Re: [tied] Animate Dual in -h3 (was: IE Roots)

From: elmeras2000
Message: 25340
Date: 2003-08-26

--- In cybalist@yahoogroups.com, Miguel Carrasquer <mcv@...> wrote:

>
> But the Skt. GL du. is -o:s (*-aus < *-ous).

Sure, but our reconstructions must pay particular attention to
languages with maximum diversity. And an old IE language
*distinguishing* the gen.du. from the loc.du. is very strong
evidence that the two were not identical in PIE. Merger, on the
other hand, needs little justification. Note that there is very
little evidence for the dual in general, so every vote counts.

[...]

> The genitive has no plural analogue: it is *-xW-s > *-ous (them. *-
oy-xW-s
> > *-oyous).

I would not exclude hat it has: It could be sitting right there in
the pronominal form *tóys-o~m which could be pl. stem *tóy + zero-
grade of genitive morpheme /-os/, supposing of course the span *-o~m
is an addition to be explained on another plane.


>
> This explains the attested forms:

[...]
> Avestan:
> G -å (*-o:s)
> L -o: (*-ou)

I do not see how the difference is accounted for in your
explanation. Could you focus on the salient point?
[...]
>
> Greek:
> GLDI -oiin < -oiun (< *-oy-h3u-m)
>
>
> The /o/ in G. *-(oy)ous (or *-(oy)o:s) is unexpected. The result
of
> syllabic */&3/ should have been /o/ in Greek, but /a/ or /i/
elsewhere.

If the vowel of "*-ous" or "*-o:s" is the thematic vowel, so that
the endings are in reality generalized o-stem or pronominal endings,
the o-timbre is inescapable before a following /H3/, no matter
whether the thematic vowel is here itself originally /e/ or /o/. I
do not see the basis for "*/&3" in endings syllabified *-o-H3-s, *-o-
H3-u, or, in extended form, *-oyH1-oH3s, *-oyH1-oH3u. The laryngeals
are clearly asyllabic here.

> I think the solution has already been provided by Jens. His *R
(the
> "causative morpheme") syllabifies as /o/ everywhere, not just in
Greek.
> The dual oblique suffix was therefore voiced *RW (*GW), as
expected in my
> "asyllabic affix voicing theory" for an asyllabic nominal suffix
(or
> prefix, such as "causative" *R) (cf. nominative *-s > *-z). [This
implies
> that *h3 was in my opinion itself not normally voiced].

Thanks, but no thanks: I believe you are compromising the good parts
of the "consonantal o" analysis by hooking it up with sheer
nonsense. To my knowledge, no phonetic rules have been found to back
any of this.

>
>
> As to the NA forms, we have animate *-o:(u), neuter *-oyh1 in the
o-stems,
> *-ih1 or *-(y)e in the consonant stems.
>
> I must say that Jens' explanation of the o-stem animate forms is
attractive
> (*-o-He > *-o:(u), as in the Skt. 1/3rd. person perfect of stems
ending in
> a laryngeal, i.e. *-VH-h2a, *-VH-e > *-o:u > -au).

Of course it is.

>
> My explanation, however, is that the dual morpheme was **-iku,
oblique
> **-iki (cf. plural **-atu > *-es, obl. **-ati > *-ey, as e.g. in
the
> personal pronouns *mésW ~ *(m)wéy, *úsW ~ *sWéy).

I confess I'm unable to follow.

> This developed by
> regular soundlaws that I have explained elsewhere into o-stem *-o-
h3, and
> inanimate (< oblique) *-íh1 (HD), *-ih1 (AD) and *-ye(:)h1 (PD) (*-
éh1
> (HD), *-(u)h1 (AD), *-e(:)h1 (PD) after *w). The PD form lost the
> laryngeal (after long vowel?, sandhi?), giving attested *-ye (e.g.
Greek
> osse < *okW-ye) and *-e. The pronominal forms are *wéh1 [< oblique
> *mu-íki, like *wéy < *mu-áti] and *(y)uh3 [< casus rectus *(t)u-
íku, like
> *(y)úsW < *(t)u-átu].

So the point of departure was not even structurally regular? I agree
in principle that earlier language stages do not have to be pretty,
they have to be right. But we need a *very good* independent
motivation for the irregular prestage (why "uns" : "ihr"?); if it is
wrong there is no point in basing phonetic rules upon it. I also
find it suspicious that I generally get more regularity by going
less far back than you.

>
[On IE 'eight', etc.:]
> But Proto-Kartvelian had no dual.

Is that a valid argument? Could it not have disappeared, so that
there would be no way of tracing it - except perhaps this way?

Jens