--- In
cybalist@yahoogroups.com, "Glen Gordon" <glengordon01@...>
wrote:
>
> Alright, I'll buy this once... What pray tell is this "shadow
> consonant" and where did it come from?
>
>
> - gLeN
It's the lingering effect, exceptions to sandhi, etc., of a
laryngeal in an earlier stage. For example, in Ossetic
stress is generally predictable, falling on the second
syllable of a phrase if the first is short, or on the
second syllable otherwise. In Old Ossetic, and still
in Digor, there was a definite article "i", which has
dissappeared in the Iron dialect, but which has a
lingering effect on stress placement, for example
"bælas" with stress on the first syllable means "the tree",
but on the second syllable "tree, a tree", because the
former is from "i bala:s" where the "i" shifts the stress
forward one syllable. Nowhere else does Iron need
an accent mark, so it would be simpler to have something
to mark the place of the lost "i" than to mark all Iron words
for stress. Of course it all depends on what type and
level of analysis one is doing. It can be insisted that no
segment itself not realized as an actual human speech
sound be represented, in which case Iron will have to
mark every phrase for stress, and a representation of
Sanskrit will have to distinguish the difference between i
and y, u and v, syllabic r and consonantal r in every place
they appear even though the majority of the time the
difference is predictable. However the original point
was to discern earlier layers, and so the argument stands.
When Ossetic still had the definite article "i", stress was
completely predictable, and when Indo-Aryan still had
a laryngeal, it could properly be analyzed as having only
/a/ as a vowel, at least if we're restricting our definition
of vowel to the resonants. This whole aspect of the
argument is really purely semantic. No one is arguing
about how many vowel sounds are in Sanskrit, only on
how to group them.
David