From: dgkilday57
Message: 68602
Date: 2012-02-23
>I like your mechanism better than opt. loss of *h2. However (and I may have missed something earlier), I am still puzzled by the lack of Olsen's h1/h2-preaspiration everywhere but in this particular suffix. We do not see it with *h1 in Lat. <-ple:tus>, Skt. <pra:ta'-> 'full', or with *h2 in Skt. <ga:tu's> 'gait, way', Grk. <be:tarmo's> 'dance', <amphisbe:te'o:> 'I stride', which have the same root as above. Proposing a new soundlaw to eliminate one or two otherwise independent suffixes seems rather heavy-handed.
> W dniu 2012-02-22 22:15, dgkilday57 pisze:
>
> > But the acute accent requires a short vowel, so it must be zero-grade.
>
> In the reconstructible PIE pattern one normally expects e-grade and root
> accent in the neuter and zero-grade and final accent in the collective,
> so the original pattern may have been /*gWéh2-tro-m/ --> *gWáh2tHrom,
> */gWh2-tré-h2/ --> *gW&2tráh2, levelled out in Greek (which generally
> does not preserve the accentual contrast), with the accent and the
> consonants of the original singular and the short vowel of the collective.