The paper takes up the question of the finiteness of the subjunctive, in particular the Greek subjunctive. Taking into account discussions within the gene­rative framework, it adopts the position that the sub­junctive is indeed finite, and examines the main stumbling block in this direction, namely its uni­versally acknowledged lack of semantic tense. It challenges this assumption, analysing a subset of its uses, that in na-complements, in an attempt to show that there is a temporal dimension to the sub­junctive even in its narrowest sense (when the only choice available is between the three aspectually opposed forms): the event it denotes is always placed at a time posterior to that of matrix verbs with a deontic meaning and contemporaneous to that of matrix verbs with a sensory/perceptual meaning.

Greek is assumed to be particularly conservative in terms of its verbal system, which is supposed to have always been organised on the basis of the grammatical category of aspect. This paper claims that, on the contrary, a major change has occurred between the early stages, as seen in the Homeric poems, and Modern Greek. Originally, the three stems of the verb represented the three main types of Aktionsart: states (Perfect), telic activities (Aorist) and atelic activities (Present). Through a gradual grammaticalization process the Aorist and Present stems came to express the binary aspectual distinction of perfectivity and imperfectivity, a change that explains the loss of the third stem. Christidis’s astute observations about the role of ideological constructs in the perception of language change explain the failure of scholars to notice this rather striking difference between the ancient and the modern system.