Three waves across a millennium
Open any Japanese dictionary and look up a common character like 行 (go, travel). You will find not one but three Sino-Japanese (音読み on'yomi) readings: gyō, kō, and an. They are not alternative spellings of the same pronunciation. They are three separate historical imports, each reflecting a different stage of Chinese phonology, borrowed centuries apart.
Layer 1 — 呉音 Go-on (5th–6th century)
The oldest layer arrived before Japan had any formal diplomatic contact with the Tang court. It came via the Korean peninsula — by tradition from Baekje — carrying the character readings associated with the Southern Dynasties capital at Jiankang (modern Nanjing), in the old land of Wu 呉. Buddhist monks carrying sutras were the main vector. The name Go-on ("Wu sound") is not evidence for that route, though: it only appears from the mid-Heian period, coined by promoters of Kan-on who borrowed the Tang capital's dismissive term for southern provincial speech. Until then these readings were simply called wa-on 和音, and the alternative traditional names 百済音 (Kudara-on) and 対馬音 (Tsushima-on) point at Korea instead. No contemporary document ties them directly to southern Chinese.
Go-on preserves archaic features of Middle Chinese that had already disappeared from the north by the Tang era: voiced initial consonants and final stops are often retained in disguised form. Examples: 経 kyō, 男 nan, 行 gyō. The voiced g- in gyō reflects the Middle Chinese 匣 initial of 行 — the voiced fricative of *ɦɨaŋ, which Old Japanese, having no voiced fricative of its own, took in as g-. By the Chang'an standard behind Kan-on that initial had already lost its voicing, giving kō.
Layer 2 — 漢音 Kan-on (7th–9th century)
From the 7th century onward, Japanese government missions (kentōshi 遣唐使) traveled directly to Tang Chang'an and brought back the prestige pronunciation of the capital. This layer — Kan-on, "Han sound" — was actively promoted as the scholarly standard and largely displaced Go-on in formal writing.
Kan-on reflects the northwestern Chinese dialect of Chang'an more faithfully: voiceless initials, different vowel qualities. 経 becomes kei, 男 becomes dan, 行 becomes kō. The devoiced k- and the different vowel both point to the Tang capital's speech.
Layer 3 — 唐音 Tō-on (12th century onward)
Confusingly, Tō-on ("Tang sound") does not reflect the Tang dynasty at all. It is a later stratum arriving from Song, Yuan, and Ming dynasty China, mainly through Zen Buddhist temples, the tea ceremony, and cuisine. The name is an anachronism — medieval Japanese called all Chinese things "Tang".
Tō-on imports are fewer but recognisable in everyday objects: 行灯 andon (paper lantern), 蒲団 futon, 椅子 isu (chair), 饅頭 manjū. The an- in andon reflects a later Chinese pronunciation in which the Middle Chinese 匣 initial of 行 — a voiced fricative *ɦ-, never a nasal — had dropped away altogether (*ɦaŋ → aŋ), leaving a syllable whose final -ŋ Japanese took in as -n — something neither Go-on nor Kan-on captured.
Why does this matter?
The three layers preserve evidence from different periods of Chinese phonological history. By comparing Go-on, Kan-on, and Middle Chinese reconstructions, linguists can triangulate what sounds existed in Chinese dialects across five centuries — evidence that Chinese textual sources alone cannot provide. Japanese kanji readings therefore complement Chinese rhyme books, modern varieties and other Sino-Xenic traditions in historical reconstruction.
- Go-on: Buddhist vocabulary, older temple readings — 経 kyō, 行 gyō, 男 nan
- Kan-on: Scholarly/official vocabulary — 経 kei, 行 kō, 男 dan
- Tō-on: Zen/cuisine/everyday objects — 行灯 andon, 蒲団 futon
The next time a Japanese learner asks "why does 行 have so many readings?", the answer is not sloppiness or accident. It is a compressed archive of a thousand years of cultural exchange, written invisibly inside a single ink-stroke.