(PDF) When Moldovan meets Russian

ISSN 2385-4138 (digital) Isogloss 2023, 9(2)/2 https://doi.org/10.5565/rev/isogloss.201 1-36 When Moldovan meets Russian: intralinguistic variation in clitic climbing Ștefania Costea University of Cambridge

Received: 09-01-2022 Accepted: 20-06-2022 Published: 26-01-2023 How to cite: Costea, Ștefania. 2023. When Moldovan meets Russian: intralinguistic variation in clitic climbing. In Romance grammars: context and contact, eds. Alice Corr & Norma Schifano. Special issue of Isogloss. Open Journal of Romance Linguistics 9(2)/2, 1-36. DOI: https://doi.org/10.5565/rev/isogloss.201 Abstract This article offers an analysis of Moldovan complex predicates, focusing on the differences between structures with aspectuals and modals. It is shown that, under the influence of Russian, a minor pattern found in old Moldovan, whereby aspectuals and embedded infinitives instantiate a monoclausal construction, was generalized. As a consequence, pronominal clitics became free to raise to the matrix aspectual predicate, since they were no longer blocked by phasal (CP) barriers. Conversely, modal predicates, which select in both Russian and Moldovan full CP-complements, do not display clitic-climbing since clitics cannot skip over phasal boundaries. Following an overview of Romance and Slavic complex predicates, Moldovan complex predicates containing both aspectual and modal verbs are analysed. In this analysis infinitival complements and subjunctive complements, which show different syntactic behaviours, especially with respect to the phenomenon of clitic-climbing, will be treated separately. Finally, the special case of Moldovan within (Daco-)Romance is 2 Isogloss 2023, 9(2)/2 Ștefania Costea discussed, since Moldovan appears to be the only variety which has started to systematically display clitic-climbing out of subjunctive clauses introduced by SĂ. Keywords: restructuring contexts, clitic climbing, aspectual verbs, modal verbs, language contact. 1. Introduction The present paper focuses on the unique features of Moldovan, the variety of Daco- Romanian spoken in the Republic of Moldova, which, unlike other Daco-Romance varieties that show strong preferences for the use of either the infinitive or subjunctive, employs both infinitives and subjunctives after functional predicates (cf. 1). (1) Moldovan Începe să plouă /a ploua start.IND.PRS.3SG SĂ.SBJ rain.SBJ.3SG A.INF rain.INF ‘It starts raining’ Moreover, unlike other Daco-Romance varieties (2a–c), Moldovan seems to be the only Daco-Romance variety to allow clitic-climbing over both the subjunctive particle SĂ (cf. 2d) and the infinitival particle A (cf. also §4. below). While the ungrammaticality of (2a–c) can be explained by the fact that SĂ (similarly to A) heads a CP,1 a phasal domain through which there is no escape-hatch for clitics to climb (cf. Chomsky 2008:143ff.), the grammaticality of (2d) remains problematic. (2)a. Daco-Romanian, Romania *Îl termin să citesc b. Čipan Aromanian *Îl dipisescu sî citescu c. Megleno-Romanian *Lă sfârșoasc s citesc d. Daco-Romanian, Moldova Îl termin să citesc CL.ACC.M.3SG finish.IND.PRS.1SG SĂ.SBJ read.SBJ.1SG ‘I finish reading it’ In this article it will be argued that Moldovan aspectual predicates (cf. 2d) enter in monoclausal structures, while modal predicates instantiate biclausal structures whose complements are full CPs. Not by chance, Russian, with which Moldovan has 1 In examples (2a–c) the subjunctive marker SĂ is taken to occur in FinP (cf. Ledgeway 2009:17; Nicolae 2015:95ff.; cf. also §2.; §4. below). As for the phasal status of SĂ- subjunctives occurring in such complex predicates, it will be argued for in §4.3.1. below. Nevertheless, it is important to note that more than one possible analysis have been put forward; for example, Alboiu (2007) remarks that FinP can be both phasal and non phasal in (Daco-)Romanian. I thank an anonymous reviewer for making me aware of Alboiu’s (2007) analysis. When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 3 been in contact for over 200 years, displays a similar syntactic behaviour (cf. 3–4), inasmuch as aspectual predicates and their complements give rise to monoclausal structures (Stepanov 2007:87–92), while modal predicates and their complements yield biclausal structures (Krejci et al. 2016:18f.). (3)a. Moldovan [CP [Asp inceptive S-a început a rupe punga]] CL.REFL=AUX.PERF.3SG start.PPLE A.INF break.INF bag b. Russian [CP [Asp inceptive Načala lomat’sja sumka]] start.PAST.F.SG break.INF bag ‘The bag started breaking’ (4)a. Moldovan [CP [Mod epistemic Trebuie [CP să [IP le citesc]]]] need.IND.PRS.3SG SĂ.SBJ CL.ACC.F.3PL read.SBJ.1SG b. Russian [CP Mne [Mod epistemic nužno [CP [IP jx pročitat’]]]] I.DAT is.necessary them read.INF ‘I need to read them’ In other words, as the dominant language in the Republic of Moldova during the second half of the last century, Russian helped retaining some structural patterns (cf. linguistic convergence), while inevitably changing others. In particular, I argue that it has come to reshape the internal configurations of Moldovan complex predicates, triggering, for example, in the case of complex predicates containing an aspectual verb, a series of other syntactic phenomena, such as clitic-climbing and reanalysis of the markers A and SĂ as T-related heads. 2. The view from Romance 2.1 From Latin to Romance Although Latin did not have clitic pronouns, it had both weak and strong pronouns (cf. Salvi 2004:123ff.). The former, from which Romance clitics emerged, tended to occur after the first (focalized) element within the colon (cf. Adams 1994), hence the common claim that they were second position (Wackernagel) elements. Oversimplifying, in the case of a CP (cf. 5), for example, (weak) pronouns were to be found at the leftmost edge of the I-domain, immediately following a constituent in SpecCP (Ledgeway 2017:189; cf. also Salvi 2004:123–44; Ledgeway 2012a:191f.).2 (5) Latin (Pescarini 2020:24) [CP [DP alium] [IP [DP illa] amat]] another that loves ‘she loves another man’ 2 Another possible analysis, put forward by Pescarini (2020:23–7), is that (weak) pronouns bearing a [+familiar topic] feature followed focused constituents and occupied a projection called W(ackernagel)P(hrase), within the C-domain. 4 Isogloss 2023, 9(2)/2 Ștefania Costea In the case of Latin complex predicates, where functional verbs such as POTERE ‘can’ were followed/preceded by infinitives (cf. 6; cf. also Ledgeway 2012a:120–5, 2017:188ff.; Adams 2013:825–7), (weak) pronouns, despite being licensed by lexical verbs, acted syntactically as dependents of the matrix predicate occupying the clause- second position (Salvi 2004:150). (6) Latin (Salvi 2004:150) Tu eum videlicet non potuisti videre you him evidently not can see.INF ‘You evidently could not have seen him’ Late Latin and, indeed, early Romance display V(erb)-raising to a C-related position in matrix clauses (viz. Verb Second), with pronominal clitics generally occurring in proclisis when SpecCP was filled (cf. 7) (Ledgeway 2017:189–91).3 With respect to complex predicates (cf. 8), the syntactic behaviour of pronominal clitics in early Romance resembles that found with Latin (weak) pronouns, inasmuch as they act syntactically as dependents of matrix (functional) verbs (Salvi 2004:150). (7) Old Portuguese (Ledgeway 2012a:161) [CP [AdvP Entom] [C lhe disse [IP lhe disse then CL.DAT.3SG say.PRET.3SG CL.DAT.3SG say.PRET.3SG nostro Senhor]]]]] our Lord ‘Then our Lord said to him’ (8) Old French (Salvi 2004:150) [CP [DP nu] [C le irrums [IP le irrums we CL.ACC.M.3SG go.FUT.1PL CL.ACC.M.3SG go.FUT.1PL àsaillir fierement ú qu’il seit]]] attack.INF proudly where that=he be.SBJ ‘we are going to proudly attack him wherever he is’ Modern Romance verbs generally only move as high as the I-domain (cf. Schifano 2018), while clitics can target C-related, I-related, and V-related positions (Tortora 2014). For example, in a variety such as Borgomanerese (Piedmont, northern Italy) object clitics occur in the V-domain, i.e., the Aspterminative projection, (9; Tortora 2014:122ff.), whereas in most other Romance varieties they generally surface in the I- domain (10). (9) Borgomanerese (Tortora 2014:85) i porta-la SCL bring.IND.PRS.1SG=CL.ACC.F.3SG ‘I’m bringing it’ 3 More technically, enclisis is interpreted as the result of phonological cliticization of a weak pronoun (a phrase) to the verb, whereas proclisis is the result of syntactic cliticization of a clitic pronoun (a head) onto the finite verb in C (Ledgeway 2017:189). When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 5 (10) Megleno-Romanian (Capidan 1934:27) au arisí CL.ACC.F.3SG like.IND.PRET.3SG ‘(he) liked her’ As for complex predicates, there are two cases which need to be taken into account: (i) where functional verbs are followed by (bare) infinitives, and (ii) where they are followed by irrealis, i.e. subjunctive, complements. In the first case, clitic climbing to the functional predicate can be compulsory, as in Neapolitan (11), optional, as in Italian (12), and impossible, as in modern French (13).4 (11) Neapolitan (Ledgeway 2012a:120) ’e gghjammo a aiutà CL.ACC.3PL go.IND.PRS.1PL A help.INF ‘we are going to help them’ (12) Italian (Cinque 2006:134) a. verrò a parlarti come.FUT.1SG A talk.INF=CL.ACC.2SG b. ti verrò a parlare CL.ACC.2SG come.FUT.1SG A talk.INF ‘I will talk to you’ (13) French (Roberts 2010:86) Jean veut le manger Jean want.IND.PRS.3SG CL.ACC.M.3SG eat.INF ‘Jean wants to eat it’ Focusing on the internal configuration of these kinds of complex predicate, Cinque (2004; but cf. Rizzi 1978:117ff.; Cardinaletti & Shlonsky 2004) argues that examples such as (12) always instantiate a monoclausal, i.e. restructured, structure (approximately corresponding to a ‘lightly bi-clausal structure’, in Tortora’s 2014:137–46 terms), irrespective of the surface position of clitics. In short, when clitics remain low (cf. 12a), they are to be found in the V-related landing site, whereas when they occur before the matrix predicate (cf. 12b), they occupy an I-related position.5 4 Cf. Rizzi (1976:4, 1978:113); Jones (1993:142–54); Ledgeway (2000:83–176, 2012a:120–7, 2012b:470f., 2015:158); Ledgeway & Lombardi (2005:104f.); Parry (2005:227–30); Cinque (2004:17; 2006:151f.); Roberts (2010:79–86); Tortora (2014:137–46); Pescarini (2020:41f). 5 For an overview of Sardinian functional predicates which enter into both restructuring (i) and non-restructuring (ii) contexts, see Jones (1993:149–51) and the discussions in Kayne (2000:66) and Roberts (2010:87). (i) Sardinian (Jones 1993:142) L’appo cumintzatu a fákere CL.ACC.3SG=AUX.PERF.1SG start.PPLE A do.INF 6 Isogloss 2023, 9(2)/2 Ștefania Costea When functional predicates are followed by subjunctive complements, as happens in varieties that display a dual complementizer system and a general avoidance of the infinitive, pronominal clitics cannot climb out of the subjunctive clause over the subjunctive marker (cf. Cinque 2004:n.21; Ledgeway 2012b:469, 2015:157), e.g. CU in Salentino (14), M(U) in southern Calabrian (15), and S(I) in Aromanian (16).6 The only exception here seems to be Moldovan (17),7 which allows clitics to climb to the matrix verb. (14) Salentino (Ledgeway 2015:155) Vogghiu cu llu fazzu want.IND.PRS.1SG CU.SBJ CL.ACC.M.3SG do.SBJ.1SG ‘I want to do it’ (15) Calabrian (De Angelis 2017:52) Putimu m’u cattamu can.IND.PRS.1PL MU.SBJ=CL.ACC.M.3SG buy.SBJ.1PL ‘We can buy it’ (16) Čipan Aromanian Batu s-lu-adarǔ try.IND.PRS.1SG S.SBJ=CL.ACC.M.3SG=make.SBJ.1SG ‘I’m trying to make it’ (17) Moldovan Îl încep să îl cos CL.ACC.M.3SG start.IND.PRS.1SG SĂ.SBJ CL.ACC.M.3SG sew.SBJ.1SG ‘I start sewing it’ (ii) Sardinian (Jones 1993:149) Appo cumintzatu a lu fákere AUX.PERF.1SG start.PPLE A CL.ACC.3SG do.INF ‘I have begun to do it’ 6 Some exceptions from Italian and Spanish, where clitics climb over a complementizer (albeit not a specialized subjunctive one), are presented in Rizzi (1982:36) and Cinque (2004:19–21), respectively; cf. also the discussion in Roberts (2010:230 n.38). 7 Although Moldovan seems to be unique within Romance with respect to the phenomenon of clitic-climbing in such contexts, it patterns with other varieties such as Serbian(/Croatian) (cf. (i)) (Cinque 2006:142; cf. also Krapova & Cinque 2018:184 n.35). Note, however, that examples such as (i) are regarded as marginal in Serbian (cf. Progovac 1993:119), while examples such as (17) are generally (though not always) accepted by Moldovan speakers. (i) Serbo-Croatian (Progovac 1993:119) ?Milan ga želi da ga vidi Milan CL.ACC.M.3SG want.PRS.3SG DA.SBJ CL.ACC.M.3SG see.SBJ/PRS.3SG ‘Milan wishes to see him’ When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 7 2.2. Daco-Romance 2.2.1. Daco-Romanian, Istro-Romanian, Megleno-Romanian, Aromanian Examining structures in which functional predicates occur, the first distinction that needs to be made within Daco-Romance is between varieties that are in contact with languages that actively employ subjunctives in these contexts and varieties that are not (cf. Table 1 below). Table 1. Complex predicates within Daco-Romance + Contact language functional V + functional V + infinitive subjunctive Daco-Romanian – (generally) no yes (Romania) (only allowed with a putea ‘can’) Aromanian + (standard) Albanian/Greek (generally) no yes (functional V + subjunctive; (only allowed with cf. Turano 2017) va ‘must’ and lipseaște ‘must’) Istro-Romanian + Croatian yes (generally) no (functional V + infinitive;8 cf. Katičić 1986; Čamdžić & Hudson 2002:326) Megleno-Romanian + Bulgarian/Macedonian (generally) no yes (functional V + subjunctive; (only allowed with cf. Turano 2017) puteari ‘can/may’, trăbuiri ‘need to/should’, and țireari ‘want’) Leaving Daco-Romanian aside for the moment (but cf. §2.2.2.; cf. also (18) below), Megleno-Romanian (19) seems to pattern with Aromanian (20), displaying an extensive use of the subjunctive at the expense of the infinitive (cf. Capidan 1925:169; Atanasov 1976:144f.; Mišeska Tomić 2006:540 for Megleno-Romanian; Capidan 1932:548–50; Saramandu 1984:460; Mišeska Tomić 2006:559 for Aromanian), a syntactic feature they share with their contact language(s). By the same token, Istro- Romanian (21), which is in contact with Croatian which usually, but not exclusively, employs infinitives after functional predicates (Kovačević & Milićev 2018:149), shows a higher frequency of the infinitive and a very limited use of the subjunctive (cf. Dragomirescu & Nicolae 2020:7; Corbeanu 2020). (18) Daco-Romanian, Romania Încep să alerg start.PRS.1SG SĂ.SBJ run.SBJ.1SG ‘I start running’ 8 While it is true that functional predicates can be followed by both DA + present indicative and infinitives, the latter are reported to be more frequent in these contexts in Croatian (Kovačević & Milićev 2018:149; cf. also Bailyn 2010). 8 Isogloss 2023, 9(2)/2 Ștefania Costea (19) Megleno-Romanian, Romania Ăncepui si gutuvesc țiva dulți, start.PRET.1SG SI.SBJ cook.SBJ.1SG something sweet că-i zua lui that=is day.DEF his/her ‘I started to cook something sweet because it’s his/her birthday’ (20) Čipan Aromanian, Romania Ahurhi s antreabî start.PRET.3SG S.SBJ ask.SBJ.3SG ‘He started asking’ (21) Istro-Romanian (Sârbu & Frățilă 1998:257) åm poșnit lucrå AUX.PERF.1SG start.PPLE work.INF ‘I started working again’ Within Daco-Romance, it seems that clitic-climbing is possible when functional verbs are followed by bare infinitives (cf. 22)9 and blocked when they are followed by the subjunctive, i.e., S(I)/S(Ă)-clauses (cf. 23). Formally, this syntactic behaviour can be captured by assuming that, unlike bare infinitives which are not phasal domains and hence do not block the spreading of the feature [finite] and therefore allow clitics to raise to the matrix verb (cf. Tortora 2015:121ff.), S(I)/S(Ă)- clauses instantiate CPs which are phasal barriers that block clitic-climbing (Chomsky 2001:12–4, 2008:143ff.).10 (22) a. Megleno-Romanian (Capidan 1925:169) nu u putu junziri b. Daco-Romanian, Oltenia nu o putu ajunge not CL.ACC.F.3SG can.IND.PRET.3SG reach.INF ‘he couldn’t reach it’ (23) a. Daco-Romanian, Romania *Îl termin să citesc b. Čipan Aromanian *Îl dipisescu sî citescu c. Megleno-Romanian *Lă sfârșoasc s citesc CL.ACC.M.3SG finish.IND.PRS.1SG SĂ.SBJ read.SBJ.1SG ‘I finish reading it’ 9 For a discussion of Istro-Romanian clitic-climbing, see Dragomirescu & Nicolae (2020). 10 For a different analysis, see Alboiu (2007); Krapova & Cinque (2018:158–64). When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 9 2.2.2. Daco-Romanian: Romania and Moldova Following functional predicates old Daco-Romanian actively allowed both infinitives, with (24a) or without (24b) the marker A, and subjunctives (25) (cf. also Nicolae & Niculescu 2016:60). (24) a. Old Romanian (SVI.~1670:243v, Wallachia) poate a plăti datoriia can.IND.PRS.3SG A.INF pay.INF debt.DEF ‘he can pay the debt’ b. Old Romanian (CLM.1700–50:252v, Moldova) așea să poate dzice likewise SE.REFL can.IND.PRS.3SG say.INF ‘the same can be told’ (25) Old Romanian (DÎ.IV.1573, Wallachia) care egumeni nu vor căuta să ţie those monks not AUX.FUT.3PL try.INF SĂ.SBJ take.care.SBJ.3PL acest iaz this lake ‘those monks who will not try to take care of this lake’ Clitic-climbing out of bare infinitival clauses (26) and, more rarely, out of A- infinitival clauses is attested (27). (26) Old Romanian (NL.~1750–66:3, Moldova/Wallachia, Nicolae 2019:104) nu le pot oamenii not CL.ACC.F.3PL can.IND.PRS.3PL people.DEF crede believe.INF ‘people can’t believe them’ (27) Old Romanian (CLM.1700–50:317v, Moldova) i-au început a slăbi CL.ACC.M.3PL=AUX.PERF.3PL start.PPLE A.INF leave.alone.INF ‘they started to leave [our armies] alone’ Nicolae & Niculescu (2016:60f.) note that old Romanian pronominal clitics were allowed to climb over the subjunctive complementizer SĂ (cf. 28), remarking, though, that it happened ‘rarely’. Although a quantitative analysis is needed for the whole of old Daco-Romanian to verify this conclusion, I did not identify any such instances of clitic-climbing in my corpus, which covers the Moldovan area. The only pattern I identified is that in (29) where the clitic remains within the subjunctive clause. (28) Old Romanian (CC2.1581:220, Transylvania) nu o poate să adaugă not CL.ACC.F.3SG can.IND.PRS.3SG SĂ.SBJ add.SBJ.3SG ‘he cannot add it’ 10 Isogloss 2023, 9(2)/2 Ștefania Costea (29) Old Romanian (FD.1582–604:511v, Moldova) nu pot să-l scoață den not can.IND.PRS.3PL SĂ.SBJ=CL.ACC.M.3SG get.out.SBJ.3PL from moarte death ‘[they] cannot bring him back to life’ Finally, a pattern found in old Daco-Romanian is that where clitics are realized twice, namely once in the domain of the modal/aspectual verb and once in the domain of the infinitival/subjunctive verb (30; cf. Nicolae & Niculescu 2016:60f.). (30) Old Romanian (Bert.1774:14r, Moldova) Cine-m va vre să-m who=CL.DAT.1SG AUX.FUT.3SG want.INF SĂ.SBJ=CL.DAT.1SG fie mie prietin be.SBJ.3SG me.DAT friend ‘Who wants to befriend me’ Similarly to old Romanian, both (dialectal) Romanian (31) (cf. Reinheimer & Tasmowski 2005:203 n.100; Dragomirescu 2013:201) and Moldovan (32) also show structures in which the clitic is realized twice, once on the functional verb and once on the subjunctive(/infinitive) (cf. also Ledgeway 2016:8).11 Nevertheless, it is worth mentioning that, at least in the case of old Moldovan, which is covered in my corpus, instances of such multiply spelt-out clitics prove rather rare (cf. Table 2), whereas in present-day Moldovan it is a widespread phenomenon. (31) Daco-Romanian, Oltenia Puloverul l-am început sweater.DEF CL.ACC.M.3SG=AUX.PERF.1SG start.PPLE să-l tricotez, (nu fularul) SĂ.SBJ=CL.ACC.M.3SG knit.SBJ.1SG not scarf.DEF ‘I started knitting the sweater(, not the scarf)’ (32) Moldovan Cozonacul a să-l încep sweet.bread.DEF A.SĂ.FUT=CL.ACC.M.3SG start.IND.PRS.1SG să-l fac acum, (nu prăjitura) SĂ.SBJ=CL.ACC.M.3SG make.SBJ.1SG now not cake.DEF ‘I’ll now start cooking the sweet bread(, not the cake)’ 11 Similar patterns are also attested elsewhere in Romance, e.g. in Cosentino (i) (Ledgeway 2000:298 n.8; Cinque 2004:17; cf. also Tortora 2014:148f. for a discussion on Piedmontese dialects). (i) Cosentino (Ledgeway 2000:298) l’amu pruvatu a ru leja CL.ACC.3SG=AUX.PERF.1PL try.PPLE A CL.ACC.3SG read.INF ‘we tried to read it’ When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 11 Table 2. Multiple spelt out clitics and clitic-climbing in old Daco-Romanian functional verbs functional verbs functional verbs without multiple spelt with multiple spelt with clitic-climbing out clitics out clitics FD.1592–1604 170 2 0 CLM.1700–50 244 3 16 ULM.~1725 225 5 9 Bert.1774 153 2 2 Although to my knowledge there are no formal explanations for this phenomenon, Dragomirescu (2013:201) notes that it ‘shows the possibility of the embedded subjunctive structure undergoing the process of argument composition’. In more technical terms, we can apply Tortora’s (2015:122ff.) theory of feature harmony12 and combine it with Ledgeway’s (2015:n.6) proposal about how phasal domains may act as barriers13 for feature spreading. For both Romanian and Moldovan it would mean that, although SĂ still heads a phase, the phasal status of this is deficient. On the one hand the fact that clitics are to be found in the embedded domain clearly proves that SĂ still heads a phase (albeit a degraded one), i.e. it is a ‘barrier’, that initially inhibits the spreading of the [finite] feature down the embedded clause (cf. the informal representation in (33a) below), but on the other the fact that the feature [finite] is now gradually allowed to percolate down the embedded clause, which, in turn, forces the pronominal clitic to attach to the matrix verb, highlights the weakened status of the phase (cf. (33b)). (33) a. [CP [TP T[finite] … [CP să [TP T […] matrix clause embedded clause b. [CP [TP T[finite] … [CP să [TP T […/finite] matrix clause embedded clause Indeed, the degraded status of phases embedded under functional predicates has previously been discussed in the literature. Among others, Krapova & Cinque (2018:160) note that such subjunctives are untensed in Balkan languages (Daco- Romanian included), receiving their relevant [T] features from the matrix C. 12 This theory states that the T head in the I-domain provides the feature [finite] to the next lower head (say, F1), and then F1 provides this feature to the next lower head (say, F2), and so on; if the [finite] feature iterates all the way down (e.g. including to the V-related cliticization site), pronominal clitics cannot attach to this V-domain, as they cannot attach to a host bearing the [finite] feature. Thus, pronominal clitics are forced to surface in the I-related position. However, if there occurs a barrier (as in Borgomenerese) at the level of V-domain, then the iteration of the [finite] feature is blocked, rendering the V-domain (‘Z head’ in Tortora’s terms) a possible placement of object clitics (Tortora 2014:122ff.). 13 Discussing the Salentino data (cf. §2.1.), Ledgeway (2015:n.6) notes that ‘the phase- head status proposed for CU […] immediately translates as a barrier in Tortora’s system. […] [I]t inhibits the spreading of the [finite] feature from the matrix T, thereby allowing the clitic to adjoin to the Z head in the embedded I-domain’. 12 Isogloss 2023, 9(2)/2 Ștefania Costea Nevertheless, although the embedded C is somewhat defective with respect to [T] features, it seems to still be able to host, albeit rarely, focused constituents in dialectal Romanian (cf. 34).14 As for Moldovan examples resembling (35), there are two possible interpretations: the fronted elements are either hosted in the left periphery of the embedded subjunctive clauses, or are scrambled within the IP (cf. the discussion in §4.3.2. below). (34) Daco-Romanian, Oltenia Cocoșul l-am început [FocP de ieri] roaster.DEF CL.ACC.M.3SG=AUX.PERF.1SG start.PPLE of yesterday să-l fierb SĂ.SBJ=CL.ACC.M.3SG boil.SBJ.1SG ‘As of yesterday I’ve started cooking the roaster’ (35) Moldovan L-am început [FocP cu copilul] CL.ACC.M.3SG=AUX.PERF.1SG start.PPLE with child.DEF să-l citesc SĂ.SBJ=CL.ACC.M.3SG read.SBJ.1SG ‘I started reading it with my child’ Nevertheless, before proceeding to a formal analysis of Moldovan complex predicates, an overview of complex predicates in Old Church Slavonic and Russian needs to be provided in order to assess the role, if any, of linguistic contact. 3. The view from Slavic 3.1. Old Church Slavonic Old Church Slavonic had only a few cases of ‘true’ pronominal clitics, i.e. dative clitics for first and second persons, with reflexive sę and accusative personal pronouns (e.g. mę, tę) acting instead as ‘semi-clitics’ (Vaillant 1977:173;261; Večerka 1989:42; cf. Migdalski 2015:186f.). The difference between the two consists in the fact that, while the former arguably tend to occur in second position (36), i.e. following the fronted verb or C-related elements such as complementizers (e.g. jako ‘as’, a(g)da ‘that…not’, ašte ‘if’) (Krapova & Cinque 2018:171; cf. also Krapova & Dimitrova 2016), the latter can also occur lower in the structure (37) adjacent to the verb (Radanović-Kocić 1988:143; Migdalski 2015:187; cf. also Vaillant 1977:172). (36) Old Church Slavonic (Krapova & Cinque 2018:171) [Foc div’no čoudo] ti imam sъpovjadati wonderful miracle CL.DAT.2SG have.1SG tell.INF ‘I have to tell you a wonderful miracle’ 14 Note that, unlike Moldovan, which seems to display scrambling within the IP of (weak) topics and a series of focalized constituents similarly to Russian (cf. Dyakonova 2009:124–31), dialectal Romanian does not have such a feature. When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 13 (37) Old Church Slavonic (Radanović-Kocić 1988:154) Ašte desnaě tvoě rōka sъblažbaetъ tę if right your hand sin.PRS.3SG you.ACC ‘If your right hand causes you to sin’ Within complex predicates, pronominal clitics can (optionally) climb, for example, to modal functional predicates such as moč ‘can’ and iměti ‘have (to)’, following the first CP element (38). In such cases, the fact that the clitic is an argument of the lower verb but raises to the matrix verb provides a clear indication of the restructuring character of the construction. Nevertheless, the more frequent pattern is that whereby the clitic occurs in between the modal and the lexical verb (39) (Krapova & Cinque 2018:171f.; cf. Pancheva 2005). (38) Old Church Slavonic (Krapova & Cinque 2018:171) čto ti mogǫt dati? what CL.DAT.2SG can.PRS.3PL give.INF ‘what can they give you?’ (39) Old Church Slavonic (Krapova & Cinque 2018:172) imamъ ti něčъto rešti have.(to).PRS.1SG CL.DAT.2SG something tell.INF ‘I have something to tell you’ 3.2. Russian Unlike Old Church Slavonic, (modern) Russian does not display any clitic (or ‘semi- clitic’) pronouns, with dative pronouns instantiated by strong pronouns, e.g. OCS mi was replaced by Rus mne ‘me’, and accusative ‘semi-clitics’ undergoing a similar evolution, e.g. OCS mę was replaced by Rus menja ‘me’. The only exception is represented by the Old Church Slavonic ‘semi-clitic’ reflexive sę, which has as Russian counterpart the particle -sja (that becomes -s’ after vowels),15 which is obligatorily enclitic to the verb (cf. 40) (Vaillant 1977:174;262). (40) Old Church Slavonic Russian bojati sę ⇒ bojat’sja be.afraid.INF REFL be.afraid.INF=REFL ‘to be afraid’ Hence, cases of restructuring cannot be verified through the placement of pronominal clitics (cf. Krapova & Cinque 2018:172). Nevertheless, other tests indicate whether Russian functional verbs enter into monoclausal or bi-clausal structures. For example, in the case of aspectual predicates (e.g. prodolžat’ ‘continue’, načinat’ ‘start’), their monoclausal character is proven,16 among other things, by the fact that 15 Russian -sja/s’ are used for all persons and numbers. 16 However, not all Russian aspectual verbs enter into monoclausal structures. For example, putat’sja ‘try’ (lexicalizing Aspconative) takes a CP complement (cf. Stepanov 2007:96 n.11). 14 Isogloss 2023, 9(2)/2 Ștefania Costea Long Object Preposing can take place (41), and that adverbs such as vsegda ‘always’ cannot occur twice (42) (Stepanov 2007:93f.; cf. Rizzi 1978:113; Cinque 2004:5f.). (41) Russian Kniga prodolžaet čitat’sja mal’čikom book.NOM continue.PRS.3SG read.INF=REFL boy.INSTR ‘The book continues to be read by the boy’ (42) Russian (Stepanov 2007:89;94) *On vsegda načinaet vsegda opazdyvat’ na he always begin.PRS.3SG always be.late.INF on zanjatija (kogda emu len’ vstavat’) classes when he.DAT lazy get.up.INF ‘He always begins to always be late for classes (when he doesn’t feel like getting up)’ By contrast, modal verbs such as dolžen ‘must/ought/should’ and moč ‘can’ enter into bi-clausal structures,17 as proven, for example, by the impossibility of Long Object Preposing (43) (Krejci et al. 2016:19; but cf. Schoorlammer 1994:417; Stepanov 2007:99). Additional evidence for a bi-clausal analysis is provided by contrastive topics18 occurring in the (higher) left periphery of the embedded infinitive, irrespective of whether they are introduced by non-verbal (cf. 44a; Burukina 2019:16 n.12) or verbal deontic modals (cf. 44b).19 (43) Russian (Krejci et al. 2016:19) *Kniga budet dolžna čitat’ (Ivanom) book.NOM AUX.FUT.3SG should.3SG.F read.INF Ivan.INSTR ‘The book will need to be read by Ivan’ 17 Krejci et al. (2016:18f.) convincingly argue against a restructuring analysis of such structures in Russian, stating that modal predicates ‘select complements that are at least the size of TP’ (my underlining). Moreover, even Schoorlammer (1994:417f.) admits that deontic modals are not base-generated in a functional projection, inasmuch as they can combine with aspectual prefixes (which they incorporate in the Asp-field) (cf. i). (i) Russian (Schoorlammer 1994:418) Ona smogla prijti she can.past.F.PF come.INF.PF ‘She was able to come’ Also, relevant here is the fact that perfective aspect is marked twice, i.e. once on the functional verb and once on the lexical verb. As Rizzi (1978:155f.) observes, monoclausal structures cannot contain more than one perfective aspectual marker. 18 Russian contrastive topics occur exclusively in the C-domain and bear the marker ‘- to’ (Dyakonova 2009:129–31). 19 We deliberately leave aside the Russian modal xotet’ ‘want’, which has a more complex syntactic behaviour (cf. also the discussion in Cinque 2004:21f.). Cf., however, Schoorlammer (1994:407); Stepanov (2007:91–6); Krejci et al. (2016:17), who unanimously argue for a bi-clausal analysis of complex predicates containing it. When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 15 (44) a. Russian (Burukina 2019:16 n.12) Možno v sredu-to projektu zakončit’sja? allowed on Wednesday=TO project.DAT complete.INF ‘As for (this) Wednesday, is it allowed for the project to be completed by then?’ b. Russian ?On možet projekt-to ran’še zakončit’, no…20 he can.PRS.3SG project.ACC=TO earlier complete.INF but ‘He is able to/can complete (this) project earlier, but…’ All in all, it is highly relevant for the discussion of Moldovan complex predicates to note that, in general, Russian aspectuals such as načinat’ ‘start’ or prodolžat’ ‘continue’ are found in monoclasual structures, while (deontic) modals such as dolžen ‘must/ought/should’ and moč ‘can’ tend to select for CP-complements. 4. Clitic-climbing in Moldovan 4.1. Complex predicates: the state-of-the-art Diachronically, old Moldovan texts (16th to 18th c.) show that infinitives and subjunctives were both employed after functional predicates, although to varying degrees (cf. Table 3). For example, it seems that a vrea ‘want’ selected almost exclusively a subjunctive complement, whereas aspectual verbs mainly took infinitival complements. Table 3. The structure of old Moldovan complex predicates a trebui a vrea aspectuals a putea ‘must’ ‘want’ ‘can’ +subj. +inf. +subj. +inf. +subj. +inf. +subj. +inf. FD.1592–604 2 0 56 0 4 14 55 25 AOD.1675–6 0 0 4 0 0 4 1 2 CLM.1700–50 0 0 51 1 3 59 24 85 21 ULM.~1725 2 0 68 5 1 31 20 88 Bert.1774 15 0 58 0 1 19 48 1 20 Notably, my Russian informants (from both Moldova and Russia) generally accepted (44b) but rejected a structure such as (i) (which they reformulated as (ii)). Indeed, this represents another argument in favour of the fact that aspectual verbs instantiate a monoclausal structure, thus not displaying a (higher) left periphery for the embedded infinitive, while (deontic) modals display a bi-clausal structure. (i) Russian *Ja načnu projekt-to pisat’, a… I start.PF.1SG project=TO write.INF but (ii) Russian Projekt-to ja načnu pisat’, a… project=TO I start.PF.1SG write.INF but ‘I will start to write (this) project, but…’ 21 Out of 5 occurrences, 4 are with bare infinitives, i.e. without the infinitival marker A, and one with a ‘long’ infinitive ending in -re. 16 Isogloss 2023, 9(2)/2 Ștefania Costea Being in contact with a language that, with very few exceptions, generalized the use of infinitives after functional predicates, i.e. Russian, Moldovan has retained infinitives in certain complex predicates (for a more nuanced analysis, cf. §5 below). Nevertheless, it must be noted that infinitives are ungrammatical after modals such as a trebui ‘must’22 or a vrea ‘want’ (cf. 45), a situation which contrasts with Daco- Romanian varieties spoken in Maramureș, Crișana, and Cernăuți, where a trebui ‘must’ and a vrea ‘want’ can be followed by the infinitive (cf. 46; Marin et al. 2000:86). Alongside infinitival complements, Moldovan regularly employs subjunctives after a series of functional predicates (cf. 47), thus making it unique amongst (documented) Daco-Romance varieties (cf. Table 1 above). (45) Moldovan *Vreau /*Trebuie a pleca repede want.IND.PRS.1SG /must.IND.PRS.3SG A.INF leave.INF quickly ‘I want to/must leave quickly’ (46) Daco-Romanian, Maramureș (Marin et al. 2000:86) dac-ar trebui a da if=AUX.COND.3SG must.INF A.INF give.INF ‘if one must give’ (47) Moldovan Începe să plouă /a ploua start.IND.PRS.3SG SĂ.SBJ rain.SBJ.3SG A.INF rain.INF ‘It starts raining’ Before proceeding to a proper formal analysis of clitic climbing in Moldovan complex predicates, it is useful to have a full list of functional verbs found in this variety. As can be seen in (48) below (adapted from Cruschina & Ledgeway 2016:562), some are enriched by alternative Russian loans which have been adapted to Daco-Romanian morphology (cf. Rus. probovat’ and Md. a probui ‘try’).23 22 Only SĂ-subjunctives and/or clausal DE-supines (cf. (i); cf. also Dragomirescu & Nicolae 2016; Costea 2021) can follow the modal a trebui ‘must’; interestingly, clitic climbing over SĂ and DE is always blocked in these contexts, suggesting that indeed SĂ and DE head (phasal) CPs. (i) Moldovan Trebu(ie) [CP[FinP de [NegPnu [ClPle [IPmai dat apă]]]]] must.IND.PRS.3SG DE.SUP not CL.ACC.F.3PL anymore give.PPLE water ‘One should stop giving them water’ 23 In many cases, Moldovan clitics mirror Russian strong pronouns and/or the reflexive particle (but not at a syntactic level); cf. Rus. (Moldova) mne polučilos’ (I.DAT succeed.PAST.N.SG=REFL) and Md. mi s-a primit (CL.DAT.1SG SE.REFL=AUX.PERF.3SG succeed.PPLE) ‘I succeeded’. When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 17 (48) [Modepistemic/alethic a trebui ‘must’, a (se) putea ‘can’ [Modvolitional a vrea ‘want’ TP [Aspterminative a se opri, a se ogoi (cf. a (se) rotunji, a (se) zacrugli) ‘stop’ [Aspcontinuative a continua ‘continue’ [Modobligation/ability a trebui ‘must’, a avea ‘have to’, a putea ‘can’ [Aspfrustative/success a reuși (cf. a i se primi, a i se poluci, a dovedi) ‘succeed’ [Modpermission a putea ‘can’ [Aspconative a încerca (cf. a probui) ‘try’ [Aspinceptive a începe ‘start’ [Aspcompletive a termina ‘finish’ [VP V … 4.2. Complex predicates with infinitives: two phases > one phase 4.2.1. Old Moldovan Nicolae (2015:66) convincingly argues that the infinitival marker A (< Lat. AD ‘to(wards)’) heads in (old) Daco-Romanian a CP, simultaneously realizing Fin° and Force° (cf. 49). Indeed, knowing that CPs are phases (Chomsky 2001:12, 2008:143ff.), with their domains generally not being accessible to operations outside the phase, albeit with the exception of their heads and their edges in accordance with the Phase Impenetrability Condition (PIC; Chomsky 2000:108, 2001:13f.), we are provided with a straightforward explanation as to why clitic-climbing cannot take place in old Moldovan out of infinitival clauses such as those in (50).24 (49) (Old) Daco-Romanian (Nicolae 2015:66) [CPa [NegP nu [ClPîl [IPciti]]]] A.INF not CL.ACC.M.3SG read.INF ‘for it not to be read’ (50) a. Old Moldovan (CLM.1700–50:255v) au început [CPa-i spune toate] AUX.PERF.3PL start.PPLE A.INF=CL.DAT.M.3SG tell.INF all.F.PL ‘they started to tell him everything’ b. Old Moldovan (CLM.1700–50:231r) n-au putut nice silihtariul not=AUX.PERF.3SG can.PPLE not.even henchman.DEF [CP a-i trimite alte isprăvi] A.INF=CL.DAT.3SG send.INF other news ‘not even the henchman was able to send him any news’ 24 As for the cases in which the infinitival marker is absent (which are attested in (old) Daco-Romanian, as well as in present-day Romanian and Moldovan), i.e. when the bare infinitive follows the functional verb a putea ‘can’, Nicolae (2015:14ff.) proposes that bare infinitives instantiate vPs (cf. also Pană Dindelegan 2013:218–20), as informally sketched in (i) below. (i) Romanian/Moldovan [NegP nu [ClPîl [IP poate [vP citi]]]] not CL.ACC.M.3SG can.IND.PRS.3SG read.INF ‘he cannot read it’ In what follows, however, we will not discuss this particular situation, focusing only on A- infinitives. 18 Isogloss 2023, 9(2)/2 Ștefania Costea Nevertheless, cases in which the clitic can climb over the infinitival marker A can be found in old Moldovan (cf. 51), although extremely rarely. However, if it is assumed that Moldovan infinitivals can only instantiate CPs (in the case of A- infinitives; or vPs, in the case of bare infinitives), examples such as (51), where we have an A-infinitive with clitics raised to the matrix verb, are unexpected, inasmuch as pronominal clitics cannot skip over phasal boundaries if the escape hatch, the phase head, is already lexicalized. Hence, I propose a more nuanced analysis of old Moldovan infinitival clauses. In particular, alongside the formal analyses proposed by Nicolae (2015), according to which infinitival complements can instantiate either CPs (or vPs), I will also take infinitival clauses to be able to also instantiate IPs.25 This way, the climbing of pronominal clitics proves unproblematic, as the PIC imposes no restriction on accessing elements from non-phasal domains (Chomsky 2001:14).26 (51) Old Moldovan (ULM.~1725:A-2.f.18v) [CP în zilelé acestui Bogdan vodă s-au in days.DEF of.this Bogdan voivode SE.PASS=AUX.PERF.3SG început a da dajde] start.PPLE A.INF give.INF tributes ‘when Bogdan voivode reigned, they started to give tributes’ Indeed, the ‘weakened’ phasal status (cf. §2.2.2. above) of the CPs headed by A is proven by instances where pronominal clitics are given multiple realization, once on the functional head and once on the embedded infinitive (cf. 52). In other words, the fact that clitics are to be found in the embedded domain clearly proves that a still heads a phase, i.e. it is a ‘barrier’, that initially inhibits the spreading of the [finite] feature down the embedded clause, such that clitics adjoin to the embedded infinitive. However, this phase is starting to ‘weaken’, inasmuch as it (marginally) allows for the feature [finite] to percolate down the embedded clause, which, in turn, forces the pronominal clitic to (also) surface on the matrix verb (cf. Rizzi 1976, 1978; Tortora 2014:122ff.; Ledgeway 2015:n. 6). (52) Old Moldovan (Bert.1774:51v) și cu această mijlocire îi and with this way CL.ACC.M.3PL începe a-i număra start.IND.PRS.3SG A.INF=CL.ACC.M.3PL count.INF ‘and this way he started to count [the big coins]’ In short, if we analyse, for example, the evolution of periphrases containing the aspectual a începe ‘start’, we observe an evolution from a periphrasis which has two (strong) phasal CPs (cf. 50a) to a construction with one strong and one ‘weak’ phasal 25 Glossing over a series of formal differences, this analysis is similar to that of Italian infinitivals headed by the complementizer di ‘of’, which can lexicalize C-, T-, and v-related heads (cf. Ledgeway 2012b:470f., 2015:158). 26 Note, however, that the idea of analyzing the infinitival marker A as an inflectional head of the sentential core is not new (for an overview, cf. Pană Dindelegan 2013:212–4; cf. also Jordan 2009). When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 19 CP (cf. 52), and finally a construction which has only one phasal CP (cf. 51), as informally sketched below, in (53). (53) [CP [TP T[finite] … [CP a [TP T […] [CP [TP T[finite] … [CP a [TP T […/finite] matrix clause embedded clause ⇒ matrix clause embedded clause (‘strong’ phase) (‘strong’ phase) (‘strong’ phase) (‘weak’ phase) [CP [TP T[finite] … [ a [ T [finite] ⇒ one phase However, it must be noted that old Moldovan was at the beginning of this ‘rephasing’ process, inasmuch as the vast majority of complex predicates have the clitics in the embedded infinitival clauses, with only a few cases of multiply spelt-out clitics and clitic-climbing attested (cf. Table 4 below). Table 4. Old Moldovan complex predicates containing infinitives functional verbs functional verbs functional verbs without multiple spelt with multiple spelt with clitic-climbing27 out clitics out clitics FD.1592–604 38 1 0 CLM.1700–50 157 2 1628 ULM.~1725 123 2 8 (929) Bert.1774 37 1 2 4.2.2. Present-day Moldovan 4.2.2.1. Aspectual verbs: two phasal CPs > one phasal CP Moving now to present-day Moldovan, it must be noted that, at least in central and northern regions (where old Moldovan was spoken),30 complex predicates containing aspectual verbs such as a termina ‘finish’ (cf. 54a) and a începe ‘start’ (cf. 54b) followed by A-infinitives are extremely common.31 27 Notably, however, aspectual complex predicates such as a începe ‘start’ and a căuta ‘try’ are attested with clitic-climbing over the infinitival marker A (for the importance of this observation, cf. §4; §5). 28 Out of 16 complex predicates, 15 were with the verb a căuta ‘try’. 29 One structure which displays clitic-climbing, given below as (i), presents a bare infinitive. (i) Old Moldovan (ULM.~1725:M.f.68r) lucrul său îl știia purta business.DEF his CL.ACC.M.3SG know.IND.IMPF.3SG wear.INF ‘he knew how to manage his businesses’ Again, the majority of examples (7 out of 8) are with a căuta ‘try’. 30 In the south of the Republic of Moldova the Wallachian dialect was/is spoken. Our discussion of clitic-climbing in this and following sections refers only to central and northern regions of the Republic. 31 Though it must be mentioned, along the lines of Rizzi (1978:117f. n.6), that there is considerable variation concerning the acceptability of clitic-climbing (especially with respect to ‘conative’ verbs such as TRY). 20 Isogloss 2023, 9(2)/2 Ștefania Costea (54) a. Moldovan, Ungheni Nu mă mai termin a uita not CL.ACC.1SG (any)more finish.IND.PRS.1SG A.INF watch.INF la film at movie ‘I won’t finish watching this movie.’ b. Moldovan, Chișinău Nu l-am început a citi not CL.ACC.M.3SG=AUX.PERF.1SG start.PPLE A.INF read.INF încă yet ‘I haven’t started to read it yet’ The fact that pronominal clitics are free to raise before the matrix verb clearly proves that these complex predicates instantiate a monoclausal structure (cf. Rizzi 1978:118f.). Indeed, this conclusion is further supported by the fact that Long Object Preposing can take place (cf. Rizzi 1976:20, 1978:113; Cinque 2004:16). Importantly, agreement between the embedded object, raised to the subject position of the matrix verb, is always compulsory in Moldovan (cf. 55). (55) a. Moldovan, Ungheni Voturile se termină a număra votes.DEF SE.PASS finish.IND.PRS.3PL A.INF count.INF rapid quickly ‘The votes will be done counting in no time’ b. Moldovan, Chișinău Bârfele prin sat s-au bad.words.DEF through village SE.PASS=AUX.PERF.3PL început a răspândi start.PPLE A.INF spread.INF ‘The bad words have started to spread through the village’ Likewise, focalized and topicalized constituents are generally banned from occurring in between the functional and the lexical verb, which clearly proves that the matrix verb does not have a lower left periphery (à la Belletti 2004) and that the lexical verb does not display a higher left periphery. Nevertheless, it must be noted that some, though not all, of my informants marginally allowed for ‘weak’, i.e. not [+aboutness], topics (topPs à la Dyakonova 2009:131) and focalized constituents marked as [+informational] (cf. 56) to intervene between the two verbs. However, rather than contradicting the assumptions regarding the monoclausal character of the constructions, this tendency can easily be explained through the influence of Russian, whereby ‘weak’ topics and a series of focalized constituents can freely scramble within the I-domain (Dyakonova 2009:123–5).32 32 Another test which proves that A does not head a (phasal) CP domain comes from the fact that, when clitics are raised before the matrix verb, A-infinitives cannot be questioned like CPs (cf., for example, the case of a începe ‘start’ in (i) below). When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 21 (56) Moldovan, Ungheni ?Nu l-am reușit [FocP în casă] a not CL.ACC.M.3SG=AUX.PERF.1SG succeed.PPLE in house A.INF despacheta pe ista mic undress.INF DOM this little.one ‘I didn’t manage to undress my baby while we were in the house’ Thus, it seems that present-day Moldovan continues the process of ‘rephasing’ which was incipient in old Moldovan (cf. §4.2.1.), displaying now a preference for constructions with clitic-climbing (cf. Table 5). Indeed, this would translate as a monoclausal (or monophasal) structure of complex predicates (cf. 57), whereby, given the lack of phasal barriers, the [finite] feature is free to percolate down the embedded infinitive, forcing the clitics to surface before the matrix verb. Table 5. Changes in the structure of Moldovan predicates containing infinitives functional verbs functional verbs functional verbs without multiple with multiple spelt with clitic-climbing spelt out clitics out clitics old Moldovan default option rare option very rare option present-day Moldovan rarer (?) option context-specific option default option (57) [CP [TP T[finite] … [ a [ T [finite] one phase 4.2.2.2. Modal verbs: two phasal CPs When a modal verb such as a putea ‘can’ (58a) or a trebui ‘must’ (58b) precedes an A-infinitive, clitic-climbing is completely ungrammatical in Moldovan. Indeed, other tests for monoclausality such as Long Object Preposing are also correctly predicted to fail (cf. 59). (58) a. Moldovan *Mă pot a pregăti singură CL.ACC.1SG can.IND.PRS.1SG A.INF prepare.INF by.myself ‘I can prepare by myself’ b. Moldovan *Îl trebuie a plăti CL.ACC.M.3SG must.IND.PRS.3SG A.INF pay.INF ‘One needs to pay him’ (59) Moldovan *Problemele se pot a rezolva ușor problems.DEF SE.REFL can.IND.PRS.3PL A.INF solve.INF easily ‘The problems can be solved easily’ (i) Moldovan *– Ce te-ai început? –A enerva. what CL.ACC.2SG=AUX.PERF.2SG start.PPLE A.INF get.angry.INF ‘What did you start? To get angry.’ (lit.) 22 Isogloss 2023, 9(2)/2 Ștefania Costea In order to understand this rather surprising behaviour, it is useful to briefly present the syntax of Moldovan modals. Discussing modal predicates such as a putea ‘can’, Nicolae (2018, 2019:104) notes that they can only take phasal complements in (old) Daco-Romanian. Indeed, this assumption can be easily verified in both old Moldovan, where a putea ‘can’ combines with either CPs (cf. 60a) or vPs (cf. 60b) (cf. also the ‘weak’ phasal CP in (52) above), and modern Moldovan, where it shows a very similar pattern (cf. 61). (60) a. Old Moldovan (CLM.1700–50:231r) n-au putut nice silihtariul not=AUX.PERF.3SG can.PPLE not.even henchman.DEF [CP a-i trimite alte isprăvi] A.INF=CL.DAT.3SG send.INF other news ‘not even the henchman was able to send him any news’ b. Old Moldovan (AOD.1675–6:8v) nu l-am putut [vPafla] not CL.ACC.M.3SG=AUX.PERF.1PL can.PPLE find.INF ‘we couldn’t find him’ (61) Moldovan îl pot [vP citi] CL.ACC.M.3SG can.IND.PRS.1SG read.INF ‘I can read it’ Now, if these combinatory requirements are put together with what was noted above (cf. §4.2.2.1.) about Moldovan A-infinitivals, i.e. that the infinitival marker a lexicalizes an I-related position, it follows naturally that the derivation crashes for Moldovan speakers when clitic-climbing occurs in structures where modals are followed by A-infinitives (cf. 58): the modal needs to combine with a phasal complement, and A-infinitives do not instantiate phases. 4.3. Complex predicates with subjunctive: two phasal CPs (> one phasal CP) 4.3.1. Old Moldovan The (old) Daco-Romanian subjunctive (irrealis) marker SĂ has previously been analysed as lexicalizing the Fin head (Ledgeway 2009:17; Nicolae 2015:95ff.; cf. also Hill & Alboiu 2016:244). Indeed, the fact that SĂ heads a phasal domain (cf. 62) is further confirmed by old Moldovan examples such as that in (63), whereby clitic- climbing out of subjunctive complements is blocked.33 As known, phasal domains are not accessible to operations outside the phase, albeit with the exception of their heads and their edges in accordance with the Phase Impenetrability Condition (PIC; Chomsky 2000:108, 2001:13f.). 33 But cf. Alboiu (2007) and Hill & Alboiu (2016:232ff.) for a different analysis, whereby SĂ does not always head a phase. The explanation put forward in Alboiu (2007) for the lack of clitic-climbing out of embedded subjunctives relates to the clitic nature of the subjunctive particle SĂ, i.e. the subjunctive marker and pronominal clitics form a clitic cluster and, as is known, excorporation is banned from clitic clusters. When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 23 (62) Daco-Romanian [CP [FinPsă [nu [ClPîl [IP citesc]]]] SĂ.SBJ not CL.ACC.M.3SG read.SBJ.1SG ‘that I wouldn’t read it’ (63) Old Moldovan (FD.1582–604:511v) nu pot [CP să-l scoață not can.IND.PRS.3PL SĂ.SBJ=CL.ACC.M.3SG get.out.SBJ.3PL den moarte] from death ‘they can’t bring him back to life’ Similarly to the case of the infinitival marker A (cf. §4.2.1. above), the marker SĂ also headed ‘weakened’ phasal CPs in old Moldovan, as proven by examples where pronominal clitics are given multiple realization, i.e. once on the functional head and once on the embedded subjunctive (cf. 64). In this kind of structure, the fact that clitics are attached to the embedded subjunctive proves that SĂ (still) heads a phase, namely a ‘barrier’ that generally (but not always) inhibits the spreading of the [finite] feature down the embedded clause, allowing, in turn, for clitics to adjoin to the embedded subjunctive. On the other hand, this phase is starting to weaken, inasmuch as it marginally allows for the feature [finite] to percolate down the embedded subjunctive, thus allowing pronominal clitics to (also) attach to the matrix verb (cf. Rizzi 1976, 1978; Tortora 2014:122ff.; Ledgeway 2015:n.6). (64) a. Old Moldovan (ULM.~1725:A-2.f.35r) de nu-i putiia să-i that not=CL.ACC.M.3PL can.IND.IMPF.3PL SĂ.SBJ=CL.ACC.M.3PL strângă gather.SBJ.3PL ‘such that they couldn’t gather them’ b. Old Moldovan (Bert.1774:14r) Cine-m va vre să-m fie who=CL.DAT.1SG AUX.FUT.3SG want.INF SĂ.SBJ=CL.DAT.1SG be.SBJ.3SG mie prietin me.DAT friend ‘who wants to befriend me’ In the case of periphrases containing the modal predicate a putea ‘can’, for example, the complements can be either (strong) phasal CPs, as in example (63) above, where the [finite] feature cannot percolate at all down the embedded subjunctive, or ‘weak’ phasal CPs (cf. 64a), whereby the [finite] feature can marginally percolate down the embedded subjunctive, forcing the pronominal clitic to be spelt-out twice, once on the matrix verb and once on the embedded subjunctive (cf. also the informal representation in (65) below). 24 Isogloss 2023, 9(2)/2 Ștefania Costea (65) [CP [TP T[finite] … [CP să [TP T […] [CP [TP T[finite] … [CP să [TP T […/finite] matrix clause embedded clause ⇒ matrix clause embedded clause (‘strong’ phase) (‘strong’ phase) (‘strong’ phase) (‘weak’ phase) Nevertheless, it is worth mentioning that ‘weak’ subjunctive phases were quite rare in old Moldovan (cf. Table 6), the subjunctive marker SĂ generally heading strong phasal CPs, which do not allow for pronominal clitics to vacate the embedded domain. A second important observation is that, at least in my old Moldovan corpus, there are no instances of ‘rephasing’ (cf. 66), namely examples in which the functional verb and the embedded subjunctive instantiate a single CP with clitics climbing to the matrix predicate. Table 6. Old Moldovan complex predicates containing subjunctives functional verbs functional verbs functional verbs without multiple spelt with multiple spelt with clitic-climbing out clitics out clitic FD.1592–604 132 1 0 CLM.1700–50 87 1 0 ULM.~1725 102 2 0 Bert.1774 116 1 0 (66) [CP [TP T[finite] … [să [ T [finite] one phase 4.3.2. Present-day Moldovan In present-day Moldovan, pronominal clitics are often spelt-out multiply in complex predicates containing subjunctives. While this phenomenon can be associated mainly with aspectuals such as a termina ‘finish’ (67a) and a începe ‘start’ (67b), it can also occur, although more rarely, with modals such as a putea ‘can’ (67c) and a vrea ‘want’ (67d).34 (67) a. Moldovan, Ungheni Nu l-am terminat să-l not CL.ACC.M.3SG=AUX.PERF.1SG finish.PPLE SĂ.SBJ=CL.ACC.M.3SG cos la lucru sew.SBJ.1SG at work ‘I haven’t finished sewing it at work’ b. Moldovan, Ungheni L-am început să-l CL.ACC.M.3SG=AUX.PERF.1SG start.PPLE SĂ.SBJ=CL.ACC.M.3SG cos la lucru sew.SBJ.1SG at work ‘I’ve started sewing it at work’ 34 Notably, when complex predicates contain a trebui ‘must’, the clitics occur only once, with the embedded subjunctive. When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 25 c. Moldovan, Bălți ?Îl pot să-l CL.ACC.M.3SG can.IND.PRS.1SG SĂ.SBJ=CL.ACC.M.3SG termin la timp finish.SBJ.1SG at time ‘I can finish it in time’ d. Moldovan, Bălți ?Îl vreau să-l CL.ACC.M.3SG want.IND.PRS.1SG SĂ.SBJ=CL.ACC.M.3SG fac now do.SBJ.1SG acum ‘I want to do it now’ Nevertheless, with respect to tests for monoclausality, aspectual and modal verbs do not pattern alike, inasmuch as the former allow Long Object Preposing (cf. 68a,b), while the latter do not (cf. 68c,d). However, it must be noted that in utterances resembling (68a,b) SE-passive needs to be spelt out twice (on the embedded subjunctive and on the matrix predicate). (68) a. Moldovan, Ungheni Voturile s-au terminat să votes.DEF SE.PASS=AUX.PERF.3PL finish.PPLE SĂ.SBJ se numere SE.PASS count.SBJ.3PL ‘The votes are done being counted’ b. Moldovan, Ungheni Voturile s-au început să votes.DEF SE.PASS=AUX.PERF.3PL start.PPLE SĂ.SBJ se numere SE.PASS count.SBJ.3PL ‘The votes have started to be counted’ c. Moldovan, Bălți *Faptele bune se pot să se deeds.DEF good SE.PASS can.IND.PRS.3PL SĂ.SBJ SE.PASS facă oricând make.SBJ.3PL always ‘Good deeds can be made at any time’ d. Moldovan, Bălți *Schimbările nu se vor să changes.DEF not SE.PASS want.IND.PRS.3PL SĂ.SBJ se facă SE.PASS make.SBJ.3PL ‘They don’t want to make any change’ Similarly to the case of infinitives (cf. §4.2.2.1), focalized and topicalized constituents are not usually to be found between the aspectual predicate and the embedded subjunctive, which seems to point towards the monoclausal nature of the construction. However, ‘weak’ topics (topPs à la Dyakonova 2009:131) and focalized 26 Isogloss 2023, 9(2)/2 Ștefania Costea constituents (cf. 69) can potentially intervene between both verbs, a tendency which can be explained through the influence of Russian, where such constituents can scramble within the I-domain (Dyakonova 2009:123–5). Alternatively, the presence of topics and foci can be accounted for by assuming that the embedded subjunctive is, in fact, a (weak) phasal CP, which can still display its own higher left periphery. (69) Moldovan, Ungheni ?L-am început [FocP cu copilul] CL.ACC.M.3SG=AUX.PERF.1SG start.PPLE with child.DEF să-l citesc SĂ.SBJ=CL.ACC.M.3SG read.SBJ.1SG ‘I’ve started reading it with my child’ While the second option, whereby SĂ-subjunctives instantiate CPs, seems to be confirmed by the fact that the SE-passive needs to be spelt-out twice when Long Object Preposing takes place (cf. 68a,b), the first option, where they do not instantiate CPs, is supported by the fact that the embedded subjunctive cannot be questioned when the clitics occur twice (cf. 70).35 This syntactic ambiguity is mirrored by the usage, where, for some speakers, the aspectual predicate and the lexical verb form a single phasal CP, as proven by the fact that pronominal clitics can optionally be spelt out only on the matrix verb (cf. 71),36 while, for others, this construction is regarded as ungrammatical. (70) Moldovan *– Ce îl începi? what CL.ACC.M.3SG start.IND.PRS.2SG – Să-l fac cât mai repede. SĂ.SBJ=CL.ACC.M.3SG make.SBJ.1SG how more quickly ‘What do you start? / To make it as quickly as I can.’ (lit.) (71) Moldovan, Chișinău Raportul de vânzări îl încep report.DEF of sales CL.ACC.M.3SG start.IND.PRS.1SG să fac mâine dimineață SĂ.SBJ do.SBJ.1SG tomorrow morning ‘I’ll start doing the sales’ report tomorrow morning’ 35 When the clitics are not raised before the matrix verb, the embedded subjunctive can be questioned (cf. i). (i) Moldovan – Ce începi? what start.IND.PRS.2SG – Să-l fac cât mai repede. SĂ.SBJ=CL.ACC.M.3SG make.SBJ.1SG how more quickly ‘What do you start? / To make it as quickly as I can.’ (lit.) 36 It is important to mention that the speakers who accepted utterances resembling (71) did not make any (semantic/pragmatic) difference between constructions which display only the higher copy of the clitic and constructions with multiply spelt-out clitics. When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 27 Consequently, in the case of structures where pronominal clitics are spelt out exclusively on the matrix predicate, we can put forward an analysis of SĂ which resembles the one proposed for the infinitival marker A above (cf. §4.2.1.). In particular, SĂ is able to also instantiate (various portions of) the IP in Moldovan. Indeed, this would translate as a monoclausal (or monophasal) structure of complex predicates (cf. 72), where, given the lack of phasal barriers, the [finite] feature is free to percolate down the embedded subjunctive and forces clitics to surface on the matrix verb. (72) [CP [TP T[finite] … [ să [ T [finite] one phase However, unlike cases of infinitives, old Moldovan complex predicates containing subjunctives do not show any clear instances of ‘rephasing’. As can be seen in Table 7 below, the preferred option was to only pronounce the clitic with the embedded subjunctive, with only a few cases of multiply spelt-out clitics. It is thus only present-day Moldovan that has started to show a tendency towards ‘rephasing’ in such contexts. Table 7. Changes in the structure of Moldovan predicates containing subjunctives functional verbs functional verbs functional verbs without multiple with multiple spelt with clitic-climbing spelt out clitics out clitics old Moldovan default option rare option – present-day Moldovan default option frequent option rare(r) option The trigger of this ‘rephasing’ process, which did not happen elsewhere in Daco-Romance, can be related to the (almost) complete equivalence between subjunctives and infinitives in Moldovan. Hence, under the influence of infinitives, which enter into monoclausal (monophasal) constructions with aspectual predicates (cf. §4.2.2.1.), subjunctives started to replicate this same pattern. In other words, they now tend to form one (phasal) CP with matrix aspectual verbs, thus allowing the clitic to be spelt out only on the functional verb, but display a different behaviour if preceded by modals, when clitic climbing to the matrix verb is generally blocked since such functional predicates take CP-complements (cf. Table 8; cf. also §4.2.2.2.).37 37 Only one Moldovan speaker allowed for structures resembling (i) and (ii). (i) Moldovan îl vreau să fac CL.ACC.M.3SG want.IND.PRS.1SG SĂ.SBJ do.SBJ.1SG ‘I want to do it’ (ii) Moldovan îl pot să fac CL.ACC.M.3SG can.IND.PRS.1SG SĂ.SBJ do.SBJ.1SG ‘I can do it’ 28 Isogloss 2023, 9(2)/2 Ștefania Costea Table 8. Clitic-climbing in Moldovan complex predicates containing subjunctives îl *trebuie */?pot */?vreau încerc reușesc încep termin să fac CL.ACC.M. must can want try succeed start finish SĂ.SBJ 3SG IND.3SG IND. IND.1SG IND. IND.1SG IND.1SG IND.1SG do.SBJ. 1SG 1SG 1SG ‘I must/can/want/try/manage/start/finish doing it.’ 5. When Daco-Romance meets Slavic: the case of Moldovan It was previously mentioned in §4.1. that Russian played an important role in preserving infinitives after certain functional predicates in Moldovan. Nevertheless, the influence of Russian goes beyond this superficial level, inasmuch as it has also boosted the frequency of monoclausal configurations involving aspectual predicates. While this type of construction was extremely rare in old Moldovan (cf. 73a), the presence of Russian, where aspectual verbs always enter into a monoclausal structure with their complements (cf. 73b), has considerably intensified it, thus rendering it today the norm (cf. 73c). Consequently, in such cases the change has multiple causation (cf. Thomason 2010:32), i.e. the pattern existed in old Moldovan prior to the contact with Russian (internal cause), but it is the contact with Russian which has generalized this already existing pattern (external cause) (Siemund 2008:9; cf. also Heine & Kuteva 2005:40; Backus et al. 2011:740–8). (73) a. Old Moldovan (ULM.~1725:A-2.f.18v) [CP s-au început a da] (rare pattern) SE.PASS=AUX.PERF.3SG start.PPLE A.INF give.INF b. Russian [CP načali davat’sja] (frequent pattern) start.PAST.M.PL give.INF=REFL c. Moldovan [CP s-au început a da] (frequent pattern) SE.PASS=AUX.PERF.3SG start.PPLE A.INF give.INF ‘they started to give’ On this point, it must be mentioned that, although old Moldovan displayed both infinitives and subjunctives, the fact that the infinitive was selected in such contexts can also be structurally explained (cf. also Matras 2011:151), and does not represent a pure syntactic borrowing from Russian (as can also be seen by the fact that it was preserved with its Daco-Romanian marker A). In order to generalize the monoclausal configuration for the structures containing and aspectual verb, the embedded verb needed to instantiate (a portion of) an IP, not a (phasal) CP. As has been shown, the subjunctive marker SĂ always headed a CP in old Moldovan, while the infinitival marker A lexicalized both I- and C-related positions. Thus, only A-infinitives represented a suitable candidate. With respect to (deontic) modal verbs, it may be the case that Russian and (old) Moldovan overlapped in their syntactic features, inasmuch as they do not instantiate monoclausal configurations in any of the varieties mentioned (cf. 74). Thus, in this situation, the presence of Russian simply further reinforced the already generalized syntactic option in (old) Moldovan. When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 29 (74) a. Old Moldovan (Bert.1774:29r) n-ați putut [CP să țineți] (frequent pattern) not=AUX.PERF.2PL can.PPLE SĂ.SBJ keep.SBJ.2PL b. Russian vy ne mogli [CP uderžat’] (frequent pattern) you.PL not can.PAST.PL keep.INF c. Moldovan n-ați putut [CP să țineți] (frequent pattern) not=AUX.PERF.2PL can.PPLE SĂ.SBJ keep.SBJ.2PL ‘you couldn’t keep’ 6. Conclusion There is no doubt that contact with Russian, where aspectuals and their complements instantiate a monoclausal structure, helped Moldovan to preserve and generalize the monoclausal configuration of complex predicates containing aspectuals and infinitives (cf. modern (Daco-)Romanian, where these structures are extinct). As a consequence, the Moldovan infinitival marker A was generalized as lexicalizing an I-related position, a change which, in turn, made it possible for pronominal clitics to climb to the matrix predicate. A further, and more recent, development concerns the evolution of subjunctives embedded under aspectuals which, under the influence of infinitives, began to allow clitic-climbing, thus signalling a monoclausal configuration. This last change makes Moldovan unique within the Daco-Romance context, inasmuch as all the other varieties block clitic-climbing over the subjunctive marker, generally interpreted as instantiating a phase head. With respect to modal predicates, however, the influence of Russian is much more subtle. Both (old) Moldovan and Russian (deontic) modal predicates select for a CP-sized complement, thus entering into bi- clausal structures. Therefore, given that the infinitival marker A became generalized with an I-related position, A-infinitives ceased to be suitable candidates for modal complements. By contrast, SĂ-subjunctives were generalized as complements to modals. Nevertheless, in this case clitic-climbing is blocked since SĂ is a phase head (which cannot change its position via analogy with the infinitival marker A) and clitics cannot skip over phasal boundaries. Acknowledgments I would like to thank two anonymous reviewers for their extremely helpful comments and corrections. Any outstanding errors are the responsibility of the author. Sources AOD.1675–6 Alexie, omul lui Dumnezeu. Maria Stanciu Istrate (ed.), Cele mai vechi cărți populare din literatura română 5, Bucharest: 30 Isogloss 2023, 9(2)/2 Ștefania Costea Fundația Națională pentru Știință și Artă, 2001, 89–101. (Moldova) Bert.1774 Bertoldo. Magdalena Georgescu (ed.), Cele mai vechi cărți populare în literatura română 3, Bucharest: Minerva, 1999, 157–239. (Moldova) CC2.1581 Coresi, Evanghelie cu învățătură. Sextil Pușcariu, & Alexandru Procopovici (eds.), Diaconul Coresi, Carte cu învățătură (Vol. I. Textul), Bucharest: Socec, 1914. (Brașov) CLM.1700–50 Miron Costin, Letopisețul Țărâi Moldovei. Petre P. Panaitescu (ed.), M. Costin, Opere, Bucharest: Editura de Stat pentru Literatură și Artă, 1958, 41–201. (Moldova) DÎ Documente și însemnări românești din secolul al XVI-lea, Gheorghe Chivu (ed.), Bucharest: Editura Academiei Române, 1979. FD.1592–604 Floarea darurilor. Alexandra Roman Moraru (ed.), Cele mai vechi cărți populare în literatura română 1, Bucharest: Minerva, 1996, 119–82. (Moldova) NL.~1750–66 Ion Neculce, Letopisețul. Iorgu Iordan (ed.), Letopisețul Țării Moldovei și O samă de cuvinte, Bucharest: Editura de Stat pentru Literatură și Artă (Second Edition), 1959, 31–388. (Moldova and Wallachia) SVI.~1670 Varlaam și Ioasaf. Maria Stanciu Istrate (ed.), Reflexe ale medievalității europene în cultura română veche: Varlaam și Ioasaf în cea mai veche versiune a traducerii lui Udriște Năsturel, Bucharest: Editura Muzeului Național al Literaturii Române, 2013, 82–325. (Wallachia) ULM.~1725 Grigore Ureche, Letopisețul Țării Moldovei. Petre P. Panaitescu (ed.), Bucharest: Editura de Stat pentru Literatură și Artă, 1955, 57–210. (Wallachia, original from Moldova) References Adams, James Noel. 1994. Wackernagel’s Law and the Position of Unstressed Pronouns in Classical Latin. Transactions of the Philological Society 92: 103–78. https://doi.org/10.1111/j.1467-968x.1994.tb00430.x Adams, James Noel. 2013. Social Variation and the Latin Language. Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO9780511843433 Alboiu, Gabriela. 2007. Moving Forward with Romanian Backward Control and Raising. In William D. Davies, & Stanley Dubinsky (eds.), New Horizons in the Analysis of Control and Raising. Studies in Natural Language and Linguistic Theory When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 31 (Vol. 71), 187–211. Dordrecht: Springer. https://doi.org/10.1007/978-1-4020-6176- 9_8 Atanasov, Petar. 1976. Infinitivul meglenoromân. Studii și Cercetări Lingvistice 4: 137–50. Backus, Ad, Seza Dogruöz, & Bernd Heine. 2011. Salient stages in contact-induced grammatical change: Evidence from synchronic vs diachronic contact situations. Language Sciences 33: 738–52. https://doi.org/10.1016/j.langsci.2011.02.004 Bailyn, John Frederick. 2010. To what degree are Croatian and Serbian the same language? Evidence from a translation study. Journal of Slavic Linguistics 18: 181– 219. https://doi.org/10.1353/jsl.2010.0000 Belletti, Adriana. 2004. Aspects of the Low IP Area. In Luigi Rizzi (ed.), The Structure of CP and IP: The Cartography of Syntactic Structures (Volume 2), 16–51. Oxford: Oxford University Press. Burukina, Irina. 2019. Mandative verbs: between obligatory control and overt embedded subjects. Ms. Čamdžić, Amela, & Richard Hudson. 2002. Serbo-Croat-Bosnian clitics and Word Grammar. UCL Working Papers in Linguistics 14: 321–53. https://doi.org/10.2478/v10015-007-0001-7 Capidan, Theodor. 1925. Meglenoromânii. Istoria și graiul lor (I). Bucharest: Cultura Națională. Capidan, Theodor. 1932. Aromânii. Dialectul aromân. Studiu lingvistic. Bucharest: Monitorul Oficial și Imprimeriile Statului (Imprimeria Națională). Capidan, Theodor. 1934. Dicționar meglenoromân (III). Bucharest: Monitorul Oficial și Imprimeriile Statului (Imprimeria Națională). Cardinaletti, Anna, & Ur Shlonsky. 2004. Clitic Positions and Restructuring in Italian. Linguistic Inquiry 35: 519–57. https://doi.org/10.1162/0024389042350523 Chomsky, Noam. 2000. Minimalist inquiries: The framework. In Roger Martin, David Michaels, Juan Uriagereka, & Samuel Jay Keyser (eds.), Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik, 89–155. Cambridge (Mass.): MIT Press. Chomsky, Noam. 2001. Derivation by Phase. In Michael Kenstowicz (ed.), Ken Hale. A Life in Language, 1–52. Cambridge (Mass.): MIT Press. https://doi.org/10.7551/mitpress/4056.003.0004 32 Isogloss 2023, 9(2)/2 Ștefania Costea Chomsky, Noam. 2008. On Phases. In Robert Freidin, Carlos P. Oterno, & Maria Luisa Zubizarreta (eds.), Foundational Issues in Linguistics Theory, 133–66. Cambridge (Mass.): MIT Press. https://doi.org/10.7551/mitpress/9780262062787.001.0001 Cinque, Guglielmo. 2004. Restructuring and functional structure. In Adriana Belletti (ed.), Structures and beyond: The cartography of syntactic structures, 132–91. Oxford: Oxford University Press. Cinque, Guglielmo. 2006. Restructuring and Functional Heads. Oxford: Oxford University Press. Corbeanu, Ramona Cătălina. 2020. Conjunctivul în istroromână. Statutul lui „neca”. Paper presented at Tradition and Innovation in the Study and Preservation of Istro- Romanian (Workshop), University of Bucharest, 21/11/2020. Costea, Ștefania. 2021. Clausal de-supines in Moldovan: The State-of-the-art. Revue Roumaine de Linguistique, 199–210. Cruschina, Silvio, & Adam Ledgeway. 2016. The structure of the clause. In Adam Ledgeway, & Martin Maiden (eds.), The Oxford Guide to the Romance Languages, 556–74. Oxford: Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199677108.003.0031 De Angelis, Alessandro. 2017. Microvariazione, diacronia e interferenza: due case- studies dall’Italia meridionale estrema. Archivio glottologico italiano 102: 40–69. Dragomirescu, Adina. 2013. Complex predicates. In Gabriela Pană Dindelegan (ed.), The Grammar of Romanian, 191–201. Oxford: Oxford University Press. Dragomirescu, Adina, & Alexandru Nicolae. 2020. Particular features of Istro- Romanian pronominal clitics. Studia UBB Philologia 65: 147–58. https://doi.org/10.24193/subbphilo.2020.4.09 Dragomirescu, Adina, & Alexandru Nicolae. 2016. Originea formelor verbale «nonfinite» cu flexiune: infinitiv vs supin. Diacronia 4: 1–16. https://doi.org/10.17684/i4a54ro Dyakonova, Marina. 2009. A phase-based approach to Russian free word order. Ph.D. thesis, University of Amsterdam: Utrecht. Farcaș, Mircea. 2006. Infinitivul în subdialectul maramureșean. In Nicolae Saramandu, Maria Marin, & Iulia Mărgărit (eds.), Lucrările celui de al XII-lea simpozion național de dialectologie, 195–202. Cluj-Napoca: Mega. Heine, Bernd, & Tania Kuteva. 2005. Language Contact and Grammatical Change. Cambridge: Cambridge University Press. https://doi.org/10.1017/cbo9780511614132 When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 33 Hill, Virginia, & Gabriela Alboiu. 2016. Verb Movement and Clause Structure in Old Romanian. Oxford: Oxford University Press. https://doi.org/10.1093/acprof:oso/9780198736509.001.0001 Jones, Michael. 1993. Sardinian Syntax. London: Routledge. https://doi.org/10.4324/9780203993804 Jordan, Maria. 2009. Loss of Infinitival Complementation in Romanian Diachronic Syntax. Ph.D. thesis, University of Florida. Katičić, Radoslav. 1986. Sintaksa hrvatskoga književnog jezika: Načrt za gramatiku. Zagreb: HAZU. Kayne, Richard. 2000. Parameters and Universals. New York: Oxford University Press. Kovačević, Peda, & Tania Milićev. 2018. The nature(s) of syntactic variation: Evidence from the Serbian/Croatian dialect continuum. In Denisa Lenertová, Roland Meyer, Radek Šimík, & Luka Szucsich (eds.), Advances in formal Slavic linguistics 2016, 147–67. Berlin: Language Science Press. https://doi.org/10.1057/9781137390608.0010 Krapova, Iliyana, & Guglielmo Cinque. 2018. Universal Constraints on Balkanisms. A Case Study: The absence of Clitic Climbing. In Iliyana Krapova, & Brian Joseph (eds.), Balkan Syntax and (Universal) Principles of Grammar, 151–91. Berlin/Boston: De Gruyter Mouton. https://doi.org/10.1515/9783110375930-008 Krapova, Iliyana, & Tsvetana Dimitrova. 2016. The genitive-dative syncretism in the history of Bulgarian: towards an analysis. Studi Slavistici 12: 181–208. https://doi.org/10.13128/Studi-Slavis-17976 Krejci, Bonnie, Vera Gribanova, & Boris Harizanov. 2016. Agree-dependent A- movement and low copy pronunciation in Russian. Paper presented at Formal Approaches to Slavic Linguistics 25, Cornell University, 14/05/2016. Ledgeway, Adam. 2000. A Comparative Syntax of the Dialects of Southern Italy: A Minimalist Approach. Oxford: Blackwell. Ledgeway, Adam. 2009. Aspetti della sintassi della periferia sinistra del cosentino. In Diego Pescarini (ed.), Studi sui dialetti della Calabria, 3–24. Padua: Unipress. Ledgeway, Adam. 2012a. From Latin to Romance. Morphosyntactic Typology and Change. Oxford: Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199584376.001.0001 Ledgeway, Adam. 2012b. Contatto e mutamento: complementazione e complementatori nei dialetti del Salento. Studi Italiani di Linguistica Teoretica e Applicata 41: 459–80. 34 Isogloss 2023, 9(2)/2 Ștefania Costea Ledgeway, Adam. 2015. Reconstructing complementizer-drop in the dialects of the Salento: A syntactic or phonological phenomenon?. In Theresa Biberauer and George Walkden (eds.), Syntax over Time: Lexical, Morphological, and Information- Structural Interactions, 147–62. Oxford: Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199687923.003.0009 Ledgeway, Adam. 2016. Minunile limbii române: lecții pentru romaniști, Prima Conferință Internațională a Centrului de Studii Românești, University of Bucharest, 29/09/16–1/10/16. Ledgeway, Adam. 2017. Late Latin Verb Second: The Sentential Word Order of the Itinerarium Egeriae. In Jaume Fontanals, & Renato Oniga (eds.), Generative Approaches to Latin Syntax, special issue of Catalan Journal of Linguistics 16: 163– 216. https://doi.org/10.5565/rev/catjl.212 Ledgeway, Adam, & Alessandra Lombardi. 2005. Verb Movement, Adverbs and Clitic Positions in Romance. Probus 17: 79–113. https://doi.org/10.1515/prbs.2005.17.1.79 Marin, Maria, Iulia Mărgărit, & Victorela Neagoe. 2000 [1998]. Graiuri românești din Ucraina și Republica Moldova. In Maria Marin, Iulia Mărgărit, Victorela Neagoe, & Vasile Pavel (eds.), Cercetări asupra graiurilor românești de peste hotare, 42–119. Bucharest: Bucureștii Noi. Matras, Yaron. 2011. Explaining convergence and the formation of linguistic areas. In Osamu Hieda, Christa König, & Hiroshi Nakagawa (eds.), Geographical typology and linguistic areas, 143–60. Amsterdam: Benjamins. https://doi.org/10.1075/tufs.2.11mat Migdalski, Krzysztof. 2015. On the loss of tense and verb-adjacent clitics in Slavic. In Theresa Biberauer, & George Walkden (eds.), Syntax over Time: Lexical, Morphological, and Information-Structural Interactions, 180–95. Oxford: Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199687923.003.0011 Mišeska Tomić, Olga. 2006. Balkan Sprachbund Morpho-syntactic Features (Studies in Natural Language and Linguistic Theory 67). Dordrecht: Springer. https://doi.org/10.1007/1-4020-4488-7 Nicolae, Alexandru & Dana Niculescu. 2016. Pronominal clitics: clitic ordering, clitic clusters. In Gabriela Pană Dindelegan (ed.), The Syntax of Old Romanian, 52–70. Oxford: Oxford University Press. https://doi.org/10.1093/acprof:oso/9780198712350.003.0002 Nicolae, Alexandru. 2015. Ordinea constituenților în limba română: o perspectivă diacronică. Bucharest: Editura Universității din București. When Moldovan meets Russian: intralinguistic variation in clitic climbing Isogloss 2023, 9(2)/2 35 Nicolae, Alexandru. 2018. Romanian modal verb configuration and phases. Paper presented at ICUB Lunchtime Seminar, University of Bucharest, 22/03/2018. Nicolae, Alexandru. 2019. Word Order and Parameter Change in Romanian. A Comparative Romance Perspective. Oxford: Oxford University Press. https://doi.org/10.1093/oso/9780198807360.001.0001 Pană Dindelegan, Gabriela. 2013. The infinitive. In Gabriela Pană Dindelegan (ed.), The Grammar of Romanian, 211–22. Oxford: Oxford University Press. Pancheva, Roumyana. 2005. The Rise and fall of second position clitics. Natural Language and Linguistic Theory 23: 103–67. https://doi.org/10.1007/s11049-003- 2510-y Parry, Mair. 2005. Parluma ’d Còiri. Sociolinguistica e grammatica del dialetto di Cairo Montenotte. Savona: Liguria. Pescarini, Diego. 2020. La microvariation syntaxique dans les langues romanes. Un modèle paramétrique. Habilitation thesis, University of Côte d’Azur. Progovac, Ljiljana. 1993. Locality and Subjunctive-like Complements in Serbo- Croatian. Journal of Slavic Linguistics 1: 116–44. Radanović-Kocić, Vesna. 1988. The grammar of Serbo-Croatian clitics: a synchronic and diachronic perspective. Ph.D. thesis, University of Illinois, Urbana–Champaign. Reinheimer, Sanda, & Liliane Tasmowski. 2005. Pratique des langues romanes. II. Les pronoms personnels. Paris: L’Hartamann. Rizzi, Luigi. 1976. Ristrutturazione. Rivista di Grammatica Generativa 1: 1–54. Rizzi, Luigi. 1978. A restructuring rule in Italian syntax. In Samuel Jay Keyser (ed.), Recent transformational studies in European languages, 113–58. Cambridge (Mass.): MIT Press. Rizzi, Luigi. 1982. Issues in Italian Syntax. Dordrecht: Foris. https://doi.org/10.1515/9783110883718.1 Roberts, Ian. 2010. Agreement and Head Movement. Cambridge (Mass.): MIT Press. https://doi.org/10.7551/mitpress/9780262014304.003.0002 Salvi, Giampaolo. 2004. La formazione della struttura di frase romaza. Ordine delle parole e clitici dal latino alle lingue romanze antiche. Tübingen: Niemeyer. https://doi.org/10.1515/9783110945508 Saramandu, Nicolae. 1984. Aromâna. In Valeriu Rusu (ed.), Tratatul de dialectologie românească, 423–76. Craiova: Scrisul Românesc. 36 Isogloss 2023, 9(2)/2 Ștefania Costea Sârbu, Richard, & Vasile Frățilă. 1998. Dialectul istroromân. Texte și glosar. Timișoara: Amarcord. Schifano, Norma. 2018. Verb Movement in Romance. A Comparative Study. Oxford: Oxford University Press. https://doi.org/10.1093/oso/9780198804642.001.0001 Schoorlemmer, Maaike. 1994. Aspect and verbal complementation in Russian. In Sergey Avrutin, Steven Franks, & Ljiljana Progovac (eds.), Formal approaches to Slavic linguistics: The MIT meeting, 400–22. Ann Arbor (MI): Michigan Slavic Publications. Siemund, Peter. 2008. Language contact: Constraints and common paths of contact- induced language change. In Peter Siemund, & Noemi Kintana (eds.), Language Contact and Contact Languages, 3–11. Amsterdam: Benjamins. https://doi.org/10.1075/hsm.7.01sie Stepanov, Arthur. 2007. On the Absence of Long-Distance A-Movement in Russian. Journal of Slavic Linguistics 15: 81–108. Thomason, Sarah. 2010. Contact Explanations in Linguistics. In Raymond Hickey (ed.), The Handbook of Language Contact, 31–47. Malden (Mass.): Wiley-Blackwell. https://doi.org/10.1002/9781444318159.ch1 Tortora, Christina. 2014. A comparative grammar of Borgomanerese. Oxford: Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199945627.001.0001 Turano, Giuseppina. 2017. Modal particles in Albanian subjunctive, infinitive and supine constructions: presence vs absence of clitic climbing. Quaderni di Linguistica e Studi Orientali 3: 61–86. https://doi.org/10.13128/QULSO-2421-7220-21339 Vaillant, André. 1977. Grammaire comparée des langues slaves (Tome V. La syntaxe). Paris: Klincksieck. Večerka, Radoslav. 1989. Altkirchenslavische (Altbulgarische) Syntax. Freiburg i. Br.: U.W. Weiher. Vulpe, Magdalena. 1963. Repartiția geografică a construcțiilor cu infinitivul și cu conjunctivul în limba română. Fonetică și Dialectologie 5: 123–56.