To butt in a bit, I can be specific about Celtic (specifically Old Irish) preterite formations - although I'm not sure (in all the details) about Latin and other Italic stuff.

In Old Irish, there are the following preterite types:

-s preterite
-t  preterite
reduplicated preterite
á-preterite (that's long a-preterite)
í-preterite (that's long i-preterite)

S-preterites are probably Indo-European (or at least Pre-Celtic) s-aorists that have been formed from laryngeal final roots (or stems, in the case of derived verbs).

For example:   (Note in all Old Irish examples below, -i- before the final consonant indicates 3rd person singular)
ana-   'wait for'     >   ana-s  (anais 'He waited for')   root: *HenH- (laryngeals unspecified)

móra- (with long-o)  'magnify', 'make great'  >  móra-s-   (mórais 'He magnified')   stem: *meHr-eH-  (laryngeals unspecified)


T-preterites are also from Indo-European (or at least Pre-Celtic) s-aorists that have been formed from roots ending in certain consonants (mostly r, l, m, g).  What happened here though, is that   the root final combination  C-s-ti (in the 3rd singular) became  C-ti,  which was then reinterpreted as Ct-i, with the 't' being part of the tense stem not the personal ending.  This new tense stem was then spread to other persons and numbers. 

For example:
ber- 'carry, take, bear'   >   ber-t-  (birt 'He bore',   biurt 'I bore,  etc).      *ber-s-ti 


Reduplicated Preterites are mostly Indo-European reduplicated perfects. Importantly, most of these forms had o-grade of the root with the first consonant reduplicated followed by 'e'   so something like  C1eC2  would be  C1eC1oC2 in the perfect. This became  C1eC1aC2 in Old Irish. 

For example:  (In the following, I have omitted Indo-European reconstructions) 

simple verb:
lig- 'lick'   >   lelag-  (lelaig 'He licked')    

compound verb:
con-rig 'bind'  >   con-rerag-  (con-reraig 'He bound')     

Many roots which formed an infixed n-present in Old Irish belong to this class:

for-ding 'oppress'  >   for-dedag- (for-dedaig 'He oppressed') 

Some oddities occur, in which the reduplicated vowel is the same as the root vowel in the present:

bong- 'break'  >   bobag-  (bobaig 'He broke') 

The other two classes (the long vowel preterites) are a bit more unclear in my understanding. I'm not sure where they came from, but maybe someone else on the list knows?

Anyway, for completeness, here are two examples of these classes:

tech-  'flee'  > tách- (long-a)    (táich 'He fled')
er-n- 'grant' (with an n-present)  >   ír- (long-i)  (ír 'He granted') 


----- Original Message ----
From: Lars Finsen <[log in to unmask]>
To: [log in to unmask]
Sent: Monday, May 4, 2009 1:33:45 PM
Subject: Re: Greek miscellanea (was: Null aspect)

Den 23. apr. 2009 kl. 19.10 skreiv Edgard Bikelis:

> In greek and indo-iranian, past was made from the aspectual stems adding the augment, an é- that took the accent. Italo-celtic made with an -a:- instead.

Could you be a little more specific about those Italo-Celtic past formations? I have mentioned before that I thought Urianian was most closely related to Germanic, however, it seems I find traces both of augments and of reduplicated perfects, things that otherwise belong to the east. But recently I have noticed some phonological features that seem to be more or less in common with Italic, so maybe I should look into proto-Italo-Celtic as well (if such a thing ever exists) for inspiration regarding verbal morphology.