Mongolian Ja ᠵ

Ja writes the voiced palatal affricate, the sound English spells j in judge, and it is the twenty-first letter of the alphabet. It stands with the voiceless affricate before it as one of the pairs the alphabet arranges by voicing, alongside the two dental stops, the two bilabial stops and the two velars. Unicode encodes ja at U+1835. Unlike the dental stops, the affricates are clearly distinguished in writing, so this is a pair the reader does not have to resolve. That unevenness is characteristic of the script: some distinctions are marked reliably, some are marked by dots that may be omitted, and some are not marked at all, and there is no single principle behind which is which.

The pattern is historical rather than systematic. Where the inherited alphabet happened to supply two shapes, two shapes were used; where it supplied one, one was kept and the reader took up the slack; where the tradition felt a distinction was worth marking it added dots. An alphabet is an accumulation of decisions taken at different times for different reasons, and Mongolian displays the fact more plainly than most because it has never been comprehensively reformed. The letter is common in native vocabulary and frequent in Tibetan loanwords, and it is one of the shapes a learner can rely on early, since neither of the affricates has to be recovered from context the way the dental stops and the rounded vowels do.

Encoding the traditional Mongolian alphabet was the easy half. The Unicode block at U+1800 to U+18AF lists the letters, and a text stored in it is unambiguous in a way a manuscript is not. Displaying that text correctly is the hard half, and it took much longer. Each letter has different shapes at the beginning of a word, inside it and at the end, and choosing among them requires examining the neighbours; some letters have more than one acceptable shape in the same position, so the text needs explicit selectors to say which is intended; the whole word has to be assembled onto a continuous vertical spine; and the column has to run down the page in a layout system built for horizontal lines.

For years the result was that valid Mongolian text displayed as a broken column of disconnected shapes on most systems, or did not display at all, and a great deal of Mongolian material circulated as images to avoid the problem. Font and layout support in current systems is much better, and the standard's own handling of the script has been revised more than once in response to what implementers and users found. It is a useful case study in the difference between recording a writing system and reproducing it.