--- In
cybalist@yahoogroups.com, "tgpedersen <tgpedersen@...>"
<tgpedersen@...> wrote:
> In Witzel's Substrate Languages in Old Indo-Aryan I find that he
> mentions k/s´ alternations in Para-Munda(?) loans in Vedic:
>
> S´arkot.a / Karkot.a "snake demon"
> karkat.a "crab"
> Kambota / S´amban "name of demon"
> kabara / s^abara
> ki:sta / s^i:s.t.a "shell"
> kirata / Cilata "a mountain tribe"
> Kimidin / S^imida "demon" / "a demoness"
> Kiknasa "ground grain" / cikkasa "barley meal"
>
> where he assumes an original Para-Munda phoneme that became /k/
> and /c/ in Vedic. But the easier assumption would be that Vedic had
k
> > c in that period and re-borrowed the roots, or?
The change k > s' is the Satem change, so the simple borrowing
hypothesis would imply Satem or earlier out of India.
I must admit I can't see how given the Sanskrit opposition k ~ c ~
s', a sound would become /k/ or /s'/ on borrowing, but not /c/.
However, let us consider the evolution of oppositions from PIE to
Sanskrit. State 1 is PIE, Stage 6 is Sanskrit.
1. kWa ~ kWe ~ kWi ~ k^a ~ k^e ~ k^i
2. ka ~ ke ~ ki ~ ca ~ ce ~ ci (Satem)
3. ka ~ ke ~ ki ~ s'a ~ s'e ~ s'i
4. ka ~ ce ~ ci ~ s'a ~ s'e ~ s'i (Law of palatals)
5. ka ~ ca ~ ci ~ s'a ~ s'i
6. ka ~ ca ~ ki ~ ci ~ s'a ~ s'i (/ki/ by analogy?)
Witzel proposed that the Para-Munda phoneme was [k^].
At Stage 2, [k^a/e/i] would be borrowed as [ca/e/i], yielding
[s'a/i], or [ka/e/i], yielding [ka/ca/ci].
At Stage 3, [k^a/e/i] would be borred as [ka/e/i], yielding
[ka/ca/ci].
At Stage 4, [k^a] would be borrowed as [ka] or [s^a], [k^e/i] as
[ce/i], the latter yielding [ca/i].
At Stage 5, [k^a] would be borrowed as [ka] or [ca], [k^e] as [ka] or
[ca], and [k^e] as [ci].
At Stage 6, [k^a/e/i] would yield [ka/i] or [ca/i].
Thus:
[k^a] yields ka, s'a or ca.
[k^e] yields ka, s'a or ca.
[k^i] yields ki, s'i or ci.
However, if we assume that by Stage 5, Para-Munda [k^] has hardened
to [k], we then get:
[k^a] yields ka or s'a.
[k^e] yields ka, s'a or ca.
[k^i] yields ki, s'i or ci.
However, this assumes that Para-Munda [e] (or something similar) was
borrowed as [e] while pre-Sanskirt had such a vowel. If there were
no such borrowing, then we would have the alternations ka ~ s'a and
ki ~ s'i ~ ci that we see.
Can anyone enviage a better mechanism? I must admit I don't like
this explanation of the phenomenon; it suggests a very long period of
borrowing.
Richard.