At 2:41:11 PM on Monday, January 21, 2013, Richard
Wordingham wrote:
[...]
> Does -rkt- survive well in Celtic. If not, -rtk- > -rkt-
> > -rt- looks a pretty reasonable progression.
Matasović:
The development of the syllabic resonant *r > ar before a
stop is probably to be explained by assuming that *tk'
first changed to PCelt. *xt, and that *x behaved like *s
when the syllabic resonant developed vocalis prothesis.
Thus we had *h2rtk'os > *(H)rxtos > *arxtos > *artos.
Another, less probable explanation would involve positing
a special rule *h2rC > *arC before CRC > CriC
Brian