This article documents and analyzes a pattern of backward subject control in the Nakh-Daghestanian language Tsez. In backward control two subject arguments are coindexed but it is the higher subject that is unpronounced: ⌬ i tried [John i to leave]. The principles-and-parameters framework (Chomsky and Lasnik 1993) explicitly rules out backward control. In contrast, recent minimalist analyses of control (e.g., Hornstein 1999) permit backward control because they allow movement from one thematic position to another. Backward control results if this movement takes place covertly. We argue that the phenomenon thus provides interesting evidence for the reduction of control to movement.

. This paper presents unusual patterns in raising and control and offers a syntactic account which would validate such patterns. On the empirical side, we present evidence for backward control (data from several languages), backward raising (in Adyghe), copy control (Zapotec, Assamese, Tongan) and copy raising (data from a number of languages). All the unusual cases of raising and control seem to involve an A‐movement chain in which the lower, not the higher copy is pronounced, or both copies are spelled out. We also review some constructions which, while superficially resembling backward or copy structure, do not provide any evidence for a movement chain. On the theoretical side, assuming that the more unusual patterns are an empirical reality, linguistic theory should be capable of analyzing them. We present mechanisms from the current Minimalist Program which we believe allow the attested variation.

No abstract

The Indonesian verbs mau and ingin 'want' look like typical control verbs. When they are followed by a passive predicate however, an additional, unexpected interpretation arises. The sentence Siti mau/ingin di-cium oleh Ali means 'Siti wants to be kissed by Ali' but also 'Ali wants to kiss Siti'. We call the latter interpretation Crossed Control (CC). In CC, the wanter is not the surface subject of 'want' but an oblique element in the complement clause and the surface subject is the theme of the embedded predicate and not an argument of 'want'. For the syntax of CC, we reject clause union and backward control analyses and propose that 'want' in this construction is an auxiliary/raising verb that does not assign an external u-role. We then propose that the control interpretation is encoded in the lexical semantics of the auxiliary. 'Want' takes a propositional argument but forces the volitional participant in this event to be construed as an experiencer of wanting. We hypothesize that this approach can be extended to volitional constructions in other languages. #

No abstract