Ditransitive verbs in Czech (and elsewhere)

Background. The internal structure of ditransitive verbs like give or send has been given a lot of attention in the generative literature. Two main approaches to these verbs in English can be distinguished: in the first approach, the double object construction (DOC) (1-a) is derivationally related to the structure with a dative PP (1-b) by an NP-movement accompanied by “dative-case absorption” (Larson 1988) or Preposition Incorporation (Baker 1997). On the other hand, Harley (2002) posits two different underlying structures for the DOC versus the NP-PP construction. While the DOC is characterized by embedding (under the little v) a possessive small clause/PP which takes a goal as its subject and a theme as its object, the latter structure embeds a PP with a theme as its specifier and a goal as its complement. Thesis. 1. Czech ditransitive verbs belong, in contrast to English, to two non-overlapping classes exemplified in (2-a) and (2-b): Dative-Accusative verbs and Accusative-Dative verbs, both of which are base-generated. While the former class is to be analyzed on a par with Harley’s possessive small clause approach (3-b), the latter class consists of verbs that have a PP complement with a null head but with marked morphological case (3-a); see McFadden (2004) for a similar analysis of Acc-Dat verbs in German. 2. Contra McFadden (2004) and Miyagawa and Tsuijoka (2004) who posit a unique structural position for an indirect object of Dat-Acc verbs, corresponding to Spec,vAppl of Marantz (1993), I argue for two high dative positions: a possessor (in a broad sense) of a direct object as in (2-b), introduced below V in Spec,PPoss, and a beneficient of the whole event as in (2-c) which is introduced in Spec,vApplP; in this way my analysis follows Pylkkänen’s 2002 distinction between high versus low applicatives. 3. In order to account for the possibility of flexible word order of Czech ditransitives exemplified in (2), I employ the concept of G(ivenness)-movement introduced in Kučerová (2007) which correlates with a requirement that Given (presupposed) elements linearly precede elements that are new (non-presupposed) in the discourse. I show that the structural distinction between Acc-Dat and Dat-Acc verbs is reflected under G-movement exactly as predicted. (Some) Empirical Generalizations. 1. Both word orders, Acc≫Dat as well as Dat≫Acc, are possible in Czech for all ditransitives as shown in (2). However, in neutral context only the Acc≫Dat word order is unmarked for the verbs like (2-a), while the Dat≫Acc word order is unmarked for (2-b) and (2-c) verbs. 2. Also constituent fronting under topicalization splits ditransitive verbs into two classes: those treating a verb and a theme as one constituent and those treating a verb and a goal as one constituent. 3. Ditransitives preferring Dat≫Acc word order in neutral context and [Verb Theme] fronting under topicalization are grammatical even if their indirect object is missing. If a dative argument corresponds to a possessor/recipient of the theme, it is existentially quantified, i.e. interpreted as “somebody”, see (4-a). On the other hand, verbs with an applied argument which benefits from the whole event described by a verb plus direct object do not have such entailment if a dative DP is not present overtly, see (4-b). 4. In contrast, verbs preferring Acc≫Dat word order in neutral context and [Verb Goal] fronting under topicalization cannot appear without their dative argument. They become either ungrammatical as in (4-c), or they receive a different, often nonsensical interpretation. 5. Two indirect objects in non-prepositional dative cannot co-occur (5-a). (Two datives with a single verb are grammatical only if (at least) one of them is assigned by an overt (non-null) preposition (5-b).) Theory vs. Generalizations. If Czech ditransitive verbs have one of the two base-generated structures in (3), distinguished by a type of a complement selected by the verb, word order preference in neutral context as well as under topicalization follow naturally. The semantic difference between a possessor of an internal argument and a benefactive event participant – as well as the tacit presence of a possessor when a dative DP is not present overtly as in (4-a) – is accounted for if only a possessor represents a true verbal argument while the beneficient has an optional (adjunct) character in Czech and is introduced above VP as proposed above. My analysis thus extends McFadden (2004) in that there are three positions in the verbal structure associated with dative Case: the null P[Dat], the Specifier of PPOSS and the Spec of vAppl.