Abstract: It will be argued that Quotative Inversion (QI) in Hungarian, i.e., inversion of the finite verb and a verbal modifier within a reporting clause, requires incorporation of an abstract operator, OpQ, into Pred°. This accounts among other things for the fact that Hungarian QI is incompatible with unbounded dependency formation, i.e., incompatible with OpQ placement in Spec,FocP. The overall head initiality of clauses undergoing QI will be derived in two steps. First, a PF-linearization mechanism in the spirit of Fox and Pesetsky (2005) guarantees strictly Pred°-initial PredP. Second, information structural impoverishment "shuts down" TopP and FocP, the phrases dominating PredP. The latter idea will be grounded in particular assumptions about the narrative force of QI constructions. OpQ will be argued to be a covert counterpart of overt demonstratives incorporated into the Hungarian verb mondja (say). A semantics of demonstrative incorporation is shown to shed interesting light on exhaustive interpretation in the presence of communication predicates having undergone QI. Considerable efforts are made to weigh the language-specific choices for the analysis of Hungarian against the options available for deriving varieties of QI in languages like English, French, Spanish, and Dutch, as our analysis is developed against the backdrop of the approaches by Collins and Branigan (1997), Collins (1997), Suner (2000), and de Vries (2006).
Keywords: Inversion, reported direct speech, parentheticals, demonstratives, information structure