The Lithuanian dative of possession (ex. 1) is a particular manifestation of the cross-linguistically robust grammatical phenomenon known as ‘external possessor’ (EP; Payne & Barshi 1999), which poses a challenge for grammatical theory. Syntactically, the dative is fully integrated into the clause structure, as if it was a regular argument of the verb, and yet, it is not lexically projected by that verb. Semantically, it is construed as the possessor of another constituent within the same clause.
Existing descriptions of the Lithuanian EP remain sketchy (Grabauskas 1971, Ambrazas et al. 1985, Sawicki 1996) and yet, its grammatical properties and distribution raise issues of general interest. The functionally oriented analysis presented in this paper is motivated by two general questions: (i) what kind of linguistic knowledge is presupposed for native-like production and comprehension of EP sentences and (ii) how does the Lithuanian pattern correspond to comparable constructions in other European languages and particularly Slavic.
I show that an adequate description of the EP requires reference to several layers of information: structural limits on the possessum, verb meaning, place of the possessum on the possessive hierarchy, and affectedness of the possessor. Empirical evidence leads to the conclusion that the EP cannot be treated merely as a syntactic variant of genitive-marked possessors in (2) (a ‘possessor-raising’ analysis); each form represents a semantically and pragmatically distinct pattern, as has been shown for many other languages (e.g. O’Connor 1994, Croft 1985, Berman 1982). Nor is the EP a special interpretation of other datives (recipient, beneficiary, etc.); instead, it represents a grammatical entity in its own right, shaped by the interaction between several factors (syntactic, semantic, pragmatic).
The data highlight the need for a framework that can accommodate instances in which sentential structure is not just a projection of the head predicate (standard assumption in generative accounts) but may integrate features contributed by a variety of sources (syntactic clause type, argument structure of the verb, inherent semantic properties of nominals, and pragmatic conditions). ). I argue that such associations are best captured through the notion of CONSTRUCTION, defined as a conventional association between morphosyntactic properties and meaning/ function (Fillmore 1989, Croft 2001, Fried & Östman 2004).
The analysis is also relevant to the study of the European areal prototype of EPs proposed by K[nig & Haspelmath 1997. The Lithuanian pattern ranks very low both on the Animacy Hierarchy (unlike most European languages, including Slavic) and the Inalienability Hierarchy (like Slavic). In particular, the scope of ‘inalienability’ in Lithuanian appears to include anything that is in a conventionally highly salient part/whole relationship; as a result inanimate ‘possessors’ are possible, thus presenting a conspicuous departure from the prototype.
(1) Pavoge jam mašiną. ‘They stole hisi car on himi.’
they.stole he.DAT.SG car.ACC.SG
(2) Pavoge jo mašiną. ‘They stole his car.’
they.stole his car.ACC.S