egyptien ancien en walaf en peul etc sont attestes en serere The consonant alternations found in Ancient Egyptian in Wolof in Fula etc are attested in Serer This is followed by a Serer example 0 pad a slave fad ne slaves The pf alternation here can be found in many word pairs in West Atlantic languages such as Wolofpo game f0 to play Fulapullo a Fula person ful5e Fula people As Greenberg 19662527 has noted these unusual yet highly specific alternations form a strong piece of evidence for the relatedness of these languages On the other hand the only example I can find in Diop 1977 of such an alternation in Egyptian is per house fart Pharaoh the latter putatively a plural of the former This is nonsense not only from a semantic point of view but also from the point of view of structure Nominal plural in Egyptian is marked by w ie prw houses In fact when one looks at Diop s list of supposed consonant alternations pp 7378 the alternations he speaks of are not within Egyptian but between Egyptian and Wolof eg Egyptian pP d be turned upside down Woloffaxaj fahad dislocation ofjoint the NigerKorfanian noun classes than the interrogatives In the Egyptian hieroglyphic writing system the written representation of almost every word includes a determinative This is a symbol which tells the semantic class of a word eg a hieroglyph representing a house for words referring to dwellings a tree for the word tree and words for types of trees a child for words referring to children a schematic map for words referring to places a scroll referring to abstract concepts such as peace a phallus referring to male creatures walking legs referring to motion Gardiner 1979438543 lists several hundred such determinatives However this is only a typological parallel showing the human tendency to put things and actions into semantic categories Unlike NigerKordofanian noun class markers the hieroglyphic determinatives played a role only in the written representation of words not in their pronunciation 14Strangely this chapter is not listed in the Table of Contents of the book African language relationships 19 In short the rather extensive discussion devoted to consonant alternation in the sections on Serer Diola and Fula provides evidence for the genetic affiliation of these three West Atlantic languages but is irrelevant to establishing a relationship with Egyptian which had nothing remotely resembling consonant alternations Diop s chapter on languages other than Wolof thus provides evidence that a few NigerKordofanian languages are related to each other a fact long recognized by Africanist scholars But what of the more than 1000 other languages of subSaharan African languages which Diop presumably lumps together as langues n gro africaines Greenberg 1966 divides them into four families NigerKordofanian NiloSaharan Afroasiatic and Khoisan15 He places Egyptian in Afroasiatic ie in a different family from the few languages which Diop has specifically has included in his discussion However Greenberg does include within his Afroasiatic family several groups of people who are n gro afrz39cains in the sense that they are black and their homeland is south of the Sahara viz the speakers of the Cushitic languages of northeast Africa such as Somali and Oromo the speakers of the Semitic languages of Ethiopia and Eritrea such as Amharic and Tigrinya and the speakers of Chadic languages of central and west Africa such as Hausa and Margi I will limit discussion here to the Chadic languages because this is the family with which I am most familiar because speakers of these languages unequivocally count as n gro africams and because they are essentially as geographically remote from the Egyptian homeland as are the languages which Diop discusses 41 Lexical evidence for an Egyptian Chadic connection If Greenberg is correct in grouping Egyptian and Chadic into a single family we should be able to link the Chadic languages to Egyptian on the basis of specific lexical and morphological evidence of the type that is lacking in Diop 1977 Such evidence exists Consider the following wordlists of Egyptian two Chadic languages and for comparison Wolof T h e hyphenated t s in Egyptian are feminine suffixes The hyphenated w in water is a plural suffix The hyphenated k s in Ngizim are nominal suffixes Egyptian Ngizim Hausa Wolof what In ta m me lan two 16 snw sirin biyu aar three xmt kw an uku ett four fdw fodu hudu ent bone ks awuk kasi yax tongue ns marnyi halse l mmi excrement hs sau kasi duul wate m w am ruwa ndox fire x t aka wuta safara sun sky p t sky afa sun rana sama jant asamaan oil mrh t mora k mai diw lion rw wura k leopard zaki gaynde die mt motu mutu dee eat t bread ta ti lekk eat am food 15Since Diop provides no overall classification of African languages nor of Africans by race it is difficult to tell what he means by NEGROaf caines Speakers of the Khoisan languages of southern Africa are by and large racially distinct from speakers of languages from the other families This issue is not important for present purposes however 16Ehret 1995273 does not consider the sn Egyptian and sr Chadic roots to be cognate However I believe that there is too much similarity for this resemblance to be a result of chance African language relationships 20 knOW sni resemble zogaw sani xam snw companion jOin gmi find ndagomu meet gama join daje meet fekk find come ii iw yi go ya come w come dem go The parenthesized forms in the Chadic languages are not claimed to be cognate with the corresponding Egyptian forms They do however raise an issue in comparative linguistics viz comparing languages pairwise as Diop did is fraught with problems17 On the one hand if one has enough imagination and does not use other languages as a check one can probably find apparent resemblances between any two languages to demonstrate relationship as in the list of fictitious WolofEnglish cognates above If one were to compare only Hausa and Egyptian one would be tempted to relate Hausa rand sun to Egyptian r sun sun god Jinju l9936 does exactly this However looking across the Chadic family one finds no other language with a word in the meaning sun sky or related concepts which resembles the Hausa word The origin of Hausa rana is a mystery but one thing is certain Hausa did not inherit this word in this meaning from protoChadic and hence could not have inherited it from protoAfroasiatic it therefore is unrelated to the Egyptian word in question On the other hand comparing languages in groups helps confirm relationships which would otherwise be less clear If we were to draw up a large list comparing Ngizim and Hausa we would find many cognate pairs confirming a relatively close genetic relationship Both these languages are also related to Egyptian though far more distantly than they are to each other As time passes different words are replaced in different languages eg Ngizim has retained the original Chadic word for two reconstructable as something like sar whereas Hausa has replaced it with a NigerKordofanian borrowing biyu conversely Hausa has retained the original word for bone reconstructable as something like Ja u which Ngizim has replaced see Newman 1977 for reconstructions The existence of the reconstructable roots in many languages across the family gives us confidence that we can relate them to Egyptian even though a particular root may be missing from a particular language Returning to the comparative list above it should be obvious which languages are genetically related The Chadic languages show repeated resemblances with Egyptian a set of resemblances which could be extended by extending the list of words and the number of Chadic languages we compared to Egyptian On the other hand there is not a single good match between Wolof and Egyptian or Wolof and the Chadic languages Note moreover that the words here are from basic vocabulary the part of a language s lexicon which we expect to be most stable and the part of the lexicon where even claimed correspondences were notably absent between Egyptian and Wolof 42 Grammatical evidence for an Egyptian Chadic connection Turning to grammar Diop proposed that Egyptian had developed a series of interrogatives which varied in form but not meaning and that the system of noun classes in Wolof was historically related to the Egyptian interrogatives and extended by euphony with the consonants of nouns I showed that both the claim of meaningless variation in the Egyptian interrogatives and the euphonic nature of noun class marking are false and moreover that any functional link between interrogatives and noun classfication is dubious at best In contrast to this farfetched scenario linking Egyptian interrogatives and Niger 17What follows is basically an exposition of the principle of mass comparison extensively explicated in Greenberg 1966Chapter l and elsewhere in Greenberg s work African language relationships 21 Kordofanian noun classes the systems of nominal classi cation and nominal morphology in Egyptian and Chadic languages resemble each other in a number of details 18 In Egyptian all nouns fall into one of two genders masculine and feminine and plurality is a separate category from gender though in some cases the gender distinction is neutralized in the plural Gender and number show up in a variety of grammatical constructions including the following on noun cup ye ar cups rm ny wrw n w these birds years rm ny n t wrw In its general outline this is a system found in many Chadic languages singular nouns are grammatically masculine or feminine gender and plural is a separate category from gender Compare the Egyptian system above with the following examples from two Chadic languages Hausa H and Warji W19 Marking 0n noun Demonstrative Possessive link Masc sing H ba o guest m H wan can ba o H ban n su that guest their m guest W zhiBa na ram W sanda n tana W zhiBa n sara this stick ram of the chief Fem sing H ba u wa guest f H wac can ba uwa H ba uwa r su that guest their f guest W soma i knife W danga i cina W somi to sara this pot knife of the chief Plural H baE i guests H wadan can baEi H baEi n su W Zhi awina ltramsgt those guests their guests sgmawina NO Warji examples With No Warji examples with a plural knives plural demonstratlves possessed noun In terms of system the only difference between Chadic and Egyptian is that no Chadic language marks a masculinefeminine distinction in the pluralithe plural of a noun root can refer to all males all females or mixed sexes unless of course the root could only refer to males or females such as stallions or mares In pronouns Egyptian and Chadic share even the gender neutralization feature in the plural see below More important than the abstract system which is not unlike that of a number of European languages such as Spanish or French are the specific markers of gender and number in particular I feminine and n plural The ubiquitous marker of feminine gender in Egyptian is t which shows up as a suffix on virtually all nouns of feminine gender and on all modifiers of feminine nouns The nominal suffix and the agreement on demonstratives are seen in the table above Numbers also show the agreement eg ijw one m ijt one f snyy two m snty two f etc The n plural in Egyptian shows up primarily in the demonstrative system seen in the table above 18Greenberg 19664648 mentions most of the resemblances here 19Warji data come from Newman nd African language relationships 22 Chadic languages have the same tfeminine n plural pattern Warji cites all nouns with a suffix na for masculine singular and plural and az39 for feminine singular The feminine 139 comes orginally from t which changed to 139 in Warji and closely related languages except at the beginning of words The t feminine n plural pattern is seen in the Hausa demonstrative elements wag feminine from wat and wadtm plural There are no examples with plural demonstratives in the Warji data available to me but in the very closely related Miya language we find forms such as tak9n am this woman mm tavam these women Schuh 1989 The t feminine genitive linking element is seen in Hausa f originally t which regularly becomes 7 at the end of a syllable and Warji t9 Hausa has plural n genitive Miya again can supply the n plural in tamakwiy my Vaziya Vaziya s sheep pl 20 Another specific resemblance in noun marking between Egyptian and Chadic is a u w plural ending This is the regular plural in Egyptian see table above It is also fairly common in Chadic languages eg Hausa layalayg charmcharms Miya kamkamamaw househouses and the wi suffix in the Warji nouns above used together with the n also associated with plural Unlike the tfeminine n plural which represent a pattern which is indisputably inherited into both Egyptian and Chadic I have less confidence that this u w plural is an inherited feature but the pluralization type certainly is the same in Egyptian and Chadic in contrast to NigerKordofanian class languages which do not mark pluralization on nouns as a separate category at all In the previous paragraph I emphasized the word pattern in referring to t feminine n masculine Individual resemblances between languages as in vocabulary items or the possibly plural ending LM w are good evidence for genetic relationship but any particular resemblance could be a result of simple chance as in English die Wolof dee On the other hand resemblances which form an interlocking pattern greatly reduce the odds that the overall resemblance could be a result of chance rather than genetic inheritance The pronominal systems of Egyptian and Chadic present such a pattern and indeed when compared with pronominal systems in Semitic Berber etc these systems provide the strongest evidence for the unity of Afroasiatic Consider the following paradigms of pronouns In Egyptian and Chadic the forms on the left side of the slash are essentially the independent form of the pronounithe pronoun that would be used in an exchange such as Who s there e 21 The forms on the right of the slashes are possessive clitic pronouns In Wolof the forms on the left are the independent pronouns the forms in the middle are preverbal subject clitics and the forms on the right are possessives The parenthesized forms in Egyptian and the Chadic languages are probably not cognate 20Greenberg 1960 describes an Afroasiatic pattern n masculine tfeminine n plural The masculine singular form varies more throughout the Afroasiatic family than the feminine and plural forms Thus in Egyptian pf are associated with masculine singular Callender 19751516 mentions a neuter n which he lumps with plural This may be a remnant of the Afroasiatic n masculine though another possibility is that it is simply the plural in impersonal reference something like English they as in they say In Chadic s and k are often associated with masculine singular The variation in Afroasiatic masculine singular forms is probably related to the fact that masculine singular is usually the default form when gender andor number are unknown or irrelevant See Schuh 1983 for gender marking elements in Chadic 21The Hausa forms are actually the subject pronouns used in the Perfective aspect However they correspond to the independent pronouns in closely related languages African language relationships 23 Egyptian Ngizim Hausa Wolof memy wi i iyu a na a manmasama youyour mSg w k l5l ka ka yowngasa m or f youyour fsg n k9m kom kin ki himhis sw f a i ri ya shi moommu am m or f herhers sy s atu ra ta ta usours n n Ga wa ja wa mun mu nunnusunu youyourS p1 tn Vin kun kun kun ku y nngeenseen theytheirS sn sn aksi k i sun su oomn seen Egyptian and Chadic share a number of points of detail First person singular This form is vocalic whereas all the other person have the form ConsonantVowelConsonant Second person singular Both Egyptian and Chadic as well as Semitic and Berber distinguish male and female addressees The masculine form can be reconstructed kV This was probably ka though Ngizim and it appears Egyptian have changed to i probably via some unstressed neutral vowel The change in vowel has caused the kto palatalize to d misleadingly represented in standard Egyptian transliteration More striking than the k masculine pronoun correspondence is its pairing with a feminine pronoun which has two forms k N probably more specifically kim in the free pronoun but ki in the clitic form Correspondences at this level of detailipaired masculine and feminine both with initial k and a feminine with two forms one ending a nasal consonant and one lacking itican have no explanation other than inheritance from a common ancestor Second and third persons plural Both Egyptian and Chadic have the consonant series k n second person s n third person where the n is probably the general n plural discussed above First person plural The West BiuMandara and Masa branches of Chadic have lost the n first person plural forms seen in Egyptian However n first person plural is retained in the East branch of Chadic and presumably is reconstructable for protoChadic eg Dangaleat has lst plural subject pronouns ni we inclusive 111 we exclusive Ebobisse 1979 p 30 Third masculine singular Though the archetypal pronoun for 3rd masculine singular in Egyptian is pf the free pronoun has s which is one of the more widespread Chadic forms cf Hausa possessive clitic In short Egyptian and Chadic match in detail throughout the entire system of personal pronouns Indeed the only apparent non match is 3rd feminine singular where Chadic has forms with the ubiquitous tfeminine but Egyptian has forms in s Even here one could speculate on a development in Egyptian something like ti her gt E i gt si gt St though I know of no independent evidence for this Contrast the correspondences between Egyptian and Chadic with the non correspondences between Egyptian and Wolof The one person where Wolof looks better than the sample Chadic languages is the n first person plural But a chance resemblance or two is not unexpectedieven English me and you resemble the corresponding Wolof pronouns In any case we saw that an n is reconstructable for Chadic though it is lost in the sample languages cited here Diop 197725 compares the Egyptian and Wolof pronouns without much comment apparently assuming that resemblances are obvious He does mention one resemblance which does not emerge in the table above viz Wolof 3rd masculine singular f a striking resemblance with Egyptian pf However this form in Wolof appears as part of only one tense aspect marker dafa as in da fa bay he FARMED African language relationships 24 Diallo 198363 refers to this form as Emphatique du Verbe The canonical consonant associated with 3rd singular where there is any consonant at all is m seen in the examples in the table In short there is a parente39 gene39tique ale l e gyptien pharoaonique et ales languages ne39gro africaines but this parente is not the one that Cheikh Anta Diop proposes viz one between Egyptian and Wolof Rather it is between Egyptian and the languages of the subSaharan branches of the Afroasiatic family ie the Cushitic EthiopianEritrean Semitic and Chadic groups I must make very clear the claim that I have been making in comparing Egyptian and Chadic and that I also could have made for Egyptian and the Cushitic and Semitic languages Egyptian and the Chadic languages are related in that they have all descended from an ancient ancestral language for which we have no historical recordithe Chaalic languages did not descend from Egyptian Diop also did not claim that Wolof and the other languages he mentions are direct descendants of Egyptian It is actually quite likely that these languages are related to Egyptian but if they are it is a relationship far more remote than that between Egyptian and Chadic I return to this point below 5 Conclusion All African Languages Probably are Related So what In this paper I have gone to some lengths to show the untenability of a particular language origin theory but more important is a general lesson about the use of linguistic data in history the only data relevant for lingustic classification is linguistic alata anal one must let the data leaal where it will without preconceptions about what the results should be There is nothing original in this statementithese are points made repeatedly in Greenberg s work on classification I have no question that Cheikh Anta Diop was sincere in his attempt to show a relationship between the Egyptian language and what he calls les langues ne39gro africaines In fact I have no question that Egyptian and these languages are related The uman species had its origin in Africa and human language had its origin in the species These facts mean that human language has existed on the African continent longer than anywhere else in the worldiperhaps in a time frame measured in hundreds of thousands of years Greenberg s classification of African languages into four great families is now universally accepted and has been repeatedly confirmed by more detailed study and additional data As far as I know Greenberg himself has not proposed a yet higher grouping of these families but the following supertree of the African families is not inconceivable Dotted lines connecting to a node indicate that a proposed grouping based on empirical evidence has been made Gregersen 1972 dashed lines mean that no proposals supporting a genetic relationship have been made in print as far as I know protoAfrican protoHuman A I SupraEquatoria I I x l I CongoSaharan Khoisan Afroasiatic NiloSaharan NigerKordofanian Greenberg applying the italicized principles in the first paragraph of this section proposed a single linguistic family Afroasiatic which includes Egyptian the language of African language relationships 25 one of the greatest civilizations in human history and Arabic and Hebrew the liturgical languages of two of the world s great religions on the one hand and on the other hand the 150 or so Chadic languages of central and west Africa many of which are spoken by people who lived in isolated mountain villages wore no clothes and practiced animist religions well into this century Judging by the diversity of the languages across the major branches of AfroasiaticiEgyptian Semitic Cushitic Berber and Chadicithe ancestral language of all these languages must have been spoken at least 1015000 years ago This is two to three times the depth of the IndoEuropean family which itself comprises languages as structurally different and as geographically distant from each other as English and Hindi The implication of these observations is that if the tree above presents a reasonable hypothesis of relationship the time depth between any given language of the Niger Kordofanian family eg Wolof and Egyptian must be immenseiperhaps 20000 years or more Over such a time period it would be astonishing to find any linguistic resemblances that could be unequivocally attributed to genetic inheritance much less the large number of resemblances that Diop claims to exist between Egyptian and Wolof The only conclusion that one can reach is that Diop assumed these languages were related then looked for ways to connect features of the languages by using leaps of faith and disregarding comparative evidence from other languages which would have cast doubt on the claimed resemblances As someone who has devoted his professional life to the study of African languages I have long been bemused even constemated by the fact that group after group of African people have fanciful histories that trace their origins to somewhere in northeast Africa or the Arabian Peninsula I have seen published works which trace the Yorubas the Ewes and the Hausas to Egypt by showing putative lexical and grammatical similarities between the respective languages and Egyptian though none attain Cheikh Anta Diop s detail and sophistication of argumentation and unlike Diop who claims that Wolof and other languages share an origin with Egyptian these works claim the respective languages to have descended from Egyptian22 Another theory shared by some Hausas is that they come from Ethiopia called Habasha in Hausa whence the name Hausa and I have collected oral histories from a people speaking other Chadic languages who claim to have come from Yemen or Saudi Arabia One wonders what origin myths existed in subSaharan Africa prior to the arrival of Arab intellectuals and European explorers One strongly suspects that the widespread fascination with an eastern origin must be an attempt by people to connect their ancestors to a civilization which has been legitimized by the establishment but what does ancient origin have to do with modern worth The great civilizations of Egypt Mesapotamia Greece Rome the Yucatan Central Mexico and the Andes have long ago disintegrated thanks to a variety of human and natural forces The ancestors of Western Europeans and their North American cousins whose civilization for better or worse now dominates the world stage were illiterate tribal barbarians a mere 2 millenia ago when Rome ruled the Western world The modern descendants of any of these societies regardless of the state of the society where they currently live cannot take credit for the achievements of their ancestors nor be blamed for their evils Egypt was home to some of the world s great intellectual artistic and architectural achievements but it also had a lousy human rights record with brutally despotic rulers concentration of wealth power and education in the hands of a tiny minority and slavery for untold numbers Today s Africans did not design the pyramids nor did they enslave the thousands who must have suffered to build them The same applies to today s western Europeans even though 22For Yoruba see Lucas 1964 I regret thatI cannot supply an exact reference the Ewe study referred to I looked through this book on Ewe in 1988 when I saw it on sale at the Village du Benin a language institute in Lom Togo For claims of EgyptianHausa connections see Jinju 1993 and other references to his own work which Jinju cites African language relationships 26 Meinhof con dently claimed the Hamites whose prototype was the Egyptians were of Caucasian origin Examination of Egyptian linguistic data proves only one thing the Egyptian language is an Afroasiatic language and hence is a descendant of the same ancestral language as all other Afroasiatic languages regardless of who speaks them REFERENCES Alexandre Pierre 1967 Langue et language en Afrique noire Paris Payot Callender J B 1975 Middle Egyptian Malibu Undena Publications Diallo A 1983 Elements syste39matiques alu wolof contemporain Dakar Centre de Linguistique Appliquee de Dakar Diop C A 1977 Parente genetiquerale l egyptien pharaonique et ales langues negro africaines Dakar Les Nouvelles Editions Africaines Ebobisse C 1979 Die Morphologie ales Verbs im Ost Dangaleat Guera Tschaal Berlin Dietrich Reimer Ehret C 1982 Linguistic inferences about early Bantu history In C Ehret and M Posnansky ed The Archaeological anal Linguistic Reconstruction ofAfrican History pp 5765 Berkeley Los Angeles London University of California Press Ehret C 1995 Reconstructing Proto Afroasiatic Proto Afrasian Berkeley Los Angeles London University of California Press Fal A R Santos amp JL Doneux 1990 Dictionnaire wolof francais Paris Karthala Faulkner R O 1976 A Concise Dictionary of Middle Egyptian Oxford Printed for the Grif th Institute at the University Press by Vivian Ridler Gardiner Sir A 1979 Egyptian Grammar 3ral eal Oxford Griffith Institute Greenberg J H 1960 An AfroAsiatic pattern of gender and number agreement Journal ofthe American Oriental Society 80317321 Greenberg J H 1966 The Languages ofAfrica Bloomington IN University of Indiana Greenberg J H 1978 How does a language acquire gender markers In J H Greenberg ed Universals ofHuman Language pp 4782 Stanford Stanford University Press Gregersen E A 1972 KongoSaharan Journal ofAfrican Languages 1116989 Jinju M H 1993 Asalin Hausawa da harshensu The origin of the Hausas and their language In A Rufa i I Y Yahaya amp A Y Bichi ed Nazari aKan Harshe ala Aalabi ala Al aalu na Hausa Littafi na Uku pp 110 Kano Cibiyar Nazarin Harsunan Nigeria