But (for once) I am not so much interested in etymology as in linguistic typology. The “dogma” to which I am objecting is the insistence that there is something intrinsic in the neuter gender that means that neuter nouns cannot distinguish between a subject and an object, whether through case endings or postpositions.
Hi fdb, I had understood your typological objection, and I agreed to it. However, I wanted to point out that a counter-assertion that the Gujarati accusative form of (neuter) nouns is distinguished from nominative is also typologically problematic. There is no
form in the "average modern IA language" (including Gujarati) that can be meaningfully considered either homologous (i.e. cognate - as you explicitly pointed out) or analogous (i.e. having the same function) to the "quintessential" IE accusative case. It is important to distinguish between the morphological and syntactic cases (approx. the Sanskrit terms vibhakti and kāraka respectively) as two separate categories in the grammar of modern IA languages in order to obtain a clean description. The syntactic accusative function can be realized either by an unmarked (=rectus) form or a -ne form. But the -ne form can also realize syntactic dative and absolutive functions. I don't see why this morphological -ne form should be called "accusative" any more than the unmarked rectus form (which is shared with syntactic nominative and absolutive as well).
Some examples (taken from "The Indo-Aryan Languages" of the Routledge Language Family Series, Ch. 18 "Gujarati"), transliterated into IAST (but showing schwa-deletion):
A. Rectus as syntactic absolutive (controlling verb agreement):
rameś-e keṭl-ī copḍī kharīd-ī? 4.1.5
Ramesh.M-ERG how many-F book.F buy-PERF.F
How many books did Ramesh buy?
B. Oblique (OBL)+ne form as syntactic absolutive (controlling verb agreement):
urmilā-e tamār-ā dīkr-ā-ne jo-y-o. 4.1.7
Urmila.F-ERG your-M.OBL son-M.OBL-ne see-PERF-M.SG
Urmila saw your son.
Putting these sentences into imperfect (my own contribution):
AA. Rectus as syn. nominative (controlling verb agreement) and rectus as syn. accusative (no verb agreement):
rameś keṭl-ī copḍī kharīd-t-o?
Ramesh.M how many-F book.F buy-IMPERF-M.SG
How many books did Ramesh used to buy?
BB. Rectus as syn. nominative (controlling verb agreement) and oblique+ne as syn. accusative (no verb agreement):
urmilā tamār-ā dīkr-ā-ne jo-t-ī
Urmila.F your-M.OBL son-M.OBL-ne see-IMPERF-F
Urmila used to see your son(s) -> Here the number of son(s) is unspecified in the form, and also not derivable from the verb agreement as in example B.
======
The usage in B is probably a very Gujarati thing, not an "average modern IA usage". But this makes naming Gujarati -ne as accusative even more arbitrary in my opinion.