How can a language have double passives but lack antipassives?

Passivization in Turkish may target both internal and external argument positions, for passives of unaccusatives, unergatives and transitives are possible in the language. Interestingly, Turkish also has double passives where the passive morpheme is iterated, and both argument positions are saturated. However, the language does not allow for antipassives, where only the internal argument position of a transitive predicate is saturated by passivization. Importantly, under plausible compositional analyses, antipassives are fully predicted, contrary to the fact. Hence, the unavailability of antipassives in a language that has double passives (and passives of unaccusatives) remains as a puzzle, which this paper sets out to solve. In particular, we argue that antipassives are ruled out on the grounds that the passive morpheme is visible to the case calculus in terms of Dependent Case Theory but has to remain caseless throughout the derivation. We show that this account also correctly predicts that internal argument positions that happen to be lexical case positions can never be targeted by passivization. Finally, we provide a fine-grained analysis of internal argument suppression accounting for the aspectual restrictions it has been reported to have, and compare our work to the existing proposals.