TOWARDS A DIALECT GEOGRAPHY OF DANGME PART I #* E. O. Apronti and * M. E. Kropp Dakubu O. Introduction. At the time of writing, the first summer's field work for the Institute's Dangme Area Studies Project is finished, but the data has not yet been analyzed. This first phase of field work was carried out in the Ada district, and involved the collection of data in the fields of social anthropology, oral literature, linguistics, Islamic studies, health and nutrition. The work in linguistics consists of administering two questionnaires. One is directed to socio-linguistic information, in an attempt to find out what linguistic groups are represented in the area, the extent of polyglotism among Dangmes and non-Dangmes in the area, the main languages of inter-and iRtra-group com- munication, whether or not any language shift te taking place, and if so, what kind. The second questionnaire is purely linguistic in nature, and has as its aim the definition of the isoglosses that divide Dangme-land into dialect areas. In this article the authors would like to review the state of our present knowledge of Dangme dialectology, on which the survey will build. There are several reasons why the Dangme-speaking area is an attractive one for dialectological study. Physically it is quite well defined, stretching along the coast from Pram pram on the west (I) to Ada on the east, and from the sea on the south across a very dry and sparsely-inhabited plain to the Akwapim range on the north. Krobo traditional area extends through the hills and forest north of Koforidua, and is quite different geographically from the rest. The area as a whole is not too large for fairly intensive study, but large enough to provide interesting variety, linguistic and otherwise. * Senior Research Fellow in Linguistics ** Research Fellow in Linquistics *- * ** 36 The Dangme area is recognized as being culturally and linguistically closely knit. Its people recognize a common heritage and speak one language that is intelligible to all. On the other hand, it is well known to all Dangmes that there has been no political unity sJnce the old days when the clans agreed to disagree at Lolovor, and dispersed to set up independent political units at their present locations. The commonly recognized linguistic distinc- tions coincide with these political ones: Gbugbla (Prampram), Ningo, Ada, Osudoku, Krobo and Shai. Yilo Krobo and Manya Krobo are political divisions which are not commonly regarded as coinciding with linguistic divisions. Assuming for the time being that these divisions are indeed linguistically meaningful, they are dialectal in nature. That is, they reflect differences in sound, grammar or lexicon, or two or all of these aspects of language, which are obvious to the speakers of Dangme. Yet they are not so serious as to pre- vent or seriously disrupt communication between speakers of, say, Ada and Krobo, or Gbugbla and Shai. Thus the difference between the Dangme dialects is different from the differences between these dialects and their very close relative Ga, for if a Ga-speaking person has not had considerable experience of Dangme he cannot understand it, nor can a Dangme-speaking person understand Ga if he has not gone through some kind of learning process. The Dangme dialects are supposed to correspond to political divisions, and the political divisions in turn to a historical event: the dispersal, whether sudden or in stages, from Lolovor. We might hope therefore that a better understanding of the relationships between these speech types will lead to a better understanding of early Dangme history. A certain amount has already been written on the Dangme dialects. Professor Jack Berry took note of some of the differences (2) and D. A. Puplampu also commented on them. (3) One of the present authors attempted to account for some of the phonetic differences between Ada and Krobo, and noted a few morphological variations between these dialects. (4) The other author has noted a number of lexical variants between Ada, Krobo, Shai, Ningo and Prampram. (5) Thus we are not starting completely from scratch. We will summarize present knowledge and opinion about dialect differences under four headings: PART I PART II 1. The Dialect Areas: a Preliminary Hypothesis. 2. Phonological Variation. 3. Grammatical Variation. 4. Lexical Variation. (6) •i 4+ 4 A •4k i I. The Dialect Areas: a Preliminary Hypothesis. 37 Several /ears ago the second author designed a questionnaire to elicit forms which were expected to show variation among the Dangme dialects. The questionnaire consisted of sixty-five single words, almost all of them nouns, and thirty short sentences. The questionnaire now being administered Is essentially a much revised, extended and, hopefully improved version of the first one. At various times the first questionnaire was administered to persons who were speakers of one of the traditional dialects, and this discu- ssion is an analysis of the data thus obtained. Eleven respondents were inter- viewed : fourfrom Krobo, including onefrom Bukunor, two from Odumase, and one from Somanya; two from Shai, including one from Dodowa and a group of old ladies at Doryumu; one from Akuse, as representative of Osudoku; one from Pram pram (Gbugbla); one from Ningo; and two from Ada, one from Ada itself and a group of fishermen at Ocansey Korpe. Except for Ada, Ocan- sey Korpe and Doryumu, all the respondents were interviewed at Legon. Of the latter, the respondents from Gbugbla, Ningo, Somanya and Bukunor and one of the Odumase respondents were university students. The follow- ing discussion is therefore strictly tentative, and the hypothesis is presented subject to support or destruction by the results of the comprehensive field investigation now being undertaken. About a hundred and thirty isoglosses (7) were derived from the mate- rial of the questionnaire. However, many of them included all the varieties of Dangme represented, and a few included Ga as well. Here we shall take it for granted that many important isoglosses unite all varieties of Dangme, and at the same time separate them from Ga. Most such isoglosses reflect innova- tion in Ga, and conservatism in Dangme, but a few, such as the borrowing of ogbeteS 'hyena' from Ewe, change of 8 to D after certain labialized consonants (as in dami, duami, Proto-Ga-Dangme *du£ - 'rubbish') and the loss of a palatal element in many words that now have d (as in do 'dance', da'valley', from *dio-, *di> respectively) are innovations that make the Dangme dialects a positively identifiable unit. In the present discussion we are only concerned with those isoglosses which unite some dialects in opposition to others. Therefore not all dialect samples are included in the same number of isoglosses. Ada and Ocansey Korpe were both included in forty-three, while Somanya for which only an incomplete response was available, was included in only twenty-nine. Most were included in between thirty and thirty-five isoglosses. The question- naire was constructed in such a way that a majority of isoglosses are of diffe- rent pronunciation of the "same word", a smaller number are morphological, and some are lexical. It is possible that an analysis based on one type only, or with a different relative weighting, would give different results. This is another reason why it must be stressed that these remarks are strictly tentative. & is* * 38 The technique employed for this analysis was to abstract from the responses which dialects were united by the same isogloss. The number of isoglosses uniting each dialect to every other dialect were then counted. For instance, if two isoglosses unite Ada with Akuse, and three unite Ada with both Akuse and Ningo, and one unites Ada with Ningo, then five isog- losses unite Ada with Akuse, but only four unite Ada with Ningo. For each dialect, the percentage of its total isoglosses that unite it with each other dialect was calculated. Since each dialect does not enter intothe same number of isoglosses, the percentage of all isoglosses including Ada that link Ada with Ningo is not necessarily the same as the percentage of all isoglosses including Ningo that link Ningo with Ada. However, in most cases the two percentages were close enough to be equivalent for our purpose. Exceptions will be noted. The following hypothesis about Dangme dialects and their interrelationships is constructed on the basis of these percentages. I.I Members of the "Same Dialect". As far as the data gathered is concerned, the traditional division of dialects has meaning, in that idiolects (8) from the same traditional dialect area are joined by 80% or more of their isoglosses, and idiolects of different traditional dialects are joined by not more than 68% of their isoglosses. Moreover, with idiolects from different traditional dialects, if the link is more than 60% in one direction, it is usually less than that in the other. Thus 67.6% of Nlingo's isoglosses include Gbugbla, but only 54% of Gbugbla's include Ningo. 65.8% of Akuse's isoglosses link it to Ada, but only 55.8% of Ada's link it to Akuse. 60% of Doryumu's isoglosses link it to Ningo, but less than 33% of Ningo's link it to Doryumu. Identity of Ada and Ocansey Korpe is nearly complete, since about 95% of the isoglosses of each include the other. These two together will hence- forth be referred to as the Ada dialect. Bukunor, the two Odumase idiolects and Somanya are slightly more divergent, but each is linked to the other by about 85% of its isoglosses, and will be referred to together as the Krobo dialect. Odumase is linked to Somanya by only 78%, but this is probably because the data for Somanya is incomplete. Inclusion in the other direction is 86%. Doryumu and Dodowa diverge a little more than the varieties of Krobo, being linked by 80% of isoglosses in each direction. They will be referred to jointly as the Shai dialect. Akuse, Ningo and Gbugbla are not joined by more than 68% of their isoglosses to any other speech type repre- sented, and hence are regarded as three distinct dialects. Several isoglosses distinguish Ada from all other dialects. Only Ada. (in this material) had bafono 'European', bsfo for 'maize', pumi for 'mosquito', ho for 'hunger', tsi for 'ladle', honyu for 'soup', and voli for 'foam froth'. Only Krobo has okunu for 'shea butter', sone for 'pillow', lakusu for 'smoke\ and dabodabo (in Somanya and Odumase) for 'duck*. Gbugbla is unique in *, i * 4 4 m *r 39 having emi for T, wa for 'grey hair', the form momi for 'okro', and menofor 'tomorrow'. The forms muna and mane'for 'tomorrow' are peculiar toAkuse and Ningo respectively, but otherwise these dialects seem to be distinguished mainly by the different isoglosses that link them with other dialects. This is also true of Shai. For example, blSfono 'European' occurs everywhere except in Ada. pami 'mosquito' occurs in Shai, Krobo and Akuse. imi T occurs in Shai and Krobo, and lasuku 'smoke' occurs everywhere except Krobo. Thus neither blSfono, pomi, imi, nor lasuku occurs only in Shai, but only Shai has all of them. Similarly pami 'mosquito' occurs in Akuse Shai and Krobo and ami "I" occurs in Akuse and Ada but only Akuse has both. Ningo and Gbugbla both have pumu 'mosquito' Ningo Ada and Akuse have sunue 'pillow' and Ningo Shai and Krobo have ngua 'grey hair' Only Ningo has all three. 1.2 Closely Linked Dialects. Examining which dialects are linked by 50% or more of their isoglosses produces a fairly clear picture of which dialects are most closely linked, in the diagram below linked dialects are joined by 50% or more of their isoglosses. A rather interesting chain pattern emerges whereby each dialect is linked to two dialects those on either side of it in a sequence encircling the Accra Plain in the order Krobo - Shai - Akuse - Ada - Ningo - Gbugbla. One might therefore expect Krobo and Gbugbla to be the most divergent dialects, but actually this is not quite so, as is indicated in 1.4 below. In the cases of Krobo and Shai, Akuse and Ada, Gbugbla and Ningo, 50% sharing is reciprocal: that is, 50% or more of Krobo's isoglosses link it to Shai and vice versa. These three closely linked pairs of dialects are linked together by pairs where linkage is much stronger in one direction than the other: 50% of Akuse's isoglosses link it to Shai, but only about 40% of Shai isoglosses link it with Akuse. Similarly 59% of Ningo's isoglosses link Ningo with Ada, but only 44% of Ada's link it with Ningo. In the diagram below, broken lines indicate that 50% inclusion obtains in one dirction only. Fig. I 40 •0 Several isoglosses link Krobo and Shai only, eg. those forfmiT, nangm? 'navel', ahlu 'tortoise', osrami 'moon', aklowa 'village', akate 'groundnut'. There are also several that Imk Akuse and Ada only: atsta 'pineapple', gigs 'groundnut', nr>mui 'okro', ami 1', sugba 'earth'. The words katie 'for 'ground- nut', pumu for 'mosquito', duami for 'rubbish', and ns in the second syllable of the word for 'today' are unique to Ningo and Gbugbla. (9) But at least in our data, there is only one isogloss that joins Ningo and Ada without including other dialects as well, namely that for anyazia 'story*. (Akuse and Bukunor have initial a-, and Shai and Somanya have ny rather than n, but only Ada and Ningo combine these features.) Between Akuse and Shai there are no unique isoglosses. These two dialects are joined by such isoglosses as that for blgfono 'European', which also includes Krobo, Ningo and Gbugbla, that for psmi 'mosquito', which also includes Krobo, that for lasuku 'smoke', which includes all but Krobo, and for ayanu 'shea butter', which also includes Ada and Ningo. Thus reciprocal 50% inclusion, represented by solid lines on the diagram, seems to co-incide with the occurrence of several uniquely characterizing isogtosses. Unidirectional 50% inclusion, represented by- broken lines, seems to concide with absence or near absence of such isoglos- ses. Another interesting feature of this pattern is that Akuse (probably Osudoku generally) seems to act as a link between the dialects of the coast and those inland. Although it is geographically closest to Odumaseand Bukunor, it seems to be linguistically much closer to Doryumu and Dodowa. Akuse is very near the traditional dispersal point of the Dangme in northern Osudoku. One is led to wonder whether this pattern of linked dialects does not reflect the original pattern of migration outward from near Akuse, one stream moving inland in stages, and another moving down the Volta and westwards along the coast. 1.3 Secondary Dialect Areas. When we examine those dialects that are linked by more than 33% of their isoglosses, a secondary pattern of overlapping dialect groupings is obtained. In this grouping, Krobo is still linked only to Shai. Ningo and Akuse 41 are linked by at least 33% of their isoglosses to Shai on the one hand and Ada and Gbugbia on the other, but Shai is still not linked to Ada or Gbugbia (see Figure 2). Gbugbia is excluded from the group that includes Shai because inclusive isoglosses were less than 33% of the total for Gbugbia — Dodewa, Gbugbia —• Doryumu, and Dodowa — Gbugbia. However, it is more than 40% for Doryumu — Gbugbia. Doryumu — Ningo linkage is more than 50%. This suggests that although Doryumu and Dodowa are related as varieties of one dialect, their relations with other dialects may be quite different. It is possible that the presence of Ningo herdsmen on the Shai plains between Doryumu and Agomeda is a factor in the situation, but until more information is available this is only a possiblity. 1.4 Minor Dialect Links. Other groups or pairs of dialects are linked by 15% or more of their isoglosses. This is the highest level at which Krobo is linked with any dialeGts other than Shai. For instance, it is linked with Akuse by blgfo 'maize', pomi 'mosquito', and with Ningo by for instance blefota 'pineapple'. At this level also Shai is linked with Ada. for instance by ayonu for 'shea butter', and with Gbugbia, for instance by sune for 'pillow'. Gbugbia is distinctively linked with Ga, mostly in the form of borrowings from the latter. Only Gbugbia and Ga have wobi for 'bee', w as the initial sound of the word for 'grey hair', bbfonyo for 'European', adesa for 'story' tale', bbfonme for 'pineapple', klakunu for 'turkey' and nku for 'shea butter'. The groupings linked by 33% or more of the isogjosses of each dialect are shown in the diagram below. Fig. 3 It may be observed that at none of these levels is there any link between Krobo and Ada or between Krobo and Gbugbia. Linking isoglosses between Gbugbia and any of the varieties of Krobo is consistently below 15%, and Odumase-Gbugbla is also below 15%. However, Somanya — Gbugbia and Bukunor — Gbugbia are linked by 20% and 2 1% respectively (6 to 8 isoglos- 42 ses). Between Krobo and Ada, however, there are consistently only one or two linking isoglosses (domi 'rubbish'). Thus the most extreme divergence among the Dangme dialects is demonstrably between Krobo and Ada. 2. Phonological Variation. In this section, discussion is limited to certain features resulting from phonetic change that distinguish Ada and Krobo. No mention will be made of other dialects for three reasons: I) As shown in section I, Ada and Krobo represent the extremes of dialectal variation in Dangme, and most of the important differences are differences between these two. 2) Ada and to a lesser extent Krobo seem to be the principal foci for innovation and the diffusion of innovations. 3) Quite a lot of reliable and relevant data is curren- tly available for these two dialects, but very little for the others Several types of phonological or phonetic difference distinguish Krobo and Ada. Of the following types, three are essentially innovations in Ada, one is the result of an innovation in Krobo, and one (labialization of initial consonants) involves change in both. 2.1 De-labialization. 2.11 There are several words that contain a syllable bua in Ada, but bo in Krobo: Ada bus nubrn Krobo nib} bami 'grow' (verb) 'stranger' 'hair' Others have wo in Ada but ho in Krobo: wo ho 'honey', also 'bee'. 2.12 There are also words that have a syllable ma, h(l)a or ho in Ada but muD, muE huo or hue in Krobo: Ada nromui hbmi honyu Krobo muDmi mueno huD huami huenyu •okro" (first syllable) 'today* 'hunger' 'coughing' 'soup' 2.13 On the basis of comparison with Ga, it would seem that in each case the more conservative form is the one with u or w, and that the dialect without this has lost it. Concerning the absence of h in the Ada word for 'bee', see 2.3 below. Concerning the vowels of the words for 'today' and 'soup, see 2.4 below. 4'i * V •1 * 9*' 2.2 Intrusive Labialization. 43 A phonetic difference which looks very similar to the first is probably to the introduction in Ada of u labialization, in second syllables with close or half-close front vowels, and nasal consonants or I, when the first syllable has a back vowel: * Ada sunue sulue rmmui Krobo sune sule miorai pillow' 'penis' 'okro' (second syllable) 2.3 De-aspiration. (10) A number of words that have hu in Krobo have w in Ada: Ada WD3 WD woli WO Krobo ho 'tomorrow' 'sleep' (verb) 'foam' 'honey', also 'bee' It seems that in Ada, the proto-syllable *hua became WD, except when it was nasal or affected by an infix - I -. In that case it was de-labialized, as in 2.12 above. Ada has also lost the h element from the proto-syllable *huo, where Krobo has lost the labializing element (see 2.11 above). ~ 2.4 Backing of Vowels. In certain two-syllable words, where the second begins in a nasal con- sonant, the original front vowel of the first syllable has become a back vowel of the same height in Ada: Ada nono honyu Krobo mu8no huenyu 'today' 'soup' 2.5 Initial Nasalization. There are two words, both names of parts of the body, in which the initial consonant is nasal in an apparently irregular way: Ada langnrD mua Krobo nangma ngua 'navel' 'gray hair' It is suggested that the initial nasal consonants are relics of old nasal prefixes which have been lost or assimilated in different ways to the original stem- initial consonant, *l in the first word and *w in the second. Comparison with the Ga words (lanms and wan) supports the hypothesis that the original stem-initial consonants were non-nasal. 44 3. Grammatical Variation. In this section we will remark on a few aspects of the grammar of Dangme where there seems to be dialectal variation, either in the realization of mor- phs expounding a structure or in the grammar of the structure itself. There are further areas where there seems to be variation, but more investigation is required before they can be discussed. 3.1 The Stative Construction, ( i l) 3.11 The Shape of the Stative Suffix. The main verb in a stative construction has a characteristic suffix, which varies somewhat. Examples collected by means of the first survey question- naire indicate that the variation is probably essentially dialectal, but rather unstable. In the Krobo idiolects, the suffix was realized by vowel length if the verb stem vowel was e, £, or D, but after a, u, and o it is usually ye or e, sometimes £ after u, and sometimes length. There is variation both between and within idiolects. Thus, Bukunor e bi lae, e bi laa o nge no nuS o nge doo 'he would not sing* 'you are drinking* you are dancing' Odumase (2) i nge laa, i nge laye i bi nuye o nge doye 'I am singing' ' would not drink' 'you are dancing' In Akuse, Ada and Gbugbla the suffix is vowel length after e, s, o, a, a, but e or e after u. Doryumu is the same, but the Dodowa idiolect also had vowel length after u, possibly conditioned by the choice of modal verb: Dodowa o nee nyu nuu 0 nge nyu nuu 1 be nue ma hi nuu 'you are drinking water' 'I would not drink' 'I will be drinking' The Ningo idiolect consisteatly had vowel length as the only realization of the stative suffix. It is curious that Ningo should most resemble Dodowa 4n this feature, since it generally tends to show more affinity with Doryumu (see I. above). 3.12 The Future Negative Modal Verb. Verbs with the stative suffix occur following one of the three modal verbs, hi, nge and bi/be/be. It seems that the last mentioned consistently has the shape bi in Krobo and be everywhere else except in Doryumu, where it is be. * Urn •<* -it "A -•*» •* •* A, -4 ffc- 1ft ft. 45 3.2 Formation of the Negative. There seems to be variation in the formation of the negative of certain classes of verbs, namely low toned disyllables and mid toned verbs. The pattern underlying this variation is not clear. In a set of data collected by the second author, the negatives of bole 'surround', sele 'swim', du 'bathe' and mis 'wait' may be bolee we, selee we, du we and mis we, or busli, sili, duiand mis. busli, sili, du we and mle we appear to be characteristic of Krobo, and bolee we, selee we, dui and mle are Ada. However, the first author reports busli and sili from Ada, and bole we and sele we from Krobo, and considers bo lee we and selee we to be more liksly to be Gbugbla. There is also a difference in the placing of the negative particle we, in those instances where it occurs with the same verb in both dialects. In Krobo, we always immediately follows the verb, but in Ada it comes at the end of the clause: Krobo Ada e la we lahi e la lahi we 'he did not sing songs' 'he did not sing songs' 3.3 Nominalization by Reduplication. In Ada, adjectives and nouns are derived from monosyllabic verb stems by reduplication plus a suffix consisting of a high toned front vowel showing cross-height harmony with the stem vowel, for instance: gba si ko 'marry 'fry' 'knock' gbagbe sisii kokoe 'marriage' •fried* •broken* This structure has been thoroughly described elsewhere by the first author, (12) and its affinity with an Ewe structure has been discussed by the second author. (13) It definitely does not occur in Krobo, and it remains to be seen whether it extends appreciably beyond the Ada area, kokoe 'broken' appa- rently occurs in Akuse. Notes. 1. We have not considered Kpone in this article, because at present we have no relevant data on the Dangme spoken there, and because today its first language seems to be Ga. 2. In Jack Berry,Structural Afpnities-of the Volta River Languages, PhD Thesis, University of London, 1952. 3. In D. A. Puplampu, Dangme Manner of Speech, London, 1953. t St. *• 46 4. In M. E. Kropp, A Comparative Study of Ga and Adangme, PhD Thesis, University of London, 1968. 5. In E. O. Apronti, The Writing of Dangme, Institute of African Studies, Legon, 1970. 6. Part I is primarily the responsibility of Dr. Dakubu. Dr. Apronti will be primarily responsible for Part II. 7. An isogloss is "a line on a map in a dialect atlas delimiting areas within which certain linguistic features, as pronounciation, vocabulary, etc. are exhibited in common". (Funk and Wagnalls Standard Dictionary). Since no dialect atlas of Dangme exists, we are here using the word to refer to isoglosses that we could draw on a map if we were to compose such an atlas. 8. An idiolect is "the totality of speech habits of a given person at a given time". (C. F. Hockett, A Course in Modern Linguistics, New York, 19518, pg. 321). Here of course we are only concerned with a small part of the idiolect of each informant. 9. In this paper "ng" is digramph representing the velar nasal consonant. 10. The first author reports the form duaml from Ada, although it did not occur there in the responses to the questionnaire on which this dis- cussion is based. 11. For a fuller treatment of the rather complicated correspondences involving h and w, see M. E. Kropp Dakubu, A Problem in the Recon- struction of the Ga-Adangme Consonant System, in the Journal of African Languages Vol. 10 No. 3 1971. 12. This construction is described, with the terminology used here, in M. E. Kropp Dakubu, The Adangme Verb Reconsidered, Journal of African Languages Vol. 9 No. I, 1970, pp. 19-26. 13. In E. O. Apronti, A Phonetic and Phonological Study of the Nominal Piece in Dangme, PhD Thesis, University of London, 1967. 14. In M. E. Kropp-Dakubu, 1970 note 4. •0 T0 4 * •t