Séminaire de Recherche en Linguistique

Using terms from morphosyntax (e.g. Corbett 2006), we can say that while syntactic features such as person, number and gender are inherent features on nominals and nominal phrases (DPs), structural case is contextual, as its value on a DP is determined by whatever relevant head in the syntactic context the DP agrees with (T, v, etc.). This analytical distinction corresponds to some degree to Chomsky’s interpretable and uninterpretable features. I will present an argument that inherent and contextual features are managed differently by the syntax, and the evidence suggests that they are not all represented in a simple set description on a given DP.

In some languages, raising a DP subject from an embedded clause into a matrix clause leads to a situation where apparently two different cases are assigned to the DP: one case in the lower clause and a different case in the higher clause (e.g. Bejar and Massam 1999 on Niuean, Polinsky and Potsdam 2002 on Tsez). On most current approaches, this is problematic. The facts show that inherent features of DP are preserved under raising, but contextual ones are not, and this suggests that they should be treated differently.

The talk focusses on issues of where and how contextual information about structural case is available and referred to in the syntax. Contextual features are syntactically active and present in a given (limited) domain, but never pass out of that domain. I will suggest that syntactic contextual features are therefore properties of the domain, but not necessarily properties of the constituents (e.g. DP) in that domain. In some language families (Athapaskan, Oto-Manguean), it is reported that case marking is present on the verb – cross-referencing the argument DPs not in terms of Phi features, but case.