Download as pdf
Download as pdf
You are on page 1of 28
‘Sprache und Geschtante in Afrika 6 (1994/85) 253-308 Tue Swaatet DiaLects oF KENYA IN RELATION TO MIVIKENDA AND To THE Bantu ID1oMs oF THE TANA VALLEY by ilhela 9.6. Méhlig University of Cologne conTENTs 1. Introduction 1.1, Aims and perspectives 1.2. Previous classifications of Kenyan Coastal Bantu 1.3. The baste patterns of historical development guage 2. the actus? dialectal situation 2.1. Introductory remark 2.2. The geographical distribution of Bantu Idioms of the Kenyan Coast 2.1. The Swahili diatects 2.2, The MiJikenda dialects 2.3, The Pokowo dialects .2.4, Tiwana 2.3, The diatectonetrical structure of the dialects 2.3.1. The structure of internal relationships of Swahili 2.3.2. The external group relations of Swahili 2.4. Discussion of the divergent dialectonetrical ingfces 2.4.1, Preliminary renark 2.4.2, Comments on the group internal indices of Swahili 1.3.1, Notes on their presentation 1.3.2, Mode? no. 1: Unilinear monogentic relationship 1.3.3, Model no. 2: Stratificational genetic relationship 1.3.4. Rodel no. 3: Shift of genetic affiliation 1.3.8. Model no. 4: Areal spreading 1.3.6. Model no. 5: Paratiel borrowing froma third lan- 3 3. 3 4 4 ruaeoe 2.6. wiz 2.4.3. Comments on the group external indices of Swahili Qualitative historical analysts 1. Lexical analysts 2. Phonological historical analysis, 3.2.1. Methodological remark 3.2.2. Description of the chronologies of soundshifts 21, Tana witatuw Buu Chonyi and Digo Tikuw Siyu and Pate mw Rvita and Jomvy 9. Chifundi (Chirazt) -2.10. Vunba 2.2.3, Comparative conclusions drawn on the language history drawn from the indiviguel chronologies of soundsnifts A general hypothesis of the history of the Swanili dislects of Kenya in relation to the other Santu {dions of the coast- al area 1. The internal history of the Swahili dialects S11. Thuy, Siye ané Pate 4.12. kay 4.1.3. Chifund! and Vunbe ALA. Mite and Jonve 2. The external Bantu history of the Swahili dialects 4.2.1, The relations between Swahili and Pokono, 4.2.2. The relations between Swahili and WiJikenda 4.2.3. The relations between Swahili and IIwana 3. The general traits of the language history of the Tan- guages and dialects of the Kenyan coast L. Introduction 1.1, Aims and perspectives In this paper, it 15 my aim to discuss the historical Yin- quistic relations of the Northern and Central Swahili dialects within the network of the neighbouring Bantu idions, namely the NiJikenda dialects in the immediate hinterland of the Kenyan Coast, and the languages and dialects of the Tana valley known as Ilana (Elwana, Malakote) and Pokono. Since there are no written documents of any significant time depth available to reveal the previous stages of linguistic history in the coastal region of Kenya, ay historical approach is based on the compar- ison of symchronical Vinguistic data, which can be verified at any time by empirical methods. It deviates from the traditional comparative method in two basic points. Firstly, in contradistinction to the principle of “test languages" as applied, for instance, by Guthrie (1967-71), my comparison is guided by the so-called "dialectological princi- Dle*. This means that the comparative work is done at the dia Tectal Yevel in contiguous areas, leaving no dialecta? noman's Tand out of consideration. The reason for this procedure is the empirical observation that, in the various dialects of a ata- Ject cluster or 2 language, the clocks of linguistic develop- Pent go at different speeds. Thus, when looking into related dialects for specific comparable features, one may find, side by side, several forns which are only different stages of the sane historical process. With respect to one dialect, these stages would occur in a chronological sequence which is, of course beyond the reach of our empirical observation. At the dialectal evel, however, the same stages appear in a spatial sequence or distribution, where they can be investigated with the empirical methods of dialectology. For the historical perspective, we only need to arrange the various stages of linguistic development in 4 chronological order. This order, as all historical computa tions based on synchronical data, is hypothetical. However, the 256 ween 9.6. noaxe stages a5 such are proven. Secondly, whereas the traditional comparative method leads to an inventory of more or less isolated historical equations of the type tp corresponds to , *e corresponds to = etc.y the historical diatectological method (cf. the detailed deserip- tion in Heine and MohIig 1980: § 3.14) leads to the formulation of dynamic soundshifts which, to a great extent, apply to whole classes of sounds and which are often interrelated in the sense that they condition each other 1.2. Previous classifications of Kenyan Coastal Bantu Several aethods of classification have already been applied to the Bantu languages and dialects of the Kenyan coast. The first to mention is 2 referential classification by Malcolm Guthrie in 1948. It is based on synchronical Vinguistic and geo- graphical features. According to this classification, the major Swahili dialects known to Guthrie at that period were grouped together in group 40 of zone G. In the same zone, we find other groups of dialects and Tanguages which are mainly spoken in Tanzania, such as Gogo, Shambala, Ziguia, Zarama, Pagolo, Sango, Hehe, Pangwa and Kinga. whereas Mijikenda and Pokomo, which 1 consider to be the clotest relatives of the Swahili diatects (Won) ig 1978:187), are classified by Guthrie under group 70 of Bone £, a zone which covers the whole of the northeastern area of Bantu There is @ second kind of synchronical classification which is based on the actual linguistic distances of the idions con cerned, without taking into account the historical factors which night have led to the actual networks of dialectal interrela- tions. Since this kind of classification is achieved by measur- ing and weighting areal linguistic features, 1.e. their quali- ties, quantities and patterns of distribution, 1ts underlying Fethod ‘s called déateotometry (MbHIig 1980a). Such a dialecto- metrical analysis limited to Northern Swahili and Comorian was published by MohIig, Phitippson, Rombi, and Winter in 1980. & [SIE SMILE DIALECTS OF KEMYA 27 ore conprehensive dialectonetrical analysis including also Some of the Tana valley Tanguages and the major Mijikenda dia- Vects was done by a working group of the ATP Intemationate au CnRS at Ivry-sur-Seine in 1961.! So far, the results have not yet deen published. Sone of the Findings will, however, de used as arguments in the following analysts A third kind of language classification applied to Swahili together with other Coastal or Northeastern Bantu Tanguages aims at the reconstruction of the hypothetical genealogical Vines which connect these languages. The first classifications of this sort were published by Hinnedusch (1976), and by myself (2978). Here, Pokono and Mijikenda form a closer subgroup as against Swahili. In my own classification (1978), Tiwana is tentatively linked to this group at a rather remote level (as to its actual position see section 3.2.3.). Another genetic classification was done by Nurse and Philippson in 1980 an the basis of lexicostatisticel methods. According to these authors, Swahili and Wijikenda form a common subgroup at an average level of 64.75% within a larger group called Sabakt, where, at the level of 59.5% also Southern Pokomo 1s Joined. In their classt- fication, Uwana and the Upper Pokona dialects are not consider ©4, el tnough Upper vokono appears to be tacitly Included (et. Hinnebusch, Rurse, and Mould 1981:10). A more sophisticated genealogical approach to classify the Northeastern Bantu languages and dialects, and among these Swahili and {ts Bantu neighbours, is applied by Hinnebusen (Hinnebusch, Nurse, and Mould 1981: chapter 2). In a first step, lexicostatisties is used to form a working hypothesis as to the Genetic relations among the Northeastern Bantu languages. In a Second step, the relevant subgroups, according to the lexico- statistical results, are defined and/or verified by ascertain- ing their shared historical linguistic innovations. These are mostly formulated as "sound shifts", although this term is used in an untechnical sence different from my ovn use. As a result, contrary to the previously mentioned classification by Nurse 28 maLsEN 9.6, wae and Phitippson 1980, Mijikenda and Pokome form again 2 closer Subgroup, whereas the Swahili dialects are affiliated at a high- er nistorical level. Hinnebusch (1976:118) explains the discrep- ancy between the lexicostatistical and the historical phonolo- gical results as effects of the cultural predominance of Swahili over most of the Mijikenda dialects and as a consequence of its geographical nearness, in other words, as areal effects or phe- nomena of contact. In his genetic classification, Hinnebuseh gives evidently aore weight to the phonological data than to the lexical ones. My own attempts at classifying the Bantu languages and, in particular, Northeastern Bantu, with 2 historical perspective are mainly based on historical-phonological methods (MBhIig 1977; 1978; 1981). Yet, it differs from the method applied by Hinnebusch et al. in three essential features. Firstly, as 1 nentioned already, the sound shifts as I understand them are not mere equations between sone hypothetical Proto-Bantu phonenes and the phonemes found in the actual languages, but they are meant to be formulas which reflect the dynamic processes of Phonetic or even phonological changes with all their intermedi- ate stages. Secondly, I try to arrange the sound shifts for each odern language or dialect in their chronological order. Since this procedure will be discussed in more detail in the following chapters, it may here suffice to say that @ straigthforard chro~ nology is not always possible. However, in 80% or even more of the cases which I have worked on, I have been able to arrive at Plausible results. Thirdly, the sound shifts which are shared between lenguages or dialects are not automatically interpreted by me as evidence for unilinear genetic relationship of the tra- ditional type. I also take into consideration other than genetic factors of language history which may have led to the actual dialectal situation. In the following section, f shall briefly introduce the five basic patterns of historical development Which, so far, I have been able to distinguish empirically (MonTig 1978:182¢¢.). (mie Swmuizez DEALECES OP KENYA 259 1.3, The basic patterns of historical development 1.3.1, Notes on their presentation These patterns are rendered here in the form of five dia~ grams showing the historical processes in tsolation. In reality however, it often happens that these patterns co-occur and that therefore the picture is mare complicate. The top lines in the diagrans show the actual state of re- lationship between two languages, exemplified by simple sound shifts and/or e typical case of linguistic correspondence. The bottom lines show the reconstructed, hypothetical proto-states. 1.3.2. Model no. t: Unilinear monagenetic relationship characterized by ‘exceptions Plagean ao. 1 The diagram reflects the old model of genetic development. This model has been dominating historical linguistics for the Vast hundred years, ot least in African studies. It has been used to explain always renoter relations among languages so that the classifications of this kind contained Jess and less historical information in the true sense of the word (e.g. Greenberg 1963). If this model is hovever applied to cases with comparatively little tine depth, the amount of historical in- RSL 3.6. wi formation automatically increases. In other words, then only does the model become useful in the true sense of language his- tory. Despite the fact that this model has been dominating his- torical linguistics for the Tast hundred years, it evidently is marginal as far as its relationship with real historical events is concerned. Particularly in the larger Bantu field, it has not led to any plausible results, i.e. results which are in agreenent with the great dynamics of ethnic and cultural history in the Bantu region, as ascertained by other methods and disciptines 1.3.3, Wodel no. 2: Stratificational genetic relationship In the historical comparison of the Bantu languages, we frequently find regular sound correspondences which, at @ closer Took, prove not to be the effects of the same sound shifts, a5 one would expect according to the unilinear mono genetic model no. 1, but as parallel strains of developnent. Since the traditional way of applying the so-called comparative method is to work rather with historical phonological equations than with reel sound shifts, these phenomena usually remain un- Jiscovered. However, 11 we work on the basis of sound shifts oF even chronologies of sound shifts, as it is demonstrated in this article, we recognize that, in fact, many regular sound correspondences are based on parallel strains of historical evelopment When working on sound shifts instead of historical sound equations, there is yet another fact to be discovered: the sane language may belong to different strains of development, de- pending on the features compared. These findings have induced me to introduce the concept of Stratification into Bantulstics (MShIig 1977; 1979; 1981) as an alternative methodological basis for historical linguistic comparison in this field. After this first step, it soon be- came evident that there are different types of stratificational relationships. They are presented in this and in the following three models of linguistic historical developaent ‘non sound ante, Topeent, being culy BiniLar oot regular This model is characterfzed by similar, but not identical Sound shifts deriving from similar but different strata. - Host of Guthrie's so-called “osculant comparative series" (Guthrie 1967-71: vol, 1, chap. 7) exemplify this model. - If these strata can be related by regular or quasi-regular sound equations, 1 assume that they are genetically interrelated, but only at a remote level which is beyond the reach of our comparative meth- ods. As to the mature or character of this renate language Situation, there are cogent reasons to assume that the Vin- guistic and dialectal diversity in Africa two thousand or even ten thousand years ago did not much differ fron the complex vatterns Of linguistic relations which we find at present. It would there- fore be highly artificial to conceive the stratificational ge~ netic relations as having originated from one individual proto Vanguage such as the highly fictitious Proto-Bantu. Instead, it seems to be more in agreement with the actual patterns of dia lectal diversification, and therefore also more realistic, to assume that the parallel genetic strains rather originates from proto-language families or stratified clusters of proto- Tanguages and proto-dialects then from one individual proto- Vanguage 262 wrnMeLe 3.0. sere 1.3.4, Nodel no. 3: Shift of genetic affi1station --+ m2 or only s part of tne | sane pales of phonenns [regular correspondences This model represents another case of stratification, which is caused by Tanguage shift. The basic situation is as follows: A former genetic affiliation was given up and a new one adopted. However, during the process, a significant amount of the phono- logical characteristics of the former language were retained. Language shifts are comparatively easy to discover, if the changes toke place between languages of different genetic fi ‘lies like Cushitic and Bantu. shifts may, of course, also occur between languages of the same genetic family. Such cases are mostly overlooked when exposed to the traditional comparative sethod. In ay opinion, only dialectological methods are capable to make them overt. 1.3.8. Model no. & Areal spreading ne? Common features which are archaic in a, bit recent HE swausnr orALECRS OF KENYA 253 If we compare two languages A2 ané 62 and discover that they show some regular correspondences in their phonological inventories, we would expect, according to the genetic model, that the positions or ranks which the connen features take in the chronologies of sound shifts are the same in both languages If, however, the chronological ranks are different (as to the procedure of defining chronological ranks, see section 3.2. be~ low), we can be certain to be faced with a case of areal spread- ing. In the diagram no. 4, it 15 assuned that the spreading takes place from idiow AZ into idiom 82. The languages Al and B1 at the bottom line of the diagram represent the state before the areal spreading. Wistorical processes according to the model no. 4 are very Frequent in Bantu. The source languages of such areal spreadings have often a higher prestige, and their speakers are culturally nore innnovative than their neighbours, or these languages are Used as trade languages or Tingua francas. Many recent examples for this can be found within the sphere of influence of Standard Swahili in East Africa 1.3.6. Model no. 5: Parallel dorrowing froma third Tanquage Sintlar features sxe (gular ia both systems The case of parallel borrowing from a third language is, to a certain extent, a sub-case of model no. 4. If we compare two languages, we sonetines discover similar features which, because of their irregularities in doth systems, can neither be explain- 264 MeL 9.6. Habaig ed as the heritage froma common ancestor language nor as an influence from one of the languages of comparison upon the other Venguage. In such cases, the reason often is that the common features have been caused by a third language which influenced both languages of comparison. In an area where there are power- ful Vingua francas, as in the case of Swahili for East Africa, this type of language history may be very important 2 The actual dialectal situation 2.1, Introdvetory remark The actual dialectal situation of the languages which have been chosen for historical linguistic comparison is practically ‘the only enpirical basis for this kind of study. there are several parameters to analyze and to describe the ectus) dia- lectal situation. One is the geographical distribution of the ‘dions of comparison. Another is the pattern of dialectal split ting. These paraneters will be used in the following section 2.2 Furthermore, there is a systen of dialectal distances in terms of linguistic features shared between the idioms. These san be measured by dialectometrical methods (Mbh1ig 19808) These methods have been developed to the extent that lexica? a5 well as phonological data can be used for this purpose. with respect to Swahili and its Bantu neighbours, this will be done in section 2.3. of this article Often the lexically based diatectonetrical indices of near ness differ markedly from the phonological indices. Such dis crepancies have not only a historical reason, but, by their nature, they also indicate which historical factors were at work to produce these discrepancies. In section 2.4., 1 shall eal with this aspect of dialectal comparison HE SwaueL DIALECES OF xxIA 265 2.2. The geographical distribution of the Bantu idions of the Kenyan Coast 2.2.1, The Swahits dialects As the map shows, the Swahili dialects of Kenya are spoken along the coast of the Indian Ocean from the Sonat ian border inthe north down to the Tanzanian border in the south. The dia- ectal areas of Swahili do mostly not cohere geographically with each other. Since until recently the road connections were 4ifficult, the interdialectal communication was principally effected vie the sea routes. In Kenya, we can distinguish three areal groups of Swahili dialects: (1) a northern group, rough Ty between the river Sabaki and the Somalian border, including the islands offsnore, (2) a central group south of the Sabaki up to Mombasa and its hinterland, (3) a southern group around the peninsula of Shimoni and the island of Masini and Funzi. In detail the dialectal distribution is as follows: (1) Northern Group of Kenyan Swahili 1. Tikuuz spoken on the coastal strip from an area oppo- Site Lamu up to the Somalian border, in the north eastern part of the island of Pate (Kizingitini, Faza) and on the small islands north of Pate 2. Pate: spoken in the small township of Pate in the southeast of the island of Pate. 3. Siyu: spoken in a smatl town in the centre of Pate island 4. Rau: spoken on the island of Lamu with three diatectal varieties: Lamu Town, Shela and Matondoni, on the coastal strip opposite Lamu in pockets up to Witu, Mambrus and Malindi (2) Central Group of Kenyan Swahili 1. Mvita: the ialect of the area between KiTifi and Monbasa, in pockets up to Malindi. 266 rues 3.0. menare Ama ley 8 exouo ak ee, i? a ” Lf Seoaraphical distribution of the Bantu Lasone of the ‘ee swasiee rALECTS OF KEIN 267 2. Jonvu: spoken in the hinterland of Mombasa (3) Southern Group of Kenyan Swahili 1. Chirazs (chi funds): spoken in some small villages north of Shimont, on the {sland of Funzi and the eastern part, of the island of Masini 2. Yunba (Vanga)+ spoken along the coast from Shimoni up to the Tanzanean border 2.2.2. The MiJikenda diatects Geographically the nearest neighbours of the Swahili are the Mijikenda. As the name of "nine settlements" already sug- gests, the Mijikenda, according to their own concept of inter- nal structure, are subdivided into nine dialects. these are: Giryama, Kauma, Chonyi, Jibana, Kambe, Ribe, Rabai, Durum and Digo. tn contradiction to the ethnonym and perhaps 4150 against the social feelings of the people concerned, there appears to be a tenth dialect of MiJikenda called Chaka, It is spoken on the peninsula of Shimoni. Dialectotogically it shows closer Links with Digo and Durusa, It differs, however, from the Misikends dialects by sharing a considerable amount of the Jexical inventory with the southern group ef Kenyan SwahiTé (see sections 3.2.3, and 4.2.2.) 2.2.3, The Pokomo dialects The third group of Bantu dialects in the region is the Pokono dialect cluster in the river oasis of the middle and ower Tana. The people thenselves distinguish between “Lower Pokono" and "Upper Pokona". Dialectotogically this grouping ts Justified in so far as both groups can be defined by bundles of shared isoglosses. However, whereas the six units of Lower Pokono i.e. Kalingi, Buu, Ozunza, Kulesa, Ngatana and Mwina Gan be referred to in terms of dialectal varieties, the inter nal diatectal relationship of the Upper Pokomo idions is markedly remoter. We can distinguish three units, namely: the Néura-Ndera dialect cluster consisting of the sub-units Néuré Kinakonba, Gwano ané Kéera, and the two independent dialects 268 rumen 9.6. wonuxe Zubaki and MiTalulu. 2.2.4. Hana In the Tane river valley north of the Pokomo, there is an- other Bantu language called rivana.? ay their Pokono-specking neighbours the Ilwana are called ualakote. This name has a pe- Jorative connotation and should therefore be avoided. 2.3, The dialectometrical structure of the dialects 2.3.1, The structure of internal relationships of Swahi1i In this section, 1 shall present the results of both types of dielectonetrical analysis, lexical and phonological, «ith regard to the internal relations of the Swahili dialects. In the following section 2.2.2., the corresponding dialecto- metrical indices for the external relations of the Swahili dialects will be presented: A survey of the internal retations based on lexical dia- lectometry is given in diagram no. 6. The corresponding phono Vogical indices are presented in diagram no. 7. The figures in the diagrams indicate percentages with reference to the Tin guistic material tested. the word aaterial for the lexical die- Tectonetrical analysis consists of 2 100-word list which was 2150 used by the international working group for the testing and developing of the dialectonetrical method at Ivry in the years 1981-1983 (Guartsma and Minlig forthcoming). For the Phonological dialectonetrical analyses, all phonemes which are interrelated within the frenework of Guthrie's Connon jantu (Guthrie 1967-71) were taken into account. secause of Vimited space, only the results can be given here. The full saterial will be published within the Dialect Attas of the Coastal and Teita Groups of Kenyan Gantu (Néh1ig forthcoming). wet 3.6, Ome rut swauzz DIALECES OF REA PE SwaeEL DIALECES OF KENTA an On the basis of these Indices, there are several dialectal groupings possible, depending not only on the nature of dia- lectonetrical results, whether phonologically or lexically bas- ed, but also on the choice of the nuclei of the groups. Some of the dialects, like Anu and Mvite-donvu, cen even be made men- bers of different dialectal groupings at almost the sane levels of dialectal relation. The diagrams no.8 and no.9 represent this State of dialectological constellation. Whereas all SwahiTs dia~ lects can be grouped together lexically at a level of relation Ship of 77%, the same grouping can only be achieved at 2 level Of relationship of 50x, if the scale of the phonological indices 15 used. The historical factors which sey be responsible for this picture will be discussed in section 2.4. of this article, Dialecta grouping of Swahili on a lexical basis sm senate nmin Sesh _ an ar BO eg —$-ars » a 7 Diagram no. 8 i ' a2 rumen 3.6. womxe Diatectal grouping of Swahili on phonological basis =, ToS + 1 %0 F876 a @ ee © Diagran no. 9 2.3.2. The externa) group relations of Swahili The dialects which have always been looked at as the closest relatives of Swahili are the Pokono and the MiJikenda dialects. Because of their geographical nearness, I also include tere ITwana and Chaka as partners of comparison. The external group relations Of the Kenyan Swahili dialects within this frame of Kenyan Bantu areas follows (The symbol “I" means lexical éialectometrical index and the symbol *p" means phonological dialectonetrical index Dialectal relationships of Tikvu Hwang 1:53 ps a9 N. = Pokono 1 + 63/84 p+ 50-57 (Zubaki: 1 : 62, 9: 57) S. = Pokomo en ps 58 N. = wijikenga 1: 67 p + $4-59 (Giryama: p : $9) S. - Migikenga 1 : 66-70 p : 52/54 (Digo: 1: 70, p : 54) Chwake len ps2 Dialectal relationships of Siyu Tana 1:80 pial N. = Pokomo 1: 63/64 ps 37/39 5. = Pokome 1: 69 pi: a6 No = WiStkenda 1: 67 ps 48/50 S.- Nidtkenda 1: 66-71 p + 46/48 (Digo: 1: TL, p+ 46) Chwake ven ps 46 Diatectal relationships of Pate Nwana vest pal WN. = Pokore 1: 62/64 ps 37/39 5. = Pokome 1: 68 ps a6 N= Migikenda 1: 68 ps 48/50 S. = Migikends 1: 67-72 p + 46/48 (Digo: Ys 72, p= 46) Chwaka lin ps a6 Diatectal relationships of Anu Tiwana 1: st pr ae N. = Pokano 1: 65/66 ps 35/37 am veruenn 3.0. wenr0 5. ~ Pokano lin pias N= Migikends 1: 71 ps 46/48 S.- Wijikenda 1: 70-75 p : 48/46 (Dig: 1: 75, p + 44) Chwaka in pe Diatectal relationships of Mvita and Jonvy Tiwana 1253 ps 48 N. = Pokono ea) ps 44/46 5. ~ Pokono 1:70 p82 N. = Midikenga 1: 71/72 ps 4-57 (Giryana: 1: 71, ps 57) S. + Mijikenda 1: 72-77 p : 52/54 (Digo: 1: 77, 9: 82) Chwaka 1: 8182p: 4B Diatectal relationships of chi fundi Tiwana 50 1 pits N. = Pokomo 1: 64/65 p= 80-54 (Mitatutu: 9 : 54) S. = Pokome 1: 69 ps9 Ne + Wigtkenda 1: 70 ps 65/67 S. = Mijtkenda 1 71-75 ps 72 (Digo: 1: 75) Chwaka 1:82 p70 Dislectal relationships of Yumba Tiwana dase ps N, = Pokomo 1: 66 pile S. = Pokome yin ps 63 No + Wijtkends 1: 72 pt 65/67 S. > Misikends 1: 72-75 p : 67-70 (Digo: 1: 75. p Chwaka 1: a8 p67 2.4, Discussion of the divergent diatectonetrical indices 2.4.1, PreVininary remark If the phonological and the lexical dialectonetrical values 4iffer, the phonological values are either higher than the 1ex- ical values or lover. In most cases, lower phonological and higher lexical values are indicators for lexical borrowing in

You might also like