From: dgkilday57
Message: 70476
Date: 2012-11-20
>Anatolian and Tocharian are ruled out, but Venetic would work, with vocalization of PIE syllabic */r./ to *-or- as in Latin. The ending of *okti recalls the protoform of Lat. <ursus>, whose vocalism indicates an early borrowing from Sabine, and whose assibilation points to the adj. in *-jo-, not the noun, i.e. (formally) PIE *h2r.tk^jos 'bearish one' > Italic *orktjos > Sab. *or(c)s(o)s, possibly North Ven. *orktis > MPF *okti-. No previous */a/ is required in 'bear' by this route.
> > > > On the other hand, beside <olut> Jouppe cites Fi. <ohra> 'barley', Middle Proto-Finnic *os^tra- referred to Balt. *as^tra- < PIE *h2ak^- 'sharp'; Fi. <ohdake> 'thistle', MPF *os^ta-kkes^, root referred to Balt. *a(k)s^ta-; and Fi. <onki> 'fishhook' referred to Gmc. *angan-.
> > > >
> > > > If these are IE loanwords, they could equally well be from North Venetic with retained */o/, namely <ohra>, MPF *os^tra- from Ven. *hozd- cognate with Lat. <hordeum>; <ohda-ke>, MPF root *os^ta- from Ven. *ozd- 'branch' (cf. Grk. <o'zos>, HG <Ast>, etc.); and <onki> from Ven. *onk- 'crooked' (cf. Lat. <uncus>).
> > >
> > > The usual explanation here has been to date these loans as sufficiently old that they would precede *o > *a in Baltic/Germanic, an approach which has been used for several other odd IE > F substitution patterns as well.
> >
> > That would seem to rule out PIE *h2ak^- in these words.
>
> On re-checking some of Koivulehto's writing he has actually suggested an initial phonetic develoment into an open labial vowel *o, *a > *å and that (some of) these loans would be later than this development & erlier than *å > *a. Which seems to require awfully close timing, but recalling another *o-loan yet: at least *okti "bear" from *h2rtk´os (exact loangiver not identified AFAIK) appears to require *o being derivable even from previous IE *a anyway; and must also be assigned to a layer distinct from the one with *H- > *k-.
> > Mea culpa. Latin <hordeum> is referred to PIE *g^Hr.zdH- and under my assumptions North Venetic should have had *horzd-. The question is then whether the cluster would have undergone metathesis in MPF, *os^tra- for *ors^ta-.Or perhaps *-rzdr- if the NVen protoform was *horzdr(o)m from *g^Hr.zdHro- or *g^HorzdHro-. If samprasa:rana (as in Oscan) produced NVen *orktis from *orktjos, perhaps NVen syncope of the *-o- in ordinary */o/-stems led to MPF acquiring *-a- as an Ersatzstammvokal rather than the original *-o-. This would explain the retention of */o/ in roots but not finals.
>
> That's a little better, yes. There is no precedent for a metathesis of this kind (rather CCC clusters tend to drop any initial resonants, as also in "bear"), but since a few other instances of *-s^tr- must be posited, it's not unimaginable this group would have managed to attract a loan original with *-rzd-.
> It may be worth mentioning that the Baltic etymology for "barley" has a similar weakness in that the cluster should probably rather be reconstructed as *-str- than *-s^tr-, given the reflexes -sr- in Karelian, -zr- in Votic, -z- in Veps. -hr- in West Finnish looks like an innovation parallel to a known change *-sl- > -hl-. Still, this does not help in assigning this layer to North Venetic: if we here had a substitution *-rzd- > *-str-, then surely "thistle" should have *-zd- > **-st-?If *os^ta- 'thistle' is from North Venetic, my earlier suggestion relating it to Greek <o'zos> 'branch' etc. is probably wrong. Another possibility is connecting it with Grk. <oxu's> 'sharp', <o'kris> 'crag', Lat. <ocris> 'mons confragosus', Umbrian <ukar> 'citadel', with the NVen protoform *okst(o)s, parallel to the presumed Baltic *a(k)s^ta-, leading to MPF *os^ta-. In this view the sandwiched sibilant became *s^ (allophonic in NVen, phonemic in MPF), admittedly an ad-hoc makeshift, but it avoids a double outcome of *-zd-.
> > > OTOH it is not clear to me if this particular pattern can be divorced from the so far basically unexplained issue of *o-loans from Indo-Iranian into the western Proto-Uralic dialect(s). This would include at least:With *s'ola we also have the matter of PIE *-l- retained as *-l-. Is that expected from Iranian loans? Cf. *ora above.
> > >
> > > *s^oj-ta- "to take care of" < Iranian *sca:ya:-
> > > *poc^a(w) "reindeer" < Iranian *pacu < *pek´u
> > > ---
> > > *ons´a "part" < *anc´a- < *h2onc´o-
> > > *ora "thorn, awl" < *a:ra: < *e:la:
> > > *orja "south"; "slave" < *a:rya-
> > > *s´ola "intestine" < *c´a:la:- < *k´olo-
> > > *s´orwa "horn" < *c´rva < *k´erw-o-
> > > *tojwo "hope" < *da:yva- < de:ywo-
> > >
> > > possibly also the likes of:
> > > *orpa "orphan" << *orbho-
> > > *pors´as "pig" << *pork´os
> >
> > Would early Proto-Baltic (with retained */o/) work here, including Osthoff's shortening but no Brugmann's lengthening?
>
> Possibly for some of them, but at least *ora and *orja look patently II in origin, similarly the substitution *k´ >> *c^ in *poc^a seems to point to Iranian in particular? A couple examples might run into a lack of an attested Baltic o-grade reflex too, and a number of cases here (*s´ola, *s´orwa, *ons´a, *orpa) have a distribution extending to Ugric where no Baltic loans are otherwise known nor should be expected. Also about as many of this list's words are found in Permic and I'm only aware of Finnic-mediated Baltic loans known from there.
> I have the impression a full reinvestigation of the big picture of the oldest IE loans into Uralic is necessary one of these days.Yes, and the difficulties cannot be resolved simply by positing one or two additional branches of IE as sources. Since much of the research in this area has been published in Finnish, obviously I need to acquire a reading knowledge. Merely examining the comparative charts in Finnish papers, like an illiterate child looking at the pictures in a storybook, is insufficient for serious work.