The Indo-European Hypothesis [was Re: The NeolithicHypothesis]

[X99Lynx at aol.com]

In a message dated 4/2/99 9:59:31 PM, Larry Trask wrote:

<<The family-tree model is *not* universally accepted as the only possible
model of the rise of languages.  But it *is* universally accepted today
as the best model of the rise of the IE family of languages....we can
reconstruct so much intricate and complex grammar for PIE that it simply
*must* have existed.  A "language mixture" scenario is just not consistent
with the elaborate grammatical system which can be reconstructed for PIE and
which is substantially preserved in at least the earlier IE languages.>>

Without contradicting any of the above, I think it might be worthwhile to
point something out.  The divergence/family tree model may explain the "rise"
of the IE family, but it doesn't completely explain how the IE groups got to
where they are today.

Two other factors are worth consideration.  One is the possible removal of
early "branches" that could affect the accuracy of reconstruction.  The other
is the convergence that occurred when later dialects were standardized or
heavily borrowed from.

For example in a message dated 3/15/99 5:21:05 PM, mcv at wxs.nl wrote with
regard to early PIE dialects:
<< ...any initial dialect gradients that came into being have been
destroyed by later language spreads.  Celtic has been largely
swallowed up by Romance and Germanic, the Slavic and Hungarian
spreads have replaced whatever gradients there were in Eastern
Europe with new dialect gradients.  Etcetera.>>  This kind of "swallowing up"
might lead us to accept the trait of an intermediate language as the trait of
an earlier ancestor.

Another example is John Green's statements in TWML's that Romance could not
have evolved directly out of standardized Latin, while yet citing the
extensive borrowings from Latin, which implies that at some point Romance and
Latin diverged and converged.  "Family-tree classifications,...give only
crude indications of relationships in Romance and tend to obscure the
convergence brought about by centuries of Latin borrowings and criss-crossing
patterns of contact [which resulted] in a high degree of lexical overlap in
the modern Romance languages."

Another example is the observable convergence of Russian and OCS.

It makes sense that such convergences, if mistakenly seen as commonalities
resulting from a common ancestor or PIE itself, could skew reconstruction
away from a true triangulation.  This would be especially true in large
wholesale characterizations such as centum/satem, where convergences of later
diffused terms of trade or uniformity might give a false impression of early
ancestry.  Given the massive amount of words involved however it would seem
that numbers alone favor the reliability of reconstruction in general.

Regards,
Steve Long