Before I respond, I want to state that it was my impression that only
the thematic dative singular ending was *-o:i and that the thematic
nominative plural ending was either *-oi or *-o:s < *-o-es.
> Because the thematic vowel becomes /o/ before a vowel.
So you say.
> Presumably, the suffix *-ey was unstressed, and was reduced
> to *-y by zero grade, but not before the *e had caused
> lengthening (i.e. o-grade) of the thematic vowel.
> A better example (because it involves voiced and unvoiced
> sounds) is the gen.sg. *-osyo, from thematic vowel + *esyo.
> the order of the rules is:
This is a sidetrack, but something I find interesting: the dative
suffix is identical in form to the root for 'go.' I don't think that
this is a coincidence.
> 1) *-a- > *-a:- before voiced (*-a-asy- > *-a:-asy-)
> 2) zero grade (-a:-asy- > -a:-sy- > -osy(o)).
So, in effect, there would be a triple-length vowel there at some
point? I find this unrealistic. It is more realistic to presume
that *-a-asy- > *-a:sy-. (Of course, you are well aware of my
amateur status :) )
> But the lenthening of the thematic vowel comes _after_ the
> rule that deletes *i (and also after the final voicing rule,
> whatever its exact formulation):
Presuming that there was an *i at some stage.
> nom.sg. *-a-iz > *-a-z > *-a:-z > *-os
Is it likely that the *i would just disappear? The most likely
change resulting from *-a-i- would be the formation of a diphthong,
which may then smooth to *-e-, or lead to *-a:- which could then
become *-o(:)-.
> fem.sg. *-a-ih2 > *-a-h2 > *-eh2
Although using 'e' may be sound from a typological point of view, I
find it unlikely that *a would first become *e and then (rather
quickly) become *a again in the presence of *h2 (which I presume to
be /x/). Rather, I think it's more likely that *a retained its
central character when adjacent to the presumed velar fricative.
> >Why would *-o-es > *-o::s and not *-o:s?
>
> For the same reason that in the dat.sg. *o-ei yields *-o::i,
> and not *oi. I suspect the ins.sg. *-o-eh1 should have
> yielded *-o::h1, but that merged with *-oh1 and *-o:h1.
If *o was monomoraic, why would combining it with *ei yield a
trimoraic vowel plus *i? Where did the extra mora come from?
Another sidetrack of sorts: isn't it true of language that /h/
disappears very easily when in word-final position?
> In other words:
>
> sg.
> nom. *-á-iz > *-á-z > *-á:-z > *-os
> acc. *-á-im > *-á-m > *-á:-m > *-om
> n. *-á-id > *-á-d > *-á:-d > *-od
> voc. *-á > *-á > *-á > *-e
> gen. *-á-asya: > *-á-asya: > *-á:-&sya: > *-osyo
> dat. *-a-á(i) > *-a-ái > *-a:-ái > *-oéi > *-o::i
> loc. *-á-a(i) > *-á-ai > *-a:-&i > *-oi
> ins. *-a-át > *-a-át > *-a:-át > *-oéh1 > *-o:h1
> abl. *-á-a:t > *-á-a:t > *-á:-a:t > *-oot
Did the collapsing of vowels happen very quickly?
It appears that you have made changes in your theory from what you
have published online (although I have not reviewed it lately).
The "strong cases" (nominative and accusative) have suffixes
containing an *i element, while the rest have suffixes containing an
*a element. Are all of the case suffixes, then, presumed by you to
derive from pronominal elements?
As you know, I attribute the sigmatic nominative to an earlier
genitive which had gained ergative force. Of course, there is much
to be explained, such as why a marked accusative came to be used if
there was already a marked ergative. Also, why there came to be a
difference between nominative and genitive singular in root nouns. I
think this distinction came from using a pronominal element for the
formal genitive, while the nominative continued the original
genitive. This is akin to saying "the horse its rider" instead
of "the horse's rider." If we presume so, the question is, when did
this occur in the broader scheme of things?
> pl.
> nom. [*-á-asW > *-á:&sW > *-osW]
> *-a-ásW > *-a:ásW > *-oes > *-o::sW
> obl. *-á-aty > *-á:&y > *-oy
>
> The other forms are based on the oblique, approximately:
>
> acc.pl. *-á:-&y-&m-&sW > *-oyms > *-oems > *-o:ms
I thought the acc. pl. was *-oms > *-ons.
> gen.pl. *-á:-&y-&m > *-oym > *-oem > *-o::m
I thought it was *-o:m.
> dat.pl. *-a:-&y-á-:sW > *-oyósW
[snip]
The problem with the plural cases in PIE, as reconstructed, is that
they violate a language (near-)universal: that plural suffixes
(almost) always precede case inflections. Yet, we see things like
accusative plural *-oms, analyzed as accusative singular *-om plus
plural *-(e)s. However, there are other languages, such as Classical
Arabic, which also show the reverse of the supposed universal (case-
plural instead of plural-case). There are other instances where PIE
seems to break universals, as in having tense endings *after*
personal endings (e.g. 1sg 'primary' (present) ending *-m-i). I
wonder how these are to be reconciled, if at all.
- Rob