Re: [tied] Bader's article on *-os(y)o

From: Miguel Carrasquer
Message: 32502
Date: 2004-05-08

On Sat, 08 May 2004 14:44:47 +0200, Mate Kapovic'
<mkapovic@...> wrote:

>From: "Miguel Carrasquer" <mcv@...>
>
><Since Gaul. <sosio> is NA n. sg. "this"(*), and not a
><genitive, it suggests that whatever the Gaulish G.sg. form
><was, it probably wasn't *sosyo.
>
>(*) BUSCILLA SOSIO LEGASIT IN ALIXIE MAGALU
>"Buscilla placed _this_ in Alisia for Magalos"
>
>Then how do you explain this sosio which means "this" but doesn't have
>anything to do with *sosyo?

The form appears to be a compound of so- and siod (for
*tyod). The corresponding acc.sg. masc. is sosin (*so-sim
or *so-siom?). Old Irish has neuter forms based on *sod
(-so) and *siod (-se), besides the normal article *sindo-
(but n.sg. *som > sa(N)). I don't quite know what to make
of these forms, but the common factors are: (1)
generalization of *s- from the old nom. masc. and fem. at
the expense of *t- from the neuter and the oblique (cf. also
Celtiberian n. soz < *sod for *tod); (2) important influence
from the pronoun *syo, *tyom (f. syah2, n. tyod) [e.g. Vedic
syá, tyám, tyásya, etc.].

>It would be really strange if it didn't. No
>matter that the meaning is accusative, I would still bet it is an old
>genitive... In some languages (like Slavic or Finnic), genitive instead of
>accusative is not so strange in some contexts. I wonder if something like
>that could be shown for Celtic (Gaulish)?

Not to my knowledge.

><We don't have that problem if the o-stem G.sg. was in fact
><-o:, with a long vowel.
>
>The problem I have with your -o < -osyo is thatnot only does *o: yield -u-
>in Celtiberian in the -oC position (D. sg -ui, Ab. -uz, g. pl. -um, a.
>pl. -us ?) but also that I. sg *-oh1 gives -u in Celtiberian. You would have
>to claim a different long *o: or different relative chronology to account
>for that.

Of course: the chronology is vastly different. The ins.sg.
in *-oh1 > -o: had a long vowel centuries, if not millennia,
before *-os(y)o was reduced to *-oo in Celtiberian. The
Proto-Celtic was still *-osyo (as shown by old Lepontic
-oiso), while the others were Proto-Celtic *-o:i, *-o:d,
*-o:m, *-o:(n)s (I would not reconstruct a long vowel
already for PIE, at least not in the ablative, since PIE
*-õ:d would have given Lith. -uõ, instead of attested -õ <
*-a: < *-aa < *-ood).

Proto-Celtic *o: gave Celtiberian /u/ (at least in a final
syllable). Only then was *-os(y)o reduced to *-oo > -o:.
There is no more of a problem with that as there is with
Greek D. -O:i, Gpl. -O:n (written with omega) vs. Gsg. -o:
(written -ou). The only difference is that loss of
intervocalic -s- is regular in Greek, while it isn't in
Celtiberian. On the other hand, we have the same thing in
Latin, where *esyo > eius, *kWosyo > cuius, *is-tosyo >
isti:us, despite the fact that -s- normally gives -r- in
Latin. This can be explained in many ways: special
development of /sy/ as opposed to plain /s/, special
development of pronouns as opposed to other words (cf.
Germanic þ > ð in pronouns, þ- elsewhere, or Slavic Gen.
*-oo(d) > -ogo in pronouns, but *-oo(d) > -a: > -a in
nouns), etc.


=======================
Miguel Carrasquer Vidal
mcv@...