Incorporated nominals as antecedents for anaphora, or How to save the thematic arguments theory

The paper discusses the incorporation facts, mainly from Hungarian and Hindi, and the extended version of DRT developed by Farkas and de Swart (2003) to capture the properties of incorporated nominals across languages. It is shown that though the framework as defined by Farkas and de Swart does not derive all predictions it was aimed to derive by the authors, it can be easily extended into a version which does predict the relevant data. However, after examining more closely the data from Hungarian-type incorporation languages, I put forward a preliminary hypothesis alternative to Farkas and de Swart’s proposal, namely, that the possibility of anaphora to singular incorporated nominals is not a parameter of a language, but arises in languages like Hungarian if and only if the context makes it clear that the maximal entity denoted by the incorporated nominal is at most atomic. It is left for future empirical research to find out whether this hypothesis can actually account for the whole range of relevant data.