26 Dec 2009

Battling the Indo-European axe

The Proto-Indo-European root *péleḱu- 'axe' seems to me to be contrived. While the correct centum-satem correspondence between Greek -k- and Sanskrit -ś- is reassuring, it by no means validates the reconstruction. First, we have two fullgrade vowels in two consecutive syllables which immediately gives the alleged root an un-Indo-European appearance. Second, the root rests solely on the strength of a comparison between just two cognates: Greek πέλεκυς pélekus and Sanskrit परशु paraśú-. Third, their respective word accents don't match.

On the other hand, there seems to be a controversy concerning the skeptic's attempt to relate the above lexemes with Assyrian pilaqqu which has been long claimed to mean 'axe' but which, given the new reading of 'spindle', is thereby discredited. Ironically however there remains Assyrian palāqu 'to fell, to slaughter' to contend with, a verb on which nominal derivatives like naplaqtu 'knife' were built. Surely this is fundamentally not an Indo-European word and a Semitic source like this one remains preferable over a PIE root that begs even more questions than it's worth.

Yet, while we're on the subject. Are we sure that Sanskrit परशु paraśú- is really connected to the Greek? Consulting the online Monier-Williams Sanskrit-English Dictionary, an alternative form, पर्शु parśú-, is also provided. Unless I'm overlooking something, this only adds further doubt to the comparison. I wonder in what way any IEist can provide reassurance that these two words should be related. Note also Akkadian parāsu 'cut off, cut into pieces, separate' which should warn us that, considering many competing sources of this word available, these cognates could be red herrings. It would help greatly if we could at least establish the antiquity of this word in Proto-Indo-Iranian by way of further evidence.


  1. It's hard to argue from parasu/parsu without knowing which is earlier; later Sanskrit not-infrequently dropped a's in various positions, having already reduced them to a vowel that was probably close to schwa.

  2. I feel that parśu- is probably a later innovation, either because of a tendency to make it look more 'Indic', or as the above mentioned syncope.

    Especially because we find some productive ablaut derivations of this word like पारशव pāraśava- 'of a warrior-tribe' I can imagine people getting thoroughly confused whether the second full grade was supposed to be there in the original word or not.

    I think the Greek ~ Sanskrit correspondence is quite convincing, and am also convinced that this cannot be an indo-european word, because it looks nothing like an Indo-European word. I like the comparison to pilaqqu. As for the word accent, there is probably something going on that Lubotsky wrote about in his dissertation.


    And now I think about it, the Sanskrit accent actually makes a point in favor of the bisyllabic version of the word parśú- which is, due to the consonant 'grade', expected to be oxytone I think. Does that mean that parśú- was the original word, or that that variant transferred its accent to the original word?

    I'm not very clear on Greek accentuation, so I have no clue whether this accent could be secondary, but from an Indo-European point of view it seems odd to have the accent that far back. Interesting stuff.

  3. Ethan: "[...] later Sanskrit not-infrequently dropped a's in various positions [...]"

    This seems a reasonable interpretation.

    Phoenix: "[...] and am also convinced that this cannot be an indo-european word, because it looks nothing like an Indo-European word."

    Yes, and Roger Woodard agrees with us as well (Woodard, The ancient languages of Asia and the Americas (2008), p.247). Note however that if there's a connection between the Greek and Sanskrit accents, Woodard doesn't know about it when he states:

    "Thus, *péleḱu-s (Gk. pélekus, Skt. paraśús) exhibits two full-grade *e's in its root, whereas we expect to find only one (and note that the attested reflexes do not agree on the position of the accent)." (boldface mine)

    Zooming out to macro, Woodard is not just questioning the validity of the *péleḱu- reconstruction but of all 'axe' words. I find his skeptical survey of the extent terms rather convincing. One way or another, it's hard to avoid the Wanderwort issue despite the seeming sound correspondences between the two cognates provided and I suspect there's something I'm missing about the true source of the Sanskrit term. An IE etymology isn't working.

  4. On more pondering, I suppose that borrowing between *post*-IE dialects around, say circa 2800 BCE, might work with a clockwise movement around the Black Sea from Turkey of a term that might, let's say, have referred originally to a general metal tool for cutting/splitting/stabbing (ie. dagger, stiletto, axe).

    In this scenario:
    * Akkadian pilaqqu (Fertile Crescent)
    * > Aegean *pilaku (Crete/Turkey)
    * > Hellenic *peleku-s (NW Pontic)
    * > Indo-Iranian *peleću-s (Eurasian steppes)

    That's the best I could come up with to explain this so far.