12-08-2020, 03:58 PM
(12-08-2020, 03:17 PM)nablator wrote: You are not allowed to view links. Register or Login to view.Yes, but there are a few instances of standalone EVA-S. Information is lost when converting to lowercase EVA, in these rare cases.
Yes, that's why we cannot treat S as s, but still can treat Sh as sh if needed.

(12-08-2020, 03:05 PM)-JKP- wrote: You are not allowed to view links. Register or Login to view.So... the closer the transliteration represents the text without having to do this kind of parsing, the kinder it is to those who are not coding everything from scratch.
Even if your transliteration is 100% precise you have to make some choices (at least about aiin, ch, sh, ckh...) before feeding this transliteration to app/algorithm (do nothing is also a choice). So producing statistical analysis of any form and keeping it theory-neutral at the same time is impossible.

