8.

Diffusion

8.1 Introduction

The diffusion of linguistic features from one area or social class to another and their adaptation to the system of the receiving language or dialect are complementary aspects of one and the same event. Both aspects must be considered in dealing with this phenomenon. Diffusion is socio culturally motivated. The manner in which spreading elements are adapted depends upon the structure of the recipient language or dialect.

The spreading of linguistic features from class to class and from community to community presupposes communication by means of the spoken or the written word, i.e., by hearing or by sight. Although all kinds of linguistic features can spread in either way, phonological traits are diffused largely by hearing.

Frequent and intimate communication facilitates diffusion. However, communication is only the necessary condition for the spreading of linguistic features. It does not determine the direction of the “flow,” which results from sociocultural factors and the socioculturally conditioned attitudes of the communicating speakers.

Within a socially divided or graded community, whatever its extent, diffusion from the higher to the lower social levels is the rule. The prestige and the usefulness of the “upper” dialect are the determining factors. A social revolution within the community may of course alter the direction of the spreading.

The sociocultural prominence of a speech area often induces a receptive attitude in neighboring communities, an inclination or desire to imitate the usage of a neighbor of superior standing. The motive may be admiration or the hope of improving one’s lot. From this results the interregional spreading of prestige dialects. Since the upper classes of society normally have the wider range of contacts, interregional diffusion is mediated largely by this element in the population.

When the balance of sociocultural prominence shifts from one center or area to another, the direction of diffusion may also change. Thus the northwestward spreading of Austro-Bavarian features during the Middle Ages and Early Modern times is followed after the Reformation by transregional diffusion from East Central Germany throughout the German-speaking area.

Viewed from the angle of the receiving dialect, diffusion is a process of adapting foreign linguistic elements to the native system. Meaningful word stems are freely introduced, but their phonemic shape is usually brought into line with the native system, and they are made to conform to the indigenous morphology.

In this process the phonemes of the donor dialect (or language) are replaced by their nearest counterparts of the receiving dialect. The extent of such replacements depends upon the degree of divergence between the phonemic inventories of the two dialects (languages) and upon the divergent use of distinctive features in the two phonemic systems.

Thus two phonemes of the donor dialect may be merged in one, and allophones of one of its phonemes may be rendered by two different phonemes in the receiving dialect. Whether the donor dialect has contrastive vowel length or not, a recipient dialect having contrastive quantity will inevitably impose it upon the imported elements. Conversely, functional vowel quantity of the donor dialect is discarded, if the receiving dialect lacks this feature.

The “tyranny” of the native system is equally evident in the prosodie features. Stress accent replaces pitch accent, and vice versa; tonal patterns of words are imposed or dropped, fore stress takes the place of end stress, and vice versa. In this manner the foreign elements are adapted to the native prosodie system.

Native word classes and their grammatical categories are regularly imposed upon the adopted word stems and expressed by native inflections. Bilinguals may of course retain some foreign inflections, and a few of them may ultimately enter the usage of monolinguals, as the Latin plurals alumn-i, -ae in English.

This manner of adapting foreign elements to the native system is characteristic of monolingual speech communities, whatever their size. Since monolingualism prevails in large parts of the world, this phenomenon has often been observed. Nevertheless, the structural point of view has rarely been applied to it with any consistency.

The dialect or language of another region or another class can, of course, be acquired fully or at least to such an extent that it can serve as a satisfactory medium of communication. If so, it may replace the indigenous dialect or language or gain currency alongside of it. In the latter case the roles of the two speechways are socially determined, the acquired dialect or language being used in transacting public affairs, the other in the home and in communication with one’s neighbors. Bidialectalism and bilingualism are not uncommon in Europe and in other parts of the world.

When after a period of bilingualism one of the languages is given up, the surviving one may exhibit some traits of the one that was submerged. Thus English has retained some phonological features derived from Medieval French in addition to a considerable body of French words; and the West Romanic languages (North Italian, French, Spanish) are said to show certain reflexes of the replaced Continental Celtic (Gaulish). Whether the replaced language was an overlay (superstratum), as French in England or Frankish in northern France, or an underlay (substratum), as Celtic in southwestern Europe, is immaterial. What counts is a more or less prolonged period of ethnically and/or socially conditioned bilingualism.

The area linguist is in a highly favorable position to trace the regional diffusion of linguistic features and the sociocultural dynamics underlying it. Dealing with living speech, he has potentially unlimited access to basic factual evidence. Also, he can observe in great detail the manner in which alien features are adapted to the native system, and can face the intricate problem of distinguishing imported innovations from parallel developments in dialects and in genetically related languages with some promise of success.

For these reasons the study of living speech has the potential of providing insights into the dynamics of linguistic change, insights that are relevant to problems confronting the scholar who concerns himself with earlier periods of linguistic history for which evidence is more limited or circumstantial.

One of the problems of area linguistics that has rarely been broached as yet is the interrelation between social dialects. Broad general statements are not wanting; but well-documented studies of larger areas will not be possible until we possess the results of sampling surveys of cultivated and of middle class speech comparable in scope with the existing atlases of folk speech.

The rate and the extent of the diffusion of linguistic features from community to community and from area to area is conditioned by the “density” of intercommunication. Ease of travel and lack of sociocultural barriers favor transregional diffusion; political, economic, and ecclesiastic boundaries tend to impede or to prevent it. Hence the frequent congruence of dialect and language boundaries with sociocultural barriers.

Since the upper classes of society enjoy the widest range of contacts with their neighbors, interregional diffusion is largely mediated by them. Their speechways “vault” from the dominant center to the elite of the subsidiary centers, whereupon they may infiltrate the lower social levels and spread out into the hinterland.

Free and frequent communication between the social groups of a speech community facilitate the spreading of innovations, social barriers retard or prevent their diffusion throughout the speech community. Usage within the community tends toward uniformity when communication is free; otherwise social dialects come into being and tend to be preserved.

The critical sketches of well-documented cases that follow are intended to illustrate the processes of diffusion and adaptation of linguistic features under a variety of sociocultural circumstances. The following factors are taken into consideration: (1) the geographic dissemination (spread) of the feature in question at a given time, i.e., the synchronic evidence; (2) the chronology of its appearance from place to place, i.e., the diachronie evidence; (3) the location of socioculturally dominant centers or areas from which the diffusion is presumed to radiate; (4) communication routes: highways, waterways, etc.; (5) barriers to communication: features of the landscape, political, and economic boundaries; (6) the adaptation of the spreading feature to the recipient dialect.

8.2 Diffusion of Reflexes of Postvocalic / r / in the Upper South

The method of dealing with diachronie phonological change in area linguistics can be effectively demonstrated by tracing the behavior of the reflexes of postvocalic / r / in the Upper South. This fascinating and intricate problem has been analyzed and discussed in great detail by William R. Van Riper [1957] as part of his treatment of the history of postvocalic / r / in four dialect areas along the Atlantic coast: Eastern New England, Metropolitan New York, the Upper South, and the Lower South. In the discussion that follows I shall rely largely on his findings.

The circumstances for handling this problem are unusually favorable. The basic areal data of the present (ca. 1940), both spatial and social, are ample, and the sociocultural setting of the present and the past is known in considerable detail. Although diachronie evidence for the change of / -r / to // in the Upper South is still lacking, the British background is fairly clear. Besides, the westward spreading of // through the cotton belt along the Gulf States and the valley of the lower Mississippi from 1810 onward presupposes that // was fully established in the Lower as well as the Upper South before 1800.

The areal data for the Upper South are furnished by the field survey for the Linguistic Atlas of the Eastern States carried out by G. S. Lowman, Jr. Some 200 communities were investigated in this area, and each community was usually represented by two speakers belonging to different social levels or age groups. Folk speech and common (middle class) speech were recorded in most of the communitie s, cultivated speech in about one fifth of them (mostly in cities). On the map reproduced in Figure 40 the responses are presented in the sequence folk-commoncultivated for each community, the numbers standing for cultivated speakers being underlined.

The evidence for the heterophones / -r / ≠ / -/ is summarized by Van Riper on the basis of their incidence in 33 words. Some speakers consistently use the postvocalic / r /, others the derivative //. Others again fluctuate between them, favoring one or the other of the variants.

The areal patterning of the variants is very striking. The western sections of Maryland, Virginia, and North Carolina consistently have / r /; eastern Virginia and adjoining parts of Maryland and north central North Carolina have //, with few exceptions; fluctuation and divided usage characterize the greater part of eastern and central North Carolina.

The focus of the derivative // in such words as ear, beard, chair, care, door, poor is clearly in eastern Virginia. In the same area we find also [ɔ<ɔ] in corn, forty and [<a] in car, barn, which we need not consider here any further.

Turning to the area of fluctuating and divided usage in North Carolina, we find that all the cultured informants interviewed—as far west as Charlotte and as far south as Wilmington—consistently use //, whereas other speakers either fluctuate between / r / and // or use only / r /, the latter especially on the points of land jutting out into the Atlantic and south of the Cape Fear River. Here the social dissemination of // clearly shows that this variant has prestige and is being diffused southward and eastward from the focal area. Inconsistent and divided usage along the lower Potomac River and in Baltimore also suggest diffusion of //. North and west of Baltimore, however, along the important dialect boundary between the South and the Midland, the trend seems to be toward / r /. It is worth noting, in conclusion, that in Virginia the variant // has not crossed the Blue Ridge Mountains.

FIGURE 40: The Upper South: Loss of Postvocalic / r /

In the shaded area / r / is lost in this position. On the periphery of this prestigious focal area speakers vary and fluctuate. Here some cultured speakers consistently LACK / r / in this position, others not. Middle class speakers fluctuate uneasily.

The percentage of the LOSS of / r / in 33 test words is shown, speaker by speaker, as follows:

Adapted from M. R. Van Riper, The Loss of Postvocalic / r / in the Eastern United States.

The search for sociocultural drives that may underly the dissemination patterns of // from postvocalic / r / in Virginia leads to two discoveries. (1) All of the colonial seaports in Virginia lie within the focus of the // area: Alexandria on the Potomac River, Fredericksburg on the Rappahannock, Richmond on the James, Petersburg on the Appomattox, and Norfolk on Chesapeake Bay. It is here that commercial relations and close cultural contacts were maintained with England throughout the Colonial period. (2) The plantation country served by these seaports in the export of staples and the import of manufactured goods coincides to a remarkable extent with the focal area of // from earlier / r /. We obtain a reliable picture of the area dominated by the plantation economy from the percentage of slaves in the total population. From 1790 to 1860 Negro slaves outnumbered the whites in the piedmont of Virginia, the center of the present // area. (See Paullin-Wright 1932: plates 67 and 68; 68-B is reproduced here in Figure 41.)

The close social and cultural ties of the planter class of Colonial Virginia with England is well known. If English fashions in architecture, furnishings, and dress were admired and imitated, why not also the speechways of cultured Englishmen? That the loss of postvocalic / r / as such was fairly widespread in eastern Virginia before 1800—as also in South Carolina—can be safely inferred from the fact that // is characteristic of all of the plantation country along the Gulf of Mexico and in the lower Mississippi valley, settled by westward expansion between 1810 and 1840.

Any notion that the “weakening” of postvocalic / r / to // could be attributed to the Negro majority in the plantation country is untenable on two grounds: (1) // is clearly a prestige pronunciation in the South; (2) it appears in New England and Metropolitan New York, areas that had few Negroes.

The date of the loss of postvocalic / r / as such in words like ear, fair, far, four, poor has not yet been established for the American South. N. E. Eliason [1956: 209-10] cites such telltale unconventional spellings as farther for father, storke for stock, polk for pork, caulk for cork, fouth for fourth and forth, vey for very for North Carolina, but unfortunately without date. For New England the early loss of / r / before consonants is clearly documented in such seventeenth and early eighteenth-century spellings as horsers for hawsers, northen for nothing, pasneg for parsonage, and Bostorn for Boston [Krapp 1925: 228-30]. Such scattered instances can, of course, not tell us how widespread this feature was in New England.

Before turning to the British background we must consider the areal patterning of the postvocalic variants // and / r / in North America as a whole. The variant // appears in four geographically separated areas along the Atlantic seaboard, each of them centered on one or more Colonial seaports. Except for the cotton belt along the Gulf of Mexico, occupied by the westward expansion of the plantation economy after 1800, this // is confined to the Atlantic Slope. From this situation we may infer, tentatively, that in the North and the Midland this phonological trait did not yet have general currency in its present domains when the settlements expanded westward; and furthermore, that // for earlier /r / spread rather late, and independently in each of the four areas. Whether this development was indigenous in the several areas or was prompted from overseas cannot be decided on American linguistic and sociocultural evidence alone. A decision can only be reached by a consideration of the British background, which is fortunately rather clear.

FIGURE 41: The South: Concentration of Slaves in 1860

From C. O. Paullin and J. K. Wright, Atlas of the Historical Geography of the United States.

In the British Isles the / r / after vowels is now regularly preserved in Scotland and Ireland, whatever its phonic realization. England is rather sharply divided: /r / survives in the west from the English Channel to Lancashire and also south of the lower Thames; the derivative // dominates all of eastern England from the Thames to Durham (see Figure 25). It is important to note that the London area is peripheral to the geographic domain of //. Little is known about the chronology of this areal patterning, except that // is spreading into Surrey and Kent.

We have some chronological evidence for the development of postvocalic / r / in Standard British English. The dramatist Ben Jonson (1573–1637), a native of Westminster, observed that /r / was “firm” at the beginning of words and “more liquid” in other positions, which clearly recognizes the existence of positional allophones. “Mute r” is mentioned by grammarians in mart, parlor, partridge before 1750. John Walker states (1775) that aunt, haunch sound as if written arnt, harnch, and (1791) that lard, card are nearly like laad, caad, especially in London, where / r / “is sometimes entirely sunk” [Jespersen 1922: 318, 360]. On the basis of this evidence, the “weakening” of / r / after vowels was taking place, or had taken place, in the London area before the American Revolution and was being accepted in cultivated speech.

In Colonial America, this innovation of SBE would be adopted in the seaports by leading families who had close contacts with England, infiltrate the lower social levels, and then spread to the hinterland, a process that continued after independence and is still in progress in some of the coastal areas.

But this is not the whole story. The spelling evidence referred to above shows that in eastern New England postvocalic / r / had been lost as such in the speech of some New Englanders by the middle of the seventeenth century—long before this feature was recognized in Standard British English. One can hardly escape the conclusion that it was brought to America by early colonists from the east of England, the area that lacks postvocalic / r / at present. If usage was divided, some New Englanders pronouncing an / r / and others not, the diffusion of // as a prestige feature supported by SBE usage from the latter part of the eighteenth century onward becomes understandable.

Although spelling evidence for early loss of / r / after vowels in the Upper and the Lower South is still lacking, developments in these areas may well parallel that in New England.

We shall briefly consider the effect of this change of / r / after vowels upon the phonemic system in American English.

In dialects that preserve / r / after vowels, it is articulated in that position by constricting the body of the tongue laterally, retracting it, and raising it close to the roof of the mouth, the tip of the tongue remaining inactive. The degree of constriction varies greatly from region to region. Full relaxation of the constriction results in an unsyllabic //-like sound in the coastal areas dealt with above. Simple as this change in articulation actually is, it has marked phonemic implications. Some have considered it a positional allophone of prevocalic / r / so that rear and roar, articulated as [ri and ro], would not only begin with / r / but also end in it. Others [Trager-Bloch] have taken it as a positional allophone of /h /, so that here and hair, articulated as [hi] and [he ~ hæ], would have / h / both at the beginning and at the end. Neither of these interpretations is acceptable, because [] has no distinctive feature in common with prevocalic /r / or with /h /. Daniel Jones [1956] appears to treat these ingliding diphthongal sounds as phonemic units. In my own discussion of this problem I have found it convenient to treat unsyllabic [] as a positionally restricted semivowel // [Kurath 1964; Kurath-McDavid 1961].

It should be noted that in dialects that do not preserve postvocalic / r / as such the sequence / ar > a/ has produced the new phoneme //, as in far, hard.

8.3 The Spreading of Diphthongal Reflexes of Middle High German /i, ū, ǖ/

This phonological phenomenon is discussed by K. Wagner in Deutsche Sprachlandschaften [1927]. His diffusional interpretation has been widely accepted by German dialectologists.

At the time of the Sprachatlas survey (1876-85), diphthongal reflexes of the Middle High German long high vowels / i, ü, ü /, as in Eis, Maus, Mäuse “ice, mouse, mice,” respectively, were current in German as far north as a line running somewhat to the north of Koblenz (on the Rhine), Kassel (on the Weser), Magdeburg (on the Elbe), and Berlin (on the Havel), but not in the greater part of the Alemannic southwest (Switzerland and adjoining sections of Elsace, Baden, and Wurtenberg). See Figure 42.

The evidence of spellings in localized and datable documents and regional literary texts is said to show that these three phonological features have been current in this entire area since the sixteenth century. However, the diphthongal reflexes of /ī, ū, ǖ/, whatever their precise phonic character, make their appearance within this extensive domain at very different times: first (1200-1300) in the Bavarian dialect area along the Danube (Vienna to Regensburg and Munich); next (a 1400) in the East Frankish area along the Main (Bamberg, Würzburg) and in Bohemia (the Emperial Chancelry at Prag); a century later (a 1500) in the East Central German area of Saxony (Meissen, Dresden, Leipzig), in the Rhine Frankish area (Frankfurt, Mainz), and in Swabian (Augsburg, Stuttgart); finally (a 1600 or later) along the present margin of the diphthongal area from the valley of the Mosel (Koblenz, Trier) to northern Hesse (Kassel) and Brandenburg (Berlin).

FIGURE 42: Diffusion in Germany: The Spreading of Diphthongal Reflexes of MHG Long Vowels i, ù (a 1200-1600)

Cities: B(erlin, D(resden, F(rankfurt, K(assel, Ko(blenz, Kö(ln, L(eipzig, M(agdeburg, Mü(nchen, P(rague, R(egensburg, St(rassburg, Stuttgart, W(ien.

Adapted from K. Wagner, Deutsche Sprachlandschaften

If the rough chronology of the regional appearance of diphthongal reflexes of the long high vowels /ī, ū, ǖ/ in such words as Eis, Haus, Mäuse is accepted, as it generally is, we seem to have here an instance of the gradual diffusion of a striking phonological feature from the Bavarian dialect area northward to the Frankish area along the Main and the Rhine and westward to the adjoining part of the Alemannic area, a process extending through several centuries. Add to that the secondary spreading of this feature from Frankish to East Central German in the fifteenth century—long after the German colonization of Saxony—and you have an impressive instance of transregional diffusion.

To account for this striking phenomenon in sociocultural terms, German dialectologists adduce the following factors.

(1) The extensive Austro-Bavarian area (some 300 miles from east to west) was unified early, both politically, culturally, and linguistically. Under the Hapsburgs and the Wittelsbachs this area had great prestige and influence. In fact, before the Reformation the Bavarian dialect bid fair to become the literary language of the German nation. Hence the adoption of the diphthongs in the neighboring Frankish and Alemannic areas.

(2) The important trade routes leading from the valley of the Danube to those of the Main and the Rhine facilitated the northward spreading of Bavarian dialect features. It is clear that from the Middle Ages onward, trade with the Near East and Italy (Venice) followed this route.

This interpretation of the spreading of the diphthongal reflexes of MHG /ī , ū, ǖ/ is rather convincing. And yet, the possibility of independent parallel development in the several dialect areas cannot be discarded outright.

For one thing, the long high vowels /ī, ū, ǖ / changed to upgliding diphthongs in other West Germanic languages in which outside influence is out of the question. For Netherlandish, Kloeke [1927] shows that this process started ca. 1200 in the south (Antwerp) and that by ca. 1500 the diphthongal reflexes had gained currency in the north (Amsterdam). In English the diphthongization of ME /ī/ and / ū /, as in ice, mice, and house, took place from ca. 1400 onward [Luick 1914–27: 563].

Secondly, in the Middle Ages all of the German dialects in which diphthongization has made its appearance had essentially the same system of syllabics, including the long vowels /ī, ǖ, ū; ē, ȫ, ō; ā / and the upgliding diphthongs /ei ~ ai, öü, ou ~ au /. This might well support the possibility of similar changes in the pronunciation of the long high vowels in the several dialects without outside influence.

Finally, for the Swabian dialect of Alemannic in which the reflexes of /ī, ū / are / ei, ou / over against Bavarian / ai, au /, K. Bohnenberger [1928: 279] maintains that the diphthongization is indigenous rather than imported. In support of his view he cites the important fact that in morpheme final position, as in speien “spew,” bauen “build,” /ei, ou / appear not only in Swabian but in all the Alemannic dialects as far south as the Bernese Alps in Switzerland [see now Hotzenköcherle 1962: vol. I, Maps 148, 152]. Remote from the Bavarian dialect area, this positional diphthongization is surely indigenous.

It is of considerable interest to observe that the diphthongs derived from the old long high vowels /ī, ǖ, ū / do not seem to merge with the reflexes of the MHG diphthongs /ei ~ ai, eu ~ öü, ou ~ au / in any of the German folk dialects [Kluge 1918: 29]. In Standard German, based largely upon the East Central German dialect, they are nevertheless merged: /ī/ with /ei / as in Wein - Bein, / ū / with / ou / as in Haufen - laufen, and /ǖ/ with / eu ~ ȫu / as in Häute - Leute. This matter deserves to be investigated from the structural point of view.

8.4 Northward Diffusion in the Rhineland

The sociocultural interpretation of the complicated linguistic situation in the Rhineland by Theodor Frings [1956 < 1922: part I, 1–54] is one of the landmarks in German area linguistics. His findings are summarily presented in Figure 43. Between the diagnostic heterophonic lines 4 and 3 lies the territory of the archbishopric of Cologne (Kurköln), between lines 3 and 2 that of the archbishopric of Treves (Kurtrier), south of line 2 that of the archbishopric of Mainz and of the Palatinate (Rheinpfalz).

Frings chooses the historically interrelated consonant shift lines as convenient indicators of the location of heteroglossic bundles, i.e., of dialect boundaries. Except for machen ≠ maken (line 4), they are of little importance in themselves from the synchronic point of view. Their significance lies in their diagnostic function.

The das ≠ dat line (2) is paralleled by the heterophones korb ≠ korf “basket,” uns ≠ ūs “us”; the dorf ≠ dorp line (3) coincides roughly with haus ≠ hùs “house,” wein ≠ wing “wine”; the machen ≠ maken line (4) agrees rather closely with offen ≠ open “open,” Wasser ≠ water “water,” han ≠ hebb “have,” fönf ~ fünf ≠ fïf “five”; and the heteroglosses öch ~ üch ≠ ow “you,” wir ≠ wej “we,” mir ≠ mej “me,” exhibit the same general trend as the ich ≠ ik line (5). See Maps 3, 4, 5, 6, 7 in Frings.

The regional dissemination of the High German reflexes /f, s, x / of the postvocalic Proto-Germanic plosives /p, t, k / furnish the clue to the interpretation of their history in the Rhineland. As far north as the S machen ≠ N maken line (4), which crosses the Rhine south of Düsseldorf (at Benrath), most of the words had the “shifted” HG consonants at the time of G. Wenker’s survey for the Sprachatlas (1876). But “unshifted” /־t / survived in dat “that,” wat “what,” allet “all,” as far south as the valley of the Mosel River (the domain of Trier). On the other hand, “shifted” /x / appeared north of the machen ≠ maken line in the pronoun ich “I” and some other pronomenal forms. The relics of “unshifted” /T~/ in the valley of the Mosel and the outposts of “shifted” /x / north of Düsseldorf clearly point to northward diffusion of the “shifted” High German (i.e., South German) consonants.

FIGURE 43: Diffusion in the Rhineland

2. Southern das ≠ Northern dat “that”

3. S dorf ≠ N dorp “village”

4. S machen ≠ N maken “make”

5. S ich ≠ N ik “I”

[1. S apfel ≠ N appel “apple” crosses the Rhine south of Worms]

Cities: A(achen, B(onn, Düsseldorf, K(oblenz, Kö(ln, M(ainz, S(iegen, T(rier.

Adapted from Theodor Frings, Sprache und Geschichte I: Maps 2 and 14.

Frings asserts that in the Rhineland the dissemination of High German /f, s, x / from West Germanic /p, t, k / revealed by the Sprachatlas for the latter part of the nineteenth century had been reached in the early Middle Ages (ca. 1000), at least essentially. However, although northward diffusion is beyond question, its chronology can hardly be reliably established, because local documents in the vernacular do not make their appearance until ca. 1250.

Linguistic evidence presented by Frings shows that the dialect area set off by the diagnostic heterophones 3 and 4 falls within the boundaries of the ecclesiastic domain of Köln, and that between lines 2 and 3 lies the domain of Trier. These ecclesiastic-political domains (set off in Figure 43 by broken lines) were relatively stable from the Middle Ages until ca. 1800. As such they played an important role in shaping the dialectal structure of the Rhineland. Each of them has some unique linguistic traits [Maps 13 and 6] and a distinctive configuration of regional features.

Frings envisions the transregional northward diffusion of southern features in the following terms. (1) The sociocultural predominance of the politically and linguistically unified southeastern (Austro-Bavarian) area along the Danube created a receptive attitude in the areas along the Main and the middle course of the Rhine. (2) The highways of commerce along the Rhine, leading from the Mediterranean to the North Sea, mediated communication with the southeast. (3) Southern traits were adopted in the centers along these highways and then diffused to the limits of the several political domains, thus creating the regional dialectal patterning of the Rhineland. Presented with considerable ingenuity and eloquence, his interpretation carries conviction, although the evidence of the Sprachatlas has serious limitations and chronological linguistic evidence for the crucial period is all but lacking.

In passing, it should be pointed out that Frings is not interested in evaluating heteroglosses or in pointing out that the adoption of the High German fricatives /f, s, x / does not introduce any phonemes foreign to the Rhineland. In native words this area actually had /f / in af “off,” / s / in glas “glass,” /x / in dach “day,” lachen “laugh,” etc.

8.5 Northward Diffusion in Eastern Germany

In eastern Germany, High German features have been spreading northward into the Low German area—from Saxony into Brandenburg — since early modern times. At the time of the Sprachatlas survey (ca. 1880) this process was in full swing. Using Atlas data, K. Wagner [1927: 66-69] describes this northward diffusion between the Saale- Elbe and the Oder in considerable detail and in sound perspective. From the divergent behavior of the heteroglossic lines, their spacing from south to north, he infers that High German features are adopted piecemeal and spread into Low German territory at different rates.

A striking example of the word by word adoption of High German consonants is shown in Figure 44. The areal divergence in the lexical incidence of the High German fricative /x / in kochen “cook,” machen “make,” and ich “I,” respectively, is ample evidence of the fact that the shift from /k / to /x / is not indigenous to Brandenburg. The spacing of these /x ≠ k / heterophones from south to north makes it equally clear that Brandenburg is adopting this feature from Saxony, though aided by the German Schriftsprache taught in the schools.

Other examples of the northward diffusion from Saxony along the Elbe and the Oder are presented by A. Bach [1950: 211–13].

8.6 The Southward Diffusion of Central Bavarian Phonological Features

A clear case of socioculturally motivated diffusion leading to revolutionary changes in the system of syllabic phonemes of the recipient dialect is found in the Bavarian section of German-speaking Europe.

For the Bavarian dialects we possess a considerable body of reliable evidence secured by direct field observation and several structurally oriented monographs. Additional data reported to the Sprachatlas and the Wortatlas at Marburg by correspondence are also available. The discussion that follows relies primarily on the evidence presented by E. Kranzmayer in his Historische Lautgeographie des gesamtbairischen Dialektraumes [1956] with its thirty-one sketch maps.

The Bavarian dialect area comprises the greater part of Bavaria and nearly all of Austria. It has four major subdivisions [Kranzmayer 1956: Hilfskarte 1]:

(1) A large Central area extending from the environs of Vienna to Munich, including the valley of the Danube and the lower course of its southern tributaries issuing from the Alps.

(2) A Northern area in Bavaria extending northward to the water-shed between the Danube and the River Main (including Regensburg and Nuremberg).

(3) A Southern area in Austria, extending from Tyrol (north and south of the Brenner Pass) through Carinthia into the southwestern part of Styria.

(4) A transition zone between the Central and the Southern areas along the northern slope of the Alps and in the province of Styria.

Kranzmayer’s outline of the dialectal structure of the Bavarian area is based on phonological data presented in twenty-six separate maps with reference to their Middle High German sources. These data are largely phonic, although phonemic implications can fairly easily be inferred in some instances. In his text of 143 pages the author frequently distinguishes between phonic and phonemic entities, especially in dealing with diachronie problems. All in all, some of the more striking systemic differences between Central and South Bavarian can be gathered with some assurance from his presentation and discussion.

FIGURE 44: Eastern Germany: Northward Diffusion of the High German Fricative /x /

1. in kochen ≠ koken “cook”

2. in machen ≠ maken “make”

3. in ich ≠ ik “I”

Cities: Berlin, F(rankfurt, H(alle, L(eipzig, M(agdeburg.

Adapted from K. Wagner, Deutsche Sprachlandschaften

Fortunately, we have a number of investigations devoted to the phonemic structure of several Central Bavarian dialects and of one South Bavarian dialect to guide us in comparing the phonemic system of Central Bavarian (CB) with that of South Bavarian (SB). More than half a century ago, A. Pfalz [1913] described a CB dialect spoken northeast of Vienna in phonemic terms—a truely remarkable performance that has had a strong influence upon the Vienna group of dialectologists, as did his later publications [1925, 1936]. A phonemic analysis of the dialect of Vienna, written as a doctoral dissertation under the direction of E. Kranzmayer, has been published by the American B. J. Koekoek [1955]. Another American-trained scholar, H. L. Kufner, has dealt in structural terms with the dialect of Munich [1961, 1962] and of a village about fifty miles to the east of it [1957]. For South Bavarian we have only one structurally oriented phonology, that of a family dialect spoken in a village near the city of Villach in the Austrian province of Carinthia [Kurath 1965]. Situated south of the formidable crest of the Alps, Carinthia has faithfully preserved many old SB characteristics. In my brief comparison of this SB dialect with CB, I shall rely primarily upon the work of Pfalz and Koekoek for the CB dialectal type.

Central and South Bavarian have nearly the same system of consonants: résonants: /m, n, η, l, r, w, j /; fricatives: /f, s, š, x /; lenis plosives: /b, d, g/; fortis plosives: /p, t, k /. But there are rather marked differences in the distribution of some of these shared phonemes: (1) in CB the résonants /n, l, r / are restricted to prevocalic position, but not in SB. (2) In SB / b / is restricted to postvocalic position, but not in CB. (3) In CB the fortis plosives /p, t, k / are restricted to postvocalic position, but not in SB.

The only structural difference in the consonants is found in the fricatives, where CB has fortis /ff, ss, šš, xx /, as in the words corresponding to Standard German offen, Wasser, waschen, machen. In SB these merged with lenis /f, s, š, x /.

In the system of syllabic phonemes, CB and SB differ markedly.

They share the following syllabics: (1) the monophthongs /i, e, ε, a, ɔ, o, u /, as in Fisch, Bett, lecken, zäh, Graf, oft, Luft; (2) the upgliding diphthongs /ai, oi, au /, as in weiss, bleuen, Maus; (3) the ingliding diphthongs / ia /, as in SB lieb, CB Wirt; / εa /, as in SB Zehe, CB wert; / ɔə /, as in SB rot, CB Garten; /uə /, as in SB gut, CB Wurst.

As indicated by the examples, CB and SB often differ in the lexical incidence of these diphthongs.

Whereas South Bavarian has no qualitative syllabic phonemes peculiar to it, Central Bavarian has a considerable number of them:

(1) The rounded front vowels /ü, ö, /, as in Bild, Hölle, hell, which result from the “vocalization” of /1 /. SB preserves the old sequences /il, el, εl /.

(2) The nasal syllabics /ĩ, ẽ, ã, õ, ũ, /, as in Sinn, schön, Bein, schon, Sohn, Wien, which result from the “vocalization” of / n /. Here SB preserves the old sequences /in, en, an, on, un, εεn /.

(3) The upgliding diphthongs /ui, oe, ɔε /, as in Pult, Wolf, Walzer, derived from the sequences /ul, ol, ɔl /, which SB retains.

The system of monophthongal oral syllabics in CB and in SB are shown below: CB has rounded front vowels, SB does not. SB has phonemically short and long vowels, CB only positional variation in duration.

The duration of the syllabic phonemes is regulated in different ways in CB and in SB.

In CB all stressed syllabics are prolonged unless they are followed by a fortis plosive or fricative. Examples containing short syllabics before fortis /p, t, k / and /ff, ss, šš, xx / are:: Köpfe /ghepf /, Witze /wits /, Ecke / ekx / and offen / offə /, Fässer / fessə /, fischen / fišša /, machen / mɔxxə /. Since in CB the duration of syllabics is regulated by position, it is not phonemic.

In SB the duration of stressed syllabic phonemes depends largely, but not wholly, upon context. Here vowels are long unless they are followed by a consonant cluster. They are also short before the fortis plosives /p / and /k /, as in täppisch /tεpat /, eckig / ekət /, before the velar nasal / η /, as in lang / loη /, and irregularly before / m /, as in dumm / tum /, zusammen / tsom /. On the other hand, long vowels occasionally appear before morphologically complex consonant clusters owing to leveling in the paradigm of verbs. Hence vowel duration is phonemic in SB, although largely controlled by the context.

The loss of the postvocalic resonants /n, r, 1 / and the consequent rise of numerous new syllabics constitute one of the most important phonological innovations in the Central Bavarian dialect area since the Middle Ages. At present (ca. 1925-50) these features are current in the valley of the Danube from Vienna to Regensburg and in the valleys of its tributaries, except for the upper reaches of the rivers issuing from the Alps (Salzach, Inn, Isar). In the province of Styria, south of the Semmering pass, postvocalic / n / is lost throughout and postvocalic / r / in the northeastern section; but postvocalic /I / is preserved (see Figure 45).

FIGURE 45: Central Bavarian: The Loss of Postvocalic /n, r, 1 / and its Southward Diffusion

1. Loss of / n / [Kranzmayer: Map 23]

2. Loss of / r / [ibid.: Map 6]

3. Loss of / I / [ibid.: Map 7]

Cities: G(raz, I(nnsbruck, L(inz, M(unich, P(assau, R(egensburg, S(alzburg, V(illach.

Adapted from E. Kranzmayer, Historische Lautgeographie des gesamtbairischen Dialektraumes.

The heterophonic lines resulting from the loss vs. the preservation of /n, 1, r / in postvocalic position run close together along the crest of the Alps that separate the province of Salzburg from Carinthia; but in Styria (south of the crest of the Alps) and in Upper Bavaria and Tyrol they are more or less widely spaced. In these graded areas, one may safely infer, change is still in progress. An investigation of the usage of age groups and of social classes in these transition zones would presumably show that /n, r, 1 / are receding. Since such evidence is as yet not available, we must be content with an inference from the location of the socioculturally dominant centers of the Danube valley and conclude that Central Bavarian usage is spreading into the South Bavarian dialect area and that this process has been in progress for some time. The Danube valley with its ancient population centers has dominated the Alpine valleys in economic, political, and cultural matters from the early Middle Ages onward. Ancient trade routes lead from the Vienna area across the Semmering pass into Styria and from Bavaria via the valley of the Inn across the Brenner Pass. Along these routes Central Bavarian features have been diffused.

In the valley of the Danube the phonological development discussed here is very old. According to Kranzmayer [1956: pp. 120, 123] loss of postvocalic /l / and /r / is attested in the Danube valley, especially in Lower Austria, shortly before 1300. Evidence for the loss of / n / seems to be even earlier.

The Central Bavarian loss of the résonants /n, r, 1 / as such in postvocalic position changed only the distribution and the lexical incidence of these consonants. On the other hand, their “vocalization” revolutionized the system of syllabic phonemes in CB. The magnitude of these changes can be gathered from the following list in which the innovations in the Viennese variant of CB [Koekoek 1955: 37-43] are set over against the synchronically corresponding forms of a SB dialect [Kurath 1965: 25-27].

As the result of a change in the articulation of / n, 1, r / after vowels, a total of 13 new units are added to the CB system of syllables:

(1) The loss of /-n / as such produced the 6 nasal vowels /ĩ, ẽ, ã, õ, ũ, /;

(2) the loss of /-r / as such created the 4 ingliding diphthongs / iə, uə, εə, ɔə /, the first two of which merged with old phonemes exemplified in lieb / liəb / and gut / guəd /;

(3) the loss of /-l / as such resulted in the 5 new syllables /ü, ö, , ɔi, oi /.

It should further be noted that two new distinctive features are thus introduced in the CB system of syllables: nasalization and lip rounding. Rounding of the lips was automatic in the older stage of CB, as it still is in SB, i.e., it was confined to back vowels.

We may finally ask this pertinent question: when speakers of South Bavarian give up their native articulation of postvocalic /n, r, 1 /, as in Styria and Upper Bavaria, do they invariably adopt the new syllables of Central Bavarian at the same time? No answer can be given until the behavior of speakers belonging to different generations of the several social groups is investigated in the transition belts.

In a recent study, P. Wiesinger [1967] shows that the province of Styria is a typical transition area, graded from northeast (Lower Austria) to southwest (Carinthia) in its phonology. See his Map 1 on page 121. His data are apparently taken from Kranzmayer’s collections in Vienna.

8.7 The Southward Diffusion of North Italian Consonants

In Southern Italy and in Tuscany, the medial voiceless plosives /p, t, k / of Latin and the Latin geminate consonants are preserved. In Northern Italy the voiceless plosives of Latin became voiced in this position and the germinates were reduced to short consonants (as also in French and Spanish). The heterophonic lines run in a close-knit bundle along the crest of the northern Apennines, the northern boundary of Tuscany. Along the Adriatic the lines are spaced; and in a corridor leading through the Papal States from the Adriatic (Ancona) to the Tyrrhenian Sea (Rome), they behave in a very complicated way, which points to diffusion rather than indigenous phonological change.

This problem is dealt with in exemplary fashion by R. A. Hall, Jr. [1943], on the basis of the evidence furnished by the linguistic atlas of Italy [Jaberg-Jud: 1928–40]. The location of the heterophones Northern / d / ≠ Southern / t /, as in N fradel ≠ S fratello ~ frater (< L frāter “brother”), is shown for ten different words; the course of the heterophone N /g / ≠ S /k /, as in N fig ≠ S fico (< L ficus “fig”), for ten lexical items, and so forth.

Northern / g / from Latin /k / exhibits the widest dissemination (see Figure 46). It is current in a belt running from the Adriatic (Ancona) to the Tyrrhenian (Rome), all within the confines of the Papal States (Marche, Umbria, Latium), varying somewhat from word to word. Northern / d / from Latin /t / is widely used within this corridor but falls short of Rome and its environs [Hall: Map 4]. Northern /k /, from Latin /kk / as in bucca “mouth,” is largely confined to the shore of the Adriatic in the province of Marche [Hall: Map 1]; so is Northern /t /, from Latin /tt / as in catta “cat” [Hall: Map 3].

FIGURE 46: Italy: Southward Diffusion of Northern / g / from Latin /k / Exemplified in Ten Words

From R. A. Hall, Jr., The Papal States in Italian Linguistic History.

Divergence in the lexical incidence of the several “shifted” Northern consonants as shown by the behavior of the heterophonic lines and the gradation of the lines from the Adriatic to the environs of Rome unmistakably point to southward diffusion from the Romagna (Ravenna, Bologna)—the northmost of the Papal States—in the direction of Rome. This spreading of northern features follows the Via Flaminia and the Via Salaria, ancient highways connecting Rome with the North.

Hall assumes that the diffusion probably started in the thirteenth century and that with the later rise of Tuscany as the dominant cultural center of Italy the Northern “shifted” consonants receded, especially in sections of the Papal States adjoining Tuscany. No philological evidence is presented in support of this chronology; it is probably not available.

The passive linguistic role of Rome within the political domain of the Papal States may seem surprising. Hall attributes it to the economic superiority of the North and the sharp decline of Rome during the early Middle Ages.

In conclusion, Hall suggests that sporadic Northern consonants in Tuscan, as /g / in pagare “pay” (< L pacāre), ago “needle” (< L acus), spread into Tuscany by way of the Papal States rather than across the northern Apennines.

8.8 Diffusion between Unrelated Languages in India

Structural similarities between genetically unrelated languages spoken in one and the same area or in contiguous areas have been pointed out repeatedly in the last fifty years: in the Amerindian field by F. Boas [1940 < 1917: 202-7] and E. Sapir [1921: 217-20]; in Europe by H. Schuchardt [1928 < 1912: 248-53], and by N. S. Troubetzkoy and R. Jakobson [1949 < 1931: 343-65]. In his article on “India as a Linguistic Area” [1956], M. B. Emeneau discusses this phenomenon in the light of ample linguistic and sociocultural evidence.

The subcontinent of India is indeed an ideal domain for tracing cross influences between languages of different stocks. Indo-Aryan(Sanskrit and its congeners) is documented both areally and chronologically, and some of the Dravidian languages are well enough known to gage their potential influence upon the Indo-Aryan overlay. For the Munda languages the evidence appears to be sufficient to compare some of its structural features with those of the other two linguistic stocks of India.

As far as Indo-Aryan is concerned, its position in the family of Indo-European languages and the consistent internal developments of its phonemic and morphological structure from Proto-Indo- European onward have been traced in great detail. When features not paralleled in any other Indo-European language make their appearance, the question arises whether they are peculiar indigenous developments or imported from a foreign linguistic system, i.e., from the Dravidian and/or the Munda languages, with which speakers of Indo-Aryan have been in intimate contact through the ages.

Emeneau shows that certain morphological features in which Indo-Aryan deviates from the other IE languages are shared by Dravidian and Munda languages and were in all probability adopted from them. Speakers for whom Indo-Aryan was a “second” language presumably mediated this change.

Four such morphological and/or syntactic features are pointed out [9-19]: (1) Distinct word stems for the singular and the plural of nouns preceding the case morphemes; (2) the extensive use of the so-called “gerund,” an uninflected verbal form; (3) a peculiar type of compound consisting of the stem of a past participle and a noun; (4) the formation of so-called echo words. None of these formations actually upsets the basic structure of Indo-Aryan.

A further morphological feature, discussed by E meneau at some length [10-16], is the use of “classifying” or “quantifying” words or morphemes in certain types of noun phrases, which some branches of all three linguistic stocks of India have in common. Since the “classifying” morphemes occurring in Dravidian and Munda are clearly adopted from Indo-Aryan, the direction of the diffusion of this feature within the linguistic sphere of India is clear. Here the superimposed language is the donor. In Indo-Aryan itself the use of “classifiers” is an innovation modeled on the usage of a vast linguistic sphere comprising “East, Southeast, and South Asia” [16] and such diverse linguistic stocks as Burmese, Chinese, and Japanese.

E meneau deals briefly with the retroflex consonants /ṭ, ḍ, ṇ, ṣ/, which contrast with the dentals /t, d, n, s / in all three linguistic stocks and characterize India as a “linguistic sphere” (Sprachbund). Since no other branch of Indo-European has these contrastive sets of consonants and Proto-Dravidian had them [7], the retroflex set in Indo-Aryan looks like an importation. But how did it come about? What started it?

The answer is rather clear. In the indigenous vocabulary of Indo-Aryan, the retroflex consonants appear almost exclusively in complementary distribution with the dentals. Thus retroflex / s / occurs only after /i, u, e, o, ṛ, r, k /; /ṭ, ḍ / only after/ ṣ, ẓ /; /ṇ / after / ṣ, r,, ṛ/. In all other positions / s, t, d, n / are normal [Thumb-Hirt’1930: §§ 83, 122, 147]. The presence of retroflex allophones of the dentals in Indo-Aryan thus facilitated the adoption of retroflex consonants in other positions when words were imported from Dravidian. In the process they became phonemic.

This problem is discussed with extensive documentation and full consideration of the sociocultural setting by F. B. J. Kuiper [1967].