I was studying rhotacism and I came across the word corpora (plural of corpus). I would reconstruct the proto-italic form as *korpoza, but I saw the entry on Wiktionary and it says that the actual reconstruction is *korpezā.
Is this reconstruction accurate? If it is, what causes this vowel alternation, considering the singular form *korpOs and it's later result corpOra?
The e in the oblique stem seems to go back to Proto-Indo-European: compare the Germanic cognate, nom sg *hrefaz, nom pl *hrifizō.
Even if the PIt form had been **korpozā, -e- would be the expected development within Latin, due to medial vowel reduction. The -o- results from the o vowel in the nom sg spreading to other forms by analogy. Since this was analogy rather than a regular sound change, we see regularized words like Old Latin corpos, corporis next to unregularized ones like latos, lateris. Finally, regular sound change turned o in final syllables into u, giving the nom sg corpus.
This word corpus ~ corpor- belongs to the same type as genus ~ gener-.
PIE neuters with an alternation *g^enos ~ *g^enes-.
As regards your question about relative dating, I would say that schwatization of unstressed medial vowels in Latin (> e/i) is older than rhotacism. So I would say that the path *korpeza > *korpera > (late leveling) corpora seems more likely than the path *korpeza > *korpoza > corpora, as the second path conflicts with schwatization of unstressed medial vowels in Latin. Vowel e in *korpeza is the regular reflex of an unstressed medial vowel in Latin.
