:. . ‘4 ' ‘ ., ,.J- - ‘ - r- - , .Dr' D ”p I n a“ - ~o...._.r ,- . -~' .4 #“M 1’ .0- .— ”—5" -.-ao-f .r ’ .x. m ”3'" M _ ,,.._...-.--—-=:;:—"‘ __ .- “a..- H' wot”.-. ..—-~.—~s._, .mp0, r- . “ml - i“- m L‘Z'A‘nk. WW... H- .4 M ::.:a: “b-F‘O :L“ H... {.2 ”Q.‘ *- ,O—O “a .. M ‘_‘.'~ 9 p...- - 0 v - ‘ I - O , -. - 3 . - V w . coo-a— ‘ -.‘ - 0 ‘ ~ I —. ”fin—w "up... n.— ‘5. 5 1 ,2'5‘5 5155' i :5 =5 25:" "1.1151147515' . . $15.55 : 5. .1- 1.....1. H...4 .m 5 55 25 '3 555' '151313‘ '555'5 5 ‘5 5 5 51 5‘5! 1'5 - o ' .5555}: {"5 1115555 W15} 55.. .2 1mm“ 555 51‘" 555555.15 I5'55' '55 '55551515'5555‘11I11'55fi5 '5 55555155:595195155553" 31-1.. .. 551:} 555,315,5555115113 15555.55 55555555523131 «2‘513'554'55' ”Lg-T,“ 5*“,1'311‘ 551‘5‘, 55555? ' ‘m 5555* 3.1 151,32 ‘5 5:55:55 '5‘ 555555515! .-’55'5555555555555515511I511.1 45151111111 1'55 "5521155115111 .11.. 15:555 I 55‘" .55 ' .5 5:51: 355 35 1 5. .5 51‘55 5,551.55'5i ’15s 555 5 ”All 5:15 . ., ”11' 5: 5 55"5555515J55. 565.5 5'5! 55545555555 5 '55 2555 5555555555 5115 51 $5555 .. 1.1..15 11m! 1 ‘9"? 5, 5:5‘51111551551111111511'11131‘1‘; '51 I“ 111 51551. 55115 W5 mm. 51.1 .155" 5515.55 "' 5 5'~" 55515111551 3555“ 11.1] 113111155115" ‘55“ 11.111'1'5555 5'55' '55' 5 .5 5555555 55 ”LA .._ "*5; ”2;.; 51! m .1555 5‘51'5515'”? .L 1,; H '1 5'511'5 151 1 . '5" 1H '1,51551:511mlI1151111‘515555555'13:11:53 355' 5 :5 .1 515. .2‘ ‘ 5155 5 5 15 5.1 5155;“. 5.1151 '5"%"‘1 15515555315555.1511 .. . 1.15 51‘ 11511 .1515 5:51 '5 ’1 5. 5555' 5555 55555.5. 15555555115559.1114 .1. 555 55 m 5' 55535555555 5'5 5; 55 511511551511 :15 55.515 1.1. 555.55 1.1.. 55' 7 H37 "‘4 :1... 1-: a". nav' 122..-;‘4 trait Pg'._’.:a—;—~1 Q'n'" 1--"“.. 52442925414 ' “EM’IJZQL Bar—7’7 @322" i‘J-S‘Jt This is to certify that the thesis entitled THE WEST-A-BRANCH OF THE CHADIC LANGUAGE FAMILY: A Comparative Study of Hausa, Sha, Angas, Karekare, and Dera presented by Beverle Michaele Lax has been accepted towards fulfillment of the requirements for M.A. Linguistics and Languages M Major professor degree in Date u/Mé an; 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution RETURNING MATERIALS: IIV153I_] Place in book drop to LJBRARJES remove this checkout from ‘=!,,~,,_._ your record. FINES wiII - be charged if book is returned after the date stamped beiow. THE WEST-A-BRANCH OF THE CHADIC LANGUAGE FAMILY: A Comparative Study of Hausa, Sha, Angas, Karekare, and Dera BY Beverle Michaele Lax A THESIS Submitted to Michigan State University. in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Linguistics and Languages 1986 ABSTRACT THE WEST-A BRANCH OF THE CHADIC LANGUAGE FAMILY: A Comparative Study of Hausa, Sha, Angas, Karekare, and Dera BY Beverle Michaele Lax This thesis is a comparative study of five languages from the west-A branch of the Chadic language family as classified by Paul Newman (1977). The comparative method is employed to derive a proto-lexicon for the Chadic West-A branch, using the Hausa, Sha, Angas, Karekare, and Dera languages. The main purpose of the thesis is to test whether or not Newman's 1977 proto-Chadic reconstructions adequately reflect the lower levels of his Chadic classi- fication as well as the Chadic language family as a whole. A comparison of the Chadic west-A reconstructions, termed 'Provisional Reconstructions,‘ with Newman's 1977 proto- Chadic reconstructions demonstrates the outcome of this test. Contrary to expectation, more similarities than dissimilarities are shown, suggesting that Newman's reconstructions do adequately reflect the lower levels of the Chadic classification. But, due to the scarcity of cognates revealed in the comparative study of the five languages, this finding has conflicting implications. To My Beloved Grandfather, Alfred Lax, Sr. (1898-1986) ii l:oer:1i.Ht:lH 5.:7‘Icolsil ,H “I p u .- 9 ’ ’O- : fu-'"‘ ll - :‘.a ‘ ;" . \ O O ACKNOWLEDGMENTS A special thanks is extended to Dr. James Wang, chairperson of the Linguistics and Languages Department, for his faith in me as a graduate student. I thank my advisor and chairperson of my committee, Dr. Grover Hudson, and the other members of my committee, Dr. Ruth Brend and Dr. John Eulenberg for their suggestions, comments and patience in the writing of this thesis. I would also like to acknowledge Dr. Patrick R. Bennett and Neil Skinner of the Department of African Languages and Literature, at the University of Wisconsin-Madison, for their invaluable suggestions and comments. I thank my mother, Frances, for drawing Figures 1, 2, and 3 in this thesis. I extend my gratitude to my family, especially Frances, and my father, Mickey, and to my friends, especially Abu, Bobbye Faye, Doug, Heide, and Lamont for their encouragement, confi- dence, and moral support, who are as proud and relieved as I over the completion of this thesis. iii TABLE OF CONTENTS CWTER I O O O O O O O O O O O O O 0 INTRODUCTION Chadic Linguistics Purpose of Thesis. Method . . . . . . Procedure. . . Expectations . Results. . . . O O C O O O O O O O O O O O C O O O O O O O O O O O O 0 O O O O O O CWTER II. C O O O O O O I O O O O O HAUSA, SHA, ANGAS, KAREKARE, AND DERA: GENERAL AND LINGUISTIC INFORMATION General Information. . . . . . . Linguistic Information . . . . . Hausa Phonemic Inventory. . Sha Phonemic Inventory. . . Angas Phonemic Inventory. . Karekare Phonemic Inventory Dera Phonemic Inventory . Chadic Phonological Traits . General Traits. . . . Hausa Traits. . . Angas Traits. . . Dera Traits . . . CHAPTER I I I O O O O O O O O O O O O O PROTO-FORMS FOR CHADIC WEST-A Phonemic Inventory of Provisional Reconstructions. . . . . . . . . Consonants. . . . . . . . . Vowels. . . . . . . . . . . iv PJH mawmmw 37 38 Discussion of Phonemic Inventory of Provisional Reconstructions . Consonants . . . . . . . Vowels . . . . . . . . . 'Regular' Sound Correspondences Consonants . . . . . . . Vowels . . . . . . . . . Initial Vowels. . . Medial Vowels . . . Final VOwels. . . . Discussion of 'Regular' Sound Consonants . . . . . . . waels . . . . . . . . Initial Vowels. . Medial Vowels . . Final Vowels. . . Verbs. . . . Nouns. . . . CHAPTER IV C O I O O O O O O O O O WEST-A PROVISIONAL RECONSTRUCTIONS 1977 PROTO-CHADIC RECONSTRUCTIONS Discussion of Similarities. . orr 000000000000 VS. ooooooomoOooooooo spond coco-000000000090 00000000000000... Comparison of Provisional Reconstructions for Chadic West-A with Newman's 1977 Proto-Chadic Reconstructions . . . . . . . CHAPTER V. . . . . . . . . . . . . CONCLUSION APPENDIX A O O O O O O O O O O I O COGNATES AND PROVISIONAL RECONSTRUCTIONS CHADIC WEST-A Cognates Across Five Languages. Cognates Across Three Languages Cognates Across Four Languages. . Cognates Across Two Languages ”PENDIX B C O O O O C O O O O O O COMPARATIVE WORD LIST LIST OF REFERENCES . . . . . . . . FOR 87 88 92 100 100 101 101 103 107 130 LIST OF FIGURES Figure l . . . . . . . . . . . . . . CHADIC FAMILY (Taken From Greenberg 1963) Figure 2 O O O O O O O O O O O O O O CHADIC FAMILY (Taken From Newman And Ma 1966) Figure 3 . . . . . . . . . . . . . . CHADIC FAMILY (Taken From Newman 1977) Figure 4 O O O O O O O O O O O O O O LANGUAGE MAP (Taken From Ikwue 1984) Figure 5 O O O O O O O O O O O O O O ORGANIZATIONAL CHART OF RESULTS vi 20 94 ABBREVIATIONS Angas Dera Hausa Jungraithmayr Karekare Newman noun Provisional Reconstruction Sha Schuh verb vii CHAPTER I INTRODUCTION Chadic Linguistics Chadic languages are spoken south of the Sahara in northern Nigeria, northern Cameroon, around Lake Chad in the Republic of Chad, and in Niger. For years linguists have attempted to classify the many languages spoken in these areas. Most of them are now recognized as a linguistic unit now called the Chadic language family, which is generally considered as one of the six branches of the Afroasiatic family of languages.1 Johannes Lukas was among the first to attempt a classification of these languages. According to Newman and Ma (l966), Lukas (1936)2 classified some of them into two different groups, Chado-Hamitic and Mandara, based chiefly on the typological criterion of grammatical gender.3 Languages such as Hausa which is characterized by the presence of grammatical gender were classified as Chado-Hamitic, while languages without grammatical gender, such as Margi, were classified in the Mandara group. As noted in Newman and Ma (1966), Lukas (1952)4 expanded each 2 group by including more of the languages from the geograph- ical areas concerned. There were similarities between the two groups but Lukas "contended that these were not ade- quate to justify combining them into a single family" (Newman and Ma 1966:218). Not until Greenberg's 1950 publications in which he attempted a classification of these languages, were they recognized as forming a single linguistic unit (Newman and Ma 1966). Coining a name from the geographical regions in which these languages are spoken, Greenberg termed this linguistic unit the 'Chad' language family, which is referred to today in the literature as the 'Chadic' language family. Greenberg's classification included more languages from the geographical areas concerned than did Lukas's. He classified the languages into nine groups (see Figure 1) versus Lukas's two-group classification. While Greenberg presented linguistic evidence:to illus- trate the genetic relationship of the Chadic languages with the so-called Hamito-Semitic languages, he did not present separate linguistic evidence to establish that the Chadic languages in fact formed a single linguistic unit (Newman and Ma 1966:219). ”His proof of the unity of the Chad family was thereby rendered weaker than it need have been” (Newman and Ma 1966:219). Consequently, in their 1966 publication, Newman and Ma set out to "demonstrate conclu- sively that the Chad family as postulated by Greenberg does indeed constitute a valid linguistic unit" (p. 219). (415m. TMMM 30“,“ (u “on, w..- ,MJ‘; RIM)“ ”Md ' “mm... ”M,“ gmmlfia-ghhg‘ NM‘. g “A,“ a, Gilda- !” “Light’s,” 0 ”39mm,“ NW, 8...,“ 09., MM Jimmrfi. 'h" Gammvhh, (on... may... m Odn,“ fish‘s-silk» 3 Duo div-‘0' 6w 03.“.M-nmmma-m but. ll.) MM (5) If.» WWW, Jada '4 U4) “jun—IMMAWJM «7‘81 ,MMdM when ,lanpnn,l‘ino,h‘om,l~l A. Figure l CHADIC [AMI 1.! (Taken Iro- Greenberg 1963) 4 Following Lukas's framework for his classification, Newman and Ma reclassified the languages into two groups which they named Plateau-Sahel and Biu-Mandara. They also postulated subgroupings for the languages within the classification (see Figure 2). Subsequently, Newman and Ma presented proof that the Chadic family formed a valid linguistic unit by identifying cognates and positing recon- structions for one hundred and forty-five Chadic lexical items, based on regular phonological correspondences established between their postulated Plateau-Sahel and Diu- Mandara groups. The reconstructions were rated as either first level confidence reconstructions for which there was little doubt concerning their accuracy, or second level confidence reconstructions which were based on insufficient data. The Newman and Ma 1966 reconstructions did not include vowels, only consonants, and, as Newman states in his 1977 publication, were sometimes based on questionable data. Upon the availability of more reliable data, Newman (1977) re-evaluated the lexical reconstructions posited in Newman and Ma (1966). Newman omitted some, added others, and revised more than half of the reconstructions posited in Newman and Ma (1966). This re-evaluation yielded a total of one hundred and fifty proto-Chadic lexical reconstruc- tions which include vowels as well as consonants, and all are described by Newman as having a 'high degree of confidence.‘ These reconstructions are based on regular Vudala (Nadya) “WMMMJ 5% Burn) 0"“: Uhfivfi.m Vt 9< V R 7 \ Q fi§ Fa li-Lirla, Ilfgt, rid/z l N r! a § Wt, mm,lzu‘.;&ur, M g hhndafhhh),l{ana,5m,7."w£flm) 9:: Jsgu ,fluh’, Joloro,Jma¢', Elna-i Q a; 5‘. 3;: Angas, Am,m,mml, 9) Sara , Q v: 3 R: a .Bolm,hu(m1,wt; Y 3 g Kankan u, o k 9d ‘s 3% \1 AV izm. & 5‘) 3 Hana. flaw» I611” Figure 2 CHADIC FAMILY (Taken From Nemnan And Ma 1966) 6 sound correspondences, and what Newman calls sound laws are given to account for the present-day reflexes in individual languages and language groups. In proposing these improved proto-Chadic lexical reconstructions6 Newman presented a revised classification of the Chadic languages which demonstrated '...greater and more precise internal struc- ture than provided in any previous classification" (l977:2). Newman's 1977 Chadic classification consists of four branches,7 the Masa, East, Bin-Mandara, and West,8 which include approximately one hundred and twenty-five lan- guages9 (see Figure 3). Though there remains room for improvement regarding the classification of the Chadic languages, Newman's 1977 classification is the most compre- hensive and most widely accepted to date. It is Newman's proposed subbranch A of the west branch with which the present study is concerned. Purpose of Thesis This thesis is a comparative study of five languages from the West-A branch of the Chadic language family as classified by Newman (1977). The hypothesis of this study is that a proto-lexicon for Chadic may be derived which is more reflective of Chadic as a whole, and of the lower level structures (e.g. branches and subbranches) within the classification, by first positing proto-lexical forms . “It, 3m “mm; ammm,mhnmmflwmm a rm, M,“’ ”at“! all... an, m, 8%“ M)M,h, ”so,“ am, an, hm,mm 31’ l u, mm,“ m, ha, 80, mm MEN“. M031. 7356“. ”meh~f" run, Totem, Gannutmf c ”as,“ u "I...“ 0 "has...“ a m-hu,auw,am I‘Uuwgmun mu, has (an-u : mtml,~,m,mb~.6voko Mom,.mfium,m,m~m Jan Jan, ma, Aim m ouch“, can Mulls” CM), Cue Eu. mu“ m, tea-Ink“ 1M; ‘1“ Ito-nu, 7m, m, was, magma. inhuman, “.mu Ian, M l .3me man, m 3“,“, In»! M a“, Manna, M, Ann. M I “(as Mm, Sou 3 Ma. 11a. Ava-n. nun, may § ”I.“ Figure 3 r: [s G H >1 :4: 5% an“?! 38 H 3a.. LI: 0) .2 fl 3‘. 8 for each branch or subbranch and subsequently comparing these proto-lexica, rather than the method of comparison employed by Newman (1977). In this publication, Newman used a limited number of citations from sometimes as few as two of the major branches in the Chadic family as a basis for his lexical reconstructions, which are meant to repre- sent the entire Chadic language family. The purpose of this comparative study is then two- fold. It suggests the viability of positing reconstruc- tions at the lower levels first, and subsequently comparing these proto-lexica for reconstructing proto-forms for Chadic. Secondly, it tests the viability of the suggestion by using the West-A branch to answer the question, 'Do Newman's 1977 reconstructions which purportedly represent the entire Chadic family adequately reflect its individual branches and subbranches?‘ The lexical reconstructions derived from the comparison of five west-A languages are compared with Newman's 1977 lexical reconstructions in an attempt to provide an answer to this question. The hypothesis proven, subsequent work would include positing proto-lexical forms from comparative studies of languages from each individual branch or subbranch to derive a proto-Chadic lexicon reflective of the entire Chadic family as well as each individual branch and sub- branch. 9 Method The goal of this thesis is approached through the comparative method. Proto-West-A lexical forms are reconstructed based on 'regular' sound correspondences deduced from the comparison of five languages from the west-A branch (Hausa, Sha, Angas, Karekare, and Dera (Kanakuru),10 and regular sound correspondences noted and cited by Newman and Schuh in their various publications for the appr0priate languages. These reconstructed lexical forms which represent a lexicon from which the languages in the West-A branch may have derived are then compared with the corresponding proto-Chadic lexical items posited by Newman (1977). The outcome of this comparison, illustrated by highlighting the similarities and differences between the two sets of hypothesized forms, serves as the criterion used to deter- mine the answer to the question posed in the purpose of this study. Procedure The West-A branch is divided into four groups of languages as shown in Figure 3: the Angas, Bole, Hausa, and Ron groups. Based on the assumption that languages classified together at the group level are closely related enough, such that it is possible for any given language of 10 its respective group to represent that group, the intention was to choose one language from each of the four groups of the West-A branch. However, in fact, five languages were chosen to be included in this study; a second language from the Bole group was added. Languages were chosen from each group for which the best documented materials were available: Hausa, by far the best documented Chadic language, was chosen from the Hausa group over the only other member, Gwandara. Avail- able sources for the Ron languages are all equally scarce; Sha was chosen from among them. Angas was chosen from the Angas group, subgroup a (Angas-a group), Dera from the Bole-b group, and the second language from the Bole group included in this study is Karekare, which is a member of the Bole-a group. While there exists more available documented material for Dera, Karekare is included mainly because of the author's interest in the language due to a previous investigation of the language for a study project.11 The primary sources consulted for the corpus of data for the languages concerned are Kraft's Chadic wordlists (1981) for Angas and Karekare (used as a secondary source for Hausa and Dera), Abraham's Dictionary of the Hausa Language (1962) and Newman's The Kanakuru Language (1974) for Hausa and Dera, respectively, and for Sha, Jungraithmayr's Die Ron-Sprachen (1970). The word list for this study was compiled with three ideas in mind; to include glosses for which Newman (1977) ll reconstructed proto-Chadic forms, secondly, to include enough other basic vocabulary items in an attempt to exclude loan words and to maximize the chance of finding cognates across all five languages, and thirdly, to include those basic vocabulary items for which citations could be found from the available sources for at least four of the five languages concerned. Therefore, the glosses that appear in the word list are taken from Newman (1977) and Kraft (1981), which included vocabulary items used by Joseph Greenberg in his 1963 African language classifica- tion. A total of two hundred and ten items appear in the word list of this study.12 Eighty-four of the items used in Newman (1977) are included in the list. Some of the items used by Newman were not included in the word list in this study due either to the fact that lexical citations were not available for at least four of the five languages for the respective glosses, or because some of the glosses used by Newman (1977) proved to be semantically dubious. For example, items with glosses such as 'grass' and 'calabash', for each of which there are many types, were omitted because the identity of the type referred to and the type referred to in the corresponding lexical citations in the sources used for the languages concerned is not always clear.13 Only some of the four hundred and thirty-four items from the Kraft (1981) list were included in the word list 12 in this study. Kinship terms were not incorporated though they are words which are considered basic vocabulary items. Although the words for 'father' and 'mother', in Angas 2222 and 2223, in Karekare 2323 and gang, and in Dera babe and gang, are 'cognate', such forms are nearly ”universal” and are not seen as significant for relationship. Some other glosses were not included because the corresponding lexical citations are derivatives of other words, and yet others like 'grass' and 'calabash' from Newman (1977), were not included because of dubious semantics. A further con- straint, as mentioned above, was the effort to include glosses for which citations were available for at least four of the five languages. The difficulty of finding citations for at least four of the five languages was further compounded by the fact that semantically related items were not conscientiously considered in the comparison of the lexical items for reconstructions, i.e. most of the reconstructions were based on the comparison of respective lexical items with an equivalent gloss. For example, a semantically related item, 'haze', was not considered in the comparison for a possible reconstruction for #166 'smoke'. Consequently, conservative reconstructions are presented in this study. Unless it is meant to be implied otherwise by virtue of the comparative method, evidence in his 1977 publication indicates that Newman took semantically related items into consideration in his reconstructions, but only to a small extent. 13 The word list in its entirety appears below as Appendix B, in which the languages are listed in left-to- right geographic order corresponding to their west-to-east position. Hausa is listed first, Sha, Angas, Karekare, and then Dera. Languages in the word list and in the examples given throughout the discussion are designated by an abbreviation using the first letter of each name: H, S, A, K, D. For each gloss, words appear for at least four of the five languages concerned. Items in the word list are as they appear in the sources consulted, i.e. the phonemic symbols, vowel length and tone markings are those of the sources from which they are taken (with the exception of Hausa; vowel length and tone markings are taken from Abraham (1962) but the convention used for these markings is that used in Kraft (1981). When an item is taken from a secondary source, the name of the author of the source appears directly below it in abbreviated form along with the date of the source in parentheses in the comparative word list in Appendix B. Consult the list of abbrevia- tions. Newman's proto-Chadic lexical reconstructions appear as the last item, after the lexical reconstructions posited in this study. Newman's reconstructions are designated by a single asterisk (*). Where appropriate, reconstructions posited in this study are found directly after the lexical citations. Because of the nature of the data upon which these reconstructions are based, they are l4 referred to as 'provisional reconstructions' and are designated in the appendices and in the examples given throughout the thesis by a double asterisk (**). These provisional reconstructions (PR's hereafter) for the west-A branch posited in this study include consonants. An attempt is also made at reconstructing vowels. Though it is recognized that tone and vowel length play a very important role in Chadic phonology, no attempt is made to reconstruct these in this study. Vowel length would be very difficult to reconstruct because of inconsistent marking of vowel length for citations in varying sources. This also holds in many cases for tone markings. Moreover, many Chadic words have alternate tone patterns depending on the context in which they appear. Therefore, due to the absence of consistent and reliable data on vowel length and tone markings, these aspects are not considered in the PR's. Tone and vowel length are not marked in the examples given throughout the discussion, but, as far as the sources will allow, are included in the citations in the compara- tive word list found in Appendix B. The PR's make use of the symbols given in the proto-Chadic west-A phonemic inventory of Chapter Three. Expectations Because of the proposed close relationship of the lan- guages, implied by their position at the subbranch level in 15 the classification, one might expect to find numerous cog- nates, and regular sound correspondences from which proto- lexical forms for the West-A branch may be posited. Because of the hypothesis espoused in this study, a low degree of consistency is expected from a comparison of the PR's for the West-A branch and the corresponding lexical reconstructions posited by Newman (1977). It is hypothe- sized that Newman's reconstructions do not adequately reflect the individual lower structures within his Chadic classification. Results First of all, contrary to expectation, the comparison of the five West-A branch languages has revealed very few cognates. Sound correspondences exist. However, few are very regular. The scarcity of cognates and the sporadic nature of the 'regular' sound correspondences, makes it difficult, if not impossible, to posit reliable and valid lexical reconstructions for the west-A branch. 0f the two hundred and ten items on the comparative word list, PR's were possible for only one hundred and thirty-three. Because of the scarcity or lack of cognates found among all five of the languages concerned, more than half of these PR's are based on as few as two of the five languages. Seventy-two of the one hundred and thirty-three PR's correspond to Newman's reconstructions. Contrary to 16 expectation, the comparison of the PR's with the corres- ponding proto-Chadic lexical forms posited by Newman reveals more similarities than dissimilarities. This finding suggests that Newman's proto-Chadic lexical recon- structions do adequately reflect the west-A branch of the present Chadic classification. However, given the dubious nature of the PR's due to the poor quality of the cognates and sound correspondences upon which they are based, this finding has conflicting implications as discussed in the conclusion. The structure of the remainder of this thesis is as follows. Chapter Two includes general and linguistic information about each language included in this study. Chapter Three includes a phonemic inventory for proto- Chadic west-A, regular sound correspondences for consonants and vowels revealed by the comparison of the languages, and a discussion of these proto-phonemes and sound corres- pondences. Chapter Four deals with the comparison of the west-A PR's with the corresponding Chadic reconstructions posited by Newman. The conclusion is presented in Chapter Five, followed by the appendices; Appendix A lists each set of cognates, its associated gloss and its respective PR. The cognates and accompanying PR's are presented in alpha- betical order according to gloss by descending order of the number of cognate items in each set; Appendix B contains the comparative word list which is outlined as explained above. NOTES 1Greenberg (1963) presented linguistic data in the form of lexical and morphological cognates as evidence of the genetic relationship of the Chadic languages with the so-called 'Hamito-Semitic' languages, which consisted of Berber, Semitic, Egyptian, and Cushitic languages. Greenberg named this new five-branch system of languages the 'Afroasiastic' family, thus abandoning the term 'Hamitic' which had develOped racial connotations. A more recent theory concerning the relationship of Chadic to other branches of Afroasiatic suggests that “the co-equal or coordinate branches of Greenberg's Afroasiatic are no longer co-equal' (Fleming 1983:23). Omotic (formerly grouped with Cushitic), is now generally con- sidered a sixth branch of the Afroasiatic language family. 2Lukas, Joannes, 1936. ”The Linguistic Situation in the Lake Chad Area in Central Africa." Africa vol 9. 332-349. 3Concerning Lukas's classification, in their 1966 publication, Newman and Ma state ”some languages were incorrectly classified because of the inordinate importance ascribed to certain typological features (notably grammatical gender), but on the whole, Lukas presented a reliable, conservative classification based on considera- tions of phonology and vocabulary as well as grammar” (p. 218). 4In Westerman, D. and M.A. Bryan, 1952. Langgages of west Africa. London: International African Institute. TLater edition 1970). 5Greenberg, Joseph H., 1950. ”Studies in African Linguistic Classification, IV. Hamito-Semitic." South west Journal ofAnthropology, vol. 6, 47-63. This article appears in Greenberg (I963) The Languages of Africa. 6Newman (1977) also offers an improved inventory of proto-Chadic phonemes which are used in the re-evaluated proto-lexical reconstructions. 17 18 7Newman (1977) states that he relies on Caprile and Jungraithmayr (1973), E. Wolff (1971), and Hoffmann (1971) for the nomenclature of the languages, and for the relationship of the lower level groups and clusters, while he himself is responsible for the 'higher groupings showing the general structure of the sub-classification within Chadic' (p. 4.). 8Newman's East and West branches represent a division of the former Plateau-Sahel group which was coordinate with the Bin-Mandara group in Newman and Ma's 1966 classification. 9More recent Chadic theory suggests that the Chadic language family is comprised of some one hundred and forty languages (Fleming 1983:19). loDera and Kanakuru are alternate names for the same language and for the group of peOple who speak the language. I have chosen to use the term Dera in this study since “Dera is the peeples' own name for themselves" (Newman 1974:ix). 11This study project involved the elicitation of words from a Karekare informant for phonemic analyses for partial fulfillment of a course on Afroasiatic linguistics with professors Neil Skinner and Patrick Bennett at the University of Wisconsin-Madison. 12Ninety of the two hundred and ten words included on this word list correspond to vocabulary items used on the Swadesh list. Nearly all of them (one hundred and ninety- six) correspond to the items used on the word list compiled for Chadic languages by Jungraithmayr (1975). 13Because of the need to compile a sizeable word list for comparative purposes, it was not possible to omit all of the semantically dubious items. CHAPTER II HAUSA, SHA, ANGAS, KAREKARE, AND DERA: GENERAL AND LINGUISTIC INFORMATION General Information All of the languages concerned in this study are spoken in northern Nigeria (see Figure 4). Hausa, also spoken in Niger, northern Cameroon, and in geoqraphical pockets throughout west Africa, has as many as twenty million speakers. Hausa is spoken in all of the northern states of Nigeria (i.e., Sokoto, Kano, Borno, Niger, Kaduna, Bauchi, Plateau) and has many dialects depending on the region where it is spoken; e.g., Sokoto, Zaria, Kano. The Kano dialect is considered the Standard. It is Standard Hausa that is referred to in this study. Though there are many languages spoken throughout northern Nigeria, Hausa is the major language of the area and functions as the lingua franca. Thus speakers of minor languages, e.g., Sha, Angas, Karekare, and Dera, typically speak Hausa as a second language. Hausa has linguistically influenced minor languages as evidenced by lexical borrow- ings from Hausa. Hausa is the most thoroughly studied and documented Chadic language, thus the quantity of linguistic l9 20 Gongola Cameroon Figure 4 LANGUAGE MAP (Taken From Ikwue 1984) CAPITAL LETTERS = LANGUAGES Lower Case Letters = States 21 literature on Hausa abounds in comparison to the paucity of available data for minor languages like Sha, Angas, Karekare, and Dera. The quality and reliability of Hausa data also exceed the quality and reliability of data for the other Chadic languages. Sha is spoken near the southern edge of the Jos Plateau in Plateau state. Other languages belonging to the Ron group, e.g., Daffo-Batura, Bokkos, Kulere, and Fyer, are spoken in surrounding areas. Sha is spoken by approxi- mately one thousand and five hundred peOple (Jungraithmayr 1966:169). Angas is Spoken in Pankshin and the surrounding areas including Amper, WOkkos, Ampan, Kabwir, and Garram (Shimizu 1974) in Plateau state. There are two main dialects, Plains Angas and Hill Angas. Lexical and phonological evidence indicate that the dialect used in this study is Plains Angas.1 Angas speakers number 138,224 according to the 1963 census (Nigeria Handbook 1973). Besides Hausa, languages neighboring Angas, include those belonging to the Chadic family as well as those belonging to the Benue-Congo branch of the Niger-Congo language family. Karekare is spoken in the western area of Borno state in Potiskum and Fika, and in Bauchi state, in Dagauda and Jalam, extending just east of Kano. Dialects of Karekare include Jalalum in the west, Pakaro in the north, and Ngwajum in the east (Hansford et a1. 1976). According to the 1963 census, there are 128,802 Karekare people 22 (Nigerian Handbook 1973) presumably first or second language speakers. Neighboring languages include Bolanci, Ngamo, and Ngizim, all of which are Chadic languages. Kanuri, a Nilo-Saharan language, is a major neighboring language spoken in the area north of where Karekare is spoken. Dera is spoken along the Hawal and Gongola rivers at Shellem and Kiri near Numan in Gongola state and in Shani and Gasi near Biu in Borno state. The two major dialects are spoken in Shani and Shellem. The dialect of this study is the Shellem dialect. There are approximately 11,000 speakers of Dera according to westerman and Bryan 1970.2 Linguistic Information A phonemic chart for each language is followed by pertinent phonoloqical descriptions. Hausa Phonemic Inventory Consonants Labial Dental Palatal Velar Glottal Stops t c k ' b d j 9 23 Labial Dental Palatal Velar Glottal Steamed I; J I; Glottalized Fricative ts Fricatives f 3 sh h z Nasals m n Liquids l r,r Semivowels w y Glottalized Semivowel 'y Vowels 1 u e o a l. /f/ is pronounced more like a voiceless bilabial fricative in some dialects. 2. /'/ has phonemic status in Hausa. It usually precedes what would otherwise be described as vowel initial words. 3. Palatalization occurs with dental and velar consonants when they are followed by front vowels. 24 4. Labialization occurs with labial and velar consonants when they are followed by back vowels. 5. Only one glottalized consonant occurs within a word unless it is a geminant glottalized consonant. 6. Most Hausa words do not end in consonants. 7. All vowels occur long and short; vowel length is contrastive. /o/ and /e/ are not common vowels and usually occur long in final position. 8. There are two basic tones, high and low. Sha Phonemic Inventory Consonants Labial Dental Palatal Velar Laryngeal Glottal Steps t c k <1 ' b d g I} Prenasalized at... 9 s 2 9 seamed r, 4‘ Fricatives f 3 sh h v z Nasals m n Liquids l r Semivowels w y 25 VOwels l. /C/ and /h/ may be allophones of the same phoneme according to Jungraithmayr (1966) . E occurs inter- vocalically and before some vowel initial words. /h/ occurs word initially, finally, and intervocalically in a few examples. As illustrated in Jungraithmayr (1966) /h/ often corresponds to /r/ and /A> in other Chadic languages. 2. /'/ occurs preceding vowel initial prefixes. 3. /b, d, j/ and /g/, represent a full range of prenasalized consonants; they occur in initial position. 4. Most of the Sha citations end in consonants, monosyllabic and disyllabic in structure. 5. vowel length is not marked except in a few examples in the source from which the data were taken. 6. There seem to be two basic tones, high and low, and a falling tone. In a few cases rising tones are marked. 26 Angas Phonemic Inventory (Taken from Kraft 1981) Consonants Labial Dental Palatal Velar Glottal Stops p t c k (') b d j g Glottalized Stops b cf 5 Fricatives f s s (h) v z z Nasals m n Liquids l r [£1 Semivowels y w VOwels i i u e o a l. ['1 occurs phonetically preceding vowel initial words according to Kraft (1981). In Kraft (1981), it is included parenthetically in the phonemic inventory. 27 2. /5/ represents a palatal implosive stop accord- ing to Kraft (1981). In Shimizu (1974) this phoneme is described as a palatalized velar implosive.3 3. [r] is not included in the phonemic inventory in Kraft (1981). According to this source it is a phonetic variant of /r/ occurring in initial and final position. However, a flap [r] is also found in these positions. 4. Voiced stops do not occur word finally. They are sometimes replaced by voiceless stops in initial position. 5. Most of the words in Angas end with consonants. 6. Burquest (1971) states that long vowels rarely occur in Angas; contrastive vowel length only occurs in a few examples, e.g., 'tooth' ggg and 'dog' 33. There are only a few citations which are marked with long vowels in Kraft (1981). However, in some cases where Kraft does not indicate a long vowel, Jungraithmayr (1963) does. 7. Burquest (1971) states that there are three tones in Angas, high, mid, and low. However, in Kraft (1981) (the main source consulted for Angas data) only high and low tones are indicated. If no tone mark is used to indicate mid tone, it is not explicitly stated in Kraft (1981). 8. The vowel /i/ represents a high front closed unrounded vowel (Burquest 1971). 28 Karekare Phonemic Inventory (Taken from Kraft 1981) Consonants Labial Dental Palatal Velar Glottal Stops t c k b d j 9 3332““ I: A‘ Fricatives f s s (h) Lateral Fricative 1 Nasals m n (ny) Liquids l r r Semivowels w y Glottalized Semivowels 'y 'w Vowels l u e (e) o a 1. /i/ is not included in the phonemic inventory in Kraft (1981). It occurs word initially and finally. /r/ 29 also occurs in these positions. 2. /1/ represents a voiceless lateral fricative. It occurs word initially and medially, however, it does not frequently occur in the language. 3. /'w/ and /'y/ represent glottalized velar and palatal semivowels, respectively. They are included in the phonemic chart in Kraft (1981). However, it is also suggested in Kraft that perhaps ['w] is an allophone of /'y/, occurring before a back vowel. 4. /ny/ and /h/ occur word initially. Both phonemes occur only in a few items in Kraft (1981). 5. According to.Schuh (1978) all vowels have contrasting length. However, as in Hausa, a phonemic contrast between long and short /e/ and /o/ is unusual.4 wael length is not indicated in the Karekare forms in Kraft (1981), but are indicated in Schuh (1984). 6. Schuh (1978) states that there are two contrastive tones which occur in Bole languages, high and low. These tones are marked in the Karekare items in Kraft (1981), as well as an occasional falling tone (a high tone followed by a low tone on the same syllable). 7. According to Schuh (1978) and (1984), /a/ is analyzable as a vowel reduction from /i/, /u/, or /a/. 30 Dera Phonemic Inventory (Taken from Newman 1974 and Kraft 1981) Consonants Labial Dental Palatal Velar Stops p t k b d j 9 Glottalized Stops f3 ci‘ Fricatives s (h) Nasals m n 3 Liquids l r Semivowels w y VOwels 1 u e e o a l. /h/ is a voiced velar fricative. It is a phonetic result of the weakening of intervocalic velar obstruents. It occurs between two non-identical vowels. 2. The only sibilant which occurs in present day Dera is /§/. This can be accounted for by a sound law posited by Newman (1970) in which all sibilants shifted to 31 /y/, followed by the shift 5 + g. Therefore, there is no voiceless palatal stop in Dera. 3. All vowels occur short and long except schwa, /a/. 4. There are two contrastive tones, high and low, and a falling tone (a high tone followed by a low tone) occurs within a single syllable. Chadic Phonological Traits General Traits Newman and Ma (1966) cite four phonological traits which are common among Chadic languages. They are as follows: 1. Palatalization: Palatals are most likely allophones of their corresponding alveolars, e.g., g < *3 and g < *E. 2. /w/ and /y/ interchange: For comparative purposes, these phonemes can often be interpreted as phonetic variants. 3. Neutralization of prosodic contrasts: The full range of prosodic features (voiceless, voiced, and glottalized) in syllable final position does not occur any- where in Chadic. 4. /47 + /r/ in syllable final position. 32 Specifically for Hausa and Dera, numerous linguistic properties and regular sound correspondences (which Newman refers to as sound laws) are noted in various works by Newman (1970, 1974, 1976, 1977) and Schuh (1976, 1978). These points are presented here because they aid in recog- nizing cognates and sound correspondences in the data presented in Chapter Three. Hausa Traits 1. *£ > y. This is a regular phonological shift postulated by Newman (1970). It occurs in all environments including intervocalically and in syllable final position. Secondary phonoloqical shifts include y_+ g/___g,g and y + i/___S (in syllable final position). 2- *.1_.>£ 3. *l,’ y. This is not a regular phonolOgical shift. 4a. *N > fl/___#. Nasal consonants, *m or *n, are lost in word-final position. Also, according to Schuh 1976 *N + 3/__g#, i.e., *m_or *n_may appear as a velar nasal [3] word finally. 4b. /n/ and /m/ assimilate to the point of articulation of a following consonant. Complete assimila- tion takes place before liquids. /m/ is subject to weakening to [w] before /r/ and /n/. 33 5. In word final or syllable final position, *K + [u] or [w] where *K represents the velar obstruents /k/, /g/, and /k/; *T + [r] or [1] where *T represents the alveolar obstruents /t/, /d/, and /d>; *P + [u] or [w] where *P represents the labial obstruents /p/, /b/, and /i§/.5 6. Palatalization of alveolar consonants occurs in the environment preceding front vowels as mentioned above; 7. *éTI g following a long vowel in intervocalic position. This shift does not occur in initial position or following a short vowel. 8. /f/ + [h] /__;V +back 9a. A regular sound correspondence occurs between initial /h/ followed by a short /a/ and initial n in closely related Chadic languages. This can be accounted for by Newman's proposal in his 1970 publication, that this /h/ is derived from nothing, i.e., /h/ < *5, "by phonemi- cization of a phonetic feature of word onset along the same lines described by Greenberg (1947) for the origin of /'/' (p. 173) .6 9b. According to Newman (1976), sometimes a glide or semivowel occurs phonetically as onset before words which are otherwise analyzed as vowel-initial; [w], or [y], is used instead of [h] or [']. The phonotactic constraints in Hausa disallowing vowel-initial words can be schemati- cally summarized: 34 + *flVlCV 'V1CV2, where -glottalized 2 C + hV1CV2, where V1 short /a/ V1 = /u/ + yVICVZ, where V = /i/ + wV1CV2, where 10. Newman (1977) notes that Hausa A/ corresponds to 4y, gy, 'g, k, 'g, 'y, L, or H, in other Chadic languages. 11. Hausa makes use of an epenthetic syllable final /r/. Angas Traits l. Newman and Ma (1966) point out that devoicing of obstruents takes place in initial and final positions in Angas. 2. Newman and Ma (1966) also claim that /n/ in Plateau languages often correspond to *5. Dera Traits l. Intervocalic weakening of obstruents to sonorants. *P.+ E, where *P represents labial obstruents, /p/, /b/, and /£/. A secondary shift occurs in some Dera dialects where the weakening process is reversed and /B/ is used intervocalically instead of /w/; *T + 5, where *T represents alveolar obstruents, /t/, /d/, and /d>; *K + Q, where *K represents velar obstruents, /k/ and /g/. Two secondary shifts take place concerning the weakening of 35 the velars. /h/ + fl/Vl___V1, i.e., /h/ is deleted in the environment between two identical vowels, and /h/ + [w] / (V+back)___(v+back), i.e., /h/ + [w] preceding, or follow- ing a back vowel.7 2. Final position hardening. /r/ + [t] in final position. According to Newman this is a diachronic shift. Intervocalic /r/ in Dera corresponds to /t/ in related languages whereas /t/ in final position corresponds to /r/. 3. 5 + c! synchronically. 4. Loss of sibilants. *S + y_where *8 represents the sibilants /s,z,s/ and /i/. This shift operates in all environments. A secondary shift, 1 + g /___V+back, some- times occurs. 5. Reappearance of /§/. *9 > 5. Dera é corres- ponds to g/k/E_in other related Chadic languages. 6. wael harmony occurs in trisyllabic verbs. If the first vowel of a verb is +hi, then the penultimate vowel is +hi and agrees in frontness with the first vowel. If the first vowel is not /i/, or /u/, the penultimate vowel is /a/. There is one exception iibere 'buy' (Frajzyngier 1976). 7. /a/ is the only vowel which occurs in word initial position. NOTES 1It is not explicitly stated in Kraft (1981) which dialect of Angas is used in the data presented. However, the two dialects are distinguishable phonologically and lexically. In many words Plains Angas has the vowel aa (long a) where Hills Angas has 00 (long 0), e. 9., 'stand up' in Plains Angas is /yaal/, _hile in Hills Angas it is /yool/ 2Given that the number of speakers cited in this source for Angas and Karekare are off by a count of about one hundred thousand compared to the more recent figures cited for these languages in the 1963 Nigerian census, it is suspected that this figure may grossly underrepresent the current number of Dera speakers. 3Burquest (1971) treats /6/ as an allophone of /'/ occurring only with the feature of palatalization syllable initially, with /'/ occurring elsewhere syllable initially. 4Schuh (1978) states that short /e/ and /0/ may be variants of short /a/ conditioned by consonantal environ- ment. 5This set of phonological shifts is referred to as the Klingenheben laws after August Klingenheben for his article "Die Silbenauslautgesetze Des Hausa" Zeitschrift fur Eingeborenen-Sprachen, Band XVIII, no. 4, . 6Because an initial /a/ commonly occurs in body part terms in Chadic languages, there is a theory which suggests that this initial /a/ may be a prefix. However, Newman asserts that /a/ was not a prefix in proto-Chadic; rather, it was an inseparable part of the individual lexical items. 7The intervocalic weakening of obstruents to sonorants does not take place when the obstruent precedes /e/, when the obstruent follows a short vowel, and when the obstruent is the consonant in a syllable of C3 of trisyllabic words. 36 CHAPTER III PROTO-FORMS FOR CHADIC WEST-A In this chapter a phonemic inventory is presented for proto-Chadic West-A consonants and vowels based on the comparative data. 'Regular' sound correspondences are illustrated for consonants and vowels for each set of cognate items to exemplify each reconstructed phoneme postulated in this study. A discussion of the proto- phonemes and 'regular' sound correspondences are included also. The cognates and the reconstructions for Chadic West-A are listed in Appendix A. Phonemic Inventory of Provisional Reconstructions Consonants Labial Dental Palatal Velar Stops t c k b d g Glottalized Stops g d‘ K,(CX) Fricatives f s 5 37 38 Labial Dental Palatal Velar Nasals m n Liquids l r Semivowels w y VOwels Initial waels: i, Y, a, V Medial vowels: i, e, Y, V +hi, u, o, W, a, V Final VOWBIS: i, e, Y, 11, O, W, V, (-V) Discussion of Phonemic Inventory of Provisional Reconstructions Consonants A voiceless bilabial stop is not included in the phonemic inventory. There is no evidence in the data which warrants that a **/p/ be reconstructed. /p/'s occur initially in Angas and Dera. However, they correspond mostly to initial underlying /f/'s in Hausa, Sha, and Karekare. Since the phoneme /f/ occurs in the majority of the five languages this is chosen as the symbol in the reconstructions rather than /p/. There is no evidence for reconstructing a voiced palatal stop, a voiced labial fricative, or a voiced palatal fricative. 39 **/K/ is used as a symbol which represents some glottalized velar phoneme with its varying reflexes occurring in each language under unidentified conditions. In the few examples given in the list of sound corres- pondences, it occurs both initially and medially. Its reflexes include /k/, /g/, /k/, /'/, /J}/. /kw/, /w/, and /6/, for the various languages. The symbol **Ex is not a phoneme. Rather, it is a symbol used to represent some sort of phonetic onset of word-initial vowels. This parallels Newman's scheme for the phonetic onset of Hausa word-initial vowels. **Qx corresponds to Newman's *E. The various reflexes in the languages include In]! ['1 I [W] I [Y] I [kW] I [gwlI and H- Vowels The symbols /Y/ and /W/ represent a front vowel and a back vowel, respectively. The symbols /V/ and /(-V)/ both represent unspecified vowels; the latter symbol applies only to final vowels of verbs. The initial vowels /i/ and /a/ occur with verbs and nouns, and /Y/ and /V/ occur only in a few nouns in the data examined. The vowels which occur in medial position are noted above. /e/, /o/, and /W/ do not occur medially with the verbs, and each phoneme occurs in only a few of the items in the data. The other medial vowels occur with all of the grammatical categories represented in the word list. The only final vowel which 40 applies to verbs is /(-V)/, where the hyphen indicates that the vowel is a suffix with phonological or morphological implications (depending on the language), and the parenthe- ses indicate the uncertainty of the existence of such a final vowel in proto-Chadic West-A verbs. 'Regular' Sound Correspondences In the following list of 'regular' sound corres- pondences, the position of corresponding phonemes within a word is noted by the placement of hyphens with each phoneme. A hyphen following a phonemic symbol indicates that the sound occurs word initially; a hyphen preceding a phonemic symbol indicates that the sound occurs in word final position; a hyphen on either side of the phonemic symbol indicates that it occurs medially. In some examples medial phonemes correspond with final phonemes. This happens where the form for a particular language for a particular item ends with a vowel and the corresponding citation(s) in another language(s) is consonant final. A plus sign (i.e. +) indicates a morphological boundary. If it is placed preceding a phoneme, this indi- cates that a morphological unit precedes it, be it the root or a prefix; if it is placed after a phoneme, this indi- cates that a morphological unit follows it, be it the root or a suffix. 3. 9. 16. 21. 32. 43. 64. 71. 77. 92. 97. 106. 116. 121. 128. 132. 156. 195. 87. 187. **b ashes beard blind man bow charcoal divide five give groundnuts hunt knee liver mouth night open place shoot two “I, horn tie 41 Consonants H S b- b- -w- _b- _p b- b... b- b- b- b- b- -u- fw- ‘P p- p- 3. 23. 26. 41. 51. 62. 70. 72. 74. 103. 114. 122. 146. 155. 161. 162. 167. 175. 184. 190. 210. 13. 17. **t ashes break into pieces burn die eat fire gather gO go out lie down moon nose rope sheep sit skin speak sun thigh tooth woman *td beans bite blood 42 +-c- +-t- +-c- -r- -r- -r- 36. 54. 66. 79. 85. 88. 95. 134. 137. 153. 159. 162. 190. 197. 11. 18. 24. 25. 37. 45. S7. 64. 68. 103. **d (cont'd) come eye fly guinea fowl hit horse kill pull quiver 88W sickle skin tooth urine a! bird blow (Woman's) breast build cook dog fall five four lie down 43 -j- -r- -r- -h- ’1' -r -r -r -r- 111. 121. 128. 142. 153. 177. 187. 98. 113. 152. 171. 19. 21. 25. 32. 37. 38. 44 “'4‘ (cont'd) H S millet night open -4: refuse ‘H- sew swallow +-d: d- tie A; V **c knife sh- monkey seize steal **k arrow baobab tree k— body +-k +-k bow -k- -c- build charcoal cook -g- cooking pot -r C- c- sh- sh- 50. 63. 67. 75. 79. 80. 81. 84. 85. 88. 91. 95. 98. 112. 116. 118. 140. 142. 151. 155. 172. 185. 194. 202. 204. **k (cont'd) ear fish forge goat guinea-fowl hair head hen hit horse hunger kill knife moisten mouth navel rat refuse seed sheep stomach three turtle wash weave 45 +-j- +-k- c- -c- +-k- sh- 32. 60. 130. 139. 161. 182. 183. 8. 20. 56. 190. 13. 44. 45. 52. 65. 75. 94. 122. 128. 172. **g beard charcoal fill person ram sit tear (v.) ten **K beans bone faeces tooth **C bite do dog egg flour goat intestines nose open stomach 46 -g- 1(- +42- +-g- g— 9— n9- .9- 39' -g- -9- g- y- 177. 189. 190. 203. 18. 34. 57. 63. 68. 72. 74. 120. 162. 175. 184. 207. 46. **Cx (cont'd) swallow tongue tooth water **f arrow blow close fall fish four go go out new skin sun thigh white **8 arm arrow bone charcoal dream 47 -3h- fy- -811.- p- p- mp- ”/'- -s- ‘3 -8- gw- p- —w- p- -w- p- 52. 56. 62. 65. 80. 99. 101. 106. 110. 117. 138. 140. 145. 156. 189. 204. 19. 84. 91. 94. 108. 112. 136. **s (Cont'd) 399 faeces fire flour hair know leg liver medicine name rabbit rat roast shoot tongue weave **2 body hen hunger intestines man moisten put down 48 -811- -8h- +-j- -j- SY" +-y- "C 2- z- -z- z- -y— -sh- y- -y- y- w- 146. 161. 47. 98. 126. 8. 16. 17. 32. 41. 50. 60. 67. 79. 91. 97. 108. 109. 111. 117. 118. 124. **z (Cont'd) rope sit *ag drink knife old **m beans blind man blood charcoal die ear fill forge guinea-fowl hunger knee man meat millet name navel oil 49 sh- -n -m- -n— m- sh- -m -111 '1'! m- -m -m -111 -m -m -m +-m- -m -m -m mw- 2- -sh- -m- -1“ -m m- m- '3 w- -j- -m -m -m -m -m- -5- -m -m -m- 126. 138. 140. 151. 162. 167. 183. 203. 205. 210. 21. 31. 32. 37. 46. 66. 99. 109. 110. 122. 126. 134. 137. 50 **m (Cont'd) H old rabbit -m- rat -H seed skin speak ten -m- water what m- woman m- **n call charcoal cook dream fly (n.) know -n- meat n- medicine nose -fl old pull quiver -m- -n -n -m m- +-m- -n .11- “1'1 '1'! -n -m -m -m m- '3 '11 ‘11 '3 -m- '3 -m- -m -n- -n- -n- ‘11 '11 -n- -m -m— -m -n— -n- -n -n -n '3 -n- -r- 146. 148. 150. 154. 155. 161. 165. 184. 185. 197. 202. 12. 26. 38. 67. 87. 109. 113. 119. 137. 148. 159. 170. **n (Cont'd) rope saliva see shadow sheep sit sleep thigh three urine wash **1 give birth burn cooking pot forge horn meat monkey neck quiver saliva sickle stand up 51 -n- -n- -n- -y- -r- -y- '11 -n ny- -n '3 n- -n '3 -n '3 “n n- -n- -n- -3- -n- -n— '3 -3- “n -n- -n- ‘11 -r- 177. 189. 195. 207. 21. 32. 43. 63. 70. 71. 77. 92. 97. 106. 110. 114. 124. 138. 146. 152. 154. 163. 171. **l (Cont'd) swallow tongue two white **r arm bow charcoal divide fish gather give groundnuts hunt knee liver medicine moon oil rabbit rope seize shadow sky steal 52 H -y- +-r- -y- -r- -y— H/-i -r- fl/-i '3 +-r- -r- -r -r- -r- -r- -r- -r- -r- -r- -r- -r- -r- -w- r- -r- -n- -r- -r- -r- -r- -r— hr- -r- 190. 194. 62. 119. 123. 145. 190. 202. 208. 11. 24. 66. 112. 170. 182. 197. **r (Cont'd) tooth turtle **w fire neck obtain roast tooth wash who **y beard bird (woman's) breast fly moisten stand up tear (v.) urine w- y- 53 -r y- Y- w- w- -j- -Y- -y- Y- -w-/_g- y- y- 44. 54. 52. 13. 94. 120. 122. 172. 189. 190. 203. 45. 65. 75. **i do eye **Y egg **a bite intestines Open nose stomach tongue tooth water **V dog flour goat 54 vowels yi- ha-+ ha-+ ha-+ ha-+ Initial vowels Ya' 'a-+ 'a-+ 'a-+ 'a-+ ha- yi- e- a- Ia- yi- ye- gwa- Ye- 63. 66. 122. 77. 139. 146. 11. 20. 87. 108. 112. 195. 18. 91. 110. 117. 154. **i fish fly (n.) 11088 **e groundnuts ram rcpe **Y bird bone horn man moisten two **V +hi blow hunger medicine name shadow 55 Medial Vowels -i- -i -a- —i- -i- -i- -u- -u- -o- -e- -i- —u- -i- -e- -e- -e- -i- -i- -i- -i- -a- -i- -e- -e- -3- -e- -8- -e- -u- -u- -1,- -u- —u- -i- -i- -i- -e- -a- -i- -e- -i- -e- -i- -i- -i- -i- 162. 171. 7. 16. 32. 41. 46. 50. 79. 85. 95. 98. 118. 119. 140. 142. 145. 185. 190. 194. **V +hi (Cont'd) skin steal **u baobab tree blind man charcoal die dream ear guinea-fowl hit kill knife navel neck rat refuse roast three tooth turtle 56 H S -u- -u- -u— -u- -u- -u -u- -u- -u— -0- -u- -u- -u- -u- -u- -a- -u- -u- -u- -u- -u- ‘11 -e— -u- -u- -u- -u- -u— -u- -a- -9- -u- -u- -u- -i- -i- -u— -u- ’11 -u- -uwa- -u- -u- -u- -u- -u- -u- -u- -u- -u- 88. 110. 137. 138. 146. 24. 159. 163. 183. l. 2. 3. 9. 21. 23. 25. 31. 43. 60. 70. 80. 57 **o H S blood -i- -o- horse -o- medicine quiver rabbit -o -o- rope **w breast sickle -au- sky ten -o- **a arm arrow ashes beard how -a- -a- break into pieces build call divide -a- fill gather -a- -a- hair -0- -o- -u- —o- -a— -a- -a- -a- -i- -a- -0- -0- -0— -e- -0- -8- —w- -a- -a- -3- -a- -a- -a- -0- -0- -0- -0 -0- -0- -u- -u- -e- -a- -8- -a- -a- 92. 109. 114. 124. 126. 152. 153. 161. 175. 177. 182. 202. 204. 207. 21. 26. 32. 37. 38. 56. 58 **a (Cont'd) H hunt meat -a- moon oil -a- old seize sew sit -au4 sun swallow -i- tear (v.) -a- wash -a- weave -a- white -a- **V beans beard bow burn -o- charcoal -a- cook cooking pot faeces -ae -a- -a- -a- -u- -a- -e- -a— -a- -i- -a- -a- -a- -i- -a- -e- -0— -e- -a- -a- -a- -a- -a- -a- -a- -a- -a- -a- -i- ‘a- -a- -i- -u- -i- -a- -9- -0- -a- -a- -a- -0- -8- -0- -0- -0- ‘11- -3- -i- -i- -i- -u- 57. 62. 63. 64. 66. 67. 68. 74. 77. 79. 84. 87. 97. 99. 103. 106. 111. 113. 121. 128. 134. 137. 138. 151. 155. **V (Cont'd) fall fire fish five fly (n.) forge four go out groundnuts guinea-fowl hen horn knee know lie down liver millet monkey night open pull quiver rabbit seed sheep 59 H S —u- -e- -i- -a— -u- -9- ‘u- -u— -e- -i- -u- -u- -e- -0- -a- -e- -e- -a- -e- -a- -a- -a- '0 -e- -a- -8- -a- -e- “a. -u- -a- -a- -e- -8- -a- -a- -u- -a- -i- -a- -o- -e- -u- -i -9- -u— -0- -a- -i- -a- -i- 156. 161. 162. 163. 165. 167. 170. 184. 187. 189. 190. 197. 19. 24. 75. 94. 108. 121. 132. 146. 60 **V (Cont'd) H shoot sit skin -a- sky sleep (n.) -a- -u— -0- Final Vowels speak stand up thigh tie -au- tongue -B- -e tooth -o- urine **i body -i-+ (woman's) breast goat -iya intestines -i man -i night place rope -8+ -u- —o- -i- -3 -a- -a— -i- -i—+ -i -a- -i- -a- -u- -9- -a- -u -i -i -i -i -i -i -0- "O -0- -i- -u- -0- -i- -i- -e- -i- -i -i -i -i -oi 155. 175. 38. 114. 120. 159. 163. 205. 68. 109. 195. 50. 54. 77. 79. 111. 116. 61 **i (Cont'd) H S A sheep -iya sun -fl **e cooking pot -e moon **Y new sickle -e sky -e what -e -a -e **u four -u -E -H meat -11W two -u **0 beard ear -e -9 -fl eye -o -ay groundnut guinea-fowl millet mouth -a-+ -o -i -i -e -e -i -i -i -iya ‘11 '0 ‘11 '0 “"0 ‘0 "O '0 "'0 -a -i -e -e -i '11 -o -o -o -o -o -o -u 172. 184. 190. 7. 45. 56. 62. 63. 80. 81. 84. 106. 119. 126. 130. 163. 190. 207. 208. **o (Cont'd) stomach thigh **W tooth **V baobab tree dOg faeces fire fish hair head hen liver neck old person sky tooth white who ** (-V) all verbs 62 -i -a -i -a -i -ai ‘3 '3 -e -i -aw -i -a '3 ‘0 ‘6 '0 -e 'U -i -a ‘8 ‘U -u -a -a -w -i -a -e-+ ‘0 '0 -e -u -i -o -ai -oi -i ‘0 '0 63 Discussion of 'Regular' Sound Correspondences Depending on the criteria used to determine the 'regu- larity' of sound correspondences, the sound correspondences illustrated for each reconstructed phoneme may or may not be considered regular. The minimal requirement ascribed for the term 'regular' determines what is considered as a regular sound correspondence; in this study a regular sound correspondence is one which occurs at least twice. The sound correspondences illustrated comply to this criterion with varying degrees. Consonants 1. **g Correspondences upon which the phoneme *fg is recon- structed comply minimally; there exists only one cognate set which demonstrates a sound correspondence between initial /£/'s in Karekare and Dera, (#87 'horn') and one medial sound correspondence, gg~bg~£_in Hausa, Angas, and Dera, respectively (#187 'tie'). Because of this, the reconstruction of the phoneme **g for either set may appear dubious, especially given the fact that the first cognate set includes only two of the five languages investigated, and the other includes three. However, the second consonant in the first cognate set (#87 'horn') under the given definition is a regular sound correspondence between 64 the two languages involved, and the corresponding /g/'s are phonemically identical. Therefore, /g/ is recon- structed as a proto-phoneme. This cognate set may represent a loan word, one language borrowing from the other. On the other hand, perhaps initial /£/'s do regularly correspond across the two languages, but by chance, evidence of this is not incorporated in the data. There are no other examples in the data of medial /g/ correspondences. 2. **K Correspondences for the reconstructed symbol **§ are few and inconsistent; the only correspondence which does occur more than once is /£/«~/g/‘~/6/ in Hausa, Sha, and Angas, respectively, occurring in both initial and medial position. There is not enough data in this study to determine whether or not there exist specific environments for the various reflexes in the languages. 3. **Ex As mentioned above, Ex is a symbol which represents phonetic onset of otherwise word-initial vowels. It manifests itself as various sounds in the languages. The reflexes in Hausa include [y], [h], and ['1: the alternate reflexes may be accounted for by the scheme for phonetic onset preceding word-initial vowels explained in Chapter Two, where **Qx corresponds to Newman's *8 for Hausa. In the other languages in this study (e.g. Angas) fl may also be a reflex of this phonetic symbol. 65. [h] occurs preceding word—initial short /a/ resulting in the phonetic realization [ha] in Hausa. Short /a/ or [ha] happens to occur initially in some body part terms in Hausa.l This alleged prefix in Hausa, according to the definition set forth, regularly corresponds with initial ['a] in Sha. However, since ['a] occurs as a prefix designating nouns in general in Sha, it is not clear what morphological function initial [ha] in Hausa corresponds to in Sha. Does it correspond to a phonologically similar morpheme with the same function as a body part prefix as in Hausa, or is it corresponding to a phonologically similar morpheme with a different function- that of designating nouns in general? Perhaps there are two functionally distinct prefix morphemes that are phonetically similar in Sha, ['all— indicating body part terms, and ['alz- indicating nouns in general. Hausa [h] in body part terms corresponds to [kw] and ['/H] in Karekare. There is only one example of a corres- pondence between Hausa and Dera for body part terms in which initial Hausa [h] corresponds to initial [w] occurring before /a/. There are no examples of Angas cognates which occur for body part terms included in the data. Elsewhere (other than with body part terms) concerning **Qx, the only correspondences which can be said to occur 'regularly' are ['1 and [y] between Karekare and Dera respectively. Other reflex correspondences occur 66 sporadically, or inconsistently. However, as in the case for **/Q/, **Ex is reconstructed for these items because other evidence identifies them as cognates, though there is not enough evidence in the data to say that the **9x reflexes represent regular sound correspondences. It is this sporadic and inconsistent nature which characterizes the correspondences of other cognate sets upon which the reconstructed phonemes in this study are based. For example, while Hausa initial /b/ regularly corresponds with /b/ in Sha, Karekare, and Dera, and with /p/ in Angas, the data does not indicate that Sha initial /b/ 'regularly' corresponds to the said phonemes for Angas, Karekare, and Dera. Some of the inconsistencies in the sound corres- pondences can be accounted for by phonological explanations outlined for the various languages in Chapter Two. For instance, in #106 'liver', Karekare.medial /b/ corresponds with Dera /w/ instead of Dera /b/. This can be accounted for by the phonological process explained above, whereby intervocalic bilabial obstruents are weakened to [w] in Dera; i.e. *P + ! intervocalically. Thus [w] is presumably the surface representation of the underlying phoneme /b/. Other phonological processes not mentioned in Chapter Two can account for some other irregular correspondences, e.g. in Angas the initial [f] in #97 'knee', may occur instead of the expected [p] due to the environment in which it occurs, namely preceding /u/, as might be the case for 67 Karekare [f] in #132 'open'. The [f] in Angas for #3 'ashes', may be due to the following /a/. Based on these data [v] cannot properly be accounted for; perhaps it was an [f] resulting from the environment in which underlying [p] occurred, and it somehow became voiced. Other phonological processes such as epenthesis may be operating. /b/ in the cognate items for Dera in #118 'navel' and #162 'skin' is not considered reconstructable. It is treated as an epenthetic consonant in the following examples: A K D ** 118. 'navel' kum kumbi kum 162. 'skin' 81m dimbi dV+hi m 4. **t Hausa and Sha [c] correspondences are due to the palatalization of /t/ in the environment preceding a front vowel. The corresponding Dera [r] which occurs may be accounted for by the sound shift noted by Newman, discussed in Chapter Two, whereby *T + r, intervocalically. The medial [d] in Karekare in #210 'woman', and for Dera in #184 'thigh', cannot be accounted for based on the available data. They are however phonetically similar with the phonemes in their cognate sets. 5. **g. The correspondence for **/d/ in medial position occurs as [r] in #54 ‘eye' and #8 'beans' in accordance with the 68 same sound shift *5 + 5, intervocalically. The [r] in #8 'beans' for Angas cannot be accounted for from the data. It is a phonetically similar corres- pondence though it does not occur regularly. There is the possibility that this is not a cognate item. Hausa [j] in #17 'blood' and #159 'sickle' may be due to palatalization in the environment of a front vowel. Sha [h] is noted by Jungraithmayr (1966) as corres- ponding to [r] and other alveolars in related languages. The justification for this is not apparent; see #54 'eye' and #197 'tooth'. While [2] is phonetically similar to the other sound correspondences for Sha in #17 'blood', its environment is not conducive to the palatalization of **/d/. 6. **€ Hausa intervocalic [r] can be accounted for by the sound shift noted by Newman, *éf+ r following a long vowel in intervocalic position. The final Hausa [r] in #64 'five', exemplifies the sound shift *T + E in syllable or word-final position. The reflexes of **éffor Sha include [d], [d5, [d], and [h]. Angas reflexes include [43 initially, and [r] and [t] in final position. Karekare reflexes are almost consistently [43 except in #18 'blow' for which the reflex is [t], and #45 'dog' for which the reflex is [d]. 69 Dera [r] can be accounted for by the shift *T + g, intervocalically. Dera final It] may be accounted for by the final position hardening shift, 5 + t, in final posi- tion. 7. **6 The Dera reflex of **/é/, [sh], is accounted for by the sound shift noted by Newman, 9 + g. 8. **E Palatalization of **k can account for the occurrence of [c] in Sha and Karekare in items #7 'baobab tree', #21 'bow', #75 'goat' and #158 'sheep' whereas the underlying g in Angas and Karekare for items #38 'cooking pot' and #63 'fish', respectively, is phonemic. Dera [sh] in item #63 'fish' can be accounted for by the sound shift mentioned above, 9 + g. Besides [k], Dera reflexes include [h] and 5, both of which result from the sound shift *K + h, intervocalically; the latter reflex is due to a secondary sound shift which takes place, namely, [h] + n / V1___‘V1 (i.e., [h] is deleted when it occurs between two identical vowels). [9] occurs twice as a reflex for Dera [w] also occurs as a reflex in the environ- ment of back vowels. Hausa ejective [k] occurs three times in the data and corresponds to [k] most of the time in the other languages. 9. **£ As noted in Chapter Two, Hausa /f/ + [h] /__;V +back, i.e., /f/ + [h] when followed by a back vowel. 70 Reflexes for Angas vary; they include [p], [f], [mp]. Reflexes for Dera include [p] and [w], the latter being the realization of the sound shift in Dera *P + w, intervocalically. 10. **g Dera reflexes include [y], [j], [w], and [sh]. [y] and possibly [j] can be accounted for by the sound shift *S + y where *8 includes all sibilants. It is real- ized as n in #2 'arrow', perhaps where [y] assimilated with the front vowels in its environment. The reflex is real- ized as [w] in #110 'medicine' where a secondary shift takes place; y + w / __;V +back, i.e. [y] is realized as [w] when followed by a back vowel. 11. **z The same conditions hold true for this sibilant as for the above sibilant **g for Dera; *S + y accounting for the occurrence of [y] and [j] in Dera. /y/ is realized as [w] in accordance with the secondary sound shift. Hausa reflex /j/ is due to the palatalization of /z/ in the environment of a front vowel. In Angas, devoicing takes place both word initially and finally. 12. **m The sound correspondences which exemplify the proto- phoneme **m are numerous across the languages and they are quite regular. It is sometimes realized as [n] in Hausa, Karekare, and Sha. It also occurs word finally in these 71 languages, and occurs medially in Hausa. This may be due to the weakening of /m/ in these positions. /m/ has been deleted in a few examples in Hausa, Karekare, and Dera in final position. [3] also occurs in final position in Karekare, and occurs once in medial position in Dera. 13. **g The sound correspondences for this proto-phoneme are more sporadic and inconsistent than the correspondences which occur for **m. Hausa initial reflexes include /n/, final reflexes, fl, and medial reflexes [n] and [3] where the latter represents nasal assimilation with a following velar consonant. *N > p /___#, i.e. nasals, /m/ and /n/, are deleted word finally. This phenomenon occurs in a few examples in the other languages also. Angas reflexes include both In] and [3] word finally. Karekare reflexes include 0, [n] and [3] in various positions. Nasals followed by other consonants in the recon- structed form in medial position are interpreted as two consecutive consonants as opposed to a single unit. Some nasals in the citation forms are considered here to be epenthetic nasals as opposed to being reconstructable nasals. Cognate items for which the phoneme **/n/ is reconstructed where /n/ is followed by another consonant include the following: 32. 37. 126. 155. 161. 184. 197. 202. include a nasal appear below. 72 H S A K charcoal kugkusu cook dyen dunku old manshi sheep tunkiya tanci sit tog tiggu thigh fantau urine 'yandeg wash wanke vwaB kuggum ddnge manjo tiga pundo yendin ** kugkusum JVnk ( -V) magsv tvgki thg(-V) anto yVnan wank(-V) Items for which the reconstructed form does not The nasals which appear in the citation forms are considered to be epenthetic phonemes. Karekare items. 21. 52. 122. 153. 210. *1 + H S A K bow rigka egg es 'igsa nose hagci 'aton wantan sew dandh woman mace mat mendo l4. **1 raha wariB dade As shown below, epenthetic n tends to occur in ** rVka Ys atin dad(-V) The [y] reflexes in Hausa result from the shift y, which according to Newman is not considered a regular phonological shift; see #26 'burn', #119 'neck', 73 #148 'saliva', #177 'swallow' and #195 'two'. Also in Hausa, the sound shift #1 + r is exemplified; #67 'forge', #189 'tongue', and #207 'white'. 15. **r The correspondence -i§/fl in #63 'fish' and #124 'oil' in Hausa is accounted for by the regular phonological sound shift postulated by Newman *5 + y and the secondary sound shift, 1 + i /___ S, i.e. y + i in syllable final position. In #71 'give', the correspondence 6 may be due to these sound shifts and vowel coalescence, where the [i] is absorbed by the preceding vowel. An epenthetic [r] occurs in the cognate item for Karekare in #111 'millet' as shown below: K D ** 111. millet magdo modo vaB Dera final [t] is another example of the hardening rule, 5 + E in final position. 16. **y The reflex [w] for Angas and Dera is due to the environment in which it occurs, i.e. before a back vowel. 17. **w An epenthetic [w] occurs in the cognate item for Dera in #163 'sky': 74 H S A K D ** 163. sky 'are rawi rV Other phenomena occurring in the cognate sets include vowel deletion, use of masculine markers, and metathesis: Forms with a deleted vowel include the following: H S A K D ** 8. beans girm. 'idem worom KVde 189. tongue hagghe 'ales lis lusan yilik le The use of k as a gender marker for masculine nouns in Chadic languages may explain the Angas form #2 'arrow': H S A K D ** 2. arrow pas fasku peek fask The process of metathesis which commonly occurs in Chadic languages is demonstrated in the following examples: H S A K D ** 32. charcoal gawayi gegp gVEQ 162. skin fata fukut thk 187. tie éaure bat Joge 4Ng(-V) 75 VOwels The vowel correspondences occur even more sporadically and inconsistently than do the consonants. It is impossi- ble to determine the quality of the reconstructed vowel for most of the vowel correspondences. Few of the recon- structed vowels occur word initially: Initial Vowels 1. it; Only two cognate sets occur upon which this **i is reconstructed, the verb #44 'do' and the noun #54 'eye'. The initial vowel given for Karekare in both words is 'if. Dera and Hausa show what is interpreted here, a glide [y] as phonetic onset for #44 'do', and no phonetic onset is indicated in the citation form for #54 'eye' for these languages. These phonetic sounds preceding the vowels are interpreted here as legacies of the recon- structed phonetic manifestation, **gx. 2. **2 Only one cognate set occurs initially supporting this reconstruction, namely, #52. 'egg'. 3. **2 This reconstructed phoneme has the most cognate sets for initial vowels. However, the correspondences are still sporadic in nature. There being few consistent regular 76 correspondences for these vowel initial reconstructions for the verbs, the process for arriving at these recon- structions (and others) is somewhat arbitrary guesswork. 4. **V The reconstructed initial unspecified vowel is a catch-all phoneme. There exists some vowel but because of its different reflexes in the various languages and the lack of regularity between the correspondences (due in part to the scarcity of cognates found in the data) it is impossible to say what the quality of the proto-form may have been. All of the initial vowels are realized with the recon- structed non-phonemic **Ex preceding them, be the reflex ['l, [h], [y], [w], [gw], or H. The examples which occur with reconstructed **V initially occur with nouns in the data. Medial VOwels Guesswork is also involved in reconstructing the medial vowels. Most of the cognate sets show no consistent regular correspondences for the vowels. Therefore, the reconstructed vowels are based on the majority of the vowel qualities which occur in the correspondences. The sound correspondences which occur for the reconstructed medial vowels, **i, g, g, X +hi, o, y, are few. 77 5. Z +hi I Perhaps this vowel is actually **i, but is realized as u in various examples due to the environment of a labial or velar consonant in which it occurs. 6. **u and **3 These are the two most frequently reconstructed medial vowels for which an actual vowel quality can be identified from the data. 7. **V As with the reconstructed initial vowels, most of the reconstructed vowels in medial position have been recon- structed as **V. Most of the medial vowels have varying non-recurring correspondences across the languages. Final VOwels Verbs The final vowels have been analyzed in two groups, verbs and non-verbs, i.e. the nouns, and the few adjectives which are included in the data. For both cate- gories it is difficult to identify vowel qualities for reconstructions or to even ascertain that a final vowel is reconstructable. For the non-verbs, various vowel quali- ties have been reconstructed as final vowels, while only **(-V) has been reconstructed for the final vowel of verbs. 78 Most of the verbs for Sha and Angas are monosyllabic having the structures CV, CVC, or VC. The most prevalent structure in the corpus of the data for both languages is the CVC pattern. The corresponding cognate forms in Hausa, Karekare, and Dera, have the pattern CVCV, with the final vowel signifying the class in which each verb belongs. Depending upon the meaning being conveyed and the context in which it is used, the final vowel of the verb in Hausa may have a range of differing qualities, e.g. [a, i, e, o, u, or 8.] Also in the citation forms given, there does not seem to be any consistent correspondence between the first vowel (the root vowel) of the verb and the final vowel. Among the cognate items the correspondences between the vowels in the root and the ending vowel for Hausa include the following: £995 ending -u- -u, -e, -a -i- -a -a- -a, -i -au- -a Also found in the corpus of the data, in non-cognate sets, are the following correspondences: root ending -i-, -a-, -au- -e In Karekare the majority of the final vowels of CVCV patterned verbs have the qualities -3 and -a, which do not 79 seem to convey any specific meaning. However, Schuh (1978) states that their presence or absence is dependent upon the context in which they are used. There does not seem to be consistent vowel correspondences between the root vowel and the final vowel in Karekare either. Most of the cognate verbs with final -g_have g in the root. However, there are examples where the root vowel is -g: with -3 final vowel. Also examples occur where the final vowel is -g and the root vowel is either -23, or -gr. There are a few examples where root vowels -33 and ‘1? correspond with final vowel -g, summarized: root ending -e- p -a- p '11- p ‘i- ‘11 Moreover, in the corpus of the data the following corres- pondences can be found: {225 ending -a- -a, -u -o- -u, -o -e- -u Most of the verbs with the CVCV canonical structure in Dera, have -i and -g as final vowels. Among the cognate items most of the examples have -g: and '2? as root vowels with -i final vowel. Also occurring with -i final vowel are root vowels -g-, -i§, -§-. Most of the root vowels which correspond with the final vowel -g, are -gf, -i—, and -g—. -gf and -3- also occur as root vowels with the final 80 vowel -3. Elsewhere in the corpus of the data (other than cognate items) the root vowel -35 is found with -i final vowel. -§f and -ge root vowels are found with -3 final vowel. Also a few examples of “ET and -ge root vowels are found corresponding with -g and -3 final vowels, respec- tively, summarized: £923 ending -e-, -u-, -e-, -o-, -i, -e -i-, -a- -a-, -u— -o, -a The Dera final vowels are also present or absent depending on the context in which they are used. Because the presence or absence of the final vowel with the verb structure CVCV is dependent upon the context in which the verb is used for Hausa, Karekare, and Dera, and due to the inconsistent pattern of vowel correspond- ences between root and final vowels, there cannot be said to be a consistent pattern of vowel harmony. Also because the function of the final vowels in Karekare and Dera is not clear, and in any case does not seem to parallel the function of the final vowel in Hausa verbs whose quality depends on the intended meaning as well as the context in which it is used, it is impossible based on the data presented here to reconstruct the quality of the final vowel of the CVCV verb structure for Hausa, Karekare, and Angas. Moreover, because the corresponding pattern for Sha and Angas is CVC, the possibility of the proto-forms for 81 verbs in Chadic-West-A having a final vowel is therefore questionable. Thus, the final vowel for verbs with the CVC pattern for Sha and Angas, and the CVCV pattern for Hausa, Karekare, and Dera is reconstructed as (-V), with 2 indicating no vowel specified, the hyphen indicating that it is an affixed element, and the parentheses indicating the uncertainty of its existence in the proto-form. Perhaps its presence in Hausa, Karekare, and Dera is due to an innovation and its presence cannot be reconstructed in the proto—form. Or, perhaps its presence is recon- structable and is deleted in Sha and Angas. If its pres- ence is reconstructable, its quality based on the data presented here, in any case, is impossible to identify. Of the five languages presented here, it is only Hausa which has a complex and intricate system of the function of the final vowel. This might indicate that this system and the presence of the final vowel is a Hausa innovation. Karekare and Dera perhaps developed final vowels inde- pendently with a system of vowel harmony based on and indicated by the majority of the root and final vowel correspondences, but through processes of analogy and borrowing in the separate languages the systems have become opaque. The vowel in the CV pattern of verbs seems to indicate the same verb endings for Hausa and Karekare that were represented in the final vowel of the CVCV pattern for Hausa, Karekare, and Dera. It is not known if the final 82 vowel of the CV pattern for Sha and Angas serve a gram- matical or semantic function, or if it is etymologically part of the CV verb. In Dera the final V of the CV pattern seems to be part of the root because the endings -i and -g are suffixed to the CV pattern. In Karekare, it is not clear whether the final V in the CV pattern represents an etymological vowel or an absorbed vowel ending with the root vowel, because all CV verbs in the data end with -gr or -g. It may very well be that the vowel represents assimilation of a root vowel and a vowel ending which may be the same indicating vowel harmony. In #12 'give birth' 1333 represents a CV pattern 13 with -g ending, a case for which vowel harmony may be skewed between the root vowel and vowel ending, so that vowel assimilation is not allowed to take place, and thus is accommodated by a semi-vowel making the would be CV-V pattern into a CVCV pattern. This may be evidence that the CV pattern is actually a CV-V pattern in Hausa, Karekare, and Dera. For the same reasons that it is impossible to reconstruct the quality for the final vowels of the CVCV pattern, it is also impossible to reconstruct the final vowel for CV patterned verbs, whether the V is considered etymologically a part of the word, or if it is merely added for vocalization, or it is an ending with grammatical and semantic implications. With CV or V pattern verbs, reconstructed (C)V(-V), it is unsure whether the final vowel represents vowel coalescence of the verb root and the verb ending, or the 83 vowel ending itself. Karekare 1233, #12 'give birth' indicates that perhaps others must have a root vowel and a vowel ending, -3 or -g, which coalesce, vs the supposition that the CV or V verbs have no vowel ending as they do in the CVCV pattern. Some CV/V patterned verbs include the following: H S A K D ** 12. give birth 1e la-wu lo-i lV(-V) 34. close vu pe fa fV(-V) 36. come 20 ji ndu do-i dV(-V) 44. do yi 'i yi-e i(-V) 47. drink sha swe sa sa(-V) 51. eat ci ci tu tu-i tV(-V) 123. obtain wa wa-i wa(-V) 136. put down za wu-i zV(-V) 150. see ni na nV(-V) Nouns The final vowel in Hausa nouns marks gender: final -§ indicates feminine nouns as does the suffix ~1yg used with some of the nouns, e.g. #‘s 75 'goat' and 155 'sheep'. There are a few exceptions for the feminine final -a rule of thumb- some masculine nouns end with -g. All other nouns with other final vowels are grammatically considered masculine. Other than the fact that Hausa masculine nouns 84 seem to have mostly final -1, there is no apparent system for predicting the final vowel, nor is there any evidence, other than the gender distinction, of the various vowels in the masculine forms carrying any other sort of grammatical or semantic meaning. Most of the Hausa nouns in the data end with -1 and -g. Grammatical gender is not morphologically marked in Karekare as it is to some extent in Hausa. Most of the final vowels of nouns are -g and -1. There is no apparent way of determining what the final vowel will be. Grammatical gender in Dera is not marked either. The vowels occurring most frequently in word final position are -1 and -g. There also does not seem to be any way to predict the final vowel in Dera nouns. There is no consistent evidence of vowel harmony between the root vowel of the nouns and the final vowel in Hausa, Karekare, or Dera. A few regular vowel corres- pondences for the final vowels across Hausa, Karekare, and Dera occur, but for the majority of them no specified vowel quality can be reconstructed. Also, as with the final vowels for the verbs, Angas and Sha nouns generally do not end in vowels, thus having the pattern CVC (with a few exceptions). Because of this the existence of final vowels for nouns in the proto-forms is questionable. Since Hausa is known to add the final vowel -1 for vocalization, per- haps a similar innovation took place in Karekare and Dera whereby final vowels were added for phonological purposes. 85 The problem is knowing which nouns have affixed vowels and which ones have vowels as etymologically a part of the word. The decision is made according to the information contained in each cognate set, thus the inventory for reconstructed vowels in final position for nouns. See Appendix A for cognate sets and the reconstructed lexical forms according to each gloss. A total of one hundred and thirty-three reconstructions are given. Thirteen of the reconstructed items are based on sound correspondences from all five languages; thirteen across four languages; thirty across three, and seventy-seven across two languages. NOTES 1Though Newman (1976) suggests that g is a prefix for body part terms in Hausa, he suggests that short initial 3 was etymologically a part of the words in which it occurs in proto-Chadic. Likewise, in the provisional recon- structions posited for this study this 3 is not designated as a prefix. Initial g or 22 in the Hausa verb 'to swallow' is treated as an anomalous prefix. It does not correspond to anything in the cognate set for the gloss, and it does not occur with any other verb in the data. 86 CHAPTER IV WEST-A PROVISIONAL RECONSTRUCTIONS VS. NEWMAN'S 1977 PROTO-CHADIC RECONSTRUCTIONS Discussion of Similarities PR's posited for the West-A branch in this study are compared with Newman's 1977 reconstructions representing the entire Chadic language family. Seventy-two of the one hundred and thirty-three PR's posited in this study corres- pond with proto-forms posited by Newman (1977). Of these seventy-two hypothetical forms, more than half, i.e. forty-five, show similarities with each other, with varying degrees. The remaining twenty-seven do not show any similar characteristics. Those proto-forms considered to show similarities are indicated with a juxtaposed x for each set in the following list. Of those PR's which show similarities with Newman's reconstructions, twenty-nine of them are based on less than four languages; twelve are based on data from three languages and seventeen are based on data from two languages. Only six of the PR's are based on data from four languages and only ten are based on the data from five languages. 87 88 Because metathesis is a common process in Chadic languages, the reconstructed forms for #11 'bird' are considered similar. The reconstructed forms for #101 'leg' are considered similar; the initial /a/ in Newman's recon- struction may represent the so-called body-part prefix. #13 'bite' and #45 'dog' indicate a possible regular correspondence between proto-Lax /fl/ and proto-Newman /k/ in initial position. However, Newman suggests that the reconstruction for #45 'dog' *kgr, may represent a wide- spread loan word in Chadic languages, replacing *ggg, "the true proto-Chadic word for dog" (1977:25). Comparison of Provisional Reconstructions for Chadic West-A with Newman's 1977 Proto-Chadic Reconstructions ** 'k 3. ashes bat betu x 7. baobab tree kukV kuka X 11. bird de\ dey x 12. give birth lV(-V) wa 13. bite ad(-V) kade 17. blood dom bar 18. blow deX-V) fi 19. body zi-k zi X 20. bone KYs Jagu X 21. bow rVka raga X 24. (woman's) breast ywdd waid X 31. 34. 36. 37. 41. 44. 45. 46. 47. 50. 51. 52. 54. 60. 62. 63. 64. 66. 68. 71. 72. 74. 75. 79. 80. call close come cook die do dog dream drink ear eat 999 eye fill fire fish. five fly four give go go out goat guinea-fowl hair 89 ** na(-V) fV (-V) dV(-V) (i‘Vnk (-V) mut (-V) i(-V) VdV sun sa(-V) kumo tV(-V) Y3 ido gam (-V) thV wVS erVfV va‘ dinn fVAL bar(-V) taf(-V) th(-V) Vki deumo sakV wa fu (-)se da meta ye/ca kar sawne sa semi ti a§i ide n-(y-) aku/akwa kerfi bad diwa fwade bars 6'3 pata a (w) ku zaban gasi XNNNN X 81. 85. 91. 97. 98. 99. 101. 103. 109. 113. 114. 116. 117. 119. 121. 122. 124. 130. 139. 142. 145. 146. 150. 165. head hit hunger knee knife know leg lie down meat monkey moon mouth name neck night nose oil person ram refuse roast rope see sleep (n.) 90 4* kV duk(-V) kV zV m +hi +hi erVm cuk sVn(-V) sio tV< XKNNNX >< 170. 171. 175. 183. 185. 189. 190. 195. 202. 203. 205. 208. stand up steal sun ten three tongue tooth two wash water what who 91 * * yVl(-V) éyhir ('V) fati gwm kun le aKVrV wutW bYlu wank(-V) am mY WV * Jars xsre fati 9 am k‘w)an ahlasi sank) sar(-) c-Q. mi/me W8. XXNN CHAPTER V CONCLUSION This thesis was based on the hypothesis that Newman's reconstructions for the Chadic language family may not adequately reflect the lower level structures of the classification because of his method of comparison employed (i.e. using a limited number of citations from sometimes as few as two of the major branches in the Chadic family). It was further hypothesized that positing proto-lexical forms for each branch or subbranch first, then comparing these proto-lexica would result in reconstructions which would better reflect the branches and subbranches of the Chadic classification as it stands, as well as represent the Chadic language family as a whole. A comparative study of five languages from the West—A branch of the Chadic family was presented to test this hypothesis by comparing the reconstructions posited in this study with Newman's 1977 lexical reconstructions. The languages used in the study, Hausa, Sha, Angas, Karekare, and Dera were taken to be representative of their respec- tive groups in the West-A branch. The vocabulary items used in the word list were selected to include vocabulary 92 93 items which Newman used in his 1977 publication, and to include basic vocabulary words to maximize the chance of finding cognates across the languages and to minimize the chance occurrences of loan words. The reconstructed phonemes and the sound correspondences upon which they are based were illustrated in Chapter Three. The lexical reconstructions and cognate sets upon which they are based appear in Appendix A. In Chapter Four the PR's posited in this study for the West-A branch were compared with Newman's 1977 lexical reconstructions for Chadic in an attempt to prove or disprove the hypothesis set forth in this study. The results of the comparison of the five languages, and of the comparison of the PR's with Newman's 1977 reconstructions can be summarized in the organizational chart shown in Figure 5. Of the two hundred and ten vocabulary items included in the word list, 40% (84) of them correspond to items used by Newman (1977), while the remaining 60% (126) are comprised of other basic vocabulary items. For 37% (77) of these items, the data was such that no PR's could be posited for them. However, PR's were posited for 63% (133) of the items. Of these PR's, 10% (13) are based on data across all five languages, 10% (13) are based on data across four of the five languages, 22% (30) across three languages, and 58% (77) of them are based on data from two languages. 94 mBADmmm ho Bm