Origin of Latin's Passive Voice

[Terra]

Where did Latin's passive voice (http://en.wiktionary.org/wiki/fero#Latin) originate from? Looking at Proto-Celtic (http://en.wikipedia.org/wiki/Proto-Celt ... uage#Verbs), it looks like this form was common to Proto-Italic-Celtic.

Could this suffix originally have been "-s-" (from the same source as "sum/esse"), which was then rhotacized into "-r-" (with the final vowel later dropping)?

[Mecislau]

It's not just Proto-Italic or Proto-Celtic; the form comes from the PIE mediopassive, and cognate forms have been attested in the Italic, Celtic, Anatolian, Tocharian, Greek, Indo-Iranian, Germanic, and Albanian branches. The original endings were something like *-h2er, *-th2er, *-(t)or, *-medhh2, *-dhue, *-ro/-ntor (primary) and *-h2e, *-th2e, *-(t)o, *-meddh2, *-dhue, *-ro/-nto (secondary).

[Salmoneus]

It's not just Proto-Italic or Proto-Celtic; the form comes from the PIE mediopassive, and cognate forms have been attested in the Italic, Celtic, Anatolian, Tocharian, Greek, Indo-Iranian, Germanic, and Albanian branches. The original endings were something like *-h2er, *-th2er, *-(t)or, *-medhh2, *-dhue, *-ro/-ntor (primary) and *-h2e, *-th2e, *-(t)o, *-meddh2, *-dhue, *-ro/-nto (secondary).

Huh. So you don't believe in the -y mediopassive form in the central dialects (Greek, II, Albanian, Germanic)?

[Mecislau]

Huh. So you don't believe in the -y mediopassive form in the central dialects (Greek, II, Albanian, Germanic)?

Less that and more "I didn't really want to delve into that mess" since it's not directly relevant to the question at hand. I have no idea where current opinion tends to side on which endings were original and where the non-original set must have come from.

[hwhatting]

Huh. So you don't believe in the -y mediopassive form in the central dialects (Greek, II, Albanian, Germanic)?

You mean the -y in endings like Greek -mai / -toi? But that's clearly the same "primary" ending *-i as in the active forms and, as such, not part of the original active - mediopassive contrast. Or do you mean something else?

[Salmoneus]

Huh. So you don't believe in the -y mediopassive form in the central dialects (Greek, II, Albanian, Germanic)?

You mean the -y in endings like Greek -mai / -toi? But that's clearly the same "primary" ending *-i as in the active forms and, as such, not part of the original active - mediopassive contrast. Or do you mean something else?

It's relevent because they (apparently) used the same 'primary' ending in both active and mediopassive - rather than, as in the peripheral languages and as under discussion here, replacing it with -r. I wasn't saying -y mediopassives to distinguish them from the actives, but distinguish them from -r mediopassives.

[hwhatting]

Huh. So you don't believe in the -y mediopassive form in the central dialects (Greek, II, Albanian, Germanic)?

You mean the -y in endings like Greek -mai / -toi? But that's clearly the same "primary" ending *-i as in the active forms and, as such, not part of the original active - mediopassive contrast. Or do you mean something else?

It's relevent because they (apparently) used the same 'primary' ending in both active and mediopassive - rather than, as in the peripheral languages and as under discussion here, replacing it with -r. I wasn't saying -y mediopassives to distinguish them from the actives, but distinguish them from -r mediopassives.

I didn't say it's "irrelevant", I just said that it's not part of the original active - mediopassive contrast. It seems clear that the primary - secondary distinction was transferred to the mediopassive at different times in different languages, and in some perhaps*1) not at all. Anatolian is a good case of having both a primary *i and r-endings in the mediopassive, with the *i being added to the r-endings, so the transfer clearly happened later. The big difference between the Graeco-Aryan-Germanic system on one side and the Italo-Celtic, Tokharian, and Anatolian systems on the other side is not the absence or presence of primary *i in the mediopassive endings, but the absence or presence of r-endings.
*1) Due to the fact that those languages who do not show primary *i in the mediopassive all have gone through various apocopes and AFAIK treat *-ri# and *-r# the same, I wouldn't venture to say that they never had forms comparable to Anatolian -ari / -tari etc.