Clitic Doubling and Differential Object Marking1 Alina-Mihaela Tigău University of Geneva & University of Bucharest 1. Introduction: CD across languages and the problems it raises How Romanian contributes to answering CD-related puzzles Towards an account of Cl+ pe DPs with a focus on: o o o The relation between the clitic and its double as a reflex of argument licensing i.e., taking abstract case as part of argument licensing (Sigursson 2007, 2012a, 2012b) to amount to a relation with a functional head, we will argue that case marking of the doubled DP is realized by means of its relation to the clitic Contra Kayne’s generalization, we will propose that the differential marker pe is not a case marker. The clitic and its double are licensed as one event participant > they represent “the same argument”, saturating one position in the theta structure of the predicate: as such the clitic gives rise to an unsaturated λ-abstract. The referential doubled DP would then be a semantic subject of predication saturating the λ-abstract in order to give rise to a proposition (Delfitto 2002, Aoun 1999) 2. Background Clitic Doubling (CD): a construction in which a pronominal clitic co-occurs with a full DP (1) L-am ajutat pe Ion. Him.cl.-have.I helped pe John. ‘I have helped John.’ Differential Object Marking (DOM): a mechanism by means of which prominent direct object DPs are marked > Romanian pe2 (a preposition similar to a in Spanish) to mark prominent direct objects. CD is dependent on DOM: only DOMed DPs may be CDed This dependence is unidirectional in the sense that DOM does not require CD 1 This work is supported by the Sciex post-doctoral fellowship 12 203/1.03.2014-28.02.205; project: SciexNMSch - Contribution suisse à l’élargissement de l’UE 2 We have employed the word ‘prominence’ because we have not yet established which is the exact factor that triggers DOM in Romanian and because languages differ with respect to what types of objects exactly they overtly case mark. Prominence is measured along several scalar dimensions such as animacy and definiteness (Aissen 2003) i.e., the more definite a DP is the more likely it is for it to be pe marked; similarly, [+human] DP are always pe marked as opposed to [- animate] ones, which are never pe marked. Farkas (2002a), Farkas & von Heusinger (2003) point out that stability of reference is also relevant for pe marking: DPs with unconditional dynamic stability (proper names and pronouns) are obligatorily pe marked, while dynamically non-stable DPs (indefinites) are less likely to get pe. The account in Farkas & von Heusinger (2003) has the advantage of accounting for the matters of obligatoriness vs. optinality or impossibility concerning pe. 1 3. The importance of CD and the problems it raises across languages Kayne (1975): complementary distribution between the argumental counterparts of full DP objects which have to move out of their merge position within the VP and the corresponding full DP (French): (2) a. Je vois Jean. b. Je le vois. c. *Je le vois Jean. Jaeggli (1982): Clitic Doubling languages pose a problem for this account: as the clitic and the associate DP may co-occur (Romanian, Spanish etc) (3) a. L-am ajutat pe Ion. Him.cl.-have.I helped pe John. ‘I have helped John.’ b. Lo Him.cl vimos a Juan. saw.we a Juan Kayne’s Generalisation: the clitic absorbs the case and the DP double can only occur within a prepositional phrase where it receives case from the preposition Suner (1988) > Porteño/Rioplatense Spanish: DO clitic doubling is possible without a (4) Yo lo voy a comprar el diario justo antes de subir I it.cl. will buy the newspaper just before of appear ‘I am going to buy the newspaper just before coming up.’ (Suñer 1988: 400) Balkan languages like (Greek, Albanian, Bulgarian): a bare Accusative was clitic doubled. (5) Ana mëi pa muai në rrugë. Ana.thenom mecl saw meFP in road ‘Ana saw me on the road.’ (Albanian, Kallulli & Tasmowski 2008: 2) (6) nu-lu ávdu fiĉórlu not-him heat boy.the ‘I don’t hear the boy.’ (Aromanian, Caragiu-Marioţeanu 1975: 237 in Kallulli & Tasmowski 2008: 5) (Some) Problems to solve 2 3.1 Movement vs. base generation the CD data come to question the movement analysis: if a full DP may co-occur with a clitic, doesn’t this point against a movement analysis? base generation: the full DP occupies the internal argument position of the verb, while the clitic is merged either as a sister of V (Jaeggly 1982, Rivas 1977) or as part of the V head (Borer 1984), or a functional head (Sportiche 1996) etc. still movement: 1. if the full DP is an adjunct: Aoun (1981), Philippaki-Warburton (1987), Androulakis (2001), Coene & Avram (2009) a.o > movement of the clitic 2. if the clitic and the associate DP both merge as one complex argument of the verb after which movement applies (BigDP): Uiragereka (1995), Belletti (1999), Papangeli (2000), Cornilescu (2006), Boeckx (2001). 3.2 The status of the clitic & the status of the double 1. the clitic is the real argument whereas the associate DP is an adjunct 2. the associate DP is an argument, while the clitic is an agreement marker/ a functional element/a determiner 3.3 The status of the differential object marker According to Kayne’s Generalisation put forth in Jaeggli (1982), the preposition preceding the full DP is a case assigner saving the derivation from crashing: the clitic absorbs the accusative case from the verb which may no longer assign case to the full DP. Hence a preposition is introduced to save the derivation. There are, however, problems with this analysis: Languages which do not require DOM with CD Romanian: the fact that DOM may be used in the absence of CD, with a verb that can assign the Accusative (which is no longer absorbed by the clitic) > if the Accusative case assigned by V is no longer absorbed by a clitic in simple DOM cases, why do we need a case assigner? 3.4 The relation between CD and DOM (for Romanian) Romanian CD requires DOM but not the other way round: why this dependency? 4. The answers Romanian provides to these puzzles: in what follows we address each of the five major puzzles that the literature on CD has identified and present the way in which Romanian may contribute to solving (some of) these puzzles: 4.1. Movement vs. base generation > in favour of a movement analysis there are several arguments supporting a movement analysis: both XP movement and X movement across languages 3 o XP movement Greek < Alexiadou (1999): objects may bind subjects to their right (7) Sinodhepse to kathe pedhii i mitera tui. Accompanied the every child the mother his ‘His mother accompanied every child.’ Standard Spanish: Zubizarreta (1998) (8) Lo castigo a el la madre de Juan. Cl-acc punished a him the mother of John ‘John's mother punished him.’ Romanian: strengthens the arguments in favour of a movement analysis: A. inverse binding > evidence that cl+pe DPs leave the VP (9) a. *Colegii luii au ajutat pe Mihaii. Coleagues.the hisi have.they helped pe Mihaii. ‘Hisi colleagues helped Mihai.’ b. Colegii luii l-au ajutat pe Mihaii. Coleagues.the hisi him.cl.-have.they helped pe Mihaii. ‘Hisi colleagues helped Mihai.’ (10) Pe un profesor buni orice elev de-al PE a teacher good any pupil of-ART.GEN ‘Any pupil will admire a good teacher.’ luii îl admiră. his him.cl admire.3SG > the clitic doubled and pe-marked indefinite pe un profesor binds the subject DP which it appears to c-command. The same relation of coreference between the two DPs is maintained when the word order changes from OSV to SOV: (11) Orice elev de-al luii îl admiră pe any pupil of-ART.GEN his CL.3SG.M.ACC admire.3SG PE ‘Any pupil admires a good teacher.’ un a profesor teacher buni. good - the unmarked DP may only bind into the subject if it is in a c-commanding position with respect to this DP. (12) a. *Profesorul lori doar câţiva elevii nu va teacher theiri only some pupilsi not AUX.3SG putea lăuda la sfârşitul anului, can praise at end.DEF year.GEN.DEF ‘It is only a few pupils that their professor will not be able to praise at the end of the year.’ b. Doar câţiva elevii only some pupilsi putea lăuda la profesorul lori teacher.DEF theiri sfârşitul anului, nu not va AUX.3SG 4 can praise at end.DEF year.GEN.DEF ‘It is only a few pupils that their professor will not be able to praise at the end of the year.’ B. Parasitic gaps (Cornilescu 2002) - (13) Cornilescu (2002) > Heavy NP Shift (HNPS) of undoubled object DPs may license parasitic gaps, as opposed to their doubled counterparts. a. Am examinat t fãrã a intrerupe t fiecare concurent Have.I examined t without interrupting t each competitor 'I examined without interrupting each candidate separately separat. separately b. Am examinat t fãrã a intrerupe t pe fiecare concurent separat. Have.I examined t without interrupting t pe each competitor separately 'I examined without interrupting PE each candidate separately (14) a. *L-am întîlnit fãrã a saluta t însã pe Ion him.cl-have.I met without to greet t though pe Ion. 'I met Ion without greeting him though.' b. L-am Him.cl-have.I (15) întîlnit fãrã a-l saluta însã met without to-him.clgreet though a. *L- am examinat t fãrã Him.cl-have.I examined t without pe Ion pe Ion. a intrerupe t pe fiecare concurent separat. interrupting t pe each candidate separately b L- am examinat t fãrã a-l intrerupe t pe fiecare concurent separat. Him.cl-have.I examined t without to-him.cl interrupt t pe each candidate separately. >HNPS is an A’ movement which targets a position inside vP; the operator in the PG adjunct clause should be a copy of the antecedent, structurally identical with it> only DPs which are not clitic licensed may be antecedents in a PG construction as they remain inside the vP. > If the antecedent is clitic licensed, i.e., it is a BigDP, the operator in the PG is also a BigDP licensed by a clitic. But the clitic moves out of the vP to T hence the adjunct clause should be itself a TP, adjoined to the main TP, and should also contain a clitic (as the operator should be a copy of the antecedent) > HNPS is possible with CD-ed object DPs without the licensing of PGs C. Focus and Focus Projection (Gierling 1997) (16) a. Ce cauți?/ Ce faci? b. Caut o CARTE (17) a. Pe cine cauți? b. Il caut pe ION (18) a. Ce faci? b. *Il caut pe ION (18b) can only be a felicitous answer to a question inquiring about the argument as in (17). In order for (18b) to be a felicitous answer to (18a) it must contain an additional accent on the verb. This is 5 due to the impossibility for focus to project from a DP which is doubled by a clitic to a higher constituent (like the VP). The solution lies in the fact that the doubled DP has left the VP. Selkirk (1995:555) Focus Projection: a. Focus marking of the head of a phrase licenses the F-marking of the phrase b. F-marking of an internal argument of a head licenses the F-marking of the head This licensing mechanism correctly ensures that F-projection is not possible from an accented adjunct. For a VP to be focused in this case, an accent on the verb is needed as well. (19) [F Il CAUT pe ION.] But why should the focus feature instantiated by the accent on an argument not be able to license the F-marking of the verb and thus of the VP? That the argument in (18b) behaves like an adjunct is quite unexpected. Apparently, with CDed DPs focus cannot project from the DP. >> with CD, the associate DP has left the VP. o X0 movement Kayne (1989b) > clitic climbing: sensitive to the presence of an intervening whcomplementizer but not sensitive to the presence of an intervening wh-phrase (20) a. ?Mario, non loi saprei [a chi affidare ti]. Mario, not Cl-acc I-would-know to whom entrust ‘Mario, I wouldn't know to whom to entrust him.’ b. *Su questo problema, non loi saprei [se consigliare ti]. On this problem, not him I-would-know whether advise ‘On this problem I wouldn't know whether to advise him.’ The wh-phrase a chi does not interfere with clitic climbing while the wh-complementizer se does. The position and order of clitics the interaction of clitics and the syntax of negation, mood, complementizers Kayne 1991, 1994; Zanuttini 1991, 1997; Rivero 1994; Rivero and Terzi 1995 a.o o o Rivero (1994) clitics left-adjoin to the inflectional head to which the verb raises, in finite environments T or AgrS in finite clauses Terzi (1996) clitics adjoin to T and/or afunctional head F Romanian: evidence from supine (Cornilescu & Cosma 2014 ms.): the verb may take a pe DP but not a cl+pe DP in prepositional supine clauses (21) Nu pot să vin la tine azi, am de vizitat *(pe) cineva. Not can.I să come to you today, have.I de visited *(pe) somebody ‘I cannot pass by today, I have to visit someone.’ - the indefinite pronoun ranging over humans is necessarily marked by pe and never clitic doubled. This is why, it is a perfect candidate for the verbal supine - the prepositional supine does not allow personal pronouns as they are obligatorily clitic doubled: 6 (22) a. *(I-)am ajutat pe ei să reușească. *(Them.cl-) have.I helped pe them să succeed. ‘I have helped them succeed.’ b. *E ușor de ajutat pe ei Is easy de helped pe them ‘They are easy to help.’ -thus, the prepositional supine in (b) seems to reject internal arguments that are necessarily clitic doubled and to only accept DPs which do not require double or which disallow it > as pointed by Cornilescu & Cosma (to appear) this is so because the prepositional supine is a reduced clause lacking Agreement. If the clitic were to remain inside the vP it would then have no problem in being licensed inside a prepositional supine. whenever there is a clitic there is movement: - (23) additional evidence: doubled indirect object DPs > may bind the subject (Tigău 2014, Cornilescu 2014 ms.) a. Prietenii lori/j le ajută multorai. Friends.the their them.cl help.they many.Dat ‘Their friends help many.’ b. Prietenii lor*i/j ajută multorai. Friends.the their help.they many.Dat ‘Their friends help many.’ The clitic is the trigger for movement > why? Why is there movement? What moves and where? 4.2. The status of the clitic & the status of the double the status of double: 1. Unilateral dependence of CD on DOM: CD is only possible with direct object DPs that have been differentially object marked vs. CLRD which allows all types of accusative DPs: (24) a. Mihai l-a ajutat pe copil/*copilul (CD) Mihai him.cl-has helped pe child/clid.the ‘Mihai helped the child.’ b. Mihai l-a ajutat, #pe copil/copilul (CLRD) Mihai him.cl-has helped pe child/clid.the ‘Mihai helped the child.’ 7 2. The CD-ed DP may serve as the antecedent of an anaphor and anaphors must be bound from Apositions according to Principle C. (25) Îi ştiam pe copiii supăraţi pe ei înşişii pentru că Them.cl. knew.I PE children upset PE them themselves because pierduseră meciul. had lost.they game.the. ‘I knew that the children were upset on themselves because they had lost the game.’ (26) La concursul de matematică am văzut copii mulţumiţi de ei înşişi, At contest of Mathematics have.I seen children satisfied with them themselves dar şi copii supăraţi de performanţa lor. but also children upset of performance their. ‘At the contest on Mathematics I saw children that were pleased with themselves and children that were upset about what they had done.’ 3. Small clauses: the DP double may function as the subject of a small clause and these subject DPs are known to occupy argument positions. (27) O consider [SC pe Maria fată deşteaptă]. Her.cl. consider.I PE Mary girl smart. ‘I consider Mary a smart girl.’ the status of the clitic 1. Rigid Order: fixed order of pronominal clitics in the cluster Cornilescu (2006), Săvescu (2009) (28) a. Alexandru mil trimite astăzi Alexandru CL.DAT CL.ACC sends today 'Alexandru sends it to me today' b. * Alexandru îlmi trimite astazi Alexandru CL.ACC CL.DAT sends today 2. Syntactic Unit with the Host > clitics end up forming a unit; the clitic may no longer be stranded from its host when it (further) moves in syntax (29) a. Maria a citit cartea cu uşurinţă. Maria has read book.the easily. ‘Maria easily read the book.’ b. Maria a citit-o cu uşurinţă. Maria has read it.cl.3.sg.fem easily ‘Maria easily read it.’ c. A citi- o Maria cu uşurinţă sau nu? Has read-it.cl.3sg.fem Mary easily or not ‘Has he/she read it easily or not?’ 8 3. the clitic does not always seem to saturate the internal argument position of the verb by itself. (30) a. (L-)ai văzut pe Ion? (Him.cl-) have.you seen pe John ‘Have you seen John?’ b. L-am văzut. Him.cl-have.I seen. ‘I have seen him’ (31) a. Pe cine ai văzut? Pe who have.you seen ‘Whom did you see?’ b. (L-)am văzut pe Ion. Him.cl-have.I seen pe John. ‘I have seen John’ c. *L-am văzut. Him.cl-have.I seen. ‘I have seen him’ the clitic and the double are licensed as one event participant > “the same argument”, saturating one position in the theta structure of the predicate (Delfitto 2002, Aoun 1999) A. Cl+pe DPs start out as a Big DP 1. the clitic and the double represent “the same argument”, saturating one position in the theta structure of the predicate (Uriagereka (1995), Torrego (1995,1998), Fisher (2000) Boeckx (2001), a.o.) 2. the selectional restrictions that the clitic imposes on its double: - only complete pe DPs may be doubled by clitic which is complete>> this is why Bare Quantifiers are excluded from doubling : (32) Nu *(l)-am văzut Not *(him.cl)-have.I see ‘I haven’t seen anybody.’ pe nimeni. pe nobody. - negative determiners nici un/nici o, which are complete are perfect candidates for clitic doubling: (33) Nu l-am văzut Not *(him.cl)-have.I see ‘I haven’t seen any one.’ pe nici unul. pe none. 9 B. the BigDP is stranded from the base-generated position 4.3 The status of the differential object marker Romanian has been analysed as a well behaved case of Kayne’s generalization: the clitic absorbs Case and thus CD is only possible in those languages that have special prepositions which can license case on the double Kayne’s generalization has been shown to encounter difficulties: o Balkan languages like (Greek, Albanian, Bulgarian) in which a bare Accusative was clitic doubled. o Suner 1988 > Porteno/Rioplatense Spanish: DO clitic doubling is possible without a o pe also occurrs in the absence of clitic doubling > all CD implied the use of pe, but not the other way round. ▼ In line with Suner 1988: pe is not a case marker What does pe do then? pe may have a semantic role. o Niculescu (1965) it has been observed that pe marking is sensitive to person denoting DPs: obligatory with person denoting proper names and personal pronouns and optional with person denoting nouns > this is why it has been interpreted as a mark of personal gender (= semantic gender, which distinguishes between non-neuter gender i.e., personal gender and neuter gender i.e., non-personal gender) (Cornilescu 2001). Bare quantifiers (34) a. nimeni vs. nothing N-am văzut pe nimeni. vs. Not-have.I seen pe nobody ‘I haven’t seen anybody’ N-am văzut (*pe) nimic. Not-have.I seen (*pe) nothing ‘I haven’t seen anything’ b. cineva vs. ceva Am văzut pe cineva. vs. Am văzut (*pe) ceva. Have.I seen pe somebody Have.I seen (*pe) something ‘I have seen somebody’ ‘I have seen something.’ c. oricine vs. orice Am văzut pe oricine. vs. Am văzut (*pe) orice. Have.I seen pe anybody Have.I seen (*pe) anything ‘I have seen anybody’ ‘I have seen anything.’ Upgrading to person status 10 (35) Iau pe rață și fac așa (child language Avram & Coene 2006) Take.I pe duck and do.I like this. ‘I take a duck and do like this’ o Bleam (1999), Cornilescu (2001): pe acts as a filter on the DP denotation: direct mapping between pe-marking and the semantic type of the DP: pe DPs may only have the argumental, object-level denotations <e> (entity, object) and <<et>t>> (generalized quantifier), while they lack the predicative, property denotation <et>. pe marking obligatory with proper names and pronouns: (36) L-am văzut *(pe) Matei/*(pe) el.. Him-have.I seen pe Matthew/pe him ‘I have seen Matthew/him.’ the individual-level reading of the verb a avea (have) (37) I have a car. a. I own a car (individual-level) b. I have a car (with me today) (stage level) (Bleam 2005) (38) a. Am mașină/*mașina. Have.I car/ car.the. ‘I own a car’ b. Am mașina (cu mine). Have.I car.the (with me) ‘I have the car (with me)’ (39) a. Maria Mary Individual-level Stage-level are (*pe) o soră. has (*pe) a sister. b. Am pe o soră (de-a lui Mihai la mine luna asta). Have.I pe a sister (of Michael at me month this) ‘I have one of Michael’s sisters living with me this month.’ Non-incorporating verbs love/hate (40) *Ion John urăște profesori. hates teachers. (41) Ion urăște profesorii/ pe profesori.. John hates teachers.the/ pe profesori ‘John hates teachers.’ o Tigău (2014 ms.) pe related to a partitive reading: (42) a. Șoferul poate duce câte trei profesori o dată. Mașina e cam mică. Driver.the can drive câte three teachers once. Car.the is pretty small ‘The drive can drive home three teachers at a time. The car is pretty small.’ b. Șoferul poate duce pe câte trei profesori o dată. Mașina e cam mică. 11 Driver.the can drive pe câte three teachers once. Car.the is pretty small ‘The drive can drive home three teachers at a time. The car is pretty small.’ 5. Integrating the data into an analysis: Elements to account for: evidence for movement case assignment: how does the clitic and the associate DP get case? Is pe a case assigner? Is the attested movement related to case assignment? unilateral dependence between the clitic and the associate DP 5.1 The Stranding Derivation3: (Boeckx 2001, Cornilescu 2002 a.o.) > the clitic is attracted to T, checking the Acc feature of the verb, the doubled is stranded below > since clitics are deficient pronominal arguments without an articulated internal structure they may not be interpreted as arguments if they are not linked to some substantive root (T(ense)) at the LF interface (Rouveret & Nash 2002). Thus, the clitic (which is merely a bundle of features) will have to move from the lexical domain onto a functional head so as to become interpretable: clitics are interpretable and syntactically active only when they reach Tense. > clitic movement is initiated only when T merges (under the assumption that movement only takes place when the head driving it has merged). However, the clitic will not move directly to T, but will first land together with its double (as an XP), into the closest potential landing site SpecvP, an intermediate position where Acc case checking takes place. The clitic moves to T as an X 0 , checking the case feature of v and leaving the double behind. The double does not have to move further up as it is -inert. Furthermore, it has been granted case by the preposition. > Thus, the clitic moves in two steps, as an XP to Spec vP and as an X0 to T, the double will be stranded out of the vP >> CD is thus a means of scoping the DO out if its vP internal position, a movement which is in agreement with the strong readings of the CD-ed constituents. 5.2 Cornilescu (2014 ms.) assumes that the clitic and its double are licensed as one event participant proposes a slightly different analysis which regards both the clitic and the double as phrases licensed in a small clause configuration headed by an empty D, with the clitic as a specifier and the associate as a complement (Uriagereka 2005, Krivochen 2014) (43) [DP DPcl [D] DPass] The BigDP is assigned a theta role in the a-structure of the verb. The clitic is left with unvalued case and person features, which it values through the derivation. o To check their unvalued person feature clitics move as phrases up to a Person field above Tense (Bianchi 2006, Savescu 2009) The Standing derivations is based on the following assumptions derived from Chomsky (1998, 1999): Case & Agreement represent the reflex of a single operation which a DP may undergo only once. The implementation of the Case-Agree mechanism amounts to valuing u features through Agree between an agreeing head (T or v) and a DP; the process of valuing the u features of the verbal head concomitantly values the K(ase) feature of D. Tense values the Nom feature of a DP, while v values the Acc feature. 3 12 o clitics value their case on the way to the Person field, they systematically pass through structural case positions. 5.3 Steps towards a proposal Basic tenets 1. the BigDP moves out of vP, the clitic aiming for a suitable projection inside the Person field in line with Bianchi (2006), Săvescu (2009, 2011) and in agreement with recent developments viewing grammar as an event-speech matching computational device Sigursson (2012)) 2. movement takes place in two steps: there is first movement of the BigDP to the edge of vP followed by further movement of the clitic to the Person field 3. In terms of current phasal locality principles (see Richards, 2004 a.o.), the internal argument bound by the clitic must be on the edge of the lower vP, so as to be still available for computation after the spell-out of the vP. 4. in agreement with recent developments in case theory, case is not a driving force, it is not responsible for NP movement and it is licensed vP internally. In agreement with Sigursson (2012) structural cases are seen as ‘indices’ assigned to roles vP-internally, c, thereby making the roles visible to relations with elements outside of vP. Extending on the basic tenets of the proposal > Clitic movement to PersP and the Person field a. A first argument that clitics target Agreement comes from the realm of prepositional supine clauses which do not allow obligatorily clitic doubled internal arguments in their structure, on account of their lacking an Agreement projection which might host the clitic (see above) >> we have thus a clear connection between Agr and clitic placement. This seems intuitively correct if we consider that clitics are fully specified for Person and Number b. it has also been shown above (inverse binding, parasitic gaps, focus projection) that the clitic and its double undergo movement out of the vP c. another argument which strengthens the hypothesis that Person is a driving force is the case of quirky subjects in Icelandic which seem to interfere with the matching of Person (but not with nominative case): (44) a. Honum mundu alltaf líka Þeir him.DAT would.3PL always like they.NOM ‘He would always like them.’ b. *Honum mundu alltaf líka Þi him.DAT would.2PL always like you.NOM.PL c. *Honum mundum alltaf líka vi him.DAT would.1PL always like we.NOM Sigursson (2007: 123) - as pointed out by Sigursson, first and second person angreement with the nominative object is ungrammatical (b,c), while third person nominative triggers plural agreement with the verb. Sigursson argues that this type of examples suggest that the Infl complex of the verb splits into (at least) Person, Number and Tense. This structure would account for the phenomenon in (44): the dative DP matches Person, moving to its immediate c-command/feature matching domain, allowing the nominative object to match Number. 13 (45) [CP C ….[IP Pers [Num DAT, Num….[vP……DAT…..NOM……… - thus, Person agrees with the dative DP and Number agrees with the nominative DP. Notice that Person may not agree with the nominative object as it is already engaged in a matching relationship with the dative subject. This is in fact what leads to the ungrammaticality of (b, c) > as it seems, Person is a driving force while structural case is not > it seems thus natural to argue that the movement of the clitic targets the AgrP (PersonP, more specifically) with the aim of checking an uniterpretable Person feature >This is in line with Bianchi (2006), Săvescu (2009, 2011) a.o. who convincingly show that clitics head for specific projections in a Person field which is part of the functional structure of the clause. - Săvescu (2009, 2011)’s account, the Person field is comprised of several Person projections, each specified for a particular person feature. -Thus, each Person projection only attracts the clitic with the right person specification. Agree takes place between features that have the same value: the probe Person1 will thus see only the clitic argument that has 1st person specification etc - The hierarchy of projections in the Person field is arrived at after testing various word order combinations and looks like (16) below: (46) Person1P>Person3P>Person2P>Refl3P>Person3P>TP>K-dat>K-acc>..>V This ordering accounts for several person restrictions in Romanian with respect to clitic ordering: a) 1stP & 2ndP clitic clusters are fine as long as the 1stP clitic comes first (47) a. Te cunosc, Ion mi te-a prezentat deja. You.Acc know.I, John me.Dat you.Acc-has introduced already. ‘I know you, John has already introduced you to me.’ b. Nu mă cunoști? *Ion ți m-a prezentat ieri. Not me.Acc know.you? * John you.Dat me.Acc-has introduced yesterday. ‘Don’t you know me? John has already introduced me to you yesterday.’ b) there are two 3rdP projections accounting for two possible clitic orders: - 3rdP clitics may target a projection above the position hosting 2ndP clitics (48) I te-au prezentat. 3P.Dat 2Psg.Acc-have.they presented. ‘They presented you to him/her.’ - two 3rdP clitics may freely combine (the order has to be Dat Clitic>Acc Clitic) (49) I l-au prezentat. 3P.Dat 3Psg.Acc-have.they presented. ‘They presented him to him/her.’ Very importantly, the low 3rdP clitic position may not host a 3rdP clitic in the dative, it seems to be specialised with respect to case: in example (20) below the dative 3rdP clitic may not combine with the 2ndP accusative clitic: (50) * Maria te i-a ajutat. Mary 2P.Acc 3P.Dat-has helped 14 - Notice also that Săvescu’s account presupposes the existence of a K(ase) field: our approach only shares with hers the idea that a Person field is part of the functional structure of the clause; case is solved vP internally in line with Sigursson (2007, 2012a,b)4. 3.2.2. Why would a PersonP be so important? > Sigursson (2007) points out the Person is, just like Tense, a basic computational element which computes or interprets features of the propositional event in relation to features of the speech event. (51) EventComputation/GrammarSpeech >There is thus a clearcut correspondence between grammatical features (Mood, Tense, features) and the features of the speech event (time/location of speech, speech participants) since grammatical features compute event features in relation to speech features. - just as Tense computes the Event Time in relation to Speech Time, in the same way Person (=grammatical participant features) relate Event Participants with Speech Participants (52) EPPSP - event participants will thus have to match Person 5 which will interpret them in relation to the speech participants - thus, all arguments enter this kind of matching and a clause may have objectPerson (hosting clitics) in addition to SubjectPerson 3.2.3. Why isn’t case a movement triggering force? > as we have seen in the case of the examples from Icelandic, case is not a causal factor in the DP-Infl Connection i.e., the so-called nexus problem about how subjects relate to their predicate (Jespersen 1924 a.o.), Person is (Person has, to some extent, the status that has been commonly attributed to case). > a natural question would then be: why isn’t case a movement triggering force and what does it do exactly. > the basic reason behind the idea that structural case is a movement driving force has to do with the fact that it is considered uninterpretable (it does not alter the interpretation of nouns). As such, it must delete under feature matching/agree between a head (v or T) and a DP, otherwise the derivation will crash since LF cannot interpret the uninterpretable feature. In case the head has an EPP feature, there is movement of the DP to the respective head. > Sigursson observes that structural cases are interpretable in as much as they have relative interpretability i.e., they are meaningful in relation to each other6. As such, structural cases may be envisaged as features which distinguish between event participants at a vP-internal level. Thus, event participants are cased roles, C: (53) Nominative: 0Case=event participant1 Accusative: 1Case=event participant2 Sigursson (2007:131) 4 When it comes to solving matters concerning the hierarchy of two clitics the entire responsibility will be transferred to the nature of the Person projections which specified for a particular person feature (including the two 3rdP projections) 5 According to Sigursson the Speech Event is syntactic and inside clausal structure; any utterance is a Speech Phrase which contains elements of the speech event and which dominates the clause (p. 137) 6 Sigursson (2003) speaks about what he calls the Sibling Correlation, showing that the structural accusative is preconditioned by structural nominative. It is in this sense that structural cases are said to be interpretable, through the meaningful relation that holds between them. 15 - thus structural cases are assigned vP-internally to roles, making them visible to relations with elements outside the vP. Under this account, it becomes clear why structural cases are not movement driving forces. 3.2.4. How is then case assigned? As seen above, Case is vP-internally interpretable: it will therefore be analysed in terms of event licensing and variable PF marking of the licensing relation (Sigursson 2007, 2012a, 2012b) > Chomsky (2001) claims that in regular Acc systems, Acc is the responsibility of phi-complete v, designated as v*. Case-star augmentation yields v* in addition to plain v. Case star augmentation is involved when a first marked case emerges. Defective v cannot assign any case value, thereby rendering the underlying object in defective VPs accessible to the Nom case. (54) a. b. c. We sank them They were sunk. They sank. (55) … [TP….. Voice/Ag….[NP2 transitive passive unaccusative v*-V v-V v-V Nom- Acc Nom Nom v*--V NP1 | | ----------- > Case star augmentation, then yielding v* in addition to plain v, is involved when a language develops from a no case language to an accusative language, distinguishing AccNPs from non-cased (Nom) NPs. > Active Acc assignment in the v-system is based on the structure in (26), whereas the corresponding unaccusative and passive structures yield Nom, as sketched in (27); the arrows connecting v*-V and NP1 indicate a matching relation that gets interpreted in terms of non-Nominative case in PF, here Acc. (56) …[TP ….VoiceAG…[NP2….v*-V….NP1…]] (57) a. b. NP1/ACC in PF NP2/ NOM in PF …[TP ….VoicePASS…[ v-V….NP1…]] NP1/NOM in PF …[TP ….VoiceEXPL…[ v-V….NP1…]] NP1/NOMin PF > We assume that all predicates are embedded under some Voice head, passives being embedded under voice passives, while unaccusatives and anticausatives are embedded under expletive VoiceExpl (even though unaccusatives & anticausatives have different vP internal structures). > Relating syntax and PF, the decision for a non-Nominative N can be taken in three steps. First syntax transfers to PF a structure of type (26), where nothing but licensing matching relations are indicated: Voice(NP) v/V ( NP]. Language specific PF properties assign a particular type of case star to one or more than one licensing heads, depending on the particular syntactic structure. This yields the right part of (26). Thirdly, PF interprets the particular case star as a particular morphologic case of the language. > Arguments thus are event licensed by specialized heads: agentive or active subjects by Voice/AG, direct objects by v-V. 3.2.5. > Deriving cl+pe DP sentences Unlike current minimalist analyses (Chomsky 2000, 2001) which assume that structural case checking is the by-product of agreement in the complete set of phi-features between a probe and a goal, we assume that case is checked independently of person and that case does not trigger movement - Person triggers movement of the clitic which needs to check its Person feature - On its way to the Person field the clitic travels to an intermediate landing site together with its double (as a BigDP) 16 - (58) Case is assigned vP internally and amounts to event licensing: the BigDP is assigned accusative by v*-V and there is agreement between the clitic and the double. As such the clitic and the double are licensed as one event participant. [Person…..[T[vP DPSu [v’ v[VP V, BigDP]] [Person…..[T[vP BigDP [vP DPSu [v’ v[VP V, BigDP]] [PersonP cl, P[T[vP DPobj [vP DPSu [v’ v[VP V, BigDP]] References: Alexiadou, Artemis (1999). On the Properties of Some Greek Word Order Patterns . In: Studies in Greek Syntax . Artemis Alexiadou, Geoffrey Horrocks, and Melita Stavrou(eds.), 46 55. Dordrecht: Kluwer Aoun, J. (1999). ‘Clitic Doubled Arguments’. In Johnson, K. and I. Roberts (eds.). Beyond Principles and Parameters. Dordrecht: Kluwer Academic Publishers. Bianchi, V. 2006.On the syntax of personal arguments. Lingua 116:2023–2067. Boeckx, C. 2001. Mechanisms of Chain Formation PhD dissertation, University of Connecticut Chomsky, N. 2001. Derivation by phase. In Michael Kenstowicz (ed.), Ken Hale: A life in language, 1–52. Cambridge, MA: MIT Press Cornilescu, A. 2001. ‘On the Interpretation of the Prepositional Accusative in Romanian’, in Bucharest Working Papers in Linguistics Cornilescu, A. 2002.Clitic Doubling and Parasitic Gaps in Romanian, paper presented at Going Romance November 2002, Groningen Cornilescu, A./Cosma, R. to appear. On the syntax of the Romanian supine: the de-constructions. In R. Cosma/S. Engelberg et al. (eds.), Komplexe Argumentstrukturen. Kontrastive Untersuchungen zum Deutschen, Rumänischen und Englischen. Berlin: Akademie-Verlag. Delfitto, D. 2002. On the Semantics of Pronominal Clitics and some of its Consequences. Catalan Journal of Linguistics 1: 41-69 Dobrovie-Sorin Carmen. 1994. The Syntax of Romanian. Berlin: Mouton, de Gruyter Fisher, S. 2000. Diachronic perspective of the Catalan clitic system and other aspects of Catalan Syntax, PhD dissertation, University of Potsdam. Kayne, Richard.1989a. Facets of Romance Past Participle Agreement . In: Dialect Variation and the Theory of Grammar . Paola Benincà(Ed.), 85 104. Dordrecht: Foris . Kayne, Richard.1989b. Null Subjects and Clitic Climbing . In: The Null Subject Parameter . Osvaldo Jaeggli and Kenneth Safir (eds.), 239 261. Dordrecht: Kluwer . Kayne, R.1991. Romance Clitics, Verb Movement, and PRO . Linguistic Inquiry (22) : 647=686. Kayne, R.1994. The Antisymmetry of Syntax . Cambridge: MIT Press . Monachesi, P. 1995. A Grammar of Italian Clitics.PhD dissertation.Tilburg University. Richards M. D. 2004. Object Shift and Scrambling in North and west germanic: A Case Study in Symmetrical Syntax’, doctoral dissertation, University of Cambridge Rivero, María Luisa (1994). Clause Structure and V-Movement in the Languages of the Balkans . Natural Language and Linguistic Theory (12) : 63 120. Rivero, María Luisa and Arhonto Terzi (1995). Imperatives, V-Movement, and Logical Mood . Journal of Linguistics (31) : 301 332. Rouveret A. and Nash L. 2002.Cliticization as Unselective Attract, Ms. University of Paris 7 Săvescu, O. 2009.A Syntactic Analysis of Pronominal Clitic Clusters in Romance: The View from Romanian.PhD Dissertation.University of New York. Săvescu, O. 2011. A cartographic approach to clitic clusters in Romanian. Revue Roumaine de Linguistique.LVI,2, p. 97–114, Bucureşti Sigurðsson, H. Á. 2003. Case: abstract vs. morphological. In New Perspectives on Case Theory, ed. Ellen Brandner and Heike Zinsmeister, 223–268. Stanford: CSLI Publications. Sigurðsson, H. Á. 2007. Argument features, clausal structure and the computation. In Argument Structure, ed. by Tanmoy Bhattacharya, Eric Reuland, and Giorgos Spathas, 121–158. Amsterdam: John Benjamins. Sigurðsson, H. Á. 2012b. Case variation: viruses and star wars. Nordic Journal of Linguistics.35(3): 313-342. (http://ling.auf.net/lingbuzz/001458) Sigurðsson, H. Á. 2012a. Minimalist C/case. Linguistic Inquiry 43, 191–227. 17 Suñer, Margarita (1988). The Role of Agreement in Clitic-Doubled Constructions . Natural Language and Linguistic Theory (6) : 391 434. Terzi, Arhonto (1996). The Linear Correspondence Axiom and the Adjunction Site of Clitics . In: Configurations . Anna-Maria di Sciullo(Ed.), 185 199. Los Angeles: Cascadilla Press . Tigău, Alina. 2010.Syntax and Interpretation of the Direct Object in Romance and Germanic Languages with an Emphasis on Romanian, German, Dutch and English. Bucharest: Editura Universităţii Bucureşti. Torrego, E. 1999. Object Dependencies. Cambridge, MA: MIT Press. Torrego, E. 1995.On the nature of clitic doubling. In Hector Campos & Paula Kempchinsky (eds.), Evolution and Revolution in Linguistic Theory. 251-271. Washington, D.C.: Georgetown University Press. Uriagereka, J. 1995. Aspects of the Syntax of Clitic Placement in Western Romance.Linguistic Inquiry 26 (1): 79-123. Zanuttini, Raffaella (1991). Syntactic Properties of Sentential Negation: A Comparative Study of Romance Languages. PhD dissertation, Philadelphia, University of Pennsylvania. Zanuttini, Raffaella (1997). Negation and Clausal Structure: A Comparative Study of Romance Languages . Oxford: Oxford University Press . Zubizarreta, María Luisa (1998). Prosody, Focus, and Word Order . Cambridge: MIT Press . Alina-Mihaela Tigău Alina.Tigau@unige.ch 18