--- In
cybalist@yahoogroups.com, Piotr Gasiorowski <gpiotr@...> wrote:
>
> W dniu 2012-02-02 21:53, dgkilday57 pisze:
>
> > I wonder whether that Greek verb could be explained on the basis of an
> > archaic morphological pattern instead, full grade in the stative (*-eh2-
> > or *-eh4) and zero-grade in the present. I seem to recall this verb
> > having a Hittite cognate (whose pattern might shoot down such a proposal).
>
> LIV (this time Kümmel) has *weh2g(^)- for 'break'. I think the brackets
> can safely be removed, as I see no reason to deny a connection with Ved.
> vájra- 'thunderbolt, Indra's weapon' (Av. vazra-). The zero grade
> somehow never appears where one would expect to see it, and it seems
> that short *a is as weak as the vocalism can possibly get (cf. also Gk.
> agmós 'fracture'). Gk. ágnumi would have to continue the stem
> *uh2g^neu-, but *uh2 undergoes regular laryngeal breaking to Proto-Greek
> *wa:, and there's no way we could get *wa instead (even if one doesn't
> accept Francis/Normier's Law: neither Greek, nor any non-breaking
> language ever shows u:g^- < *uh2g^- in this root). Nothing but a
> powerful international conspiracy could have wiped out the regular zero
> grade so consistently.
>
> As for the perfect <éa:ga>, all that LIV has to say is this: "R(e) statt
> R(o), bei Wurzeln mit *eh2 im Gr. regulär". This kind of reasoning is
> circular if there's no independent proof of the presence of *h2 here in
> the first place (a-colouring isn't enough, as far as I'm concerned).
Apart from that, it is very odd that Greek would do this. The real explanation must be otherwise.
> The Hittite cognate is wa:ki/wakkanzi 'bite'. Those who reconstruct a
> laryngeal here have a hard time explaining the initial <wakk-> rather
> than <ukk-> in the weak stem. If it looks for all the world like
> *wa:g^-/*wag^-, why should it be anything else?
Marginal ablaut series are hard to accept. I prefer to look for a way to justify the standard ablaut series, obscured by unusual phonology or morphology. After all, that is what brought us laryngeals in the first place. For diachronists, system should take precedence over surface structure.
If the Proto-Greek zero-grade of PIE *weh2g^- was *wa:g-, it might be used to explain the Attic aorist passive <eá:ge:n> 'I was broken'. (Sihler's explanation, New Comp. Gr. §442.1, of quantitative metathesis from a protoform with long augment, *e:wáge:n, is unconvincing because he provides no basis for a long augment in this verb.) However, Epic <(e)áge:n> and the Attic compound <kateáge:n> have short /a/, and the second aor. pass. generally takes zero-grade. The long /a:/ of Att. <eá:ge:n> has probably been introduced from the intransitive second perfect <éa:ga> 'I am broken'.
The only good parallels to Greek <éa:ga> are other long-vowel second perfects like <péphe:na> 'I have appeared, I am visible' and <héa:da> 'I am pleasing'. (Sihler, NCG §516, says that the vocalism of all three is "best taken as analogical" without specifying to what.) The paradigm of <phaíno:> 'I show' includes an intr. 2nd aor. pass. <epháne:n> 'I appeared' against the tr. 1st aor. pass. <ephánthe:n> 'I was shown', and the 1st perf. <péphagka> 'I have shown' is also transitive. This last form is not ancient, since Epic Greek forms kappatic aorists only to long-vowel stems. Thus the intr. sense of this verb is older, with the tr. sense likely developing from middle forms reinterpreted as passive, 'I appear (for my benefit)' > 'I am caused to appear, I am shown', due to morphological overlap of the middle and passive in several tenses. The 2nd perf. <péphe:na> stands to the 1st aor. <éphe:na> as <dédo:ka> 'I have given' to <édo:ka>. Since the kappatic perfect is not found outside Greek, it is likely that it spread from <dédo:ka>, <téthe:ka> 'I have placed', and <-heîka> 'I have sent', modelled after the 1st aor. (singular only) suppletive forms with /k/-extension, <édo:ka>, <éthe:ka>, <hêka> (for the last two cf. Latin <fe:ci:> 'I made', <je:ci:> 'I threw'). Thus <péphe:na>, while older than <péphagka>, cannot be securely regarded as ancient, and its full-grade root-vocalism may come from the aor. <éphe:na>. Similar comments apply to <méme:na> 'I am mad'.
This leaves <héa:da> (Doric <héada> with short vowel), 2nd perf. to <handáno:> 'I please' (not used in Attic prose). The extant 2nd aor. forms have short vowel, Attic-Ionic <héadon>, Epic <hádon>, <eúadon>, Cretan 3sg. <ewade>; Dor. <héada> probably took its vocalism from the aorist. The future <hadé:so:> and 1st perf. <háde:ka> make it very difficult to argue for analogical replacement of a long vowel. It appears that the zero-grade of the root *sweh2/4d- in Proto-Greek was *swad-, not *swa:d-. Then the question is what full grade is doing in the 2nd perf. <héa:da>. Perhaps it is /o/-grade, with *-oh4- > *-a:- in Proto-Greek, the root being *sweh4d-. Then <éa:ga> could also be referred to the /o/-grade of *weh4g^-. This can explain why *wa:g- (and *swa:d-) are not reflected as Greek zero-grades. While *-uh2- becomes PGrk *-wa:-, *-uh4- becomes *-wa-. Thus there is no objection to <ágnu:mi> 'I break' continuing *uh4g^-néu-mi with the usual Greek remodelling of the infix.
We get no additional information from <agmós> 'fracture, cliff'. This word was probably formed within Greek, like many others, with the vocalism of the verb. We have normal grade in <agermós> 'assembly' and <keuthmós> 'hiding place', but the inherited root-vocalism with this suffix was probably /o/-grade as in <kormós> 'log' and <stolmós> 'clothing, dressing'. (Cf. Chantraine, Form. §§101-3.)
Greek <plé:sso:> 'I strike' has 2nd perf. <péple:ga> and 2nd aor. pass. <eplé:ge:n>, but the compounds <exepláge:n> and <katepláge:n> have a short root-vowel. This is remarkably similar to the behavior of <éa:ga>, <eá:ge:n>, and <kateáge:n> discussed above. Again I consider the simplest explanation to be that <eplé:ge:n> took the vowel of <péple:ga>, but the original zero-grade was retained in the compounds. If the PIE root was *pleh2g- as currently understood, the Proto-Greek /o/-grade should have *plo:g- (pace LIV), and the zero-grade *pla:g-. But if it was *pleh4g-, under the assumptions applied above to *weh4g^- (*-oh4- > *-a:-, *-Rh4- > *-Ra-), we get the PGrk /o/-grade *pla:g- (Att.-Ion.-Ep. *ple:g-) and zero-grade *plag-, agreeing with <péple:ga> and <-epláge:n>.
The short stem of Latin <nato:> 'I swim', frequentative of <no:>, could be explained similarly. If the root was *sneh4- and *-Rh4- produced *-Ra-, not *-Ra:-, in Proto-Italic also, *snat- would regularly result from *snh4t-. Likewise, several other roots reflecting *-Ra- as apparent zero-grade in Greek and Latin might be of this type. Lat. <lateo:> 'I lie hidden' and Grk. <lantháno:> 'I elude, escape notice' (2nd aor. <élathon>, 2nd perf. <léle:tha>) could be referred to a root *leh4- extended by *-t- in Latin and *-dH- in Greek.
The example of Sanskrit <vájra->, Avestan <vazra-> is the 12th of the 14 cited by Lubotsky, "Laryngeals before mediae in Indo-Iranian", MSS 40:133-8 (1981), in which a laryngeal appears to have vanished before an inherited voiced unaspirated stop. In this paper L. does not distinguish *h2 from *h4, and wherever he can determine the laryngeal, it is *h2 (or *h4). His form for 'break' is thus *weh2g^-. His explanation is that a laryngeal was lost in InIr before a voiced unaspirated stop plus another consonant. Apparent exceptions, in his view, resulted from later processes, primarily the thematization of originally athematic presents. It is worth noting that *sweh2d- and *pleh2g- are two of L.'s other examples. Since I now regard these roots as probably *weh4g^-, *sweh4d-, and *pleh4g-, I intend to examine the remaining examples for evidence of *h4.
Returning to *weh4g^-, it appears that Tocharian generalized the full grade <wa:k-> 'split, burst'. I do not know whether *h4 rather than *h2 can explain Hittite <wakkanzi> against <wa:ki>. If not, the expected alternation *wa:k-/*ukk- would have been unusual for a verb. The attested <wa:k-/wakk-> 'bite' might then be analogical after <za:h-/zahh-> 'hit, beat' and the like.
DGK