>I don't think there's only internal reconstruction to rely on; many words in dif. branches of IE have the same meaning, ending in -no- in one and -ro- in the other. There are many other forms of ev., even doublets like:
> W dniu 2010-07-13 00:22, stlatos pisze:
> > So is it more economic to derive words with -no- / -ro- from *-n.o- in
> > PIE, since there are so many that alternate in IE? Or does this sort of
> > logic only apply to -no- / -mo-?
> I personally favour a phonological explanation of the origin of *-n/r-
> heteroclisy in PIE (maybe even involving *-nt-), though it's difficult
> to _prove_ anything using only internal reconstruction. It's my opinion,
> for what it's worth, nothing more.
> > > > The -t- in 'eastern' may not come from -s.r.- at all since it is aThat is no ev. against my statement that PIE had both; not based on ev. peculiar to this word, but from simple examination of all words in *-wo:s (that show clear s/t alt.). Moreover, if you believe in an opt. sr > str in one l., why a non-opt. sr > str in the other? Particularly when the ev. you use aren't sure cognates; an old words only preserved in compounds isn't unusual.
> > > > derivative of 'morning', a word ending in -wó:s with f/t* (dental) alt.
> > > > (which words I believe you said probably came from T > s/t in PIE, with
> > > > no f, etc., mentioned). The -t- is also found in some l. in which there
> > > > is reason to think no sr>str took place, like Sl * utro.
> > >
> > > There are very good reasons to think Slavic <(j)utro> comes from
> > > *h2ausro- as well. First, Baltic has it (Lith. aus^ra, Latv. austra);
> > > secondly, Slavic has many traces of the by-form *(j)ustro.
> > How does any of that show sr>str over sr / str or anything similar?
> Slavic by itself can't distinguish between inherited *-str- and *-str-
> from *-sr-, but Lithuanian evidence favours *-sr-, and Greek and
> Indo-Iranian support it further.
> > WhatThat is not a reasonable possibility; the odds of two words rec. * austro- for dir./wind with dif. origins is too small to consider. The ev. from all branches supports a single origin. A change in dir. needs no explnation, especially when speakers of one l. have moved to a place w dif. wind patterns. Consider Salish: tënc^al'ëqW = w wind Saa; tënc^ë´y'ëqW = sw wind Kl. There is as much (or more, considering l' / y' instead of no alt. for * austro-) reason to think these are two words with dif. origins as auster/eastern, etc. Though they are closely related l., Saa and Kl. have only two out of many cognates sharing the same appearance and meaning for '_ wind':
> > about L auster? Wasn't an old rec. * aus-tero- made to account for all
> > these? Wouldn't your possible * aussro- leave open opt. s>t/s_r or sim.,
> > at least?
> Perhaps double -ss- was preserved in pre-Latin and did not undergo
> rhotacism, then *-s(s)r- yielded -str- as in a few other branches. But
> there's also another possibility: that Lat. auster, which means neither
> 'east(ern)' or 'dawn', is not connected with <auro:ra>.