The following ten theses, one for each chapter, try to distil the essence of all that precedes:
1. The variety and rich attestation of the verbal system of Ancient Greek make it an ideal starting point for the reconstruction of the verbal system of the Indo-European ancestor language. Virtually everything that is found in Greek has a recognisable linguistic pedigree, and few things that are found elsewhere have not left at least some trace in Greek as well.Footnote 1
2. Although there is a sea of literature on the Indo-European verb, only few scholars have sought to capture the systemic relationships between the reconstructed data. Where pertinent attempts have been made, these are often marred by incongruities and a certain disregard for typological insights and/or general principles of linguistic change.
3. Since the most basic verbal forms of Indo-European are found in both imperfective and perfective root paradigms, and since there is no coherent way of explaining the genesis of PIE aspect by reference to the objective boundedness (telicity) of many root aorists, it is likely that the latter have been perfectivised secondarily, because of their boundedness. Consequently, there must have been earlier perfectives, a role into which the historically marginal reduplicated aorists fit well.
4. If reduplication once served as a marker of aspectual perfectivity in Proto-Indo-European, the existence of non-perfective reduplicated stems must be justified. Among these, the reduplicated presents represent secondary imperfectives built on the model of the equally imperfective simple thematic presents. For the latter, a nominal ancestry has rightly been postulated.
5. A similar non-verbal background is reflected by the PIE perfect’s stative character, which is cognitively prior to any nactostatic value. After the nominal type involved had been verbalised, it too acquired a perfective (reduplicated) counterpart; and it is this innovation which accounts for the prominent generic semantics of many of the earliest perfect forms.
6. The most widespread thematic aorist type is not, as is usually thought, a thematised version of the root aorist, but a descendant of the reduplicated aorist with a simplified and generalised reduplication syllable. Refunctionalised as the ‘augment’, the latter has spread to other formations with past-tense reference.
7. That the augment was no past-tense marker to begin with is unambiguously shown by its distribution in the earliest texts. The philological data in all the relevant languages concur with the notion that its original function was that of a pre-radical perfectiviser.
8. The s-aorist as the last major aorist type of Indo-European must have competed with the root aorist already in the proto-language, the principal difference being its more pronounced semantic transitivity. Diachronic explanations of it which presuppose intransitive leanings are therefore unsustainable.
9. Any reconstruction of the earlier layers of the Indo-European verb should address the question of (Pre-)PIE alignment structure. A variety of independent arguments concerning both nominal and verbal morphology support the idea that a (split) ergative system preceded the accusative system of later Proto-Indo-European.
10. Once the mechanism of the Pre-PIE ergative → PIE accusative alignment change is identified, it becomes possible to understand not only some of the intricacies in the system of verbal endings but also the logic behind the evolution of the individual stem categories and verb forms. The fact that many previously unexplained details now fall into place indirectly corroborates the correctness of the reconstruction as a whole.
For all the breadth of the investigation, it cannot of course be claimed that every issue worth addressing has been addressed. That there will be room for discussion and, presumably, disagreement has been anticipated in the Preface. At the same time, it must be stressed again that what has been offered is a grammatical framework, not a grammar. Even when we look at Greek alone, there are matters that have been consciously ignored. For example, while we have touched on the origins and development of the subjunctive, the optative has hardly been mentioned; and although we have had occasion to say something about nearly all the majorFootnote 2 tense-stem formations, nothing has been said on the intransitive/passive aorist in -(θ)η- and, even more prominently, on the varied hypostases of the nasal presents.
These omissions are not due to a wish to ‘forget’ about potentially recalcitrant evidence. In each case it is possible to sketch genetic hypotheses which accord with what is presented here and which do not contradict what the data tell us. Thus, we are free to join those scholars who have connected the intransitive/passive aorist in -(θ)η- with the ‘stative’ suffix *-eh1- encountered in 5.44 and 8.36, and to hold that a form like 3sg. aor. (ἐ)μάνη ‘was/became mad’ (< *(h1e-)mn̥-eh1-t) either represents something very old – say, an archaic PIE tense/aspect-stem type which does not as such survive elsewhere – or an inner-Greek innovation that was backformed from a still nominal Proto-Greek stem *mn̥-eh1-nt- (> ptcpl. μανείς) and secondarily assigned a full paradigm.Footnote 3 Either way, nothing precludes a further analysis of the basic *CC-eh1 structure involved as a verbalised, and in the end aoristified, predicative instrumental, and thus as ultimately belonging to the same nominal type as the *CC-é form discussed in 5.42.Footnote 4
Similarly, whatever we may think about the mutual relationships between the subtypes of the nasal presents, including the antiquity (or not) of the type’s partial thematicity (1.11), we can agree with the long-standing notion that what ended up as a nasal root infix must have begun, as infixes commonly do, as a root affix.Footnote 5 This will then permit a reformulation of Kronasser’s idea that what is ultimately at stake here is an object pronoun fossilised from a time when the personal endings were still independent (subject) pronouns.Footnote 6 Given our identification of the nasal nominal ending *-m as an old Target case, this would square with a hypothesis that the generally telic and semantically (high-)transitiveFootnote 7 nasal present stems also go back to antipassive structures, particularly ones used in contexts with non-prototypical (animate) O’s.Footnote 8
However, the systemic relevance of any such theory is different from that of the theories we have explored in greater depth. The interconnectedness of the latter is such that non-trivial alterations would have substantial repercussions. To be sure, doubting our explanation of the origins of, for instance, the s-aorist or the i̯-present would not by itself bring down the ergative theory or our model of formal rejuvenation among the Indo-European perfective stems. But it would remove one building block which adds stability without being created purely ad hoc.
Not each of our ten theses is equally unorthodox, or equally incompatible with Indo-Europeanist orthodoxy. This is as it should be since the linguistic facts are not negotiable. And of course ‘orthodoxy’ itself is a slippery concept. Even where we did find reason to depart from a communis opinio, we were never the only ones to do so. Yet, small differences can add up, and that is why our end result, as diagrammed in Fig. A, bears little resemblance to the earlier big-picture theories reviewed at the outset.
In this respect, it is two points in particular that deserve final highlighting. Firstly, we have firmly subscribed to the idea that aspect is prior to, not a by-product of, tense in Indo-European. Like other peoples with non-tensed languages, the Proto-Indo-Europeans certainly knew how to keep apart past, present, and future eventualities, but their linguistic ordering system was not primarily geared towards such temporal categories. And secondly, the dogma of all-important PIE Aktionsart distinctions is nothing but a convenient way of avoiding reconstructive commitment. Again, different Aktionsarten will have existed in Proto-Indo-European no less than they do in modern English, French, or German – but there is no reason to believe that their grammatical status was at any point more pivotal than that of corresponding semantic groupings in the modern European languages.
*
To reconstruct a proto-language is a challenging enterprise, with difficulties lurking behind every bend of the path. One may get lost in the forest of data, glance back too often to where one is coming from, or mistake a gap between the bushes for a track. From time to time other paths branch off or meet one’s own, giving reassurance that someone has been there before but also providing opportunity to take a wrong turn. And even at the end, when one has arrived at a sunny clearing with a beautiful lake inviting to rest, there is no signpost, no absolute certainty that this is the spot one has been looking for.
On our path we set out with a detailed though incomplete set of directions: the philological data of Ancient Greek. Under way we encountered other hikers with their own directions, written in other languages, which they shared with us. Comparing these notes allowed us to choose at the next crossroads where our own indications would not have sufficed. But the deeper the woods became, the fewer people we met; for many had chosen to return as crossroads continued to appear. At that point, then, the best we could do was to make inferences, to replace comparative with internal reconstruction, to opt for those paths which seemed to point in the right direction. No doubt, our intuition alone might well have led us astray here or there, but more often than not there was something else to guide us: the safest path was the one most travelled. So we steered away from the fainter trails, the ones typological considerations advised against. And thus we finally reached ‘our’ clearing. Beyond it the forest continues, but to venture there we shall leave to others whose spirit of adventure is even greater. We know that we have already gone further than many are prepared to go, and we can only hope that some at least will follow us – or tell us where else to head next time.
Fig. A. Origins of the Greek verb (master diagram)
