Egiturazko kasuaren erkaketa euskaraz. (Structural Case Checking in Basque)

The topic of this dissertation is the Ergativity in Basque. The author adopts the Chomsky" Minimalist Program framework (1993), and it exhibits three related empirical domains: the nature of subjects, the dative case and the Split Ergativity.

The ergative case system is different form the more familiar nominative/accusative system in that it exhibits different markings of transitive and intransitive subjects: transitive subjects are marked for the Ergative case and intransitive subjects for the Absolutive case, like transitive objects. Following Bobaljik (1993) and Laka (1993) (but see Murasugi (1992) for a different proposal). The author propose the theory that the Ergative is indeed the case of subjects in Basque, in contrast to the nominative/accusative subject which bears the Absolutive case.

The author differs from Bobaljik"s proposal concerning non-finite sentences. In control-type nonfinite sentences (Ortiz de Urbina, 1986), the intransitive subject has to be null, just the same as the transitive one. To account for this phenomenon, the author proposes that non-finite Tense checks an obligatorily activated null case. Therefore, both transitive and intransitive subjects are realized as PRO.

In the framework assumed, structural cases are checked in the Specifier of Agreement heads. This structural requirement is visible in Basque as an overt agreement for Ergative and Absolutive. Since the Inflection also grees with Datives, Basque Dative is analyzed as a structural case (cf. Cheng & Demirdache, 1993; López & Austin, 1995). This increases Chomsky"s (1993) crossing paths problem by adding a third A-movement path.

In the proposal defended by the author, this problem does not arise. The author proposes that the functional head licensing Absolutive is situated lower down in the structure than the position where the subject is base-generated; i.e., the Absolutive is checked in a functional Spec between VP-shells. This proposal has also been independently formulated by Collins & Thrainsson (1993).

Apart from the general pattern sketched above, Basque shows a phenomenon known as the Split Ergativity. Typologists (Dixon, 1994), which proposes the concept of a Split Ergativity if a language combines properties of ergative and nominative/accusative systems. Basque has two kinds of Split Ergativity. One is based on the semantic nature of the predicate, the other one on the semantic nature of the NP arguments.

The split based on predicate types is analyzed as ephiphenomenal: the apparently exceptional behavior of unergative predicates is believed to proceed from an underlyingly transitive representation in Lexical Relational Structure (Hale & Keyser, 1993). If the object is incorporated in the lexicon, the predicate behaves as an intransitive in syntax and its only syntactic argument is assigned Absolutive case. If the object does not incorporate in the lexicon, it is visible in syntax (incorporating at Logical Form), and the predicate behaves as a transitive licensing Ergative case.

Bound vs. free split (Dixon, 1994) concerns the semantic nature of NP arguments. In the past tense, first and second person subjects carrying Ergative agree with an Absolutive prefix in the auxiliary, if the object is a third person. Concerning these configurations, I propose that the Ergative argument does not check its features in the canonical way. Instead, it moves towards an Absolutive agreement and to an Ergative case position. The Absolutive argument only checks its case, but no agreement. I call this mechanism Split checking of nominal-features