On Fri, 14 Sep 2001 22:24:57 +0200, "Piotr Gasiorowski"
<
gpiotr@...> wrote:
>> The argument for Common Slavic */r./ is exactly equivalent to your argument for PIE */r./. Why would Slavic */r./ not simply continue PIE */r./?
>
>Because it comes in two phonemic flavours -- "palatal" and "hard" (same for syllabic laterals). The only reasonable explanation of their existence I can think of is that the palatalised syllabic liquids derive from pre-Slavic *iR and the non-palatalised ones from *uR -- sequences evidenced by Baltic.
But why the split into /iR/ and /uR/ (etc.) in the first place?
I have argued elsewhere for a pre-PIE split of the consonantal
phonemes into plain, labialized and palatalized varieties (presumably
after the loss of pre-PIE short /i/ and /u/), so that PIE may have had
three flavours of syllabic /r./ etc. In unstressed position:
pre-PIE PIE Baltic Slavic
/CarC/, /CraC/ -> /CrC/ -> /CurC/ /CrC/, /CUrC/, /CrUC/, /C&rC/
/CirC/, /CriC/ -> /Cr^C/ -> /CirC/ /CrC/, /CIrC/, /CrIC/, /CirC/
/CurC/, /CruC/ -> /CrWC/ -> /CurC/ /CrC/, /CUrC/, /CrUC/, /C&rC/
Obviously, the original distinctions have been blurred, because of the
unstable character of such secondary articulations as labialization
and palatalization in general (cf. the sporadic and arbitrary loss of
palatalization in Polish). That's why we have Russ. <volk>, with hard
/l./ and Pol. <wilk> with soft /l^./, where Baltic <vilkas> (and OCS
<vlIkU>) seem to suggest Polish has here maintained the original
quality (while in <mërtvyj> vs. <martwy> the opposite seems to be the
case).
A form like Pol. <martwy> cannot be derived from Slavic *mUrtvU-, so
for Common Slavic we have to reconstruct syllabic */r./ and */r^./ in
any case.