No Proto-Celtic?

[David L. White]

        It looks like I am going to meet my deadline, so I can afford to
start re-indulging my vices, among which this list is chief ...
        Two words or expressions are being thrown around here in a way that
makes me a little uncomfortable.  The first is "lingua franca".  This has
(if we are not case-sensitive) three meanings:  1) a real language, "Lingua
Franca", which is (or was) a sort of Mediterranean Romance semi-creole, 2)
an international business or trade language characteristic of a certain
place/period, as for example Akkadian, and 3) something in the pidgin/creole
range.   Meanings 2 and 3, though they can overlap, do not necessarily do
so, and due to inherent ambiguity the expression "lingua franca" is probably
best avoided, as considerable confusion is likely to be sown.
           Second is "convergence".  I believe the impression some would
like to create is that when mutually unintelligible languages meet they can
converge or "mix" and form a new language, which can then mislead
circularly-reasoning linguists into wrongly imagining a proto-language.
Whether mixed languages occur is a good question, though there can be little
doubt that about 90% of proposals along those lines (often by quite
unsophisticated missionary linguists) have turned out to be wrong (for
examples of this sad syndrome, see the introductory chapter of Welmers
"African Language Structures").  But be that as it may, "convergence" is, I
believe, more properly used to describe the sort of situation detailed by
Gumperz and Wilson (as well as Emeneau?) in Kupwar India, where due to
extensive routine bilingualism the local versions of Marathi and Kannada
have undergone almost complete syntactic (and phonological?) convergence, to
the point where they are virtually the same language with different words.
(I have heard that the same might be said of Rumanian and Bulgarian.)  But
they have _not_ "mixed".  The lexicons (last I heard) are kept entirely
separate, and it is never at all problematic to determine which language
someone is speaking.  If this is the sort of situation that is supposed to
have led to "mixed" "lingua francas", and so to the "illusion" of
proto-Celtic, proto-Uralic and so on in the misty past, then it is surely a
good question why the result in the well-studied and not at all misty
present has been so utterly different.  I do not know, but I can guess:
mixed languages by and large belong to that semantic category that has been
called "fantasies of the ignorant".   There may be a few out there, but no
example I have seen is convincing.  The only ones I am aware of that are
even just-barely-possibly convincing involve cases of a language dying (e.g.
Anatolian Greek), which is not exactly the most promising way for a new
language family to get off the ground.

Dr. David L. White