Khitan Small Script

The Khitan were a people of the eastern steppe, between the Mongolian plateau and the forests of Manchuria, who in 907 founded the Liao dynasty under Abaoji of the Yelü clan and ruled a state stretching from the Sea of Japan to the Altai for two centuries. Their language, Khitan, is related to the Mongolic languages. Within a few years of founding the state they gave it two scripts: the Khitan large script of 920, a logographic system inspired by Chinese writing, and the Khitan small script, which the History of Liao credits to Yelü Diela, the emperor's younger brother, around 924–925, after he met an ambassador who showed him the Uyghur script. The two scripts were used side by side and share no characters.

The small script lived on after the Liao fell. The Qara Khitai (Western Liao, 1124–1218) carried it into Central Asia, and the Jurchen Jin dynasty that replaced the Liao kept it in official use for decades, until Emperor Zhangzong of Jin ordered it suppressed in 1191. From then on it was forgotten so completely that no dictionary, primer or printed book survives. What we have is mostly stone: about 33 inscriptions dated from 1053 to 1171, nearly all funerary epitaphs buried with Liao royalty and nobles, plus a few bronze mirrors, tallies, seals and one jade cup.

The script came back to light in 1922, when the Belgian missionary Louis Kervyn entered the Liao imperial tombs at Qingling in Inner Mongolia and copied the epitaphs of Emperor Xingzong and Empress Renyi. Decipherment is a slow, collective work: the Mongolian scholar Chinggeltei and his Khitan Script Research Group at Inner Mongolia University published the foundational study in 1985, Daniel Kane's The Kitan Language and Script (2009) summed up what was known in English, and Wu Yingzhe, Liu Fengzhu, Viacheslav Zaytsev, András Róna-Tas and others keep adding readings from newly excavated epitaphs.

How the Khitan small script works: clusters of sound signs

Unlike the large script, the small script is mostly phonetic. It has a few hundred characters, and most of them are phonograms: each writes a syllable, a consonant-plus-vowel or vowel-plus-consonant pair, or a single sound. A word is not written in a line but packed into a square block, or cluster, of two to eight characters. Inside the cluster the characters are read in pairs, left to right and then top to bottom, and an odd final character sits centred below the pair above it. The result looks a little like a Korean Hangul syllable block built from several letters, though a single Khitan sign carries more sound than one Hangul letter. Clusters then stand in vertical columns that run from right to left, as in Chinese of the time.

Mixed in with the phonograms are logograms, characters that stand for a whole word: the numerals, calendar words such as 𘭧 month and 𘲺 day, kinship terms, 𘰲 heaven, 𘲇 great, 𘮝 gold. Many of these come in two shapes, a plain form and a dotted form that marks masculine gender agreement, which is why the grid above shows 𘬣 and 𘬤 for one, 𘲝 and 𘲞 for two, 𘬦 and 𘬧 for five. Khitan, like Mongolian, also distinguished vowel classes, and some phonograms exist in pairs for front and back vowels.

How much can be read? Of roughly 380 graphs in Kane's index, about a third have an agreed meaning and about a third a reconstructed sound value; the rest are known only as shapes. Scholars can read dates, titles, names and formulaic phrases on the epitaphs fairly securely, while much of the running text is still uncertain. So the readings on this page are reconstructions, agreed enough to appear in the standard literature, not dictionary entries.

Khitan small script vs Khitan large script

The two scripts are easy to tell apart. Large-script text is a line of separate, roughly square characters, many of them looking like simplified or altered Chinese characters, and there are about 830 of them known; it is a logographic system in which one character is usually one word or morpheme. Small-script text is made of tight clusters of smaller, simpler strokes, and the same word is spelled from a small stock of sound signs. The small script could record any word, including foreign names, with far fewer characters, which is probably why it became the script of the imperial epitaphs.

Only the small script is in Unicode. The large script has been proposed (its ISO 15924 code is Kitl) but is not yet encoded, so there is no large-script collection on this site. The two CJK ideographs 𰏃 and 𰕲, which Chinese sources use as Khitan words meaning army, are Chinese characters, not Khitan script, and belong with Chinese Characters rather than here.

The bilingual key to the small script is the Langjun inscription of 1134, carved on the Wordless Stele of Empress Wu Zetian at the Qianling mausoleum near Xi'an by a Jin official. It carries a Chinese text and a Khitan small-script text that say much the same thing. For centuries it was taken to be Jurchen writing; only in the 1920s and 1930s was it recognised as Khitan, and it still anchors many readings, including the pleasing fact that the Khitan word for gold, 𘮝, is the name the Jin ("Golden") dynasty called itself.

Khitan small script in Unicode: block, filler, marks

The Khitan Small Script block, U+18B00 to U+18CFF, entered Unicode 13.0 in March 2020 with 470 characters, after a long effort led by Andrew West, Wu Yingzhe, Sun Bojun, Viacheslav Zaytsev and Michael Everson that reconciled the character lists of Chinggeltei, Liu Fengzhu, Kane and others into one repertoire. Unicode gives the characters no names of their own, only numbers: 𘬣 is simply KHITAN SMALL SCRIPT CHARACTER-18B23. Unicode 16.0 (September 2024) added 𘳿 at U+18CFF, a marker for a character that is obscured or missing in a damaged inscription, drawn as an empty box. Unicode 18.0 (2026) adds five more at U+18CD6 to U+18CDA for the handful of Jurchen small-script inscriptions written in the same system; they are too new for fonts and are not yet on this page.

Two of the encoded characters are not Khitan words at all but tools. 𘬀 at U+18B00 is the iteration mark: placed after a cluster, it means "read that cluster again". And U+16FE4 KHITAN SMALL SCRIPT FILLER, an invisible character that lives in the neighbouring Ideographic Symbols and Punctuation block, tells a font how to stack a cluster: the usual cluster pattern starts with two characters side by side, and when a cluster instead starts with one centred character the filler goes right after it. Fonts that understand clusters then draw the block correctly; everything else simply ignores the filler.

Because the filler is invisible and the glyphs have no names, the block order matters for finding a character: the code chart follows the order of the encoding proposal's table, so our last groups keep Unicode order and label each row with its first code point. Copy a character and check it with a Unicode lookup if you need its exact number, for example when citing an epitaph in a paper.

Fonts, typing and pitfalls

No operating system ships a Khitan font, so without one every character on this page shows as an empty box. Install Google's free Noto Serif Khitan Small Script (also on Google Fonts) or Andrew West's BabelStone Khitan Small Linear and BabelStone Khitan Small fonts; the BabelStone fonts are the ones that lay clusters out as blocks, while Noto and most others draw the characters one after another in a line, which is the accepted "linear" layout for discussion and dictionaries. Vertical right-to-left columns need a browser or editor that supports vertical text on top of the font.

There is no Khitan keyboard. Copy from this page, use a character picker (the block starts at U+18B00), or type the hex code and press Alt+X in Microsoft Word. Write clusters as their characters in reading order, left to right then top to bottom, with a space between clusters; modern transcriptions use an ordinary space where the stones have a narrow gap. Put the filler straight after the first character only when the cluster starts with a single centred character.

Pitfalls: Khitan characters often resemble Chinese ones but never mean the same thing, so 𘮝 is gold, not 山 mountain. Do not substitute look-alike CJK ideographs, Chinese numerals or katakana for Khitan characters; they are different code points and a Khitan font will not show them. Treat every reading on the web, including ours, as a reconstruction, and check a name or a phrase with a specialist before it is carved, printed or tattooed.