A specter is haunting Europe: The Alps as a linguistic area?

Abstract As is well-known, the Alps are a zone of long-standing, intensive contact and multilingualism among Germanic, Romance and Slavic languages and varieties. In this introductory paper, some reflections are presented on the viability of the Sprachbund hypothesis to encompass the occurring phenomena of convergence observed throughout the whole area.

lands, economically specialized (typically in herding), and economically dependent on the lowland markets and winter pastures. They are often transhumant or partly transhumant, with part or all of the society moving between highland summer pastures and lowland markets and winter pastures. (Nichols 2015: 262) This picture describes fairly well the traditional state-of-affairs holding for several villages placed on the upper slopes of the Alpine massifs with regard to their lower neighbors speaking a different language. In fact, we can find in many cases socalled 'Burushaski distributions', named after the Burushaski language isolate in the western Himalayas (cf. Nichols 2015: 265). A Burushaski distribution characterizes varieties spoken in dialect forms on adjacent mountains along separate valleys of tributaries of a single river, as found for instance in the Walser German speaking villages of Alagna, Carcoforo and Rimella, while the lowland language spoken at the end of their respective valleys and below the confluence is Piedmontese, an Italo-Romance variety spoken in the Valsesia. The Walser German villages result from a medieval colonization of the upper parts of the valleys dug by tributaries of the river Sesia due to settlers speaking a variety of Highest Alemannic and coming from the northern slope of the crest mountain, the Swiss canton Valais/Wallis (hence their name Wal(li)ser). These villages are usually found at the very end of the valleys, separated by the intervention of other Romance settlements like Fobello, Rimasco, etc. Similar examples are scattered everywhere throughout the whole Alpine region.
In such a geographic-linguistic scenario, the lowlands are the centers of spread of languages, in particular loan vocabulary, as well as of cultural and economic advances. Thus, in the central crest mountain area the real geographic periphery consisting of the surrounding lowlands acts as the center of innovation, while the real geographic center, i.e. the highest highlands, acts as the periphery which typically preserves archaisms and traditional cultural and economic customs, just like the periphery of a classic dialect zone (cf. Nichols 2015: 269).
The peculiar character of such a central crest mountain system consists in the possible spread of linguistic features from all cardinal points giving rise to possible different convergence phenomena which involve in the lowlands languages belonging to the three main European linguistic families, namely Romance (from the West and the South), Germanic (from the North and the East) and Slavic (from the East). On the other hand, in the highlands, where "the pace of spreading is sufficiently slow that considerable diversity remains ... so that the central crest area as a whole is quite diverse" (Nichols 2015: 270), a certain number of isolated linguistic islands is found displaying peculiar characters (cf. Baechler 2016 for an attempt of evaluating the diversity of isolated linguistic islands with regard to the lowlands varieties).
In the dialectic relation holding between the dynamic lowlands and the conservative highlands within the Alpine central crest mountain system, several spread scenarios have been pointed out in the literature. Indeed, unidirectional grammatical borrowing has been reported. As for Germanic varieties, Mayerthaler and Mayerthaler (1990) have made the (controversial, cf. Rowley 2017) proposal that many syntactic traits of Bavarian German are actually pattern-borrowings from (Rhaeto-)Romance. In a similar vein, Ramat (1998) draws attention to different morphosyntactic features resulting from long-lasting contacts around the Alps, and particularly to what Wiemer (2011) in a survey on passive constructions has even termed the 'Alpine passive' based on the auxiliary verb COME. As for Romance, cases of verb second (Poletto 2002;Liver 2009) or DO-periphrasis (Benincà and Poletto 2004) have been reported in Alpine varieties, i.e. constructions that are highly reminiscent of similar patterns otherwise much better known from Germanic. As for Slavic, Reindl (2008) attributes a number of morphosyntactic features of (varieties of) Slovenian to the influence of Germanic varieties. Is this enough to invoke the presence of a proper Alpine Sprachbund? The latter is usually defined as "a geographically delimited area including languages from two or more language families, sharing significant traits (which are not found in languages from these families spoken outside the area)" (Aikhenvald and Dixon 2001: 11). Since the linguistic-geographic condition is largely satisfied as we can clearly identify the central crest mountain system where three distinct language families are included, the issue amounts to question the distinctiveness of the common traits mentioned above and further illustrated in the rest of the volume, which are likely to be found in the area.
On the other hand, the structural effects of long-standing language contact may be more complex than putative unidirectional grammatical borrowing. In this light, bidirectional contact-induced change may result in Alpine shared innovations (cf. Seiler 2004 on case marking and clitic doubling): developments that (i) are plausibly explained on the basis of language contact within the Alpine region, (ii) are shared among languages/varieties of different genetic affiliations, and (iii) are much less (or not at all) found in non-Alpine varieties of Germanic, Romance, and Slavic. Shared innovations seem to be the most striking examples for areal structural convergence within the Alpine area. Bidirectional changes, however, twist the knife in the wound of the old question of how two languages come to share the same feature. Since the three language families involved in the alleged Sprachbund share the same ancestor, this is a particularly delicate issue as similar innovations can be either due to true contact-induced convergence or to parallel development (Sapir's drift, cf. Aikhenvald and Dixon 2001: 3) which is likely to have taken place independently in each language, long after their genetic separation. In the following section we will briefly discuss one example where the question of drift is strictly interwoven with the issue of the linguistic islands as the more conservative parts of a central crest mountain system.

Looking for convergence in a multilingual
context: two case studies

Passive auxiliaries in Walser German
The typical Burushaski distribution hinted at in the previous section also involves the village of Gressoney which is found in a parallel valley further west of Alagna, from which it is separated by the massif of Monte Rosa (cf. Rizzi 1993 for historical reconstructions). Gressoney closes this narrow valley which borders on the Swiss Valais and is isolated from the other Walser German village of Issime by the intervening Romance-speaking village of Gaby. In such a peculiar geographic context at least four languages constitute the repertoire of a Gressoney speaker, namely the ancestral Walser German variety called Titsch, the variety of Piedmontese spoken in the rest of the valley, Standard Italian, mediated by the school and the media, and Standard French, which is recognized as official language in Aosta Valley and similarly taught in the school. It must be added that at least some speakers have familiarity with German varieties because of long-lasting personal contacts with Germany and Switzerland, as well as with Franco-Provençal as the latter is spoken in the close village of Gaby. Such a multifarious picture should be kept in mind when one looks at the highly complex system of passive constructions in Titsch (cf. Gaeta , 2020. In fact, at least three different constructions are found involving the auxiliaries si 'to be' (1a), chéeme 'to come' (1b) -the Alpine passive hinted at above -and goa 'to go' (1c): 1 || 1 The data are extracted from a text corpus collected thanks to an ongoing project on Italian minorities in Piedmont and Aosta Valley to which the reader is referred for the exact source (cf. Gaeta to appear for a survey and Angster and Gaeta's contribution in this volume).
(1) a. However, their value is not the same insofar as the BE-passive is the unmarked construction as it can be used without any particular meaning difference in the present as well as in the past (2a): (2) a. keis Grab esch gsid verlassen-z un vargässen-z no.N.SG grave[N] is been abandoned-N.SG and forgotten-N.SG 'No grave has been abandoned and forgotten' [DOK_0100] b. *Bés hit éscht z'ganz material ém Walserzentrum until today is the.N.SG=whole material [N] in.DET Walser.center kéemet henderzochen-z come.PSTPTCP look.PSTPTCP.after-N.SG int. 'Until today the whole material has been looked after in the Walser center' c.
Of jede fall éscht z'lied gsongen-z kanget on every case is the.N.SG=song [N] sung-N.SG gone 'At any rate the song has been sung' [DOK_0202] On the other hand, the COME-passive usually has a prospective meaning (1b) and cannot be used in the past (2b). Finally, the GO-passive only has a deontic value in the present (1b) while in the past it displays a plain passive value (2c) and is actually as frequent as the BE-passive. Besides the striking richness of constructions for conveying passive in Titsch, the question arises as to the role of the contact with Romance varieties. Note that the BE-passive is likely to represent the ancestral passive type as is testified in Old High German, while the BECOMEpassive based on the fientive auxiliary werden 'to become' normally used in Standard German is not found. If we look at the Romance languages occurring in the repertoire, we clearly find possible models both for the COME-passive and for the GO-passive (for convenience examples are drawn from Italian): Walser.center 'From now on the whole material will be looked after in the Walser center.' b. L-a benedizione va scritt-a qui su. the-F.SG blessing[F] goes written-F.SG here above 'The blessing has to be written thereupon.' In both cases, we find the same semantic nuances observed in the Titsch examples, insofar as respectively the COME-passive has a prospective and the GO-passive a deontic value. Furthermore, the corresponding past sentence of the COMEpassive is ungrammatical reflecting a general restriction on this passive construction with analytic past tenses, while the synthetic preterit is admitted (4a) On the other hand, past tenses are not compatible with the GO-passive, including the deontic meaning (4b). Given these structural similarities -which are completely absent in other German varieties -we can safely conclude that the COME-passive as well as the GO-passive with deontic value represent instances of unidirectional contact-induced change in Titsch.
More complex is the case of the GO-passive limited to the past tense, which is unknown in Italian as well as in other German varieties. Therefore, it is a good candidate for being a genuine Titsch innovation. The GO-passive is likely to come from the generalization of the so-called consumption passive (see Sansò and Giacalone Ramat 2016) in which so-called consumption verbs like verliere 'to lose' are found which involve the cancellation of the object by virtue of an uncontrolled or agentless process as in the following examples: (5)  While in the examples (5a-b) involving the consumption verb no agent can be inferred and the process is portrayed as uncontrolled and unstoppable, in the example (5c) the agent is not explicitly mentioned but can easily be inferred from the context. On the other hand, the inference is favored by the fact that the washing event can be seen as a natural process induced by the intervention of natural forces like the sun.
That this is the plausible origin of the GO-passive is further shown by the occurrence of the consumption passive in the contact variety, as shown by the following Italian example: On the other hand, a similar consumption passive, displaying the same properties found in Italian, is also found in German, as well as in English and in other European languages like French, Swedish, etc. Thus, the case of the GO-passive -restricted to the past tense -is rather to be interpreted as an autonomous innovation which has surely benefitted in terms of convergence from a bidirectional contact to the extent that both Italian and German display a similar pattern.

Case-marking and prepositions
We find striking similarities between Germanic and Romance Alpine varieties in the domain of case marking, too. It is difficult, and to some degree pointless, to determine a concrete directionality of borrowing (from Germanic to Romance or vice-versa). Instead, similarities in case systems are highly suggestive of structural convergence, and must therefore be interpreted as shared innovations within a geographically contiguous, but genetically heterogeneous area (cf. Seiler 2003Seiler : 239-241, 2004. In many Upper German dialects, dative objects are introduced by a prepositional case marker that is homophonous with the local preposition in in some dialects and an in others. Examples for this kind of prepositional dative marking (Seiler 2003: 15)  There is no doubt that there is a general tendency for case syncretism (nominative and accusative) or loss (genitive) in the overwhelming majority of German dialects if compared with Old High German or the modern standard language but, as Dal (1942Dal ( /1971 notes, distinct dative case morphology is strikingly robust in most High German dialects. As for the Alemannic dialect of Glarus, the following paradigms are representative of the morphological integrity of the dative case (whereas the nominative-accusative distinction is lost in forms other than personal pronouns): Thus, prepositional dative marking cannot be analyzed as a periphrasis that circumvades the use of dative inflections, but dative inflections are part of this construction since it is the inflected forms that are combined with the prepositional dative marker, cf. (7)-(8) above. Due to the selection of case forms prepositional dative marking cannot be viewed as a direct borrowing of the type of indirect object marking found in many Romance languages, i.e., using a case marker derived from Latin ad (cf. Italian a, French à etc.). For ad governs accusative, not dative. A hypothetical direct copy of this type of indirect object marking would look thus (but it does not occur; DM = dative marker, A = accusative): Hypothetical, pseudo-Romance prepositional dative marking in Alemannic: *er git dr Öpfel a mīx he gives the apple DM me:A 'he gives the apple to me' In sum, and somewhat paradoxically, Upper German prepositional dative marking has both a Romance flavor insofar as case relations are expressed by means of a prepositional marker, and a Germanic (specifically High German) flavor insofar as fully distinctive inflectional dative case morphology is used.
A representative example is given in (10). Note the striking structural similarity to the Alemannic example (8b) above: (10) Friulian [Marchetti 1952: 135-136] Inflectional case paradigm of the 1st singular personal pronoun (full form): From the Romance perspective, the conservation of distinctive dative case morphology in personal pronoun paradigms is a remarkable property (whereas this property would be fully expected in the relatively dative-friendly Germanic/High German varieties). However, dative case morphology is combined with the use of prepositional markers in the relevant dialects, and such markers are widespread in Romance as a whole. More generally speaking, the geographical picture that emerges can be summarized thus (Seiler 2003: 243): Distinctive dative case morphology is widespread in High German (and thus Germanic). Prepositional marking of indirect objects is widespread in Romance. Crucially, both features overlap in the Alpine region, such that many Alpine dialects express indirect objects by means of a combination of dative case morphology and prepositional dative markers. This combination occurs on both sides of the Germanic-Romance linguistic boundary, such that the southernmost Germanic and the northernmost Romance varieties show, with regard to this specific combination, greater structural similarity to each other than to other Germanic or Romance varieties. This combination is neither Germanic nor Romance: it is Alpine (Seiler 2004: 489).

Perspectives of an Alpine linguistics
The results from the contributions selected for the present volume suggest that structural convergence between Germanic, Romance and Slavic Alpine varieties is a fruitful, promising field for future research. In addition to the two examples above concerning the domains of voice (passive auxiliaries) and case (prepositional dative marking), we expect that phenomena of (stronger or weaker) areal convergence are very likely to be uncovered in all areas of grammar, namely in phonology (initial sibilant voiceless fricatives; front round vowels and the absence thereof, vocalic and consonantal quantity), verbal morphosyntax (loss of a synthetic past tense, future auxiliaries, progressive periphrases, do-periphrasis, pronoun-based verbal suffixes), nominal morphosyntax (rich pronoun morphology including series of clitics and/or reinforced pronouns such as noi altri, clitic doubling, predicative agreement), word formation (special causa-tives, verb particles, semelfactive abstracts [nomina vicis]), clause structure (verb second and deviations thereof, doubly filled complementizers, relativization), and spatial deixis (adverbs, prepositions, demonstratives).
A second bundle of tasks for a future Alpine linguistics is concerned with methodological challenges related to fieldwork, language documentation, use of (electronic) resources especially with lesser-used languages (for a brief survey regarding the German-speaking islands in Northern Italy, cf. Angster et al. 2017 and Gaeta to appear), quantification, and the interpretation of change as being contact-induced or not (cf. . This is especially urgent in light of quick phenomena of language decay affecting several varieties spoken in the area.

In this issue ...
In the issue, which includes selected papers from a Workshop held at SLE 2018 organized by Livio Gaeta (Turin) and Guido Seiler (Zurich), the common background consists in research questions relating to the possible existence of an Alpine linguistic convergence area based on convincing empirical case studies of Alpine contact-induced change, in particular shared innovations (in the sense as defined above) in phonological, morphological, and syntactic structure, especially with regard to other closely located convergence zones such as the Charlemagne area or the Balkan area. In this regard, the contributions focus on empirically well-grounded case studies as well as methodological issues related to the identification of Alpine structural features, i.e. features that are shared among varieties of Germanic, Romance, and Slavic spoken in the Alps, possibly due to structural convergence within the area.
In particular, Birgit Alber, Joachim Kokkelmans and Stefan Rabanus investigate one possible phenomenon of convergence of a phonological nature, namely the retraction of the alveolar sibilant /s/ before another consonant, e.g. /st/ > /ʃt/, which is found today in several varieties of the Alpine area, in particular in the eastern side of Alps. It has to be stressed that since Trubetzkoy shared phonological changes have traditionally been considered as crucial evidence in support a linguistic area. Relying on fresh data drawn from their own fieldwork as well as from dialect atlases, they show, however, that -in spite of clear evidence of the role of contact between the involved Germanic and Romance varieties -s-retraction cannot be considered a water-tight example of Sprachbund phenomenon, since it is attested only in languages which also underwent a specific set of sound changes. Thus, language-internal factors are likely to have played a substantial role in the emergence of the process. On the other hand, this does not exclude that also language contact might have played a supporting role, in the light of the pattern similarity observed in all languages displaying s-retraction.
In Giuliano Bernini's contribution, the diffusion of particle-verbs in Rhaeto-Romance and Italo-Romance dialects is surveyed on the basis of a selection of thirteen maps of the Sprach-und Sachatlas Italiens und der Südschweiz (= AIS), presenting the responses to twelve input sentences referring to different events. In this way, the attempt is made to identify the role of communicative habits such as the deictic specification of actions/states as potential sources of particle-verb constructions in the Alpine. The latter are an inherited feature in Germanic and an innovation in the Alpine Romance dialects, which lends particular interest to the phenomenon in a convergence perspective. Bernini is able to show that -in spite of the clear role played by contact -a number of different factors have to be taken into consideration which give as a result a complex picture. Particle-verbs are more frequent in the coding of motion than nonmotion events, while the geographic diffusion appears to be biased towards the Central and the Eastern Alps and the Central Po Plain. The distribution of particle-verbs is shown to correlate with the presence of systems of topographical deixis and to result from processes of language shift from Romance to Germanic in the mountains. Especially interesting from the point of view of the Burushaski distributions hinted at above, their diffusion in the plain results from inter-dialectal contact along cross-Alpine trade routes.
On cases of isolation focuses Marco Angster and Livio Gaeta's contribution which emphasizes that despite isolation conservative patterns can also evolve into innovative strategies. To illustrate this, causative and progressive constructions in the historical Walser German minority varieties on the southern side of the Alps are surveyed. In particular, in Greschòneytitsch the remarkable development of a causative particle, tònz, as well as the grammaticalization of an adverb, eister, into a marker of progressive aspect are discussed. The interest of these developments lies on the one hand in the clear role played by contact in such an isolated but multilingual environment. On the other hand, these phenomena also reflect possible processes of convergence in the area in neat contrast with other varieties of German which happen to be in different contact contexts. What is more, these peculiar developments witness an original elaboration of different features in a clear contact situation.
Progressive periphrases are also investigated by Rossella Maraffino who focuses on their occurrence in the areas of Swiss Grisons, Trentino-Alto Adige and Friulian Carnia, trying to explain the mechanism of adaptation or re-elaboration of the borrowed structure in the replica languages. In this way, she is able to pinpoint which of this structure replication seems to be the result of an internal development witnessed in the Alpine area. As a result, the progressive expression in the Alpine area is shown to represent a continuum which overcomes the dialectal barriers, as in the same varieties endogenous and exogenous forms coexist. Even when not all the processes concerning the collected forms represent a shared innovation, but rather a unidirectional borrowing path, a certain type, centered on the source particle BEHIND, marks the progressive aspect in all the languages of the analyzed area and qualifies as possible feature corroborating the Alpine Sprachbund-hypothesis.
Oliver Schallert and Ermenegildo Bidese focus on a peculiar morphosyntactic trait found in Alpine varieties, namely the occurrence of double complementizers: the violation of the so-called Doubly-filled COMP Filter (= DFC), especially in the context of embedded questions, occurring in Germanic (e. g. Alemannic, Bavarian), Romance (e. g. Rhaeto-Romance, Friulian, Ladinian, Venetian), and even Slavic varieties (e. g. Slovenian, Croatian). Even though this phenomenon is not exclusively restricted to this contact zone, the respective varieties show interesting convergences as well as differences in its grammatical properties. A look at the more fine-grained distributional facts in Romance reveals some similarities, but also important differences to Germanic. Whereas there is some indication for a structural parallelism between Germanic and Romance in terms of the wh-items involved, both language groups are set apart when it comes to DFCs in root sentences. Furthermore, since DFCs have been reported for a number of varieties lying outside the alpine region proper, the occurrence of this trait must be regarded as a polygenetic structural option in the C-domain.
Malinka Pila's contribution is the one with the strongest focus on Slavic in the present volume. Pila investigates the venitive passive -and thus a class of phenomena that directly relates to the case study of Walser German in Section 2.1 above -in two isolated Slavic micro-varieties in the south-eastern Alps, namely Resian and Carinthian Slovene. Whereas Resian has been in longlasting contact with neighboring and dominating Romance varieties, in Carinthian Slovene Slavic-Germanic contact is more central. Whereas the origin of the Resian venitive passive is clearly due to contact with Romance, Pila shows that as for Carinthian Slovene, where the phenomenon does occur, too, though with a relatively low frequency, the decision whether it is due to Italian or Carinthian-South-Bavarian influence is much more difficult to make. The findings in the two isolated Alpine Slavic varieties are compared with Molise Slavic, i.e., another Slavic minority language in Italy, but in the south (Puglia).
Finally, Ermenegildo Bidese and Alessandra Tomaselli place the context of the Alpine varieties within the broader picture of Sprachbund phenomena pro-posing a methodological (and theoretical) differentiation between surface forms and structural differences. In particular, while continua refer primarily to surface forms -for instance, for all Romance dialects north and south of the Brenner Pass we find the obligatory realization of the 3rd person singular, unlike Standard Italian and similar to the Germanic varieties -the theoretical status of these convergent forms is very different. In the Italian dialects this can be shown to be a purely morphological phenomenon assuming its status as prefix, while these varieties can be argued to be null subject languages. On the contrary, Bavarian dialects are clearly non null subject languages with the obligatory expression of the expletive subject. In modeling the continuum the traditional Wellentheorie is revisited taking the parameter values to be the stable centers of core syntactic options. At the boundaries of the circles the languages show convergence phenomena and a consistent overlapping of surface, i.e. linear, patterns that derive from divergent parameter settings.
In sum, although such convergence phenomena might affect only single parts of the Alpine space as they mostly focus on dialectal varieties, they testify of a widespread multilingualism diffused throughout the whole area in which different centers can be identified clearly displaying shared innovations. It is important to stress that the latter can be opposed to other developments found outside of the area as for instance in the case of grammaticalization instances relating to the verbal complex. This substantially enriches the picture of convergence phenomena as it is traditionally described in Europe (see for instance van der Auwera and Van Olmen 2017, where several cases of contact of Germanic with Romance and Slavic varieties are surveyed although no reference to the Alps as a convergence area is made).
At a more general level, the issue of an Alpensprachbund has to be seen in connection with the distinction between standard and non-standard varieties which is made particularly urgent by the presence of standard varieties serving as Dachsprachen in the area at least since the second half of last century. In several cases, the convergence phenomena discussed in the contributions correspond to similar developments found in substandard varieties of the national languages, also testified across the national borders, such as for instance the development of particle or phrasal verbs.
This opens new ways for future research on language contact in Europe insofar as the traditional studies centering on convergence phenomena at a dialectal level need to be integrated by specific enquiries focusing on the role and the properties of the substandard varieties of the national languages found in the area. It is hoped that the contributions in the volume will settle the stage for such new investigations to come.