2.1. Infinitive Syntax in Standard German

SG has two main forms of the infinitive: bare infinitives (“1. Status” infinitives, according to Bech Reference Bech1955), marked with the suffix -en (e.g., les-en ‘read’), and zu-infinitives (“2. Status” infinitives), marked with the suffix -en and the left-adjacent particle zu (e.g., zu les-en ‘to read’), which may be analyzed as a prefix (Zifonun et al. Reference Zifonun, Hoffmann and Strecker1997:2159, Eisenberg Reference Eisenberg2020:382–383).Footnote ⁴ Which (status) form is used depends on syntactic factors, such as the syntactic head governing the infinitive (Statusrektion ‘status government’; Bech Reference Bech1955). Modal verbs like können (‘can’) or wollen (‘want’), for instance, govern “1. Status” infinitives (8), while propositional verbs like behaupten (‘claim’) or bestreiten (‘deny’) govern “2. Status” infinitives (9) in SG.Footnote ⁵

ICs such as in (8) and (9) also vary in relation to “coherence” (see, for example, Bech Reference Bech1955, Kiss Reference Kiss1995, Zifonun et al. Reference Zifonun, Hoffmann and Strecker1997:2186–2205, Wurmbrand Reference Wurmbrand2001, Müller Reference Müller2002:37–46, Osborne Reference Osborne2005, Haider Reference Haider2010:274–286, Rapp & Wöllstein Reference Rapp, Wöllstein, Meibauer, Steinbach and Altmann2013). Coherent and incoherent ICs differ in how “close” the connection between the matrix verb and the infinite is (Eisenberg Reference Eisenberg2020:391). While incoherent ICs are often considered a separate constituent, coherent infinitives are regarded as part of the verbal cluster (for a formalist model see Haider Reference Haider2010:274–284). Thus, examples like (8) can be interpreted as monoclausal structures, while examples like (9) are biclausal structures, which include an independent infinitive clause (often called satzwertiger Infinitiv ‘clause-value infinitive’; Duden 2022:118–120, Rapp & Wöllstein Reference Rapp, Wöllstein, Meibauer, Steinbach and Altmann2013; for a critical review see Kiss Reference Kiss1995:14-20). We will not go into the problem of this differentiation in detail, as our analyses focus on incoherent ICs in Austrian dialects.

Incoherent ICs with zu can have different syntactical functions, which we will illustrate with examples from SG. To start with, they can be an object (10) as well as the subject (11) or an adverbial (12). In the latter case, the IC is usually combined with a complementizer, such as um (‘for’) or anstatt (‘instead’). Moreover, zu-ICs can be complements of nonverbal heads like adjectives (13) or (often deverbal or deadjectival) nouns (14) (Rapp & Wöllstein Reference Rapp, Wöllstein, Meibauer, Steinbach and Altmann2013:341-346).Footnote ⁶

However, none of these functions can be completely fulfilled by zu-ICs. With regard to ICs in complement function, it is noteworthy that there are no predicates in SG selecting exclusively such constructions.

The most important alternative to zu-ICs are subordinated clauses with the complementizer dass (‘that’) (see (2) and (4) above). In SG, clauses with dass can replace ICs in many contexts (see (15) and (16)), but not all (see (17) and (18)) (Eisenberg Reference Eisenberg2020:68).

Various studies have attempted to determine which factors modulate the choice of zu-ICs and dass-constructions in such contexts in SG (see, for example, Zifonun et al. Reference Zifonun, Hoffmann and Strecker1997:1449, Wöllstein Reference Wöllstein and Eichinger2015, Rapp et al. Reference Rapp, Laptieva, Koplenig and Engelberg2017, Brandt & Bildhauer Reference Brandt, Bildhauer, Fuß, Konopka and Wöllstein2019). It has been argued that, although ICs are more economical, they are less explicit. This is because they have no overt subject, only limited morphological markers for tense, and have no morphological marker for mood (see, for example, Rapp et al. Reference Rapp, Laptieva, Koplenig and Engelberg2017:196).Footnote ⁷ Therefore, for instance, zu-ICs are preferred with (control) verbs where the covert subject of the infinitive tends to correspond to a complement of the matrix clause (e.g., in versuchen ‘try’). Where this is not the case (e.g., in erzählen ‘tell’), dass-constructions may be preferred. According to Rapp et al. (Reference Rapp, Laptieva, Koplenig and Engelberg2017:219), these preferences can be “entrenched” and thus ultimately become part of the verb valence.

2.2. Infinitives in (Upper) German Dialects

ICs already occur in the earliest stages of German, that is, in Old and Middle High German (Schrodt Reference Schrodt2004:5–9, Paul Reference Paul2007:314–315, Dal & Eroms Reference Dal and Eroms2014:108–124) – this applies to both ICs with bare infinitives (19) and ICs with zu. The latter were grammaticalized from a combination of the preposition zu (‘to’) with a nominal infinitive (Haspelmath Reference Haspelmath1989, Smirnova Reference Smirnova2017). In the historical stages of German, the infinitive regularly – but not always – showed dative inflection; see (20) for Old High German (for Middle High German see Paul Reference Paul2007:314; for Early New High German see Ebert Reference Ebert, Reichmann and Peter Wegera1993:396).

Not only is the question controversial concerning when the zu-ICs became fully grammaticalized as a verbal form, but also how this grammaticalization process took place (Haspelmath Reference Haspelmath1989, Demske Reference Demske, Müller and Reis2001, Abraham Reference Abraham2004, Smirnova Reference Smirnova2016, Reference Smirnova2017). In any case, (various) German dialects do not seem to have grammaticalized such ICs to the same extent as has SG (Schallert Reference Schallert, Abraham and Leiss2013:105–107). In what follows, we will focus on the situation in the Upper German dialects, in particular (Central) Bavarian, which have been most thoroughly researched to date – except for Bavarian dialects in Austria (see footnote 3).

In Bavarian dialects in Germany, ICs with zum, among other variants (see section 1), are frequently used instead of zu-ICs in SG. However, as Bayer (Reference Bayer, Bayer and Abraham1993:52–53) argues, there might be some “deep rooted differences” between both constructions. For example, zum-ICs allow for noun incorporation (21), and they may occur without particle separation (22). In contrast, in SG, zu always appears left-adjacent to the verb stem, and nominal constituents governed by the infinitive as well as separable verb particles appear obligatorily on the left, (23), (24).

These facts – along with some other distributional differences between zu-ICs in SG and zum-ICs in Bavarian – are taken as evidence by Bayer (Reference Bayer, Bayer and Abraham1993: 58) “that B[avarian] retains the prepositional nature of zu.” A similar claim is made by Donhauser (Reference Donhauser, Koller, Wegstein and Richard Wolf1989), who concludes that zum-ICs in Bavarian show a “more nominal character” than zu-ICs in SG. For instance, she finds that ICs in older Bavarian are mostly used as adverbials or attributes, with either directive or purposive semantics. Due to (contact-induced) change, only very recently have ICs been used as verbal complements, although in most cases still without any complements governed by the infinitive. Consequently, dass-sentences remain the main strategy to date to produce propositional complements (Donhauser Reference Donhauser, Koller, Wegstein and Richard Wolf1989:295–298).

Weiß (Reference Weiß1998) holds a different view. He argues that there are already verbal zum-ICs present in older Bavarian corresponding to zu-ICs in SG (Weiß Reference Weiß1998:235). Weiß (Reference Weiß1998) also distinguishes between two diachronically distinct “infinitive systems” in Bavarian: In the older system, zum-ICs are already frequently employed in numerous functions (Weiß Reference Weiß1998:240). Specifically, he maintains that zum-ICs are used (a) as complements of nouns and adjectives, (b) as adverbials (corresponding to adverbials with um zu ‘in order to’ in SG), (c) with noun incorporation, or (d) as part of the verbal cluster in coherent ICs with certain matrix verbs (e.g., kriegen ‘get’). Although some of these ICs may be interpreted as prepositional phrases (e.g., ICs with noun incorporation; Weiß Reference Weiß1998:252–255), zum-ICs are clearly verbal in others (e.g., in coherent ICs; Weiß Reference Weiß1998:247–252, 262–263). In the more recent system, these functions of Bavarian zum-ICs are extended by their general use as complements in incoherent ICs. According to Weiß (Reference Weiß1998:265–272), this was triggered by contact with SG, a process which expanded verb by verb (starting with phase verbs like anfangen ‘start’; see also Donhauser Reference Donhauser, Koller, Wegstein and Richard Wolf1989:297). Thus, although there are still some differences in the complexity and frequency of ICs, ultimately, the infinitive syntax in Bavarian is becoming increasingly similar to SG. The remaining differences between Bavarian and SG can mainly be attributed to modality-related factors, such that Bavarian is used nearly exclusively as a spoken variety, while SG is both a spoken and a written variety (Weiß Reference Weiß1998:262, 272).

Weiß (Reference Weiß1998:235–239) also discusses ICs with proclitic z’, which he regards as an abbreviated form of zum in present-day Bavarian (although diachronically it may have evolved from zu, as Weiß argues). According to Weiß, z’ and zum are synonymous and can often be used to fulfil the same functions. However, z’ is more restricted, and cannot occur, for example, with noun incorporation (bier z’tringa, see (7), but not *z’bier tringa). In addition, the use of z’ is constrained in certain phonological and morphological environments (for example, in stressed positions or with prefixed infinitive verbs; see Weiß Reference Weiß1998:245–248). Some further constraints in the use of the clitic are investigated in Bayer & Brandner (Reference Bayer, Brandner, Patocka and Wiesinger2004). They suggest, for example, that the clitic is restricted to transitive infinitives combined with weak indefinite quantifiers (e.g., etwas ‘something’ or nichts ‘nothing’; see also Donhauser Reference Donhauser, Koller, Wegstein and Richard Wolf1989:296). In addition, the matrix verb has to be an “empty verb,” that is, a verb which has the primary goal of highlighting a presupposition of existence (e.g., haben ‘have’ or kriegen ‘get’).

In addition to zum- and z’-ICs discussed above, there are two other forms of ICs in Bavarian corresponding to SG zu-ICs, namely ICs with bare infinitives and zu-ICs. This is already indicated by data from the Wenker corpus, reported in Schallert (Reference Schallert, Abraham and Leiss2013).Footnote ¹⁰ With respect to the sentence Es hört gleich auf zu schneien (lit. it stops soon ptcl to snow ‘It’s about to stop snowing’), zum-IC are widespread in the central and eastern parts of Austria,Footnote ¹¹ whereas zu-ICs are frequently reported in the south and west, in particular in Tyrol. In Tyrol (as well as in Alemannic Vorarlberg and neighboring Switzerland), ICs with bare infinitives are also used (Schallert Reference Schallert, Abraham and Leiss2013:120), for example, Es hört gleich auf schneien. Scheutz (Reference Scheutz2016:127) confirms these results for South Tyrol (Northern Italy).

Taken together, these findings point towards geographical variation in Bavarian dialects (see also Mayerthaler et al. Reference Mayerthaler, Fliedl and Winkler1995).Footnote ¹² That said, since most research on Bavarian dialect syntax focuses on Central Bavarian in the German federal state of Bavaria, the constraints on the use of ICs in the Bavarian dialect areas in Austria are yet comparatively unclear.

For Alemannic dialects, the situation is different. Here, the widespread use of ICs with bare infinitives is well documented. As Brandner (Reference Brandner2006:208), among others, notes, “Alemannic allows bare infinitives in lexical environments … and syntactic positions (extraposition) where SG uses a zu-infinitive.” However, while ICs with bare infinitives can replace SG zu-ICs in some contexts (for example, with phase verbs or in subject function), this is not possible with others (for example, as complements of propositional or factive verbs) (Brandner Reference Brandner2006:213; see also Schallert Reference Schallert, Abraham and Leiss2013:106-107, 119-120).Footnote ¹³ Consequently, other variants are also regularly used in Alemannic, for example, subordinate clauses or nominalizations. In addition, zum- and z’-ICs are documented, which may occur for purposive clauses or as attributes of nouns and adjectives (for purposive clauses in Alemannic see also Seiler Reference Seiler, Eggers, Erich Schmidt and Stellmacher2005). In such ICs, zum and z’ can be doubled in Alemannic (Brandner Reference Brandner2006:215, Schallert Reference Schallert, Abraham and Leiss2013:121), with zum acting as a complementizer and preceding all elements governed by the infinitive (25) (Brandner Reference Brandner2006:216-219).

Brandner (Reference Brandner2020:29) also reports on intra-Alemannic variation, and claims that in Alemannic varieties in Germany, zum generally acts “as a lexical variant of zu.”Footnote ¹⁴

Overall, the evidence to date concerning the variation of ICs in Upper German (Bavarian and Alemannic) paints an incomplete, inconclusive, and even contradictory picture, which is not least due to the extremely heterogeneous data base. Against this background, our aim in the present contribution is to draw on a larger and more consistent empirical dataset from Austria, in order to more comprehensively describe (a) the variation of ICs, (b) their linguistic constraints, and (c) their recent changes.

3.1. Research Design, Research Locations, and Informants

The data of the present study comprise dialect recordings that were carried out by trained fieldworkers.Footnote ¹⁶ The dialect survey was conducted in 40 small, rural villages throughout Austria. Figure 1 shows their locations within the dialect regions of Austria (based on the dialect classification of Wiesinger 1983, which essentially used phonetic criteria).

Figure 1. Research locations (the abbreviations are explained in table A1 in the Appendix).

As figure 1 indicates, by far the most research locations are in the Bavarian dialect area, consisting of Central Bavarian, the South-Central Bavarian transition zone, and South Bavarian. In addition, several locations are situated in the Bavarian-Alemannic transition zone and the small Alemannic dialect region in the west of Austria. Alemannic dialects differ from Bavarian dialects in many features of grammar, lexis, and phonology (for an overview on the Austrian dialects see Lenz Reference Lenz, Joachim Herrgen, Fischer and Ganswindt2019; for a recent account of syntactic variation in Austria see Vergeiner et al. Reference Vergeiner, Bülow and Elspaßforthcoming).

Four speakers of the traditional dialect from each location participated in our study.Footnote ¹⁷ Two informants were chosen from an older (65+ years) and two from a younger (18–35 years) generation, with one male and one female per age group. Based on the apparent-time hypothesis, comparing the data of these two age groups facilitates not only an investigation of dialect variation, but also of recent dynamics in dialect change.

Our sample consists of 163 informants. Apart from age and gender, traditional dialectological criteria for sampling were applied (Chambers & Trudgill 1998:13–31). The older speakers are typical NORM/Fs (= nonmobile, older, rural males/females), and the younger informants can be considered prototypical speakers of the traditional dialects as well. They were all raised in local artisanal or agricultural networks and had not completed any higher education. Their parents were also born and raised in the same location. Both their social and working lives took place in the same local environments.

3.2. Items and Method

We used a traditional dialect questionnaire for this survey. During the interviews, all 163 informants completed the same test battery comprising translation tasks, cloze tasks, and picture naming tasks. In this study, we focus on the translation tasks, in which the informants translated sentences containing an incoherent IC with zu into their own dialects; the stimulus sentences were presented to the informants in a SG variety by the explorers. The informants were encouraged to use phonetical, morphological, and syntactical features which they considered most natural in their everyday dialect usage.

Translation tasks have been employed successfully in prior studies to provide insights into the variation with regard to ICs; see Mayerthaler et al. (Reference Mayerthaler, Fliedl and Winkler1995), Scheutz (Reference Scheutz2016), or Brandner (Reference Brandner2020), among others. Mirror effects, that is the adoption of given standard language structures, can never be completely ruled out in translation tasks. However, as the results in section 4 below show, most dialect realizations deviate from the structure presented in the stimulus sentence, which indicates that the design worked well.Footnote ¹⁸

Table 1 shows the 14 sentence stimuli of interest for the present study.

Table 1. Items

A total of 2,213 tokens are included in this study. Missing data resulted either from nonresponse items or from translations that differ considerably from the stimulus, for instance, because a different matrix verb was used. For example, some informants translated item #5 not with the given phase verb anfangen (‘start’) but with the copula verb werden (‘become’) and a present participle (26).

For the sake of comparability, we excluded such constructions from our study.

As column 3 in table 1 indicates, the ICs have different syntactical functions in the stimulus sentences:

• In one item, the IC is the subject (#1) of the matrix verb freuen (‘please’).

• In seven items, the ICs can be interpreted as an object (#2–#8). Five of these items (#4–#8) contain a phase verb (anfangen ‘start’, aufhören ‘stop’) as matrix verb.Footnote ¹⁹ One item (#2) has a propositional verb (glauben ‘believe’) and one item (#3) a causative verbFootnote ²⁰ (helfen ‘help’).

• In four items, the IC is an adverbial (#9–#12) – in three cases it is a purposive or final adverbial introduced by the complementizer um ‘for’ (#9–#11), in one item it is a contrastive adverbial introduced by the complementizer anstatt ‘instead (of)’ (#12).

• In the remaining items, the IC is an attribute within the accusative object of haben (‘have’). Both ICs have pronominal heads, consisting of the indefinites etwas (‘something’) (#13) and nichts (‘nothing’) (#14).Footnote ²¹

Moreover, the items differ in terms of the complexity of the IC in the stimulus sentence. Six items contain an extended infinitive, that is the ICs consist of additional phrases governed by the infinitive verb. Among other things, this allows us to investigate noun incorporation and the incorporation of pronouns.

In terms of research question 3, which addresses linguistic constraints on the use of IC variants, one of the main aims of the present study is to identify which variants are employed in the aforementioned syntactic contexts. As indicated in section 2.2, one might expect that the use of different infinitival and noninfinitival constructions depends heavily on the structure of the stimulus sentence.

4.1. Subject Function

In item #1 (Es freut mich, dich zu sehen), the (extended) IC in the stimulus sentence is a verbal complement with subject function, governed by the matrix verb freuen (‘be happy’). The results for this item are presented in table 3, including both the general findings and the findings for each age group individually.

The dominant variant for item #1 is not an IC but an alternative construction, namely the dass-clause, which is realized in two-thirds of all cases. Consequently, only in one-third of all cases item #1 is translated with an IC with either zum, zu or z’ (ICs with bare infinitives are not used at all by the informants). Significantly, zum-ICs occur less often when compared to most other items (and when zum-ICs are used, the pronoun is never incorporated).

The results suggest that there are limitations to the use of ICs – in particular ICs with zum – with subject function in the investigated dialects. However, as table 3 shows, there are notable differences between older and younger speakers regarding the use of zum-ICs. While older speakers rarely employ zum-ICs, this variant occurs much more often with younger speakers. In turn, younger speakers use dass-clauses less often, which might indicate an apparent-time change.

The geographical patterns are shown in the upper left map (#1) in figure A1 in the Appendix. The most noticeable pattern is that zu-ICs dominate in western South Bavarian (in Tyrol and the western part of Carinthia), while zum-ICs are generally more frequent in eastern Austria.

4.2. Object Function

In item #2 (Maria glaubt, sie zu kennen), the (extended) IC is an object of the propositional verb glauben (‘believe’). Table 4 shows the results for this item.

Table 4. Results for item #2 (Maria glaubt, sie zu kennen)

Obviously, most informants use alternative constructions for item #2 – either dass-clauses or V2-clauses (27).

In contrast, ICs are rarely used – in other words, their use seems to be restricted when there is a propositional verb in the matrix clause. Notably, there is no indication of an apparent time change for this item, as the results for older and younger speakers are virtually identical (see table 4). In addition, there are no significant geographical patterns visible (see figure A1, #2).

In item #3 (Sie haben uns geholfen, das Feuer zu löschen), the (extended) IC is an object of the causative verb helfen (‘help’). Table 5 shows the results for this item.

Table 5. Results for item #3 (Sie haben uns geholfen, das Feuer zu löschen)

For item #3, most informants employ ICs in object function. The most frequent variant is the IC with zum (in all but three cases from the Alemannic dialect region without noun incorporation, perhaps because the noun phrase is specific and definite). ICs with zu or z’ are also used, but much less than the zum-IC. Moreover, informants use ICs with bare infinitives quite often, but very rarely resort to alternative constructions like dass-clauses or nominalizations.

As for change in apparent time, there are considerable differences between the responses of older and younger speakers. Younger speakers use fewer ICs with bare infinitives and z’ than older speakers, but more ICs with zum. Geographically (see figure A1, #3), the most noticeable pattern is again that zu-ICs dominate in western South Bavarian, while zum-ICs seem to appear more often in eastern Austria. In the western half of the country, there are also more ICs with bare infinitives and ICs with z’.

The patterns are similar for items #4 (Nach der Firmung fing er an, Bier zu trinken), #5 (Ich glaube, es fängt an zu regnen), #6 (Es war so kalt, dass es anfing zu schneien), and #7 (Er fing an zu tanzen). In all these items, the phase verb anfangen (‘begin’) is the matrix verb. In item #4, the IC is extended, and in items #5 and #6, the matrix phrase containing the ICs is in a subordinate position. The results for these items are shown in table 6, 7, 8, and 9, respectively.

Table 6. Results for item #4 (Nach der Firmung fing er an, Bier zu trinken)

Table 7. Results for item #5 (Ich glaube, es fängt an zu regnen)

Table 8. Results for item #6 (Es war so kalt, dass es anfing zu schneien)

Table 9. Results for item #7 (Er fing an zu tanzen)

ICs are frequently used for these items, in particular ICs with zum. Strikingly, ICs with zum are least frequent with item #4, in which the IC is extended. When zum-ICs are used in this item, the noun is incorporated in 32 of 72 instances, as in example (28).

ICs with zum are more frequent with the items #5 and #6, where the matrix phrases containing the ICs are themselves subordinate clauses, and most frequent in item #7, where the IC is neither extended nor deeply embedded. While ICs with zu occur with similar frequencies in all four items, ICs with bare infinitives occur less often in item #7 and ICs with z’ in item #6. When speakers resort to alternative constructions to ICs, they usually employ nominalizations.

Comparing older and younger speakers, the most noticeable pattern is again that younger speakers use more zum-ICs, with item #4 (containing an extended IC) showing the greatest difference between older and younger speakers. The spatial distribution of the variants (see figure A1, #4 to #7) indicates once more that ICs with zu predominate in western South Bavarian, while ICs with bare infinitives are generally more frequent in the western half of Austria, and zum-ICs in the eastern half.

In item #8 (Jetzt höre ich auf zu fragen), the IC is an object of the phase verb aufhören (‘stop’). Table 10 shows the results for this item.

Table 10. Results for item #8 (Jetzt höre ich auf zu fragen)

Again, ICs occur very often – in particular ICs with zum. As table 10 reveals, younger speakers use these zum-ICs much more often, and as the map in figure A1 (in the Appendix) shows, zum-ICs predominate everywhere, except for western South Bavarian, where zu-ICs prevail.

4.3. Adverbial Function

In the stimulus sentences of item #9 (Ich kam ins Tal, um zu arbeiten), #10 (Sie geht auf den Markt, um die Kirschen zu verkaufen) and #11 (Du bist noch nicht groß genug, um eine Flasche Wein alleine auszutrinken), the IC is used as a purposive or final adverbial introduced with um. In item #9 the IC is not extended, while in both #10 and #11, the verb governs a direct object (#10: die Kirschen ‘the cherries’, #11: eine Flasche Wein ‘a bottle of wine’). Item #10 contains a prefixed verb and item #11 a particle verb. Furthermore, items #9 and #10 include the motion verbs kommen (‘come’) and gehen (‘go’) as matrix verbs, while the matrix clause in item #11 contains a complex predicate consisting of a copula verb and an adjective. The tables 11, 12, and 13 show the individual results for the ICs used.

Table 11. Results for item #9 (Ich kam ins Tal, um zu arbeiten)

Table 12. Results for item #10 (Sie geht auf den Markt, um die Kirschen zu verkaufen)

Table 13. Results for item #11 (Du bist noch nicht groß genug, um eine Flasche Wein alleine auszutrinken)

ICs – in particular, ICs with zum – are the predominant constructions used for item #9, where the stimulus contains no extended IC. Notably, in most of the translations (71 of 107 cases) when zum-ICs are used, the informants did not employ the complementizer um (29). One instance where both um and zum co-occur is presented in (30). In contrast, with zu-ICs, um is always realized (31).

The usage of zu and zum with preceding um in (30) and (31) illustrates why zum cannot be interpreted as a preposition in such zum-ICs: The complementizer um can govern only clauses and not prepositional phrases (see the discussion in section 2.2 and below).

For item #9, alternative constructions, mostly finite clauses with dass, are quite rare. This is different for item #10 and #11, in which dass-clauses (32) – and for item #11 also V2-clauses (33) – appear more often.

However, both for items #10 and #11, informants frequently use zum-ICs. This variant occurs in about one-third of all cases, and for item #10, it is even the most frequent variant. In items #10 and #11, the complementizer um can be used (34) or omitted (35, 36, 37) in co-occurrence with zum-ICs. In the latter case, zum can precede the infinitive verb (35) or the incorporated noun phrase of the IC (36), or is doubled (37). Note that the incorporated noun phrase is in most cases indefinite (and non-specific), as in (36).

In item #10, um is not realized in 36 of 60 cases, with 28 cases showing noun incorporation and only one translation by a speaker from the Alemannic region evincing zum-doubling. In item #11,Footnote ²³ um is omitted in 23 of 53 cases. In all these instances, zum precedes the noun phrase and the adverbial and thus may act as a complementizer. In five cases, we also find zum-doubling. Notably, these forms are exclusively documented in the Alemannic dialects. In items #10 and #11, adverbial zu-ICs are used both with (38) and without um (39), although the latter variant appears less frequently – for item #10 it occurs in 3 out of 24 cases, for item #11 in 6 out of 19 cases.

While ICs with zu occur with similar frequencies, there is also an interesting difference between the two items concerning the usage of adverbial ICs with clitic z’ and bare infinitives. ICs with bare infinitives seem to be restricted to item #10 (40), while ICs with z’ practically never occur for this item. In contrast, for item #11, we find ICs with z’ (41) but hardly any bare infinitives.

This difference could be explained, first, by the fact that item #10 contains a prefixed verb and item #11 a particle verb. As for the use of z’ in co-occurrence with prefixed verbs, our data confirm Weiß’s (Reference Weiß1998) observation. Second, regarding the usage of bare infinitives, the fact that the matrix clause of item #10 includes the motion verb gehen (‘go’) as predicate seems to be relevant. Even in SG, gehen can be used with a bare infinitive, when it is not used as a full verb but a semi-auxiliary verb (see Demske Reference Demske2020 for a discussion). This restriction to the verb gehen could also explain why IC with bare infinitives are virtually absent from item #9, where the motion verb kommen (‘come’) is used in the matrix clause. In SG and other present-day varieties of German, kommen can only occur with a past participle when employed as a semi-auxiliary verb (e.g., sie kamen gelaufen ‘they came running’; cf. Vogel Reference Vogel2005).

Comparing both age groups, we find that for all three items the share of zum-ICs increases considerably from older to younger speakers while the proportions of the other variants decrease accordingly in most cases. The geographical patterns (see figure A2, #9 to #11, in the Appendix) are more diverse. Interestingly, for all three items it is the Alemannic west, where zum-ICs are most dominant. In the neighboring Alemannic dialect areas of Eastern Switzerland, the zum-IC is also the dominant variant (see Glaser Reference Glaser2021, e.g., vol. 1:333, vol. 2:207). In eastern and central Austria alternative constructions are more frequently and for items #10 and #11, in which the ICs are extended, even predominantly used. In the western South-Bavarian locations, zu-ICs are frequently reported. In addition, for item #10, ICs with bare infinitives are more often used in the west, while for item #11, ICs with z’ occur most frequently in the east.

Figure 2. Apparent-time results, geographical differences for the usage of zum-IC.

In item #12 (Jetzt geht er lieber ins Wirtshaus anstatt zu arbeiten), the IC in the stimulus sentence also has an adverbial function. Unlike in the final clauses discussed before (items #9 to #11), it is introduced with the complementizer anstatt ‘instead of’, which marks a contrastive meaning of the adverbial. According to previous research, the use of ICs is not possible in this context in Bavarian dialects. Weiß (Reference Weiß1998:264), for example, states that the IC is replaced by a dass-clause in Bavarian when introduced by anstatt. Table 14 shows the results for this item.

Table 14. Results for item #12 (Jetzt geht er lieber ins Wirtshaus anstatt zu arbeiten)

Our findings reveal that – while dass-clauses do indeed occur in more than one-third of the reported cases – zum-ICs are also used in this context (42), although less often than with the purposive or final adverbials. Apart from dass-clauses, nominalizations introduced with the preposition statt (‘instead’) (43) are frequently realized alternative constructions. Another relatively often employed variant is ICs with bare infinitives, commonly introduced with comparative particles such as als wie (lit. ‘than as’) (44). The high proportion of bare infinitives seems again (as in item #10) to be due to the fact that the matrix sentence contains the verb gehen, which is reanalyzed as a semi-auxiliary verb by our participants.

Interestingly, item #12 is one of just two items where the younger speakers do not use zum-ICs more frequently than older speakers but with a similar frequency. At the same time, younger speakers use bare ICs less often and alternative constructions more often. Regarding the geographical patterns (see figure A2, #12), it is observable that ICs with bare infinitives (as in item #10) are prevalent in the west, while zum-ICs are frequently employed in the east and alternative constructions in the northern and central regions of Austria.

4.4. Attributive Function

In the remaining two items, #13 (Du hast immer etwas zu jammern) and #14 (Habt ihr nichts zu tun?), the IC in the stimulus sentence is used with an attributive function. The head of the IC is an indefinite pronoun (#13: etwas ‘something’, #14: nichts ‘nothing’). Tables 15 and 16 show the results for these items.

Table 15. Results for item #13 (Du hast immer etwas zu jammern)

Table 16. Results for item #14 (Habt ihr nichts zu tun?)

For this context, the informants realize only ICs with either zum, zu or z’. ICs with zum are more frequently used in response to item #13 than to item #14. In contrast, ICs with z’ are frequent for item #14, but less so for item #13. While the reason for this pattern is not entirely clear, it may be due to the different head pronouns.

Interestingly, for both items the frequency of zum-ICs is again higher for younger speakers when compared with older speakers. In turn, older speakers use more ICs with z’. This effect is particularly pronounced for item #14. Geographically (see figure A2), the most noticeable pattern is again the prevalence of zu-ICs in western South Bavarian. In addition, for item #14, ICs with z’ are more or less restricted to western Austria. In the Alemannic region of Vorarlberg, the zum-IC seems to be used about as frequently as the z’-IC. This distribution appears to correspond to the usage of these two variants in the neighboring Alemannic dialect areas in the east of Switzerland (see Glaser Reference Glaser2021, vol. 2:153).Footnote ²⁴

4.5. General Patterns, Constraints, and Tendencies

In this section, we aim to summarize the general trends identified from the previous item-by-item analyses. Initially, we will focus on geographical patterns, followed by linguistic factors and lastly, we will examine age-related differences to determine language change.

Geographical variation: Our findings reveal that there are clear geographical differences in Austria for the dialect variants used for SG ICs with zu. Importantly, the geographical patterns for these syntactic phenomena only partially coincide with traditional dialect areas, which are essentially classified according to phonetic criteria (see section 3.1). This finding corroborates previous and ongoing research on dialect syntax in Austria, where east–west differences are found to be more important than north–west differences (Bülow et al. Reference Bülow, Vergeiner and Elspaß2021, Vergeiner et al. forthcoming). Although the distribution of the different IC-variants depends on the items (and thus on the linguistic context, see below), a clear overall picture emerges (see figures A1 and A2 in the Appendix):

• zum-ICs are predominant in Central and South-Central Bavarian, and also in the eastern parts of South Bavarian. In about half of the items, zum-ICs also occur frequently in the Alemannic dialect region. Interestingly, in Austria zum-ICs appear more frequently in Bavarian when the IC in the item is an object of a phase verb, while they are more often used in Alemannic when the IC in the item has an adverbial function (as can be confirmed by the evidence in Schallert Reference Schallert, Helen Christen, Haas, Montefiori and Ruef2010:143–145). This finding is consistent with the results for the neighboring Alemannic areas in Switzerland, where, for example, the zum-IC is the dominant variant for the adverbial function (Glaser Reference Glaser2021).

• A clear exception is the South Bavarian area in Tyrol and in western Carinthia. Here, zum-ICs are rarely used, which may be explained by the prevalence of another variant, namely, ICs with zu + infinitive. This result dovetails with the findings of Scheutz (Reference Scheutz2016) on the South Bavarian dialects in neighboring South Tyrol in Northern Italy. He shows that the zu-IC has been grammaticalized in these dialects, and it appears that this is also true for the geographically adjacent western South Bavarian region in Austria, which is considered a particularly conservative dialect area. Note that the zu-IC formally corresponds to SG zu-IC but is considered a genuine dialectal structure in these regions. Accordingly, ICs with zu + infinitive are rarely used in dialects outside western South Bavarian. In fact, the instances of zu-IC reported in research locations outside of South Bavarian may be explained by a methodological effect, namely, that some informants mirror the structure of the stimulus sentence presented in SG.

• Clear geographical differences are also visible for bare infinitival constructions. This variant tends to occur in the west, that is, in the Alemannic dialect region and in the Bavarian-Alemannic transition zone, but also in western South Bavarian and South-Central Bavarian. This finding amends the picture provided by other studies, which similarly located bare ICs in the southwest of the German-speaking area but considered them to be predominantly a feature of Alemannic dialects (Brandner Reference Brandner2006, Reference Brandner2020; but see Schallert Reference Schallert, Abraham and Leiss2013).

• Finally, for the clitic form z’ there are less clear-cut geographical differences as the variant occurs throughout Austria – except South Bavarian, which in its western part is a stronghold of the zu-IC. This finding seems to support Weiß’s assumption that z’ is to be regarded as a reduced form of zum and not of zu in the investigated dialects (see the discussion in section 2.2).

Linguistic factors: Our results show that the use of ICs in the Bavarian and Alemannic dialects in Austria varies greatly according to linguistic factors, such as the syntactic function of the IC, the matrix verb, the morphological complexity of the infinitive verb, the internal complexity of the IC, etc. Our analyses have revealed that the observable microvariation in and between the investigated dialects may be attributable to differences in these linguistic factors. Our results confirm some of the findings in the research literature and aid in both clarifying or even amending statements from other studies which are based on comparatively small amounts of data – and some of which claim to be generalizable to other Upper German dialects.

• ICs with zum are frequently used as verbal complements. While their usage is very common in sentences with a phase verb or a causative verb as matrix verb (section 4.2), there are particular constraints with propositional verbs as matrix verbs (item #2) or with ICs in subject function (see section 4.1, item #1). In addition, zum-ICs can be used as adverbials or attributes. Notably, when used as a purposive or final adverbial, the complementizer um is frequently – but not always – omitted. In such contexts, the doubling of zum and its usage as a complementizer are restricted to Alemannic dialects. ICs with zum frequently allow the incorporation of simple noun phrases, at least if they are indefinite and non-specific, both in Alemannic and in Bavarian dialects. In contrast, zum-ICs tend not to license an incorporation of pronouns (see section 4.1 on item #1 and section 4.2 on item #2). In general, zum-ICs seem to occur less often the more the IC has been expanded.

• ICs with zu are autochthonous variants primarily in the western parts of South Bavarian. In these dialects, the use of zu-ICs seems to be licensed in a wide range of contexts and functions. Although they may be restricted in a few contexts (for example, in object function with propositional verbs as matrix verbs; see section 4.2 on item #2), their functions more closely resemble zu-ICs in SG than zum-ICs in Austrian dialects.

• ICs with z’ is a lesser used variant in most contexts, and their usage is even more restricted (for example, with prefixed infinitive verbs, as opposed to particle verbs; see section 4.3 on items #10 and #11). These ICs are most frequently used when they have an attributive function with an indefinite pronoun as syntactical head (section 4.5).

• ICs with bare infinitives are restricted to verbal complements in object function with a phase verb or a causative verb as matrix verb (see section 4.2, items #3 to #8) and to ICs with adverbial function when the matrix clauses contains the verb gehen (‘go’) as a (grammaticalized) semi-auxiliary verb (Demske Reference Demske2020; see section 4.4 on items #10 and #12). In contrast, bare infinitives are obviously not possible when the IC has an attributive function (section 4.4) or a subject function (section 4.1) – even in the Alemannic dialects of Austria, where they are predominantly used. This constraint regarding the subject function is remarkable in that Alemannic dialects in Switzerland license the use of bare infinitives in this function (see Brandner Reference Brandner2006).

Language change: Previous research indicates that the infinitive syntax in Bavarian dialects in Germany is prone to change, resulting in an increased use of ICs, for example, with zum (Donhauser Reference Donhauser, Koller, Wegstein and Richard Wolf1989, Weiß Reference Weiß1998; see section 2.2). Our findings can clearly confirm this assumption for Austrian dialects as, in general, younger speakers use zum-ICs more often, while older speakers tend to employ more ICs with z’ and bare infinitives and also alternative constructions. This trend can be observed in 12 of 14 items,Footnote ²⁵ although not equally pronounced in all of these 12 items. Younger speakers use considerably more zum-variants in cases where older speakers seem to avoid them, for example in the function of a subject (item #1), an object with the causative matrix verb helfen ‘help’ (item #3) or an attributive with the head nichts ‘nothing’ (item #14). At the same time, younger informants increase their use of the zum-variant in contexts where older informants already employ it frequently, namely, in ICs with the matrix verb anfangen ‘begin’ when the IC is not extended (items #5 to #7) and in an attributive IC with the head etwas ‘something’ (item #13).

In sum, these apparent-time findings indicate ongoing change, resulting in the spread of zum-ICs at the expense of the other variants. We see here a form of syntactic leveling that is unusual in that it is not a standard form that is spreading, but a nonstandard variant. However, the zum-variant is also the predominantly used ICs in colloquial language in most of the Upper German language areas (see AdA, www.atlas-alltagssprache.de/runde-7/f13b-d/ [retrieved 01.08.2023]). Moreover, there is reason to believe that this change does not constitute a formal, but a functional advergence to SG (see the discussion in section 5 below).

In order to illustrate geographical differences in change, figure 2 maps the total usage frequency of zum-ICs for older and younger speakers separately.

Our findings indicate that zum-ICs have become more frequent in most research locations (in 32 of the 40 locations). Moreover, this increase affects both Alemannic and Bavarian dialects. However, the change seems to be much stronger in the Alemannic dialects where zum-ICs are overall only a minority variant for older speakers but the majority variant for younger speakers. It is possible that this sharp increase of zum-ICs in Alemannic can be attributed to dialect contact with Bavarian, to which most Austrian dialects belong.

Within Bavarian, zum-ICs are quite widespread among older speakers, although it is notable the zum-ICs are increasing among younger speakers here. As can be seen from figure 2, this change affects the dialects in the (south)east more strongly than those in the (north)west of the Central and Central South Bavarian dialect area. Interestingly, the dialects in western South Bavarian are most stable because they retain the dialect variant zu + infinitive, while zum-ICs are uncommon among both older and younger speakers. This unusual retention of a syntactic variant with only a small areal distribution can probably be attributed to the fact that the dialectal zu-IC happens to be the sole IC variant in SG so that its status as standard variant exerts a stabilizing influence on the dialect.