On the idea of 2ps *tu/*tun and plural *ti/*tit:
> In what language?
>
> Eskimo-Aleut, as I said, has lost the original free
> pronouns, but if we assume a common Uralo-Eskimo stage, they
> must have been similar to the Uralic free pronouns, which
> are PUralic *mi, *ti (oblique *mi-nä, *ti-nä; plural *m'äd
> (or *m'än), *t'äd (or *t'än), oblique *m'äj, *t'äj).
Of course the pronouns cannot be recovered in EA although the
endings still preserve some former affairs as I've stated
them. At any rate, the same pronouns you cite in Uralic show
exactly what I'm saying.
Ur *mi and *ti is to be connected to *mu(n) and *tu(n). We
see the *n-forms in Finnish oblique cases like the genitive
/min-un/ and this oblique nature of the *n-forms is even
confirmed in IndoTyrrhenian by both IE and Etruscan which
has /mi/ for the nominative and /mini/ for the accusative.
The plural forms as I said are either ending in simple *-i,
becoming *-äj in the reconstructed forms you quote, or they
end in the plural found in nominal stems, *-it, becoming *-äd.
The same oscillation of *-i and *-it is seen in IE where 1pp
*wei shows earlier *mi (with a switch of *w- for *m-), and
*ns (in verbs as *-mes) shows a contracted form of *mit. The
question isn't "In what language do we find this?" The question
is "In what language do we _not_ find this?" We even find
the *-n in Altaic (Turkish /ben/ 'I') and the correlative *-r^
in the 1pp for former *-it (Turkish /bir/ 'we').
So ample evidence shows that the 1ps/2ps was *mu(n)/*tu(n)
while the 1pp/2pp was *mi(t)/*ti(t).
= gLeN