Resilient Subject Agreement Morpho-Syntax in the Germanic Romance Contact Area

: In this work, we intend to investigate one fundamental aspect of language contact by comparing the distribution of subjects in German, Northern Italian dialects and Cimbrian. Here, we show that purely syntactic order phenomena are more prone to convergence, i.e., less resilient, while phenomena that have a clearly identiﬁable morphological counterpart are more resilient. The empirical domain of investigation for our analysis is the morphosyntax of both nominal and pronominal subjects, the agreement pattern and their position in Cimbrian grammar. While agreement patterns display a highly conservative paradigm, the syntax of nominal (vP-peripheral and topicalized) subjects is innovative and mimics the Italian linear word order.


Introduction
In this work, we intend to investigate one fundamental aspect of language contact by comparing the distribution of subjects in German, Northern Italian dialects and Cimbrian. 1 It is well known that languages that have been in contact for a long period of time develop a process of "grammar convergence", at least from the linear point of view. It is obvious that this process stretches along an ample period of time and the way it proceeds can be illuminating for a general theory of language change and ultimately, of linguistic competence. The specific research question we focus on in our investigation is precisely how the process of convergence advances: in particular, we intend to pin down the phenomena that are more prone to erosion in a minority language such as Cimbrian by comparing it with other German dialects on the one hand, and with the Northern Italian dialects it is in contact with on the other hand. Cimbrian is clearly under pressure from a dominant language, standard Italian, which can be considered as a "Dachsprache" in sociolinguistic terms. It is also in contact with the Trentino varieties that are spoken in the area. The hypothesis we argue for is rather straightforward, but has, to our knowledge, never been explicitly stated. Here, we show that purely syntactic order phenomena, such as vP peripheral subjects, are more prone to convergence, i.e., less resilient, while phenomena that have a clearly identifiable morphological counterpart, such as, for instance, the position of clitic-like elements (enclisis versus proclisis), are more resilient. This results in a clear divergence between the syntax of nominal subjects and the syntax of pronominal subjects: the first one looks akin to Italian, the second firmly remains Germanic. If this observation proves true, we could even be able to measure the advancement of the convergence process and make predictions about the phenomena that are the first target of convergence, although the development of measures of convergence is beyond the scope of this work. The empirical domain of investigation that is the testing bed for our analysis is the morphosyntax of both nominal and pronominal subjects, the agreement pattern and their position in Cimbrian, standard Italian, Trentino and Bavarian (as the closest term of comparison in the German speaking world). The syntax of the subject in Cimbrian does not display any pro drop of the Italian type (i.e., licensed by rich verbal morphology) in any person of the verb. It also does not display the typical Bavarian phenomenon traditionally referred to as "complementizer agreement" (Bayer 1984). However, it allows for low vP internal subjects of unergative and transitive verbs in addition to object-like subjects of unaccusative verbs. This property has traditionally been associated with the phenomenon of pro drop in languages such as Italian and analyzed as one of its correlates. This bizarre situation can lead us to a better understanding of the pro drop property as well. We propose that in Cimbrian, subjects can remain in a low vP peripheral position because SpecT, the position of preverbal subjects in languages such as English and French, is inactive. This means that although Cimbrian apparently displays the same linear word order of Italian, its system is radically divergent from the contact languages (in line with Bidese and Tomaselli 2018 and subsequent work).
The article is organized as follows: In Section 2, we provide a general overview of subject pronouns in the three languages under consideration (NIDs, Bavarian and Cimbrian) and show what their differences and commonalities are. We show that the system of Cimbrian does not resemble either the northern Italian, i.e., the contact one, or the Bavarian one, i.e., the genetically closest related varieties. In Section 3, we examine the possible positions of nominal subjects in the three languages and show that in this case, Cimbrian shares the same distribution as Italian, but crucially, the way the same distribution is derived is radically different. In Section 4, we conclude that language contact is not simply a borrowing of rules, but something more complex: actually, we think that languages in contact "mimic" each other's properties, allowing for the same linear order without necessarily having the same grammar. Furthermore, the rules that are more resilient and less prone to change are those that have morphological anchoring.

Subject Pronouns
In what follows, we take into account the system of subject pronouns in Northern Italian varieties, some of which are in contact with Cimbrian and Bavarian varieties, which are the closest "sisters" of Cimbrian in the German environment.

The Subject Pronoun System of the Northern Italian Varieties
While standard Italian is a regular systematic pro drop language, where verbal inflection takes care of the interpretation of the subject, Trentino, i.e., the Northern Italian dialect Cimbrian is in contact with, has a more varied situation, since it has subject clitics. Subject clitics in NIDs have been shown to occur in different positions; those of the Trentino dialects are generally analyzed as agreement and/or case heads located in their own specific position depending on the person of the verb. The standard tests (in use since Brandi and Cordin 1989) that show that they are not in the same position as complex nominal subjects are: position with respect to the preverbal negative marker, impossibility to be elided in various types of coordination, inversion in main interrogatives, conflation with the complementizer, etc. Since Renzi and Vanelli (1983), it is well known that the distribution of subject clitics follows an implicational scale, which we summarize in (1): (1) Implicational scale: Second singular 2 → third singular → third plural → first singular/plural → second plural The Trentino dialects of the area that is close to Cimbrian have second person subject clitics and third person ones (singular and plural with gender distinctions on the third person), i.e., they stop at the third step of the implicational scale. This is a very typical pattern of the Northeast and is also present in Veneto and Lombardy. The second person singular subject clitic, i.e., the first step of the implicational scale, has a special status in any case, since it presents the following set of properties: (a) It is never optional: when a dialect has a form for second person singular subject clitics, it is bound to use it. We represent this fact by using just one dialect, Venetian, although the generalization is meant to be valid for all the dialects represented in the ASIt data base (at present, 315 dialects).
(2) *(ti) vien you come (b) It is also the first clitic to require doubling with a tonic pronoun. Whenever a dialect has a second person singular subject clitic, it must double the tonic pronoun. 3 Again, we represent the pattern by means of a single dialect, although this is meant to be valid for the whole set of dialects of the ASIt.
(3) TI *(ti) vien (Venetian) YOU (you) come (c) It is the first subject clitic to fail subject clitic inversion when this phenomenon is progressively lost (see Venetian, which has residual subject clitic inversion with auxiliaries in the third person but not in the second singular) (again, we refer to ASIt data).
Cossa ga-lo? (Venetian) What has-he 'What does he have' b.
Cossa ti ga? 4 What you have? 'What do you have?' (d) It is the first clitic of the implicational scale in (1) to be repeated as in object clitics in all types of coordination, even the one that coordinates two instances of the same verb but with a different prefix (5c). Ti canti e *(ti) bali 'You sing and (you) dance' 'You sing and dance' c.
Ti lesi e ti rilesi sempre el stesso libro 'You read and you reread always the same book' 'You are always reading and rereading the same book' (e) In several dialects, only the second person singular is systematically located after the preverbal negation marker, but not other subject clitics. La un viene 'She not comes' 'She is not coming' Some of these properties show that second person subject clitics are lower in the structure than other types of subject clitics (see Poletto 2000 on this). Indeed, they share properties with object clitics (i.e., properties c, d, e: occurrence after the negative marker, lack of inversion, repetition in coordinated structures). Since it corresponds to the addressee, the second person singular also displays other special properties: for instance, the second person singular tonic pronoun has become a sentential particle found in rhetorical and surprise/disapproval special questions in several Veneto dialects (see Poletto 2003, 2006). It is also realized as the subject pronoun in Old Italian in typical pro drop contexts when it also corresponds to the addressee of direct questions. This tallies well with properties a and b, since it shows that the addressee has the tendency to be lexically realized more than other subjects.
Without entering into a detailed analysis of second person singular pronouns in NIDs, we can conclude that the second person singular has a special status in NIDs.

The Bavarian Subject Pronoun System and Complementizer Agreement
Interestingly enough, Bavarian second person singular (and in some varieties also the plural) also has peculiar properties with respect to the other persons, since it is the first person of the paradigm to display agreeing complementizers 5 across Bavarian dialects. There are Bavarian varieties that also display the phenomenon with the first person plural, as attested by the following examples (cf. Bayer 2013, p. 30 andBayer 1984, p.  However, the same implicational relation that we see in NIDs about subject clitics is present in modern Bavarian with respect to complementizer agreement, where the implicational hierarchy is the following: (10) 2sg > 2pl > 1pl.
As Weiβ put it (2005, p. 149): "Complementizer agreement is often confined to the second person singular, which seems to be the minimal form of the phenomenon and that with the greatest areal distribution". The term "complementizer agreement" is well chosen, since in Bavarian, the morpheme of the second singular and plural is identical to the verbal inflection. However, the first person plural is a weak form of the pronoun, as shown by examples (7)-(9). There are two analyses on the market concerning this phenomenon: the traditional one going back to Bayer (1984) that treats the phenomenon in Bavarian as a case of complementizer agreement, i.e., the morpheme attached to the complementizer is indeed to be analyzed as agreement morphology, and as such, it licenses a null subject. The other analysis proposed by Gruber (2008) argues for the same status of Bavarian complementizer agreement as subject enclitics, i.e., similar to the cases we find in Dutch varieties, where the forms are indeed those of a weakened version of the full pronoun. Since the focus of this work is Cimbrian, we would like to remain agnostic with respect to the status of these morphemes in Bavarian. However, we cannot fail to mention that the problem of what actually is complementizer agreement in syntactic terms is rather complex. We follow Weiβ (2005Weiβ ( , 2016, who shows that a complete typology of complementizer agreement includes cases in which the complementizer agreement morpheme is identical to verbal morphology, cases where it is a weakened version of the pronoun and cases where we see both forms. If we take morphology seriously, there must be a distinction between those forms that look like pronouns and those that look like verbal inflection. In this sense, Bavarian might also have a mixed system, as the NIDs have, where the second person morphemes are agreement inflection, while other persons are actually subject clitics. This would be the origin of the hierarchy we see in (10). Be as it may, if we compare the Northern Italian system with the Bavarian one, we notice that in both cases, the second person singular represents the odd man out. If, in Bavarian, the second person singular is a case of real agreement, then the Germanic and Romance systems are opposed to each other: in the NIDs, the first person to be realized with a subject clitic is the second person singular. In Bavarian, due to complementizer agreement, the second person singular is the first to display pro drop (i.e., to allow a gap in the paradigm of subject pronouns). The gap in the second person singular is confirmed by other German dialects, as noted by Weiβ (2005) and originally by Weise (1900, p. 56); in Thuringian, the only possible null subject is indeed the second person singular: Altenburg (Thuringian): (11) a. schreib mir einmal den Brief, kriegst pro auch einen Groschen write.IMP.2SG me once the letter, get.2SG too a penny' 'Write the letter for me and you will get a coin' b.
schreibt mir einmal den Brief, *kriegt pro auch einen Groschen write-IMP-2PL me once the letter, get-2PL too a penny On the other hand, if the Bavarian system is to be analyzed as subject clitics, then the Germanic and Romance systems are identical in requiring the second person singular to be realized through a subject clitic before the others. In a very general sense, the situation found in Bavarian is similar to the one in the NIDs, since it shows that the dichotomy between agreement verbal morphology and subject clitics is most probably only an oversimplification of a much more complex state of affairs, where pronouns pass different stages of weakening before disappearing in a way that has been described by van Gelderen (2011) in her book on cyclic historical change. Bavarian and Northern Italian reinforce the idea found in the literature (see Walkden 2013, among others, for an overview) that regular "pro drop" languages such as standard Italian, Spanish Portuguese, etc., are the result of the overlayering of different pronominal systems, the one of the deictic persons: (speaker and addressee), the one of the non-deictic persons (third person singular and plural) and those persons that are a composition of the deictic and non-deictic ones (first and second person plural).
We underline that in both cases, the second person singular is special and surmise that this is due to its status as the addressee, i.e., as the marked member of the deictic pair speaker vs. addressee. This seems to be a pretty stable and interesting property in both types of languages and must clearly be connected to a deeper explanation of this markedness hierarchy. Recent work by Sigurdðsson (2011) assumes that the speaker and addressee are represented as the pragmatic agent and theme in the left periphery of the clause. Portner et al. (2019) propose that for imperative sentences, there is an AddresseeP in the left periphery of the clause. Following Giorgi (2010), the speaker must be the highest projection in the left periphery, since it provides the deictic coordinates onto which tense, person and location can be interpreted. If we combine all the preceding intuitions, we can assume that the highest portion of the left periphery, the one that is, in general, never reached by the verb and probably stands higher than Force, must have the following layering: The need to spell out the second person subject must be related to this state of affairs. 7 However, one might expect that the second person subject is the last to be realized, since it can be bound by the addressee (in AddresseP) without any interference from the SpeakerP, which is located higher in the structure and establishes the reference of the addressee from its higher position. This is evidently not the case, since the need to double mark the second person, either as a subject clitic or as a doubling agreement marker, is found in both language groups. There is another interpretation of this observation, a much more complex one, which is based on the relations between the phenomenon of indexicality: as stated by Sigurdðsson (2011), indexical pronouns must be able to link arguments to participants of the event, so the second person singular links the subject to the addressee. We surmise that the procedure of establishing this link must be related to the reason why the second person singular stands out in all these systems. Assuming Sigurdðsson's system that sees the phenomenon of indexicality as a syntactic and pragmatic one, where a syntactic procedure feeds pragmatics, we argue that the reason why the second person (singular) stands out in all these dialects has to do with the fact that the addressee is the logophoric patient (i.e., the patient in relation to the discourse, since the speaker is assumed to be the agent, i.e., the person talking) in Sigurdðsson's terms. When the logophoric patient is the subject, which, in most cases, realizes an argumental agent, it is necessary to mark this relation, because there is a reversal between the logophoric roles and the argumental role. Since there is an agree relation between the edge linker features in the CP domain and the subject, in the case of a second person, we have a clash between the logophoric and the argumental features: the addressee is the logophoric patient but the argumental agent. This could ultimately be the reason why the second person singular stands out in all these systems. 8

The Cimbrian Subject Pronoun System
If we now consider Cimbrian, we see that it has neither the Northern Italian nor the Bavarian system. The first interesting observation to mention is that Cimbrian does not have any form of complementizer agreement, a rather rare property with respect to the system of German OV varieties. Interestingly enough, VO Scandinavian languages do not display any sort of complementizer agreement. Insofar, the absence of complementizer agreement might be related to the VO type to which both Cimbrian and Scandinavian belong. 9 This is even more relevant, since Weiβ (2005) notices that in German dialects, verbal inflection can change: when the inflected verb is in final position, it has a different morpheme with respect to the C position.
The second relevant observation is that Cimbrian has indeed a rather complex system of subject pronouns with two clearly distinct morphological paradigms plus the deictic third person d-forms also found in other German varieties, including spoken standard German: All the forms in the highest row are tonic pronouns; the forms in the second row with the -are enclitics; the d-pronouns are in the third row, which exist only for third person, only occur in sentence initial position in an analogous way to the German d-pronouns to satisfy V2. Cimbrian subject clitics (as well as object clitics) are clearly of the Wackernagel type, i.e., C-oriented: they attach to the right of either the inflected verb when it is in C or the complementizer when the verb remains lower, unlike modern Italian ones: (14) Haüt khintar atz Lusern today comes = he to Luserna (He will come to Luserna today) This configuration of properties concerning the subject is rather interesting, since it does not mimic either the Italian or the Bavarian system: Cimbrian has clitics like Italian but they are of the Wackernagel type as in Bavarian. It does not display either complementizer agreement as in Bavarian or pro drop like Italian. We can conclude that Cimbrian does not display any empirical evidence that the second person singular is special in any sense, as we have seen both in the Bavarian and Northern Italian system. Cimbrian has a perfectly regular system where only the third persons have an additional type of pronoun, which is most probably related to subjects that have already been mentioned in the discourse, as is the case of German d-pronouns. More generally, it seems that Cimbrian has not developed any complementizer agreement as in Bavarian nor any special subject clitic hierarchy as in NIDs. It is thus both different to its historical sisters, the Bavarian varieties, and the languages it has been in contact with for the last centuries. The morphosyntax of Cimbrian subject pronouns seems to be rather impenetrable to any influence and has remained quite conservative.

Nominal Subjects
There is one property that Cimbrian shares with German and Italian, namely vP peripheral focused subjects. The fact that Italian allows for postverbal subjects (actually postparticipial) is rather well known in the literature (see Belletti 1988Belletti , 2004 and constitutes one of the main correlates of the pro drop property in all the standard accounts on pro drop. It is probably less well known that standard German also has low subjects, as shown by Grewendorf (2005) on the basis of examples such as the following, which show that subjects can be located lower than modal and manner adverbs: Hence, both Italian and German comply with the observation that (both definite and indefinite) vP peripheral subjects are only found in languages which allow for expletive pro. The problem is, however, that Cimbrian has postverbal subjects with all verb types without allowing an expletive pro. No null subject is licensed either within the Tense domain such as in Italian (via rich verbal morphology or subject clitics), or by C such as in German. Actually, Cimbrian systematically uses an expletive clitic pronoun, -da, when an expletive subject pro is required in German (cf. Bidese and Tomaselli 2018;Grewendorf and Poletto 2011): (24) Haüt iz = ta khent dar nono atz Lusern today is = DA come the grandpa to Luserna 'Yesterday somebody danced' (Today the grandpa has arrived to Luserna) (25) Gestarn in balt hatt = ta gisek di diarn in has yesterday in = the wood has = DA seen the girl the hare (26) Gestarn iz = ta khent/ getånzt yesterday is = DA come/ danced (Yesterday, it was danced) Cimbrian postverbal subjects are clearly rather low in the sentence structure because they occur after the past participle as in Italian. This shows that the past participle must raise higher than the low Focus position where the subject is sitting. 12 The Cimbrian data pose two distinct problems: How come a language without null subjects allows for postverbal subjects? What is da? In this paper, we concentrate on the first question, because a detailed analysis of the particle da would lead us astray from the general research question we are dealing with here. The proposal we adopt here to explain the bizarre configuration Cimbrian displays is the one put forth in Bidese and Tomaselli (2018); Bidese et al. (2020) and Padovan et al. (2021): In Cimbrian, SpecT is not a generalized structural subject position as it is in German, because in Cimbrian, SpecT is not a position for nominative case assignment. Hence, being SpecT inactive, i.e., not projected, Cimbrian looks similar to Italian. The fact that nominal subjects never occur on the right of the finite verb (i.e., never invert ala German) represents clear evidence in favor of this hypothesis (cf. Padovan et al. 2021, ex. 1 and 2, repeated below): (27) Allz in an stroach in balt der pua hatt gesek in vuks All of a sudden in.the wood the boy has seen the fox 'All of a sudden, the boy saw the fox in the wood' (28) *Allz in an stroach in balt hatt dar pua gesek in vuks yesterday is = DA come/danced 'Yesterday, it was danced' As for Italian, we adopt the idea proposed by Poletto (2000) that Italian varieties never occupy SpecT with the lexical nominal subject, because this position is already saturated by either agreement morphology or by subject clitics. 13 Nominal subjects are located higher in the structure in different positions depending on the type of subject. A version of this view has been around for pro drop languages since Benincà and Cinque (1985), who proposed that preverbal subjects are always left dislocated in the pro drop languages, because the SpecI/SpecT position is occupied by a pro (which also licenses postverbal subjects as already proposed by Rizzi 1982). Almost forty years later, we are not even sure whether there is indeed a null maximal category with the properties traditionally attributed to "pro" (see Poletto 1995 for a proposal against the pro category and an extended discussion in Manzini and Savoia 2005). It might be the case that the pronominal features are incorporated into the verb, an idea that has been around since the etymological pronominal origin of verbal inflection was noted. Adopting this view, one might also assume that since the inflected verb in Italian already contains pronominal features provided by the inflectional morphology, it does not need to fill/project its specifier (cf. footnote 12). We intend to maintain Benincà and Cinque's intuition and, following Poletto (2000), we argue that the position for nominal subjects in Italian varieties is in the left periphery. It is most probably a different position for definite nominals and for indefinite and quantified ones (as shows in Poletto 2000, p. 173, example 5), not simply a regular left-dislocated position. The subject positions in Italian varieties are illustrated in (29), where they are signaled in bold: (29) [ Notice that this analysis also explains a series of very well-known facts that have been reported in the literature since the 1980s: (a) Why there are dialects whose preverbal subjects can occur in front of the complementizer (see Poletto 2000, p. 61, examples 44, ff., Paoli (2003).
(b) Why in dialects with a double complementation system, such as the Southern Italian ones, the preverbal subject never occurs on the right of the lower complementizer (cf. Salentino ku, as originally noted by (Calabrese 1993) and also discussed by (Ledgeway 2003) and further developed by (Damonte 2010). 14 (c) Why there is no subject inversion of the Germanic type in Romance pro drop languages, which do not allow for the sequence XP Aux Subject Participle (see Rizzi 1991).
(d) Why preverbal subjects are rather marginal after a focused constituent. All these facts descend from the fact that the subject is simply too high to invert, i.e., to occur on the right of the low complementizer in Fin.
If we compare Italian with German, we can state that in German, the low vP peripheral FocusP can be occupied by the subject, as shown by Grewendorf (2005) and reported above (cf. ex (21)-(23)). Therefore, Italian varieties and German allow for low subjects because the T requirements are satisfied otherwise and therefore, do not need to move the subject.
However, German also definitely uses the SpecT position, not being a language that allows for personal pro drop; hence, inversion is possible, indeed mandatory, in all V2 contexts. 15 The case of Cimbrian is neither identical to the Romance nor to the German system: Cimbrian preverbal subjects are also located in the left periphery as the Italian ones are, although probably in the SpecFin position. However, Cimbrian does not fill the SpecT position because this position is not active at all, i.e., for the opposite reason with respect to Italian. Therefore, in Cimbrian, the possibility to have low subjects depends on a completely different property with respect to Italian: it is neither inflectional morphology nor subject clitics that satisfy the requirements of T, since T does not have any. 16 This shows that Cimbrian mimics the properties of German and Italian but it does so through a different mechanism: low subjects are licensed by the inactivity of the SpecT area and not by an alternative way to check the pronominal features of TP (by C in German via expletive pro; by "absorption" in Italian).
We conclude that both Italian and Cimbrian are languages in which the nominal subject is never located in the TP area but either higher in the CP or lower at the vP border but crucially, for different reasons: in Italian varieties, the T requirements are already satisfied by verbal morphology/subject clitics; in Cimbrian, T has no requirements. Summing up: Cimbrian and Italian are apparently the same, because they allow for vP peripheral subjects. This is rather surprising, given that Cimbrian does not allow for expletive pro, while German and Italian do. The reason why Cimbrian allows for postverbal subjects without being pro drop is that the SpecT position is not projected. Cimbrian uses a left peripheral position for the subject, not SpecT. This means that the case of convergence we are analyzing is a "fake" one; although the output is the same in the three languages under consideration, the syntactic implementation is rather different.

General Conclusions: Resilient (Morpho)-Syntax
The syntax of Cimbrian nominal subjects mimics the one of Italian with respect to both the high preverbal position and the low postparticipial one. The high subject occurs in the CP domain, while the low one occurs at the vP border. The fact that SpecT is inactive results in a linear convergence between the two languages, which, however, still display different grammars, which respect the well-known macrovariation between V2 Germanic and non-V2 Romance languages.
Furthermore, Cimbrian has maintained two peculiar properties since its first attestations in the two Catechisms of the XVII century: (a) it is a coherent non-pro drop language, and (b) it displays Wackernagel-oriented clitics. Evidently, the pro drop/lexical pronoun system, which is clearly more connected to overt morphological properties, seems to be more resilient than the purely syntactic option of low subjects. The same is true of the Wackernagel property of clitics, which is also still rather stable (and is most probably one of the major clues for establishing a V2 system in this language). Interestingly enough, we see that the two resilient properties are those associated with a clear morphological pattern which involves either verbal inflection or clitic pronouns. The less resilient one is the purely syntactic one. Our second conclusion with respect to the grammar convergence is that languages can better mimic each other when syntactic patterns are not directly associated with morphological ones, a rather intuitive conclusion, which is, however, rather new in its present formulation.
Author Contributions: C.P. is responsible for all the aspects of the sections from Section 1 to Section 2.2 included Alessandra Tomaselli is responsible for all the aspects of the sections from Section 2.3 to Section 4 included. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement:
The data discussed in the article are taken from previous literature.

Conflicts of Interest:
The authors declare no conflict of interest.

1
The data used in this article can be found in the literature quoted or come from the ASIt database for the NIDs, and from our own field work for Cimbrian. For the purposes of the Italian academy, Cecilia Poletto is responsible for Sections 1, 2.1 and 2.2 and Alessandra Tomaselli for Sections 2.3, 3 and 4. 2 This first step of Renzi and Vanelli s generalization is violated precisely in some Trentino dialects, such as Tuenno (see AIS, Atlas Italiens und der Südwschweiz), which is, however, not the closest Trentino variety to Cimbrian, which is geographically closer to Valsuganotto (see Poletto 1993Poletto , 2000Poletto and Tortora 2016 for a discussion of the violation of the first step of the implicational scale). This is also pointed out by Manzini and Savoia (2005)  It is interesting to note that the Trentino dialects in contact with Cimbrian still have a more widespread and conservative pattern of subject clitic inversion in main interrogatives, which also includes the second person singular pronoun. 5 Some authors (see Gruber 2008) propose that complementizer agreement is actually a type of pronoun. We remain agnostic with respect to this point, simply noting that the type of pronoun found in Bavarian is identical to verbal inflection, not to a reduction of a tonic pronoun. This difference shows that complementizer agreement is not a regular case of enclitic subject pronouns but something else. 6 See among others Haegemann (1992); Shlonsky (1994); Weiβ (1998Weiβ ( , 2005Weiβ ( , 2016; Carstens (2003). 7 On the special status of the second person, there is a long tradition of studies from different perspectives (see, for instance, Lazzeroni 1994). We do not see fit to enter here a long summary of all the existing approaches, since this is not the major focus of the article. In any case, an anonymous reviewer suggests that an alternative view to capture the special status of the second person could be found in the perspective discussed in distributed morphology in work by Bobaljik (2008); Calabrese (1995); Halle (1997);Noyer (1992). 8 This is even true of standard Italian, where in the present subjunctive, a second person singular pronoun is obligatory when verbal agreement is identical for the first, second and third person singular, as originally noted by Poletto (1993). 9 We leave this point for further research. We thank an anonymous reviewer for pointing this out to us. 10 The distinction between do and to is a purely phonological one and depends on the preceding phonological context: when there is a voiceless consonant that precedes the pronoun, the /d/ assimilates for sonority and becomes voiceless.

11
For low subjects with unaccusative verbs and in passive constructions cf. also Den Besten (1989) and Tomaselli (1986). 12 There is evidence that the past participle position in Cimbrian is rather low, and most probably lower than all aspectual adverbs. For our analysis to be correct, we only need to say that it moves immediately higher than the SpecFocus position where the subject is sitting. As shown by Belletti (2004, p. 13), the vP peripheral SpecFocus position is indeed located lower than the lowest (manner) adverbs. Alternatively, the postverbal position in Cimbrian might be the argumental SpecvP one, since Cimbrian probably resolves these cases of Focus through a prosodic strategy relying either on remnant VP movement or Subject right-extraposition. 13 Whether we assume a null category pro licensed in SpecT by either verbal morphology or subject clitics sitting in a higher projection or not, i.e., we postulate that subject clitics/agreement morphology satisfy the pronominal feature of T per se, is a problem of theoretical viewpoint which should be left open for further discussion. In any case, it does not have any direct import on the analysis presented here. In this perspective, an inactive SpecT could well be interpreted as non-projected. 14 For a different view on the complementation system of Salentino, which treats ku as a modal particle similar to mi/mu of Calabrian and Sicilian, see Manzini and Savoia (2005). 15 It is probably the case that subjects in German can also be located higher, since they can occur higher than Topics, but we do not pursue these facts further here, since they would require a whole detailed investigation on this property. 16 For a detailed discussion on the properties of TP in Cimbrian, see Bidese et al. (2020). 17 The SpecT position is in italics because it is not projected at all.