US WURK LXV (2016), p. 146 [1328] Minority languages in language contact situations: three case studies on language change Andrea Padovan*, Alessandra Tomaselli*, Myrthe Bergstra**, Norbert Corver**, Ricardo Etxepare§, Simon Dold§^ Abstract. The phenomena described in this paper ideally represent the convergence of two apparently distant fields in linguistics, namely language contact and grammatical change. Taking into account minority languages in contact with standard languages, we will show that contact itself can actually play a role in affecting the speed of ongoing changes in both the “stronger” and the “weaker” language. The intuitive idea that syntactic structures per se can be borrowed from neighboring languages will be proved false – or too coarse – as only in rare cases do grammatical phenomena manifest themselves as replicas of their counterparts in the other language: our major claim is that borrowing occurs at a more abstract level, i.e. the level of formal features. 0. Introduction This paper presents results of the research carried out within the European Research Project AThEME (http://www.atheme.eu),1 that stands for “Advancing the European Multilingual Experience”. One of the main questions and objectives that the project pursues is to investigate the possible effects of multilingualism on grammar changes in a situation of contact between regional and standard languages. Therefore this article presents three case studies each of which zooms in on a specific grammar change phenomenon in which multilingual competence plays a key role - cf. Abraham & Leiss (2013), Sasse (1992) and Thomason (2001a-b). In the last decades, regional varieties and minority (even endangered) languages have proven to be precious instruments for the observation of 1 . The research leading to these results has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstration under grant agreement no. 613465. The University of Utrecht, the University of Verona and the French IKER Center are the institutions involved in this joint paper: they all belong to the “AThEME Work Package 2” devoted to regional and minority languages in different European countries; Simon Dold is a PhD student collaborating at the IKER lab. We would like to thank the reviewers from Us Wurk for their insightful comments and suggestions which helped us to improve the article. Us Wurk, jiergong 65 (2016), s. 146-174. US WURK LXV (2016), p. 147 grammar change, which normally occurs faster and is somehow easier to be observed than in standard varieties (cf. Bidese et al. 2012; Abraham and Leiss 2013 and references cited there). In this article three geographic areas featuring language contact in a multilingual context will be taken into account: (i) Friesland as an example of contact between Standard Dutch and a language genealogically related to it, namely Frisian (see section 1). (ii) the Region Trentino-Alto Adige/Südtirol as an example of contact between Germanic and Romance varieties with a particular focus on the Cimbrian minority (see section 2). (iii) the Basque Country (the central variety of Basque, Guipuscoan), as an example of two genealogically unrelated languages (central varieties of Basque vs Spanish) – see section 3. Even if the three different areas under investigation provide a manifold picture of contact situations, they are all comparable from a sociolinguistic viewpoint: all speakers of the different varieties are bi- or multilingual; moreover, they represent cases of multilingualism where local varieties face the pressure exerted by either a stronger standard (Dutch, Spanish/French, Italian) or other more prominent dialects (even of different language families). As for the grammatical aspects that might constitute the possible core of the research on regional languages, they encompass – as one might expect – a wide range of different phenomena from phonology to syntax, from morphology to lexical semantics. However, in this paper we will restrict the phenomena under investigation to the syntactic level in order for the results to be more homogeneous and hence directly comparable. In particular, the three syntactic case studies under investigation in this paper involve phenomena in the realm of the clausal structure. Specifically, the paper will center around the following topics: Germanic-Germanic contact: ongoing changes in directional verbs in Frisian (gean ‘go’ and bliuwe ‘stay’), which have taken on the typical features of functional verbs entering aspectual constructions, i.e. they imitate the behavior of their Dutch counterpart. Germanic-Romance contact: borrowing of a functional word in the Alpine area (complementizer borrowing in Cimbrian). US WURK LXV (2016), p. 148 Basque-Romance contact: attrition phenomena in the Spanish spoken in the Basque area. The syntax of Spanish wh-interrogatives seems to be affected by Basque in showing the same restrictions found in Basque. Even though the three case studies focus on different domains of the clausal system – i.e. V-Aspectual in Frisian, the system of subordinating conjunctions in Cimbrian, the structure of interrogatives in Basque – it will become clear that, at a more abstract level, the nature of grammatical change in those domains displays some similarities, confirming that language contact does not result in the transfer of morphosyntactic structure but it rather affects the (more abstract) system of functional features: this is basically what is proposed in this paper within a common syntactic framework based on an elaborate system of functional projections enhancing the basic C-T(v+V) spine.2 In section 1, which presents Frisian data, it will be shown that lexical elements already present in a language develop a propensity to take on more functional characteristics and end up being “semantically bleached”, i.e. are used to convey a specific functional property (say a temporal/aspectual feature for a verb of movement): we take contact to be at work in this case. In fact, as will be presented in the section, this is the case when such a grammaticalization path has already occurred in the model language, the replica language simply reproducing the same behavior. Section 2, devoted to the system of Cimbrian subordinating conjunctions, will focus on the behavior of a complementation system which has borrowed a new member from Romance giving an account as to how the features of this newly-introduced element are manipulated. 2. The C-T-(v+V) system is simply a commonplace shorthand for the functional structure of the clause: ‘C’ stands for Complementizer, ‘T’ for Tense, small ‘v’ and capital ‘V’ both stand for the verbal layer consisting of its lexico-semantic (V) and functional features (v). The sequence C-T-(v+V) is said to be basic per se, since it is made of just four elements, but each atom can be split in an array of sub-features, i.e. those appertaining to the Complementizer which are normally fronted in the sentence (subordinating elements per se, interrogative elements, dislocated phrases, etc.); those appertaining to Tense (temporal features per se, aspectual and mood features, etc.) and so on. Theoretical approaches within the Generative-linguistic tradition differ as to how much structure is assumed to describe a sentence properly. Here, we are proposing a richer and more fine-grained make-up of the functional features of the sentence to make sense of all the data coming from the minority languages taken into account. US WURK LXV (2016), p. 149 Finally, in section 3 it will be shown that, when it comes to possible word order patterns, language contact seems to push towards an overall reduction of the set of all possible word orders by simply favoring the pattern shared by the model language and the replica language rather than by introducing new ones: to put it in another way, language contact favors the discarding of alternative orders ˗˗ especially the marked ones. As a matter of fact, BasqueSpanish bilinguals tend to restrict the options that Spanish monolinguals have in following the pattern possible in Basque. 0.1 Our view on contact In the tradition of language contact studies, borrowings are known to be extended in their usage or – on the contrary – to appear in very restrictive contexts, lexically instantiating a specific position in the replica language. What is new in this article is the approach towards the very concept of language contact which is unanimously taken to play a role in grammatical change but in a way subtler than previously thought. Even in the process of lexical borrowing – especially when it comes to functional elements (e.g. complementizers) – one must be very cautious not to confuse the adoption of a piece of E-language with the adoption of the abstract features (or a pertinent subset thereof) appertaining to that item, which is rather an Ilanguage phenomenon.3 In other words, computational atoms (i.e. an abstract feature or a bundle of features) borrowed by another language are “less innocent” than they seem. In fact, it is the way in which formal features are manipulated in the bilingual mind that affects the very essence of syntactic objects. Basically, being bilingual means that two I-languages are in contact: this contact may be unbalanced i.e. there might be stronger competence in either language, as is often the case with minority languages. In any case, two grammatical systems in contact entail that one system affects the other in speeding up or maintaining ongoing change or even 3 . The distinction between E(xternalized)-language and I(nternalized)-language was first proposed in Chomsky (1986): the former is taken to be the epiphenomenal aspect of language, i.e. actual languages or their “visible characteristics” whereas the latter represents the biologically specified mental object that abstracts away from idiosyncratic instantiations and represents the core properties of language itself. As regards language contact, it has been emphasized that comparing languages entails more delving into abstract syntactic operations relating to general principles of U(niversal)G(rammar) and to linguistic interfaces rather than comparing superficial characteristics in the lexicon and in constructions (see also Chomsky 2000). US WURK LXV (2016), p. 150 more in introducing feature characterizations coherent with the system which imports them. On the contrary, a system never accepts erratic features – i.e. incoherent with the system itself – and is never derailed from its internal direction of change. 1. Grammaticalization of directional gean (‘go’) to an aspectual item in Frisian In this section, we will consider the change in the lexical nature and grammatical behavior of the Frisian verb gean (‘go’) and its relation to DutchFrisian language contact. Gean used to be a lexical element expressing motion, but is nowadays used as a functional item expressing prospective aspect in Frisian, while it also remains being used as a motion verb. The term Frisian in this paper refers to the West-Frisian language variety that is spoken in Friesland, a province in the northern part of The Netherlands. Frisian is recognized as an official language in The Netherlands, but it is spoken in a very limited area of the The Netherlands and mostly used in informal settings. Similar to Dutch, Frisian is a West-Germanic language. Dutch and Frisian are actually very similar both in structure and vocabulary. There used to be a “stable diglossia” situation in Friesland (De Haan, 2010), in which Frisian was used for informal matters and Dutch for formal matters. In the 20th century, however, this changed to a situation of “unstable bilingualism” (De Haan, 2010). Dutch is used more and more in the public life and at home. Dutch is the obligatory language of primary education in the Netherlands and all speakers of Frisian speak Dutch as well. This means there are no Frisian monolingual speakers anymore. Dutch is clearly the majority language nowadays and is influencing Frisian in many aspects. One example of this is the behavior of the Frisian verb gean (‘go’), which seems to change from a lexical motion verb, as in (1b), to a functional item indicating the near future, as in (2b). This replicates the behavior of the Dutch verb gaan (‘go’) (see (1a) and (2a)). In both languages, the verb also remains being used as a verb of motion. (1) a. b. (2) a. b. Ik Ik ga gean naar nei Amsterdam. Amsterdam I go to Amsterdam Ik Ik ga gean zwemmen. swimmen. I go swim “I’m going to swim” Dutch Frisian Dutch *Former Frisian / ?√Frisian US WURK LXV (2016), p. 151 1.1 The data The data which is referred to in this section was collected by means of a digital, written questionnaire. The 87 participants were all speakers of Frisian (either Standard Frisian or a Frisian variety). Their ages ranged from 21 to 84 (M = 53.3, SD = 15.5). The questionnaire consisted of a grammaticality judgment task (i.e. they had to decide whether constructions with gean and bliuwe were natural or not) and background questions section. In this section, the participants were asked about their age, gender, education level, their education in Frisian, place of birth, place of residence, native languages, the language of their parents and their use of Dutch and Frisian on an average day. As a control, a Dutch version of the same questionnaire was administered to a group of Dutch speakers (N = 61), excluding the background questions which referred to Frisian. 1.2 Analysis This section analyses the change in the Frisian gean as a case of contactinduced grammaticalization.4 It shows that language contact has very clear but subtle influences; it targets an item which, because of its lexical features, was likely to grammaticalize, and it is, therefore, not the only cause of this change. 1.2.1 Grammaticalization Grammaticalization is the process of change from a “more lexical” towards a “more functional” item. Gean in (1b) is a lexical motion verb, but in (2b), it is functional. The main verb is swimme, and gean has to be in a functional position, since the sentence is monoclausal,5 functional verbs not being 4 . The grammaticalization path and its decomposition in several implicational steps as a diachronic process within a single language has been well-known ever since Heine (2003) a.o. As we will see, in the specific case of language contact the trigger of grammaticalization is due to the transfer of functional features available in the model language. To what extent the process of borrowing in language contact mirrors the process of grammaticalization prompts an important research question to be addressed in future studies. 5 . In a mono-clausal structure, negation can only take scope over the whole sentence, not just one part of it (Erb, 2001). (i) does not have scopal ambiguity, as illustrated below, and is therefore monoclausal. (i) Sy giet net swimmen. She goes not swim “She is not going to swim” US WURK LXV (2016), p. 152 endowed with theta-features and not being associated with a v/V structure. The question is then, in which functional position gean might be and whether this is similar to Dutch gaan. An obvious candidate for the position of gean would be T, or, assuming a richer clausal hierarchy like Cinque (1999), Future Tense. However, sentences like (3), with a stative verb as complement of gean, and (4), with a modal verb as complement of gean, are judged very poorly by native speakers in our questionnaire. (3) *Wy We gean go nije next wike net week not thús at home wêze. be “We will not be home next week.” (4) *Hy giet net kieze kinnen tusken syn freonen. He goes not choose can between his friends “He will not be able to choose between his friends.” If gean were a Future Tense item, there would be no reason why these sentences are ungrammatical; the counterparts with the future tense verb sille (‘will’) are grammatical. However, if gean is actually an aspectual item, this would explain the impossibility of (3) and (4). Since aspect deals with the internal temporal structure of events, it makes sense that they are hard to combine with stative verbs; verbs that lack internal temporal structure.6 The inability for an aspectual item to embed a modal verb follows from the functional hierarchy by Cinque (1999), partly represented in (5), which shows that modal projections (Mod) are higher than aspectual projections (Asp). (5) [Moodspeech act [Moodevaluative [Moodevidential [Modepistemic [T(Past) [T(Future) [Moodirrealis [Modnecessity [Modpossibility [Modvolitional [Modobligation [Modability/permission [Asphabitual [Asprepetitive(I) [Aspfrequentative [Aspcelerative(I) [T(Anterior) [Aspterminative [Aspcontinuative [Aspperfect [Aspretrospective [Aspproximative [Aspdurative [Aspgeneric/progressive [Aspprospective [AspSgCompletive(I) AspPlCompletive [Voice [Aspcelerative(II) [AspSgCompletive(II) [Asprepetitive(II) [Aspfrequentative(II) .. * “She is going to [not swim].” . Stative verbs in Frisian can, for example, also not be put in a progressive aspect context: (ii) *Ik bin thús oan it wêzen I am at home at the be “I am being at home” 6 US WURK LXV (2016), p. 153 Selectional restrictions thus suggest that gean is in an aspectual head. We propose that gean expresses prospective aspect, i.e. it is lexicalized in the Aspprospective projection. This corresponds to the meaning that functional gean expresses: a precedence relation between the utterance time and the event (in other words, the utterance refers to a moment before the event actually started). Interestingly, this corresponds exactly to the behavior of the Dutch gaan: the Dutch counterparts of (3) and (4) were also judged poorly, and gaan is therefore assumed to be an aspectual item in Dutch as well. As stated above, prospective aspect expresses a relation of precedence (assertion time before event time) (Demirdache & Uribe-Etxebarria, 1998). A relation of precedence between two things (X before Y) can be both spatial or temporal. In its motion verb use, as in (1b), gean actually expresses precedence as well: the subject is before (moving towards) the goal in space. According to Demirdache & Uribe-Etxebarria (1998), this is the same kind of precedence relation as temporal or aspectual precedence: they all have the features [–central coincidence] and [+centripetal coincidence].7 Because gean is a very general, “empty” motion verb (it does not specify manner or direction, for example). The proposal of this paper is that the item gean contains only these precedence features and can therefore express precedence both in space and in time. Grammaticalization is the process of becoming more functional, and can therefore be seen as moving upwards in the functional structure (IJbema, 2001). This is what happened to gean: besides being a lexical verb, it can, in present-day Frisian (of many speakers) also be inserted in a higher, functional position in the structure: prospective aspect. Although we speak of grammaticalization of gean, and therefore a change in gean, actually the features of the verb itself do not change. Prospective aspect was not explicitly expressed in Frisian before, but now it lexicalized, and this is done so by means of gean, because gean is very well fitted for this role: its precedence features already match the features of aspect prospective. One aspect that should be noted here, is that gean has had the possibility of selecting a bare infinitive for a longer time, but only with a limited set of verbs: the posture verb lizze (to lie), sitte (to sit), stean (to stand) and hingje (to hang). However, in these constructions, there is always a sense of 7 . These features refer to the position of a Figure with respect to the Ground (Demirdache & Uribe-Etxebarria, 1998). [-central coincidence] means that Figure and Ground are not at the same place or time. [+centripetal coincidence] means that the Figure precedes the Ground (centripetal means “towards center”). US WURK LXV (2016), p. 154 motion, so gean is not yet an aspectual verb here. Nevertheless, having these constructions in the input might have lead acquirers to hypothesize that selecting a bare infinitve could be one of gean’s selectional properties. 1.2.2 Language contact and other factors As shown in (1a) and (2a), the Dutch verb gaan patterns similar to the Frisian gean: it has both a motion use and a functional use. It also shows the same limitations as in (3) and (4), showing that it is an aspectual item rather than a future tense item. The difference is that in Dutch, this is not a recent change, the pattern has been accepted for a long time. A natural assumption to make is then that Frisian gean changed because of contact with Dutch. However, this has never really been proven. Since go grammaticalizes in so many languages (Bybee et al., 1994), and the West-Germanic languages closely related to Frisian (Dutch, Flemish, English, German to some extent) have undergone the same change, it could also be the case that this is an internal change. We propose, however, that language contact with Dutch indeed plays a role. We find a clear indication for this in the results of our questionnaire. It turned out that there is a small but significant correlation between the amount of Dutch spoken on an average day and ratings on the items with gean as a functional item (r = .267, p = .013). In other words, people who speak more Dutch, are more accepting of the new use of gean. A second indication that contact with Dutch plays a role in the change of gean is the fact that the change already happened in contact dialects (dialects that emerged from a lot of contact between Dutch in Frisian), such as Town Frisian (van Bree & Versloot, 2008) and West-Frisian (Hoekstra, 1994). At first glance, the change might seem like a fairly superficial borrowing process, since it involved just one word. However, it is not the case that a new word has been borrowed, but the behavior of an item that was already present in Frisian has changed. The change in Frisian can therefore be seen as a process of “replica grammaticalization”, as described by Heine & Kuteva (2003: 539). However, grammaticalization is not a process happening within one speaker, it is something gradual we observe on a population level: gean is used more and more in a functional way. On the individual level of a Frisian acquirer, it is one setting of an I-language grammar: the lexicalization of prospective aspect (by gean). This speaker therefore uses gean as an aspectual item. When a lot of speakers acquire the I-language with this setting, which is different from the I-language of previous gener- US WURK LXV (2016), p. 155 ations, their output will be different from previous generations, and a change can be observed in the E-language, the Frisian language of the population. In short, although the change might seem like an E-language change, the ways in which the change could happen was constrained by the linguistic system (the possibility of a prospective aspect category and the features of gean). Moreover, we found that language contact played a role in actually initiating the change. 1.2.3 Bliuwe A similar case of replica grammaticalization happened with the Frisian verb bliuwe; it grammaticalized from a lexical verb expressing lack of motion (6b) to a functional item expressing durative aspect as in (7b). It patterns exactly like the Dutch verb blijven (6a & 7a). (6) a. Ik b. Ik I (7) a. Ik b. Ik blijf in bliuw yn Amsterdam. Amsterdam stay in Amsterdam blijf bliuw zwemmen. swimmen. Dutch Frisian Dutch *Former Frisian/?√Frisian I stay swim ‘I keep swimming’ Similar to gean, bliuwe was an item sensitive to change anyway; its [+coincidence] feature matches both with the use in (6b) (‘X stays at location Y’) and with (7b) (‘X stays at event Y’). The category durative aspect, i.e. the ASPdurative projection, can now be lexicalized in Frisian, boosted by Dutch influence, and enabled by the features of bliuwe. 1.3 Preliminary conclusion In short, language contact with Dutch has influenced the changes in Frisian, but in very subtle ways: the changes occurred in items which did not have much lexical content and were therefore the most likely candidates for change. In this case, those items were gean and bliuwe, verbs that have little semantic content except for the [±coincidence] features that are relevant for both space and time. This made them likely candidates for change, which can also be noted from the fact that these verbs grammaticalize in many other languages, too. Although language contact is therefore not the only relevant factor, it plays an important role in initiating the change. The change is also more subtle than one might think. It is not just the borrowing US WURK LXV (2016), p. 156 of an E-language item; the limitations of gean clearly show that it fits into the category of prospective aspect, a functional category that has now become lexicalized in Frisian: an I-language change. To sum up: even if there is no actual borrowing of lexical items – as these verb forms are already there in the replica language – the categorical features that they are made of go through a restructuring path: the motion verb begins instantiating an aspectual auxiliary. 2. Complementizer borrowing in Cimbrian: data from a Northern-Italian language island Borrowing is a common phenomenon in contact situations (cf. Matras 2013 and Winford 2013 a.o.): the process of borrowing of lexical items in a replica language is continuous, especially if the pressure exerted by the model language is strong enough. When it comes to the borrowing of functional words, though, there is no general consensus as to whether they maintain the set of the formal features (or a subset thereof) they are endowed with in the model language. Anyway, there is strong typological evidence that functional words like complementizers are borrowed, in which case a partial “readjusting” of the complementation system of the language is likely to occur (see also Bayer 1999 and Tánczos 2013). For our concerns here, we take into account the double system of complementizers found in a Germanic minority language, Cimbrian, which is spoken in the area between the Regions Trentino-Alto Adige/Südtirol and Veneto. Cimbrian belongs to the group of Southern Bavarian-Austrian dialects from which it has been isolated since the 11th Century (cf. Bidese 2004) preserving some morpho(phono)logical features it has in common with its medieval cognates (cf. among others Bidese 2010 ed.). Nowadays, the three major varieties of Cimbrian are spoken in Luserna/Lusérn in the Province of Trento; in the so-called area of the Tredici Comuni (lit. “Thirteen Municipalities”) in the Province of Verona (where Cimbrian is spoken in the village of Giazza/Ljetzan only); in the so-called area of the Sette Comuni (lit. “Seven Municipalities”) close to Asiago/Schlege in the Province of Vicenza (where only few speakers of Cimbrian are found in the village of Roana/Robaan). However, the only variety in which a certain degree of competence is found among younger speakers is the one spoken in Luserna. In this small village in the Trentino Region there are 300 inhabitants: they are all taken to be speakers of Cimbrian (de facto about ¾ of them, so ca. 230). Their fluency is not homogeneous since fluent-speak US WURK LXV (2016), p. 157 erness and semi-speakerness are intermingled (“semi-speaker” in Dorian’s 1981 sense). The other villages in the Veneto Region have less than 10 speakers each, all with different degrees of competence. In this dialect the Romance complementizer ke (Italian che) has been used for at least a century (at least, this is what turns up in the most ancient written texts we have, namely the “Tales of Lusern” gathered in 1905 by Josef Bacher) but it might have been present even earlier. The native complementizer of Cimbrian is az; it introduces either declarative clauses or embedded polar question, thus it can both mean ‘that’ and ‘whether’. It is typically selected by nonfactive verbs and often occurs with embedded subjunctive. Its selection is related to word order asymmetry: in fact, az-introduced sentences show the typical root vs embedded word order asymmetry i.e. in the embedded order the finite verb (henceforth Vfin) obligatorily follows sentential negation and other sentential adverbials. Moreover, az can host phonologically weak elements like clitic pronouns and the expletive subject -da (cf. Kolmer 2005 and Bidese; Padovan & Tomaselli 2012) forming compound words such as az-to, ‘that-you’, azz-ar, ‘that-he’, az-ta, ‘that-there’ and so on. On the other hand, the complementizer ke shows up in both declarative and non-restrictive relative contexts; it behaves more like a generalized “subordinator” rather than a full-fledged clause-typer as is the case of az. In fact, ke does not affect word order in subordinate clauses: in other words, it neutralizes the asymmetry main vs embedded extending the root word order pattern to embedded clauses: in ke-introduced clauses the Vfin precedes both Neg and sentential adverbials and subject-verb inversion is possible in the same fashion as in root clauses. Ke can host neither clitic pronouns nor the expletive -da (cf. Kolmer 2005 and Bidese; Padovan & Tomaselli 2012) and has to be followed by tonic pronouns like in ke du (*ke-to) ‘that you’, ke er/dar (*ke-ar) ’that he’ and so on. (8) (9) a. I bill azzar nèt gea ka Roma (az PronClit Neg Vfin) I want that=he not gosub to Rome I do not want him to go to R. b. Di Maria khött ke er geat nèt ka Roma (ke Pron Vfin Neg) The Maria says that he goes not to Rome Maria says that he does not go to R. a. Di Maria khött ke geischtar izz=ar rivart The Maria says that yesterday is.he arrived Maria says that yesterday he arrived US WURK LXV (2016), p. 158 b. dar Mario, ke z’iz a guatz mentsch, khinnt pitt üs the M. who it is a good person comes with us Mario, who is a nice guy, is coming with us 2.1 Analysis In recent work (Grewendorf & Poletto 2011; Padovan 2011; Kolmer 2012) it has been put forward that az and ke differ in their syntactic behavior since they are merged as different C heads in a split-C scenario. A low position and a high position must be distinguished: this is due to the fact that the major functional categories of the clause are not monolithic but can be broken down in specialized sub-categories, as we hinted at in the introduction.8 Recall that az can be related to embedded Mood (subjunctive) and its presence forces the finite verb to occur lower (possibly in v/V). In other words, assuming that V2 sentences feature the finite V moving to Fin0 it is reasonable to assume that az occurs in complementary distribution with the finite verb since it competes for the same position. Thus, az heads FinP blocking V-to-C(/Fin) movement. (10) a. [SubordP [ForceP [ … [FinP az-cl [TP … mood/Vfin ]]]]] b. [SubordP ke [ForceP [ … [FinP Vfin [TP … V̵fin ]]]]] As regards the abstract features of these items, it would be quite tempting to assume prima facie that a functional element like ke maintains the formal features with which it is endowed in Italian: for instance, one might expect 8 . Ever since the seminal work by Rizzi (1997) concerning the left periphery of the clause, the categorial features of the Complementizer per se have been taken to instantiate dedicated projections which host the functional categories normally associated with C, namely subordinating conjunctions and interrogative phrases but also left-dislocated constituents like topicalized or focalized XPs. In particular, the Force Phrase – the topmost C projection – encodes the illocutionary force of the clause and is assumed to be the merging position of “high” complementizers (like Italian che, ‘that’); the Fin(iteness) Phrase – the bottommost C projection – represents the boundary between C and Tense and controls tense and mood features of the embedded verb, moreover, it can also be a merging position for complementizers (the “low” ones, such as the Italian infinitival complementizer di, ‘to’). Complementizers can therefore be found in at least two different positions of the the C-layer or the left periphery of the clause. However, there are languages that do not conflate subordinating and clause typing: in fact, it can also be the case that a generalized subordinating element precedes the “real” complementizer simply indicating the “beginning” of a non-matrix clause: Bhatt & Yoon (1992) first proposed the presence of a Subordinator Phrase which has to be kept distinct from the C head (in contemporary terms, both Force and Fin). Assuming such a position allows us to explain several embedded root phenomena, such as embedded V2 (see also Julien 2008 for embedded V2 in Scandinavian). US WURK LXV (2016), p. 159 no difference between main and embedded clause, as is the case in Italian, and in fact this is borne out. However, this similarity is only apparent: in fact, Cimbrian main clauses feature v/V-to-C movement, whereas their Italian counterparts don’t. What is the treatment of a functional word in the replica language? What is it merged with? Assuming a split-CP scenario, it is quite reasonable to take a borrowed complementizer like ke to enter the Cimbrian complementation system being merged in the topmost C layer (the so-called “Subordinator Phrase” in Bhatt & Yoon’s 1992 terms) leaving the bottommost layer, i.e. Fin, free for verb movement to take place. Going back to the parallel Italian che vs Cimbrian ke, we want to emphasize that even if they both introduce a “symmetric system” where matrix clause and subordinate clause feature the same word order, there remains a subtle, yet crucial, difference between Cimbrian and Italian since ke does not introduce exactly the same symmetry found in Italian: in other words, the system is symmetrical (as in the model language) but in its own way. 2.2 The unexpected pattern ke + subjunctive In the previous section we have pointed out that az is connected with embedded mood, which is expected in a framework where Fin checks for mood downstairs. Recall that az can also cooccur with indicative; what is not expected in this system is the cooccurrence of ke and subjunctive mood. Let’s start by observing that in Italian the factive/non-factive status of the matrix verb is relevant for complementizer selection (cf. Manzini 2000 and Hooper & Thompson 1973 a.o.) whereas in Cimbrian it is the mood of the embedded clause that ultimately affects complementizer selection. Nonfactive verbs such as gloam, ‘think/believe’ select for both az and ke; on the contrary, other non-factive verbs like khün ‘say’ only select for ke (cf. Bidese et al. 2013) as shown in (11-12): (11) a. Sa gloam azzar sai gerift spet They think that=he besub arrived late ‘They think that he arrived late’ b. Sa gloam ke dar iz gerift spet They think that he is arrived late ‘They think that he arrived late’ (12) a. Dar Mario khütt ke dar Bèppe iz gånt kan(n) birt1 The M. says that the B. is gone to the pub ‘M. says that B. went to the pub’ US WURK LXV (2016), p. 160 b. *Dar Mario khütt azta dar Beppe sai gånt kan(n) birt The M. says that.da the B. issub gone to the pub ‘M. says that B. went to the pub’ If the system were limited to just [ke+IND/*SUB] and [az+IND/SUB] we could invoke a relatively simple scenario: az, heading Fin, acts as a probe checking the mood feature of embedded T; on the contrary, ke is not expected to do so since it is “blind” to embedded mood. However, if it turned out that ke can actually occur with subjunctive9 the assumption we have made so far would be problematic, then ke, being merged in SubordP, does not possess features compatible with mood selection. As a matter of fact, subjunctive turns up in ke-introduced sentences: in particular, in the translation tasks administered to our (bilingual) informants the mood of the Italian stimulus sentence seems to play a key role in affecting mood selection in Cimbrian translations and hence in triggering the presence of an “unexpected” subjunctive (13 and 14b). ’Z iz nèt khött ke dar Gianni khemm pit üs (ke + sub) It is not said that the G. comesub with us Stimulus sentence: “Non è detto che il Gianni venga con noi” (‘we are not sure that G. comes with us’) (14) a. I gloabe ke dar Gianni iz sa gerift ka Tria (ke + ind) I think that the G. is already arrived in T. Stimulus sentence: “Credo che Gianni sia/è già arrivato (a Trento)” (‘I think that G. isIND already arrived in T.’) b. I gloabe ke dar Gianni sai sa gerift ka Tria (ke + sub) I think that the G. beSUB already arrived in T. Stimulus sentence: “Credo che Gianni sia già arrivato (a Trento)” (‘I think that G. beSUB already arrived in T.’) (13) 9 . The data presented here were gathered in several fieldwork sessions in Luserna: groups of fluent speakers (up to six) were administered translations task from Italian into Cimbrian (ca. 60 sentences with distractors and other syntactic phenomena). Informants are divided according to age. US WURK LXV (2016), p. 161 Table 1. The distribution of subjunctive and indicative Italian stimulus ‘che+SUB’ credo che tu… ‘I think that you…’ credo che lui… ‘I think that he…’ credo che noi… ‘I think that we…’ credo che voi… ‘I think that youPL…’ credo che loro… ‘I think that they…’ Cimbrian translation i gloabe azto+SUB i gloabe ke du+IND i gloabe azzar+SUB i gloabe ke er+IND *i gloabe azpar+SUB i gloabe ke biar+IND i gloabe azzar+SUB i gloabe ke dar+IND i gloabe azze+SUB i gloabe ke sa+IND To account for [ke+SUB] it is reasonable to assume a system made of two grammars coexisting in the bilingual mind: the mood system in Cimbrian becomes more and more parasitic on the mood system in the model language, Italian, mood features infiltrating the complementation system of the replica language through what could be dubbed “weak points” of the structure (see table 1), i.e. points where morphological ambiguity (such as indicative vs subjunctive) might favor an unmarked option: on the one hand the unmarked option initially favors the spreading of indicative as default mood in the whole paradigm, on the other hand it also affects the abstract features of ke, which ends up taking on the role of az in selecting the subjunctive (the marked option). To sum up, we put forward a three-step contact-induced change in mood selection: 1. the topmost left periphery hosts elements (mostly) endowed with a minimum of formal features (discourse-related particles, adverbials, conjunctions) 2. in language contact, what is less marked (e.g. indicative vs subjunctive) is likely to be assigned a default value, 1st p.pl. SUB changes into 1st p.pl. IND: this accounts for the second step, i.e. the use of ke+IND with non-factives verbs 3. the third step consists in the accessibility to the lower C-layer (due to pressure of the model); there has to be an Agree relation between Fin and T US WURK LXV (2016), p. 162 (MoodIrrealis): differently from Italian, Fin0 does not host an overt complementizer but Vfin instead. 2.3 Preliminary conclusion As our data confirm, syntactic interference per se in language contact does not obtain: this is in line with e.g. Abraham (2012). Nevertheless it is possible to agree on some preliminary starting points. In the same fashion as lexical words, functional words are also inserted in their “bare” form into the replica language, discarding their original feature array; however, differently from lexical words they do not assume -˗ at least initially -˗ the morphosyntactic feature characterization of the replica language (ke does not compete with az for the same position and feature characterization but instantiates a higher position not endowed with mood features). US WURK LXV (2016), p. 163 On the syntactic level just single abstract features enter the target language ([mood], etc.). We want to emphasize in particular the fact that abstract features (i.e. pieces of I-language) can enter alone the replica language and do not need a lexical item to “hold on to”. 3. Syntactic transfer in Basque-Spanish bilingual speakers: interrogative structures In many languages, constituent questions are characterized by the fronting of the wh-phrase into clause-initial position, often accompanied by locating the verbal complex in the second position. This is especially true of languages with basic SVO and VSO order. In contrast, SOV languages generally locate the wh-constituent in-situ, i.e. in the position of the constituents for which they substitute (Ultan, 1978; Siemund 2001: 1020). In peninsular Spanish, generally classified as SVO language, constituent questions normally exhibit wh-fronting and the direct adjacency of the whphrase and the verb is even obligatory. That is, nothing is allowed to occur between the two constituents, except some few adverbs like jamás ‘never’, todavía ‘still’ (Bosque & Gutiérrez-Rexach 2011: 449) or Polarity (negation). Interestingly, Basque, an SOV language and typologically very different to Spanish, also displays wh-fronting combined with obligatory adjacency of the wh-phrase and the verb (Hualde & Urbina 2003: 495-; Aldai 2011). (17) and (18) demonstrate the phenomenon for Spanish and Basque, respectively. (17) a. ¿Qué compra Pedro? ‘What does Peter buy?’ b. * ¿Qué Pedro compra? (18) a. Zer erosten du Pedrok? ‘What does Peter buy?’ b. * Zer Pellok erosten du? Since the influential work by Torrego (1984) and others, it has become clear that Spanish shows some exceptions regarding the obligatory adjacency of the wh-phrase and finite verb. It is generally assumed that in clauses with non-argumental wh-phrases, especially por qué ‘why’, adjacency is optional and not obligatory (see Uriagereka a.o. 1988). According to Rizzi (2001), this is explained by the assumption that a wh-phrase like por qué ‘why’ is generated in a higher position than argumental wh-phrases in an extended CP layer. This accounts for why in (19) the subject Juan can surface before or after the verbal complex (Torrego 1984): US WURK LXV (2016), p. 164 (19) ¿Por qué Juan quiere salir (Juan) antes de los demas? ‘Why does Juan wants to leave before the rest?’ However, Ordoñez (1998), Zubizarreta (2012) and others claim that the decisive factor for obligatory adjacency is not so much the argumental vs. non-argumental status of the wh-word, but the complexity of the wh-phrase. That is, complex or heavy wh-phrases seem to allow non-adjacency more readily than simple ones, as shown in (20) (Ordoñez 1998): (20) ¿A cuál de las chicas que han venido tu hermana había visto antes? ‘Which of the girls who came earlier had your sister seen?’ Note that these sentences seem to allow elements other than the subject to surface between the complex wh-phrase and the verb like a direct/indirect object or other constituents (Ordoñez 1998). A comparison with another leftward movement in Spanish, namely Clitic-Left-Dislocation (CLLD), reveals some interesting parallels. Normally, CLLD differs from whmovement in some crucial characteristics (i.a. clitic-doubling (CLD) or multiple CLLD). Nevertheless, following Ordoñez (1998), it seems that heavy wh-phrases behave in some aspects like a CLLDed constituent. First, for a sentence like (20), adding a doubling clitic does not render the sentence ungrammatical, as shown in (21). (21) ¿A cuál de las chicas que han venido tu hermana la había visto antes? ‘Which of the girls who came earlier had your sister seen?’ Second, there is a parallel structure between left-dislocated complex negative quantifiers and complex wh-phrases. As a consequence, Ordoñez (1998: 347) claims that complex whphrases in Spanish “are not in SpecCP but are left dislocated”, which would explain why adjacency is not obligatory between these kinds of wh-phrases and the verb. Hence, these sentences resemble the formerly described wh-sentences with por qué ‘why’. The case of Basque is different. Basque generally displays strict whfronting and obligatory adjacency (see (18)) with all wh-words except zergatik ‘why’ and nolatan ‘how/how come’, where, as in Spanish, direct adjacency is not obligatory (Hualde & Urbina 2003: 465): (22) Zergatik Jonek gainerakoak baino lehenago atera nahi du? ‘Why does John want to leave before the others?’ US WURK LXV (2016), p. 165 The main difference to Spanish lies in clauses with heavy wh-phrases. Basque always demands direct adjacency, no matter the length of the whphrase as shown in (23). (23) Neska horietako zein (*zure arrebak) bisitatu zuen lehenago? ‘Which of these girls had your sister visited before?’ In the context of an ongoing phd thesis, Simon Dold (IKER UMR5478University of Konstanz) investigates the relative acceptance of non-adjacency of the wh-phrase and the verb in Spanish by bilingual Basque-Spanish speakers as compared to monolingual Spanish speakers. A first pretest (Dold 2015) yields empirical data suggesting that the former group judges sentences with an intervening constituent as less grammatical than the latter one. The pretest was based on acceptance tests realized among 10 bilingual speakers and an equal number of Spanish monolinguals, both from within the Basque Country and from other areas of Spain. The ongoing study follows L2 acquisition methodologies (see e.g. Montrul, 2008:18; Meisel, 2011) to classify the speakers into four different groups. The preliminary results suggest that age and sequence of acquisition play a crucial role in the acceptability of monolingual Spanish word order configurations. Simultaneous (2L1) and early Basque-Spanish bilinguals differ significantly from the other bilingual as well as monolingual speakers. The former group seems to be more restrictive than the latter in the sense that they rate whsentences without direct adjacency in general as less grammatical. An obvious exception are clauses with por qué ‘why’, which do not observe such a difference between groups. This is not surprising as this is he same context where adjacency between the wh-word and the verbal complex is not required in Basque. The working hypothesis, supported by preliminary data, is that the different ratings can be explained by the influence of Basque grammar in the bilingual mind. As a reminder, in Basque, heavy wh-phrases behave identically as simple ones. In a split CP approach (cf. Rizzi 2001), they are moved to SpecFinP first and later to SpecFocP. Their Spanish counterparts, however, which are supposed to be left dislocated, are assumed to move to a higher position. Following a proposal by Rizzi (2001), we entertain the hypothesis that the wh-phrase in Spanish targets a dedicated wh-position higher than Fin. In other words, the main difference between the Spanish wh-question strategy and the Basque one, would be that in the Basque case, the whphrase targets the focus projection, whereas in Spanish the wh-phrase US WURK LXV (2016), p. 166 targets both the focus position and a higher Wh-projection. This is in line with what we know about the parallel behavior of focus fronting and whfronting in Basque (see Etxepare and Ortiz de Urbina, 2003). This higher Wh-position allows the intervention of topical phrases between the Whphrase and the finite verb. The parameter involved in this difference is ultimately a parameter concerning the lexicalization of the relevant features involved in wh-questions : as recent works have made manifest (Hagstrom, 1998; Cable, 2007; Slade, 2011), wh-questions manipulate three basic elements. On the one hand, there’s a special C-feature which introduces a set of propositions, a Q particle which maps into a choice function (an existential quantifier over a restricted set of alternatives), and a whindefinite which denotes a set and restricts the range of the quantification to the features expressed in the indefinite. Some languages, as is the case for Tlingit (as explored by Cable, 2007) or Japanese (Hagstrom, 1998), have independent exponents for the wh-indefinite and the Q-particle. The mutual syntactic independence of those two features makes certain configurations possible. Among other things it allows the Q-particle to target the C position independently, without the whphrase being pied-piped. This possibility is not allowed for Spanish, where the wh-indefinite and the existential quantifier (the Q-particle) are lexicalized together. Assuming that a whphrase must target at least two positions, a focal position (particularly clear for languages like Basque in which the syntax of wh-questions and of focus operators is basically identical) and another one, a higher C-related position encoding the question feature, the different lexicalization options represented by the two languages have an effect in the available derivations and their resulting configurations. Wh-phrases in Spanish are attracted to a higher position than Basque wh-phrases, as they are attracted to the ultimate position of the Q-feature itself. In Basque the Q-feature (null in this language but visible in Japanese, Tlingit and other languages) can target C without the wh-indefinite. The wh-indefinite only targets the focus position, which is adjacent to the verbal complex. We represent the two options as follows (24a corresponds to Basque; 24b to Spanish, copies in italics): (24) a. [CP Q C0 [FocP [Wh-Ind] Foc0 [IP …[WhP Q [WhP Wh-ind]]…]]] b. [CP [QP Q [WhP Wh-ind]] C0 [FocP [QP Q [WhP Wh-ind]] Foc0 [IP …[QP Q [WhP Wh-ind]] …]]] In (24b) the Q particle and the wh-indefinite are bound together, as a projection of Q, and they target the position corresponding to Q in the left US WURK LXV (2016), p. 167 periphery. In Basque, as in other languages which build their interrogative forms on the basis of so-called indeterminate pronouns (Kuroda, 1969; Haspelmath, 1997), the Q-particle is adjoined to the wh-indefinite, and it is not spelled out with it. Basically, there’s a lexical parameter in the merging of the Q-particle and the wh-indefinite. In Spanish the two features are merged together (a case of Set-Merge); in Basque, they are adjoined (a case of Pair-Merge).10 We gloss over that detail, sticking to the more traditional «adjunction» operation : (25) a. [QP Q [WhP Wh-ind]] b. [WhP Q [WhP Wh-ind]] (Spanish) (Basque) Now, what happens with the bilingual speakers that differ in the acceptance of wh-questions in Spanish in which the wh-phrase is not adjacent to the finite verb? The working hypothesis of the ongoing dissertation is that speakers of the first group (early bilinguals) treat wh-phrases in Spanish in the same way as they would in Basque. This explains their reservation about clauses without direct adjacency, since there is no position for the insertion of constituents between the wh-phrase and the finite verb, located in Fin in Basque. On the other hand, the general acceptance of intervening constituents in clauses with porqué ‘why’ is explained by the fact that the wh-phrase expressing ‘why’ is assumed to be base-generated in both languages in a higher position than SpecFinP or SpecFocP (Rizzi 2001). In our terms, this means that in that case the Q-particle and the wh-indefinite must have been spelled out together. This may seem like an ad hoc hypothesis, but the truth is that zergatik « why », unlike the other wh-phrases, is independently used as a C-element in Basque in the context of causal subordination in the central dialects which have been examined in this study: (26) a. Zergatik egin duzu hori? ‘Why did you do that?’ 10 . In recent minimalist literature ‘Set Merge’ and ‘Pair Merge’ are two technical concepts relating to the broader concept of Merge. Set and Pair Merge were first proposed in Chomsky 2004 to capture in particular argument-adjunct asymmetries: Set Merge takes two syntactic objects yielding an unordered set {α, β} which is symmetrical and binary whereas Pair Merge yields ordered pairs <α, β> which are by definition asymmetrical. Moreover, the two options may be distinguished by the presence versus absence of labelling (see Hornstein, 2009). US WURK LXV (2016), p. 168 b. Zergatik bai Why yes ‘For no specific reason’ (Lit. ‘Because yes’) b’ Zergatik presaka nabil eta joan beharra nuen ‘Because I was in a hurry and I had to leave’ In any case, we know from cross-linguistic comparison, that why-questions are special (Tsai, 1994; Shlonsky and Soare, 2011; Blochowiak, 2014). 3.1 Preliminary conclusions The correspondence between the sequencing and age of the acquisition of Spanish and the differential acceptance rates regarding those configurations in which the wh-word and the verb are not adjacent should be taken with the necessary caution, as the pretest involves a small population, statistically insufficient to draw any categorical conclusion. If the facts are confirmed by testing on a larger population, they may suggest a clear case of systemic syntactic transfer in bilingual first language acquisition, as has been observed in other cases (for a particularly relevant one, see Yip and Matthews, for Cantonese-English bilinguals, 2007). Unlike other cases, in which a shift in language dominance may seem to reverse the early transfer (but see Meisel, 2011 for a critical assessment of this idea), the Basque-Spanish case would look like a successful transfer, maintained during the linguistic life of the tested speakers as part of their syntactic competence. Basque data seem to point to the convergence of and the maintaining of the structures that are shared by the two languages: what is not present in both languages (marked options) is discarded to the point that it is not acquired. 4. General Conclusions In this article we have seen that contact phenomena manifest themselves in ways that appear superficially quite different, at a first sight. However, if one abstracts away from superficial differences one can easily recognize a common pattern: language contact can be basically thought of as two abstract feature systems (i.e. I-languages) in contact inside the bilingual mind. The contact between the two systems manifests itself in either favoring the transfer of specific features or promoting a pattern shared by the two languages.11 In general, a formal feature belonging to the model 11 . Rather on the contrary, lexical borrowing (even of functional words, like the lexical complementizer ke in Cimbrian) implies the deprivation of the original functional feature characterization. In fact, a lexical loanword, i.e. an E-language item, acquires the morphosyntactic features of the target language (cf. The sun is feminine, by US WURK LXV (2016), p. 169 language does not replace a feature of the target language, but just “fills a gap” in that system; in a similar way, no new word order pattern is added to the existing ones, but the system capitalizes on the shared one disfavoring the ones that are language-specific. The former case is represented by Frisian and Cimbrian, where the formal features of functional verbs and complementizers respectively, are transferred from the model languages (Standard Dutch and Romance) to the replica languages (Frisian and Cimbrian); the latter case is represented by the contact Basque-Spanish which favors influence in the opposite direction, from replica to model language: this does not come as a surprise if one assumes contact/attrition phenomena not to widen the possible set of word order patterns but, again, to restrict them discarding the ones that are not shared by the two languages. To sum up, if on the one hand contact at the syntactic level never implies “substitution of pieces” (either single features or specific word order patterns) on the other hand it seems to affect the speed of an otherwise expected diachronic development. In all three cases taken into account in this article, contact contributes to the acceleration of ongoing change by either favoring well-known grammaticalization paths (a.o. semantic bleaching) or reducing the range of variation patterns (selection of unmarked structures). It is important to note that in language contact the process of acceleration of change that we have discussed here might also have an evident counterpart in deceleration: in other words, contact might even promote the maintenance of conservative structures, as has often been observed in minority languages studies (cf. Abraham and Leiss 2013 and references cited there). If the hypothesis that contact affects the speed in grammar change proves to be on the right track, the apparent contradiction between opposite forces (i.e. innovation vs conservativity) finally finds a natural explanation. *Università di Verona **Universiteit Utrecht § Centre de Recherche IKER UMR5478 §^ Universität Konstanz Taeschner, 1983), whereas here, a lexical complementizer gets transferred as a “devoid category”. US WURK LXV (2016), p. 170 REFERENCES Abraham, W. (2012). “Philologische Dialektologie und moderne Mikrovarietätsforschung. Zum Begriff des Erklärstatus in Syn- und Diachronie”. In: Glauninger, M.M. & B. Barabas (eds): Wortschatz und Sprachkontakt im Kontext oberdeutscher Wörterbücher, Sprachatlanten und Sprachinseln. Werner Bauer zum 70. Geburtstag, 171-190. Wien: Praesensverlag. Abraham, W. & E. Leiss (eds) (2013). Dialektologie in neuem Gewand: Zu Mikro-/Varietäten-linguistik, Sprachenvergleich und Universalgrammatik. Hamburg: Helmut Buske Verlag. Aldai, G. (2011). “Wh-questions and SOV languages in Hawkins’ (2004) theory. Evidence from Basque”. Linguistics 49, 1079-1135. Bayer, J. (1999). “Final complementizers in hybrid languages”. Journal of Linguistics 35, 233-271. Bhatt, R. & J. Yoon (1992). “On the composition of COMP and parameters of V2”. In: Bates, D. (ed): The proceedings of the tenth West Coast Conference on Formal Linguistics, 41-53. Stanford: Center for the Study of Language and Information. Bidese, E., A. Padovan & A. Tomaselli (2012). “A binary system of complementizers in Cimbrian relative clauses”. Working Papers in Scandinavian Syntax 90, 1-21. (http://project.sol.lu.se/uploads/media/ Bidese_et_al_WPSS90_02.pdf) Bidese, E., A. Padovan & A. Tomaselli (2013). “Bilingual competence, complementizer selection, and mood in Cimbrian”, In: Abraham, W. & E. Leiss (eds): Dialektologie in neuem Gewand: Zu Mikro-/Varietätenlinguistik, Sprachenvergleich und Universalgrammatik, 47-58. Hamburg: Helmut Buske Verlag. Bidese, E. (2004). Die diachronische Syntax des Zimbrischen. Tübingen: Gunter Narr. Bidese, E. (ed) (2010). Il cimbro negli studi di linguistica. Padua: Unipress. Blochowiak, J. (2014). A theoretical approach to the quest for understanding. Semantics and pragmatics of whys and becauses. PhD thesis, University of Geneva. Bosque, I. & J. Gutiérrez Rexach (2011). Fundamentos de sintaxis formal. Madrid: Akal. Bree, C. van, & A.P. Versloot (2008). Oorsprongen van het Stadsfries. Leeuwarden: Afûk. US WURK LXV (2016), p. 171 Breu, W. (ed) (2005). L’influsso dell’italiano sulla grammatica delle lingue minoritarie. Problemi di morfologia e sintassi. Rende: Università della Calabria. Bybee, J., R. Perkins & W. Pagliuca (1994). The evolution of grammar: tense, aspect and modality in the languages of the world. Chicago: University of Chicago Press. Cable, S. (2010). The Grammar of Q. Q-Particles, Wh-Movement and PiedPiping. Oxford: Oxford University Press. Chomsky, N. (1986). Knowledge of Language. New York: Praeger. Chomsky, N. (2000). “Minimalist inquiries”. In: Martin, R., Michaels, D. & J. Uriagereka (eds.): Step by step, 89-155. Cambridge: MIT Press. Chomsky, N. (2004). “Beyond explanatory adequacy”. In: Belletti, A. (ed.): The cartography of syntactic structures. Vol. 3, Structures and beyond, 104-131. Oxford: Oxford University Press Cinque, G. (1999). Adverbs and functional heads: A cross-linguistic perspective. Oxford: Oxford University Press. Demirdache, H. & M. Uribe-Etxebarria (1998). Towards a Theory of the Diversity of Temporal Systems. (Manuscript) Dold, S. (in progress) Basque-Spanish Syntactic Transfer in Interrogatives. PhD dissertation, Université de Pau et des Pays de l’Adour-University of Konstanz. Dorian, N.C. (1981). Language Death. The Life Cycle of a Scottish Gaelic Dialect. Philadelphia: University of Pennsylvania Press. Erb, M.C. (2001). Finite auxiliaries in German. PhD Dissertation, Katholieke Universiteit Brabant. Grewendorf, G. & C. Poletto (2011). “Hidden verb second: the case of Cimbrian”. In: Putnam, M.T. (ed.): Studies on German Language Islands, 301-346. Amsterdam/Philadelphia: John Benjamins Publishing Company. de Haan, G.J. (2010). “Grammatical borrowing and language change. The dutchification of Frisian.” In: Hoekstra, J., W. Visser & G. Jensma (eds): Studies in West Frisian Grammar. Selected papers by Germen J. de Haan, 233-250. Amsterdam/Philadelphia: John Benjamins Publishing. Hagstrom, P. (1998). Decomposing questions. PhD dissertation, Massachusetts Institute of Technology. Distributed by MIT Working Papers in Linguistics. http://www.cog.jhu.edu/~hagstrom/ling.html Haspelmath, M. (1997). Indefinite Pronouns. Oxford: Oxford University Press. US WURK LXV (2016), p. 172 Heine, B. 2003. “Grammaticalization”. In: B.D. Joseph and R.D. Janda (eds): The Handbook of Historical Linguistics, 575-601. Blackwell Heine, B., & T. Kuteva (2003). “On contact-induced grammaticalization”. Studies in language 27, 529-572. Hickey, R. (2013). The Handbook of Language Contact. Malden, MA: Wiley-Blackwell. Hoekstra, E. (1994). “Positie-en bewegingsaspect bij selectie van de infinitief op-E of-EN in het Westfries en het Fries”. Taal en tongval 46, 66-73. Hooper, J.B. & S.A. Thompson (1973): “On the applicability of root transformations”. Linguistic Inquiry 4, 465-497. Hornstein, N. (2009). A theory of syntax: Basic operations and the minimalist program. Cambridge: Cambridge University Press. Hualde, J.I. & O. de Urbina (eds.) (2003). A Grammar of Basque. Berlin: Mouton de Gruyter. IJbema, A. (2001). Grammaticalization and infinitival complements in Dutch. PhD Dissertation, Utrecht University. Julien, M. (2007). “Embedded V2 in Norwegian and Swedish”. Working Paper in Scandinavian Syntax 80, 103-161. Kolmer, A. (2005). “L’elemento da come espletivo della posizione del soggetto enclitico pronominale nel Cimbro di Luserna (Trentino)”. In: Breu, W. (ed.): L'influsso dell'italiano sulla grammatica delle lingue minoritarie. Problemi di morfologia e sintassi. Atti del Convegno Internazionale - Costanza, 8.-11.10.2003, 55-78. Rende: Università della Calabria. Kolmer, A. (2012). Pronomina und Pronominalklitika im Cimbro. Untersuchungen zum grammatischen Wandel einer deutschen Minderheitensprache in romanischer Umgebung. Stuttgart: Steiner. Kuroda, S.Y. (1969). “On attachment transformations”. In S. Schane and D. Reibel (eds.), Modern Studies in English: Reading in Transformational Grammar, 331-351. Englewood Cliffs, NY: Prentice-Hall. Legate, J.A. & C. Yang (2002). “Empirical reassessment of the poverty of stimulus argument”. Linguistic Review 19, 151-162. Manzini, M.R. (2000). “Sentential Complementation”. In: Coopmans, P.; M. Everaert & J. Grimshaw (eds): Lexical specification and insertion, 241-268. Amsterdam: John Benjamins Publishing Company. Matras Y. (2013). “Contact, convergence and typology.” In: Hickey, R. (ed.): The Handbook of Language Contact, 66-85. Malden, MA: WileyBlackwell. US WURK LXV (2016), p. 173 Meisel, J. (2011). First and Second Language Acquisition. Cambridge: Cambridge University Press. Montrul, S. (2008). Incomplete Acquisition in Bilingualism. Re-examining the Age Factor. Amsterdam/Philadelphia: John Benjamins Publishing Company. Ordóñez, F. (1998), “Postverbal Asymmetries in Spanish”. Natural Language and Linguistic theory 16, 313-345. Padovan, A. (2011). “Diachronic clues to grammaticalization phenomena in the Cimbrian CP”. In: Putnam, M.T. (ed.): Studies on German Language Islands, 301-346. Amsterdam/Philadelphia: John Benjamins Publishing Company. Putnam, M.T. (ed) (2011). Studies on German language-islands. Amsterdam/Philadelphia: John Benjamins Publishing Company. Rizzi, L. (1997): “The fine structure of the left periphery”. In: Haegemann, L. (ed.), Elements of grammars, 281-337. Dordrecht: Kluver. Rizzi, L. (2001). “On the position ‘Int(errogative)’ in the left periphery of the clause”. In: Cinque, G. & G. Salvi (eds.): Current studies in Italian syntax. Essays offered to Lorenzo Renzi, 267-296. Amsterdam: Elsevier. Sasse, H.J. (1992). “Language decay and contact-induced change: Similarities and differences”. In: Brenzinger, M. (ed): Language death: factual and theoretical explorations with special reference to East Africa, 59-80. Berlin: Mouton de Gruyter. Shlonsky, U. & G. Soare (2011). “Where’s Why?”. Linguistic Inquiry 42, 651-669. Siemund, P. (2001) “Interrogative Constructions”, In: Haspelmath, M., König, E., Osterreicher, W. & W. Raible (eds): Sprachtypologie und sprachlichen Universalien. Ein Internationales Handbuch 2, 339-352. Berlin: Mouton de Gruyter. Slade, B. (2011). Formal and philological inquiries into the nature of interrogatives, indefinites, disjuction and focus on Sinhala and other languages. PhD Dissertation, University of Illinois. Taeschner, T. (1983): The Sun is feminine. A study on language acquisition in bilingual children. Berlin: Springer-Verlag. Tánczos, O. (2013). A mixed-headed CP layer in Udmurt from a historical perspective. Paper presented at the Newcastle Workshop on European Languages and Diachronic Linguistics. Thomason, S.G. (2001a). Language contact. An introduction. Washington, DC: Georgetown University Press. US WURK LXV (2016), p. 174 Thomason, S.G. (2001b). “Contact-induced typological change”. In: Haspelmath, M.; E. König; W. Oesterreicher & W. Raible (eds): Language Typology and Language Universals 2, 1640-1648. Berlin: Mouton de Gruyter. Torrego, E. (1984). “On Inversion in Spanish and Some of its Effects”. Linguistic Inquiry 15, 103-129. Tsai, W.T.D. (1999). On economizing the theory of A′-dependencies. PhD dissertation, MIT. Ultan, R. (1978). “The Nature of Future Tenses”. In: Greenberg, J. H. (ed.), Universals of Human Language, 83-123. Stanford: Stanford University Press. Uriagereka, J. (1988). “On government”. PhD Dissertation, MIT. Winford D. (2013). “Contact and borrowing”. In: Hickey, R. (ed.): The Handbook of Language Contact, 66-85. Malden, MA: Wiley-Blackwell. Yip, V. & S. Matthews (2007). The Bilingual Child. Early Development and Language Contact. Cambridge: Cambridge University Press. Zubizarreta, M.L. (2012). “A note on the Spanish Left Periphery”. In: Brugé, L., A. Cardinaletti, G. Giusti, N. Munaro & C. Poletto (eds), Functional Heads, The Cartography of Syntactic Structures, vol. 7, 112125. Oxford: Oxford University Press.