Valency mismatches and the coding of reciprocity in Australian languages

Abstract

Reciprocals are characterized by a crossover of thematic roles within a single clause. So, in John and Mary wash each other, each of John and Mary is both washer and washed, both agent and patient. The competing pressures to distinguish and merge the reciprocating argument(s) are resolved by different languages in complex and illuminating ways which often create special argument configurations not found in other clause types. While some languages either encode reciprocals by clearly bivalent, transitive clauses (like Warlpiri or English), or clearly monovalent, intransitive clauses (like Wambaya or Yukulta), other languages adopt a mixed or apparently ambivalent solution.In this paper, based on an extensive sample of Australian languages, we develop a typology of apparent valency/transitivity mismatches in reciprocal constructions including: (a) monovalent clauses with a single ergative NP; (b) mismatches between case marking and the number of arguments encoded on auxiliaries or by pronominal affixes to the verb; (c) the use of ergative marking on secondary predicates and instrumentals with a nominative subject; and (d) complex clause constructions sensitive to valency. Such mismatches, we argue, result from an ‘overlay problem’ by which both divalent and monovalent predicates in the semantic representation of prototypical reciprocal scenes have had a hand in shaping the morphosyntax of reciprocal constructions through grammaticalization.