The grammaticalization chain of case functions

In the 1980s and 1990s, grammaticalization research brought forth a number of intriguing proposals concerning the directionality of extension and change between case functions, i.e. semantic roles. One of these proposals, by Heine et al. (1991), consisted in a single unidirectional chain of increasing grammaticalization of case functions, advancing from the spatial domain via an anthropocentric domain to inanimate and abstract domains, based on metaphorical extension. This bold proposal is still often referred to, but has proven to be problematic in some points, especially in the intermediate area of anthropocentric concepts. The present paper investigates this chain in detail, showing which hypothesized extensions still hold and which not. It is argued that the single chain is untenable and has to be broken up into at least two larger directionalities, one leading to the development of the expression of core participants, often triggered by constructional reanalysis, and the other leading to highly abstract, inanimate case roles. The application of the concept of metaphorical extension to the reanalysis cases is much less straightforward than that to the semantically motivated cases. However, the result of change is similar, as the semantic weight of the linguistic expression shifts practically completely to the nominal, and case marking becomes almost purely syntactic marking. Dative case functions are in an intermediate position leading to extensions both in the direction of core participants and to inanimate, abstract roles. Keywords: Grammaticalization; semantic roles; semantic maps; unidirectionality; case functions; metaphor