Licensing of Instrumental Case in Hindi/Urdu Causatives

There is a long tradition in the generative literature of interpreting the presence of the by-phrase adjunct in the English passive as the signal of a ‘demoted argument’, in this case, the Agent of the corresponding active (Jaeggli 1986, Grimshaw 1990, Baker et al. 1989) or a Agent feature in the syntax (Embick 2004). In the case of Hindi/Urdu causative constructions, an instrumental se-marked adjunct is licensed with an ‘intermediate agent/causee’ interpretation in the indirect morphological causative using the suffix -vaa (Masica 1991, Saksena 1982b, Kachru 1980, Hook 1979), inviting comparisons with the demoted agent analysis of English by-phrases. In this paper, I revisit the licensing and interpretation of instrumental casemarked nominals in Hindi/Urdu causative constructions to argue against the hypothesis that the se-marked phrase corresponds to a demoted agent. Rather, I will argue that a more unified analysis of sephrases can be achieved through an event-structural analysis, in line with the standard interpretation of other adverbials in the syntax (cf. Ernst 2002 ). Since the ‘intermediate agent’ interpretation is only possible with indirect causatives in Hindi/Urdu, the event structural analysis proposed here also has implications for the direct vs. indirect causation distinction in the syntax. The paper is organised as follows. In the first section, I lay out the basic facts about direct and indirect causation in Hindi/Urdu with respect to distribution and interpretation. I argue here, following Saksena (1982b) that the indirect causative is not derived from, and cannot be seen as the ‘second causative’ of the direct causative. In section 3, I present new facts about the licensing and interpretation of se-marked phrases in these different