Explaining Siewierska's generalization

This article presents an explanation for a cross-linguistic gap observed by Anna Siewierska: morphologically unmarked indirect objects may alternate with prepositional marking in what is sometimes called a 'dative' or 'prepositional-dative' ditransitive frame, but never with actual dative case marking. 'Dative', to the extent it alternates with accusative, is always expressed as a preposition. I show firstly that German, which has a robust dative case paradigm, also displays a double object alternation in which the erstwhile dative DP occurs in a prepositional phrase, meaning both accusative (in English) and dative (in German) indirect objects alternate with prepositional encoding. I construct an analysis in which the the indirect object may be generated as either a DP (which receives dative in German and accusative in English) or a PP in the same theta position. This characterization of the double object alternation does not admit an alternation between dative and accusative case on the indirect object, capturing Siewierska's generalization. The analysis also extends to 'symmetric' passive languages, in which either object in the double object construction can be raised to subject in the passive. Some current perspectives on this phenomenon make such languages exceptions to Siewierska's generalization, but not the analysis proposed here.