Keywords
3. Resultados
I propose that Dinka receive the same analysis as languages with anti-agreement, i.e. the agreement facts follow from the presence of an underlying bi-clausal structure. This implies that Dinka is not a V2 language in the technical sense (V-to-C movement in all main clauses), but simply exhibits second-position effects. Since anti-agreement always involves a fronted focussed constituent, I first provide an account for the syntax of sentences with fronted instrumental PPs, which, according to my data, are obligatory focussed (cf. section 2.1.3 above). I suggest that sentences with a fronted focussed PP have a bi-clausal structure, in which a CP takes a relative clause / small clause (SC) as
its complement. The fronted element is not moved to CP, instead it is merged in the complement position of the SC where it receives case via number agreement with a silent copula (cf. Ouhalla 1993): this is why it does not exhibit operator properties. In Somali, the copula can only assign the unmarked/default case, not the nominative, and predicative DPs have to bear the non-nominative case (Frascarelli 2010). I propose that the same happens in Dinka: fronted PPs lose the case they have in their base position and take the absolutive because the copula only assigns the absolutive (non-nominative) case. The subject of the SC is the Presupposition, which is realised as a free relative clause, headed by a generic NP-pro form. This is restricted to range over a limited set of denotations (person, thing, place, time) and is semantically matched with the focused PP. The interpretation of the predicate follows from movement to Spec,FocP, where it checks the strong [+foc] feature, cf. the structure of (39) in (37).35
(37)
35 Following Frascarelli (2010), I assume the structure proposed in Den Dikken and Singhapreecha (2004) and Den Dikken (2006) in which the relative DP takes an SC as a complement, having the NP-head merged in subject position and the CP as its predicate.
I suggest that the case and thematic role of the focussed PP appearing in the relativisation site are realised by a resumptive element in the clause.36 This very common relativisation strategy (cf. Bianchi 2002, Cinque 1990, 2013a, McCloskey 1990, Shlonsky 1992, Suñer 1998 a.o.) is illustrated in (38) with colloquial Italian. It can be seen that the head of the relative clause is doubled by a clitic realising case and thematic role.
(38) a. Il ragazzoj che glij ho dato il libro the boy that him have.1SG given the book
‘The boy whom I gave the book.’
b. La cordaj che cij ha tirato la vacca the rope that with-it haS pulled the cow ‘The rope with which I pulled the cow.’
I suggest that in Dinka the function of the pronominal resumptive element in (38) is carried out by the voice alternations, and the oblique voice found with PPs is derived from the incorporation of the PP pro-form into the verb morphology.37 A Dinka sentence with a fronted instrumental PP (39a) has the underlying structure shown in (39a).
(39) a. Yuêen à̰-cí dhɔɔk wḛ́ŋ miḭ̂it rope.ABS D-PRF:NS boy cow pull
36 For the idea that resumptive relativisation does not involve movement, cf. Chomsky (1977), Borer (1984), Safir (1986). The lack of weak-cross over effects in Dinka (cf. 36) follows from the presence of an underlying relative clause.
37 Van Urk (2014) proposes that what incorporates is a silent P. The idea proposed here that the pro-form incorporates is supported by Italian, where resumptive pronouns can move from the relativisation site, cf. Il ragazzoj che glij ho dato il libro tj
the boy that to him have-1SG given the book
“The boy whom I gave a book.”
‘With the rope the boy pulled the cow.’
b. It is the rope the thing which the boy pulled the cow with it
The fact that one of the objects must appear in the OV position with a fronted focussed PP I account for by assuming that objects must be the subjects of the SC because V° is associated with an EPP feature in Dinka (cf. vU&R 2015). The fact that only objects can satisfy this low EPP feature follows from specific (possibly phonological and prosodic) constraints of Dinka allowing only morphologically light (i.e. those unmarked for case and bearing the unmarked/default absolutive) constituents to be the subject of a SC.38 Locative PPs do not share the syntax of instrumental PPs, according to my informants.
My tentative explanation is that this depends on the absence of a locative pro from able to appear within the SC in the varieties under consideration, where the pro form is available for instrumental PPs only. PPs can be fronted and bear the absolutive in the Agar and the Nyarweng Bor sub-variety (cf. Andersen 1991, vU&R 2015): this indicates that a locative pro form is potentially available in other Dinka varieties.39 4.2.3 Structure of sentences with a fronted object
Sentences with focussed fronted objects must receive the same derivation as sentences with a fronted PP: the object is merged with a silent copula in the complement position of the SC and then moved to Spec,CP to check the focus feature. Unlike in sentences with a fronted PP, the OV position within the SC is not available for the non-fronted object because the relativisation site is blocked by the presence of the object resumptive
38 Svolacchia, Mereu and Puglielli (1995) show that a similar constraint is also found in Somali, a mixed OV/VO language in which the OV position is restricted to known / given arguments and to a series of morphologically light elements which can appear in the presupposition when a focus is fronted.
39 There is no mention of the discourse properties of fronted locative PPs in the literature. My analysis analysis predicts that they should be foci. This point remains open for further research.
pronoun in the SC which realises the thematic and the case features of the object and also satisfies the EPP feature on V°. The object pronoun incorporates into the verb morphology leading to the object voice.
I suggest that sentences in which the fronted object is a topic also involve a bi-clausal structure. Topicalised subjects appear in a copular structure that introduce the
presupposion taking a SC expressing the new information cf. (40).40 (40) a. wééŋ àò ku l- ɛéɛ m côò
cow.ABS D-steal.NS man.GEN
‘The man is stealing the cow.’
b. As for the cow <is> the thing that the man is stealing it The difference between (38) and (40) is both pragmatic and syntactic. Pragmatically, (40) involves a topic – comment articulation, and (38) involves a comment-topic articulation. Syntactically, this asymmetry follows from the fact that the fronted element is merged in the complement position of the SC (copula is pre-nominal) in (38), but not in (40), where the fronted object is assumed to be base-generated in Spec,TopicP as the head of the relative clause and to be doubled by a resumptive pronoun in the
relativisation site realising thematic role and case. The function of the (silent) post-nominal copula in (40) is not to realise a focus structure and to assign absolutive case to its complement, but simply to connect topic and comment. In (40), OV is assumed to be blocked by the mechanism also at play in (38), i.e. by the presence of a resumptive
40 The copular construction I assume for topicalised constituents in Dinka is the one of specificational sentences for the type “As for the best player, it/ *he is Paolo” (from Frascarelli 2010:2131,fn32). Cf. Den Dikken, 2006a; Moltmann, 2009 on this.
pronoun in Spec,VP which incorporates into the voice verb morphology. The structure of (40) is given in (41) below.
(41)
The proposed analysis implies that Dinka relies on a bi-clausal structure to realise both fronted topicalised and focussed objects, unlike languages with anti-agreeement where the bi-clausal structure is used as a focus strategy. I propose that this follows from the absence of TP in Dinka (CP selects for SC), but not in languages with anti-agreement (where TP is present and the nominative is in fact possible, cf. Frascarelli 2010).
That fronted objects (but not fronted PPs) can be both topicalised and focussed follows from the fact that they bear the absolutive case in all clause positions. This allows them to be base-generated in Spec,TopicP and to be the head of the relative clause – an option which is not available to PPs which can only be fronted when they are given case by the copula, i.e. when they are focussed. Since the two underlying structures realising fronted topicalised and focussed objects are syntactically very close, it has to be assumed that Dinka relies on intonation/prosody to indicate the pragmatic/discourse
properties of the fronted object. Therefore, the prediction to be tested in future research is that (38) and (40) involve different prosodic contours.