Yânnis aghapâi ti MARIA
4.2 The Clitic-Construction
4.2.1 Clitic-Left-Dislocation
(7), (8) and (9) are examples of CLLD. In MG, the DP in CLLD can be a direct object (example (7)), an indirect object (example (8)) or a measure phrase (example (9)).
(7) ta luludhia *(ta) éfere o Vassflis the flowers-Acc them-cl brought-he the Vassilis-Nom 'Vassilis brought the flowers.'
(8) tis Marfas *(tis) stflane luludhia the Maria-Gen her-cl sent-they flowers-Acc They sent Maria flowers.'
(9) ekatô kilâ tha (ta) ziyfzi a hundred kilos-Acc will them-cl weigh-he
He must weigh a hundred kilos.'
As the stars in examples (7) and (8) indicate, the sentences become ungrammatical, if the clitic is omitted. Interestingly, when the DP in CLLD is a measure phrase the clitic appears to be optional (cf. (9)). The generalization is that the clitic is obligatory when a specific object, whatever that means, precedes the verb. With respect to example (9), it seems that measure phrases can be read as either specific or nonspecific DP's.
Cinque (1990) has shown that the relation between the DP in CLLD and the thematic position it is associated with is sensitive to island constraints. This is illustrated below with examples from MG.
(10) *tu C6sta tha su müiso ja [ ta vivKa [ pu tu the Costas-Gen wiU to you talk-I about the books that him-Gen arésun]
please-they
'Costas, I will only talk to you about the books that he likes.'
(11) *eséna o Yôrghos me piokalôs mathitis apo 6ti su you-Gen the Yorghos is a better student than what you-cl-Gen fénete
seems
'*To you Yorghos is a better student than it seems.'
(12) *tu Yânni ton idhame prin na tu tilefonisume the Yannis-Gen him saw-we before that him-Gen phone-we '*Yannis we saw him before we phoned.'
Example (10) involves a complex NP island, example (11) a comparative clause island and example (12) an adverbial clause island.
The possible hnes of analysis for CLLD considered in the literature are two. One advocates movement of the object DP while the other advocates base- generation of the object DP in a clause-peripheral position. In the first line of analysis there are again two possibilities. According to the first possibility, CLLD is an instantiation of Clitic Doubling, where what is 'doubled' is a moved phrase. An alternative would be to argue that what is 'doubled' is the wh-trace. According to the second possibility the clitic is the overt spelling out of the pronominal features left on the variable. In the base generation analysis of CLLD the clitic is resumptive and is there to receive the theta role.
We need to identify the position hosting the object DP in CLLD. The position under examination cannot be the canonical object position, as MG is a head-first language.
Another important point is that although indirect objects in MG can be either DP's or PP's, only DP indirect objects can participate in CLLD. Consider example (13), below:
(13) sti Maria *(tis) stflane luludhia to the Maria her-cl sent-they flowers-Acc
The fact that the position occupied by the CLLDed object is reserved for DP's suggests two things. First, that the position involved is unlikely to be an adjoined position. As far as I know, adjoined positions do not differentiate between DP's and non-DP's. And second, that the position occupied by the object DP does not seem to be one of the known Specifier positions, i.e. Spec of CP or Spec of FP, for the same reason that it could not be an adjoined position.
The matching in phi-features between the object DP and the clitic reminds one of licensing mechanisms in the form of Spec-head agreement configurations. I will next present an analysis of CLLD which is in that spirit. Consider claim (14),below, and the tree in (15).
(14) CLLD is a Spec-head phenomenon. (15) ^ ^ C h d c g h r ^
DP^ 0)tic'
Clitic
DP*
I argue that in CLLD the object DP is in the Spec of a functional maximal projection, a CliticPhrase, of which the clitic is the overt head. The term CliticPhrase is Sportiche's (1992) (GLOW talk). The object DP occupies the Specifier of that projection as an instance of A'-movement from the canonical object position. The Spec of the CliticPhrase is an A'-position, being a non-Theta- position.
(16) CliticPhrase itic’ Clitic ta éfere o Vassflis 4.2,2 The Proposai
In section 4.2.1, claim (14) was advanced. Next I would like to generalise this claim to (17), below:
(17) CLC is a Spec-head phenomenon.
The change from (14) to (17) is not trivial because CLC also includes CD, where the object DP follows the matching clitic. I assume that the position of the DP in CD is still in the Spec of CIP, as in CLLD, and that there is an additional movement of the verb. See section 4.4.1 for a discussion of CD.
The reason why there is movement of the object DP to the Specifier of a CliticPhrase is to satisfy a Spec-head agreement configuration. The Spec-head agreement between an object DP and a matching clitic is a literal case of Spec-head agreement, given that the Spec and the head are identically marked for phi- features 1. The Clitic-Criterion in (18) forces movement of the object DP from its canonical position to the Spec of a CliticPhrase headed by a clitic matching the object DP in phi-features. I take the Clitic Criterion to have S-structure status only. Ijn other Spec-head agreement configurations (Rizzi’s WH-Criterion, syntactic focussing as a Spec-head agreement configuration) what we usually have is a Spec- head configuration and agreement with respect to some abstract feature.
(18) By S-structure: Clitic-Criterion In CLC:
A DP must be in a Spec-head configuration with a matching clitic. A clitic must be in a Spec-head configuration with a matching DP.
I say that the Clitic-Criterion applies by S-structure and not at S-structure because I take it that the Clitic-Criterion sometimes applies as early as D-structure. This occurs where the Spec of CIP is a theta position (See section 4.3.2.2 for data and a summary of potential theoretical problems.). If D-structure is a true representation of thematic stmcture, then we will have to say that when the Spec of CIP is a theta position the Theta Criterion is satisfied at D-structure by a chain. Or, alternatively, we drop the assumption that D-structure is a pure representation of thematic structure and we have the Clitic-Criterion applying at S-structure uniformly.
The question is raised what is the nature of the DP* when it is silent. Saying that it is pro would perhaps be all right for MG but would be too easy a way out for object clitics in French, for instance; as in French object clitics and object DP's are in complementary distribution. If I claimed that the silent DP* is pro in French, I would need either to answer the question why can't DP* be overt in French, or beg the question and assume that whether or not DP* can be overt is not of theoretical relevance. Both possibilities seem unpromising to me. The reader is referred to section 4.6 for an answer to the question why French does not have object CLC.