Tibetan Dza ཛ

Dza is the nineteenth of the thirty consonants and the voiced member of the affricate line, written as the shape of ja carrying the tsa phru, the same small mark that gives tsa and tsha theirs. In the classical language it was a voiced dz; modern Lhasa speech devoices it and carries the difference on low tone instead.

It is the least frequent of the thirty letters in native Tibetan vocabulary. A reader can go a long way through ordinary Tibetan prose without meeting it at the head of a word, and where it does appear it is very often in a transcription of a Sanskrit name or term, standing for the Indic letter of the same value.

That gives dza an unusual position. It is a full member of the recitation, learnt in its place and written with the ordinary letters, and it is at the same time doing much of the work of the transcription apparatus, which otherwise sits outside the alphabet. The reversed letters and the subjoined ha were added for Sanskrit and stayed outside; dza was inside from the start and found itself pressed into the same service.

Like the other two affricates it is a root letter only. It can carry a superscript and the wa subscript, and it never serves as a prefix, a suffix, a superscript or a subscript itself.

Its rarity has a practical side. A learner drilling the recitation will have met tsa and tsha in ordinary words long before meeting dza in one, so of the thirty letters it is often the last to become familiar in reading rather than in reciting.

Tibetan books and Tibetan buildings frequently carry writing that is not in Tibetan letters at all. On the first folio of a canonical text, above a doorway, around a mandala or along the border of a painting, the script is often Lantsa, an ornamental Indic hand of the Ranjana family that Tibet adopted for exactly these purposes.

The division of labour is clear and old. Tibetan letters carry Tibetan, including translated Buddhist texts, which is nearly all of the written literature. Lantsa carries Sanskrit, and particularly the Sanskrit of mantras and of the titles under which Indian works were known. A canonical text traditionally opens with its title in Sanskrit before giving the Tibetan, and the Sanskrit is often set in Lantsa, so the reader is told at once that the work is a translation and where it came from.

The choice is aesthetic as well as functional. Lantsa is an elaborate, calligraphic script with long ornamental heads, and it reads as formal in a way that plain Tibetan letters do not, which is why it is used decoratively on surfaces nobody is expected to parse letter by letter.

Tibetan has, then, two ways of writing a Sanskrit mantra: transliterated into Tibetan letters with the reversed forms and the subjoined ha, or set in Lantsa as Sanskrit. Both are common, and the choice is one of register.