Re: HORSA vs. EXWA

From: Joao S. Lopes
Message: 70748
Date: 2013-01-21

Why unexpected S? Shift ursus < *urxus is regular, isn't it?

JS Lopes



De: Tavi <oalexandre@...>
Para: cybalist@yahoogroups.com
Enviadas: Segunda-feira, 21 de Janeiro de 2013 12:35
Assunto: [tied] Re: HORSA vs. EXWA

 
--- In cybalist@yahoogroups.com, "Richard Wordingham" wrote:
>
> > NEC *XHVr[ts´]V 'marten; otter'
> > Altaic *karsi 'fox, marten'
> > Unfortunately, this word isn't "PIE" (i.e. Kurganic) at all,
>
> > > The evidence says otherwise.
>
> > Not "the evidence" but the *ortodox theory* of which you're such a
fervent supporter.
>
> The evidence is:
>
> Sanskrit _r.'ks.a- 'bear'
> Avestan _ars^a_ 'bear'
> Ossetian _ars_
> Armenian _arj_
> Albanian _ari_
> Greek _árktos_ (later _árkos_)
> Middle Irish _art_
> Welsh _arth_
> Gaulish _Deae Artioni_
> Latin _ursus_
> Hittite _hartka_ (presumed to mean 'bear')
>
> This looks pretty 'Kurganic' to me.
>
But "Kurganic" isn't synonym to "IE". In fact, Anatolian (Hittite) is
the *least* Kurganic of all the attested IE languages, having a more
archaic morphology and phonology than the rest (although apparently some
IE-ists seem oblivious to that). In my own interpretation of R. Adrados'
model (somewhat modulated by Villar), Kurganic would be the direct
ancestor of Greek-Armenian, Indo-Iranian and Dacian (Albanian), and
possibly Celtic as well, but only a *superstrate* to the rest of IE
languages except Anatolian.

This vision is mostly based on morphology, although there're also
lexical implications. While the mainstream PIE model is quite accurate
with regard to Kurganic, it doesn't work so well on the older layers.

> A cluster -rtk- or -rtk^- invites irregular development,
> though the development here doesn't seem so irregular.
>
Celtic and Hittite have /t/.
Greek has /kt/.
Latin, Iranian, Armenian and Albanian have /s/ ~ /S/ ~ /j/.
Indic has /ks./.

These outputs are rather strange for an ortodox "thorny cluster".

> The greatest irregularity is the initial /u/ of Latin _ursus_, which
appears to lack an explanation.
>
Not only that, but /s/ is also *unexpected* here (assuming it's the
reflex of *k'), because Latin is a "centum" language. Hence the
hypothesis of borrowing from another IE language is plausible.

From the internal IE evidence along with external data, I think the IE
word for 'bear' comes from an older layer than Kurganic and had a
palatal(ized) affricate similar to one reconstructed for NEC instead of
the "thorny cluster" of the ortodoc theory.