One Language, Two Scripts — Why?
The Khitan people founded the Liao dynasty in 907 and ruled a vast steppe-and-settled empire stretching from Mongolia to the Yellow Sea for over two centuries. Literate and administratively sophisticated, they needed writing. What is surprising is not that they invented a script — it is that they invented two, five years apart, without abandoning the first.
The Khitan Large Script (契丹大字, Qìdān dà zì) appeared in 920 CE. Modeled visually on Chinese characters, it comprised several hundred logographic symbols adapted from Han forms. The emperor Abaoji (耶律阿保機) commissioned it, and Chinese accounts praise its rapid adoption for official use.
Just five years later, in 925 CE, Abaoji's brother Diela invented the Khitan Small Script (契丹小字, Qìdān xiǎo zì). The Small Script is structurally radical: rather than logograms, it uses small graphic elements assembled into square blocks. Each block represents a syllable or word, and the elements within it encode phonetic information. In this respect it resembles the way Korean Hangul assembles consonants and vowels into syllabic squares — though the resemblance is typological, not historical.
An Agglutinative Language in Square Blocks
Khitan is classified as Para-Mongolic — not a descendant of Proto-Mongolic, but an extinct sister branch of the Mongolic languages (Janhunen). The language is agglutinative — it strings suffixes onto roots to build complex meanings — which makes a logographic system like the Large Script a poor fit. The Small Script's phonetic blocks are better suited to this morphology, which may explain why it eventually superseded the Large Script in most formal contexts.
Yet scholars are not entirely sure why both scripts continued to coexist for two centuries. One theory holds that the Large Script was prestigious for monumental and memorial inscriptions while the Small Script was the everyday administrative workhorse. Another is that they mapped onto different dialects or registers of Khitan. The honest answer is: we do not fully know, because we cannot fully read either.
The Great Decipherment Problem
After the Jurchen Jin dynasty defeated Liao in 1125, Khitan culture faded rapidly. The last dated Khitan inscription is from 1176. By the Mongol period the language was dead and the scripts forgotten. Only about fifty monumental inscriptions survive — some fifteen in the Large Script and around thirty-five in the Small Script — almost all in northeastern China, with two in Mongolia and a Khitan-inscribed bronze mirror from Kaesŏng in Korea; systematic scholarly analysis only began in the 20th century, after the 1922 discovery of the bilingual epitaphs of Emperor Xingzong and Empress Renyi at the Qingling (慶陵) Liao imperial tombs in Bairin Right Banner, Inner Mongolia.
The decipherment strategy exploits a few bilingual monuments — inscriptions in both Khitan and Chinese that commemorate the same events. By aligning the Chinese text with blocks of Khitan, scholars can infer the general meaning of Khitan passages. Crucially, Khitan names of people and places known from Chinese records help anchor phonetic values for the Small Script.
Aisin Gioro and Recent Progress
A landmark contribution came from Aisin Gioro Ulhicun (愛新覚羅烏拉熙春), a Japanese scholar of Manchurian imperial descent who has published extensively on both scripts since the 1990s. Working with co-author Yoshimoto Michimasa, she has proposed readings for hundreds of Small Script blocks and identified grammatical suffixes, personal name patterns, and honorific vocabulary.
Even so, progress remains partial: perhaps 20% of the Small Script can be read with real confidence, and significantly less of the Large Script. The field remains open: newly excavated steles are still awaiting full publication, and new comparative tools — including computational alignment with related Mongolic languages — are being brought to bear. Khitan is one of the great unsolved puzzles of East Asian historical linguistics.
Unicode encoded the Khitan Small Script in 2020 (U+18B00–U+18CFF, 470 characters, Unicode 13.0), allowing digital scholarship to accelerate; the Large Script is still only at the proposal stage — a modest technical victory for a writing system that has been silent for nearly 850 years.