> > Doesn't it also give you *mogóN, *móz^esI free of charge, so to
> > speak (with some tweaking, ie. that *-oN <- *-oH)
>
> As I said: "it doesn't work on lengths produced by laryngeals".
>
> The law doesn't work on the 1sg. of thematic verbs, as shown by the
> a.p. c 1sg. (e.g. be``roN).
>
I'm not familiar enough with Slavic phonology, I'm afraid, and the
only Slavic language I know a smattering of is Russian, but it seems
to me that stressed -ú is the typical 1sg ending in that language.
To quote Wright's Gothic Grammar:
"
In the parent language the nom. sing. [of the weak inflection] ended
partly in -e:n, -o:n, and partly in -e(:), -o(:). The reason for this
difference is unknown. The various Indg. languages generalized one or
other of the two forms in prehistoric times, as in Gr. nom. poimé:n,
shepherd; he:gemó:n, leader; acc. poiména, he:gemóna, beside nom. Skr.
rá:ja:, king; Lat. homo, man; sermo, discourse; acc. rá:ja:nam,
hominem, sermo:-nem. In prim. Germanic the two forms existed side by
side, as in Goth. hana from -e:n, -o:n (§ 87, (1)), beside tuggo:,
haírto: from -o(:) (§ 89). In Goth. the -o(:) became restricted to the
feminine and neuter, whereas in the West. Germanic languages it became
restricted to the masculine, as OE. guma, OS. gumo, OHG. gomo, man,
from -o(:), beside OE. tunge, OS. tunga, OHG. zunga, tongue; OE.
e:age, OS. o:ga, OHG. ouga, eye, from -o:n.
"
Further, the ON weak inflection has nom. -i, obl. -a.
But also, the ON Conjug. I (dömi, dömir, dömir), III (vaki, vakir,
vakir) and possibly II (tem (?<- *temj ), temr, temr) have -i in 1sg,
where you'd expect an -a <- *-oN.
All this mess might somehow have to do with PIE stress patterns, which
is why I asked.
> The verb *mogti is a special case. It started out as barytone (one
> of the few, if not the only, verb that was always a.p. I: most of
> the others became a.p. I only after Hirt's law). By Dybo's law, it
> must have shifted to mogóN, *moz^és^I. No other verb in Slavic
> showed this accent pattern, so it was analogically levelled to the
> usual model for a.p. b verbs, yielding mogóN, mòz^es^I, but with
> lengthening in Czech (mohu, móz^es^), just like in other cases of
> late accent retraction in that language (nestí > né:st(i), voljâ >
> vó:le, peró > pé:ro)
So it jumped two morae (I find it difficult to believe single-mora
stress-movement that isn't caused by analogy)? Which came first,
lengthening or movement, and is there a causal connection?
Torsten