Mongolian Ga ᠭ

Ga writes the voiced velar stop and is the thirteenth letter of the alphabet. Its written shape is closely bound to that of the back class velar before it, and the two are separated by dots. In practice this means the difference between them can be absent from the page, since dots in this script are dropped freely in ordinary handwriting and are not always supplied in print.

Unicode encodes ga at U+182D. Like the other velars, it participates in the vowel harmony, taking one pronunciation in back class words and another in front class words, so a single written letter covers a range of realisations that a phonetic transcription would have to separate. Alphabet charts usually present the velars as a small group and teach them together for exactly this reason, since learning them singly gives a learner three unrelated facts where the group gives one pattern. The letter is common, occurring in native vocabulary throughout, and its various medial forms are among the shapes a reader has to become fluent with early. Its relationship to the back class velar is the clearest example in the alphabet of a distinction that the script marks in principle and often declines to mark in practice.

Every writing system leaves something out and expects the reader to supply it. What differs is how much, and Mongolian leaves out a good deal: the difference between two rounded vowels, the difference between two dental stops, and, whenever the dots are omitted, the difference between two velars and between a nasal and a vowel. A page of traditional Mongolian is best understood as a set of strong constraints on what the word can be rather than as a specification of what it is. This works because the constraints are strong. The harmony halves the possibilities at a stroke. The position in the word rules out further candidates. The vocabulary of a language is finite and its frequent words are very frequent. A fluent reader converges on the word almost immediately and does not experience the ambiguity at all.

The situation is not unlike unpointed Hebrew or Arabic, where a script that looks impossibly underdetermined to a learner is read at speed by anyone who knows the language. It is worth saying because the alternative reading, that the script is simply imprecise and its users put up with it, misdescribes both the script and its users. The imprecision is where the language is predictable, and nowhere else.