Re: [tied] Greek thematic endings

From: Miguel Carrasquer
Message: 36780
Date: 2005-03-16

On Wed, 16 Mar 2005 08:41:17 +0000, P&G
<G&P@...> wrote:

>What do you make of [Sihler's] suggestion that B-S and Greek both show
>signs of an original primary *-ti in athematic stems, and *-i in thematic?

I wouldn't have gone through the trouble of working out a
different solution if I had thought the case convincing...

Beekes reconstructs for the thematic endings: *-oH, *-eh1i,
*-e; Jasanoff *-oh2, *-esi, *-eti; Sihler *-oh2, *-esi,
*-eti ~ *-ei. All these proposals have in common the
thought that all or some of the thematic endings are somehow
related to or influenced by the "stative" endings of the
perfect/middle/hi-conjugation.

The evidence for such a thing in the thematic 2 and 3 sg. is
however weak. In the 2sg., we see regular *-e-si almost
everywhere. Old Irish *-i: can come from *-esi with early
loss of intervocalic *-s- in Celtic (cf. o-stem Gsg.
*-os(y)o > CeltIb. -o(:)), and Greek -eis is also explained
straightforwardly as *-esi > *-ei with analogically added
-s. The Balto-Slavic ending *-eí cannot be explained by
*-esi, but *-eh1i doesn't explain it either (that would have
given circumflex *-ei~, cf. a:-stem DL *-ah(e)i > -ai~), and
the ending seems to originate in the *athematic* paradigm
(specifically that of "to be"?) anyway, as can be seen in
Slav. athematic -si (< *-eí) vs. thematic -es^I ~ -es^i (<
*-es^i). Nowhere (Tocharian aside) does the 2sg. thematic
show any traces of *-th2, which is what we would expect if
the endings originate in the "stative". We still need an
explanation for Balto-Slavic *-(s)eí (*-(s)eiH or *-(s)oiH)
and for Slavic thematic -s^- instead of -s-, but the
"stative" hypothesis doesn't really help there.

In the 3sg., Sihler's *-e-i, which works for Greek, cannot
explain Baltic -a (*-o for *-e) nor Slavic -e. Beekes' *-e
might explain Balto-Slavic, but doesn't explain the Slavic
(OCS, mod. Russ.) variant in *-etU, *-oNtU.

The Old Irish 3rd. person forms are only compatible with the
scenario that I outlined in my previous message. We have
absolute: berid, ber(a)it vs. conjunct -be(i)r, -berat. The
conjunct forms cannot be from *-eti, *-onti, wich would have
given *berid, *ber(a)it, and must be from *-et, *-ont, like
Slavic -e(tU), -oN(tU). The absolute forms have a suffix
*-es(t) [no doubt the copula], so we have:

*-o:-es(t) > *-o:s > -u
*-esi-es(t) > *-i:s' > -i
*-et-es(t) > *-et' > -id'
*-omos-es(t) > *-omi:s' > *-mi
*-ete-es(t) > *-et'e:s' > *-th'e
*-ont-es(t) > *-ont's > -at'

3sg. *-e-es > *-e:s would have given *-e (like 2pl. -th'e).
*-eti-es is possible, but *-eti doesn't work in the
conjunct, so it must also be discarded.

We have evidence then for 3sg. present *-et (and sometimes
3pl. *-ont) in Baltic, Slavic, Greek, Celtic. Also in
Tocharian, where the thematic 3sg. is *'-0 (palatalization
followed by nothing), which points to *-et (*-eti would have
merged with 2pl. *-ete > '-c).

Variation between *-et and *-eti (*-ont and *-onti) is
precisely what would be expected if the bulk of the thematic
presents derive from subjunctives.

Another reason why I cannot accept Sihler's *-e-i is that it
violates the rule that the thematic vowel is *-e- before
voiceless sounds and silence, but *-o- before voiced sounds.
This means that *-e-i cannot be ancient, or we would have
*-o-i (what we indeed have in the middle). For the same
reason, I cannot accept *-oh2 or *-oh1 as the 1sg. thematic
ending: that should be *-eh2 > *-a: or *-eh1 > *-e:. *-oh3
is possible (if *h3 was voiced we would get *-o-h3 > -o:, if
it was unvoiced we would get *-e-h3 > -o:), but doesn't fit
in with the stative ending *-h2a-.

In Anatolian, the ending *-o: is unknown. Thematic verbs
have 1sg. -ami in Hittite, -awi in Luwian. Since the 1sg.
ending *-m, *-mi must be derived from the pronoun **mu
(later replaced by *h1eg^), just like 2sg. *-s, *-si is
derived from *tu "you", the *-m was originally labialized
*-mW (1pl. *-mWén > Hitt. -wen or *-mWésW > *-mésW > Arm.
-mk`). The Luwian form -awi is then simply from *-o-mW-i
with *-mW- > -w- regularly, while Hittite -ami has restored
/m/ for /w/ analogically after athematic -mi (with
labialization regularly lost in a consonant cluster, cf.
Arm. nom.pl. *-esW > -k`, but acc.pl. *-(o)ns(W) > -s).

If the PIE 1sg. thematic ending was *-owi, like the u-stem
loc. sg. *-ow-i ~ *-ew-i, we would expect similar
developments in both forms. In the u-stem loc.sg., the *-i
gets umlauted to -u (*-ow-u), and word-final *-wu (like *-yi
in the i-stems) is eliminated with compensatory lengthening
of the preceding vowel (u-stem *-o:u/*-e:u, i-stem *-e:i).
The same must have happened in the 1sg. thematic *-ow-i >
*-o:u, and this is precisely the attested form in Tocharian
B 1sg. thematic -eu (< *-o:u), later -au.

Outside Anatolian and Tocharian, the ending *-o-mWi > *-o:u
was irregularly shortened to *-o:.

That leaves the question of where the Tocharian B
athematic(!) and subjunctive 1sg. ending -u comes from. The
soundlaws would suggest *-ou (< *-omW?).


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