Thank you, Christophe. So there is at least one Buddhist parallel and one Vedic (ṚV 8.72.5c: véti stótava ambíyam, as pointed out by Jan). That is interesting.

Best wishes,
Martin


Den 2019-03-21 kl. 09:33, skrev Christophe Vielle:
There is maybe some parallel with what Renou Gram. p. 46 §41c (dealing with the sandhi of final -e/-o + vowels) notes about the "hiatus de a (issu du nomin. aḥ des thèmes en -a) devant initiale a- suivie de deux consonnes dans un texte bouddh[ique]", referring to Pischel SBBerl. 1904, p. 812, here:
https://www.biodiversitylibrary.org/item/92836#page/890/mode/1up
ekacīvaraka asthāt instead of ekacīvarako 'sthāt
pṛṣṭa avyākṛtam instead of pṛṣṭo 'vyākṛtam
I mean (in asya) the presence of the two consonants following the initial a, which lenghtens the syllabe, and makes the initial a pronounced long, resulting in the fact that the final e here conforms with the "general" rule (becoming a(y) when followed by any vowel, except a, or diphtongue).
Conversely, the two consonants following a ā can make it considered as short, as Renou notes ibid. with the (mainly epic) occurrences of -o/-e  'tman for -a ātman.