The "Slava" system, as revised and recently
canvassed by Dolgopolsky, is indeed Gargantuan:
25 stops and affricates (3 rows
of eight + glottal stop)
12 fricatives (2 rows of 5 + pharyngeal and
glottal fricatives)
13 sonorants (5 nasals, 3 laterals, 2
rhotics, 3 semivowels)
7 vowels
---------------
Total: 57 phonemes, including 50
consonants
The reconstructed prosodic structure of
content words can be represented thus: (CV.)CV(C).CV, where dots stand for syllable boundaries and optional
material is parenthesised. Actually most roots are disyllabic, but the final
vowel is left unspecified more often than not.
The phoneme inventory is twice the
average size of PIE reconstructions and nearly three times as big as the
standard Proto-Uralic, Proto-Altaic and Proto-Dravidian systems. Even typical
Kartvelian and Afro-Asiatic systems are far less baroque (e.g. Georgian has 28
consonants and five vowels). Some North Caucasian languages could successfully
compete with this version of Nostratic, but (1) North Caucasian languages are
not included in Nostratic, and (2) those with extremely large consonant
inventories usually have just two vowels (low : non-low).
The postulation of a parent system so
much more complex than any of its daughters raises the suspicion that
protosegments have been artificially multiplied in order to obtain a
satisfactory number of matches. This suspicion is reinforced by the following
facts:
(1) Many of the proposed correspondences
have very few instantiations; this betrays their ad hoc character and the
spurious character of such matches.
(2) The relative frequency of protosegments
in reconstructions does not seem to correspond to their typological markedness
-- e.g., the inventory contains a plethora of coronal fricatives and
affricates (no fewer than twenty!), but plain *s is extremely rare.
(3) Dolgopolsky employs -- very liberally
-- a large number of "uncertainty symbols": unspecified vowels, unspecified
consonants, cover symbols for classes of segments, multiple possibilities ("X or
Y", "X or similar", "X or nil"), etc.
And since uncertainty symbols have been
mentioned ...
... many of the reconstructions look like
this: "*n|n`u|üs'V or *n|n`u|üsyV 'woman; woman of the other moiety'", where the
uncertainty symbol <|> and the word "or" are used to compress eight
different possibilities (not to mention the run-of-the-mill unspecified final
vowel) into one "disjunctive" reconstruction; even so, a good deal of
special handling is required to account for unexpected daughter
reflexes.
There are some really mind-blowing
"correspondences", e.g. where a Nostratic "egg" is reconstructed as
<glottal stop> + <a|o "or similar" _or_ u> + <uvular or
pharyngeal fricative> + <i> on the basis of the IE "egg" word (cited as
*ou(y)o-), compared with Old Japanese u 'egg' (representing the entire Altaic
phylum!), and with a solitary Syro-Lebanese Arabic word for 'white of egg'
(?a:h_), out of which an Afroasiatic protoword for "egg" is conjured
up!
To be sure, Dolgopolsky himself marks the
most bizarre of such reconstructions as "questionable" (many others deserve the
same qualification but don't get it). However, if they are bizarre and he
realises it, why the heck does he publish them?
Piotr
----- Original Message -----
Sent: Friday, January 26, 2001 9:28 AM
Subject: [tied] IS's "regular roots" and the benefits of
back-up
Glen to Miguel:
<snipped exchange of irrelevant verbal abuse between my
honourable friends MCV and GG>
The Nostratic phonology as it now stands is a farce and Illich-S. has
hardly provided many, if any, credible disyllabic roots for Nostratic. It's not
necessarily impossible for polysyllabic roots to have existed in Nostratic but
the language is still unlikely to have been Japanese-like in nature. Don't you
agree?