Re: The phonetic value of PIE *h3 and the 'drink' root.

From: elmerasdk
Message: 14216
Date: 2002-08-04

--- In cybalist@..., "sergejus_tarasovas" <S.Tarasovas@...> wrote:
> Do you accept Dybo, Nikolaev & Co's theory of Balto-Slavic accent
> (every syllable carries a tone, first high-toned syllable acquires
> accent etc.)? If you do, how would you derive these tones from PIE
> accentual system (with its expiratory stress, at least at the early
> stage)? I feel a strong sympathy to this theory, but I'm not sure
> whether it has received any support in the "West", so to say.

I accept the descriptive facts as to where the Baltic and Slavic
tones are located. I accept that IE mobile and oxytone paradigms get
mobile in BSl., and that barytones stay that way, except for the
working of Dybo's Law and Saussure's Law in Lith. and Slav., resp.
(and of a similar rule for OPruss. found by Kortlandt); I also accept
Hirt's Law which made more barytones than were inherited. That is
enough for me to bridge the gap between BSl. and the rest of IE. That
does not really answer your question, on which I am in two minds:
On one hand, I am highly impressed with the predictive power of
the "valency chemestry" of the Moscow school. It works remarkably
well in chains of derivation in Slavic and Baltic alike. I have seen
Paul Kiparsky do the same for Sanskrit, again with truly remarkable
results. It may have a strong nucleus of truth, and that, or some of
that, may have to be pushed back into Indo-European, for the branches
agree in a most striking manner on where to place the accent (despite
younger disruptions).
On the other hand, we know of minimal pairs like Greek phóros :
phorós, tómos : tomós, and Sanskrit ápas 'work' : adj. apás-
'active', where the two members of each pair apparently consist of
the very same material. There is also the indirect evidence of Baltic
metatony, as Lith. skìrtas 'separated' (mobile, IE end-stressed type,
her restored after the working of Hirt's Law) vs.
skir~tas 'separation' (barytone) or áukstas 'high' (mobile, IE end-
stressed) : au~kstas 'storey' (barytone), where the circumflex is (in
my opinion) best explained by the lateness of the differentiation:
after a certain point in time Lithuanian formed no more acutes
(you've begun again in words like bánkas because that now sounds
closer to the foreign source); so here, too, the variants consist of
the same material, only the accent has been differentiated to express
a difference of "part of speech" (substantive vs. adjective), and
when that happened late enough the newly accented syllables could
only get circumflex, whence the difference which is thus essentially
not one of intonation, but of accent placing. That indicates that the
Moscow idea is not the whole story, and, after much pondering, I must
say that I find it hard to acceot it as even *part of the story*.
Much in PIE accent works with quite strong predictability. We are
not surprised that *H1és-ti/*H1s-énti, opt. *H1s-iéH1-t/*H1s-iH1-ént
work this way, for all roots do that, except those that have
underlyingly long vowels; the same goes for nouns, as *H2nér-m/*H2nr-
ós, and *p&2-tér-m/*p&2-tr-ós. It is theoretically conceivable that
the suffix /-men-/ was recessive and did not take the accent,
while /-ter-/ (originally perhaps rather /-tel-/ of agent nouns did
take the accent, but this difference may just as well be ascribed to
the function, agent nouns being oxytone, while names of things are
barytone. Then *H1néH3-mn/*H1n.H3-mén-s 'name' could be original, and
so could Gk. ákmo:n 'stone' denoting a thing, while the end-stressed
type of Gk. poimé:n 'shepherd' may be due to the fact that it denotes
a person. One may note that in further derivatives /-men-/ and /-ter-
/ (/-tel-) operate the same, cf. Gk. stéphanos 'wreath' from *stébH2- (or *sté derived from the men-stem seen in stémma,
and the instrument noun type *H2ér&3-tro-m 'plough' derived from the
agent noun *H2r.H3-té:r (Gk. arrté:r). In these forms, both the
apparently unaccented /-men-/ and the apparently accented /-ter-/
give up their accents to the preceding root. Here the reason for the
accentuation must lie elsewhere, i.e. in the signal value it has to
show the part of speech.
Therefore, if there is to be assigned such a thing as "inherent
accent valency" to each morpheme, that can hardly be pushed back to
or beyond the IE protolanguage. It apparently did not govern the pre-
IE accent interplay that caused the IE ablaut. I therefore do not
particularly like the theory for IE. Consequently it strikes me as a
touch of mysticism if a certain predilection for having or not having
the accent is assumed to have arisen out of nothing in some early
stage of the individual branches.
In conclusion, I would much prefer to regard the "accent
properties" of Baltic and Slavic derived words as the only thing they
can be observed to be, i.e. the effects of analogical copying of the
base-word. In the terminology of Kurylowicz, one could say that
the "forme de fondation" has been utilized in the various "formes
fondées" with its full set of properties, including the accent.