Learn the kanji through components and connections.

Middle Chinese for learners of Japanese


What do I care about Middle Chinese?

After learning a number of kanji and their readings, you might find yourself noticing certain patterns. Perhaps you've been wondering about things like:

  • Why do some kanji have slightly varying on'yomi?
    人 can be read as nin or jin, 女 can be read as nyo or jo, 入 can be read as nyū or , or even ji, etc.
  • Why do so many kanji (and sound components) stand for the same sound?
    Why should 工, 交, and 高 all stand for the same sound ? And why aren't they interchangeable as sound components?
  • Why are the on'yomi all sort of the same?
    They're always one or two syllables long, and the second syllable is always ku/ki or tsu/chi, etc.

Of course, none these questions are central to learning Japanese. But knowing the reasons behind these patterns can help you make sense out of some of the knottier parts of the Japanese writing system.

Moreover, the answers to these questions may prove interesting to anyone interested in Japanese culture. The story of the on'yomi reaches back to the very foundations of Japanese civilization, when writing was first introduced to Japan. And the first written language of Japan was not Japanese, but Chinese! Specifically, it was the Chinese language as spoken around the Tang and Song dynasties, which is today known as Middle Chinese.

In Kanjijump, kanji are presented with on'yomi along with Romanized Middle Chinese readings, as a way of illuminating the relationships between characters, on'yomi, and sound components. You don't need to worry about learning Middle Chinese. But if you find yourself wondering about questions like those mentioned above, these readings are there to help satisfy your curiosity.

The origins of the on'yomi

The first step to understanding all these quirks about the kanji and their on'yomi is to realize the following:

  • The on'yomi are Japanese corruptions of Chinese sounds. Just like the more recent katakana loanwords words borrowed into Japanese, the on'yomi represent Japanese speakers' attempts to pronounce foreign words within the comparatively tight restrictions of their native sound inventory.
  • The Japanese language has changed over the centuries. This shouldn't come as a surprise; all languages change over time. Since the on'yomi were imported into Japanese so long ago, we can't understand them without also understanding the gradual shifts in Japanese pronunciation over the centuries.

Now, if you've been studying Japanese for a while, you might already know all this. But unless you've studied some linguistics, you may not realize that both these processes unfold according a highly consistent set of rules. This means that, with a little bit of knowledge of the original Chinese sources of the on'yomi, you can largely predict how any given kanji's on'yomi will sound. Even for those tricky kanji with varying on'yomi, there are often patterns governing exactly how their on'yomi can vary.

In this way, there are clear parallels between the on'yomi and the countless English loanwords in Japanese. Once you figure out the patterns of how English sounds get transformed into Japanese sounds, you can easily guess how a Japanese rendition of any given English word will sound, and you can write it in katakana. (If you're at the beginning of your Japanese studies, rest assured, you will gain this skill soon enough. Though right now you may find yourself dumbfounded at learning that something like マクドナルド makudonarudo is supposed to be mean MacDonald's, it's just a matter of being exposed to enough words for the patterns to become clear.)

Of course, you probably don't speak Middle Chinese, so the patterns governing the forms of the on'yomi won't just reveal themselves to you through exposure to lots of kanji. For that reason, all throughout Kanjijump, a simplified rendition of the Middle Chinese reading is given next to each kanji.

This notation is based largely on English spelling conventions so you don't have to worry about learning the gritty details of Middle Chinese phonology at the same time as you're learning the kanji. Here is the basic run-down of the few tricky parts.

A very rough Middle Chinese pronunciation guide

We don't know exactly what Middle Chinese sounded like at any point in its centuries-long history. The notation used here is meant to give you only a very rough idea of how the kanji might have been pronounced.


We do know that the consonant inventory of Middle Chinese probably included some sounds with no equivalent in most standard English.

<kh> as in the final sound of loch, Bach, or like Spanish J in jamón
<gh>the voiced counterpart to kh, like the G of Spanish between vowels, as in amigo, or like the /g/ in some fast Japanese speech, as in kage. A close-enough substitute would be the R sound of French or German.
<nj>something like an N sound followed by a soft J sound (as in Jacques), or just think of it like a Spanish Ñ.
<mv>this was somewhere between an M sound, a V sound, and a W sound. You can think of it as a W, if that makes things easier.
<tr> <ttr> <dr>just like <t>, <tt>, and <d> but pronounced with the tongue curled back.
Acceptable approximations would be the sounds in tree, mattress, and dragon.
<sy> <jy> <chy> <tchy>you can get by pronouncing these the same as <sh> <j> <ch> <tch>.
These are the palatal counterparts to those sounds, pronounced with the tongue a bit higher in the mouth--more like the onset of Japanese し shi than English she.

We also know that some standard English consonants that only appear in the middle of words can appear at the start of words in Middle Chinese.

<ng>as in singer, NOT as in finger.
<pp> <tt> <kk>as in in pepper, twitter, trekking,
or as in the second consonant in spoon, stop, skunk.
In other words, double consonants don't have aspiration.
<tch>the unaspirated version of <ch>, pronounced just as in kitchen, NOT as in children.
<ts>as in the consecutive <ts> and <h> in the phrase hits hard.
<tts><dz>as in putts and adze.

All the other consonant letters and letter combinations represent sounds that are roughly equivalent to common English sounds, as far as we know.

<p>as in pool
<b>as in big
<f> <ph>as in far, photograph
<v>as in vast
<d>as in dog
<n>as in noodle
<l>as in lion
<s>as in song
<z>as in zoo
<j>as in judge, or even like the G in genre
<ch>as in children
<sh>as in shoe
<y>as in yes
<k>as in kite
<g>as in go, NOT as in giraffe
<r>as in rag.

Especially where <r> shows up in a reading, you can be sure that linguists don't agree much on the details of its pronunciation. You can think of the <r> in this notation as marking anything from a consonant, a semivowel, to even a feature on adjacent vowels.


Linguists haven't settled on all the distinct vowel sounds of Middle Chinese, and so this notation doesn't attempt to represent them all. As a student of Japanese trying to understand patterns in the on'yomi, you don't need to worry about their exact values.

For the limited purposes of Kanjijump, you could get away with pronouncing the vowels like you're used to in Japanese. Alternatively, you can go with the conventions from any language with a "classical" set of five written vowels, e.g. Spanish, Italian, Swahili, Indonesian, Hawaiian. That is, anything vaguely like:

<a>as in cat, or cha-cha
<e>as in get, or melee
<i>as in kiwi
<o>as in polo (though in some places, this marks a sound probably closer to love)
<u>as in tutu.
Next to <i>, <y>, and <e>, or with two dots <ü>, this may have sounded closer to use, or something like the German über or French déjà vu.

Vowels with circumflex accent (â, î, û) represent sounds that were different than their accent-free counterparts, but these aren't important for our purposes in Kanjijump.


Middle Chinese had tone, much like its daughter languages Mandarin, Cantonese, etc. This is almost the same thing as Japanese pitch accent, in that a word's pitch contour is an essential part of that word. That is to say, if you say a word with a different pitch contour than the customary one, you may end up actually saying a different word.

Take the Japanese sentence, 関係ない。 Kankei nai. In standard Japanese, the pitch in this sentence will first rise, then stay mostly flat, and finally drop quickly.


Arrows just like these are used in the Middle Chinese transcriptions throughout Kanjijump to mark rising, flat, and falling tone.

For all syllables ending in /p/, /t/, and /k/ in Middle Chinese, it is likely they had a mostly flat pitch. Since pitch contour is never contrastive for these syllables, Kanjijump's notation does not apply tone marks to them.

Again, this is only a rough guide to Middle Chinese pronunciation. The exact pitches and contours of the different tones of Middle Chinese at any given point in time are not clearly known, and they almost certainly changed much throughout the centuries, just like the rest of the Middle Chinese sounds.

Really though, don't sweat it.

I can't stress enough that this notation is only meant to give you a rough idea of a word's Middle Chinese pronunciation. I am not advocating for any specific reconstruction of Middle Chinese sounds--I am just trying to help you make sense of the evolution of the on'yomi.

If you're interested how this notation system lines up with actual scholarly reconstructions of Middle Chinese, you can check out the details here. But be warned--the deeper you go, the murkier the waters get.

Going from a Middle Chinese reading to a Japanese on'yomi

Coming soon! Follow me on Twitter for updates. @justsilvestrein