7

I am trying to fit

ᾰ̓κήκοᾰ (active indicative perfect of ᾰ̓κούω, first person singular)

to the model of

λέλῠκᾰ (active indicative perfect of λῡ́ω, first person singular),

wherein the word breaks down to:

λέ-λῠ-κ-ᾰ, or
reduplication - stem - tense suffix - personal ending.

(Please let me know if I got that terminology wrong already.)

QUESTION

  1. Am I right to do it as follows?

    ᾰ̓κ-ήκο-ᾰ, or
    reduplication - stem - [no tense suffix] - personal ending

    To get that far, I relied on Mastronarde's Introduction to Attic Greek, which states:

enter image description here

  1. Is there any basis to think that the original reduplication was of the initial vowel and consonant and the second vowel? The idea would be that, for example:

    ᾰ̓κο-ᾰκο-ᾰ → ᾰ̓κ-ήκο-ᾰ.

    Actual attested earlier forms of the ᾰ̓κο-ᾰκο-ᾰ sort would be the best kind of answer.

    Another good answer would give a word that began with a vowel - consonant - consonant sequence, but received an Attic reduplication with the 'lengthening' of the second vowel (which would, as it were, refute the theory being proposed). Note that all four of Mastronarde's examples begin with a V-C-V sequence (tempting one to the theory).

BACKGROUND

As an aside, the theory seems to represent my mind's trying to reduce the number of primitive devices (already too many). Reduplication of some number of initial letters (up to three) seems like fewer such devices than reduplication or reduplication plus lengthening.

Catomic
  • 1,503
  • 7
  • 10
  • 4
    In Attic ο+α=ω, not η. You get η primarily by lengthening α (which gave ᾱ in other dialects). – Draconis Mar 06 '17 at 15:59
  • 3
    Same goes for ὄμνυμι, since ο + ο = ου. I have a lot of sympathy for your desire to reduce the arcane ramifications of Greek grammar into a few easily understood axioms, but unfortunately--it's not always possible! – brianpck Mar 06 '17 at 16:33
  • 2
    Cf. also the perfect ὄπωπα, stem ὄπ-, from the suppletive verb ὁράω. I have to agree with the others: the vowels η (from α or ε), ω (from ο), and ει (from ε (from ι is not lengthening)) at the beginning of a past verbal stem are commonly a form lengthening of the initial vowel, rather than contractions. – Cerberus Mar 06 '17 at 18:06

2 Answers2

9

Unfortunately -- or fortunately, depending on your point of view -- "Attic reduplication" is one of the many irregular areas of Greek morphology which can only be understood by going back to Proto-Indo-European. Here's a short version of the probable answer:

PIE *h₂ke-h₂kows-h₂e > akākowsa > akākowa > akēkoa

This is a perfect form of the root *h₂kews- "hear" (which actually also gives us the English verb hear!). The rule for perfect formation in PIE was to place the root in the pattern Ce-CoC-, where Ce- reduplicates the initial consonant of the root, and add the perfect endings. But, each of those C's can stand for more than one consonant: if the root began with two consonants, as here *h₂k, both of those could be reduplicated, hence *h₂ke-.

Now apply the following regular sound changes:

  • Initial *h₂ before a consonant > a
  • *h₂e > a
  • *eh₂ > ā > Attic-Ionic η
  • intervocalic *s is lost
  • intervocalic w is lost (in Attic-Ionic and many other dialects)

... and you get ἀκήκοα.

(Note that there is some debate both about the etymology of this particular verb and the origin of Attic reduplication generally, but the above seems to be the majority opinion.)

The explanation for other such verbs, like those in Mastronarde's table, is similar, except with different laryngeals: *h₁ for verbs with ε/η, *h₃ for verbs with ο/ω.

TKR
  • 31,292
  • 2
  • 66
  • 120
  • Very interetsing. And do laryngals also explain the augmental lengthening, as in ὤμνῠον, εἶπον, which doesn't look like ordinary contraction (or does it?)? I thought the augment was hypothesised to be from an adverb "then" that looked like he and was related to Germanic ge-, but perhaps I misremember and it was in fact a laryngal? Incidentally, now I wonder why those different verbs used different laryngals in your examples. Assimilation? – Cerberus Mar 07 '17 at 00:30
  • 1
    @Cerberus, the PIE augment is reconstructed as (h₁)e-, but is not related to Germanic ge-, which is most likely from PIE ḱóm. εἶπον is actually not an augmented form at all but a reduplicated aorist, *we-wkʷ-. As for the temporal augment (ὤμνῠον), laryngeals may be part of its explanation (see comments to this answer) but there's certainly some analogy involved too since it also appears in verbs that did not begin with a laryngeal. – TKR Mar 07 '17 at 01:15
  • 1
    As for why different verbs have different laryngeals, they're just part of those roots. Laryngeals are consonants like any others -- some roots happen to begin with h₁ just like some roots happen to begin with s. – TKR Mar 07 '17 at 01:55
  • As to ge-, I heard that somewhere, but it must be an outdated theory (although it seems that the relation with con- is not yet consensual). And εἶ- in perfect forms, is that still a lengthening of the vowel, or is it an effect of reduplication, too? Analogy is always a strong factor indeed. About those laryngals, I assumed they only appeared in the perfect forms that were mentioned, not in the roots; that what confused me. But if they are in the roots, then it makes sense. – Cerberus Mar 07 '17 at 02:05
  • @Cerberus, the ge- : ḱóm connection is somewhat problematic because PIE ḱ regularly gives Germanic h. But none of the laryngeals should give Germanic g, either. By "εἶ- in perfect forms" do you mean e.g. εἴληφα? If so, I believe that's something else again: that root had sl-, and the form seems to be from se-sl- with the first *s lost by Grassmann's Law (because of the φ later in the root) and the second by the First Compensatory Lengthening. Other perfects in εἰ- are thought to be analogous to εἴληφα. – TKR Mar 07 '17 at 02:19
  • Thank you. Am I right to understand as follows, where brackets represent an operation? *h₂ke-h₂kows-h₂e =[initial *h₂ becomes a]=> ake-h₂kows-h₂e =[*eh₂ > ā > Attic-Ionic η]=> akηkows-h₂e =[intervocalics lost]=>akηko-h₂e =[*h₂e > a]=> akηkoa? (The sequence of operations not meant to represent order of historical development.) – Catomic Mar 07 '17 at 07:35
  • If so, was my guess right to the extent it consisted of (1) a reduplication of 'three things' (*h₂, k, and e) and (2) a subsequent merger between the last of them (e) and the initial thing in the stem (h₂), but wrong in thinking (3) these operations might have taken place 'in Greek' when in truth (or probability) they had happened prior to Greek and the results were inherited by Greek? | I just don't understand why the last of the three duplicated elements, viz. e, is represented as *e in one of your lines, viz. '*eh₂ > ā > Attic-Ionic η'. – Catomic Mar 07 '17 at 07:37
  • @Catomic, yes, that's all correct. I'm not sure I understand your last question -- if it's about the asterisk, that's meant to show that the sequence *eh₂ is all part of reconstructed PIE rather than attested anywhere. – TKR Mar 07 '17 at 22:19
  • Thanks for that confirmation. (About the asterisk, I'd never seen PIE notations before and simply didn't know what it meant.) – Catomic Mar 08 '17 at 01:54
3

Discussions of proto-Indo-European, and especially those postulating multiple laryngeals, are best when copiously peppered with words like “perhaps”, “possibly”, “hypothesis”. The so-called Attic reduplication is a highly contested matter and there is no consensus about its origin. What one needs to mention in any case is that, unlike the regular reduplication of the type le-lu-ka (which is well known in Greek, Latin, Sanskrit, Gothic and other languages), the “Attic” reduplication is specific to Greek and is thus likely to be a Greek innovation, not an inheritance from IE. Also, that in Sanskrit, and mostly also in Greek, roots beginning with a consonant cluster reduplicate only the first consonant of the cluster (type: ke-kleia; *si- > hi-stāmi). This is the problem with the proposed proto-form *h₂ke-h₂kows-h₂e.

fdb
  • 17,845
  • 1
  • 23
  • 47
  • 2
    This is a fair caveat. To me, the laryngeal account works too neatly to be accidental, and I don't know of a better theory. Initial preconsonantal laryngeals are not preserved in most IE languages, so you wouldn't expect to see Attic reduplication in the other languages you mention; one might still expect a reflex of the post-reduplicant, root-initial laryngeal (e.g. *h₂keh₂- > kā-) and I don't know if that's found anywhere, but it could easily be analogically leveled out in any case. – TKR Mar 07 '17 at 22:37
  • 2
    As for reduplication of initial clusters, it's notoriously inconsistent (e.g. tiṣṭhati vs. ἵστημι vs. steti; the first shows it's not the case that in Sanskrit only the first consonant is reduplicated), but the view that the whole cluster was reduplicated in the PIE perfect is not uncommon -- this seems necessary for sT clusters to explain e.g. Latin steti, scicidi, and it's plausible that HT clusters would have behaved the same. – TKR Mar 07 '17 at 22:39
  • 2
    Looking into the Indo-Iranian evidence, consider Vedic jāgāra (:Gk. ἐγείρω, with h₁), which shows the ā I speculated about in my first comment above -- this would be the regular outcome of *h₁ge-h₁gor-, same pattern as for ἁκήκοα. The only other option is to say that only the second consonant was reduplicated (ge-h₁gor-), which is unlikely. – TKR Mar 07 '17 at 23:07
  • 2
    And note Armenian arari "I made" : ἄρηρα, which shows the "Attic" pattern including the reflex of an initial laryngeal. I'll shut up now. – TKR Mar 07 '17 at 23:23