It was a combination of both, because the concrete and abstract nouns
basically declined for different cases originally. This Ancient Arithide
case system, as I wrote up in in-universe style last year and haven't had
time to revise, is on GoogleDocs here: <> There are lots
of places that could still be improved, and all comments are welcome.


Things/Animals: Accusative, Genitive, Instrumental, Dative/Prepositional,
People: Nominative, Accusative, Genitive, Dative/Benefactive, Vocative
Places: Nominative, Locative, Allative, Ablative, Genitive

Hence the distinction between the Places class, with a greatly different
declension pattern, from the two concrete classes.

I can't remember why the Things class had a vocative caseā€”it might have been
because people were named after things and animals sometimes. Essentially I
imagine that the separate cases that these classes exhibited collated
themselves into the coherent 11-case system of Classical Arithide, by
analogy, syncretism and many other processes that characterise a society
increasingly confident with playing around with its language (i.e. poetic
synecdoche, perhaps, like bahuvrihi compounds or somesuch). The benefactive,
prepositional and the allative merged into the one single dative of the
classical language, I imagine.


2010/6/13 Alex Fink <[log in to unmask]>

> On Sat, 12 Jun 2010 22:28:31 +0100, Eugene Oh <[log in to unmask]> wrote:
> >Ancient Arithide had three basic lexical classes: persons, things and
> >places, of which the latter had a more deviant declension because of its
> >"abstract" nature".
> Because of its abstract nature?  What's the mechanism for this?  I could
> see
> a noun class of places having variant formations for some local cases,
> because of old adpositions or whatever that were only used locally; I could
> see a class of abstract nouns having a basically defective declension,
> since
> they can't be used in certain semantic roles which demand concrete nouns;
> but I don't quite see what you're getting at here.
> Alex