Rob:
> I'd like to give what I think is a possible explanation:
>
> *-mé:n < *-méns < *-máns
> *´-mo:n < *´-mons < *´-m@... (< *´-mans ?)
> *-té:r < *-térs < *-társ
> *´-to:r < *´-tors < *´-t@... (< *´-tars ?)
>
> What do you think?
It's essentially my theory. Although I add that the animate
suffixes *-m&n and *-t&r are derived from inanimate *-mn and
*-tr during the Thematicization period of Late IE. Both
*-n in the first and the *-r in the latter are one and the
same suffix that derived from a single inanimate agent marker
MIE *-an (> *-r; thus *wod-r "water", *ye:kW-r "liver"). After
the Heteroclitic Rule of MIE all instances of *-n become *-r
except after *m. Therefore *-mn, not **-mr (an idea I adopted
from Miguel; Thanks Miggy!).
However, a different MIE agent suffix *-éra was used to create
_animate_ nouns. By postSyncope eLIE, we get *-ér-s in the
nominative. It was combined with other suffixes like a
verb-substantivizing *-t- to produce *-t-ér-s. During the
the Thematicization period in mLIE, a new animate suffix *-t&r-
based on inanimate *-tr (from *-t- + MIE inanimate *-an) was
introduced to butt heads with *-tér- of identical function
(from MIE animate *-era). The word *px-tér- "father; provider"
is based on the older suffix *-tér- (*pex- "to feed, provide
for").
The oscillation of *-tér- and *-tor- was one of the things that
influenced the analogical creation of an accented *-mén- out of
*-mon-.
During Schwa Diffusion, *& lengthened before voiced consonants
producing nominatives *-m&.ns and *-t&.rs, becoming *-mans
and *-tars after Schwa Merger. Vowel Shift then produced
*-mons and *-tors to go with *-méns and *-térs.
Then Nominative Loss yielded *-to:r, *-mo:n, *-té:r and *-mé:n.
As always in all my rants, the exact same order of events occur:
Heteroclitic Rule
Syncope/a-Epenthesis/Szemerenyi Lengthening
Thematicization (*-mn, *-tr, *-x => animate *-m&n, *-t&r, *-&x)
Schwa Diffusion (*& before voice => *&.)
Schwa Merger I (*& > *e)
Schwa Merger II (*&. > *a)
Vowel Shift (most instances of *a > *o)
Nominative Loss
By the way, in my theory, *e simply derives from *e. Only
*o and those instances of 'true' *a that are not coloured *e,
come from earlier pre-VowelShift *a, so I don't like the idea
of *-máns as it is reconstructed.
= gLeN