Mongolian Da: The Voiced Dental Stop

Da writes the voiced dental stop and is the nineteenth letter of the alphabet, standing immediately after its voiceless partner and sharing its written treatment with it. The script does not consistently separate the two, so a reader identifies which is meant from the word rather than from the page. Unicode encodes da at U+1833, giving it a character of its own even where a manuscript gives it no shape of its own, which is the standard arrangement for this script: the encoding records the letters of the alphabet, and the rendering records the shapes, and the two do not correspond one to one. That mismatch is the single hardest thing about handling traditional Mongolian in software.

A correct digital text distinguishes letters that a correct handwritten text does not, so converting a scanned manuscript into encoded text requires reading it, not merely recognising shapes, and converting encoded text into a faithful image of a manuscript hand requires knowing which distinctions to discard. Neither direction is mechanical. The dental stops are a good place to see the problem in miniature, because they are common, they are genuinely distinct in the language, and the page frequently declines to say which one is present.