An Encyclopaedia of Language_10 pdf

maxən/ ‘do’ (cf. German machen) found as far north as the area round Berlin and /tsvaɪ/ ‘two’ (cf. German zwei), with the dental affricate, in a large area east of Hanover. Similar phonological variation is naturally also found in the dialects and accents of English. The Linguistic Atlas of England (LAE) and the Atlas of English Sounds (AES), based on the Survey of English Dialects (SED), show the distribution of a number of familiar geographical variants in the English dialects (cf. Figure 52). Thus, in the traditional dialects and also the accents south of a line running approximately from the Severn to the Wash, the vowels of but and butcher are distinct, as in the RP pronunciations /bʌt/ and /bʊtʃə/, as the result of a phonemic split in the original Middle English short /ʊ/ which took place in the Early Modern English period. By this change, the vowel was unrounded and centralised in many words, but retained unchanged in others. It was, however, not adopted in most dialects and accents of central and northern England, where one typically finds the same phoneme, characteristically /ʊ/, in all these words, e.g. /bʊt/ and /bʊtʃə/. Similarly restricted to the dialects and accents of southern and eastern England is the lengthening of original short /æ/ before fricatives and nasal clusters in some words, e.g. in grass, laugh and grant. South of the isogloss shown on Figure 52 these and similar words are pronounced with a long vowel, which in RP, which has adopted these forms, is typically a low, back /ɑː/, e.g. grɑːs/, laːf/, /grɑːnt/, phonemically distinct from the short /æ/ of, e.g., /gæs/ gas or /kænt/ cant. This development did not occur in the dialects and accents of the north and midlands, which have retained the original short vowel, usually with a more central quality than is usual in RP, and there /grɑs/, /lɑf/ and /grant/ have the same vowel as /gas/ and /kɑnt/. In fact, the situation to the south of this isogloss is not uniform over the whole area. Although in most south-western accents and dialects the vowel of grass, etc., is usually long, it is doubtful, as Wells (1982:345–57, especially 353–5) points out, whether there is a phonemic distinction between this and the vowel, of e.g., gas, since it is characteristic of this area that many originally short vowels have been lengthened. However, the border between this region and the area of south-eastern England where there is a clear phonemic opposition between /æ/ and ɑː/ has still to be established since this matter has not yet been investigated in sufficient detail. In traditional dialect geography, as exemplified by the maps of the DSA and the LAE, and of most other linguistic atlases, phonological variation is thus shown in terms of the differential development of certain historical phonemes as shown in the words elicited through the respective questionnaires. What is mapped, in effect, is the variation of isolated sounds in individual words. This emphasis on isolated phonetic differences seen from a historical perspective was challenged, from the point of view of modern structural linguistics, by Weinreich (1954), although similar ideas had already been voiced by Trubetzkoy (1931), cf. Petyt (1980:102–7 and 117–31). He suggested that such an atomistic approach was incapable of Figure 52 Vowel isoglosses in English (after LAE and AES) 502 LANGUAGE AS GEOGRAPHY identifying significant geographical distinctions in a systematic way. What was important were not necessarily differences in phonetic detail, but differences between the phonological structuring in different dialects. In terms of phonological structure, then, dialects may differ first in the number of phonemic contrasts they draw. Thus, the dialects and accents of northern England are distinguished from those of the south in that, as we saw earlier, they lack a distinction between /ʌ/ and /ʊ/ which is characteristic of the south and thus have one vowel fewer in their phonemic inventory. In Weinreich’s view, the important boundary is not that of any individual lexical items, but the one dividing those dialects which possess this phonemic contrast from those which do not. On the other hand, the distinction between long /ɑː/ and short / æ/ in words like grass may be considered relatively insignificant, since it does not affect the number of phonemes, but only the incidence of particular phonemes in particular words. Indeed, what now appears significant in terms of phonological structure is the possible lack of such a distinction in the western dialects, about which, as we have seen, the LAE is unable to give us clear information (cf. Trudgill 1983:36–7). There may also be significant differences between dialects not simply in respect of the phonemic inventory but also in terms of the types of relevant contrast. Thus, we may distinguish in general terms between those German dialects which have a set of front rounded vowels and those which do not (cf. Figure 53). Furthermore, this distinction is purely synchronic, since these front rounded vowels derive from a number of historical sources, from the very old conditioned fronting process known as Umlaut, as in most Low German and Swiss, from the vocalisation of postvocalic /l/, as in Austrian, or from a spontaneous, unconditioned palatalisation of original back vowels, as in Alsace; cf. Keller (1961:123–30), Lüssy (1983) and Wiesinger (1983b). Phoneme distribution may also be subject to areal variation. Goossens (1969:43–5 and 51–2) gives a striking example of this within a relatively small area, showing no fewer than twenty-seven areas with different types of restriction in the occurrence of /ɑː/ in the dialects of the Belgian province of Limburg, ranging from one where this phoneme is found in all phonetic environments except before /r/ and /j/ to one where it only occurs before /r/, medial /v/ and final /f/. In the 1960s and 1970s a number of notable attempts were made to account for geographically determined phonological variation in terms of the model of generative phonology, notably Newton (1972) for Greek, Thomas (1967) for Welsh, Vasiliu (1966) for Romanian and Veith (1972) and (1982) for German. They are reviewed in detail by Chambers asnd Trudgill (1980: 5–50), Francis (1983:171–92) and Petyt (1980:171–85). Underlying these studies is the assumption that the dialects of the same language could all be considered to possess the same underlying systematic phonemic inventory, and geographical differences could be accounted for in terms of variation in the number, order, scope or form of the rules by which surface forms are derived from these underlying forms (cf. Agard 1971). Figure 53 Front rounded vowels in German dialects (after Wiesinger 1970, 1983b and Kranzmayer 1954) AN ENCYCLOPAEDIA OF LANGUAGE 503 A typical example of this approach is provided by the account in Rein (1974) of the vocalisation of the liquids in the Central Bavarian dialects of German. Here, historical /l/ and /r/ in pre-consonantal or final position, as seen in standard German words like Holz ‘wood’, viel ‘many’, nur ‘only’ and Dorf ‘village’, are vocalised to give forms like [hoɪdṱ zṱ], [fʊɪ], [nuɐ] and [tɔɐvṱ ], for instance, in the dialect of Grossberghofen near Munich, cf. Gladiator (1971). There are significant regional variations in the phonetic results of this vocalisation, and Rein shows how, in the case of original /l/, they can be accounted for in terms of variation in the application of seven rules, which occur in seven distinct configuration types within this area. Three rules—to palatalise underlying /l/ to /ʎ/], to round the preceding vowels and to vocalise the [ʎ] to [ɪ]— are common to all the dialects except for some peripheral ones which lack the third of them, the vocalisation rule, and thus form a transitional belt. The other four rules, involving, for example, the unrounding or diphthongisation of the preceding vowel, then differ areally in their application, giving the geographical divisions shown on Figure 54. A generative phonological approach can show dialect differences in a way which demonstrates their close relationship in a quite explicit way, with the varieties involved differing slightly in the number and type of rules, and in the scope and ordering of them, but not in their underlying structure. However, doubts have been raised regarding its appropriateness in an account of geographical variation in language, cf. Chambers and Trudgill (1980:47–50), Löffler (1980:86–8) and Petyt (1980:182–5). Particular difficulties are caused by ‘persistent’ rules which present problems of ordering (cf. Newton 1971), by variations in lexical incidence, as in the case of English /æ/ and /ɑ:/, which is not amenable to statement in terms of phonological rules, and by the problems surrounding the establishment and status of the underlying representations. For, despite the insistence by its practitioners that the rules of a generative phonology are actually synchronic, many generative accounts of dialectal variation in terms of differences in the rules by which surface forms are generated from common base forms are in practice indistinguishable from traditional statements of the differential application of historical changes on common ancestral forms; such is indeed the case with Rein’s (1974) study of the vocalisation of liquids in Bavarian. However, more recent work, e.g. Glauser (1985), is concerned to evade such problems. 2.2.2 Morphology and syntax Morphological variation in language involves geographical differences in the systematic exponence of morpho-syntactic categories. By and large, it has been studied less exhaustively than phonological variation—and variations in derivational morphology have very rarely been investigated, with the exception of relatively simple cases like the diminutive suffix in German: cf. DSA (map 59), Besch et al. (eds 1983:II, 1250–5) and Toefenbach (1987). However, most of the major dialect Figure 54 Vocalisation of l in Bavarian German dialects: dialect areas identified through phonological rules (after Rein 1974) 504 LANGUAGE AS GEOGRAPHY geographical projects provide a certain amount of information, if it is largely restricted to isolated forms which are easily elicited. An important example of regional variation in the exponence of morphosyntactic categories is provided by noun plurals in the Romance dialects and this is commonly taken as a diagnostic feature dividing Gallo-Romance (French, Occitan and Rhaetic) from Italo-Romance. In the former, as in Sardinian and some Alpine Italian dialects, plurality is signalled by a suffixed morph -s, cf. French chiens, Sardinian canes, Rhaetic tgauns ‘dogs’. In Italian, on the other hand, plurality is commonly marked by a change in the final vowel, as in cani (sing. cane), cf. Devoto and Giacomelli (1974), Posner (1966:59– 63) and Wartburg (1950:20–31). These differences resulted from alternative directions of analogical levelling when the Latin case system was abandoned. Whilst in Gallo-Romance and Sardinian the original accusative plural suffixes were generalised as a marker of plurality, in Italo-Romance nominative plural forms were selected. It is thus a difference of some antiquity. Most interest in morphological variation has usually been given to the survival of historical inflectional classes or types no longer current in the relevant standard language. The SED provide numerous examples of this in English, as for instance in the case of the plural marker -(e)n, which in standard English is now only found in one lexeme, i.e. ox—oxen. Regionally, though, a significant number of other words have retained this suffix, e.g. een ‘eyes’ and shoon ‘shoes’ in northern English and Scottish and housen ‘houses’ in the West Midlands and East Anglia, cf. LAE (maps M60–64), Aitken (1984:104) and Wakelin (1977:109–11). On the other hand, many dialects have undergone analogical levelling in the paradigms of numerous words which have historical irregularities in standard English. Thus, past tense or past participle forms like catched ‘caught’, knowed ‘knew/known’ and stealed ‘stole(n)’ are widespread in English dialects, cf. LAE (maps M51–55) and Wakelyn (1977: 122–4). Regional variation in the systematic exponence of morpho-syntactic categories tends naturally to involve much larger areas and a greater degree of consistency than, for instance, the geography of isolated forms. Thus, the German dialects vary considerably in the number of case distinctions drawn in nouns and pronouns, cf. Shrier (1965) and Koß (1983). We may illustrate this from the systematic exponence of case in the masculine definite article, where standard German has the four distinct forms, nominative der, accusative den, dative dem and genitive des. The genitive has been lost in all German dialects as a morphologically distinct case, but very few dialects, mainly in Swabia and Westphalia, retain distinct forms for the remaining three. In much of the west, including Switzerland, there are two forms, a dative and a common nominative/ accusative. The north and east, on the other hand, have a nominative and a common objective case with syncretism between accusative and dative. Finally, a small area near the Dutch border has lost all case distinctions, in common with the dialects of Netherlands and Belgium. However, as Shrier herself makes clear, the exact extent of this important morphosyntactic variation is extremely difficult to ascertain either from the DSA or published dialect surveys, as traditional dialect research has tended to concentrate on individual forms and has rarely been concerned to examine complete grammatical systems. A rare exception to this is the study by Panzer (1972), which gives an account of the geographical structure of the Low German speech area purely on the basis of systematic morphological differences. Less difficult to isolate are instances where there is variation in the presence or absence of particular morpho-syntactic categories, although, as with the distribution of case systems in German dialects mentioned above, they have not always been fully surveyed. However, it is known that, for example, the dialects of German are bisected roughly at the river Main by an isogloss which marks the southern limit of the simple past tense, cf. DSA (map 79). Whilst the northern and central dialects retain both the simple past and the perfect, as does standard German, e.g. ich stahl ‘I stole’ and ich habe gestohlen ‘I have stolen’, in the south only the perfect is found. There has been a significant degree of theoretical interest in recent years, within the model of ‘Government and Binding’ (noted in Chapter 4, section 4), into the nature of ‘empty’ categories and so-called ‘pro-drop’ languages. In these latter, unlike in English or French, there is no requirement for an explicit subject in all tensed clauses and the categories of person and number may be fully expressed through verbal inflection rather than through pronominal forms. This is the case in Italian and Spanish, cf. Italian compriamo or Spanish compramos ‘we buy’—and also in the Occitan dialects of southern France by contrast to the northern French dialects and standard French, cf. Jochnowitz (1973:129–32). It is possible that a study of this areal distinction could provide material of some interest to current syntactic theory, but, as yet, it has not attracted attention. Indeed, areal syntactic variation has been little investigated in general, although it is by no means unusual. To some extent this is due to the relative neglect of syntax in linguistic theory until fairly recently, but there are also practical problems for the field-worker in eliciting the longer stretches of text or the extensive amount of material necessary for a large-scale systematic survey of syntactic variants. Nevertheless, as Kirk (1985) has pointed out, there is a not insignificant body of data on regional variation in English syntax in the material of the SED, and he reviews a number of recent studies, many of which, however, are as yet unpublished. As an example of such variation we may cite the differences in English in the order and form of direct and indirect object, where the alternatives give it to me, give it me and give me it are found in different relatively discrete areas of England: cf. Figure 55, based on LAE (map Sɪ) and Kirk (1985:131–5), where these forms are discussed in a historical context. AN ENCYCLOPAEDIA OF LANGUAGE 505 2.2.3 Lexis Geographical variation in the lexis is by contrast relatively well documented for most major linguistic communities. Glossaries of local lexical items were compiled in a number of countries from a relatively early date, cf. the account for England in Wakelin (1977:43–6) and many modern publications of a popular nature take the form of lists of local words. On a more systematic level, there are numerous large-scale dictionaries of the dialects of larger areas, cf. for Germany the survey of such work in Friebertshäuser (1986), and a number of atlases devoted largely or entirely to word geography, e.g. Kurath (1949) (eastern North America), LAS (Scotland), DWA (Germany), WGE (England) and Thomas (1973) (Wales). Furthermore most atlases covering the Romance language areas, including ALF (France), AIS (Italy) and some of the more recent regional atlases of France like ALMC, together with the RNDA for Dutch, take the form of raw lexical data given in phonetic transcription and are thus primarily orientated towards the presentation of lexical variation. Of all levels of language, the vocabulary is perhaps most subject to small-scale local variation, since, unlike the phonology, it is not a closed system and words may be adopted, transferred or replaced without major consequences for the linguistic structure or hindrances in communication. Thus, particularly in the field of domestic or agricultural words, which are mainly used exclusively within a relatively small community, traditional dialect may exhibit a great deal of variation in a limited geographical area. A typical instance is seen in the eighty-eight equivalents for ‘left-handed’ reported in the SED. It should be noted, though, that regional variation is not confined to what might be considered ‘traditional’ realms of the vocabulary; there is significant variation in the German speech area in the dialect equivalents for relatively recent innovations such as the potato, which was not cultivated widely until the eighteenth century (cf. Martin 1963 and DWA (map XI, 4–5)) and the match, which was not invented until the nineteenth century (cf. DWA (map III, 27)). Naturally, geographical variation in the lexis, as in other linguistic levels, results from the differential spread of historical change, in this case principally from divergences in the adoption of lexical innovations. In the main, the scholarly study of word geography has tended to concentrate on forms which are historically interesting in some way. Many locally restricted words are archaic survivals in the sense that whilst they are no longer current in the standard language they have been preserved in Figure 55 Direct and indirect objects in English dialects (after Kirk 1985 and LAE) 506 LANGUAGE AS GEOGRAPHY particular regions, typically in peripheral areas which have resisted innovations from a dominant cultural centre. Thus WGE (map 2) shows that the standard English terms cow-house and cow-shed, which are both relatively recent formations (cf. Wakelin 1977:71–2), are found in the dialects of most of southern and central England, whereas shippon, which goes back to an Anglo-Saxon form, is now restricted to two areas, one in the south-west (principally Devon) and one in the north-west (principally Cheshire and Lancashire), from whence it has spread into the English of North Wales (cf. Parry 1985:58–9). These are clearly the relics of what was once a much more widespread distribution of the word, and older sources suggest that it was once current over a continuous area of western England from most of which it has now been displaced. By its nature, the vocabulary also reflects innovation due to cultural influence or migrations of population more clearly than, say, the phonology. This is particularly true in the case of lexical loans, as for example in the well-known distribution of Scandinavian words in the English dialects. Although some Scandinavian words, such as sky, take and the pronouns they, their, etc., were adopted universally in English, many other Scandinavian words are found characteristically in the local dialects of northern and eastern England, reflecting the density of Scandinavian settlement in this area; cf. Figure 56. The WGE gives maps for a good number of these words, and further information is also to be found in Kolb (1965), Wakelin (1977:130– 8) and Weijnen (1978:63–8). In some instances, such lexical regionalisms may provide our sole evidence for certain links. In the Bavarian dialects of German we find the words Ertag ‘Tuesday’ and Pfinztag ‘Thursday’, which are commonly assumed to derive ultimately from Greek Areōs (hēmerā) and pemptē (hēmerā). It is now generally, if not universally, accepted that these forms reached the Bavarians through Gothic intermediaries, possibly missionaries, although there is no historical documentation of such contact (cf. Wells 1985:56). The study of word geography has concerned itself primarily with identifying the geographical distribution of regional equivalents of isolated lexemes of the relevant standard language. However, this atomistic approach is not wholly satisfactory, as it is now recognised that there is a significant structuring in the lexicon and that regional variation may not consist simply in the use of different words but in differences in how lexical subsets (or ‘semantic fields’) are structured (see Chapter 5). This was clearly shown in a number of studies based on the material of the DWA, surveyed in Goossens (1969:70– 5) and Reichmann (1983a), as for example in Durrell’s (1972) investigation of the equivalents in the German dialects for standard German warten ‘wait’. Previously, interest in this lexical set had tended to concentrate on the retention of the archaic Figure 56 Scandinavian words in English dialects AN ENCYCLOPAEDIA OF LANGUAGE 507 survival beiten, which is the main equivalent in medieval German but which now persists only in a few discontinuous relic areas in Austria and Switzerland. However, in a number of large areas the DWA material reported two or more equivalents with almost equal frequency, e.g. warten and lauern in the centre and the east and warten, passen and belangen in the southwest. These are in no way alternative synonyms, but reflect the fact that the dialects possess semantic distinctions not found in standard German and thus a different structuring of this lexical subset. Thus, in the southwest, warten is a relatively neutral term, whilst passen stresses the duration of the activity and belangen presupposes a degree of impatience. Nevertheless, the description of regional variation in the structuring of lexical fields presents extreme difficulty for a number of reasons. Much of the available material has been collected on the assumption of one-to-one semantic correspondences between dialect lexemes and those of the standard language and cannot provide a full view of the structural incongruities which may exist. Furthermore, constructing questionnaires by means of which one might elicit all the relevant material is problematic if one is not fully aware beforehand of what different types of regionally restricted semantic distinction might be present. Given these difficulties and the fact that the systematic investigation of the lexicon is a potentially vast task, it is not surprising that, although it is now widely accepted that the study of regional variation in the lexis must now take account of differences in its structure as well as of differences in the individual lexical items, most work in this area has been confined to small geographical areas, e.g. Zürrer (1975), or to small lexical subsets, as in Durrell (1975), Francis (1983:22–6), Goossens (1969:75–114) and Löffler (1980:110–17). 3. THE PROBLEM OF DIALECT BOUNDARIES One of the major aims of the DSA was to establish the geographical limits of the dialects of German, cf. Wiegand and Harras (1971) and Knoop, Putschke, Wiegand (1982). We saw in section 1 that our inherited terminology is based on a hierarchical model, in that it assumes superordinate ‘languages’ each divided into a number of ‘dialects’ which are current in specific and identifiable regions of the territory of the particular ‘language’. However, just as it has proved impossible to devise objective linguistic criteria by which one might identify the boundaries of individual ‘languages’ within a geographical continuum, so there is no agreement on criteria for the determination of boundaries of dialects. As Jochnowitz (1973:30–1) puts it: There has never been a definition of “dialect” that enabled us to determine whether two given idiolects belong to the same dialect, or, for that matter, to the same language’. An unresolved controversy has existed since the beginnings of systematic research into dialects. On the one hand we have those scholars who stand in what we might loosely call the German tradition who assume the existence of discrete dialects and regard it as one of their primary concerns to identify their boundaries, such as Bach (1950:56–73), Reichmann (1983b), Schirmunski (1962:132–3) and Wiesinger (1982 and 1983a). On the other, there are those who deny that such boundaries can exist, given that linguistic transitions across space from one region or even village to the next are gradual and imperceptible. This view was expressed as early as 1875 by Paul Meyer and 1888 by Gaston Paris, who was of the opinion that ‘il n’y a réellement pas de dialectes …’ (cf. Bach (1950:58) and Francis (1983:50–8), and many scholars have followed them in rejecting entirely the notion of dialects as relatively unified language varieties used by discrete groups within a well defined geographical area, although in some cases the theoretical standpoint may be significantly different, as with Bailey (1973, 1980) and Bickerton (1973). The crucial problem is that although speakers characteristically make use of regional labels for their particular speech forms and refer to them, for instance, as Venetian, Bavarian, Gascon, etc., and assume that these exist as relatively coherent entities, it has proved inordinately difficult to correlate these designations unambiguously with a discrete set of linguistic features. Ferdinand Wrede, Wenker’s successor, proposed a division of the German speech area into major and minor dialect regions and the results were published as map 56 of the DSA (cf. Kurath (1972:83–6) and Wiesinger (1982)). His method was to base each boundary on a single isogloss, which he took to have the function of a diagnostic indicator. Thus, the boundary of the Bavarian area is given by the geographical limit of the form enk ‘you’ (acc. pl.) as against euch or ui found in the adjoining regions. Similarly, he divided the Low Saxon area of Low German into West Low Saxon and East Low Saxon on the basis of whether the plural of the present tense of verbs was realised through the morpheme -(e)t or -(e)n. Wrede used phonological, morphological or (less frequently) lexical isoglosses without distinction and naturally did not take structural considerations into account. Selecting the isogloss of an isolated linguistic feature and ascribing diagnostic significance to it in the definition of linguistic areas may serve for an initial orientation, and Wrede’s success in this may be measured by the fact that most handbooks and accounts of German dialects have based their accounts of the German dialect regions on Wrede’s subdivisions. Nevertheless, most scholars would now regard this procedure as unsatisfactorily arbitrary. Some of Wrede’s underlying assumptions, as, for example, of the relative significance for the subdivision of the German dialects of the isoglosses stemming from the High German consonant shift (cf. Figure 51) are nowhere made fully explicit or justified, and such a procedure will naturally run the risk of making the a priori assumption that, say, such an entity as ‘Westfalian dialect’ exists (perhaps because speakers use such a designation) and then finding an isogloss which most closely corresponds to this assumption. In fact such popular designations will most usually derive not from a perception on the part of speakers of diagnostic 508 LANGUAGE AS GEOGRAPHY linguistic similarities but from the association of the speech form with a particular geographical or political territory. As with languages, the designation of dialects may rely more on political or cultural than primarily linguistic factors. Furthermore, the selection of a single isogloss to define a linguistic area is highly dubious from a purely linguistic standpoint. As Georg Wenker realised at a very early date (cf. Knoop, Putschke, Wiegand (1982) and Wiegand and Harras (1971)), isoglosses for individual words with the same phonological feature rarely coincide exactly. The isoglosses for the High German Consonant Shift (cf. Figure 51), provide many examples of this, but they are also found in the SED material for the split of Middle English u and the lengthening of a (cf. Figure 52), where, in the latter case, we see a broad belt where a is lengthened in some words only, so that we may find, say, [læst] but [grɑːs] in the data from the same informant. There is thus not a clear division between northern and southern English in respect of these features, but a fairly broad transitional band between ‘pure’ southern dialects, which have low central /ʌ/ and long /ɑː/ in a relatively uniform number of lexical items, and ‘pure’ northern dialects, which lack these entirely in the relevant words. Similarly, we also frequently find gradual phonetic transitions from one area to the next. This phenomenon may be illustrated from German-speaking Switzerland, where the SDS (maps I, 15–16) shows short [e] in numerous words in eastern Switzerland which in western Switzerland have [ε], so that, for example Zurich [setsə] corresponds to Bernese [sεtsə], cf. German setzen ‘to put’. In the central area between these two, intermediate qualities are found, so that there is a gradual progression from a closer to a more open realisation without one being able to identify a clear boundary of any kind. Isoglosses, then, as lines on a map, may be, as Markey (1979:103) says, ‘highly simplified representations of the geographic demarcation of linguistic differentiation’: cf. also Handler and Wiegand (1982). The basic notion of the isogloss as a linguistic boundary is open to doubt, since, as Trudgill (1983:47– 51) clearly points out, they quite erroneously suggest abrupt transitions from one form to another at a particular point in space. That this is not the case is confirmed by the evidence of the DSA and DWA, with their dense network of localities; Bach (1950:120–1) demonstrates how difficult it was in many cases to draw clear isoglosses from this material, as alternative variants were often current across a broad intermediate area, and Hildebrandt (1967) demonstrates the danger of establishing quite artificial lines in such instances. Many maps of the LAS show similar difficulties in respect of lexical data, with numerous words showing confusingly overlapping distributions. In practice, sudden discrete breaks are rare, and what one finds is a gradual transition from one form to another, whether one is dealing with phonological, grammatical or lexical features. Chambers and Trudgill (1980:125–42), in an interesting study of the borders between southern English /ʌ/ and northern English /ʊ/ and between southern English long /ɑː/ and northern English short /a/, identified a large region in the East Midlands with complex intermediate stages between ‘pure’ southern and ‘pure’ northern dialects. In the case of the low vowels, they found what they termed ‘mixed’ lects (mainly southern or northern, but with the ‘other’ vowel occurring at times), ‘fudged’ lects (with occasional half-long vowels as compromise forms) and ‘scrambled’ lects (with both ‘mixing’ and ‘fudging’). Although as yet there have been few extensive surveys along the lines suggested by Chambers and Trudgill, it is most likely that many other border areas will exhibit similar phenomena and reward investigation. A recent example, within the generative paradigm, is Glauser (1985). A traditional solution to the problems posed by the potential arbitrariness of defining dialect areas through the selection of isolated features has been to take all features for which one has data and identify boundaries where numerous isoglosses run together in a so-called ‘bundle’ of isoglosses. Some of these are extremely well-known, as for instance the border between High German and Low German, where a significant number of other features follow the same course as the isogloss for the High German consonant shift, notably and in the greatest concentration along the mountain ridge of the Rothaargebirge northeast of Siegen (cf. Figure 51). Following this principle, numerous regional dialect studies summarise their data on geographical variation within the area concerned in the form of ‘honeycomb’ maps, where the difference between the dialects of the individual localities is shown in terms of the size of the bundles of isoglosses between them. Such a quantification of the linguistic distance between localities appears to provide the most objective measure for the identification of dialect boundaries, but, as has been pointed out by many scholars, e.g. by Chambers and Trudgill (1980:112–20) and Löffler (1980: 134–9), serious problems still remain. There is first the often unremarked fact that such ‘honeycomb’ maps rarely present clear continuous boundaries. And, of course, the problem of potential arbitrariness, although reduced, may still be present, in that even such bundles will represent a selection of the total number of linguistic features exhibiting variance, as it is scarcely possible to take all into account. Perhaps more seriously, the question must be asked whether the individual isoglosses which are subsumed in such bundles are all of similar real significance, in other words whether some isoglosses ought to be ranked more highly than others, cf. Chambers and Trudgill (1980:112–20). There is no prima facie reason why, for instance, a single isogloss dividing two localities might not actually be more important than even a large bundle because it marks the boundary of a major linguistic feature, like, for example, the pro-drop/non-pro-drop border between French and Occitan. It is commonly assumed that the isoglosses of single lexical items are less significant than those of such major features, but we lack any widely accepted explicit and well justified metric for grading isoglosses, although there have been many attempts to develop one. Nevertheless, many scholars see quantitative measures as the most objective method for determining the degree of difference between geographical variants, and there has been substantial work again recently which aims at refining these procedures. Such methods, termed ‘dialectometry’ have been developed primarily within the Romance areas, as by Goebl AN ENCYCLOPAEDIA OF LANGUAGE 509 (1983, 1984 & 1987) and Séguy (1973), but Viereck (1985) has applied them to English and Thomas (1985) has presented work along similar lines on lexical differences within the Welsh dialects. However, the underlying problem that the same significance is thereby ascribed to all isoglosses, of whatever type, has still not been satisfactorily resolved. And, as we have already seen, the underlying assumption of all attempts to identify dialect boundaries on the basis of isoglosses that these represent discrete breaks in the spatial continuum is open to serious challenge. The problems are not wholly alleviated even if one moves away from the isoglosses of single features as the major criterion for identifying dialect boundaries. When structuralist linguistic theories were first applied to dialectology, it was hoped that these might provide a more objective method for the delimitation of dialect, cf. Ivić (1963) and Stankiewicz (1957). Thus, Catford (1957) showed how the Scottish dialects could be subdivided in terms of the number of vowel phonemes they possessed, with the dialects of the Central Lowlands, for example, having nine (though with variations in the inventory) whereas the peripheral regions of Buchan and Galloway have twelve, cf. also Aitken (1984) for fuller details. In a series of justly celebrated papers using the material of the SDS, Moulton (1960, 1961, 1963 and 1968) presented a survey of divisions in Swiss German dialects on the basis of similarities and differences in parts of their phonological systems, cf. also Goossens (1969:45–68). Subsequently, numerous other scholars have attempted to delimit dialect areas on the basis of shared aspects of phonological structure rather than on that of the isoglosses of isolated features. Thus, Panzer and Thümmel (1971) attempted to show the internal geographical structure of the Low German and Dutch dialects in terms of the extent to which inherited phonemic contrasts have been retained, and for the dialects of this region Teepe (1973) showed how large dialect areas could be established on the basis of the structural development of the inherited long mid vowels; cf. Niebaum (1980) for further details. Although most attempts to establish dialect divisions on structural criteria have been based on distinctions in phonological systems, Panzer (1972) presented a survey of divisions within the Low German dialects using differences in the inflectional morphology, albeit unfortunately without illustrative maps. In a rather similar way, linguists working within the generativist paradigm have attempted to establish dialect boundaries on the basis of differences in the rules by which surface phonological forms may be derived from a common underlying base. We have already discussed Rein’s account of the subdivisions in the Bavarian dialects which may be identified through their treatment of underlying liquids. Vasiliu (1966) demonstrated how, taking differences in the inventory and ordering of major phonological rules, two major dialect groups may be posited for Romanian, ‘Muntenian’ (in the north) and ‘Moldavian’ (in the south), and each of these may be further subdivided according to less prominent differences in rule configuration. It has been claimed, e.g. by Agard (1971), that the generative model can provide us with an explicit method for distinguishing the boundaries of languages from those of dialects. Whilst ‘dialects’ will merely differ in the constituency, inventory and ordering of rules by which surface phonological forms may be derived from a common underlying systematic phonemic base, as in the example given above, ‘languages’ will differ in the underlying base itself. Nevertheless, this claim is now widely regarded as too strong. As Francis (1983:175) points out, given the notorious problem of the lack of theoretical constraints on the abstractness of phonological rules, it would be quite possible to demonstrate that all the modern Romance languages and dialects are ‘merely dialects of a common language’. Agard (1971) himself, using the data from Moulton (1962), did indeed suggest that there were two Swiss German ‘languages’ on the basis of his criteria. More recent generative work, e.g. Glauser (1985), has moved away from these strong claims and been more sensitive to the problems of boundaries and the nature of dialect transitions. Clearly, divisions based on structural or generative criteria may avoid some of the potential arbitrariness involved in the selection of isoglosses of individual features to determine dialect boundaries, but even so, there is no fully clear theoretical reason why we might be justified in ascribing a greater degree of significance to them as postulated linguistic ‘frontiers’ than to any individual isoglosses. Nevertheless, many dialectologists still regard Gaston Paris’s view as too extreme and remain convinced that there are geographical groups which exhibit such a significant degree of uniformity in their linguistic behaviour as to justify the attempt to identify them. Nevertheless, the precise delimitation of such groups is not possible solely in terms of a particular isogloss or group of isoglosses; rather, one has to be satisfied with establishing areas of relative uniformity on the basis of sets of shared linguistic features which are not necessarily exclusive and distinguishing these from areas which exhibit a lesser degree of coherence. In practice, dialectologists have long followed this insight and spoken of ‘focal areas’ (‘Kernlandschaften’) and ‘transitional areas’ (Übergangs- or Saumlandschaften), as Bach (1950:60–3) and 114–6)) and Petyt (1980:60–1), but there have rarely been satisfactory attempts to establish these systematically for a large speech-area, perhaps in part at least because of the established tradition of representing geographical variation exclusively through abrupt lines on a map. However, Wiesinger (1983) has proposed a new systematic subdivision of the German dialects on this principle on the basis of a large number of structural phonological and morphological features. There are few clear-cut boundaries, with the exception of the divide between Low German and High German, but he was able, using a uniform methodology and uniform data, to subdivide the whole speech-area into areas of relative homogeneity separated by zones of transition. In many cases, particularly in the west and centre, his results differ significantly from earlier proposals for German, e.g. that by Wrede in the DSA (map 56), which, as we have already seen, were based on assumptions which were not always fully explicit or purely linguistic. The result is an interesting and plausible account which gives us clear insights into 510 LANGUAGE AS GEOGRAPHY the geolinguistic structure of the German-speaking regions without the distortions inherent in divisions based simply on particular selected isoglosses. Similar methods could clearly be used, for example, to establish the divide between southern and northern English, together with the transitional zone in the Midlands, from material such as that presented in Chambers and Trudgill (1980:125–42) and discussed above. 4. ISOGLOSSES: CONFIGURATIONS AND EXPLANATIONS 4.1 Isogloss patterns If geographical variation results from the differential adoption of linguistic changes, then the isoglosses which represent— albeit, as we have seen, in a relatively abstract way—the limits of particular variants reflect the dynamics of change within a speech area. Isoglosses take on particular recurrent patterns through which we are able to identify processes of diffusion of linguistic innovations from ‘active’ areas, where such innovations have arisen, into ‘passive’ recipient areas; cf. Bach (1950: 146–57) and Markey (1977:10–17), if it is not the case that such diffusion necessarily presupposes a single innovating source within the ‘active’ area, cf. Kurath (1972:131–6). In the history of particular languages, different regions may have been active and passive in turns at different times, and these fluctuations may all have left their mark on the present-day geography of the language. This is naturally rather the case for, say, German and Italian, where large centralised political entities emerged at a late stage, than for English and French, where the region surrounding the capital cities has for centuries formed the major or only ‘active’ area. Isoglosses in the form of a circle present clear instances of diffusion from an innovating centre. These are frequently associated with large urban centres (which are typically ‘active’) whose linguistic innovations have penetrated the surrounding countryside. Thus, Wiesinger (1970: II, 132, 148) shows the area round Vienna with a monophthong /a/ where other Austro-Bavarian dialects (and standard German) have the diphthong /aɪ/ in words such as weiß ‘(I) know’. In the case of older innovations which have spread more widely, the circle may develop into a more distorted shape as the diffusion proceeds irregularly due to particular factors which for various reasons have restricted or accelerated its progress in certain areas. This is the case with numerous innovations originating in Paris which have been diffused through the northern French dialects, cf. Gluth et al. (1982:489). Over longer periods, the gradual dissemination of such innovations outward from a particular centre may reach the borders of the speech-area, leaving discontinuous relic areas on the peripheries or as isolated circles in particular regions which have been ‘by-passed’, as it were, by the innovation. Thus, Jaberg (1908), using material from the ALF, showed how the original Latin initial /sk/, as in scala ‘ladder’, which in standard French and most French and Occitan dialects has a prothetic vowel, e.g. French échelle, lacks such a vowel in six disconnected remote regions, five on the periphery of the speech area and one in a remote region of the Massif Central, cf. also Petyt (1980:61–2). Relic areas of this kind are typically found on the borders of most larger speech-areas; e.g. for Dutch, western Flanders and Limburg, for German, the Alpine regions, and for English, the southwest and the north. Conversely, an isogloss in the form of an isolated circle may be an enclave resulting from the adoption of an innovation from a more distant area. Typically, this occurs in cases where an urban centre has accepted a feature from an ‘active’ area across intervening rural territory which is more conservative in its speech-form. Thus, the city dialect of Berlin has a number of characteristics which correspond to those of the High German dialects some distance to the south, e.g. haben ‘to have’, whereas the surrounding rural dialects retain Low German forms, e.g. hebben (cf. König 1978:140). Indeed, in the present century, a number of studies in different countries have shown innovations being transmitted solely or principally between urban centres, from each of which they may then be diffused outwards into the surrounding rural areas; see Moser (1954) for Swabia and Trudgill (1983:64–87) for East Anglia and southern Norway. There are a number of other recurrent configurations of isoglosses which reflect the irregular adoption of innovations as they have penetrated furthest into areas which, for a variety of reasons, may be more receptive to them. These may be illustrated from Figure 57, which shows how a number of southern features have pushed northwards along the river Rhine, giving wedge-shaped or pipe-shaped patterns in their modern distribution. As Wagner (1927:59) pointed out, we may draw the general conclusion in respect of such isogloss patterns, which occur repeatedly on the maps of the DSA, that the initial penetration is relatively narrow—in this case along the river valley—but it may subsequently broaden out into a wedge as the innovation is adopted in the lateral regions. Eventually, such a wedge may reach the limits of the speech area, leaving relic areas on each side. Jaberg (1908) shows how this is the case with the northern French form for ‘it is necessary’ (standard French il faut, deriving ultimately from Latin fallit) which has penetrated along the Rhone valley as far as the Mediterranean (cf. Petyt 1980:62–3), with what we may assume to be the original Occitan forms, deriving mainly from Latin calet ‘it is hot’, now restricted to areas to the east or west of the valley and inland from the Coṱ te d’Azur. Wedge-shaped patterns may also arise if AN ENCYCLOPAEDIA OF LANGUAGE 511 [...]... languages 2.5 Indo-European Languages The more easterly part of south-western Asia and the northern part of South Asia are dominated by languages of the Indo-Iranian branch of Indo-European, which also stretches into Central Asia and has outliers in Sri Lanka and the Maldives; the boundary between Iranian languages (to the west) and Indo-Aryan languages (to the east) runs through Pakistan The Iranian... department of Reunion, giving a total of over 1 million for Indian Ocean French Creole Spanish is the language of the Canary Islands and of the two Spanish enclaves on the Moroccan coast, Ceuta and Melilla 3.6 Austronesian languages Malagasy, the language of Madagascar, is the westernmost extension of the Austronesian language family; it has over 8 million speakers 4 OCEANIA (For explanation of the terms... the official and national language of Bulgaria The major expansion of Slavonic languages outside Europe has been the expansion of Russian to other parts of the USSR, especially northern Asia (see section 2) 1.1.5 Other Indo-European branches of Europe Only two languages of the Baltic branch survive, Lithuanian and Latvian, both national languages of the corresponding constituent republics of the USSR,... Etruscan from northern Italy and Iberian from the Iberian peninsula) The Romance languages remain dominant in south-western Europe, with Rumanian as an outlier in south-eastern Europe, but the spread of, especially, Spanish and Portuguese to the Americas means that speakers of Romance languages in Europe are now outnumbered by those in the Americas The Romance languages that are also official languages of. .. than 3 million speakers Outside Europe, Spanish is the official language of every mainland American country from Mexico to Argentina and Chile, with the exception of Belize, Guyana, Suriname, French Guiana and Brazil, and also of Cuba and the Dominican Republic A creolised version of Spanish (or possibly of Portuguese with subsequent Spanish influence) called Papiamentu is the national language of. .. variant is the official language of India, the Urdu variant that of Pakistan Bengali has about 140 million speakers, in Bangladesh and the Indian state of West Bengal Other national languages are Nepali (10 million, Nepal) and Sinhalese (10 million, Sri Lanka); an offshoot of Sinhalese, Divehi, is the language of the Maldives Other Indo-Aryan languages include the following; those marked with an asterisk... Namibia, Botswana South Africa), these are the languages of the so-called Bushmen and Hottentots Some Khoisan languages extend into Angola, Zambia and Zimbabwe and there are even two outliers in Tanzania It is probable that Khoisan languages occupied a much larger part of Africa before the southward expansion of the Bantu languages 3.5 Indo-European Languages The only part of the African continent where... official and national language of Poland Czech (over 10 522 LANGUAGES OF THE WORLD million) and Slovak (5 million) are the official and national languages of Czechoslovakia In Yugoslavia, three Slavonic languages are both official and national languages: Serbo-Croatian (17 million), Slovenian (in the north-west, 2 million) and Macedonian (in the south-east, 1 million) Bulgarian (8 million) is the official... Australian languages AN ENCYCLOPAEDIA OF LANGUAGE 531 has led to the divergence of an originally Australian language The surviving records of the extinct Tasmanian languages are insufficient to demonstrate whether or not these languages are related to the Australian languages The Australian languages with the largest numbers of speakers have around 12,000, a number greatly exceeded by some of the pidgins and... second-language speakers, that of Tajik around 4 million Pashto, the most widely spoken language of Afghanistan and a co-official language with Dari, has about 13 million speakers in Afghanistan and Pakistan Iranian languages that have large numbers of native speakers but do not correspond to political units are Kurdish (5 million, divided among Turkey, Iraq and Iran), Balochi (2 million, Pakistan), and . (Sweden and parts of Finland, 9 million). Among languages with fewer than a million speakers, Icelandic clearly merits mention as the official and national language of Iceland; another Germanic language, . Indo-European and some other languages since 1500.) (For legend see p. 960) AN ENCYCLOPAEDIA OF LANGUAGE 519 1.1.1 Germanic Germanic languages are dominant in north-western Europe. Languages with. further on this topic. AN ENCYCLOPAEDIA OF LANGUAGE 517 26 LANGUAGES OF THE WORLD: WHO SPEAKS WHAT BERNARD COMRIE The aim of this chapter is to present an overview of the range of languages spoken in