Not all Greek verbs inflect perfects

Note: I’m interrupting Andrew Keenan’s series on Wittgenstein to make a note about the Greek perfect.

A few weeks ago I put a poll up on Twitter and another one on Facebook, asking whether people thought that a particular verb had the perfect as part of its inflectional paradigm. Here are the Facebook results (looks like I typed the accent wrong):

Poll on ᾂδω in the Perfect

The question is an interesting one, in part I think, because introductory and intermediate grammar discussions of what perfects mean tend to obscure a few substantive issues. To be fair, teaching grammars are predominantly focused on fairly general issues about the meaning of the perfect rather than on questions of distribution or the significance of that distribution.

The correct answer was ‘no.’

The Ancient Greek perfect has a strong tendency for avoiding verbs that do not inherently denote a change of state. And ᾄδω one such verb. There are state predicates of course, that form perfects, but even then, their usage is a bit more complicated. Generally speaking, we can observe the following patterns:

  1. Large swaths of state predicates that never appear in the perfect.
  2. Bare activity predicates (dynamic, durational, no change of state, ) are surprisingly unusual and uncommon even in the Koine period.
  3. Semelfactive predicates (dynamic, instantaneous, no change predicates like, ‘clap’, ‘wave’, ‘flash [intrans.]’, or ‘glimmer’) cannot be found with the perfect at all.

Classicists likely are aware of these issues far more than those of us who study post-classical Greek, if only because during the Homeric and classical periods of the language, the distribution of the perfect was even smaller than it is in the Hellenistic and Roman period. The traditional narrative for Classics goes along the following lines (cf. Allan 2016, for example):

  1. The perfect was resultative in Homer (not in the Chantraine sense of object-result).
  2. It extended to transitive verbs after Homer, while still in some way referring to the state of the subject.
  3. It became an anterior-like perfect at some point after the classical period.

The problem is that this account doesn’t engage with the distribution across verbs very well either. The Greek perfect, diachronically, with its stative-like origin (Willi 2018, Clackson 2007, etc.) has a number of possible paths of grammaticalization before it. Anterior is certainly one of those. We can compare the standard definition of the Anterior from Bybee, Perkins & Pagliuca (1994) with distributional patterns in the language data.

“An anterior signals that the situation occurs prior to reference time and is relevant to the situation at reference time. Anteriors are typically translated with the English perfect and often accompanied by the relational adverbs ‘already’ and ‘just’. Anteriors may occur with past or future tense marking” (BPP 1994, 54).

Because the anterior does not require the kind of lexical distribution patterns that we find in Ancient Greek, I’m not convinced it is the best way to characterize the Greek perfect either in the Classical period or in the Early Roman period, though I am quite open to it still being a later development.

I think there are better descriptive alternatives than “anterior” and while I will not be spending as much time on particular labels for the perfect, I very much look forward to talking about usage and lexical distribution this coming weekend at Southeastern Theological Seminary.  I’m sure I’ll see some of my readers there.