P & Q, and -pt- & -kt-

From: richardwordingham
Message: 14403
Date: 2002-08-20

--- In cybalist@..., Piotr Gasiorowski <piotr.gasiorowski@...>
> Albanian, incidentally, has /k/ for inherited *kW (yielding /s/ in
palatalising environments), thus testifying that in the only well-
known "aboriginal" language of the area there was no change of old
*kW into /p/. Note also that while e.g. Latin /-kt-/ gives Albanian /-
ft-/ (similar to Romanian /-pt-/ as in <opt>), _inherited_ *-kt-
gives Albanian -t- (as in <natë> 'night'). This shows that Balkan
Latin followed its own ways rather than mimic the developments found
in the local languages.

Surely Albanian <natë> 'night' shows the development of IE -k^t-,
which I would not expect to be like Latin /-kt-/ as late as the Roman
Empire. What happened to the cluster with 'plain' (uvular?) /k/, or
indeed /kW/, or indeed -kt- developed internally after satemisation?

Is there any evidence that Latin /-kt-/ > Balkan Latin /-pt-/ was not
a substrate-induced substitution comparable to the occasional
(reverse!) change /-pt-/ > /-kt-/ (e.g. Latin capti:vus > French
chétif, English caitiff) in Gallo-Romance? (How relevant is the
parallel development of the cluster -mn- from -gn-, as in Latin
lignum 'wood'? I'm sorry I can't remember the Rumanian word - it's
twenty years since I last saw a discussion of this particular change,
which was in the context of selecting matrices of distinctive
features to define phonemes.)