How linguists rebuilt the sound of Tang-era Chinese from a rhyme dictionary
No one recorded how Chinese sounded 1,400 years ago, yet linguists have reconstructed it. Their key is the Qieyun, a rhyme dictionary of 601 that sorted characters by tone and rhyme, plus clues from Japanese, Korean and Vietnamese borrowings and from Buddhist monks transcribing Sanskrit.
Middle Chinese is the name for the variety preserved in the Qieyun and its later expanded editions. The Swedish linguist Bernhard Karlgren believed it captured the speech of Chang'an, capital of the Sui and Tang. Most scholars now read the dictionary's preface differently: it was a compromise blending northern and southern reading and poetic traditions. Scholars of the day cared intensely about reciting the classics correctly, and the Qieyun merged the distinctions of six earlier dictionaries so successfully that all six have vanished.
Its great innovation was fanqie spelling. Each group of identically pronounced characters was glossed with two other characters, one sharing the initial consonant and the other the rest of the syllable. Homophones were filed under 193 rhyme classes within four tones. In 1842 the Cantonese scholar Chen Li worked out the underlying categories by chaining these spellings together. Around 1150 the Yunjing rime tables laid the system out in 43 grids, though their authors, speaking a later form of the language, were already reconstructing a sound system unlike their own.
The dictionary nearly slipped away. When Karlgren began his pioneering work in the early 20th century, scholars had only fragments and relied on the Guangyun of 1008, a much expanded Song edition. Then large parts of a Qieyun version turned up in the Dunhuang caves, and in 1947 a complete copy of Wang Renxu's 706 edition emerged from the Palace Library.
The dictionary describes sound categories but never the actual sounds, so reconstruction leans on outside evidence. Vietnamese, Korean and Japanese borrowed huge amounts of vocabulary, though squeezing Chinese into foreign sound systems blurred distinctions. Transcriptions of Sanskrit are especially telling because Sanskrit pronunciation is known in detail: nasal consonants that rendered Sanskrit nasals in the early Tang were later used for its b, d and g, hinting at a shift in some northwestern dialects. Almost every modern Chinese variety except Min can be traced back to this system, which also illuminates Tang poetry.
Source: Middle Chinese
