Sanskrit .n (was: PIE suffix *-ro - 'similar-with')

--- In cybalist@yahoogroups.com, Sean Whalen <stlatos@...> wrote:

> --- Piotr Gasiorowski <gpiotr@...> wrote:

> > *ksud-ró- 'small, broken into small pieces' (Skt.
> > ks.udrá- beside
> > ks.un.n.á- < *ksud-nó- 'pulverised') from *kseud-
> > 'crush' (cf. Slavic
> > xudU 'meagre' < *ksoud-o-).
>
> How do you account for the retroflex n. if you don't
> reconstruct a retroflex series in PIE (it's not just
> after u/i (vi-sad- and vi-san.n.a- "sad"))?

Isn't that _vis.an.n.a_? The only forms in -s- given by
Monier-Williams is Vedic imperfect _vyasi:dat_ besides _vyas.i:dat_.
If you posit too early a /n./, you then have to explain Pali _visanna_
with /n/, not /n./.

More prosaically, in Sanskrit it's the Nati rule:

Triggers = {r s.}
Targets = the first /n/ to the right of a trigger
Blockers = {retroflexes, dentals, palatals}
The target /n/ [or nn] must be followed by a nonliquid sonorant.

There's a review of it at
http://www.cog.jhu.edu/pdf/gafos_dissertation-chapter_05.pdf .

Richard.