Whether *k^(e)rd- 'heart' and the
compositive element *k^red- 'faith, trust' in *k^red-dHeh1- should be identified
is a celebrated question. Since rhotics are prone to metathesis more commonly
than any other class of segments (cf. PIE *ters-/*tres- 'tremble' or
*perk^-/*prek^- 'ask') and the semantic match is acceptable, there are no grave
formal objections against such an identification. However, it's clear that even
if these forms are ultimately connected, they were consistently distinguished
already in PIE times (cf. Latin cord- : credo or Celtic *crid < *krd- vs.
*cred-). When Indo-Iranian *z'Hrd- replaced (for whatever reason) inherited
*s'rd-, the form *s'rad- was unaffected because it remained outside the paradigm
of the "heart" word. I only meant the latter in when I said that the
<h-> was invariable.
Here is a very similar case: Polish
<serce> 'heart' "should be" <*sier(d)ce> < *sIrd-I-ko-. Such a
form (<sierce>) in fact existed in Old Polish, but the initial palatalised
fricative /s'/ was replaced by /s/ in the 15th century, presumably due to Czech
influence (Prague was then an imperial capital and the cultural centre of the
region, and the mutual intelligibility of Czech and Polish was high). The same
change affected semantically transparent derivatives (e.g. <serdeczny>
'hearty, cordial'), but not historically obscured ones (e.g. s'rodek
'centre').
Piotr
----- Original Message -----
Sent: Wednesday, March 28, 2001 10:14 PM
Subject: Re: [tied] gr!
>while Sanskrit <hrd-> has an invariable
<h->.
But not in the compound forms which show the expected s'
! The form s'rad compounds with dha: (obviously) and also with kr and
da:na. A form with lengthened vowel also occurs in compound with dha,
namely the word sra:ddha, meaning faith.
Peter