Any family reconstruction must be based
first of all on regular sound correspondences in cognate morphemes and words --
there is no other secure starting-point. Working with semantics comes later,
when we have a firm formal foundation to build on. Lexical
similarities don't mean much in themselves, especially between neighbouring
languages -- they may be due to borrowing. Take an English dictionary: for any
5200 entries there will be at least about 4000 loans from French or Latin. This
doesn't make English a Romance language. Proto-Dravidian is harder to
reconstruct that PIE for the very good reason that IE languages are more
numerous and their historical attestation is exceptionally good. But
the genetic unity of the Dravidian languages is not doubted by any
serious linguists. You confuse areal convergence with genetic relationship.
Both produce affinities between different languages -- but they are DIFFERENT
KINDS OF AFFINITIES.
Piotr
----- Original Message -----
Sent: Monday, September 24, 2001 5:21 PM
Subject: [tied] Re: Dravidian in Persia?
I hope linguists see through this. If 4000 of the so-called
5200 etyma have IA cognates, how can Burrow and Emeneau claim that they have
reconstructed Dravidian Etyma? What is the rationale for claiming borrowing from
IA > Dravidian or vice versa? Had grammar of
Proto-dravidian been
constructed? Working with sounds is one thing and working with semantics is
another.