Latin perfects

[Peter]

Thanks, Wilmer,  for the interesting questions on the Latin perfect, some of
which have answers, some of which don't, and some of which are disputed.

> - a 'weak' perfect obtained in the most verbs using -vi/-ui

One theory, (which I think is true, but I don't know how much agreement
there is about it), is that:
(A) -vi/ui was added only to stems ending in a vowel or laryngeal.   The
vowel is sometimes obscured or lost by the time we get to meet the verb, but
often recoverable.   Here's some details:
(a) The Latin 1st, 2nd, and 4th declensions all have a vowel suffix on the
stem (eyo / -yo) hence the "choice" of the -vi / ui perfect, giving the
regular -avi, -ui, -ivi
(b) About half of the other forms which have -ui, the irregular verbs, end
in laryngeals, which of course have been lost before the time of the written
language.  Examples are:
crepo, crepui (laryngeal shown by Sanskrit -is- aorist)
domo, domui (a well known laryngeal root)
tono, tonui (laryngeal appears in the long -i:- in the imperfect in
Sanskrit:  asta:ni:t, and elsewhere)
Likewise vomui, volui, enecui, genui, colui, posui, serui
(c) I do not know whether the other half of the irregular -ui roots ended in
laryngeals or not.
(d) As far as I know, all of the irregular forms in -vi are laryngeal roots,
with the single exception of iu:vi, discussed below.  For example, fo:vi,
mo:vi, la:vi, ca:vi etc.

(B) -s- was added only to roots which ended in a consonant.   This appears
to be invariably true, although it may not be true that all roots ending in
a consoant were given the -s-.   Normally the vowel was also lengthened,
although there are some exceptions, such as coxi (short o), but this word at
least must be late in formation.

(C) In addition, there were survivals of other forms:
(a) reduplication (what you call "doubling").   In Latin this always has
zero grade, and never the -o- grade whcih is regular in Greek and Sanskrit.
(b) -o- grade.   In Latin this is never found with reduplication, and by
classical times is disguised by vowel changes.  You find it in iu:vit <
iouvit, vi:dit < voidit, vi:cit, li:quit, fu:git, fu:dit, etc.   (oi
normally > oe > u:, but after v, or between l and a labial or labiovelar it
> i:)
(c) lengthened grade.  There are traces of this throughout IE (e.g. a few
Germanic perfect plurals, a few Greek perfects, etc).  In Latin it is more
complex because long vowels were also produced through phonetic changes from
reduplicated perfects, such as e:git < h1e-h1g-it, or se:dit < se-sd-it, or
from -o- grades, such as iu:vit < iouvit.   The best example in Latin is
probably venio, ve:nit, which has parallels in the Germanic plural qe:mum.
scabo, sca:bi remains a mystery.

(D) The few short vowel perfects were originally reduplicated, but have lost
the reduplication, for example, tuli, fidi, -peri etc.

(E) Perfects without change usually have the prefect form obscured.   This
is true of the -uo, -ui forms you mention, but also of mandi, verri, and
many of the others.  It may be true of them all.   Bibo, bibi remains a
mystery.

It is important to recognise that the distinction of aorist and perfect
which is familiar from Greek is found in no other IE language in quite that
form.   Sanskrit distinguishes the forms of aorist and perfect, but the
meanings, when they are distinct, are reversed.    Most IE languages have no
such distinction.    So it is not such a surprise that Latin, too, has
merged the two forms.   In fact many verbs show a variety of forms for the
perfect, e.g. pango has pepigi, pe:gi and panxi.

As for the endings, Latin has a muddle of aorist and perfect forms.   The 3
sing is aorist, the 1 plural could be either, the 1 sing is perfect, and the
3 plural and 2 sing and plural seem to be a blend.   You m ention the 3
plural in particular.   There is dispute, and details are not clear, but it
looks like a cross between the -nt of the aorist, and an -r which is found
in some other IE languages in some forms, such as Tocharian, and the perfect
in Sanskrit.   Celtic also has a blend of -nt and -r , but the other way
round.   Generally Celtic has retained the two sets of endings, but with
asbolutely no difference in meaning.

The origin of the -vi/-ui form is disputed, and variously connected with the
perfect participle (as you do), or with a u stem noun form, or derived
phonetically from a laryngeal, or developed by analogy.   No one knows.

The 1 sing was originally in Latin not -i but -ai, though that doesn't
answer your question about the origin of the -i.  The best guess might be
that it is the same as the -i added to make the present (or primary)
endings -mi, -si, -ti, etc.

Hope that helps

Peter