From: Tavi
Message: 68830
Date: 2012-03-08

--- In, "Brian M. Scott" <bm.brian@...> wrote:
> You missed the point completely, which suggests that you
> don't understand how comparative reconstruction works.
> > I know perfectly how the comparative method works,
> This is far from evident.
See below.

> PIE is by definition the most recent common ancestor of
> the IE languages, as best we can reconstruct it.
> > According to this definition, PIE would be closer to a
> > Platonic ideal than to an attainable goal.
> It's a rather basic definition.
No, it's a HYPOTHESIS. IMHO this "most recent common ancestor" (it it actually existed) would be quite remote in the past (at least Mesolithic and possibly Upper Paleolithic), more like the "Proto-Nostratic" commonly depicted by Nostraticists.

> > IMHO the hypothesis of a monophyletic IE family (i.e. all
> > IE languages branching from a tree with PIE at the top) is
> > inconsistent with linguistic data, although surely it was
> > the only available when neogrammarians first formulated it
> > in the 19th century.
> Oh, the historical facts are obviously more complicated than
> simple branching; that's obvious even just within IE
> subfamilies. It's also clear that the IE languages do form
> a family.
Yes, but being a family doesn't imply the traditional single-tree model is still valid for IE.

> > But IMHO they can be explained as different reflexes of a
> > former sibilant affricate like the one we find in PNEC.
> > I've simply extended the comparative method outside the IE
> > family to help us understand what you regard as "purely
> > internal matters".
> Because they *are* purely internal matters.
Only that they can't be properly explained within the traditional model, thus showing its inconsistence.

> > I also don't think we have to demonstrate a genetic
> > relationship PRIOR to accepting cognacy, as this is
> > utterly inconsistent. We can only posit a genetic
> > relationship from a mass of cognates along with
> > predictable (I prefer this term to "regular") sound
> > correspondences, and nobody would do so with a single
> > cognate.
> Don't be silly. The word 'cognate' *means* that there is a
> genetic relationship. When you say that A and B are
This might be true for the words themselves, but not necessarily for the languages involved.

For phonetical reasons, I think *h2ºrtk´-o- 'bear' is actually a loanword and not a native IE word. So as far as the comparative method goes, there's no reconstructable PIE word for 'bear' besides the one found in Germanic. Of course, this doesn't necessarily mean that bears were unknown to the speakers of the proto-language, only they hadn't a specific word for them (*g´hwe:r- 'wild animal'). But at a later chronology a word 'bear' was introduced into most IE branches, possibly for trade or religious reasons.

> Any possible relationship between PIE *h2rtk^o- and PNEC
> *XHVr[ts']V becomes relevant only when those
> reconstructions become part of the evidence for a common
> ancestry for PIE and PNEC. This, of course, is
> necessarily preceded by their convincing reconstruction
> from their putative reflexes, which is a family-internal
> matter.
> > This corresponds to the isolacionist model held by most
> > IE-ists but not by macro-comparativists.
> It has nothing to do with 'models'; it's simply a matter of
> methodology.
Then IMHO the comparative method has been incorrectly applied for the IE family.

> And while there are certainly exceptions, a
> great many long-rangers are methodological dunderheads;
> Ruhlen, Bengtson, and Starostin come to mind immediately.
Then I must be one of these exceptions. :-)