Tangut Script

The Tangut were a people of the Chinese north-west whom Chinese sources call Dangxiang and who called themselves Mi, written 𗼇 (read mji). Their forebears herded in the highlands of Qinghai and western Sichuan, moved north under pressure from the Tibetan empire in the 8th century, and in 1038 their ruler Li Yuanhao proclaimed himself emperor of a state the Chinese knew as Xixia, the Western Xia. Its own name was 𗴂𗹭𘜶𗴲𗂧, the Great State of White and High. From its capital at Xingqing, today's Yinchuan in Ningxia, it held the Hexi Corridor, the Silk Road's gateway into China, for almost two centuries.

Two years before the proclamation, in 1036, the court promulgated a script of its own. The History of Song credits its design to the official Yeli Renrong, and it was put to use at once: state schools taught it, laws and documents were issued in it, diplomatic papers were written bilingually, and a vast programme translated Buddhist scripture from Chinese and Tibetan into Tangut and printed it from woodblocks and, remarkably, from movable type. Most of what survives is Buddhist, but there are also law codes, dictionaries, poems, calendars, military manuals and translations of Chinese classics.

The Mongols destroyed the Western Xia in 1227; Genghis Khan died during the final siege of the capital, and the city, its archives and the imperial tombs were sacked. Yet the script outlived the state by nearly three centuries. Tangut officials served the Mongol Yuan dynasty, a Tangut Buddhist canon was printed in Hangzhou in 1302, and the latest dated Tangut text is a pair of stone dharani pillars raised in 1502 by a Tangut community in Baoding, Hebei, naming more than eighty donors. After that the script was forgotten so completely that by the 19th century nobody on earth could read a line of it.

How Tangut characters work: components, compounds and no pictures

Tangut is a logographic script in the manner of Chinese: each character writes a word or syllable, and about 5,900 distinct characters are known. It looks like Chinese at a glance, with square characters built from the same kinds of strokes, but every character was invented from scratch and not one of them is a picture. Where the oldest Chinese characters began as drawings of a sun or a horse, Tangut 𗾔 sun and 𘆝 horse are abstract from the start, and Tangut characters average more strokes than Chinese ones, which gives a page its dense, bristling look.

A small number of characters are simple; most are compounds of two, three or four components. The 12th-century Tangut dictionary known as the Sea of Characters explains thousands of them as either meaning plus meaning or meaning plus sound. The classic example is 𗿝 wild goose, built from 𗿼 bird and 𗨜 longevity. What makes Tangut unusual is that a component does not keep one fixed value: the same element can lend its meaning in one character and its sound in the next, and the meaning it lends can shift from bone to smoke to food depending on the compound. Some pairs of characters even swap the same two components left for right and mean nearly the same thing.

The 768 Tangut components in Unicode (𘠀 to 𘫿, in the second half of the grid) are these building blocks as modern dictionaries index them, the Tangut equivalent of Chinese radicals. They are not words: a component has no reading or meaning of its own and never appears alone in a Tangut text, so copy one only when you are talking about how a character is put together. Tangut was written in vertical columns running right to left, like Chinese of its day; modern scholarship and Unicode write it left to right in horizontal lines, and regular, running, cursive and seal styles all survive on paper, stone and seals.

Deciphering Tangut: Khara-Khoto, the Pearl in the Palm and Nevsky

The first modern readings came from coins and a bilingual stele at Wuwei: in 1896 the British physician and collector Stephen Bushell worked out 37 characters, and the French scholars Gabriel Devéria and Georges Morisse added more from a Tangut Lotus Sutra. The breakthrough was archaeological. In 1908 and 1909 the Russian explorer Pyotr Kozlov reached the dead city of Khara-Khoto on the Edsin Gol in Inner Mongolia and, in a stupa outside its walls, found a whole library: more than 2,000 Tangut books, scrolls and manuscripts, now in St Petersburg.

Among them was the key. The Pearl in the Palm, 𗼇𘂜𗿳𗖵𘃎𘇂𗊏 in Tangut, is a Tangut–Chinese glossary compiled in 1190 by Gule Maocai: 414 words in nine categories, each written in both scripts with the pronunciation of each language spelled in the other's characters. Aleksei Ivanov recognised what it was, and from 1925 Nikolai Nevsky used it, together with Tibetan transcriptions of Tangut, to begin reconstructing the sounds of the language. Nevsky was arrested in 1937 as a supposed Japanese spy and shot; his notebooks were published in 1960 as Tangut Philology and won him the Lenin Prize in 1962, twenty-five years after his death.

Since then Nishida Tatsuo, Gong Hwang-cherng, Evgeny Kychanov, Shi Jinbo, Li Fanwen and others have turned Tangut from a riddle into a language that can be read. Li Fanwen's Tangut–Chinese dictionary (1997, revised 2008) numbers about 6,000 characters and is the reference behind this page: every meaning given here follows his English glosses, and the Unicode data files cite his numbers. Readings are another matter. Tangut phonology is reconstructed, which is why scholars write the words with an asterisk (𗼇 *mji, 𘓐 *dzjwo), and the systems of Li, Gong, Kychanov and Miyake differ in detail. The meanings are solid; the pronunciations are learned guesses.

Tangut numbers and the words worth knowing

The numerals are the easiest Tangut to read, because they behave like Chinese numerals: 𘈩 one, 𗍫 two, 𘕕 three, 𗥃 four, 𗏁 five, 𗤁 six, 𗒹 seven, 𘉋 eight, 𗢭 nine, 𗰗 ten, then 𘊝 hundred, 𗡞 thousand, 𗕑 ten thousand and 𗦲 a hundred million. Fourteen is 𗰗𗥃 (ten four), sixty is 𗤁𗰗 (six ten), 105 is 𘈩𘊝𗏁 and 2,362 is 𗍫𗡞𘕕𘊝𗤁𗰗𗍫. A second character for ten, 𗰭, is used above all for the days of the month in dates. 𗸕 means a half, and the suffix 𗡪 turns a number into an ordinal: 𘕕𗡪 is third, 𗨁𗡪 first. Unicode records these values in its Tangut data file, so they are the most firmly identified characters of all.

Beyond the numbers, the grid gives the words a curious reader asks for first. 𗼇𘓐 is a Tangut person, 𗼇𗟲 the Tangut language and 𗼇𘝞 the Tangut script, 𘝞 on its own being writing. 𗹦 is heaven or sky, 𗼻 land, 𗾔 the sun, 𗼘 the moon, 𗄓 the stars, 𗋽 water and 𗜐 fire. In the family 𗴺 is mother, 𗔀 father, 𗷸 son and 𘇾 daughter; 𘟙 is king and 𗼓 emperor. 𗢳 is the Buddha, 𗵮 and 𗼈 both mean god, 𘜸 temple, 𗤶 heart and 𗨜 longevity. Among the animals are 𘆝 horse, 𗵊 ox, 𗂪 sheep, 𗉠 camel and 𗗵 yak, the livestock of a steppe kingdom, beside 𗵃 dragon, 𗿻 phoenix and 𗁤 tiger.

Two names of books are worth copying whole. 𘝞𗗚, writing plus sea, is the Sea of Characters, the great native dictionary; and the title of the Pearl in the Palm reads, character by character, Tangut, Han, time, according to, palm, middle, pearl. Where Li Fanwen lists several characters for one meaning, as he does for man, dog or ghost, the grid shows one of them and the rest can be found through the search; Tangut, like any language, had synonyms, and some characters are rare words or transliterations of Chinese and Sanskrit that have no translatable meaning at all.

Tangut script in Unicode: blocks, components, fonts and pitfalls

Tangut entered Unicode in version 9.0 in June 2016, after a decade of proposals led by Andrew West and Michael Everson with Chinese and Russian Tangutologists: 6,125 characters in the Tangut block (U+17000 to U+187FF), 755 components in Tangut Components (U+18800 to U+18AFF) and the iteration mark 𖿠 at U+16FE0, which means read the previous character again. Unicode 11.0 (2018) added 5 characters and 12.0 (2019) another 6. Unicode 13.0 (2020) separated 9 characters that had wrongly been merged with look-alikes into a new Tangut Supplement block at U+18D00, and added 13 components, bringing the components to 768. Unicode 17.0 (September 2025) added 8 more ideographs and a Tangut Components Supplement block of 115 components; those newest 123 characters are too new for fonts and this site and are not on this page.

Unlike letters, Tangut characters have no Unicode names, only numbers: 𗼇 is simply TANGUT IDEOGRAPH-17F07, and 𘠀 is TANGUT COMPONENT-001. To type one in Microsoft Word, enter its five-digit code and press Alt+X (17F07 Alt+X gives 𗼇); elsewhere, copy it from this page or from a Unicode lookup. Tangutologists cite characters by Li Fanwen number as often as by code point, and the Unicode data file TangutSources.txt maps one to the other.

No mainstream operating system we know of ships a Tangut font, so without one every character on this page is an empty box or a question mark. Install Google's free Noto Serif Tangut, released in 2019 and redesigned in 2021 after Tangutologists criticised the first version, or Andrew West's Tangut Yinchuan, which descends from the typeface Jing Yongshi of Beifang Ethnic University drew to typeset Li Fanwen's 2008 dictionary and is free to download and use. Three pitfalls: components are not words and should never stand in running text; several characters differ by a single stroke, so copy from the grid rather than retyping; and the readings you meet online are reconstructions that vary by scholar, so quote them with their asterisk and source.