We Can Hear Confucius Now

Say the name Confucius out loud. In Mandarin it is Kong Qiu, two flat, tidy syllables. Now here is the unsettling part: the man himself would probably not have recognized it.

Reconstruct his surname 孔 as it likely sounded around 500 BCE and you get something closer to khˤongʔ: an aspirated k, a tightening in the throat, a little catch at the end, and no tone in the sense Mandarin means by tone. His full name, 孔丘, comes out roughly khˤongʔ khwə. The language of the Confucian classics had consonant clusters and throat-tightened syllables. It sounded like a different world, because it was one.

For most speakers of Chinese, Classical Chinese, 文言文, is something met at a desk: a fifth of the exam, a wall of characters, a thing to be decoded and then closed. It is easy to forget that behind the characters was a mouth. People said these words. They argued and joked and prayed in them. And in the last few decades, scholars have quietly recovered a great deal of how those words actually sounded.

The script is a fossil record

Chinese characters look like the least phonetic writing on earth. They are, in fact, hiding a sound archive in plain sight.

Most characters are built from a meaning part and a sound part. The character 馬 (Mandarin mǎ, "horse") turns up as the phonetic hint inside a whole family of other characters. That only makes sense if those characters once rhymed or nearly rhymed with "horse." Follow enough of these families and you can triangulate the old pronunciations, the way you might reconstruct a lost melody from the harmonies still built around it. Reconstructed, 馬 comes out as mˤraʔ, with a little r buried in the middle that Mandarin long ago swallowed.

Three streams of evidence flow together here. The first is those phonetic families inside the characters. The second is rhyme: the Shijing, the Book of Odes, is full of poems that rhyme in Old Chinese but fall flat in Mandarin, and the pattern of what once rhymed with what is a map of the old vowels. The third is loanwords, the Chinese words borrowed into and out of neighboring languages, which froze old pronunciations like insects in amber.

The standard modern reconstruction is the one William Baxter and Laurent Sagart published in 2014, with a free database of nearly five thousand items anyone can download. It is careful to the point of humility. Where the evidence runs out, the notation says so: a bracket around a sound means "something like this, but we cannot be sure." That honesty is the opposite of slop, and it is worth admiring.

A language with no tones

Here is the fact that tends to stop people. Old Chinese, the language of Confucius and the Odes, very probably had no tones at all.

The tones that feel utterly essential to Mandarin or Cantonese today were not there at the start. They grew, over centuries, out of sounds that were later lost. When a syllable used to end in a little catch of the throat, or in an s, and that ending wore away, the pitch it left behind hardened into a tone. Tones are, in a sense, the ghosts of vanished consonants.

So the deeper you go, the less the language sounds like a tonal language and the more it sounds like something with clusters and codas and a texture all its own. The compound 司馬, sīmǎ, a title meaning roughly "master of the household," is literally "manage-horse," and in Old Chinese it comes out as something like s.ə-mraʔ. You can almost feel the cluster in the mouth.

None of this is a tape recording, and honest scholars say so plainly. We reconstruct systems with real confidence: which sounds contrasted with which, how syllables were built. The exact music of it, the fine grain, is still argued over, and the Baxter and Sagart system has serious critics who prefer rival reconstructions. But the big claim holds. This was a spoken language, spoken not only by sages but by everyone, and it sounded nothing like the classroom version.

Five words, and how they may have sounded

The gap between the modern reading and the reconstructed one is the whole story in miniature:

• 孔 (Mandarin Kǒng) reconstructed khˤongʔ, Confucius's family name.
• 馬 (Mandarin mǎ) reconstructed mˤraʔ, horse, with its lost inner r.
• 丘 (Mandarin Qiū) reconstructed khwə, Confucius's given name, "hill."
• 有 (Mandarin yǒu) reconstructed ɢʷəʔ, to have.
• 司馬 (Mandarin sīmǎ) reconstructed s.ə-mraʔ, a steward, literally "manage-horse."

To study Classical Chinese only as a code on a page is to keep it at arm's length. To learn it as a language, with its own recovered sound in your ear, is to stand a little closer to the people who first said these words, and to hear, faintly, the voice of an ancestor. That nearness is exactly what PRAVIEL is built to give you: not a summary of the tradition, but a way into it in its own words, and now in something close to its own sound.

Sources

• Baxter-Sagart Old Chinese reconstruction (data and papers)
• Old Chinese phonology (overview)
• Reconstructions of Old Chinese (competing systems)
• Baxter and Sagart, draft of the method (2015)