Hittites ~ Phrygians ~ Balkan peoples?

[Sean Crist]

On Wed, 11 Aug 1999, petegray wrote:

>> an
>> algorithm developed to produce ...
>> the correct phylogeny, or family tree,

> Applied to languages, this is no better than the isoglosses you choose,

True.  The same is true when this methodology is applied to biological
species; the results are no better than the biological traits you choose.
In both cases, the characters you choose are a matter of professional
judgment.

> and
> this takes us back to an earlier debate where it became clear that an
> isoglossic approach fails to consider:
> (i) wave theory (which may underlie the satem/centum and RUKI isoglosses)
> (ii) language continua
> (iii) language convergence and Sprachbuende (is that word sufficiently
> anglicised yet for an English plural to be accepted?   If not, which German
> plural is the correct one - is it a language bunch (-bunde) or a language
> alliance (-buende)?)

These three are really one and the same.  To quote something that Don
Ringe once said when presenting this work, "Why do we make trees?  Because
we can."

Ringe et. al. are quite conscious of the problem you describe.  For
example, when they attempted to apply this methodology to the West
Germanic languages, the attempt "failed spectacularly".  Apparently, the
reason is that these languages/dialects developed in close contact with
each other, and innovations were shared in ways which cannot be modelled
in a Stammbaum.

In the case of the Indo-European tree that I gave, however, there wasn't
this kind of failure, with a few points of messiness such as the
placement of Germanic and of shared innovations between IIr and BS.

Mathematically speaking, the set of possible trees is a fairly small
subset of the set of possible wave diagrams.  If we can produce a tree
which is a fairly near perfect phylogeny, we should do so, since the
Stammbaum model is the more constrained one.

> (iv) time of attestation.

I don't understand the problem here; could you say more?

  \/ __ __    _\_     --Sean Crist  (kurisuto at unagi.cis.upenn.edu)
 ---  |  |    \ /     http://www.ling.upenn.edu/~kurisuto/
  _| ,| ,|   -----
  _| ,| ,|    [_]
   |  |  |    [_]