On Sun, 27 Jan 2002, Miguel Carrasquer Vidal wrote:
> On Sun, 27 Jan 2002 23:22:40 +0100, "Piotr Gasiorowski"
> <gpiotr@...> wrote:
>
> >> Well, Lith. N. óbuolos, Latv. âbuols point to a paradigm *h2ábo:l(s),
> *h2bél-os ~ *h2b-l-ós, which would have given Slavic *a:ba:l- / *obl-.
> >
> >Well, neither is attested in Slavic, so one must again resort to a
> "compromise" theory, which is always an embarrassment. Without liquid
> blocking, *h2bl- > *abl- > *a:bl- follows straightforwardly. Note that
> using the same kind of reasoning you explain a long vowel in a putative
> length-blocking environment and a short vowel in a non-blocking context.
> This is excessive explanatory power -- it immunises the liquid blocking
> theory against falsification, since any counterexample can be explained
> away.
Yeah, right. I have seen my name mentioned with this matter, but I have to
hand the credit to a former student of mine, Toshihiro Shintani who left
the field of IE studies som 15 years ago to become a professor of Danish
in the university of Osaka. It was Toshihiro who discovered a phonetic
rule he did not like himself at all, namely the suspension of Winter's
lengthening in accented monophthongs. In diphthong-like combinations, WL
apparently works with no restrictions. My distinguished Japanese student
wrote a paper about it in a Working Papers series from the department we
used to issue in Copenhagen, and much to my disappointment he never
bothered to get it properly published anywhere. I later worked on the
matter again (in a paper read at a congress in Kiel in 1991 in the
presence of Winter himself), finding Toshihiro's accent-based restriction
to be valid, but adding the restriction of non-lengthening before stop +
sonant. None of it works 100 % - and how could it? This is a rule common
to Baltic and Slavic, but Baltic and Slavic do not always have the same
forms, ergo the preforms the rule worked on must have differed from some
of the forms we find attested. Here nihilistic lecturing on method should
be administered with prudence; the formulation of rules has not departed
from the ambiguous cases. My paper was published in a younger series of
Working Papers from Copenhagen and later reprinted in my Selected Papers
in 1999.
For 'apple' the Baltic forms are fine, while the Slavic forms do not
conform to the restriction (but the extremely productive suff. -Uko- can
have been added at any point). For vydra I see no salvation, that looks
like a genuine counterexample. Still the cases of lack of lengthening very
plainly do come in clusters, one type being -VDR-. I have no idea at the
present moment what subrule should be added to the conditioning, if any.
Making it a post-PIE vrddhi derivative would do the trick, but that is
unattractive in view of Lat. lutra which has a short /u/ reflected in
Romance. Still, *udró-s, coll. *ú:draH2 could be what is reflected in
BSl., with or without support from elsewhere. Then, as in 'apple', the
long /u:/ form must have been generalized, so that also Lith. údras
(mobile) has taken it (from the immobile údra). I simply worked with the
indices of Fraenkel's and Vasmer's etymological dictionaries, looking up
every word compared to anything that had an unambiguously unaspirated
media anywhere, and what I found was overwhelmingly supportive of Winter's
rule - with one or two restrictions for which subrules could be
formulated.
The point my Japanese former student did not like was of course the
instinctively bewildering observation that lengthening does not occur
where it is being supported by the accent; Toshihiro himself suggested
that it worked only in the imediately pretonic syllable, and I found that
to be right. I understand that Proto-Slavic *only* preserves length in the
syllable immediately preceding the accent. This may then reflect a common
funny feature of the BSl. group.
Note: the -t- of Latin lutra is no problem if IE *dr yields Latin /tr/
by regular development. It's a funny subrule of very long standing, but it
has a number of very attractive etymologies in its favour and none that I
know of against it. The l- looks like a popular precursor of the article
(but then the -t- could also be "populaire").
Jens