A nitpick, but this is known as ‘neutral’ alignment, rather than ergative alignment. Ergativity is specifically when the transitive subject (‘ergative’ argument) is treated differently to the transitive object/intransitive subject, which are treated the same way (‘absolutive’ argument). By contrast, in the Latin case, all three arguments are treated the same way — hence, ‘neutral’ alignment.

Incidentally, English has a very similar pattern: most animate pronouns have nominative–accusative alignment, while nouns and inanimate pronoun have neutral alignment. It’s pretty common in Indo–European languages.

I was under the impression that the unification of nominative and accusative case for neuters in Indo-European languages was indeed felt to derive from an ergative-absolutive distinction in proto-Indo-European. The idea would be that neuters can't be agents and are therefore always absolutive.

I’ve definitely seen some speculation about PIE having been ergative at one point (or at least split intransitive), but I don’t know enough about it myself to have a strong opinion either way. I believe it isn’t widely-accepted though.

Isn't this common to most Indo-European languages? I think it's valid for Germanic, Slavic and Greek.

Descendants of Latin like French have lost the neuter gender and the relevant inflectional endings. So now nominative and accusative for what we would translate as "it" differ. But it's true that Germanic pronouns still confirm to the system as well (he/him vs. it/it).

(Actually, come to think of it, in German only the masculine differentiates between nominative and accusative.)

> For example, bellum is nominative and accusative singular, bella is nominative and accusative plural.

That is true, but it's not really an example of Latin neuter nouns being subject to a fundamental rule specifying that the nominative case can never be distinct from the accusative case.

The rule exists, but the second declension forms are very standardized. You could equally use them to argue that Latin has a fundamental rule specifying that the ablative case and the dative case are always identical. (Though for that argument, the first declension would present a problem.)

Latin requires that neuter nominatives be identical with neuter accusatives even when the nominative form is unconstrained by its declension (only possible in the 3rd declension), and even when the nominative form is, by the standards of Latin, utterly bizarre. The neuter plural nominative of hic is haec (identical to the feminine singular, also an interesting [but not universal] pattern in Latin), and so is the neuter plural accusative. This is a word in the second declension, but it's a wildly illegal form for that declension.

And yes, as far as I'm aware, this treatment of neuters is an Indo-European universal rather than being anything specific to Latin.