Complex Predicates

Cambridge Handbook of Romance Linguistics Chapter 21 Complex predicates Adina Dragomirescu, Alexandru Nicolae, and Gabriela Pană Dindelegan 21.1 Outline and scope This chapter first discusses the various meanings associated with the concept of ‘complex predicate’ (hereafter CPred) and sets up a battery of reliable syntactic diagnostics for the identification of a CPred, which will be invoked throughout the analysis of the Romance empirical data. We then address the shift from Latin syntheticity to the greater analyticity of the Romance languages, one of the most striking developments of which in the verbal domain is the emergence and subsequent profusion of a whole series of periphrastic verb constructions, which often add to and complement the older synthetic structures inherited from Latin to offer explicit marking of categories which were previously not distinctly marked (e.g., FECI ‘I did’ > synthetic Ro. făcui ‘I did’ alongside analytic am făcut ‘I did’; SCRIBO ‘I write/am writing’ > synthetic It. scrivo ‘I write/am writing’ alongside sto scrivendo ‘I am writing’). The transition from the morphologically-oriented structures of Latin to the increasingly syntacticallyoriented structures of Romance (Ledgeway 2012:11) involves the grammaticalization of verbs such as HAVE, BE, WANT, OWE, COME, GO, MAKE, HOLD, STAND, FOLLOW, and the transfer of many inflexional categories from the lexical verb to the auxiliary / first component of the monoclausal CPred. Taking stock of the set of syntactic diagnostics used to classify a given structure as a CPred, we turn to the analysis of the Romance facts, which are extremely well documented both in diachrony and in synchrony, and which can be used as a series of fruitful case studies to throw light on the diachronic and synchronic relationship between inflexion and periphrasis from a wider cross-linguistic perspective. 21.2 Delimitations and diagnostics 21.2.1 What is a complex predicate? A large set of constructions with distinct properties has been considered to constitute CPreds, this notion often being applied in a vague and underspecified manner. In a very broad sense, any predicate structure that ‘consists of more than one piece is complex’ (Svenonius 2008:47). Under this broad understanding, even auxiliary-verb constructions have been included in the class of CPreds (Müller 2006; Abeillé and Godard 2002; 2003; ‘verbal complex’ in Monachesi 2005), a fact which, in a certain respect, captures the intuition that Romance auxiliaries represent a heterogeneous set of elements, which share fewer morphosyntactic properties than do English or other Germanic auxiliaries (Green 1987:257; Ledgeway 2012:119). A coarser definition restricts the notion of CPred to constructions based on restructuring (Rizzi 1978) – defined as an operation by which ‘the scope of operations associated with a lower predicate [cliticization, auxiliary selection] is extended to the domain of a higher predicate’ in Cardinaletti and Shlonsky (2004:524) – and other related constructions such as: periphrastic causatives, verb-particle constructions, resultatives, consider + predicate combinations (in Amberber, Baker, and Harvey 2010), along with serial verb constructions and light verb constructions (in Svenonius 2008), and constructions with motion verbs, copula verbs, and perception verbs (in Abeillé and Godard 2002:404; 2003:125-27). The common intuition, to which we return below, is that CPreds are monoclausal, this property variously being obtained either via clause union (Aissen and Perlmutter 1976) or restructuring (Rizzi 1978; 1982), i.e., as an operation by which an underlying biclausal Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics structure becomes a simple sentence (cf. also Pesetsky 2019 in this respect), or via direct selection of a complement of a smaller size than a CP (Wurmbrand 2001). Other authors stress the fact that, monoclausality notwithstanding, there are subtler differences between different types of CPreds. In Baker and Harvey (2010), there are two types of CPreds, distinguished on the basis of the relation established between the component units: CPreds based on merger (the units share conceptual structure), the result being a predicate structure ‘whose range classes with the range of predicate structures found in monomorphemic predicates’ (Baker and Harvey 2010:13; cf. also Sheehan 2016:981 on Romance causatives) and CPreds based on coindexation, an operation which extends the conceptual structure of the predicate (e.g., perception verbs and faire-infinitive verbs, which do not obey the condition of the unique argument structure, Labelle 2017). This chapter adopts a broad notion of CPreds and focuses on monoclausal periphrases whose component units are verbal in nature. Examples of particular interest include auxiliaryverb constructions (passive periphrases, resultative perfects, periphrastic futures and conditionals) and monoclausal constructions with aspectual, modal, causative, and perception verbs, typically complemented by a non-finite form. 21.2.2 Diagnosing monoclausality From a formal perspective, the monoclausal nature of the CPred ensures that its components share one single extended projection, a property which derives the syntactic diagnostics variously proposed in the literature. The complex nature of these formations, in conjunction with monoclausality, explains why in some aspects CPreds pattern with prototypical words, but in others, with prototypical phrases (Amberber, Baker, and Harvey 2010:3); however, from a narrow syntactic perspective, only one of the component verbs (the bearer of TAM1 (and phi-feature) information) serves as a syntactic head of the CPred (Svenonius 2008:55). The following linguistic phenomena have been generally used to diagnose monoclausality: (i) negation expressed exclusively on the higher head and disallowed in the embedded domain, illustrated with an Ibero-Romance aspectual periphrasis (Ledgeway 2012:127): 1 o seu país non está (*non) buscando the his country not is not seeking unha bomba nuclear (Glc.) a bomb nuclear ‘his country is not trying to build a nuclear bomb’ construir build.INF (ii) clitic climbing, illustrated with a Fr. faire-infinitive construction (Abeillé, Godard, and Sag 1998:2): 2 Paul le fera lire Paul 3MSG.ACC= make.FUT.3SG read.INF de terminale (Fr.) of sixth.form ‘Paul will make the sixth-form students read it.’ aux élèves to.the students 1 With certain periphrastic formations, TAM marking is shared between the two components; for example, in active have/be + participle constructions, aspect is formally marked on the participle; however, the auxiliary, which is the head of the formation from a narrow syntactic perspective, bears most of the grammatical information, marking, among other things, mood, tense, and phi-feature values. Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics (iii) se/si-passive adjoined to the higher head (Burzio 1986), illustrated with a Romanian ‘semi-auxiliary’ modal configuration (Guțu Romalo 2005): 3 Romanele nu se pot novels.DEF not PASS= can.IND.PRS.3PL peste noapte. (Ro.) over night ‘Novels cannot be written overnight.’ scrie write.INF (iv) in certain languages the choice of perfective auxiliary of the lower verb is determined by the argument structure of the higher verb; e.g., in Italian, both in a prototypical restructuring configuration (4a), and in a compound passive (4b), the selection of auxiliary BE is determined by the unaccusative nature of andare ‘go’ and stata ‘been’ (Frank 1996): 4 a b Mario ci sarebbe proprio voluto andare. Mario it be.COND.3SG really want.PTCP go.INF ‘Mario would have really wanted to go there.’ Maria è stata accusata. Maria be.AUX.PASS be.PTCP.F.SG accuse.PTCP.F.SG ‘Maria has been accused.’ (It.) (It.) Other phenomena taken to be sensitive to CPred formation (e.g., past participle agreement, reflexivization) are tied to more particular language-specific rules and cannot be taken as general tests of CPred formation (this is also the case of auxiliary selection, discussed above). It is also important to highlight that there are numerous structures which, despite showing a certain degree of cohesion, do not make up CPreds, but rather admit a biclausal analysis: Romanian causatives, modals, and perception verbs followed by the subjunctive, certain Spanish verbal periphrases, and future periphrases of the Fr. aller-type or Ro. o + subjunctive are all constructions in which clitic climbing, one of the characteristic signatures of monoclausality (but cf. Cinque 2004), does not apply (Abeillé and Godard 2003:125-27). The question of which combination permits or requires clitic climbing is also subject to variation in Romance, with particular language-specific and construction-specific options (e.g., clitic climbing occurs in French with causatives and perception verbs, but not with other classes of verbs) (Legendre 2007:294). Furthermore, Cinque (2004) stresses the fact that a given verb may be associated with two distinct syntactic configurations in the very same language: witness (5) below, where It. volere ‘want’ may occur both in a restructuring (5a) and a nonrestructuring (5b) configuration, as testified by the presence (5a) or absence (5b) of clitic climbing. 5 a b b' Lo volevo [vedere subito] 3MSG.ACC= want.IPF.1SG see.INF immediately ‘(I) him wanted to see immediately.’ Maria vorrebbe già averlo Maria want.COND.3SG already have.INF=3MSG.ACC già lasciato already left. ‘Mary would already want to have already left him.’ **Maria lo vorrebbe già Maria 3MSG.ACC= want.COND.3SG already già lasciato (It.) (It.) aver have.INF (It.) Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics already left. When restructuring/CPred formation does not apply, the properties of the lower verb do not extend to the higher verb; e.g., only when clitic climbing applies does the higher verb reflect the auxiliary selection properties of the lower verb (compare 6a and 6b) (Burzio 1986; Frank 1996). There are, of course, notable and important exceptions, e.g., andare/venire ‘go’/‘come’ always take auxiliary ‘be’, while finire/cominciare ‘finish’/‘start’ always take the auxiliary ‘have’ irrespective of clitic climbing (Adam Ledgeway, p.c.). 6 a b I ragazzi si sarebbero voluti vedere the children REFL= be.COND.3PL want.PTCP.PL see.INF più spesso. (It.) more often I ragazzi avrebbero voluto vedersi the children have.COND.3PL want.PTCP see.INF=REFL più spesso. (It.) more often ‘The children would have wanted to see each other more often.’ Last but not least, while auxiliary-verb constructions and other monoclausal constructions have in common two of the core features which identify a CPred (clitic climbing and negation on the first verb), they exhibit major differences with respect to other important features; the review in Ledgeway (2012:119-50) is summarized in Table 21.1: auxiliary-verb constructions morphophonological reduction adjacency (exc. Fr) clitic forms no VP-ellipsis no stress other monoclausal constructions – no obligatory adjacency the same form as the lexical verb VP ellipsis stress (in answers) Table 21.1 Auxiliary-verb constructions vs other monoclausal constructions 21.3. Auxiliaries 21.3.1 Introduction In auxiliary-verb constructions, also labelled ‘verbal complexes’ (Ramat 1987; Monachesi 2005), the auxiliary is the bearer of TAM information, and the argument structure of the entire complex is that of the lexical verb. In contrast to Latin, where only the perfect passive and (semi-)deponent auxiliary ESSE ‘be’ is systematically used (see Ledgeway 2012:34fn9 on the apparent auxiliary usage of other verbs, e.g., UELLE ‘want’, POSSE ‘can’, DEBERE ‘must’), the Romance languages are characterized by a profusion of analytic, auxiliary-based constructions which replace or often add to and complement the existing synthetic structures (Green 1987:263; Ledgeway 2012:11,33), an empirical development reflecting the emergence of, (a), dedicated structural position(s) hosting auxiliaries to the left of the VP (Ledgeway 2012:33; 2017). A significant number of originally lexical verbs develop into a wide range of auxiliaries, following the parameters generally underlying such processes of grammaticalization: phonetic attrition, morphological specialization, morphosyntactic decategorialization, and semantic bleaching. The emergence of auxiliaries also triggers a redistribution in the marking of grammatical values (Vincent 1987; Danckaert 2016:132), i.e., Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics large-scale transfer of many inflexional categories from the lexical verb to the auxiliary (Ledgeway 2012:119). 21.3.2 Auxiliary-verb constructions based on HABERE ‘have’ The Latin verb HABERE underwent multiple grammaticalization paths, the most productive and best studied of which is the compound past, a pan-Romance development. The emergence of the HABERE periphrases (the compound past and the doubly compound structures, the future, and the conditional) occurred at different moments in the Latin-to-Romance transition, or even within the history of the Romance languages themselves. The literature documents two main points of view with respect to the emergence of the HABERE compound past: some authors (Thielmann 1885; Benveniste 1962; Harris 1982; Salvi 1987; Pinkster 1987) consider that it developed in late Latin, while others (Ledgeway 2012; Adams 2013; Roberts 2013; Haverling 2016; Legendre 2017) argue that it is a Romancespecific development. The latter represents the more plausible scenario; Adams (2013:646) stresses the fact that Latin periphrases which on the surface look like perfects are open to interpretations in which HABERE has full lexical value, hence there is a long period of ambiguity before grammaticalization took place. Also relevant in this respect is the fact that HABERE did not have a fixed position but could either precede or follow the participle and complements cold also intervene between the auxiliary and the partciple in early Romance – e.g., old French (Buridant 2000:375-77), early Italian (Rohlfs 1969:330), and even sixteenth and seventeenth century Romanian (Nicolae 2019). The emergence of the Romance compound past has been traced back to the reanalysis of an originally resultative aspectual periphrasis (7a) as a present perfective periphrasis (7b) (details below) (Ledgeway 2012:130; Haverling 2016:200). 7 a b [VP [THEME [LOC in ea prouincia] pecunias magnas in that province.ABL money.ACC.FPL big.ACC.FPL [AP collocatas]] habent] (Lat., Cic. Leg. Man. 18; in Ledgeway 2012:130) placed.ACC.FPL have.IND.PRS.3PL ‘they have large sums invested in that province’ [IP [VP [LOC in ea prouincia] [THEME pecunias in that province.ABL money.ACC.FPL magnas] collocatas] habent] big.ACC.FPL placed.ACC.FPL have.IND.PRS.3PL ‘they have invested large sums in that province’ The Romance developments do not immediately replace the Latin synthetic perfect; and still today in many parts of the Romània the synthetic perfect is very strong, unlike its compound competitor, e.g., large parts of Spanish and Portuguese-speaking Latin America, European Portuguese, southern Calabria, and Sicily; rather, the periphrasis acquired new values, ranging from present resultative (8) and iterative (9) and canonical (10) present perfect to punctual perfective (11), all variously preserved in different diachronic and diatopic Romance varieties (Harris 1982; Salvi 1987; Squartini and Bertinetto 2000; Adams 2013; examples compiled by Ledgeway 2012). 8 non m’ ha chiamatu (SCal., in Alfonzetti 1998) not me=has.AUX.PST.3SG call.PTCP ‘I don’t know what’s happened to him’ [because he hasn’t rung me] 9 aquí también ha hecho frô (Pal., in Leal Cruz 2003:132) Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics here too it.has.AUX.PST do.PTCP ‘here too it continues to be cold’ cold 10 siempre la he escuchado con atención, always 3FSG.ACC= have.AUX.PST.1SG listen.to.PTCP with attention pero nunca más (EuSp., in Penny 2000:159) but never more ‘I have always listened to her attentively, but never again.’ 11 la France a the France has.AUX.PST.3SG en 1939 (spoken Fr.) in 1939 ‘France declared war in 1939.’ déclaré la guerre declare.PTCP the war There are two main factors which favoured the grammaticalization of HABERE as a perfective auxiliary: semantic bleaching of HABERE, from expressing an action with a durative value (initially a near synonym of TENERE ‘keep’) to expressing a relation, i.e., possession (Salvi 1987:229); subject coreference between the locative subject of HABERE and the agentive/experiencer subject of the participle, causing the subject of transitive and unergative participles to be reanalysed as the subject of the entire structure (Ledgeway 2012:131f.) (unaccusatives and passives occur with ESSE ‘be’, see §§21.3.3, 5). With HABERE losing the ability to host its own thematic subject, the argument structure of the participle is extended to the entire construction (the Heir-Apparent Principle of Harris and Campbell 1995); the participle becomes the lexical head of the extended projection of the simplified structure, and thematically empty HABERE becomes a placeholder for the mood and tense values of the entire construction. From a strictly syntactic point of view, one of the crucial steps in the development of the compound past is structural simplification (Roberts 2013), i.e., the removal of the thematic subject of HABERE. Furthermore, in late Latin and in Romance there emerges an entirely new system for the future and the conditional, based on Latin periphrases involving the infinitive and the present indicative of HABERE (CANTARE HABEO ‘sing.INF have.PRS.1SG’, which gave rise to the Romance inflexional future Fr. chanterai, It. canterò, Sp. cantaré ‘I will sing’) and the infinitive and the imperfect of HABERE (CANTARE HABEBAM ‘sing.INF have.IPFV.1SG’, which is the origin of the Romance conditional: Fr. chanterais, It. canteria, Sp. cantaría) and, more rarely, the perfect of HABERE (CANTARE HABUI ‘sing.INF have.PFV.1SG’ yielding the Tuscan/modern Italian conditional canterei ‘I would sing’) (Vincent 1987:245f.; Pinkster 1987:25f.). The grammaticalization process involving futures and conditionals is radically different from that leading to compound past forms in two respects: first it emerged much earlier2 from structures where HABERE is placed after the lexical verb (as it reflects a headfinal ordering, this was taken as evidence for the claim that future and conditional auxiliaries grammaticalized earlier than compound past auxiliaries, see Adams 1991, Ledgeway 2012:33fn7)3, and, secondly, the final result is a novel synthetic form (in contrast to the compound past, which remained an analytic form all over Romance). The ‘new’ synthetic Cf. also the controversial example involving DARAS ‘you will give’ (juxtaposed to NON DABO ‘I will not give’) in seventh century Fredegarius, identified by Krusch (1888:85) (see Alkire and Rosen 2010:165). 3 Cf. early examples such as (i) given in Adams (1991:148-54), ambiguous between possibility and futurity: 2 (i) si enim sustuleris istam tertiam, remanere habent duae (Pompeius 129.26) ‘For if you take away the third [last syllable], two [syllables] will have to remain’ Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics future emerged in all the Romance varieties, except Sardinian, Romanian, and Dalmatian (Ledgeway 2012:134f.). Moreover, periphrastic future forms, which already existed in Latin from the classical period (Pinkster 1987:211) are attested in all the Romance languages. Sometimes, they are based on the auxiliary ‘have’, as in Abruzzese (ajja cantà, hî da cantà, a da cantà, avem a cantà, avet a cantà, anno da cantà – Ledgeway 2012:135) and in Romanian (am să cânt, ai să cânți, are să cânte, avem să cântăm, aveți să cântați, au să cânte ‘have.IND.PRS.1SG-3PL COMP.SBJV sing.SBJV.1SG-3PL’). However, these periphrastic formations do not show the same degree of morphosyntactic cohesion as analytic formations of the compound past type; for example, the Romanian ‘have’-future behaves incongruously with respect to the monoclausality diagnostics in that the clausal negator surfaces on the higher verb, ‘have’ (12a), while clitic climbing is impossible with the clitic surfacing in the domain of the lower verb (12b); furthermore, periphrastic formations based on the imperfect of ‘have’ with a future in the past interpretation are also possible (13) (Zafiu 2013:40f.). 12 13 nu am să (*nu) cânt (Ro.) not have.IND.PRS.1SG COMP.SBJV not sing.SBJV.1SG ‘I’m not going to sing’ b (*l-)am să-l cânt (Ro.) it=have.IND.PRS.1SG COMP.SBJV=3MSG.ACCsing.SBJV.1SG ‘I’m going to sing it’ aveam să cânt (Ro.) have.IND.IPFV.1SG COMP.SBJV sing.SBJV.1SG ‘I was going to sing’ a 21.3.3 Auxiliary-verb constructions based on ESSE ‘be’ Although the grammaticalization of ESSE had been under way since Latin (when it was used as a perfective passive and as a periphrastic future auxiliary, Green 1987:259f.), the paths taken by ESSE in Romance are less numerous than those involving HABERE (for the Romance analytic passive, see §21.4). On the one hand, verbs semantically incompatible with HABERE, i.e., unaccusatives, were absorbed into the ESSE perfective periphrasis for (semi-)deponents and passives (Aranovich 2009:21), with which they have in common their co-occurrence of an Undergoer subject (Ledgeway 2012:133), a construction such as LAPSUS SUM (lit. slipped I.am, ‘I have slipped’) being constructed on the model of the passive AMATUS SUM (lit. loved I.am, ‘I have been loved’) (Burton 2016:165). These facts gave way to the well-known phenomenon of auxiliary selection in Romance varieties such as Italian (Burzio 1981; Centineo 1986; Van Valin 1987), French (Sorace 2000, 2004), old Spanish (Lamiroy 1999; Mackenzie 2006; Stolova 2006), old Romanian (Dragomirescu and Nicolae 2013); this includes mixed systems which appear to illustrate the gradual phasing out of the protoRomance active/inactive alignment and the return to the nominative/accusative alignment (Loporcaro 2007:173-85). On the other hand, in a language such as Romanian, the auxiliary ‘be’ grammaticalized as an irrealis marker (Avram and Hill 2007), which occurs in the structure of the perfect subjunctive (să fi citit COMP.SBJV be.IRREALIS read.PTCP ‘should have read’), the perfect conditional (aș fi citit AUX.COND.3SG/PL be.IRREALIS read.PTCP ‘I would have read’), the future perfect (voi fi citit AUX.FUT.1SG be.IRREALIS read.PTCP ‘I will have read’), the perfect infinitive (înainte de a fi vorbit before of to.INF be.IRREALIS talk.PTCP ‘before I/you etc. would have talked’), and the presumptive (voi fi citind AUX.FUT.1SG be.IRREALIS read.GER ‘I would be reading’). 21.3.4 Auxiliary-verb constructions based on other verbs Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics Except for the analytic forms with HABERE (§21.3.2), other future periphrases are based on proto-Romance *voˈlere ‘want’ (Ro. voi cânta, vei cânta, va cânta, vom cânta, veți cânta, vor cânta; Friulian, southern Italian dialects), UENIRE (AD) ‘come to’ (Srs. vegnel a cantar, vegns a cantar, vegn a cantar, vegnin a cantar, vegnis a cantar, vegnen a cantar), DEBERE ‘owe; must’ (in Sardinian), IRE/AMBULARE/VADERE ‘go’ (Sp. voy a cantar, vas a cantar, va a cantar, vamos a cantar, vais a cantar, van a cantar; and in French, Occitan, and (Brazilian) Portuguese) (Ledgeway 2012:122-24,134f.). Other verbs enter periphrases expressing different temporal and aspectual values (Green 1987:259f.; Ledgeway 2012:122-24, 134f.; 2017:847): UENIRE ‘come’ expresses different values, from iterative aspect (14a) to past tense (14b), while TENERE ‘hold, keep’ expresses iterative aspect (14c) or the present perfect (14d); STARE ‘stand’ and SEDERE ‘sit’, along with ESSE may occur in passive, progressive, and present perfect periphrases (see §21.4.5). 14 a b c d l’ ai tornat 3MSG=have.AUX.PST.1SG returned a veire (Lgd., in Ledgeway 2012:122) to see.INF ‘I saw him again’ vaig anar al mercat go.AUX.PST.1SG go.INF to.the market ahir (Cat., in Ledgeway 2012:123) yesterday ‘I went to the market yesterday’ lo tenh de velhat (Occ., in Ledgeway 2012:123) 3MSGACC=hold.AUX.PRS.3SG of watched ‘she keeps watching him’ el ga invecià he have.AUX.PST.3SG aged tanto (Ven., in Ledgeway 2012:123) a.lot ‘he has aged considerably’ Many of the periphrastic structures occurring in Romance (e.g., the *voˈlere, TENERE, and STARE periphrases) do not have forerunners in Latin texts (Pinkster 1987:195, 211). 21.3.5 TAM make-up of auxiliaries Cross-Romance comparative considerations indicate that the division of labour in the marking of TAM categories in the analytic cluster varies across the Romance languages, and that the degree of the morphological richness of a given auxiliary differs from one Romance variety to another (Giacalone Ramat 2000:125; Nicolae 2015:82-84; 2019:31f., cf. also Fleischman 1983:183). Consider, for example, the contrast between the analytic paradigms with the grammaticalized descendants of HABERE in (standard) French (15) and (standard) Romanian (16). 15 a b j’ai I.have.IND.PRS.1SG ‘I have eaten/I eat’ j’avais I.have.IND.IPF.1SG ‘I had eaten’ mangé eat.PTCP (Fr.) mangé eat.PTCP (Fr.) Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics c d e 16 j’aurai mangé (Fr.) I.have.IND.FUT.1SG eat.PTCP ‘I will have eaten’ (que) j’aie mangé (Fr.) that I.have.SBJV.PRS.1SG eat.PTCP ‘(that) I have eaten’ j’aurais mangé (Fr.) I.have.COND.1SG eat.PTCP ‘I would have eaten’ eu am I have.IND.PRS.1SG ‘I have eaten/I ate’ mâncat eat.PTCP (Ro.) The contrast in (15)-(16) reveals that Romanian HAVE is unable to undergo tense variation (Dobrovie-Sorin 1994; Avram and Hill 2007; Giurgea 2011; Nicolae 2015), but does however possess an unambiguous mood specification (indicative), while its French counterpart undergoes tense variation, mood and aspect variation. Thus, Romanian auxiliaries are moodoriented, while French auxiliaries are tense-oriented (also marking mood and aspect values), a microparametric distinction which also accounts for the distinct nature of the multipleauxiliary paradigms in these two languages: Romanian possesses multiple auxiliary structures in which auxiliaries encode mood-oriented information (17) (indicative/conditional and irrrealis), while in French doubly compound structures auxiliaries have tense-oriented values (18). 17 voi / aș fi will.AUX.IND.FUT.1SG have.AUX.COND.1SG be.IRREALIS≡INF ‘I will / would have sent’ 18 j'avais eu / I=have.AUX.IPFV.1SG have.PTCP eu envoyé have.PTCP send.PTCP ‘I had sent’ trimis (Ro.) send.PTCP j’ai I=have.AUX.IND.PRS.1SG (Fr.) 21.4 The periphrastic passive 21.4.1 Synthetic vs analytic The history of the passive represents another illustration of one of the most significant typological changes in the transition from Latin to Romance, namely the passage from Latin predominantly synthetic structures (AMATUR ‘(s)he is loved’) to Romance chiefly periphrastic structures (Fr. Il est aimé, Ro. El este iubit, It. Lui è amato, Sp. Él es amado ‘He is loved’) (Danckaert 2017:217). The Romance periphrases are descendants of Latin constructions used for perfective paradigms (Danckaert 2017:216; Ledgeway 2021a:§1). The synthetic passive, restricted in Latin to imperfective paradigms (Danckaert 2017:216; Ledgeway 2021a:§1), was fully replaced in Romance by analytic formations made up of a descendant of ESSE or another auxiliary + a passive past participle, this representing a ‘functional extension of an already existing periphrasis of the classical language’ (Ledgeway 2012:16). 21.4.2 Frequency and distribution Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics There is a common place in the Romance reference literature concerning ‘unpopularity’ of the passive, especially in informal registers. While it is true that in some non-standard varieties of Italy the canonical passive is inexistent or very rarely used (see Ledgeway 2021a:§§2.1, 2.2 and references), it is equally true that in some contexts and under certain pragmatic conditions, especially in the standard languages, the passive has a wide distribution and a significant frequency. Overall, the usage of the passive is much more nuanced. The considerable differences in usage depend on different factors: (i) type of register (formal vs informal register; written vs oral register – the formal and written registers resort to the passive much more often, see also Cennamo 2016:975); (ii) pragmatic conditions (the choice of the passive is tied to a given pragmatic context; for example, a particular way of formulating a question might influence the selection of the passive); (iii) semantic and syntactic-semantic type of verb (i.e., weak transitive verbs are rarely or never used in the passive; the ungrammaticality of the passive with stative, modal, measure verbs or unergative verbs with an internal object is well-documented); (iv) lexical restrictions of the Agent and of the Patient/Theme (the [+human] vs [-animate] feature bears upon the selection of the passive); for details on the frequency and distribution of the passive, see Ledgeway (2021a:§2.1, 2.2). 21.4.3 Participle agreement Irrespective of other variables (auxiliary selection, participle agreement in the compound past, overt realization of the Agent, etc.), there is a constant feature of analytic passives across Romance: the passive past participle undergoes number and/or gender agreement with the clausal subject (19a-d). Brazilian Portuguese is exceptional in this respect: as a general tendency of agreement weakening, for some speakers, the participle does not undergo number agreement with preverbal subjects, and gender and number agreement with postverbal subjects (Ledgeway 2021a:§7.1). 19 a b c d DeputațiiMPL sunt aleșiMPL de popor. Les députésMPL sont élusMPL par le peuple. I deputatiMPL sono elettiMPL dal popolo. Los diputadosMPL son elegidos MPL por el pueblo. ‘Deputies are elected by the people.’ (Ro.) (Fr.) (It.) (Sp.) 21.4.4 The reflexive passive Alongside the periphrastic passive, the Romance languages also make use of the reflexive passive; this construction is particularly productive in Romanian and Italian (D’Alessandro 2007; Maiden and Robustelli 2007:285f.; Adams 2013:711;), but also occurs in the other Romance varieties. Without being equivalent in all occurrences, the auxiliary-based passive and the reflexive passive are used in parallel, with distinct features depending on the syntactic construction and stylistic register. The reflexive passive is specialized for the third person, singular and plural, with non-animate passive subjects and constructions with an unexpressed Agent (20); as for the linguistic register, it is preferred in popular and colloquial varieties. 20 a b Cărțile se citesc. (Ro.) book.PL.DEF REFL.PASS.3PL read.PRS.3PL ‘Books are read.’ Si distrusse Dresda. (It.) REFL.PASS.3SG destroy.3SG Dresden ‘Dresden was destroyed’ (Maiden and Robustelli 2000:285) Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics 21.4.5 Inventory of passive (semi-)auxiliary verbs In most Romance varieties, the passive periphrasis is based on ESSE, irrespective of the auxiliary/auxiliaries employed for the perfect. Exceptionally, in southern Italian dialects spoken in the Pugliese province of Bari and in southeastern Lucania, HABERE is used as a passive auxiliary, in free variation with ESSERE and UENIRE (Loporcaro 1988; Cennamo 2016:975; Ledgeway 2021a:§3.4); this represents an extension of the free variation of ESSERE and HABERE from active constructions. A different situation is found with a southern Calabrian dialect from Polia, where, in the absence of this free variation with active forms, avire ‘have’ generalized, being used also in the passive (Marchese 2016, in Ledgeway 2021a:§3.4). Besides the exceptional selection of HABERE, the Romance languages show considerable variation in the choice of the passive auxiliary. In French, Italian and Romanian (Abeillé and Godard 2003), prototypical ESSERE occurs with a full paradigm and with identical forms in the passive (21a) and in the copulative (21b) usage; note that Romanian also has a third usage of be – i.e., a perfective/irrealis auxiliary usage (cf. Avram and Hill 2007) –, with be being invariable4 (21c). 21 a b c Profesorul trebuie să teacher.SG.DEF must.PRS.3SG COMP.SBJV fie plătit. be.SBJV.3SG pay.PTCP ‘The teacher must be paid’ Profesorul trebuie să teacher.SG.DEF must.PRS.3SG COMP.SBJV fie bătrân. be.SBJV.3SG old ‘The teacher must be old’ Profesorul trebuie să teacher.SG.DEF must.PRS.3SG COMP.SBJV fi plătit pentru serviciile cerute. be.INV pay.PTCP for services.DEF required ‘The teacher must have paid for the required services’ (Ro.) (Ro.) (MRo.) Alongside prototypical essere ‘be’ (22a), Italian also employs venire ‘come’ (22b) and andare ‘go’ (22c), the former for the dynamic passive, the latter for a deontic passive (Ledgeway 2016:226). Their occurrence in compound tenses (hence their combination with another auxiliary) is disallowed. Venire incorporates a dynamic value (i.e., in contrast to La porta è aperta ‘The door is open(ed)’, which is ambiguous between a stative and a dynamic reading, La porta viene aperta ‘The door gets opened’ is unambiguously dynamic) (Maiden and Robustelli 2007:284); the andare passive is characterized by more complex restrictions (see Maiden and Robustelli 2007:282f.). 22 a Il topo the mouse(MSG) è mangiato is eat.PTCP.MSG dal gatto. by cat (It.) 4 In old Romanian, perfective/irrealis be also occurred with variable forms, and had identical forms with copulative and passive be (Nicolae 2015:120, n5): (i) să fim be.SBJV.1PL ‘for us to have loved God’ COMP.SBJV noi we iubit love.PTCP pre DOM Dumnezeu God (ORo.) Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics b c ‘The mouse is eaten by the cat.’ Il libro viene letto dal ricercatore the book(MSG) comes read.PTCP.MSG by.the researcher ‘The book is read by the researcher.’ Queste medicine vanno prese these medicines.FPL andare.IND.PRS.3PL take.PTCP.F.PL ogni mattina. every morning ‘These medicines should be taken every morning.’ (It.) (It.) Catalan mainly uses the auxiliary ser ‘be’ (23a) (Alsina 2016:379); with restrictions similar to those found in Italian, a UENIRE auxiliary is also employed as a dynamic passive (23b), its choice being also dialectally constrained (Wheeler, Yates, and Dols 1999:512; Ledgeway 2021a:§3.3). 23 a b Han estat descrits els simptomes have.3PL be.PTCP described.PTCP.MPL the symptoms.MPL (per un doctorand). (Cat.) by a doctoral.student ‘The symptoms were described by a doctoral student.’ Lus premit venian distribuits. (Alg.) the prizes come.PST.IPFV.3PL distribute.PTCP.MPL ‘The prizes were being handed out.’ Spanish employs two BE verbs, ser (24a) and estar (24b), with different aspectual properties, estar incorporating a resultative aspectual value. Furthermore, the presence of the Agent is preferred with ser but disfavoured with estar. 24 a b Los terroristas fueron arrestados the terrorists.MPL be.PRT.3MPL arrest.PTCP.PL por la policía. (Sp.) by the police ‘The terrorists were arrested by the police.’ Los terroristas están arrestados. the terrorists.MPL be.IND.PRS.3PL arrest.PTCP.MPL ‘The terrorists are under arrest.’ Besides the canonical BE-passive, which belongs to the formal register (Jones 1993:124), Sardinian also features a WANT-passive similar to that found across the dialects of southern Italy (Ledgeway 2000; 2021b), and a modal passive with kérrere ‘want’ (25), used in all registers (Jones 1993:124; Mensching and Remberger 2016:286f) 25 Sa màchina keret acconzada the car.FSG want.IND.PRS.3SG repair.PTCP.FSG dae mechanic (Srd.) by mechanic ‘This car needs to be repaired by a mechanic.’ Romansh varieties feature an analytic passive based on the auxiliary neir ‘come’ (< (Anderson 2016:177): UENIRE) Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics 26 La proposta vign the proposal.FSG come.IND.PRS.3SG ʻThe proposal is accepted.ʼ acceptada. (Rms.) accept.PTCP.FSG Besides the grammaticalized BE construction in (27), in non-standard registers Romanian also employs veni ‘come’ (28) in constructions with a modal necessity value (Iordan 1950; Pană Dindelegan 2003:133-39; Dragomirescu and Nicolae 2014). Like Sardinian and southern dialects of Italy, Romanian features a construction based on a deontic modal, a trebui ‘must’ (29), used with a full paradigm (see also Cabredo Hofherr 2017:244). 27 Copiii sunt lăudați children.DEF.M be.IND.PRS.3PL praise.PTCP.MPL ‘The children are praised by their parents.’ 28 a b 29 de părinți. by parents (Ro.) Cratița în care se coace cozonacul vine pan.DEF.F in which CL.REFL.PASS bakes cake. DEF comes unsă cu unt. (Ro.) smear.PTCP.FSG with butter ‘The pan in which the pound cake is baked has to / must / ought to be / is smeared with butter.’ Celălalt bec vine slăbit. (Ro.) the.other bulb.MSG comes loose.PTCP.MSG ‘The other bulb must be loosened.’ Cartea / cărțile trebuia / vor book.F.DEF books.FPL.DEF must.IPFV will.AUX.FUT.3PL citită / citite de elevi read.PTCP.FSG read.PTCP.FPL by students ‘The book(s) were to be read / will have to be read by the students.’ trebui must.INF (Ro.) Among the sub-Danubian varieties, the veni ‘come’ and rămâne ‘remain’ passives occur occasionally in Istro-Romanian (30) (Sârbu and Frățilă 1998:66). Given the circulation of these passives in areas of contact with Venetan, it is presumed that Venetan might have played a role in the existence of these constructions, yet it is hard to decide whether language contact is the sole source for these constructions (Dragomirescu and Nicolae 2014:79). 30 a b ie vire ucis he comes kill.PTCP ‘he is killed’ ie ramas-a he remain=have.AUX.PST.3SG ‘he was killed’ ucis kill.PTCP In several Romance languages (Fr., Ptg., Sp., It., Cat., and Ro.), especially in the journalistic and literary styles, the verb ‘see’ accompanied by the reflexive pronoun is used in passive structures, followed by a participle or an infinitive (31) (Giacalone Ramat 2017:170-3; Ledgeway 2021a:§3.5). 31 a Il se voyait envahir / envahi Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics b he REFL see.IMPF.3SG overrun.INF overrun.PTCP par un sentiment d’échec. (Fr.) by a feeling of-failure ‘He was being / was taken over (lit. saw himself to invade / invaded) by a sense of failure’ El s-a văzut abandonat și he REFL=have.AUX.PST.3SG abandon.PTCP and uitat de Dumnezeu. (Ro.) forget.PTCP by God ‘He saw himself abandoned and forgotten by God’ 21.4.6 The double passive A special construction attested across non-formal varieties of modern Spanish, but deemed absent from the other Romance languages, is the ‘double passive’ (32a-b) (Bosque and Gallego 2011). The construction consists of a cluster made up of two verbal complexes: a passive periphrasis with the lexical verb and the auxiliary ser ‘be’ and a doubling periphrasis, made up of a tensed form of ser and an aspectual verb. Bosque and Gallego (2011) analyse this construction as a type of syntactic doubling, and stress the expletive nature of some of the components (the low auxiliary and the high, aspectual, participle). A similar phenomenon is found in colloquial French with the aspectual verbs finir ‘finish’ / commencer ‘begin’ (32c); here, the passive reading of the infinitival complement is formally marked on the aspectual verb (Ledgeway 2021a:§2.3). 32 a b c La ermita fue empezada a ser the hermitage be.IND.PRT.3SG start.PTCP.FSG to be construida en el siglo XIV (Sp.) build.PTCP.F.SG in the century 14 ‘The hermitage was started to be built in the fourteenth century’ El misil fue acabado de ser armado ayer. (Sp.) the missile be.IND.PRT.3SG finish.PTCP of be arm.PTCP yesterday ‘The missile was finished being armed yesterday’ La maison est finie de construire. (Fr.) the house be.PRS.3SG finish.PTCP.FSG of build.INF ‘The house has finished being built.’ 21.4.7 The position of constituents in the passive periphrasis The preferred word order in Latin passive ESSE-constructions places the auxiliary in postposition (participle > ESSE), but the reverse word order is also possible (ESSE > participle) (Ledgeway 2012:223). In a statistical analysis, Bauer (2006:294) convincingly demonstrates that the ESSE > participle word order gradually extends in the history of Latin. In accordance with the passage from a head-final to a head-initial grammar, the word order ESSE > participle continues to gradually surpass the reverse word order in the diachrony of the Romance languages (for the word order preferences of late Latin, see the percentages in Danckaert 2017:226). For example, the patterns with pre-auxiliary participles (33) were much more frequent in old Romanian than in modern Romanian, where the auxiliary + participle word order, productive since old Romanian (34), has generalized. 33 Deaci dzise do say.PST Adusu fu aducă Pavelu. COMP.SBJV bring.SBJV.PRS.3SG Paul el de ceia… (ORo.) se Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics bring.PTCP be.PST.3SG he by those ‘So he meant to bring Paul. He was brought by those…’ 34 nu sunt dumnedzei ceia ce-su not be.IND.PRS.PL gods those who=be.IND.PRS.3PL cu mârule fapți (ORo.) with hands.DEF make.PTCP.MPL ‘They are not gods who are made with the hands.’ On the basis of word order evidence, the recent literature documents an attempt towards reconsidering the emergence of the Romance analytic passive (e.g., It. sono amato ‘I am loved’). Thus, according to this hypothesis, based on the word order difference between the Latin pattern (AMATUS EST ‘he has been loved’) and its Romance counterparts, which feature the word order auxiliary > participle, the analytic passive does not descend from a Classical Latin structure, but rather represents a new formation, not older than the fourth century AD (Danckaert 2017). 21.4.8 Monoclausal properties In contrast to tense and mood auxiliaries, which present many syntactic and morphophonological signs of decategorialization (grammaticalization), the periphrastic passive formation has greater autonomy, ensuring, among other features, the possibility of employing multiple (semi)auxiliaries in the passive (this represents one of the reasons why the term ‘semi-auxiliary’ is more appropriate than ‘auxiliary’ with reference to the analytic passive). This also accounts for the fact that, in the absence of a broader linguistic or extralinguistic context, these periphrases are systematically ambiguous: constructions like (35a) may be read either as passives (35b), or as [copula + predicative] constructions (35c). 35 a b c Pâinea este coaptă (Ro.) bread.DEF.FSG is bake.PTCP.FSG ‘The bread is baked.’ Pâinea este coaptă cu grijă. (Ro.) bread.DEF.FSG is bake.PTCP.FSG with care ‘The bread is carefully baked.’ Pâinea este foarte coaptă, aproape arsă. (Ro.) bread.DEF.FSG is very bake.PTCP.FSG almost burn.PTCP.FSG ‘The bread is well baked, almost burned.’ Also, there are significant differences in the behaviour of the passive auxiliaries: the BEauxiliary behaves in all aspects like a copula verb, while auxiliaries based on motion verbs exhibit clear signs of grammaticalization, the most important of which is the loss of the motion semantics. Using for illustration material from Romanian, the diagnostics below synthesize the monoclausality vs autonomy properties for the analytic BE-passives. Diagnostics for monoclausality: (i) negation expressed exclusively on the auxiliary: 36 Nu mi-au not 1SG.DAT=have.AUX.PST.3PL ‘The books were not given to me.’ fost be.PTCP date cărțile. give.PTCP.FPL books.DEF.FPL Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics (ii) (dative) clitic climbing (accusative clitics are excluded by default in the passive): 37 Mi-au 1SG.DAT=have.AUX.PST.3PL ‘Five books were given to me.’ fost be.PTCP date cinci cărți. give.PTCP.FPL five books.F (iii) occurrence of the passive in restructured modal configurations (see §21.6): 38 a b Cartea poate fi citită de elev. book.DEF.FSG can be.INF read.PTCP.FSG by student ‘The book can be read by the student.’ Cartea îmi poate fi furată oricând. book.DEF.FSG 1SG.DAT= can be.INF steal.PTCP.FSG anytime ‘My book can be stolen from me at any time’ Diagnostics for autonomy: (i) semantically, passive BE behaves like any other copula; (ii) passive BE has a full paradigm (like existential BE): it shows no sign of morphophonological erosion and may co-occur with other (TAM-expressing) auxiliaries (39); 39 (Aș fi dorit) să have.AUX.COND.1SG be.INF want.PTCP COMP.SBJV ajutat de prieteni. (Ro.) help.PTCP by friends ‘I would have wanted to have been helped by friends.’ fi fost be.INF be.PTCP (iii) full constituents may be interposed between passive BE and the participle (40a), and the participle may be even topicalized (40b), giving rise to the word order participle > passive auxiliary; 40 a b A fost have.AUX.PST.3SG be.PTCP ‘She was praised today as well.’ Știut este că… know.PTCP is that ‘It is known that…’ și also astăzi lăudată. today praise.PTCP.FSG (iv) passive BE may be replaced by other auxiliaries (41a) or be elided (41b). 41 a b Cratița este / rămâne / trebuie / vine unsă pan.DEF.FSG is remains must be comes smear.PTCP.FSG cu unt. with butter ‘The pan is/remains/must be/comes smeared with butter.’ Obligată de părinți, a renunțat force.PTCP.FSG by patents have.AUX.PST.3SG give.up.PTCP la facultate. at faculty ‘Forced by her parents, she gave up university.’ Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics 21.5. Aspectual periphrases In Romance, aspectual values are often (but not always) syncretic with temporal values (Bertinetto and Squartini 2016:939; Maiden 2016:501f.) but there also exist numerous verbs which lexically encode aspect, and aspectual periphrases with verbs having different meanings, which show a CPred behaviour. These aspectual periphrases are made up of a verb (such as CONTINUE, BE, STAND, GO, HOLD, etc.) bearing the inflexional information and a lexical verb in the infinitive or the gerundive. Verbs lexically encoding aspect in Romance have a different behaviour. Similarly to other classes of verbs, in standard Italian, clitic climbing is not obligatory with aspectuals5 (Rizzi 1982:4), hence their CPred status is contextual (42); in Romanian negation and clitic climbing are obligatorily hosted by the aspectual verb when it is followed by a supine (43a) (Dragomirescu 2013:196f.), but when the second verb is in the infinitive (43b) or the subjunctive (43c-d) the construction receives a biclausal analysis, as negation may occur either on the higher verb (43c) or on the lower verb (43d) (with scope differences), and clitic climbing is disallowed (43b). 42 a b 43 a b c d Gianni continua a/ sta per raccontargli Gianni continue.PRS.3SG to stay.PRS.3SG for tell.INF=3DAT stupide storie. (It., in Rizzi 1982:4) stupid things Gianni gli continua a/sta per raccontare Gianni DAT.3= continue.PRS.3SG to stay.PRS.3SG for tell.INF stupide storie. (It., in Rizzi 1982:4) stupid things ‘Gianni is continuing / going to tell him stupid stories’ Cărțile nu le termină de books.DEF.F not 3FPL.ACC= finish.PRS.3SG of.SUP citit. (Ro.) read.SUP Cărțile (*le) nu termină books.DEF.F 3FPL= not finish.PRS.3SG a le citi. (Ro.) to.INF F.3PL= read.INF ‘He doesn’t finish reading the books.’ Nu continuă să vină not continue.PRS.3SG COMP.SBJV come.SBJV.3SG la ore. (Ro.) to hours ‘(S)he doesn’t keep on attending classes’ Continuă să nu vină continue.PRS.3SG COMP.SBJV not come.SBJV.3SG la ore. (Ro.) to hours ‘(S)he keeps on not attending classes’ 5 In actual usage, clitic climbing is always obligatory in the regional Italian of southern speakers, but usually absent in the speech of northern speakers, this distribution reflecting the options found in the dialects spoken in each area in turn (Adam Ledgeway, p.c.) Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics Aspectual (progressive) BE-periphrases are specific to Gallo-Romance, Italian, and Sardinian (44a) (Bertinetto and Squartini 2016:947-50; Andriani 2017:197). Romanian also possesses an imminential periphrasis made up of fi ‘be’ and the subjunctive (44b), which disallows clitic climbing (and allows for negation both on the higher, and on the lower predicate, with scope differences), therefore not qualifying as a CPred. Squartini (1998:27-29) indicates that the aspectual BE-periphrases with a gerund/present participle are also attested in Catalan (estar), French (être), Galician (estar), Portuguese (estar), and Spanish (estar). Imminential and proximative periphrases may be also found in Catalan, French, Italian, Portuguese, and Spanish (Bertinetto and Squartini 2016:951). 44 a b so travallande (Srd., Jones 1993:83) be.IND.PRS.1SG work.GER ‘(I) am working.’ era să te lovesc (Ro.) be.IPFV COMP.SBJV 2SG.ACC= hit.SBJV.1SG ‘I almost hit you.’ Aspectual (progressive) STAND-periphrases are specific to Italo-Romance and IberoRomance, where they represent a widespread option for expressing progressive aspect (45a,b); diachronically, in many Romance varieties (but not in all of them; cf. standard European Portuguese and Romanesco) the infinitive was replaced by the gerund in these constructions (Andriani 2017:198f.). The infinitive is still an option in certain varieties, being introduced by the non-finite subordinator a ‘to’ (46a,b). The infinitival STAND-periphrasis (alternating with the gerundial one) is also attested in Neapolitan, western Abruzzese, Laziale, and Tuscan varieties (see Andriani 2017:ch.5 for an overview). 45 a b 46 a b ma non vedi che sto but not see.IND.PRS.2SG that stand.IND.PRS.1SG guidando? (It., Andriani 2017:199) drive.GER ‘can’t you see I’m driving (right now)?’ istan fakende su mándicu (Srd., Jones 1993:84) stand.IND.PRS.PL make.GER the food ‘they are preparing the meal’ tu duorme e Ninno tuo you sleep.IND.PRS.2SG and Ninno your a penarǝ (Nap., Rohlfs 1969:133) to suffer.INF ‘you’re sleeping and your Ninno is suffering’ estou a falar com stand.IND.PRS.1SG to speak.INF with você (EuPt., Mendes Mothé 2006:1554) you ‘I’m speaking to you’ stacǝ stands The GO-andative periphrasis with infinitives is specific to certain Italian varieties, where it is found in an early grammaticalization stage, without (yet) encoding temporal and aspectual information (Andriani 2017:205) (47); this type of structure underwent further grammaticalization as an intentional future in Ibero-Romance (48). GO is also attested in Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics progressive constructions with the gerund in Catalan (anar), French (aller), Galician (ir), Italian (andare), Occitan (anar), Portuguese (ir), Romansh (ir), and Spanish (ir) (Squartini 1998:27-29; Bertinetto and Squartini 2016:949f.). 47 a b 48 a b vado a dormire (It., Andriani 2017:205) go.IND.PRS.1SG to sleep.INF ‘I’m going to sleep’ […] chesto che te vaco this that to-you go.IND.PRS.1SG a dì? (Nap.: Ledgeway 2009:454) to say.INF ‘[you’ll certainly be surprised at] what I’m going to tell you?’ ¿dónde van a estar a las dos? (Sp., Zagona 2002:33) where go.IND.PRS.3PL to be at the two ‘where are they going to be at two (o’clock)?’ vou (*a) compra-lo diario (Glc., Pérez Bouza 1996:72) go.IND.PRS.1SG to buy-the newspaper ‘I’m going to buy the newspaper’. Other verbs which may be part of aspectual (progressive) gerundial periphrases have been recorded by Squartini (1998:27-29): COME (Cat. venir, Gal./Pt. vir, It. venire, Sp. venir), WALK (Glc., Pt., Sp. andar), and CARRY (Glc. levar, Sp. llevar). 21.6. Modal complex predicates Unlike auxiliaries (§21.3), but similarly to aspectuals (§21.5), modal verbs (reflexes of protoRomance *poˈtere ‘be able’, Lat. DEBERE/HABERE DE-AB ‘must’, proto-Romance *voˈlere ‘want’, Lat. QUAERERE ‘seek, require’) selecting an infinitive or a non-finite complement do not generally show signs of morphophonological erosion, yet they pass to variable degrees the generally accepted CPred diagnostics (§21.2). Clitic climbing has been extensively investigated across Romance. In modern Romance, the availability of clitic climbing with modals ranges from impossible in French (49a), heavily preferred in Occitan (49b), optional in Spanish (49c), Catalan (49d), and Italian (49e) (with a preference for clitic climbing in the spoken language) to obligatory in Neapolitan (49f) and Romanian (non-finite configurations) (50) (Guțu-Romalo 2005:161f.; Rizzi 1982:4; Picallo 1990:285f.; Motapanyane and Avram 2001:152-54; Abeillé and Godard 2003:127, 129f.; Ledgeway 2012:120f.). 49 a b c d e nous pouvons we can.IND.PRS.1PL ‘we can help them.’ los anam 3MPL= go.IND.PRS.1PL los vamos a 3MPL= go.PRS.1PL to ‘we are going to help them.’ El Pere ho the Pere 3NSG.ACC= ‘Pere must explain it.’ Mario lo vuole / les 3.MPL= aider help.INF (Fr.) (los) ajudar (Lgd.) 3MPL= help.INF ayudar (los) (Sp.) help.INF =3MPL deu explicar (-ho) (Cat.) must.PRS.3SG explain.INF= 3NSG.ACC sa risolver(lo) da solo Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics f 50 a b c Mario 3MSG.ACC= wants knows solve(=3MSG) by himself (questo problema). (It.) this problem ‘Mario wants to / can solve it by himself (this problem).’ ’e gghjammo a aiutà (Nap.) 3MPL.ACC go.IND.PRS.1PL to help.INF ‘we are going to help them.’ îi pot (**îi) ajuta (Ro.) 3MPL.ACC= can.IND.PRS.1SG 3MPL.ACC= help.INF ‘I can help them’ le trebuie (**le) date cărți fetelor (Ro.) 3PL.DAT= must 3PL.DAT= give.PTCP.FPL books girls.DEF.DAT ‘Books must be given to the girls’ îl are de (**le) citit. (Ro.) 3MSG.ACC= has K.SUP 3MSG.ACC= read.SUP ‘(S)he has to read it’ Romanian modals present a more complex picture, as they may take a wider range of complements: they can select either a non-finite complement – a bare short infinitive (50a) (the modal putea ‘can, be able’, in a diachronically stable construction, Hill 2012), a participle or a prepositionless supine (50b) (the modal trebui ‘must, have to’) or a supine (50c) (avea ‘have’, in its modal usage) –, or a subjunctive clausal complement (51) (a Balkan Sprachbund property, cf. Sandfeld 1930:173-80). The subjunctive configuration is unambiguously biclausal (Nicolae 2013:ch.4), as shown by impossibility of clitic climbing and by the availability of negation in the embedded domain. Reflexes of *voˈlere / QUAERERE may also take a finite clausal complement and, in Spanish and Neapolitan, a participial complement (52) (Ledgeway 2012:121). 51 52 (*Îl) pot 3MSG.ACC= can.IND.PRS.1SG citesc. (Ro.) read.SBJV.1SG ‘I can not read it’ a b să COMP.SBJV nu not (îl) 3.MSG.ACC= quería que preparasen todo / preparado vuleva ca preparassero tutta cosa / preparata want.PST.3SG that prepare.PST.3PL everything prepare.PTCP todo (Sp.) tutta cosa (Nap.) everything ‘He wanted for them to prepare everything / everything prepared’ The availability of negation in the embedded domain is also subject to variation: while in languages such as French (53a) and Spanish (53b) (Zagona 2002:195) the clausal negator may freely occur in the embedded domain, in Romanian non-finite configurations (53c) (Nicolae 2013:ch.4) and in Neapolitan (53d) (Ledgeway 2000:168) negation exclusively occurs in the domain of the modal verb: 53 a il peut ne pas dire toute la vérité (Fr.) Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics b c d he may.PRS.IND.3SG not not say.INF all the truth ‘He may not say all the truth’ La cámara no puede leer el nivel de la carga (Sp.) the camera not can.IND.PRS.3SG read.INF the level of the charge ‘The camera cannot read the level of the charge’ nu poate (**nu) spune tot adevărul (Ro.) not can.PRS.IND.3SG not say.INF all truth.DEF ‘He cannot say all the truth’ dubbeto ‘e num puté venì (Nap.) doubt.IND.PRS.1SG of not be.able.INF come.INF ‘I doubt that I won’t be able to come’ Therefore, there appears to be a strong correlation between clitic climbing and the availability of negation in the embedded non-finite domain: languages/configurations with obligatory clitic climbing (Romanian, Neapolitan) disallow negation in the embedded domain, while languages/configurations in which clitic climbing is blocked (French) or optional (Spanish, Italian) freely allow the negator to occur in the embedded domain. This shows that in the former type of languages, the modal verb and its non-finite complement make up a monoclausal configuration, ensuring that CPred formation is obligatory. Of the Romance languages which show auxiliary selection, only in Italian (54a) and Occitan (54b) (but not in French and Neapolitan), is the auxiliary which surfaces on the modal verb sensitive to the transitive / unaccusative nature of the embedded infinitive (Ledgeway 2012:121). 54 a b non mi sono potuto me soi pogut not myself=am been.able ‘I couldn’t fall asleep.’ pas not addormentare (It.) dormir (Lgd.) fall.asleep.INF To sum up, modal verbs show variable degrees of concatenation with their non-finite complement across Romance; while the tight monoclausal nature of the [modal verb + infinitive/non-finite complement] complement is evident in Romanian and Neapolitan, the same is not true of the other Romance languages, in which optional or impossible clitic climbing and the availability of negation in the embedded domain testify to a looser concatenation of the modal verb and the infinitive. 21.7. Causative complex predicates 21.7.1 Introduction Most of the Romance languages, except for modern Daco-Romance (Benucci 1993:53), feature causative constructions with the Romance reflexes of Lat. FACERE ‘make, do’ (55a), LAXARE ‘let’ (55b) and MANDARE ‘command’ (55c) followed by an infinitive (examples from Sheehan 2016:981). 55 a b c La ragazza fece cadere la caraffa. the girl made fall.INF the carafe ‘The girl made the carafe fall.’ Elle a laissé entrer le petit garçon. she have.AUX.PST.3SG let enter.INF the little boy ‘She let the little boy enter.’ O Presidente da Assembleia mandou votar (It.) (Fr.) a lei Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics the president of.the assembly ordered vote.INF the law aos deputados. (EuPt.) to.the MPs ‘The President of the Assembly made the MPs vote through the law.’ CPred formation is diagnosed using the same tests as those used for auxiliaries; however, in contrast to auxiliary-verb constructions, causative constructions (and constructions with perception verbs, see §21.8) contribute to the argument structure of the new overall construction/periphrasis with an external argument, and may also add an adjunct (Labelle 2017:309). The constructions in (55) are characterized by E(xceptional) C(ase) M(arking) on the subject of the infinitive (which surfaces with the accusative – or with the dative, see §21.7.2.1 –, instead of the nominative). When the subject is cliticized, it raises to the first verb, leading to restructuring (55b becomes 56a); negation is also hosted by the first verb (56b). Therefore, constructions such as (55) do not match the CPred diagnostics (§21.2.2), whereas those in (56) do. 56 a b Elle l’a laissé entrer. (Fr.) she 3MSG.ACC=have.AUX.PST.3SG let.PTCP enter.INF ‘She let him enter’ Elle ne l’a pas laissé entrer. (Fr.) she not 3MSG.ACC= have.AUX.PST.3SG NEG let.PTCP enter.INF ‘She does not let him enter.’ 21.7.2 FACERE causatives The causative structure with FACERE had been attested since classical and postclassical times, and it proliferated in early Romance. There is, however, a significant difference between the Latin biclausal construction and the Romance monoclausal one (Vincent 2016). There is major diachronic variation across the Romance languages. In Portuguese and Spanish it appears that clitic climbing, which was previously obligatory, has become optional quite recently, so that Spanish and Portuguese causative constructions have gradually undergone a change from monoclausality to biclausality (Davies 1995; Sheehan 2016:993). By contrast, while modern Romanian disallows the FACERE causative as a CPred (because the verb face ‘make’ selects a subjunctive clausal complement), old Romanian attests the Romance monoclausal infinitival construction (Nedelcu 2016:244) (57). 57 Șî va fi ceriul nou șî pământul and will.AUX.FUT.3SG be.INF sky.DEF.NOM new and land.DEF.NOM nou, cari eu le-am făcut a new which I 3FPL.ACC=have.AUX.PERF.1SG make.PTCP to.INF trăi naintea mea (ORo.) live.INF in.front.of my ‘And the sky and the land that I made come to life in front of me will be new.’ Since the pioneering work on French by Kayne (1975), the literature on causatives has distinguished two types of monoclausal FACERE constructions: faire-infinitif and faire-par (see also Guasti 2006; Ciutescu 2013). Both these types are also attested in Italian, in some varieties of Spanish, and, to a lesser extent, in European Portuguese and Catalan (Sheehan 2016:985), with subtle semantic differences: in French, Italian, and Spanish, the faire-infinitif, unlike the faire-par, involves a sense of obligation exerted on the subject of the second verb Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics (Folli and Harley 2007:212), whereas in the faire-par construction the subject of the first verb has to be an agent (not a cause), and therefore it is obligatorily animate. 21.7.2.1 The faire-infinitif construction In this structure, illustrated in (58), the subject of the embedded verb surfaces in the dative when the verb is transitive; clitics and negation attach to the first verb (Labelle 2017:306). 58 a b c Il fera boire un peu de vin he will.make drink.INF a bit of wine à son enfant. (Fr., Sheehan 2016:987) to his child ‘He’ll make his child drink a bit of wine.’ On la lui laissa réparer. (Fr., Labelle 2017:307) we 3FSG.ACC= 3SG.DAT= let.PST.3SG repair.INF ‘We let him repair it.’ La hizo funcionar. (Sp., Labelle 2017:307) 3FSG.ACC= make.PST.3SG function.INF ‘He made it work.’ The Romance languages show variation with respect to clitics; for example, the reflexive clitics (Zubizarreta 1985:274): se/si may attach to the embedded verb in French and Spanish (59a,b), but not in Italian (59c): 59 a b c On a fait se we have.AUX.PST.SG make.PTCP 3REFL= Pierre. (Fr., Zubizarreta 1985:274) Pierre ‘We made Pierre shave himself.’ Le hicimos lavarse 3MSG.ACC= made.PST.1PL wash.INF=3REFL a Pedro. (Sp.) to Pedro ‘We made Pedro wash his hands.’ **Mario ha fatto accusarsi Mario have.AUX.PST.SG make accuse=3REFL ‘Mario made Piero accuse himself.’ raser shave.INF las manos the hands Piero. (It.) Piero 21.7.2.2 The faire-par construction In the faire-par construction, the subject of the embedded verb is realized as an adjunct PP introduced by the same preposition as the by-phrase in the passive construction (60) (Labelle 2017:319). In Brazilian Portuguese and Catalan, this construction seems to be possible when the prepositional phrase is suppressed (Sheehan 2016:985) 60 a b Elle fera manger cette pomme she make.FUT.3SG eat.INF that apple ‘She’ll have that apple eaten by Jean.’ Gli fecero sparare CL.DAT.3SG make.PST.PL fire.INF da un agente. by an agent par Jean. (Fr., Kayne 1975:235) by Jean addosso upon (It., Burzio 1986:270f.) Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics ‘They had him fired upon by an agent.’ It is a monoclausal construction allowing clitic climbing of the embedded objects, including datives (60b); the reflexive/passive clitic se cannot be hosted by the embedded infinitive (61) (Labelle 2017:326). 61 **On fait / laisse se laver IMPERS make.IND.PRS.3SG let.IND.PRS.3SG 3REFL= wash.INF avant de s'asseoir à table. (Fr., Zubizarreta 1985:264) before of 3REFL=sit.INF at table (intended) ‘We make / let wash oneself before sitting down at table.’ 21.7.3 LAXARE causatives LAXARE causatives with ECM complements are permitted in French, European Portuguese, Spanish, and (marginally) in Italian. In these structures, clitic climbing is permitted, the subject of the infinitive surfaces to its right, and when the embedded infinitive is transitive its subject appears in the dative (Sheehan 2016:984): 62 a b J’ ai laissé lire I have.AUX.PST.1SG let.PTCP read.INF à Jean. (Fr., Sheehan 2016:984) to Jean ‘I have let Jean read the book.’ Je le lui ai I 3MSG.ACC= 3SG.DAT= have.AUX.PST.1SG laissé lire. (Fr.) let.PTCP read.INF ‘I have let him read it.’ le livre the book 21.7.4 MANDARE causatives MANDARE is the canonical causative verb in European Portuguese, where it allows both restructuring/CPred formation (63a) and ECM constructions (63b). 63 a b O João mandou à Ana procurar the João ordered.PST.3sg the Ana fetch.INF o livro. (EuPt., Sheehan 2016:984) the book O João mandou procurar o livro à Ana. (EuPt.) the João ordered find.INF the book to.the Ana ‘João made Ana find the book.’ 21.8. Complex predicates with perception verbs Most of the Romance languages (except for modern Daco-Romance) feature monoclausal constructions with ‘see’ and other perception verbs which select an infinitival complement with a preverbal subject. Perception verb constructions have many features in common with causative constructions. The subject of the embedded verb is exceptionally case marked (with the accusative) (64a) and may surface as an accusative clitic hosted by the main verb (Labelle 2017:300) (64b). Just as in the case of modal verb configurations in certain languages (§21.6), there is variation: clitics and negation may also surface on the embedded verb (64c) (Labelle 1996:91). Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics 64 a b c Jean a vu Marie réparer la voiture. (Fr.) Jean have.AUX.PST.3SG seen Marie.ACC repair.INF the car ‘Jean saw Marie repair the car.’ Jean l’a vue réparer Jean 3FSG.ACC=have.AUX.PST.3SG seen.FSG repair.INF la voiture. (Fr.) the car ‘Jean saw her repair the car.’ J’ ai cru voir Pierre ne pas I have.AUX.PST.1SG believe.PTCP see.INF Pierre not NEG s’arrêter au feu rouge. (Fr.) 3REFL=stop.INF at.the light red ‘I think I have seen Pierre not stop at the red light.’ In Romanian, perception verbs are followed by a gerund (Nicula 2012), in constructions with subject-to-object raising (and negation on the first verb) (65a); evidence for subject-toobject raising comes from the availability of the passive voice (65b), a context in which the raised subject bears nominative and acts as the subject of the passive verb. Note that there is no clitic climbing in this construction (cliticization of the embedded subject on the perception verb in (65a) is the effect of subject-to-object raising). 65 a b Nu îl văd pe not 3MSG.ACC= see.IND.PRS.1SG DOM ‘I do not see Ion hitting her.’ El nu a fost he(NOM) not have.AUX.PST.3SG be.PTCP lovind-o. (Ro.) hit.GER=3FSG.ACC ‘He has not been seen hitting her.’ Ion Ion lovind-o. (Ro.) hit.GER=3FSG.ACC văzut see.PTCP Across Romance, these constructions behave differently (Sheehan 2016:982f.). In Spanish, Italian, French, Catalan, Galician, and European Portuguese (to a lesser extent) the perception verb and its complement make up a CPred. French, Spanish, European Portuguese, Catalan, and Romanian display ECM complements where the lexical verb is a gerund(ive); these constructions involve Exceptional Case Marking; ECM complements do not generally display restructuring phenomena such as clitic climbing, though the subject of the embedded verb behaves like an argument of the matrix perception verb and must be expressed on the matrix verb when it is realized as a clitic. 21.9. Conclusions: what Romance languages tell us about complex predicates The concept of CPred is, to a certain degree, fuzzy, yet it brings under the same umbrella a variety of syntactically and semantically distinct constructions, which have in common monoclausality (mainly diagnosed by clitic climbing and the exclusive realization of negation on the high verb). The examination of the empirical data has shown that the distinction between CPreds based on merger and CPreds based on coindexation (Baker and Harvey 2010; see §21.2.1) holds with respect to the Romance languages too: on the one hand, in CPreds with auxiliaries, modals, and aspectuals, the argument structure of the formation reflects the argument structure of lexical verbs, hence auxiliary/modal/aspectual verbs are ‘athematic’ predicates Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics (which ‘do not assign either an external or internal argument but, rather, inherit their argument structure directly from their non-finite verbal complement’, Ledgeway 2007:121), but rather mere placeholders for TAM values; on the other hand, CPred formations with causatives and perception verbs are characterized by an argument/conceptual structure distinct from that of the component verbs. Morphophonological erosion affects only a subclass of CPred constructions, those with (non-passive) auxiliaries. By contrast, modals and aspectuals (as well as passive auxiliaries) do not generally undergo morphophonological erosion / weakening, yet they share with auxiliaries the fact that they are devoid of thematic properties; the term ‘semi-auxiliary’ appears to be an appropriate label for this class of fully inflecting, yet athematic, verbs. Furthermore, modal and aspectual semi-auxiliaries may also occur in biclausal configurations and preserve their athematic nature; identity of subject between the higher and embedded predicates (technically obtained via subject raising or obligatory control) gives the appearance of a more cohesive structure, yet a CPred formation is not at play, given biclausality. The diachronic paths taken by the various types of Romance CPreds are also diverse. Some of the formations undergo grammaticalization, the most extreme case being that of the western Romance futures and conditionals; in this case, a new synthetic system was created on the basis of analytic formations with postverbal auxiliaries. Other CPreds are extremely stable form a diachronic point of view: the Romanian modal CPred made up of the verb putea ‘can, be able’ plus a bare infinitive did not undergo any significant change in the five centuries of the attested history of Romanian (Hill 2012). Opposite diachronic developments are also attested. For example, rather than undergoing further structural simplification, the Spanish and Portuguese causative constructions appear gradually to revert to a biclausal structure, as shown by the fact that previously obligatory clitic climbing has quite recently become optional. Global changes affecting the syntax of a given language may have local effects on the grammar of CPreds: the replacement of the infinitive by the subjunctive in Romanian (a more general Balkan Sprachbund phenomenon) has led to the replacement of the infinitival monoclausal causatives attested in old Romanian with subjunctive biclausal causatives, the only option in modern Romanian. To sum up, the phenomena investigated point to the fact that ‘complex predicate’ is a heteroclite concept in any accepted sense, as every construction in its turn features different properties and has non-converging diachronic developments. References Abeillé, A. and Godard, D. (2002). ‘The syntactic structure of French auxiliaries’, Language 78:404-52. Abeillé, A. and Godard, D. (2003). ‘Les prédicats complexes dans les langues romanes’. In Godard, D. (ed.), Les Langues romanes. Problèmes de la phrase simple. Paris: CNRS Éditions, 125-84. Abeillé, A., Godard, D., and Sag, I. A. (1998). ‘Two kinds of composition in French complex predicates’. In Hinrichs, E., Kathol, A., and Nakazawa, T. (eds), Complex Predicates in Nonderivational Syntax. San Diego: Academic Press, 1-41. Adams, J. (1991). ‘Some neglected evidence for Latin HABEO with infinitive: the order of the constituents’, Transactions of the Philological Society 89:131-96. Adams, J. N. (2013). Social Variation and the Latin Language. Cambridge: Cambridge University Press. Adriani, L. (2017). The Syntax of the Dialect of Bari. University of Cambridge: unpublished doctoral thesis. Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics Aissen, J. and Perlmutter, D. (1976). ‘Clause reduction in Spanish’. In Thompson, H. (ed.), Proceedings of the Second Annual Meeting of the Berkeley Linguistic Society. Berkeley: Berkeley Linguistic Society, 1-30. Alfonzetti, G. (1998). ‘Passato prossimo e passato remote: dimensioni di variazione’. In Ruffino, G. (ed.), Atti del XXI congresso internazionale di linguistica e filologia romanz. Centro di studi filologici e linguistici siciliani, Universtà di Palermo 18-24 settembre 1995, II, Tübingen, Niemeyer, 346-60. Alkire, T. and Rosen, C. (2010). Romance Languages. A Historical Introduction. Cambridge: Cambridge University Press. Alsina, A. (2016). ‘Catalan’. In Ledgeway, A. and Maiden, M. (eds), The Oxford Guide to the Romance Languages. Oxford: Oxford University Press, 363-81. Amberber, M., Baker, B. and Harvey, M. (2010). ‘Introduction’. In Amberber, M., Baker, B., and Harvey, M. (eds), Complex Predicates. Cross-linguistic Perspectives on Event Structure. Cambridge: Cambridge University Press, 1-12. Anderson, S. (2016). ‘Romansh (Rumantsch)’. In Ledgeway, A. and Maiden, M. (eds), The Oxford Guide to the Romance Languages. Oxford: Oxford University Press, 169-84. Aranovich, R. (2009). ‘From esse to ser: diachronic mismatches in the selection of perfect auxiliaries’. In Reinheimer Rîpeanu, S. (ed.), Studia linguistica in honorem Mariae Manoliu. Bucharest: Editura Universităţii din Bucureşti, 21-35. Avram, L. and Hill, V. (2007). ‘An irrealis BE auxiliary in Romanian’. In Aranovich, R. (ed.), Split Auxiliary Systems: a crosslinguistic perspective. Amsterdam: Benjamins, 47-64. Baker, B. and Harvey, M. (2010). ‘Complex predicate formation’. In Amberber, M., Baker, B., and Harvey, M. (eds), Complex Predicates. Cross-linguistic Perspectives on Event Structure. Cambridge: Cambridge University Press, 13-47. Bauer, B. (2006). ‘“Synthetic” vs “analytic” in Romance: the importance of varieties’. In Gess, R. and Arteaga, D. (eds), Historical Romance Linguistics: Retrospective and Perspectives. Amsterdam: John Benjamins, 287-304. Benucci, F. (1993). ‘Temporal periphrasis and clitics in central Romance languages’, Catalan Working Papers in Linguistics 3:51-83. Benveniste, É. (1962). Hittite et indo-européen. Études comparatives. Paris: Adrien Maisonneuve. Bertinetto, P. M. and Squartini, M. (2016). ‘Tense and aspect’. In Ledgeway, A. and Maiden, M. (eds), The Oxford Guide to the Romance Languages. Oxford: Oxford University Press, 939-53. Bosque, I. and Gallego, Á. J. (2011). ‘Spanish double passives and related structures’, Revista de estudos linguísticos da Univerdade do Porto 6:9-50. Buridant, C. (2000). Grammaire nouvelle de l’ancien français. Paris: SEDES. Burton, P. (2009). ‘Analytic passives and deponents in classical and later Latin’. In Adams, J. and Vincent, N. (eds), Early and Late Latin. Continuity or Change?. Cambridge: Cambridge University Press, 163-79. Burzio, L. (1981). Intransitive Verbs and Italian Auxiliaries. MIT: unpublished doctoral thesis. Burzio, L. (1986). Italian Syntax. A Government-Binding Approach. Dordrecht: Foris. Cabredo Hofherr, P. (2017). ‘Voice and voice alternations’. In Dufter, A. and Stark, E. (eds), Manual of Romance Morphosyntax and Syntax. Berlin: De Gruyter, 230-71. Cardinaletti, A. and Shlonsky, U. (2004). ‘Clitic positions and restructuring in Italian’, Linguistic Inquiry 35:519-57. Cennamo, M. (2016). ‘Voice’. In Ledgeway, A. and Maiden, M. (eds), The Oxford Guide to the Romance Languages. Oxford: Oxford University Press, 967-80. Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics Centineo, G. (1986). ‘A lexical theory of auxiliary selection in Italian’, Davis Working Papers in Linguistics 1:1-35. Cinque, G. (2004). ‘Restructuring and functional structure’. In Belletti, A. (ed.), Structures and Beyond: the cartography of syntactic structures, III. Oxford: Oxford University Press, 132-91. Ciutescu, E. (2013). ‘Micro-parametric variation in Romance causative constructions’, Bucharest Working Papers in Linguistics 15:45-60. D’Alessandro, R. (2007). Impersonal si constructions. Berlin: De Gruyter. Danckaert, L. (2016). ‘Variation and change in Latin BE-periphrases: empirical and methodological considerations’. In Adams, J. and Vincent, N. (eds), Early and Late Latin. Continuity or Change? Cambridge: Cambridge University Press, 132-62. Danckaert, L. (2017). ‘The origins of Romance analytic passive: evidence from word order’. In Mathieu, E. and Truswell, R. (eds), Macro-change & Micro-change in Diachronic Syntax. Oxford: Oxford University Press, 216-34. Davies, M. (1995). ‘The evolution of causative constructions in Spanish and Portuguese’. In Amastae, J., Goodall, G., Montalbetti, M., and Phinney, M. (eds), Contemporary Research in Romance Linguistics. Amsterdam: Benjamins, 105-22. Dobrovie-Sorin, C. (1994). The Syntax of Romanian. Berlin: Mouton de Gruyter. Dragomirescu, A. (2013). ‘Complex predicates’. In Pană Dindelegan, G. (ed.), The Grammar of Romanian. Oxford: Oxford University Press, 191-203. Dragomirescu, A. and Nicolae, A. (2013). ‘Urme ale selecţiei auxiliarului de perfect compus în română’. In Cepraga, D. O., Lupu, C., and Renzi, L. (eds), Hommages offerts à Florica Dimitrescu et Alexandru Niculescu. Bucharest: Editura Universităţii din Bucureşti, 33852. Dragomirescu, A. and Nicolae, A. (2014). ‘The multiple grammaticalization of Romanian veni ‘come’: focusing on the passive construction’. In Devos, M. and van der Wal, J. (eds), COME and GO Off the Beaten Grammaticalization Path. Berlin: De Gruyter Mouton, 69-100. Fleischman, S. (1983). ‘From pragmatics to grammar. Diachronic reflections on complex pasts and futures in Romance’, Lingua 60:183-214. Folli, R. and Harley, H. (2013). ‘The syntax of argument structure: evidence from Italian complex predicates’, Journal of Linguistics 49:93-125. Frank, A. (1996). ‘A note on complex predicate formation: evidence from auxiliary selection, reflexivization, and past participle agreement in French and Italian’. In Butt, M. and Holloway King, T. (eds), The Proceedings of the LFG96 Conference. CSLI On-line Publications. Giacalone Ramat, A. (2000). ‘On some grammaticalization patterns for auxiliaries’. In Smith, J. C. and Bentley, D. (eds.), Historical Linguistics 1995: selected papers from the 12th International Conference on Historical Linguistics, Manchester, August 1995. V.1. General Issues and Non-Germanic Languages. Amsterdam: Benjamins, 125-54. Giacalone Ramat, A. (2017). ‘Passives and constructions that resemble passives’. Folia Linguistica Historica 38:149-176. Giurgea, I. (2011). ‘The Romanian verbal cluster and the theory of head movement’. In Herschensohn, J. (ed.), Romance linguistics 2010. Selected Papers from the 40th Linguistic Symposium on Romance Linguistics (LSRL), Seattle, Washington, March 2010. Amsterdam: Benjamins, 271-86. Green, J. N. (1987). ‘The evolution of Romance auxiliaries: criteria and chronology’. In Harris, M. and Ramat, P. (eds), Historical Development of Auxiliaries. Berlin: Mouton de Gruyter, 257-77. Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics Green, J. N. (1991), ‘The collapse and replacement of verbal inflection in late Latin/early Romance: how would one know?’. In Wright, R. (ed.), Latin and the Romance Languages in the Early Middle Ages. London: Routledge, 83-99. Guasti, M.-T. (2006). ‘Analytic causatives’. In Everaert, M. van Riemsdijk, H. (eds), The Blackwell Companion to Syntax, I. Oxford: Blackwell, 142-72. Guţu Romalo, V. 2005. ‘Semiauxiliarele de mod’. In Guțu Romalo, V., Aspecte ale evoluţiei limbii române. Bucharest: Humanitas, 156-84. Harris, A. and Campbell, L. (1995). Historical Syntax in Cross-Linguistic Perspective. Cambridge: Cambridge University Press. Harris, M. (1982). ‘The ‘past simple’ and the ‘present perfect’ in Romance’. In Harris, M. and Vincent, N. (eds), Studies in the Romance Verb. London: Croon Helm, 42-70. Haverling, G. (2016). ‘On the use of habeo and the perfect participle in earlier and later Latin’. In Adams, J. and Vincent, N. (eds), Early and Late Latin. Continuity or Change? Cambridge: Cambridge University Press, 180-201. Hill, V. (2012). ‘Romanian ‘can’: changes in parametric settings. In Galves, C., Cyrino, S., Lopes, R., Sandalo, F., and Avelar, J. (eds), Parameter Theory & Linguistic Change. Oxford: Oxford University Press, 265-80. Iordan, I. (1950). ‘Note sintactice’. Studii si cercetări lingvistice 1:269-79. Jones, M. A. (1993). Sardinian Syntax. London: Routeledge. Kayne, R. (1975). French Syntax: the transformational cycle. Cambridge, MA: MIT Press. Krusch, B. (1888). Monumenta Germaniae Historica, Scriptores Rerum Merovingicarum, II. Hanover. Labelle, M. (1996). ‘Remarques sur les verbes de perception et la sous-catégorisation’, Recherches linguistiques de Vincennes 25:83-106. Labelle, M. (2017). ‘Causative and perception verbs’. In Dufter, A. and Stark, E. (eds), Manual of Romance Morphosyntax and Syntax. Berlin: De Gruyter, 299-331. Lamiroy, B. (1999). ‘Auxiliaires, langues romanes et grammaticalisation’, Langages 135:3345. Leal Cruz, P. (2003). El español tradicional de La Palma. La Laguna: Centro de la Cultura Popular Canaria. Ledgeway, A. (2000). A Comparative Syntax of the Dialects of Southern Italy: a minimalist approach. Oxford: Blackwell. Ledgeway, A. (2007). ‘Old Neapolitan word order: some initial observations’. In Lepschy, A. L. and Tosi, A. (eds). Languages of Italy. Histories and Dictionaries. Ravenna: Longo Editore Ravenna, 119-46. Ledgeway, A. (2009). Grammatica diacronica del napoletano. Tübingen: Niemeyer. Ledgeway, A. (2012). From Latin to Romance. Morphosyntactic typology and change. Oxford: Oxford University Press. Ledgeway, A. (2016). ‘Italian, Tuscan, and Corsican’. In Ledgeway, A. and Maiden, M. (eds), The Oxford Guide to the Romance Languages. Oxford: Oxford University Press, 206-27. Ledgeway, A. (2017). ‘Syntheticity and analyticity’. In Dufter, A. and Stark, E. (eds), Manual of Romance Morphosyntax and Syntax. Berlin: De Gruyter, 839-86. Ledgeway, A. (2021a). ‘Passive periphrases in the Romance Languages’. In Gardani, F. and Loporcaro, M. (eds), The Oxford Encyclopedia of Romance Linguistics. Oxford: Oxford University Press. Advance online publication. https://doi.org/10.1093/acrefore/9780199384655.013.639. Ledgeway, A. (2021b). ‘Indirect object WANT-passives in southern Italy’. In Grohmann, K, Matsuya, A., and Remberger, E.-M. (eds), Passives Cross-linguistically: theoretical and experimental approaches (empirical approaches to linguistic theory). Leiden: Brill, 138170. Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics Legendre, G. (2017). ‘Auxiliaries’. In Dufter, A. and Stark, E. (eds), Manual of Romance Morphosyntax and Syntax. Berlin: De Gruyter, 272-98. Loporcaro, M. (1988). Grammatica storica del dialetto di Altamura. Pisa: Giardini. Loporcaro, M. (2007). ‘On triple auxiliation in Romance’, Linguistics 45:173-222. Mackenzie, I. (2006). Unaccusative verbs in Romance languages. New York: Palgrave Macmillan. Maiden, M. (2016). ‘Inflectional morphology’. In Ledgeway, A. and Maiden, M. (eds), The Oxford Guide to the Romance Languages. Oxford: Oxford University Press, 497-512. Maiden, M. and Robustelli, C. (2007). A Reference Grammar of Modern Italian. London: Routledge. Marchese, F. (2016). Il lessico del dialetto di Olia (VV). University of Florence: unpublished doctoral dissertation. Mendes Mothé, N. G. (2006). ‘Gerúndio versus Infinitivo Gerundivo: Brasil e Portugal em contraste nos séculos XIX e XX’. Estudos lingüísticos 35:1554-63. Mensching, G. and Remberger, E.-M. (2016). ‘Sardinian’. In Ledgeway, A. and Maiden, M. (eds), The Oxford Guide to the Romance Languages. Oxford: Oxford University Press, 270-91. Monachesi, P. (2005). The Verbal Complex in Romance: a case study in grammatical interfaces. Oxford: Oxford University Press. Motapanyane, V. and Avram, L. (2000). ‘The syntax of putea and its mixed typology’, Belgian Journal of Linguistics 14:149-65. Müller, S. (2006). ‘Complex predicates’. In Brown, K. (ed.), Encyclopaedia of Language and Linguistics. Oxford: Elsevier, 697-704. Nedelcu, I. (2016). ‘The infinitive and the infinitival construction’. In Pană Dindelegan, G. (ed.), The Syntax of Old Romanian. Oxford: Oxford University Press, 232-49. Nicolae, A. (2013). Types of Ellipsis in Romanian. University of Bucharest and University of Cambridge: unpublished doctoral thesis. Nicolae, A. (2015). Ordinea constituenţilor în limba română: o perspectivă diacronică. Bucharest: Editura Universității din București. Nicolae, A. (2019). Word Order and Parameter Change in Romanian. A Comparative Romance Perspective. Oxford: Oxford University Press. Nicula, I. (2012). Modalități de exprimare a percepțiilor fizice. Verbele de percepție în limba română. Bucharest: Editura Universității din București. Pană Dindelegan, G. (2003). Elemente de gramatică. Dificultăţi, controverse, noi interpretări. Bucureşti: Humanitas Educaţional. Penny, R. (2000). Variation and Change in Spanish. Cambridge: Cambridge University Press. Pérez Bouza, J. A. (1996). El Gallego. Munich / Newcastle: Lincom Europa. Pesetsky, D. (2019). Exfoliation: towards a derivational theory of clause size. MIT, unpublished MS. Picallo, C. M. (1990). ‘Modal verbs in Catalan’. Natural Language & Linguistic Theory 8:285-312. Pinkster, H. (1987). ‘The strategy and chronology of the development of future and perfect tense auxiliaries in Latin’. In Harris, M. and Ramat, P. (eds), Historical Development of Auxiliaries. Berlin: Mouton de Gruyter, 193-223. Ramat, P. (1987). ‘Introduction’. In Harris, M. and Ramat, P. (eds), Historical Development of Auxiliaries. Berlin: Mouton de Gruyter, 3-19. Rizzi, L. (1978). ‘A restructuring rule in Italian syntax’. In Keyser, S. J. (ed.), Recent Transformational Studies in European Languages. Cambridge, MA: MIT Press, 113-58. Rizzi, L. (1982). Issues in Italian Syntax. Dordrecht / Cinnaminson: Foris. Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023 Cambridge Handbook of Romance Linguistics Roberts, I. (2013). ‘Some speculations on the development of the Romance periphrastic perfect’. Revue roumaine de linguistique 58:3-30. Rohlfs, G. (1969). Grammatica storica della lingua italiana e dei suoi dialetti: sintassi e formazione delle parole. Turin: Einaudi. Salvi, G. (1987). ‘Syntactic restructuring in the evolution of Romance auxiliaries’. In Harris, M. and Ramat, P. (eds), Historical Development of Auxiliaries. Berlin: Mouton de Gruyter, 225-36. Sandfeld, K. (1930). Linguistique balkanique: problèmes et résultats. Paris: Champion. Sârbu, R. and Frăţilă, V (1998). Dialectul istroromân. Timişoara: Amarcord. Sheehan, M. (2016). ‘Complex predicates’. In Ledgeway, A. and Maiden, M. (eds), The Oxford Guide to the Romance Languages. Oxford: Oxford University Press, 981-93. Sorace, A. (2000). ‘Gradients in auxiliary selection with intransitive verbs’, Language 76:859-90. Sorace, A. (2004). ‘Gradience at the lexicon–syntax interface: evidence from auxiliary selection for unaccusativity’. In Alexiadou, A., Anagnostopoulou, E., and Everaert, M. (eds), The Unaccusativity Puzzle. Explorations of the Syntax-Lexicon Interface. Oxford: Oxford University Press, 243-68. Squartini, M. (1998). Verbal Periphrases in Romance. Aspect, Actionality, and Grammaticalization. Berlin: Mouton de Gruyter. Squartini, M. and Bertinetto, P. M. (2000). ‘The simple and compound past in Romance languages’. In Dahl, Ö. (ed.), Tense and Aspect in the Languages of Europe. Berlin: Mouton de Gruyter, 403-39. Stolova, N. I. (2006). ‘Split intransitivity in Old Spanish: irreals and negation factors’, Revue roumaine de linguistique 51:301-20. Svenonius, P. (2008). ‘Complex predicates and the functional sequence’, Tromsø Working Papers on Language & Linguistics: Nordlyd 35:47-88. Thielmann, P. (1885). ‘Habere mit dem Part. Perf. Pass.’, Archiv für Lateinische Lexikographie 2:372-423; 509-49. Van Valin, R. D. (1987). ‘The unaccusative hypothesis vs. lexical semantics’. In McDonough, J. and Plunkett, B. (eds), Proceedings of NELS 17, 641-61. Vincent, N. (1987). ‘The interaction of periphrasis and inflection: some Romance examples’. In Harris, M. and Ramat, P. (eds), Historical Development of Auxiliaries. Berlin: Mouton de Gruyter, 237-56. Vincent, N. (2016). ‘Causatives in Latin and Romance’. In Adams, J. and Vincent, N. (eds), Early and Late Latin. Continuity or Change? Cambridge: Cambridge University Press, 294-312. Wheeler, M., Yates, A., and Dols, N. (1999). Catalan. A Comprehensive Grammar. London: Routledge. Wurmbrand, S. (2001). Infinitives: restructuring and clause structure. Berlin: De Gruyter Mouton. Zafiu, R. (2013). ‘Mood, tense, and aspect’. In Pană Dindelegan, G. (ed.), The Grammar of Romanian. Oxford: Oxford University Press, 24-64. Zagona, K. (2002). The Syntax of Spanish. Cambridge: Cambridge University Press. Zubizarreta, M. L. (1985). ‘The relation between morphophonology and morphosyntax: the case of Romance causatives’, Linguistic Inquiry 16:247-89. Dragomirescu, Adina & Nicolae, Alexandru & Pană Dindelegan, Gabriela. 2022. Complex Predicates. In Adam Ledgeway & Martin Maiden (eds.), The Cambridge Handbook of Romance Linguistics, 571–603. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/9781108580410.023