Kanji are characters that originated in ancient China. In the very beginning, writing in China was not created to represent sounds — it was created to represent meanings, objects, like a tree, a person, or the sun.
However, as the system developed, they began using symbols for their sounds as well.
At first, a character like 木 (tree) was simply a drawing meaning “tree.” Later, the same character could also be used as part of another word because it had the sound “mu.”
This shift gave birth to the idea of phonetic borrowing: using a character not for its meaning, but for the sound it represented. This system is called phono-semantic compounds, and the majority of Chinese characters (and therefore kanji) work this way today. Many characters combined both aspects:
One part of the character suggested meaning.
Another part hinted at pronunciation.
Before writing was introduced, the Japanese language existed only as an oral language. People spoke Japanese, but there was no system for recording it in writing. This changed when Chinese characters were introduced to Japan around the 4th–5th century CE.
Later, they adapted the characters to represent Japanese words. This lead to characters that have a Chinese reading (On Reading) and a Japanese reading (Kun Reading).
Note that over time, some kanji were simplified into purely phonetic symbols which lead to Kana. They are actually old kanji.
Kanji: 暗 (dark / darkness)
Chinese Origin
Radical / Semantic part: 日 (“sun / light”). It indicates that the meaning has something to do with light—or in this case, the absence of light, which gives the sense of darkness.
Phonetic part: 音 (“sound”). 音 was pronounced roughly "im".
Japanese Origin
The oral word for dark in japanese is くらい "Kurai". To write it, they used the chinese character 暗, as 暗い.
"Im" is not a sound that exist in japanese so it was adapted to あん "an".
Modern Times for 暗
Meaning : Darkness
On’yomi (Chinese Reading) : あん
Kun’yomi (Japanese Reading) : くら.い
Modern Mandarin pronounces 暗 as àn (fourth tone), which is different from the older pronunciation that Japanese borrowed.
We’ve seen that katakana is usually used for foreign words. That includes words borrowed from China. This is why kun’yomi (native Japanese readings) are written in hiragana, while on’yomi (Chinese-derived readings) are often written in katakana.
When a kanji is used on its own (as a noun, verb, or adjective), it usually takes the kun’yomi reading. The Japanese word already existed, and the matching Chinese character was later assigned to represent it.
When a word is made of multiple kanji, it usually takes the on’yomi readings. These are compound words created using Chinese characters, and they kept their original Chinese-based sounds.
東 (ひがし) = east (kun’yomi)
京 (みやこ) = capital (kun’yomi)
東 (トウ) + 京 (キョウ) = 東京 (トウキョウ) → “Tokyo" = "Eastern Capital” (on’yomi)
When combining kanji in compounds, pronunciation often shifts to make words easier to say:
火山 = 火 (カ) + 山 (サン) → pronounced カザン, not カサン. (The s sound becomes z between vowels.)
楽器 = 楽 (ガク) + 器 (キ) → pronounced ガッキ, not ガクキ. (The k sound doubles to form a small pause or glottal stop.)
It’s important to remember that language evolves over time. This is why kanji can have multiple meanings, on’yomi (Chinese-derived readings), and kun’yomi (native Japanese readings).
It also means that a kanji’s radical may sometimes seem illogical, because its meaning doesn’t always match the character as a whole.
There are patterns that are worth learning, as they apply to many kanji, but there are no strict rules — exceptions are common.
There’s no single “best” way to learn kanji. Kanji involve meaning, common components, readings, vocabulary usage, and frequency. Different methods emphasize different aspects.
I have created two resources that can help you.
This is my kanji deck. I've been working on it for years. It is without a doubt the most complete kanji deck that you can find. I regrouped every informartion that could be found and compiled them in this deck. It's used to be the most downloaded kanji deck on ankiweb before being taken for copyright issues since it contains information from the books mentioned below. But you can still download it here.
This is the public spreadsheet, you can find all the information you need. The kanji deck is based on that spreadsheet.
Kanji are often associated with a keyword, which represents the meaning they usually convey in different words.
Some can be easy like 日 means sun, 山 means mountain. But when the keyword of a kanji become more abstract, the link between the kanji's keyword and the meaning of the different words using that kanji will not be as evident. When studying keywords, always pay attention to the vocabulary to see how each kanji is used in context.
Many learners use mnemonics — stories linking parts to the meaning — to remember characters. It's important to remember than in kanji there's one part that gives meaning, and the other pronunciation. Mnemonics usually try to combine the meaning from all it's parts.
If you learn that 火 means “fire” and 山 means “mountain,” you might guess that 火山 means “volcano.” But you wouldn’t know how to read any of these words unless you also learned the on’yomi and kun’yomi for both kanji.
Because a single kanji can have multiple readings, it’s not practical to memorize every possible one in isolation. A better approach is to learn readings through vocabulary. By seeing the same kanji in different words, you’ll naturally start noticing patterns.
Some people argue you should learn each kanji individually with all its readings before moving on to vocabulary, while others say to skip kanji study and just focus on words. In reality, the two approaches complement each other — it’s best to combine them.
Take 私, made of 禾 (grain) + 厶 (personal). Its assigned keyword is “private,” which can be tricky to remember from the components alone. But 私 is also the pronoun “I,” one of the very first words you’ll learn, and it’s so common that you’ll never forget it.
Learning kanji in context, through verbs, adjectives, and nouns, makes them easier to remember. Conversely, if you already know the kanji inside a compound, it becomes much easier to guess its meaning and pronunciation.
The Japanese government’s official list, the Jōyō kanji (常用漢字), contains 2,136 characters taught in schools and used in newspapers, official documents, and daily life.
However, if you look at the 2,000 most frequent Japanese words, you’ll encounter only about 700 distinct kanji (depending on the source).
The downside of studying kanji strictly by frequency is that it can feel random. It’s often easier to learn kanji in logical groups — for example, characters sharing the same radical, or compounds built from kanji you already know. This way, new kanji connect to what you’ve already learned, making them stick more easily.
Heisig’s Remembering the Kanji (RTK) teaches kanji by breaking them into components and linking them to a single keyword. The method focuses on memorizing mnemonics to recall the keyword and write the kanji.
It does not cover origins, readings, vocabulary, or usage. As Heisig himself writes: “You will read nothing about how kanji combine to form compounds. Nor is anything said about the various ways to pronounce the characters. Furthermore, all questions of grammatical usage have been omitted.”
When Remembering the Kanji first came out (1977), there was nothing like it in English. Traditional methods taught kanji by rote: stroke order + readings + vocab all at once, which overwhelmed learners.
While groundbreaking, Heisig’s method is riddled with inaccuracies. He often alters the decomposition of kanji to fit his mnemonic stories, rather than basing the stories on genuine components. Since learners are supposed to recall the kanji’s form from these stories, this can result in writing characters incorrectly—defeating the method’s stated purpose. Later parts of the book have no stories at all, just the component's nicknames :
笑 (Laugh). Rtk gives the following components: "Bamboo . . . heavens." (⺮) + (天), but the lower part is actually 夭 (calamity), not 天. The top stroke is different.
It may sound trivial but if your purpose is to write kanji, strokes and strokes order does matter.
He also give the same components different names, and some components that are not the same, share the same name...
行く is the verb "to go". Easy enough to remember. The radical is 彳 (going). But Heisig gives that simple radical 3 differents names.
"Going": 復 徒 待 径 彼 役 徳 徹 徴 微 街 桁
"Columns" : 律 復 得 従 徒 待 往 征
"Line": 徒 待 径 徹 徴 微 街 衡 後 御 術 衛 徐 衝 循
RTK also introduces inconsistencies: the same component may be given multiple names, while entirely different components sometimes share the same name. After going through the book multiple times, it can feel less like you’re learning kanji themselves and more like you’re memorizing Heisig’s personal reinterpretation of their components and meanings.
To ease the writing method, most people doing RTK usually focus on recognition of the kanji, still using the same break down technique.
It was the first structured English method to tackle kanji systematically.
Writing by hand reinforces memory and if you can write a kanji, recognizing it becomes much easier.
Some keywords are obscure—especially for non-native English speakers—and can feel arbitrary.
The way kanji are broken down is often inconsistent.
The book provides no explanation of the actual origins of kanji, even when the historical derivation would be simpler than the invented mnemonic.
The sequence in which kanji are introduced is not always logical.
191 of the characters included are not part of the official jōyō list.
In today’s digital age, handwriting is less essential. Writing kanji also takes significantly more time than simply recognizing them.
Following the approach popularized by RTK, the Kodansha Kanji Learning Course (KKLC) builds on the same component-based method but improves the execution. Note that the goal here is kanji recognition, not writing kanji.
The learning order is more structured, with a consistent sequence of components, which creates smoother transitions between kanji—even if the difference isn’t dramatic.
Although the course still instructs learners not to focus on readings initially, it provides vocabulary examples that use only kanji you have already studied, reinforcing recognition and context.
Keywords are arguably better chosen than in RTK. There's usually a main keyword, and secondary ones to add nuances.