>a-mRtá- doesn't contradict the pattern of <a-> + <CVC-> = <aCC->. Pardon my ignorance, but, which root is the <-zatru> of <a-zatru> from?
> On Wed, Apr 7, 2010 at 4:39 PM, gprosti <gprosti@...> wrote:
> > --- In email@example.com, Piotr Gasiorowski <gpiotr@> wrote:
> >> One of the most beatiful examples, for its simplicity, is RV á:sat-
> >> 'unreal', the negated pres.part. od 'to be' < *n.-h1sn.t-
> >> Piotr
> > Couldn't this be due to analogy? A CVC root plus the negative prefix would have the form aCCat-, with two moras in the initial syllable. A VC root would normally only have one mora in this syllable, but maybe the initial vowel was lengthened (giving a:sat-, etc.) in order to maintain the two-mora pattern.
> If the alpha privativum had an intermediary state as [*an-], where did
> that nasal go, and the rhythm, in words like a-mRtá-, a-zatru?
> your case, why the sequence -ans- would not merge with all the otherIt's not clear to me what you mean. Why would *n.-sont- have merged with other cases of -aNs- if none of the above cases merged with them, either?
> cases of -aNs-, like
> aMsa- < *H?oms
> taMs < *tens
> haMsa < *g'h(H2)ans
> &c &c; instead we have:
> ásta < *.ns-to-
> asmé < *.nsmei
> gaccha- < *gW.m-sk'e-
> There are several other cases of lengthening due to laryngeal, likeThe second element of dvi:pa- begins with a vowel in its independent form (ap-). The second element of dvidhA- begins with a consonant in its independent form (dha:-). The two-mora pattern may not have applied to the latter type of case.
> dvi:pá- from *dwi-H2p-ó-, but dvidhA- from *dwi-dheH1-, so no rhythm
> here to be preserved.