--- In
cybalist@yahoogroups.com, Piotr Gasiorowski
<piotr.gasiorowski@...> wrote:
> But it has been done for a long time. Romanicists have always
emphasised the difference between Proto-Romance and written Latin. To
do the reconstruction literally "without any knowledge of Latin" is
of course impossible, and a historical linguist well-versed in the
Romance languages but completely ignorant of Latin is hard to find,
even nowadays :-). We can't erase what we know about Latin from our
brains for the purpose of the experiment, but as far as I can see
Romance linguists have no trouble avoiding any bias in this repect
and don't patch the gaps in the reconstruction with Classival Latin
material. The differences are quite revealing, but don't necessarily
have much bearing for the actual/reconstructed PIE question, apart
for the trivial observation that all comparative reconstruction is of
necessity partial and very incomplete. Classical Latin is not the
direct ancestor of Romance, so by comparing it with Proto-Romance we
don't actually test the reconstruction. Instead, we learn more about
Latin and its internal differentiation. The PIE counterpart of
Classical Latin may have never existed, for all we know. Unless some
PIE documents are found (and I don't think that's a realistic
expectation), all that we shall ever know about PIE is what
comparative analysis can yield.
>
> Piotr
>
In some article I just read (I can find it if anyone wants the
reference) they used an algorithm to compute "distances" between IE
languages and to cluster them. Invariably, reconstructed PIE ended up
in the Romance cluster. When they removed the Romance languages, PIE
ended up in a cluster with Greek. So much for the impartality of PIE
reconstruction.
Torsten