The perfect in Homeric Greek: A functional account

The function of almost all perfects in Homer can be inferred from the function of the present/aorist of the same verb according to a model which combines time reference, Aktionsart, and thematic roles. This makes it possible to explain why some perfects are resultative (e.g. όλωλε ‘it is destroyed’) and some are not (e.g. τέτριγε ‘it squeaks’) and why some active perfects have a different argument structure from that found elsewhere in the active (e.g. perfect όλωλε ‘is destroyed’ vs present όλλυμι ‘I destroy’). The exceptions to this pattern are few enough for language learners to have been able to acquire the function of the perfect as a rule. We hypothesize a pre-Homeric stage at which the perfect predicated the present state of the subject but was voice-neutral. The Homeric data can be explained as arising from this earlier stage with minimal reanalysis.*