PIE alternating stems

I mentioned earlier an idea I had that PIE had alternating stems,
which were later regularised differently in different languages,
thus explaining the centum/satem split:

PIE had a labiovelar *kW (+ *gW, + *ghW) with allophones *kW/*k (+
*gW/*g, + *ghW/*gh), and
a plain velar *k (+ *g, + *gW) with allophones *k/*c^ (+ *g/*d3, +
*gh/*d3h).

The latter of the allophones was before front vowels. The
alternation became pervasive because of the e/o ablaut.

Those phonemes were regularised by picking one allophone over the
other.
For kentum languages the result was the two velar series *kW (+ *gW,
+ *ghW) and *k (+ *g, + *gh).
For satem languages the result was the two velar series *k (+ *g, +
*gh) and *c^ (+ *d3, + *d3h).

That's what I thought.

Here's a modification.

The satem languages did not regularise the kW, gW, ghW series.
Instead they palatalised their two allophones further:
From *kW/*k, *gW/*g, *ghW/*gh to
*k/*c^, *g/*d3, *gh/*d3h, not uniform *k, *g, *gh as I proposed
above.


Nice, huh?
This accounts for secondary palatalisation in IndoIranian, and the
first palatalisation in Slavic. They are thus reflections of the
state of affairs in PIE, not einzelsprachlich in those two branches.

So, that should take care of that. Be my guest, shoot away.


Torsten


BTW, since the centum/satem split can be traced to the ablaut e/o
alternation, I wonder if the third grade, zero-grade might explain
the decem/taihun split?