08 March 2013

To Be or to Be: Aspects of Existence (Act II)

Proto-Indo-European verb roots had an obligatory aspectual value: they were either perfective (punctual, discrete) or imperfective (durative, continuous). Their inherent aspect could be switched by adding appropriate suffixes, but it was preserved in simple root verbs (in which inflections were added to the bare root). Inflectional endings could contain information about the tense of the verb (to be precise, they could carry present-tense markers), but perfective stems were always unmarked for tense and were normally interpreted as a perfective past tense referring to a punctual or completed action. By contrast, imperfective stems could form both a past tense and a present, referring to an action in progress or to repeated/habitual actions. The traditional terminology of IE studies uses the following, somewhat confusing terms for combinations of aspect and tense:
  • present (= imperfective, present)
  • imperfect (= imperfective, non-present)
  • aorist (= perfective, non-present)

The confusion is made worse by the fact that there was a third aspect with a stative (non-eventive) value (e.g. referring to a state resulting from a previous action), called the “perfect” (not to be confused with “perfective”). The perfect was not the inherent lexical aspect of any root; it was derived (from both perfective and imperfective roots) by means of certain morphological operations, and had its own special personal endings.

The formal aspect of a PIE verb can be quite surprising. The root *gʷʰen- ‘kill, strike’ was imperfective, though most people would regard killing as a punctual, non-durative activity. Remember, however, that the meanings we attach to PIE roots and words are only approximate. Perhaps it would be more correct to gloss *gʷʰen- as ‘to deal blows’ (a prolonged or repeated action).

Sometimes we meet pairs of distinct but nearly synonymous roots, the main semantic difference between them being one of aspect (how the action is viewed in relation to the flow of time). For example, *h₁ei- means ‘walk, be on the move’ (imperfective), while *gʷem- means ‘go, come, advance a step’ (perfective). But, beside a root verb, *gʷem- could also form derived stems with imperfective semantics: *gʷm̥-je/o- (3sg. *gʷm̥-jé-ti, 3pl. *gʷm̥-jó-nti) and *gʷm̥-sḱe/o-, with the approximate meaning ‘to be going/coming’. Note that those derivatives, unlike root verbs, had a fixed accent on the stem-forming suffix.

Non-present eventive verbs (imperfects and aorists) had so-called secondary personal endings. In the third person, they differed from presents by lacking the final *-i, which seems to have functioned as a marker of here-and-now (and therefore of the present tense). Thus, *gʷem- had the following forms: 3sg. *gʷém-t, 3pl. *gʷm-ént. In some branches (for example Greek and Indo-Iranian) the past tense was explicitly characterised by the accented prefix *é- (called the “augment”), so that ‘(s)he came’ was expressed as *é-gʷem-t, and ‘(s)he was walking’ as *é-h₁ei-t (as opposed to the present *h₁éi-ti). Without the augment, the non-present forms could be interpreted as “timeless” (neutral with respect to tense).

The imperfective verb *h₁es- ‘be’ also had a perfective counterpart, the root aorist *bʰuH-. The use of a capital *H as a cover symbol for any of the three laryngeals means that the evidence is insufficient to make the reconstruction more specific. Unlike most other verb roots *bʰuH- is usually cited without a full vowel in the basic form, because we do not even know for sure if it was originally *bʰeuH- or *bʰweH- (each reconstruction has its fans, but the evidence is inconclusive). The reduced “zero grade” of either root shape would have been *bʰuH-, and that’s the only form widely attested across the branches of the IE family. Quite possibly the weak form was generalised already in the common ancestor of the Indo-European languages. If so, the ancestral aorist displayed no vowel alternation even if its accent was mobile: 3sg. *bʰúH-t, 3pl. *bʰuH-ént.

Not exactly an action
If *bʰuH- was at the same time perfective and semantically related to *h₁es-, it must have referred to some transitional aspects of existence like ‘come into being, arise, appear, happen’, or ‘get, grow, become’, i.e. entering a state rather than remaining in it. We find copious evidence of the root having those meanings, as well as many similar ones, in various IE languages. But, as we have seen, aorists could be transformed into presents by adding imperfective suffixes. For example, the derivative *bʰuH-jé-ti would have meant ‘is becoming’, ‘is happening’, etc., converging semantically with ‘being there’. Let’s suppose that *h₁es-, because of its rather special existential function (‘to be’ is hardly an action or an event), was defective in some respects. Judging from the comparative evidence, it did not form any stative (“perfect”) or perfective (“aorist”) stems; no verbal adjective analogous to English past participles was derived from it. If any descendant of PIE evolved gramatically in a way which made those gaps inconvenient, existing forms of other verbs with a similar meaning could be co-opted to make the paradigm complete. In the next post I will try to show how this process operated in Germanic.


  1. Would you agree that the traces of PIE *gʷʰen- (with different ablaut grades) can be found in English bane, gun (f. ON), de-fend (f. Lat.) and Polish gonić, gnać 'chase' and żąć 'reap'? It would mean that PIE *gʷʰ > PGmc *b (E bane), but not in the form that yielded ON gunnr...

  2. Not żąć 'reap', which is probably a different root (Balto-Slavic rather than IE), but definitely gnać and its old inflected forms: żonę, żeniesz, etc. Gonić is a regular o-grade iterative of the same root.

    As for Germanic, *gunþ- 'fight' most likely belongs here (with normal loss of labiality before *u, even if it reflects the prop vowel of a former syllabic nasal). I'm less sure about *ban-an- 'death, bane', first because of the paucity of examples illustrating the development of initial *gʷʰ, and secondly because I do not understand the derivation of such a nasal stem with an o-grade in the root.

  3. As regarding the augment, it's interesting to notice a large class of Basque verbs have a similar prefix *e- in their non-finite and past forms. The main difference is the Basque isn't stressed and apparently it also appears in some indefinite pronouns such as e-zer 'something', from zer 'what'.

    A possible interpretation of this correspondence (which I don't think is a chance resemblance) is that (under the Vasco-Caucasian hypothesis) both prefixes had a common origin, possibly related to NEC *=a/=i 'to be (an auxiliary verb)', where = is a wildcard for a "class-prefix" such as the glottal stop ʔ.

  4. Perhaps it would be more correct to gloss *gʷʰen- as ‘to deal blows’ (a prolonged or repeated action).

    An obvious parallel would then be German erschlagen, which generally has its literal meaning "beat to death", but is also in more poetic use for "slay" (generally involving a sword rather than a club or bare fists).

    (And slay is a cognate of schlagen, right?)

    1. Yes, OE slēan (*slahan, cf. Goth., OHG slahan), slōh, slōgon, slæġen still meant primarily 'strike, deal a blow'. To quote one of the laws of King Æthelbirht of Kent,

      Gif man ōðerne mid fyste in naso slæhþ...

      'If someone strikes another (person) on the nose with his fist...'