-r passive archaic or innovating?

[Carol F. Justus]

>In any case, what you may be remembering is that Celtic, Anatolian, and
>Tocharian all share all share a passive formation in /r/, as does Italic
>(and Phrygian?).  All four are also "centum"-languages.  The r-passive seems
>to be a shared retention of a fairly archaic formant.  Of course, this
>should not be interpreted to indicate any closer relationship between these
>languages in the PIE-period, beyond the fact that they were all
>centum-dialects.

>Sincerely,
>Jim Bilbro

It is, however, peculiar that in Hittite the -ri of the present tense -r
passive in innovating. Older Hittite medio-passives have the -ri optionally
tacked on. One scenario suggests the Old Hittite lost a final *-R passive,
then it began to be restored with the present -i as -ri. There is a nice
study by Yosida in the early 90s (?) that plots the progress of this new
-ri through historically dated Hittite texts, from OH without it to NH with
increasingly more uses of -ri in e.g., forms such as arta(ri) 'stands',
kisa(ri) 'becomes', esa(ri) 'sits', kitta(ri) 'lies' (cf. Greek keitai,
Sanskrit saye, sete).

Interesting that this *-R passive / middle, medio-passive got lost in OH
and everywhere except Italic, Celtic, Tocharian, and later Hittite. Holger
Pedersen and Annelies Kammenhuber thought that the opposite was true, that
this form, as many lexical items too argued for a closer relationship among
Hittite, Italic, Celtic, and Tocharian. A closer relationship among these
branches, of course, is problematic for an variant of an Indo-Hittite or
first branching off of Anatolian.

Carol Justus