Processes of horizontal and vertical convergence in present day Germany

0. Introduction This paper attempts to give a brief account of the major processes of convergence in the German language area, processes which may be called horizontal (if they affect the inter-dialectal dimension) or vertical (if they affect the standard-dialect dimension). The first section presents some preliminary thoughts on convergence and divergence in social dialectology, while section two outlines four developments in the German language area (focussing on the situation in Germany and Austria).


Convergence and divergence in social dialectology: some preliminary remarks
The notion of convergence is used in a structural sense here, i.e. as the loss of contrasts between two linguistic varieties/systems.It is therefore to be distinguished from the following processes: -Short-term interactional convergence, which, for reasons of clarity, should be called (positive) accommodation, a process which implies a temporally limited assimilation of two or more interactionists' behaviour within an interactional episode; it may affect just one, but usually works on various levels of symbolic expression, and non-grammatical parameters such as rhythm and tempo, mimics or gesture, seem to lend themselves more easily to accommodation than phonology, morphology or syntax, in the framework of H. Giles' accommodation theory, this type of convergence receives a socio-psychological explanation: The accommodating speaker is said to wish to gain the co-participant's "approval" by assimilating his or her behaviour (cf.Giles, Coupland & Coupland 1991).
-Long-term (dialect) accommodation, which affects the linguistic habits of a person who assimilates another person's or group of persons' habits, for instance as a consequence of migration across dialect communities within a language community (Trudgill 1986, Auer/Barden/Grofikopf 1998).This process has consequences beyond the limits of short-term accommodation/convergence but nevertheless is restricted to the behaviour of individuals, while the repertoire of the community at large remains unaffected.
In short, the three developments refer to changes in different domains, which may be summarized as follows: • Short term accommodation: affects behaviour within an interactional episode • Long term accommodation: affects a (group of) person's speech habits • Structural convergence: affects a linguistic system (variety/Zangue) The relationship between these three types of 'convergence' remains to be investigated; there is good reason to believe that they should be kept apart analytically.In particular, although the three levels obviously stand in a hierarchical relationship of some kind to each other (starting from the 'lowest', i.e. most restricted level, and ending with the 'highest', i.e. the most general level), and although they proceed from the most 'individual' to the most to the least manipulatable one), they do not seem to 'feed' into each other, in the sense of the lower level changes being the prerequisites for the higher level ones, finally accomplishing change in the language system.There are numerous examples for interpersonal accommodation to occur without language change, and, more importantly, for language change to occur without interpersonal accommodation (cf., e.g.Gilles 1999).The same holds for the relationship between short-and long-term accommodation (cf.Auer & Hinskens, in prep., for a summary).
Since convergence was defined as the loss of contrasts between two linguistic varieties/systems, focussing on the changes in one linguistic system can of course not to be an adequate way of looking at convergence: There are always (at least) two linguistic varieties involved by definition.Convergence is a phenomeon in which two varieties (dialects, languages) are in contact; and this also holds for those not unfrequent contexts in which only one variety is changing (assimilating) while the other remains unchanged.It may, however, be more adequate to speak of advergence in these cases (Mattheier 1998).
Two further theoretical issues need to be commented on here: (1) Does convergence lead to an increase or a decrease of variation?and (2) does convergence equal levelling?Ad (1):' In discussing the relationship between variation and convergence, it is necessary to distinguish between the process of convergence and its outcome.Obviously, and by definition, the outcome of any process of convergence will be less distance between the varieties or variants in the repertoire or variational space, comparing them to what they were like before convergence set in.Thus, if we consider the repertoire of a community in which, for instance, a standard and a dialect existed as relatively distant varieties clearly separated from each other, convergence within this repertoire implies that, at the termination of this process, these varieties will be structurally more similar.If one refers to the magnitude of the variational space within a Cf.Hinskens 1993Hinskens (1996)): 20ff for further discussion.repertoire as defined by its extreme points (maximal contrasts), then the outcome of convergence implies that this magnitude is reduced, and variation within the repertoire diminishes together with it.However, the picture is more complicated, since convergence within a repertoire often co-occurs with the emergence of compromise forms or varieties of various kinds.In phonology/phonetics, these may be -Phonetically intermediate forms (Trudgill 1986 speaks of "fudged lects") which can be introduced into the repertoire whenever the difference between the two original forms is continuous (as in centralization, velarization of /a/, unrounding and rounding, lowering and raising, etc.); -Structural convergence without phonetic convergence; for instance, the distribution of a variant may be assimilated to that of the structurally corresponding one in the converged-to variety, while the phonetic surface exponent of that phoneme remains the same (cf.Hinskens 1998:46 for ich/«c/i-Laut-allophony in easternmost Dutch as an example: The allophony itself is lost as a consequence of advergence towards standard Dutch, but the generalized regional exponent [9], i.e. the former front variant, continues to establish a phonetic contrast with Std. Dutch [x]).
-Lexical erosion: The feature of the converging dialect which is given up in certain lexical contexts but is retained in others (cf. Auer 2000a, in press, for examples).
-Hypercorrection: In convergence towards another variety, speakers overshoot the target and generalize the 'new' form to environments in which it is not used in the converged-to variety.(For instance, avoidance of /g/-coronalizationi.e., convergence towards standard phonologyhas led in a number of German dialectsmost notably in the area of Aachento hypercorrect /g/ in words such as Menschen 'people' = std./mcnfn/.)-The variant used in the converging variety is not replaced directly by that of the converged-to variety (e.g., the standard), but rather, a more prestigious form is borrowed from a third variety (for instance, a neighbouring dialect).In this case, 'psychological' convergence towards one variety leads to structural convergence towards another (cf.Thakerar, Giles & Cheshire 1982 and Auer 1997 for an example).
All these cases result in a situation in which the variational space between the varieties in a formerly strictly compartmentalized repertoire is replenished with new forms; they are the outcome of convergence between the 'old' varieties, and they lead to assimilation between them, but at the same time, they imply the emergence of a plurality of new forms which did not exist before.Thus, while the variational space (as defined by the structurally most distant variants) becomes more restricted in the end, the number of alternative forms in the speaker's and/or community repertoire may increase.
If we look at convergence as a process, it is obvious that convergence does not lead to a reduction, but rather an increase in variation; for the loss of contrasts between two varieties does not necessarily imply that the old, more distinct forms disappear entirely and all of a sudden.Thus, variation within the reportoire increases instead of decreasing at the beginning of a process of convergence, simply because new and old forms co-exist.
Ad (2): The equation of levelling and convergence obviously hinges on the definition of levelling.If levelling is defined as the "reduction or attrition of marked variants" (Trudgill 1986:98), the notion of markedness becomes central, which may receive very different interpretations as well.In one sense, which equals Victor Shirmunskij's distinction between primary and secondary dialect features (cf.Auer 2000b, in print), markedness (or saliency) is a psychological feature which is probably quite independent of structural considerations (cf.Auer/Barden/ Grofikopf 1998 for a discussion).Since the term 'saliency' is available to cover this meaning of markedness, it seems preferable to reserve the latter term for structural changes.But in this case, whatever one's theory of structural markedness may look like, there surely are processes of convergence in which the structurally more marked feature is being accommodated.Very simple examples for an increase in structural markedness which results from convergence are phoneme splits as they regularly occur in dialect convergence towards the standard variety.(For instance, in German, there is a change from dialectal /i/ to std./i y/ in various areas.)Levelling in this sense then is a specific form of convergence, but must not be equated with it.

Four tendencies in present-day German
The fundamental sociolinguistic development of the German language area in this century (with the possible exceptions of Switzerland and the Low German area) is without a doubt the process which Bellmann (1983;1998) calls dediglossation, i.e., a transition from a diglossic to a rfmglossic situation.The diglossic situation as it must have existed in Germany at least until the latter half of the 19th century was characterized by a nationwide uniform standard language on the one hand, which started out as a purely written variety and later came to be spoken by certain classes of the population (in a heavily regionalized phonological form), and the base rural dialects on the other hand, which were structurally distant from this standard as well as being exclusively oral, without significant intermediate forms.The transition to a diaglossic situation implies the structural replenishing of the variational space between these extremes.The sociolinguistic (and especially social-dialectal) developments which can be associated with these dynamics are those of vertical convergence as outlined in section 1 above.In Germany, they have included the formation of so-called regional dialects as well as of regional standards.Processes ot horizontal convergence between dialects are implied by vertical standard convergence between dialects and the standard; however, they can occur independently as well.
Standard-dialect-convergence has not been particularly well regarded by dialectologists nor by most language speakers.Both suspect that this process generally leads directly to dialect attrition: Regional dialects are not considered to be 'real' dialects and as such are 'worthless'.2The accusation is particularly directed at the idea that standard-dialect-convergence leads to the disappearance of old base dialects, therefore destroying language forms which have developed naturally over the centuries.The implicit or even explicit argument here is therefore puristdealing with pure versus corrupted languages or language varieties.
In the following pages we will attempt to develop a somewhat more pleasing picture of standard-dialect-convergence.As will be shown, despite possible structural losses at the extremes of the standard-dialect continuum, standard-dialect convergence can even be said to have had a positive effect on the maintenance of regional language forms; this effect is due to the fact that it has been working against the decreasing contexts of usage most modern societies (and surely the German one) offer for regionalized speech on the basis of the base dialects, and by somewhat defusing the choice 'base dialect or standard language'.At least the following four tendencies can be identified.

Enrichment of the repertoires
A fundamental step from the traditional to the social dialectological point of view entails understanding individual varieties (such as base dialects) as parts of the linguistic repertoires of speakers or groups of speakers, rather than in isolation.As outlined in section 1, the linguistic developments in Germany and Austria in this century have undoubtedly led to an increased complexity in the repertoire of forms at the disposal of individual speakers, which they use appropriately in a given situation.Thus, from a repertoire point of view, no loss of any kind can be ascertained, but rather the opposite, an enrichment.It is quite likely that there is scarcely one speaker of German today who does not have several ways of speaking at his or her disposal which can be characterized according to their proximity been activated since the beginnings of the discipline in order to convince financial benefactors of the necessity to research base dialects on the verge of extinction.
to standard or dialect.The pure dialect speaker (in sociological terms typically the immobile villager employed in traditional branches of the economy such as farming, fishing or viniculture), is a thing of the past.As rarely as we find one-dimensional social identities these days, do we find one-dimensional speakers: identities are no longer claimed by and attributed to individuals across time and situation, but rather constantly re-negotiated in different settings.In this way, identity dissolves into a multitude of context-dependent categorisations, or rather: into acts of identity.These acts and their co-comitant self-and othercategorisations are based on various symbolic resources, including the use of the appropriate means from the linguistic repertoire.
Up to now, we know rather little about how linguistic repertoires are structured in the various parts of the German language area.It is assumed though that there are mostly continua in the south, and more clearly separated varieties in North Germany (or rather, those parts of the North in which Low German is still spoken), as well as in German-speaking Switzerland (where this internal compartmentalization is strongly supported by "medial diglossia" between an almost exclusively written standard and spoken dialects); or in other words: the enrichment of the repertoires has proceeded further in the middle and southern part of Germany and in Austria than elsewhere.This, however, is surely not the whole story.In some areas the regional standard in the repertoire is quite distant from the codified national language, but the repertoire reaches across the continuum to the very base dialects.In other regions, the standard-nearest way of speaking is closer to the codified national language, but the repertoire does not include more than a regional dialect or even a regional standard on the dialectal side.Compare, for example, a city like Regensburg in the Bavarian dialect areawhich represents the first case, with a place such as Freiburg in the Alemannic area which represents the second case: Both are located in southern Germany but belong to different repertoire types.The structure of individual speakers' repertoires continues of course to be influenced by sociodemographic factors such as rural/urban (still!), age, sex and professional mobility, in addition to the differences in the repertoires on the community level.
The result of standard-dialect-convergence is therefore not an attrition of the verbal resources, but rather an increase in heterogenity which enables the speaker to design his or her speech to the demands of the particular situation.As is also true for multilingual repertoires, this contextual design is by no means determined by the parameters of the situation; rather, the link between linguistic resources and the definition of the situation allows room for interpretation and negotiation.Speakers can utilise the repertoire's resources actively and creatively (as contextualisation procedures; cf.Gumperz 1982), in order to differentiate between meanings, to determine the relationship to the co-participant and to portray themselves as social personae.
Moreover, standard/dialect-convergence has made obsolete the social stigmatisation of the once large groups of speakers who only spoke a traditional dialect.Despite the so-called dialect renaissance of the 70s, there is little reason to believe that there has been a fundamental change in attitudes towards these base dialects.Even today and even in Southern Germany or Austria, they continue to be illegitimate languages, in Bourdieu's sense (1972), which are by no means tolerated outside close-knit rural (mainly family) networks.(Again, Switzerland may be an exception.)Usage of the base dialects in any other situation will entail the speaker being categorised as 'rural', as well as a whole series of associated, mostly negative stereotypes (from 'cute' but 'backwards' to 'limited' and 'stupid' or 'uneducated').Tolerance towards dialect usage in the Upper and to some extent the Middle German language areas only extends as far as regional dialects are concerned.From this it follows that it is only because of the fact that almost all Germans speak such intermediate varieties or regiolects (instead of, or in addition to a base dialect) today, which unlike the base dialects are accepted in many situations, that regional (non-standard) language use outside close-knit rural communities has remained acceptable at all.In contrast to this, in diglossic situations with a strict separation between the standard and base dialects (as in the Low German language area), the negative prestige of these dialects and the lack of an alternative regionalized way of expression has led to the disappearance of the dialectal component in the repertoire in toto.Dialects here have developed into a linguistic means of expression with an extremely limited range, and are at the verge of extinction, while all remaining communicative events are now served by the standard variety or an approximation of it (formerly some kind of Missingsch, today a regional standard)?Since in these sociolinguistic contexts, the regional standard is the only way of expressing local identity, strong destandardization tendencies are likely to occur, which imply a certain divergence from the neighbouring areas.4' Only in recent times Low German has been reappraised; in most areas this has come too late to be able to stop the disappearance of Low German dialects in everyday communication ■* there is little research on the divergence ot regional standards, unfortunately, but it seems that certain features of the Hamburg city vernacular are a case in point: Both in the older generations (nonpalatalisation of morpheme initial /s/ before obstruents as in Stein ([stain], std.German [Jtain]) and in the younger generation (flapping of intervocalic /t/), certain local features are prestigious and spread; cf.Auer (1998b) for some preliminary research.1 Cf.Mattheier & Radtke (ed.) (1997).

Destandardisation
Despite this general expansion of the repertoires through the inclusion of intermediate forms, it is obviously the case that the extreme poles of the standard-dialect continuum have suffered from standard-dialect convergence.Yet, the cutting off of extreme forms takes place not only at the base dialect pole of the continuum (where in fact certain local dialectal features have disappeared), but also at the standard extreme where destandardisation of the non-regionalized national language has occurred?
Let us consider the latter aspect first.The high-days of a nonregionalized, national standard language in Germany must have been the period between the end of last century and the Second World War.Throughout the 19th century, standardization of the spoken language proceeded, particularly in the lower middle and the upper working classes; but the apex of this development does not seem to have been reached earlier than around the turning of the century, and lasted until the 2"d World War.There is a good reason to assume that the German and Austrian (upper) middle classes of the pre-war period spoke a variety closer to the national standard and less regionalised than their counterparts of today.This is clearly demonstrated, for example, in recent recordings made with emigrant German and Austrian Jews in Israel (Betten & Hecker 1995).Untouched by the sociolinguistic changes of the post-war period in their countries of origin, these speakers have retained a variant of German only minimally influenced by regional varieties.
In contrast, present-day language reality in the Federal Republic as well as in Austria is characterised by a prestige shift from the one uniform national standard language to the regional standards, at least in phonology but also partly in lexicon and grammar.For example, all German politicians speak regional standards in public.6(The fact that this is more noticeable in some like former Chancellor Kohl or former President Herzogthan in others -like present-day Chancellor Schröder or present-day President Rauhis easily explained by the differing degrees of deviation of the regional standard languages spoken in their respective places of origin from the uniform national language.)The objective distance between the regional standards and the codified national standard varies in fact from region to region: Presumably, it is relatively pronounced in Bavaria as well as in Baden-Württemberg, Rhineland-Palatine, the Saarland, Saxony and Thuringia, but less in some parts of Niedersachsen, Hessen, the Ruhr and even in the Rhineland.From the point of view of the speaker, i.e., in an ethnodialectal perspective, the perceived distance of the respective regional standard from normative Standard German also varies: Northern German speakers normally look upon their own way ot speaking as pure standard German despite the obvious regional forms they use.7 7 A good example is the merger of the middle high front long vowel /e/ (of normative, orthoepic phonology) with /e:/, which is rated as pure standard by all north German speakers.8 This lexical loss is not quite as superficial and in the end not as irrelevant for dialect maintenace as is sometimes suggested: The old phonological classes disappear with the words.Lexical change thereby also promotes phonological change through lexical diffusion.
In sum, dediglossation has led to the establishment of regional standards which can be used for expressing regional identities.The pluricentricity of the present-day sociolinguistic situation corresponds to the federative and non-centralistic self-image of the Federal Republic of Germany.Acts of identification with larger regions such as Hessen, Bavaria, Franconia, and so forth, play an essential role for many Germans.This can be demonstrated linguistically in the extension of regional standard usage to official, formal situations (where dialect usage would be inappropriate).

Loss of base dialect features
Since most social dialectological studies have used apparent time methodology, not too much indisputable information on the actual shifts at the deepest dialectal level is available.However, the opinion of some dialectologists (cf.Ruoff 1997) that the (southern German) dialects have not changed at all, but that it is only their speakers which change while they move from one age cohort to the next, can be refuted empirically.The actual changes in the base dialects are not limited to vocabulary, where of course words whose denotations have disappeared from a rural daily culture die out.8They come to light whenever more recent data can be compared with those of older studies.Although the latter have followed traditional dialectological methods and only recorded older speakers, they offer a way to investigate changes in real time at least in this age (and social) group.On the basis of a combined real/apparent time method, distinct changes were found, for example, in the city of Constance (cf. Auer 1990(cf. Auer , 1997)).They show clear loss in the base dialect phonological system.One striking example is the almost complete disappearance of the high long vowels corresponding to Middle High German /i:, u:, y:/, which were still common in the 20s within the group examined, but which in the meantime have as good as disappeared even in the older group and been replaced by near-standard diphthongs.
The attribute 'as good as' indicates that the loss of such dialectisms generally proceeds word by word; loss of the old form and replacement with a form closer to the standard is not a purely phonological matter, but often one of lexical diffusion as well.Thus, relic words can still remain in use.The urban vernacular of Constance provides interesting examples.Despite the general standard-dialect convergence, the remains of older phonological regularities have been maintained in some words and are even surprisingly stable, as the comparison with the 1920s study shows.One example is the realisation of MHG ei as /□/, a very small-scale development of the western Lake Constance region, which, however, is preserved in a handful of highly frequent words: ein 'one' (/on/), kein 'no' (/kon/), mein 'my' (/mon/) and weißt 'know (you)' (/voj7)9.The preservation of such lexical shibboleths appears to be typical for the development of modern local repertoires.On the one hand, it allows speakers to exhibit local identity, and on the other hand it also enables them to adapt the base dialect structurally to a regional variety with its wider communicative range.
' The high frequency of the latter word is due to the fact that it is also used as a tag question.

Horizontal convergence: standard-independent dynamics in the regional dialects
In the emergence of regiolects as a consequence of the advergence of the base dialects towards the standard the large cities such as Munich, Stuttgart or Cologne play an important role.
There is evidence that they influence the base dialects to a greater degree and more directly than the standard (cf.e.g.Biicherl 1982).However, since the regiolects by definition are closer to the standard than the base dialects, convergence towards the regiolect often indirectly unintended implies convergence towards the standard as well, although this may be more or less easily discernible (cf. the discussion of psychological and linguistic convergence on p. 4 above).For instance, if a North Bavarian speaker uses the Middle Bavarian diphthong /ub / (MHG no) as in bua (= standard German Bub 'boy') instead of the more traditional Northern Bavarian falling diphthong bou, we surely have evidence of convergence towards the Bavarian regiolect centred on Munich which is spreading into the North.However, and at the same time, there is an indirect convergence towards the standard, for the diphthong /uy / is phonetically closer to the standard long vowel /u:/ than the diphthong /on/, due to the similar onset.However, the spreading regional dialects/regiolects also include linguistic features which cannot be explained within the standard/dialect dimension but only within the dialect/dialect dimension (interdialectal convergence, koineisation).10Thus, while the convergence of dialect and standard explains many of the changes in present-day German, it does not tell the whole story.For example, if the same North Bavarian speaker takes over the vocalisation of syllable-final /l/ from Middle Bavarian (as in Wald, halt, will), this does not imply convergence towards, but rather a divergence from the standard variety (like the standard, Northern Bavarian does not vocalise the lateral).This, in fact, is what we observe.Further examples of this type are not rare: Convergence towards Swabian (North Alemannic) has brought the unrounded front vowels /y/ and /o/ to the area north of Lake Constance and thereby displaced the older (High Alemannic) form identical to the standard; /f/-coronalisation is expanding within the entire middle German region (Herrgen 1986); in the central Rhine area, older base dialectal diphthongal forms are replaced by regional dialectal monophthongal forms, like /e / for MHG /ei/, For details of these dimensions, cf.Auer & Hinskens, 1996.ID as in the word Kleid (cf.Bellman 1998), etc.

Conclusion
In sum, the dediglossation of the linguistic situation in Germany and Austria has led to extremely dynamic developments in the regional repertoires over the 20th century.Incidentally, the intergration of the GDR into the Federal Republic has rather strengthened than weakened the overall trend to regionalization 'above' the level of the base dialects.Although the base dialects were treated with some suspicion during most of the existence of the GDR (both the bourgeois revolution of 1789 and the workers' movements of the 19lh century looked upon the base dialects as a remnant of the dark ages of peasants' oppression, and the GDR seems to have followed this tradition), the linguistic situation there was characterised by the rivalry of two regiolects, both of which represent levelled varieties (koinai), and both of which are rich in tradition: the Upper Saxonian Vernacular and the Berlin Koine.11Divergence through these regiolects from West German regional standards and codified standard was tolerated if not functional for a conscious political and cultural separation.After the reunification, these regiolects (and their corresponding regional standards) continue to be used.
" The prestige of these varieties in the GDR is shown for example in the wellinvestigated fact that a variety closer to the standard was spoken in West Berlin than in East Berlin.Cf. , among others, Schonfeld & Schlobinski, 1995.
Finally, it should be stressed that while there is both interdialectal and standard-dialectal convergence going on in the German language area, there are also divergent forces at work.These, however, do not seem to be linked to the traditional dialects at all, but rather imply the use of ethnic and/or youth styles, of code-mixing andfusing (with a variety of languages), and of special registers (such as technical or professional jargons).*