On Fri, 31 Jan 2003 00:08:53 -0000, "deshpandem <mmdesh@...
>Thanks. If the lack of vowel alternation in Sanskrit compounds
>matching the alternation of accent is an indication of a shift from
>stress to pitch accent (This was my hunch), that would put this
>shift in pre-Vedic phase of Indo-Iranian. How far back can one
>push this shift to pitch accent? Another way to look for would be
>to see how far back can one trace the lack of vowel alternation in
>compounds? Before Indo-Iranian split from its parent? During
>the spit of Iranian from Indo-Aryan?
For Greek, Balto-Slavic and (indirectly) Germanic, a pitch
accentuation which is basically in agreement with the Vedic one can
also be reconstructed, so my hunch would be that the "transition to
pitch accent" (alternatively, the loss of emphasis on stress
accentuation [pun intended]) had already taken place in
Proto-Indo-European itself, at least in that part of it that was
ancestral to Indo-Iranian and the language groups mentioned above
(which, in my opinion, can leave out only Tocharian and Anatolian).
Using compounds in general as a dating mechanism seems difficult, as
they are continuously created anew during the history of a language.
Such compounds as discussed recently on this list (Skt. -ks.u, -bda-
from the roots *pek^u (pas'u) and *pod- (pad-)) do show reduction
(even more so than the nominal paradigms they are related to), and are
perhaps very ancient, while a dvandva compound like ma:tára:-pitára:
was no doubt a recent creation in Sanskrit, with both components still
independent enough to retain their own accent. Still, it is in
principle possible to separate the different historical layers, and
the evidence from the largest "middle group" of Sanskrit compounds,
with no or little reduction of the unaccented part, is that stress
accent had ceased to be an important factor at the time they were
Miguel Carrasquer Vidal