The Sinosphere's Rosetta Stone
Every country that adopted Chinese characters also adopted Chinese pronunciation, but they borrowed from different eras and different dialects. The result is a linguistic Rosetta Stone hidden inside four modern writing systems. The borrowed reading-layers of Japanese, Korean and Vietnamese are known collectively as Sino-Xenic (a term coined by Samuel E. Martin in 1953); set beside Mandarin, Middle Chinese's own direct descendant, they give four cognate reflexes of a single Chinese source that diverged after borrowing.
A Character-by-Character Comparison
Take 山 (mountain). In Middle Chinese it was reconstructed as *ʂan. Today:
- Mandarin: shān (tone preserved, final nasal preserved)
- Korean (漢字音): san (conservative nasal, no tone)
- Japanese (漢音 kan-on): san (same layer as Korean)
- Vietnamese: sơn (vowel shifted, final nasal preserved)
Now 國 (nation), Middle Chinese *kwok:
- Mandarin: guó — final -k lost, tone shifted
- Korean: guk — final -k preserved
- Japanese (漢音): koku — -k reflected as -ku (vowel insertion)
- Vietnamese: quốc — final -k fully preserved
The Final-Stop Hierarchy
This pattern repeats across every character with a Middle Chinese checked-tone (入聲) ending. The hierarchy of conservation is remarkably consistent:
- Cantonese/Hokkien: preserve -p, -t, -k intact as unreleased stops (most completely in Cantonese; Hokkien also has a colloquial glottal-stop layer)
- Vietnamese: preserves all three (-p, -t, -c/-ch)
- Korean: keeps -p and -k, but Middle Chinese -t appears as -l (一 il, 發 bal)
- Japanese: converts them with vowel epenthesis (-pu/-tsu/-ku)
- Mandarin: erases all three, leaving only tone traces
For 學 (study, *ɣwak in Middle Chinese): Korean hak, Japanese gaku, Vietnamese học, Mandarin xué. The -k is alive in three out of four.
Why Japanese Has Multiple Readings
Japan borrowed Chinese pronunciation in at least three distinct waves: 呉音 Go-on (Wu-based, 5th–6th c.), 漢音 Kan-on (Tang-court standard, 8th–9th c.), and 唐音 Tō-on (Song–Ming era). For 行:
- Go-on: gyō
- Kan-on: kō
- Tō-on: an (from southern dialects)
Korean by contrast received a comparatively unified transmission: the main layer of Sino-Korean (漢字音) is generally traced to the Chang'an standard of Tang China, carried home by the Silla students and envoys at the Tang capital in the 8th and 9th centuries. Resting on a single dominant source is what makes Sino-Korean so much more internally consistent than the Japanese layers — drawn, as it happens, from much the same Tang standard as kan-on.
The Tonal Preservation Ladder
Beyond consonants, tonal preservation follows a clear gradient. Middle Chinese had 4 tone categories that split into 8 in some languages:
- Cantonese: 6–9 tones (best preservation)
- Vietnamese: 6 written + 2 subphonemic (excellent)
- Korean Sino-Korean: historically 3 pitch accents, now mostly levelled in Seoul dialect (good historical record)
- Japanese on-readings: no lexical tone, but pitch accent preserves some traces
- Mandarin: 4 tones, with massive mergers — the weakest preservation
The ladder is what makes the comparison useful: Cantonese and Vietnamese together preserve most of what Middle Chinese distinguished, and reading a character across all four systems recovers more of the original than any one of them holds alone.
The Sinosphere's borrowed layers aren't just curiosities — they are the phonological time machine that lets scholars reconstruct a language no living person has spoken for over a thousand years.