2.1. Brief description of the current language situation

In our study, we concentrate on three German linguistic islands in northern Italy: Sauris as well as Gressoney and Issime. The speakers of Sauris and Gressoney/Issime have no contact with each other because of the geographical distance, the former located in the easternmost part of northern Italy and the latter in the westernmost part. In all three communities, an Upper German dialect is spoken: in Sauris this is a Bavarian one, while in Gressoney and Issime it is an Alemannic one (Walser German).

Sauris, or “Zahre” in the local dialect, is the highest municipality in the Friuli-Venezia Giulia region (Province of Udine) (with Sauris di Sopra at 1,400 meters) and is situated in an isolated valley of the Carnic Alps (see e.g. Geyer Reference Geyer, Eller-Wildfeuer, Rössler and Wildfeuer2018). Today, it is a very well-kept village, famous for its bacon and popular among hikers. Sauris nowadays has around 400 inhabitants, of whom about 200 speak and use Zahre German (see Rabanus et al. Reference Rabanus, Kruijt, Alber, Bidese, Gaeta and Raimondi2023–). According to a survey conducted in 2018, 77.8% of the inhabitants always speak Zahre German with their parents, 52.4% with their peers, 13% with their children, 41.4% with their neighbors, and 33.3% with their colleagues at work (Costantini Reference Costantini2023:79).

Gressoney and Issime are in the Lys Valley, in Valle d’Aosta, close to Monte Rosa. Located at the end of the valley, Gressoney consists of two separate municipalities (though, in this case, one speech community), Gressoney-Saint-Jean and Gressoney-La-Trinité, at an altitude of 1,385 meters and 1,630 meters respectively; Issime is lower at 953 meters (Zürrer Reference Zürrer1999: 25). Today, Gressoney is a well-known ski resort, while Issime is largely unknown when it comes to winter tourism (Zürrer Reference Zürrer1999:28). According to Zürrer (Reference Zürrer1999:386), there are about 500 active German speakers in Gressoney (56% of the total population of Gressoney-Saint-Jean and Gressoney-La-Trinité) and 240 in Issime (65% of the total population of Issime). However, this estimate dates back to 1992 and it can be assumed that the number of speakers has declined even further since then, especially in Gressoney, whose ski resort mainly attracts Italian visitors (see Vicquery Reference Vicquery2006:127–129).

Today, the dialects of all three communities (Sauris, Gressoney, Issime) are on the verge of extinction (see Dal Negro Reference Dal Negro2004:42–45). Although efforts are being made to preserve their idioms (i.e. teaching the dialect in elementary school), current figures show that the dialect is hardly being passed on to the next generation (see Vicquery Reference Vicquery2006:131–133, Costantini Reference Costantini2023:79).

2.2. Historical bi- and multilingualism: a (short) overview

Sauris, settled in the thirteenth century from neighboring, Bavarian-speaking Austria, has been bilingual with Friulian and the German idiom since at least the mid-nineteenth century (Geyer Reference Geyer, Eller-Wildfeuer, Rössler and Wildfeuer2018:97–98, Moser Reference Moser2025:47–50). Compulsory education was introduced in the nineteenth century, which meant that Italian was learnt as an official and standard language and was used in formal contexts (Protto Reference Protto, Heller, Prader and Prezzi2009:280). In the first half of the twentieth century, Zahre German was spoken in the family and among the inhabitants of Sauris, while Friulian was used in contact with neighboring communities, and Italian was necessary in official and formal contexts (Denison Reference Denison and Sture Ureland1979:28–29). In the twentieth century, Zahre German lost much of its prestige; however, since the 1970s, the dialect has regained prestige, and there are efforts to preserve the culture and the dialect (Protto Reference Protto, Heller, Prader and Prezzi2009:269–271).

Although the dialect of both Gressoney and Issime is called “Walser German” and although the two communities are located in the same valley, it is important to note that they differ from each other in terms of their (historical) language contact situation (see below). This difference has also had consequences for the linguistic structure today: The dialects of Gressoney and Issime are quite distant – Gressoney is indeed more similar to Formazza or to other Walser dialects than to Issime – and people from Gressoney speak Italian with people from Issime.Footnote ² For this reason, they are also treated as two different speech communities in this article.

The population of Gressoney was, from 1400 until the twentieth century, monolingual in the German idiom; there was no permanent settlement when the first settlers from the Valais in Switzerland arrived in the Middle Ages (cf. Zürrer Reference Zürrer2009:87–91, 97; see also Giacalone Ramat Reference Giacalone Ramat1979). Issime, however, was a mixed group of settlers, with two coexisting groups speaking different idioms: Walser German and Francoprovençal (cf. Zürrer Reference Zürrer2009:86–92). In contrast to Gressoney, Issime was settled before the arrival of German settlers. Thus, since the colonialization of this area, Issime has consisted of two different groups of speakers; it was not until the eighteenth century that the Francoprovençal part of the population joined with the German-speaking part (Zürrer Reference Zürrer2009:96). In the nineteenth century, in formal contexts, written (Standard) German was used in Gressoney and written (Standard) French in Issime (Zürrer Reference Zürrer2009:88). From the 1920s onwards, however, written German and written French were gradually replaced by (written) Italian (Zürrer Reference Zürrer2009:99-100). Furthermore, since the end of the nineteenth century, Piedmontese has also been used as a lingua franca (e.g. in contact with local traders) (Zürrer Reference Zürrer2009:98–99). Italian, on the other hand, has become an everyday language in Gressoney for the first time among speakers born in the early 1950s (Zürrer Reference Zürrer1999:33, Dal Negro Reference Dal Negro2004:55).

2.3. Language contact and language attrition

We define language contact as “a bilingual practice reflecting the processing/use of language in a bilingual speaker’s repertoire” (Riehl Reference Riehl, Schmid and Köpke2019:314). The contact situation in Gressoney, Issime, and Sauris is characterized by a long-standing bi- or multilingualism for generations: the speakers use the German dialect side by side with Italian, the official language, which is present in education, the media, and in administration. The dialect is used in the domestic context and, depending on the language skills of the interlocutor, in further oral interactions as well (e.g. with friends or with colleagues at work) (see Zürrer Reference Zürrer2009:100–117, Costantini Reference Costantini2023). In the past, many speakers also spoke a local Romance variety (Friulian, Piedmontese, or Francoprovençal), but our survey indicates that this is hardly the case today or for our speakers (see section 4.1). In terms of language structure, the Romance contact varieties require a lexical subject pronoun (in contrast to standard Italian); they differ from each other with regard to the degree of obligatoriness/optionality of the subject clitic and in relation to the grammatical person in which the subject clitic is used.

The bi- or multilingualism attested in both Zahre German and Walser German has grown historically, that is, the speakers have been in close contact with the surrounding Romance varieties for several generations. It is therefore likely that these varieties are susceptible to contact-induced language change: “intense contact is more likely than casual contact to result in typological restructuring of the receiving language” (Thomason Reference Thomason, Haspelmath, König, Osterreicher and Raible2001:1640). It is also possible that the typological proximity or distance of the languages in contact play a role. If the languages/varieties are too similar, typological change cannot occur: “Clearly, not all contact-induced changes alter the typology of the receiving language. Some of them are typologically inert: either they fit neatly into the typological patterns of the receiving language, or they are too minor to constitute a significant typological change” (Thomason Reference Thomason, Haspelmath, König, Osterreicher and Raible2001:1640). Note that while the Germanic varieties under scrutiny and Standard Italian are typologically distant (null-subject language vs. subject language), other Romance contact varieties are somewhere in between, as they do have subject clitics (see sections 2.4 and 2.5). Instead of using the term of contact-induced typological change (which entails the difficulty of defining when precisely typological change has taken place), Matras & Sakel (Reference Matras and Sakel2007) prefer to speak of “grammatical borrowing” to describe language contact phenomena. Part of this is the borrowing of words and morphemes as well as the change of structures on the morphological and syntactic level. Grammatical borrowing is triggered by “the need to reduce the cognitive load when handling a complex linguistic repertoire” (Matras Reference Matras, Matras and Sakel2007:67). As far as changes in syntax are concerned, it is particularly important to ascertain whether the change is contact-induced or if internal factors may have played a role (see Bidese Reference Bidese2023:1–68). Word order changes, for instance, are susceptible to language contact (cf. Heine & Kuteva Reference Heine, Kuteva, Aikhenvald and Dixon2001:395–396 for African languages; see also section 2.5). Conversely, there are (morpho)syntactic phenomena that are more resistant to language change. An illustrative example is that of subject clitics in speakers of linguistic islands in northern Italy (see Poletto & Tomaselli Reference Poletto and Tomaselli2018, Rabanus & Tomaselli Reference Rabanus, Tomaselli, Christen, Gilles and Purschke2017).

The language attrition approach is closely linked to research in language contact: Both “investigate the impact of a second language (L2) on the first (L1)” and “the linguistic features found in individual attrition and typical language contact settings are similar” (Riehl Reference Riehl, Schmid and Köpke2019:315). The two usually differ in terms of the speaker populations: “Whereas typical attrition research analyses first generation migrants in an L2 environment, language contact research focuses more on the use of languages in multilingual speech communities” (Riehl Reference Riehl, Schmid and Köpke2019:315).

Changes taking place in small communities like those mentioned above can be triggered not only by mechanisms linked to language contact and language attrition but also by second language acquisition strategies (see Kolmer Reference Kolmer2010:148). The interface hypothesis is known from second language acquisition research (see Sorace Reference Sorace2011). According to Sorace (Reference Sorace2011), phenomena at the syntax–pragmatics interface (so-called phenomena with interpretable features) are more prone to language change than other, for example, purely syntactic, phenomena. In their study, Tsimpli et al. (Reference Tsimpli, Sorace, Heycock and Filiaci2004) examined subject pronouns in multilingual speakers of languages that differ in the pro-drop parameter: The study encompassed Italian and Greek speakers (L1) and near-native speakers of English (L2) and yielded evidence for attrition in L1 in interpretable features of subjects, among others in topic continuation vs. topic shift. No attrition was observed in noninterpretable, syntactic features, such as pro-drop and non-pro-drop. In the context of L1 attrition, the interface hypothesis predicts that after processes of generational transmission “there may be changes that concern the whole pronominal system and/or structures that involve different interface conditions” (Sorace Reference Sorace2011:26). Studies in the field of language acquisition (L2) but also in the field of language contact and diachronic change indicate that overt subject pronouns, in contrast to null subjects, enjoy an “unmarked default status” (Sorace Reference Sorace2011:20).

2.4. Pronominal subjects

Standard Italian and German (with its varieties) differ in the way they express pronominal subjects. German is a language “in which simple sentences with pronominal subjects normally if not obligatorily contain a pronoun in subject position. … [S]ubject position is defined as the position in the sentence in which nominal subjects appear” (Dryer Reference Dryer, Dryer and Haspelmath2013). In the following examples from Standard German, the pronoun is obligatory. Moreover, in both (1a) and (1b) the subject is in the first position because it is a topic. Note that the structural position of the subject (i.e. the position in which it gets the agreement with the finite verb) is the postverbal one for both the nominal and the pronominal subject:

In certain registers of German (e.g. computer-assisted communication) and under certain extra-linguistic contextual conditions (familiarity of the interlocutors, informal context), sentences without a subject are grammatical as well: The subject can thus be omitted if it can be reconstructed from the conversational/discourse context (see Helmer Reference Helmer2016). Note, however, that this type of V1 sentence (with a null subject in initial position) does not represent the phenomenon of Topic-drop (at least for V2 varieties).

Italian, on the other hand, is a language “in which the normal expression of pronominal subjects is by means of affixes on the verb” (Dryer Reference Dryer, Dryer and Haspelmath2013). In theoretical terms, Italian is a null-subject language (cf. e.g. Dryer Reference Dryer, Dryer and Haspelmath2013) or a pro-drop language (cf. Chomsky Reference Chomsky1981:240–242). Example (2b) illustrates the dropping of the pronominal subject (“pro” refers to the null subject); as in example (1), the structural position of the subject is the postverbal one for both the nominal and the pronominal subject:

Put simply, the subject can be omitted in Italian if it can be inferred from the context, whereas “[t]he pronoun is more likely to be overtly expressed in sentences where there is a shift in topic with respect to the previous discourse” (Frasson et al. Reference Frasson, D’Alessandro and van Osch2021: 6). This difference is illustrated by Frasson et al. (Reference Frasson, D’Alessandro and van Osch2021:6–7) with the following examples from Spanish (which has pro-drop as well). Example (3) shows topic continuation: “the topic of the second sentence is the same as in the first sentence, which makes a null subject the most felicitous option”; by contrast, example (4) shows topic shift: “the subject changes from the 1st person to a 3rd person, thus requiring an overt pronoun (or a full DP)” (Frasson et al. Reference Frasson, D’Alessandro and van Osch2021:6–7):

In Upper German dialects – including our linguistic islands in northern Italy – as well as in spoken German the subject pronoun can be realized in the form of a clitic. This is possible in a context of subject–verb inversion where the subject clitic sticks to the finite verb (see e.g. Lessiak Reference Lessiak1963, Merkle Reference Merkle1993, Zürrer Reference Zürrer1999, Cattarin Reference Cattarin2020). The subject–verb inversion is found in interrogative sentences (see example (5)) and in sentences with an XVS word order (with scl for “subject clitic”):

Unstressed pronominal forms in German (and in general in Germanic varieties) are realized in the so-called Wackernagelposition, which is on the top of the so-called Mittelfeld. Wackernagelposition is “right-adjacent to the finite verb in main clauses and to the complementizer in embedded clauses” (Bidese Reference Bidese and Putnam2011: 363). Germanic varieties (and all Germanic linguistic islands) have not developed proclisis at all, that is, a subject clitic is expected just in the inversion context X Vfin+clitic.Footnote ³ Object clitics cannot appear as proclitic to the finite verb either (see Bidese Reference Bidese and Putnam2011). Furthermore, the linguistic islands in northern Italy all have structural V2 (see Bidese et al. Reference Bidese, Padovan, Tomaselli, Woods and Wolfe2020), i.e. subject–verb inversion is still possible even if the linear restriction is lost.Footnote ⁴ In order to obtain a declarative clause, at least one position before the finite verb must be realized (see, e.g., Tomaselli & Bidese Reference Tomaselli, Bidese, Bidese, Casalicchio and Caterina Moroni2019:52–53). In the following example from Cimbrian, we have a context of subject–verb inversion, and the position before the finite verb is filled (X Vfin+clitic):

On the other hand, the Romance contact varieties of interest – Friulian, Piedmontese, and Francoprovençal – are characterized by (optional) subject clitics occurring in proclisis to the finite verb in declarative clauses (not so, however, in certain other contexts such as interrogative clauses, where they appear in enclisis; for a comprehensive account of Romance subject clitics in general see, e.g., Poletto & Tortora Reference Poletto, Tortora, Ledgeway and Maiden2016).

Friulian displays a complete paradigm of subject clitics that occur in all syntactic contexts, in addition to the inflected verb (Benincà Reference Benincà and Mucignat2014:44–45; for (a few) exceptions, see Benincà & Vanelli Reference Benincà, Vanelli, Ledgeway and Maiden2016; see also Hack & Gaglia Reference Hack, Gaglia, Kaiser and Remberger2009:157). According to Benincà (Reference Benincà and Mucignat2014:43), Friulian “has evolved more than any other [northern Italian] dialect and has removed practically all restrictions to the appearance of subject clitics.” For instance, in Friulian a subject clitic is almost always obligatory in main clauses, even when the sentence already has a lexical subject (see (9c); Benincà Reference Benincà and Mucignat2014:43–44; Benincà & Vanelli Reference Benincà, Vanelli, Ledgeway and Maiden2016:150). In this case, one could also speak of subject doubling (see section 2.4). The following examples illustrate the obligatory use of the subject clitic (here third person singular) in main clauses in Friulian:

Also in Piedmontese subject clitics usually precede the finite verb in declarative sentences (see example (9)): they are “mandatory in the written language (except perhaps in poetry for prosodic reasons) and in many varieties in main clauses … and tensed dependent clauses” (Tosco et al. Reference Tosco, Miola and Duberti2023:177). However, first-person subject clitics in particular may be absent in local varieties and are often omitted in allegro speech (Tosco et al. Reference Tosco, Miola and Duberti2023:177–80; see also Parry Reference Parry1993 for a diachronic account of subject clitics in Piedmontese).

As for Francoprovençal, the usage of subject clitics varies considerably from one variety to another (Kristol Reference Kristol, Ledgeway and Maiden2016; see also Kristol Reference Kristol2009). In Val d’Aosta, subject clitics are always obligatory for the second person singular but may be optional or absent in the remainder (cf. Diémoz Reference Diémoz2008).

Note, in conclusion, that in all these varieties subject clitics for the second person singular are always compulsory; see, in this respect, the generalizations on the different systems of subject clitics in Romance varieties formulated by Renzi & Vanelli (Reference Renzi and Vanelli1983).

2.5. Pleonastic occurrence of subject clitic and subject pronoun

Standard Italian shows no subject doubling because of the lack of a subject clitic paradigm (Tomaselli & Bidese Reference Tomaselli, Bidese, Bidese, Casalicchio and Caterina Moroni2019:49). In Germanic and Northern Italo-Romance varieties, however, a clitic subject can also appear in a declarative sentence, together with a full subject pronoun. If the co-presence of a subject clitic and overt subject (either in the form of a personal pronoun or a lexical NP) is not pragmatically motivated, one can speak of “pleonastic occurrence.” Consider, for instance, the following example from Walser German:

This pattern is not restricted to German linguistic islands in northern Italy but is possible also in other Germanic varieties such as Bavarian (see e.g. Bayer Reference Bayer1984) or Dutch dialects (e.g. West Flemish or the dialect of Wambeek, cf. Haegeman Reference Haegeman, van Riemsdijk and Rizzi1991; van Craenenbroeck & van Koppen Reference van Craenenbroeck, van Koppen, van Koppen, Thrift, van der Torre and Zimmermann2002):

The pleonastic occurrence of two subjects in Zahre German is briefly mentioned by Denison (1988:29–30), who ascribes it to the influence of Friulian (but see Moser Reference Moser2025:71–73, who argues in favor of a convergence process). In Zahre German, the pleonastic occurrence is almost exclusively attested in context of subject–verb inversion, mainly in interrogative sentences and not in declarative sentences with XVO:

In Bavarian, as well as in Cimbrian, a Bavarian-speaking linguistic island in northern Italy, a lexical subject or a subject pronoun that occurs to the left of the finite verb cannot be doubled (Tomaselli & Bidese Reference Tomaselli, Bidese, Bidese, Casalicchio and Caterina Moroni2019:50). Consider the following example:

We speak of “subject doubling” provided the case that there are some pragmatics involved (pragmatic strengthening). In the example in (15) we find subject doubling as endschandre ‘WE’ is a stressed pronoun, which occurs together with a clitic subject pronoun -wer (see Bauen Reference Bauen1978: 69):

In northern Italian varieties, on the other hand, subject clitics occur in proclisis and may occur together with a subject pronoun/lexical NP that is left to the verb. In Friulian or Piedmontese, for instance, the occurrence of a subject clitic that agrees in number and person with the full subject is possible or, depending on the variety, even obligatory. The word order in declaratives is S + Sclitic + Vfin (D’Alessandro Reference D’Alessandro2009:1).

Full pronouns can be doubled as well, but only under certain pragmatic conditions: In the following example, the full subject pronoun jo ‘I’ is only placed under emphasis, and is otherwise omitted (cf. Haiman Reference Haiman, Traugott and Heine1991:138):

In summary, our speakers could possibly realize the pleonastic occurrence of a subject and subject clitic in two different ways: (i) the Romance pattern with (S) + Sclitic + Vfin and/or (ii) the Germanic pattern with (X) + Vfin + Sclitic + (S). The results are discussed in section 5.

2.6. Theoretical status of subject clitics

In the literature, the status of subject clitics has been repeatedly discussed, as they are somewhere between (independent) words and morphologically bound affixes (see, e.g., Zwicky & Pullum Reference Zwicky and Pullum1983, Spencer Reference Spencer1991:350–392, Nübling Reference Nübling1992:56–86). Subject clitics in German dialects (including in German linguistic islands in northern Italy) are so-called “special clitics” as “the clitics do not have the same distribution as the corresponding full forms” (Zwicky & Pullum Reference Zwicky and Pullum1983:510). The clitics usually represent a phonologically reduced form of the full form, as exemplified in table 1 (for the sake of clarity, we have not included all the variants for each form, but have instead opted for one or a maximum of two variants per form; an overview of all variants can be found in Zürrer Reference Zürrer1999:207–208 for Walser German and Cattarin Reference Cattarin2020:50–51 for Zahre German).

Table 1.

Full forms and clitic forms of subject pronouns

Table 1 Long description

A table comparing full forms and clitic forms of subject pronouns in Walser German and Zahre German. The table has four rows and four columns. The columns are labeled ‘Walser German full’, ‘Walser German clitic’, ‘Zahre German full’, and ‘Zahre German clitic’. The rows are labeled with singular and plural pronouns. Row 1: 1.ps. iich, i, i, -i. Row 2: 2.ps. dou, de, t, ø, du, -te. Row 3: 3.ps.m./f./n. eer/dschi/ääs, er/dschi/äs, ar/si/is, -ar/-se/-is. Row 4: 1.ps. wiir, wir, ber, bier, -ber. Row 5: 2.ps. iir, ir, der, ier, -er. Row 6: 3.ps. dschii, dsch, si, -se.

Subject clitics in German dialects are usually regarded as subject clitics and not as verbal affixes (Kolmer Reference Kolmer2012:74–79). The two types of clitics differ in their distribution (Kolmer Reference Kolmer2012:74–79): subject clitics have a certain position in the sentence and are therefore syntactically distributed (second-position clitics); verbal affixes, on the other hand, are morphologically distributed, that is, bound to a certain part of speech (verbal clitics). Clitics in northern Italian dialects (and in Romance varieties in general) are referred to in this way (Kolmer Reference Kolmer2012:74–79; certain subject clitics are even obligatory in some varieties, see Benincà et al. Reference Benincà, Parry, Pescarini, Ledgeway and Maiden2016:195–196). Word order changes from V2 to SVO in German linguistic islands in northern Italy have led to a discussion regarding the extent to which these clitics are still syntactically distributed or, conversely, whether their distribution is rather morphologically governed (Kolmer Reference Kolmer2012:223). The pleonastic occurrence of subject pronouns is interpreted as an indication of the clitic element approaching the status of an inflectional affix, i.e. the clitic is getting more grammaticalized, fulfilling agreement functions rather than referential functions (see Giacalone Ramat Reference Giacalone Ramat, Kastovsky and Szwedek1986:324, Reference Giacalone Ramat and Rizzi1989:42–43, Reference Giacalone Ramat and Spinozzi Monai1990:88, Reference Giacalone Ramat, Gerritsen and Stein1992:327 for Walser German dialects; following Giacalone Ramat, see e.g. Dal Negro Reference Dal Negro2004:165, Kolmer Reference Kolmer2012:79–80, 223–226). According to Kolmer (Reference Kolmer2012:203), the grammaticalization status of the clitic is more advanced in Walser German than in German linguistic islands of Bavarian origin (Cimbrian in particular). She therefore suggests that in Walser German, in contrast to Cimbrian, subject clitics will evolve into morphological markers (see Kolmer Reference Kolmer, Bidese, Dow and Stolz2005, Reference Kolmer2012: 203). This is evidenced by the obligatoriness of subject clitic pleonastic forms in contexts where the full pronoun is doubled by a subject clitic.

A similar, quantitative study on the grammaticalization of the enclitic subject pronoun in linguistic islands in northern Italy, namely in Walser German spoken in Formazza, has been carried out by Dal Negro (Reference Dal Negro2004:159–180). Remember that, as mentioned in section 2.2 (note 2), Gressoney is more similar to Formazza or to other Walser dialects than to Issime. The study was based on freely structured interviews, with a total of 23 informants (see Dal Negro Reference Dal Negro2004:103–105). According to this study, third-person plural enclitic pronouns “present the highest percentage values of grammaticalized forms” (Dal Negro Reference Dal Negro2004:176). Three types of verb–subject clitic sequences were counted as grammaticalized forms: “verb initial utterances, cases of pleonastic subject pronouns and cases of inversion after a coordinative or subordinative conjunction” (Dal Negro Reference Dal Negro2004:172). Cases in which subject–verb inversion was syntactically motivated (see section 2.7) were not included, since they are “particularly ambiguous in that it is not possible to decide whether the inversion is triggered by the syntactic context of whether it conceals a reanalysis of the pronoun” (Dal Negro Reference Dal Negro2004:172). According to Dal Negro (Reference Dal Negro2004:176), the high number of evidence in third-person plural enclitic pronouns can be attributed to the fact that third-person subjects can often be full nouns which helps “to mask the pronominal nature of grammaticalized clitics and to favour the occurrence of pleonastic forms.” Dal Negro (Reference Dal Negro2004:176–178) also observes that grammaticalized forms are used more frequently by less traditional dialect speakers (younger speakers born between 1970 and 1979) as well as in informal contexts with friends or at work.

Therefore, Dal Negro’s (Reference Dal Negro2004) study suggests that the entry point for the enclitic pronouns as grammaticalized affixes is the third person plural. By contrast, according to other authors such as Hotzenköcherle (Reference Hotzenköcherle, Bindschedler, Hotzenköcherle and Kohlschmidt1971:93), second-person plural pronouns can be seen as an entry point to the grammaticalization process, followed by first-person plural pronouns (note that this assumption is not based on a quantitative study).

2.7. Word order

German (including its dialects) is a strict V2-language, that is, the finite verb always appears in second position in a declarative clause. If a constituent other than the subject appears in the first position of the clause, German creates a XVS word order, with the verb appearing in second position. This subject–verb inversion avoids ungrammatical V3. Strictly speaking, German is an asymmetrical V2-language in that in subordinate clauses the verbal complex (including the finite verb) appears in last position (SOV).

In German dialects spoken in northern Italy, word order is more variable, most likely due to language contact: From a typological point of view, these dialects can be classified as standing between German (word order with V2 and non-null-subject language) and Italian (word order without V2 and null-subject language) (see Bidese & Tomaselli Reference Bidese, Tomaselli, Cognola and Casalicchio2018). In these dialects, the first position in a declarative sentence can indeed be occupied by more than one constituent, and there is no V2-effect; this word order is known from the Romance contact varieties. In the following example, both an AP and a NP appear before the finite verb. It is also noteworthy that, as illustrated in this example, the German bracket construction is missing in these dialects: the finite verb hatt and the part participle gisek are not separated (i.e. they do not form a bracket), but appear next to each other:

4.1. Participants

Our study involves 32 speakers: 16 speakers for Sauris, 9 for Gressoney, and 7 for Issime (= 16 for Walser German in total). The speakers were provided by our local contacts (native dialect speakers who live there and know the community). Sauris nowadays has around 400 inhabitants, of which about 200 speak and use Zahre German (see Rabanus et al. Reference Rabanus, Kruijt, Alber, Bidese, Gaeta and Raimondi2023–). Our speakers thus represent 8 percent of the population. Unfortunately, there are no current figures for Walser German. Zürrer (Reference Zürrer1999:386) states that 740 people actively speak Walser German: 240 people speak the dialect of Issime (65 percent of the population of Issime) and 500 the dialect of Gressoney (56 percent of the population of Gressoney). According to this, the number of speakers of our study would therefore represent 3.3 percent of the Walser German speaking population of Issime (with 240 speakers of Walser German in Issime as 100 percent) and 1.6 percent of the Walser German speaking population of Gressoney (with 500 speakers of Walser German in Gressoney as 100 percent) Nevertheless, it can be assumed that the number of speakers has continued to decline since that time: In fact, a study from 2006 on the active dialect competence of primary school children revealed that only three children out of a total of 53 spoke the German dialect (Vicquery Reference Vicquery2006: 131–132). This is due to numerous mixed marriages in the 1970s and 1980s, in which the dialect was not passed on (Vicquery Reference Vicquery2006:129). Thus, we suggest that 16 speakers of Walser German should represent about 8 percent of the population as well.

Each speaker had to provide information on his/her sociolinguistic background; see tables 2–4 for an overview. All of them have acquired the dialect as L1 (with their parents being native dialect speakers) and learned Italian as a surrounding language (as L1 too), or at the latest as L2 at primary school (see table 2). One person from Sauris also indicated Friulian as L1 in addition to Zahre German and Italian.

Table 2.

Language acquisition

Table 2 Long description

A table comparing three dialect communities – Gressoney G., Issime German, and Zahre German – each with two sub-columns: L1 (first language) and L1 parents. Rows show: ‘dialect only’ L1 speakers numbering 5, 5, and 8 respectively; ‘dialect and Italian’ L1 speakers numbering 4, 2, and 8; ‘both parents native dialect speakers’ numbering 8, 7, and 11; and ‘one parent native dialect speaker’ numbering 1, 0, and 5.

Table 3.

Language use^a (with I = Issime, G = Gressoney, S = Sauris)

Table 3 Long description

The table has 9 rows and 6 columns. The columns are labeled ‘With the family’ and ‘With friends’, each with sub-columns ‘G’, ‘I’, and ‘S’. The rows are labeled with different language groups. Row 1: Walser/Zahre G., G 2, I 3, S 8, G 0, I 1, S 4. Row 2: Italian, G 2, I 0, S 2, G 1, I 0, S 1. Row 3: Walser/Zahre G. and Italian, G 5, I 3, S 6, G 5, I 3, S 7. Row 4: Walser/Zahre G., Italian, Francoprovençal/Friulian, G 0, I 1, S 0, G 0, I 1, S 2. Row 5: Walser G., Italian, Piedmontese, G 0, I 0, S 0, G 0, I 0, S 2. Row 6: Walser G., Italian, German, English, G 0, I 0, S 1. Row 7: Walser G., Italian, French, G 0, I 0, S 1. Row 8: Italian, Friulian, G 0, I 0, S 1. Row 9: NA, G 0, I 0, S 1, G 0, I 1, S 1.

^a We also asked about the language(s) used with colleagues at work. However, we received hardly any feedback in this respect (presumably because they were either already retired, self-employed in their own company, or only marginally employed).

Table 4.

Level of education

Table 4 Long description

A table comparing levels of education across three locations: Gressoney, Issime, and Sauris. The table has five rows and four columns. The columns are labeled Gressoney, Issime, and Sauris. The rows are labeled Primary school, Secondary school, Further education, University, and NA. Row 1: Primary school, Gressoney 0, Issime 3, Sauris 4. Row 2: Secondary school, Gressoney 1, Issime 1, Sauris 3. Row 3: Further education, Gressoney 7, Issime 3, Sauris 8. Row 4: University, Gressoney 1, Issime 0, Sauris 0. Row 5: NA, Gressoney 0, Issime 0, Sauris 1.

Table 5.

Frequency of enclitic pronouns in declarative sentences in the three communities (with I = Issime German, G = Gressoney German, Z = Zahre German)

Table 5 Long description

The table presents the frequency of enclitic pronouns in declarative sentences across three communities: Issime German, Gressoney German, and Zahre German. It has 5 rows and 12 columns. The columns are grouped into three sections: to (do) drink, to go, and can/may, each with sub-columns for Z, G, and I. The rows are labeled as enclitic only, pleonastic, full subject pronoun only, finite verb only, and total. Each cell contains a value representing the frequency and percentage of occurrences. For example, in the to (do) drink section, the enclitic only row shows 11 (9 percent) for Z, 11 (16 percent) for G, and 0 for I. The table provides a detailed breakdown of how different types of pronouns are used in declarative sentences across the three communities.

The great majority of our speakers is older than 60 years and less than a third is younger than 60. Only three speakers (two from Sauris and one from Gressoney) are 40 years old or younger. On average, the speakers of Issime German are somewhat older than the speakers of Zahre German and of Gressoney German: the mean age is 72 years for speakers of Issime German, 68 years for those of Zahre German and 59 years for those of Gressoney German. In both populations we interviewed 13 female speakers and three male speakers. Among the speakers of Walser German, two men and seven women were interviewed in Gressoney, and one man and six women in Issime.

The German idiom is mainly spoken within the family, especially in the case of Zahre German (see table 3). In contact with friends, either the dialect or Italian is spoken, depending on the conversation partner; here, too, more speakers from Sauris claim to speak the dialect only. As far as the use of Romance varieties (Friulian, Piedmontese, and Francoprovençal) is concerned, the data show that very few speakers use them in everyday life (with the family or with friends). Note that one speaker from Gressoney and one speaker from Sauris did not answer the question about the language(s) used with friends.

As far as education/schooling in the three speech communities is concerned (see table 4), approximately one-fifth of the speakers only attended primary school (elementari); slightly fewer went on to secondary school (scuola media). More than half of the interviewees have completed further education, that is, practical training, apprenticeship or technical college (istituto professionale, liceo), while only one person from Gressoney attended university. Altogether, the educational level of the interviewees is highest in Gressoney, followed by Sauris and Issime. Note that one person from Sauris did not wish to provide any personal details in relation to this question (“NA” in table 4).

4.2. Experimental design

The goal of our experiment was to test the use of enclitic subject pronouns and full subject pronouns. Note that our elicitation method tends to elicit speakers’ competence and their interiorized norm, while free speech, as analyzed by Dal Negro (see section 2.6), elicits performance. Possible differences in the results could therefore also be related to the design of our survey.

The questionnaire consisted of different text types. To begin with, we tested the presence/absence of subject clitics in affirmative declarative sentences (see examples (21)–(23)). Importantly, the stimuli were embedded in a context and the subject of the stimuli represented a continuation topic and not a topic shift: In (21), for instance, the friends in question are already introduced in the preceding sentence. The stimulus itself then contains the subject pronoun. For this type, we asked different grammatical persons (n = 7, with third person singular masculine and feminine). Four possible options are known from the literature (cf. Zürrer Reference Zürrer1999, Dal Negro Reference Dal Negro2004), which in the following are called (i) “enclitic only,” that is, the subject pronoun is realised in the form of an enclitic pronoun; (ii) “pleonastic,” that is, there are both a full subject pronoun and an enclitic one; (iii) “full subject pronoun only”; (iv) “finite verb only”. Remember that both (i) and (ii) are counted as grammaticalized forms.Footnote ⁶ As illustrated by the following examples, these options were tested with different verb types (auxiliary, modal verb, short verb), in order to determine whether the use of subject clitics is influenced by different verb types.

A further question type was designed to test whether a subject doubling form occurs more often in the context of topicalization and emphasis on the subject or not. This could then be interpreted as an indication that the pleonastic occurrence of two (pronominal) subjects has an emphasizing or reinforcing function. This sentence type was tested in two versions, both of which are documented in a very similar form in the literature (see Zürrer Reference Zürrer1999:335, 341):

Finally, we wanted to test the participants’ use of enclitic subject pronouns, depending on the word order. In all cases there was a continuation topic, that is, the subject was already known from the context (previous sentence). The test consisted of one sentence type, presented in two variants, each covering all grammatical persons. In variant 1, the sequence was S (pronoun) V O (see example (23) above), while in variant 2 the sequence was X V S (pronoun), as in the following example: