--- In
cybalist@yahoogroups.com, Mate Kapoviæ <mkapovic@...> wrote:
> Are you sure that 3rd palatalization would indeed work here even if it
> were already *i?
Nobody is ever "sure" of anything in historical linguistics. We are
devising stories that attempt to account for the attested facts in
terms of certain ground rules. We call those stories hypotheses or
theories. We're never completely sure there isn't a better story
waiting around the corner. We are never completely sure the ground
rules are to be trusted 100%. We are never completely sure the facts we
try to account for deserve the proud title of "facts".
> I ask this because there are man examples where we get
> double forms, so-called "exceptions" etc. Many people consider 3rd
> palatalization as "sporadic", [...]
Those people are few and IMHO they are wrong. The relevant material is
accounted for almost perfectly if it is assumed that palatalization
took place only after high unrounded front vowels (short, long and
nasalized *i) not followed by high rounded vowels (short and long *u)
at a period after the monophthongization of diphthongs but before the
uncoupling of long and short vowels and the delabialization of long and
short *u.
This is not only phonetically simple, but accounts for all the examples
with the single exception of *lice* as long as that is regarded as
reflecting *leik-.
The assumption that palatalization was blocked by "long and short *u"
accounts for the most important fluctuations because it sets up
alternations that were analogically levelled differently in different
areas. The Russian-Belorussian area, for instance, always generalized
unmodified stem-final g in nouns, for instance, so that you get
Russian -jaga vs. Polish jeNdza 'witch'.
Most of northern Slavic generalized the unmodified velar in derived
imperfectives in -ati, which is a highly productive category anyhow, so
what is all the fuss about?
The trouble is that slavists are not used to dealing with material that
does not immediately yield its secrets. Since they are raised on
blindingly obvious sound laws, they tend to panic when things are a tad
less obvious. If OCS has *dvidzati* 'shift' (modified velar) but Old
Czech *pozdwihati* 'raise' (unmodified velar), they tend to think the
universe is conspiring against them and "something must have happened",
let us *pulleease* forget about our ground rules and start operating
with sporadic (i.e. untestable) rules. Indo-Europeanists would laugh
their collective asses off at the kind of things slavists panic
about. ;)
Willem