From: Tavi
Message: 68855
Date: 2012-03-09

--- In, Piotr Gasiorowski <gpiotr@...> wrote:
> > - The IE word for 'bear' is very difficult to reconstruct as a PIE root
> > because of its phonetic irregularities.
> > - This word has a likely NEC cognate.
> What is so irregular about it? It's true that Arm. arj^ and Alb. ari are
> hard nuts to crack, but only because our knowledge of the early history
> of those two particular branches is very imperfect. In Hittite, Latin,
> Greek, Celtic, Indic, Iranian (plus possibly a trace attestation in
> Baltic) we find regular reflexes or *^os.
But the problem lies precisely on this clumsy PIE reconstruction, as we got an internal cluster *rC where C has various reflexes: t, s, s^, kt, ks\ which can't be properly explained by the traditional model. Fortunately, thanks to external comparison we can posit a sibilant affricate here.

> > - Germanic has a different 'bear' word linked to other IE roots (either
> > 'wild animal' or 'beaver'/'brown').
> Precisely. So it quite simpoly *isn't* the original 'bear' word, just as
> English <deer> (though derived from a PIE root) does not reflect a PIE
> (or even PGmc.) 'deer' word.
I'm afraid this is a very different case, so it can't be a valid example.

> By contrast, *^o- is represented in
> Anatolian and in several branches of Core IE, which to all intents and
> purposes demonstrates its PIE status.

> > - The IE word 'horse' is irregular in some languages such as Hittite and
> > Greek, pointing to they being loanwords.
> The Greek reflex is only a little problematic, not enough, IMO, to
> warrant invoking an external source.
In all likelihood, the Greek word must have been borrowed from another language.

> Everywhere else (Latin,
> Indo-Iranian, Baltic, Germanic, Celtic, Tocharian) the reflexes are as
> regular as anything. I used to believe the Anatolian horse terms were
> Mitanni Aryan or Iranian borrowings, but now, having read Melchert on
> the matter, I have to concede that the Luwian and Lycian reflexes of
> *h1ek^wos (> PAnat. *ák^k^wos) are completely regular and inherited, not
> borrowed.
In my model, the word 'horse' belongs to a more recent layer (i.e. superstrate) than other IE words. Some IE-ists have put too much emphasis on this layer, up to the point of identifying it as the "true" PIE.

IMHO the traditional PIE needs a major revision because it has become obsolete.

> > - This word has a likely NEC cognate.
> > - Horses were first domesticated in the Pontic-Caspian Steppes.
> > - There're other 'horse' words in IE languages, including one which is
> > exclusive of Germanic.
> There are also exclusively English words for 'pig', 'dog', 'badger' and
> many other animals, replacing older terms. Does this fact support any
> fairy-tales about long-range contacts?
Once again, the example you're quoting is irrelevant for the matter. Archaeological data tell us when and where the horse was domesticated, and linguistic evidence gives us a 'horse' word native to that area. Unfortunately, it looks like most IE-ists are too busy (or perhaps too lazy) to look at any data outside it own field.