Re: [tied] IE classification: a lexicostatistical experiment

From: fournet.arnaud
Message: 49899
Date: 2007-09-15

 
----- Original Message -----
From: mkelkar2003
To: cybalist@yahoogroups.com
Sent: Wednesday, September 05, 2007 10:39 PM
Subject: [Courrier indésirable] [tied] IE classification: a lexicostatistical experiment

A fourth purpose (of this monograph) is to present some lexicostatistical
results which bear on the controversial question of the higher level
groups in
Indoeuropean. The answers to these questions reached by
lexicostatistics do NOT
bear on the validity of lexicostatistics. Rather, to the extent one
believes
this method has been validated by the results in lower level
classification and
by other evidence, the results produced by the lexicostatistical
method simply
provide one more view. The most notable contribution to the higher
level groups
are as follows:

(1) there is clear evidence of a Baltoslavic branch;

A.F : obviously so. It can/must be used as a backbone for any classification method.

(2) evidence is lacking for an Italoceltic branch;

A.F : I definitely disagree.

what about the future -bo, genitive in -i and some words like wespertilio "evening flyer" that exists only in Italic and Celtic. Etc.

(3) there is evidence that
Romance, Germanic, and Baltoslavic are most closely interrelated among
the distinct branches of Indoeuropean,

A.F :  I definitely disagree.

Connection of Balto-slavic and Indo-Iranian plus Germanic in a sub-groupe is ok.

Connection of this sub-group with Romance is absurd.

Italic and Celtic share both a set of distinctive innovations and a set of common conservations, which means common period and separation from the other set of innovations shared by Balto-slavic and Indo-Iranian plus Germanic, such a genitif in -os-yo.

thus suggesting, though the
evidence is far
from conclusive, that these three divisions form a single separate
branch, for
which the term Mesoeuropeic is introduced. Whether Celtic can
be attached to this group so that it conforms with the Northwestern
Indoeuropean group suggested by
Meillet (1922) is not indicated by our lexicostatistical data, though
there are
reasons for believing that such an association is not contraindicated.

As Indoeuropeanist recognize, there are strong indications of much
dialectalization at the highest level of Indoeuoropean, so that no
classification whatever can be expected to provide a good description
of the
historical reality. These effects agree well with the traditional
hypothesis of
a highly differentiated, deeply dialectalized Proto-Indoeuropean.
Certainly
this conclusion would be consistent with the many contradictory
classifications
for which weak evidence, of a traditional nature can be found. While
these
contradictions provide only a negative kind of evidence for diffusion
effects,
multidimensional scaling methods such as that of Black (1976) may be
able to
provide positive evidence (Dyen, Kruskal, and Black 1992, pp. 5-6,
parenthesis
added, emphasis in the original)."

"Perhaps the most notable discrepancy between the lexicostatistical
classification and generally accepted hypotheses is its failure to
distinguish
an Indoiranian group. The difficulty is not on the Iranian side, for the
Iranian Cluster has its highest percentage (18.1) with the Indoaryan
Cluster.
However, the latter group exhibits like percentages with the
Mesoeuropeic Hesion
(18.6) and the Baltoslavic Subfamily (18.1). An *Indoiranian Hesion
would have
been established if the Indoaryan-Iranian percentage had been at least
21.1
(i.e., 18.6+2.5).

The evidence that the Indoaryan and Iranian languages continue a
secondary
protolanguage, Proto-Indoiranian, is evaluated by Meillet (1922.251)
as follows:
"..Indoiranian offers an entire series of peculiarities of detail
which are
found nowhere else and which continue from the common period peculiar
to that
group." This conclusion is based on a study of older stages of
Indoaryan (i.e.,
Sanskrit) and Iranian (i.e. Avestan, Old Persian) and is difficult to
challenge. Thus the lexicostatistical classification in this
monograph misses a
group that clearly existed (Dyen, Kruskal, Black 1992, pp. 47-48)."

"It is important to realize, however, that the failure to find the
Indoiranian
group is not due to the lexicostatistical method as such, but is due
rather to
the self-imposed restriction to CONTEMPORARY word lists (Dyen,
Kruskal, Black
1992, p. 48, emphasis in the original). "

Perhaps the explanation for the discrepancy lies in the fact that the
entire
Indoiranian area, particularly the Iranian area, was increasingly
subject to
invasion and domination by nonnative groups from the end of the first
millennium
B.C., the Iranian area by speakers of Arabic among others, and the
north Indian
area by Iranian speakers led by Mongol rulers. These conquests could
have led
to intimate borrowing on a sufficiently large scale to have considerably
deflated the percentages between Indoaryan and Iranian (Dyen, Kruskal,
Black
1992, p. 49)."

Dyen, I., Kruskal, J. and Black, P. (1992). An Indoeuropean
classification: a
lexicostatistical experiment. Transactions of the American
Philosophical Society, 82 (5), Independence Square, Philadelphia.

M. Kelkar