Towards a theory of the meaning of complementizers in classical Attic

Abstract The application of TG-oriented methods to the study of complementation in Latin (Lakoff 1968) and Greek (Lightfoot 1975) has led to the belief that complementizers are empty of meaning, and that they are distributed more or less at random. Complementary to this is the view that independent uses of subjunctives and optatives can only be explained by the adoption of abstract governing verbs, which have the same idiosyncratic properties as surface governing verbs. I argue that this view of complementation overlooks the semantic regularity underlying the choice of complementizers. It can be shown that the system of complementation in classical Attic is, in fact, well structured with respect to the presence or absence of an existential presupposition in the dependent clause. Moreover, cooccurrence of different complementizers after the same governing verb is in my view not due to this particular verb's idiosyncracy, or to historical coincidence, but to the particular value each complementizer bears. Second, I argue that Lightfoot's view on complementation fails to see the real factors (viz. semantic affinity and partial functional equivalence between some of the complementizers) which determined the change of complementizers in the evolution towards Modern Greek.