Ok, now for the nominal morphology. I'll leave accent-ablaut for later, since it's such a difficult question (and there are lots of difference answers to consider). For this post, we'll do endings.
The surviving cases at this point are nominative, accusative, genitive, dative, locative, and instrumental. I don't believe that PIE had an ablative case, and the allative was apparently already on its way out. The PIE terminations were as follows:
| athematic | thematic |
---|
sing. | plur. | sing. | plur. |
---|
nom. | -s | -es | -os | -oes > -ōs |
---|
acc. | -m̥ | -m̥s | -om | -oms |
---|
neut. | -Ø | N/A | -om | N/A |
---|
gen. | -s ~ -és | -Hom | -os | -oHom |
---|
dat. | -ei | -os | -oei > -ōi | -oos > -ōs |
---|
loc. | -i | -su | -oi | -osu |
---|
ins. | -h ~ -éh | -is | -oh | -ois |
---|
For the dat. and ins. plurals, see
this article. I also assume that the conventional gen.pl. ending -oHom was generalised from the o-stems, which isn't implausible, leaving an athematic -Hom.
The full-grade alternants of the ablauting endings are generalised through athematics, and with regular sound changes this yields:
| athematic | thematic |
---|
sing. | plur. | sing. | plur. |
---|
nom. | -s | -es | -os | -ōs |
---|
acc. | -əm | -əns | -om | -ons |
---|
neut. | -Ø | -es | -om | -ōs |
---|
gen. | -es | -ʔom | -os | -oʔom |
---|
dat. | -ei | -os | -ōi | -ōs |
---|
loc. | -i | -su | -oi | -ohu |
---|
ins. | -eʔ | -is | -oʔ | -ois |
---|
And here's what the system looks like by the merger of o into a:
| athematic | thematic |
---|
sing. | plur. | sing. | plur. |
---|
nom. | -s | -es | -as | -ās |
---|
acc. | -əm | -əns | -am | -ans |
---|
neut. | -Ø | -es | -am | -ās |
---|
gen. | -es | -am | -as | -aam |
---|
dat. | -ī | -as | -āi | -ās |
---|
loc. | -i | -su | -ai | -ahu |
---|
ins. | -ī | -is | -ū | -ais |
---|
It is at about this point that the opposition between dative and locative breaks down, putting their endings into competition. Generally, the locative endings are lost, except locative -i wins out over dative -ī, due to pressure from the instrumental. In the thematics, either locative -ai directly wins out over dative -āi, or the latter ending was remodelled to -ai, in order to preserve the general pattern that the thematic endings are the athemtic endings suffixed to the formant -a-.
I'll finish this post with a discussion of the neuter endings. It's fairly uncontroversial that the neuter didn't originally have a plural, and that one was created partly from a collective (providing the nom-acc plural), and partly by importing the masculine endings (providing the remaining cases). In Methinat, the nom-acc was instead created by an analogy with the vocative. In the athematic singular, both were endingless. On the basis of that identity, the vocative plural -es (which happens to be identical to the nom.pl.) was imported to provide a neuter nom-acc plural ending. For the thematics, the same case ending (in this case, -oes > -ōs) was imported. The other cases were simply imported from the masculine. At some point after this, the vocative then merged with the nominative.