On 2006-04-24 16:59, Rob wrote:
> Fair enough. Where does *(xW)re:gs 'king' fit into this picture?
It doesn't :). Seriously, *h3reg^- seems capable of yielding Narten
forms, cf. RV 3sg. ré:s.t.i 'rules', 3sg. inj. ré:t. < *h3re:g^-t . For
an originally static root noun I would expect the following forms:
nom. *h3ro:g^-s
acc. *h3re:g^-m.
gen. *h3reg^-s
voc. *h3re:g^
Typically, the accusative would have developed an analogical o-grade
form modelled on the vocalism of the nominative (like *pod-m. for
**pe:d-m., the proportional equation being *h2ne:r : *h2ner-m. = po:d-s
: X). The reason why the 'king' word does not behave typically is, I
think, the vocative, with its atypically high frequency of use (in any
imaginable IE social setting "O king!" would have been several orders of
magnitude more frequent than "O foot!"). The vocative-supported *e: not
only remained in the accusative but also managed to spread to the
remaining cases already in PIE.
Piotr