These are perfect examples of usage of "nearest-neighbor" shifts in reconstruction.
I can now show by example what kinds of things I am proposing.
Piotr Gasiorowski wrote:
09:37, Richard Wordingham wrote:
> Vedic Sanskrit almost completed [l] > [r], and I believe some
> Iranian languages did complete the merger. Piotr could name
> examples. Otherwise, I'll have to hunt through the archives on
Well, Avestan had no /l/ at all, and later Iranian /l/'s are of
[and intervocalic *-t-] > l in Pashto,
How about th>l, th>dh>d instead
How about -rth->rsh> rj and rth>rl>l
Obviously the principle(s) has/have repercussions on reconstruction.