The rise (and possible downfall) of configurationality

In research on Warlpiri, Hale (1983) called attention to a number of features of non-configurational languages which point to the absence of hierarchical relations among constituents, notably the non-existence of the VP. Further research has shown that the most relevant features of non-configurationality are discontinuous constituents, free use of null anaphora, and free word order (Baker 2001). An array of features of non-configurationality exists in the ancient Indo-European languages (Devine, Stephens 1999:143-148, Schäufele 1990, Sigurðsson 1993); the IE languages present a common trend toward increasing configurationality (Hewson, Bubenik 2006). Among features of non-configurationality, free occurrence of null objects in the IE languages (Luraghi 2004) has been paid little attention, even though some studies indicate that increasing obligatoriness of object pronouns reflects increased transitivity (Johnson 1991). Within formal frameworks, free occurrence of null objects in non-configurational languages is viewed as allowed by obligatory occurrence of zero arguments, which fill in the argument positions when no NPs occur; possible NPs are then considered as adjuncts, predicated of zero arguments (Jelinek 1984). In Luraghi (forthcoming) a different explanation is suggested, whereby a distinction is made between semantic and syntactic valency (Luraghi, Parodi 2008: 197-199). Following this approach, there is no need to set up zero arguments for non-configurational languages of the dependent-marking type, such as the ancient IE partly were: possible null objects occurred on account of the purely semantic nature of verbal valency. The rise of configurationality reflects a shift toward increasing syntacticization of valency, which in part corresponds to observed increase in transitivity. This explanation has the merit that it keeps distinct dependent-marking non-configurational languages from head-marking ones, which also present other relevant differences (e.g. they do not usually allow discontinuous constituents, Baker 2001). In this chapter I elaborate on the grammaticalization of the verbal valency in the early IE languages, and to address the question why configurational syntax arose. In doing so, I would like to compare formal and functional explanations, in order to evaluate the explanatory power of different theoretical frameworks.