On Tue, 25 Nov 2003, tgpedersen wrote:
>
> In other words, the Caland system differs from the rest of IE in
> time, not space. How do you know that? Suppose I were to tell you
> that English has X-system, which is characterised by the suffixes -
> ate, -ation, -able which are the result of some very old derivations
> in English, but they are _not_ loanwords?
Then I would be astonished to see that you are ignorant of the obvious
source of the borrowing which in this case is such that no discussion was
ever necessary.
The IE Caland thing is different. You are in a sense asking me to let the
cat out of the bag before the time is really ripe. But I started this
myself. The core of the matter is that the morphological "replacements"
observed in the ramifications of the IE adjectives in what seems to be
their most archaic form are repeated elsewhere in the language. We do not
*only* find replacement of -e/o- by -i- in adjectives used as first parts
of compounds, we also see o-stems turn into i-stems when final in
compounds (Lat. insignis), and we see -i- turning up for e/o in enclitic
stems (*im, *id : *ésyo; interr. *kWó-/kWé- : indef. *kWi-), and we find
something very comparable in reduplications, e.g. *dhi-dhéH1-ti,
*dhé-dhH1-nti. And it is not only in adjective stems we observe reductions
of *-Ce/o- to *-e/o- (and further *-i-), we see something of the kind at
the end of compounds too (*newo-g^nH3ó-s for **-g^nH3-tó-s). Reduction of
vowels in the first member of compounds is not restricted to forms where
they are adjectives, cf. compounds in *n-, *H1su-, *tri-, *kWtru- which
are generally even accented (cf. Piotr's recent posting).
Therefore, if an adjective stem *H2rg^-ró- 'bright, swift' when used as
the first member of a compound is reduced to *H2g^i- (Gk. argí-pous
'swift-footed', Ved. rjí-s'van- 'having swift dogs', then this is not
necessarily just a matter of mysterious replacement of one allomorph with
another. It may be a simple event of sound change. For the first part of a
compound is reduced anyway, and the reduction of a composite morpheme -Co-
is plain -o- anyway, and the reduced form of -o- is -i- in a very old
layer of this language anyway.
Now, saying that the change of *-ro- to *-i- is here due to simple sound
change is nothing short of sacrilege in our field, and by Jove it hurts to
say it. For it belongs to the most famous results produced by the late
Jochem Schindler that the *-i- of these formations is akin to the *-i-
appearing in abstract formations like Lat. ravis 'hoarseness' : ravus
'hoarse', OCS zUlI 'an evil' : zUlU 'evil' (adj.). Still, I would say that
this is in reality rather the effect of the contrastive accent, i.e. pairs
like Gk. tomós 'cutting, sharp' : tómos 'slice, a cut', only formed in an
earlier period antedating a change of unaccented "thematic vowel" to /i/.
I have the highest respect for the combinations pointed out by Schindler
and his pupils and colleagues, but I simply fail to see the abstract
function of the i-forms into which adjective stems are sometimes
transformed. Why must it all of a sudden be "having swiftness-feet", when
"having swift feet" does just as well, and all other compound types use
the original part of speech unchanged?
I therefore say that the Caland alternation which is nowhere productive
and so must represent archaisms wherever it is met must be seen in
connection with what other examples there are of the same alternations
appearing throughout the language. And in fact the very same changes occur
in all corners of the language, showing that this is not a functional
phenomenon at all. There is certainly no "abstract substantive meaning" in
the -i- of insignis which has "replaced" the *-o- of signum in a compound
where the function of the noun has been pragmatically changed into an
adjective. This is the opposite of the ravus : ravis pair. Now, why is
that? I think it is very simple: In adj. *-ó-s : abst. *'-i-s and in
*sek-nó-m : *én-sek-ni-s the -o- is accented and the -i- is unaccented.
It appears to be common opinion that such very old matters in PIE are
always and entirely due to function. I guess the appeal this has to
scholars' minds is due to the fact that it is safer. A theory held
exclusively in functional terms cannot be checked and falsified, while a
claim of phonetic change mostly can. Sound changes are things we know how
to control, so little is accepted and everything is criticized; but
functional claims are above the law. A phonetic account need only go awry
on one minor detail to end up in complete discredit, whereas a functional
theory stands tall because there are no criteria by which i could be
disqualified. So functional theories rule unjustly.
Jens