Kanjis are Chinese characters imported into Japan from China, each one representing a particular concept. Kanjis were the first Japanese written form. The Japanese and Chinese grammars are very different, though, so the Japanese had to be resourceful at adapting the Chinese characters.
one symbol, many sounds
Many kanjis are difficult to tell apart, most kanjis are difficult to write, and there are many kanjis; to be able to read a newspaper, we need to learn about 2,000 kanjis, but there are thousands more, well over 50,000. Daunting as all of this sounds, the number of kanjis is not the main difficulty of learning them. What makes them challenging is that their meaning and sound changes depending on the context.
For example, consider the kanji 日 in the following words:
two days ago
day after tomorrow
1st of the month
2nd of the month
きょう, こんち (kyou, konchi)
あした, あす (ashita, asu)
As amateur decoders, given these many words that contain 日, together with their sounds, we would expect it would be easy to figure out how to pronounce 日. Good luck with that!
日, the kanji for ‘sun’ sounds ‘hi’; however, 日 sounds ‘ni’ when we use it to write 日本 (ni-hon, Japan), and 日 appears two times in the word 日曜日 (nichi-you-bi, Sunday), where it sounds ‘nichi’ and ‘bi’. In other contexts, 日 will have other meanings and sounds. These changes are not just simple variations in pronunciation, but completely different sounds. The table has more words that contain 日, and none of them contains the sound ‘hi’, ‘ni’, ‘nichi’, or ‘bi’. 😓
There is a large variation of the meaning and sound of a kanji depending on the context in which it appears, which makes kanjis challenging to learn. In essence, we have to memorize each different kanji in each different context in which it may appear. Although this sounds daunting, English has the same issue with the pronunciation of its letters. For example, the sounds of the ‘s’ in ‘sun’, ‘sugar’, ‘aisle’ and ‘wolves’ are completely different [TWD], and this variability is the norm in English. Think of words like ‘knight’, ‘choir’, ‘muscle’, ‘have’, ‘tomb’, ‘psyche’, ‘colonel’, ‘Carlisle’, etc. There are hundreds of words in English whose pronunciation we just have to memorize.
kun-yomi and on-yomi
Part of the reason for these variations in sound is that some pronunciations come from the original Japanese sound of the word, while others come from the Chinese sound of the word. Usually a kanji has at least two ‘yomi’ (readings): the ‘National-country (kuni) reading’, or kun-yomi, is the Japanese sound of what the kanji represents, while its ‘sound reading’, or on-yomi, is the Chinese sound of the character. For example, ‘person’ in Japanese is ‘hito’, and in Chinese is ‘ren’; hence, the kanji for person, 人, has the kun-yomi ‘hito’, and the on-yomis ‘jin’ and ‘nin’, which sound like ‘ren’ [jisho.com].
We usually use kun-yomis when the word is by itself, e.g., ‘sono hito’ (その人 – that person), and on-yomis when the kanji is part of a compound word, e.g., ‘america-jin’ (アメリカ人 – American person), or ‘san-nin’ (三人 – three persons). This is similar to English having words from different origins meaning the same thing; for example, ‘water’ (germanic root), ‘aqua’ (latin root), and ‘hydro’ (greek root) all mean ‘water’, but we use them in different contexts: while we usually use ‘water’ by itself, like in ‘clear water’ or ‘water park’, we use ‘aqua’ and ‘hydro’ only as parts of compound words, like ‘aquamarine’, or ‘hydroelectric’.
We might get the impression that a kanji would have only one on-yomi (Chinese reading) but, actually, the Chinese characters were imported into Japan at three different times, from different regions of China:
- the go-sound pronunciation was imported in the 5-6th centuries from the Wu region
- the kan-sound pronunciation was imported in the 7-9-th centuries from the central Chang’an region
- the tou-sound pronunciation was brought by monks into Japan in the 12-13th centuries.
Hence, a given kanji might have many on-yomis depending on what was the pronunciation of the character each time it was imported. For example, for 日 we have the following readings [Jisho]:
that explain the sound of many words that use 日, like 日本 [にほん] (に here is an abbr. of にち), 日曜日 [にちようび], 本日 [ほんじつ], and 二日 [ふつか].
The kun-yomi and on-yomi readings of a kanji explain the sounds of many of our sample words. However, how do we explain the sound of, for example, 昨日 [きのう], 今日 [きょう], or 明日 [あした]? We cannot match their sounds to any of the kun-yomis or on-yomis of 日.
All languages borrow words from other languages. Japanese borrowed a large amount of characters from Chinese. The Chinese representation for ‘Sun’ was 日, while the Japanese sound for ‘Sun’ was ひ, so the Japanese borrowed 日 and called it ひ. Likewise, the Chinese character for ‘now’ was 今, while the Japanese sound for ‘now’ was いま, so the Japanese borrowed 今 and called it いま.
Now, let’s consider the words for ‘yesterday’, or ‘today’, or ‘tomorrow’, which the Japanese call きのう, きょう, あした. It turns out that the Chinese have no single characters to represent those concepts but, instead, they represent them as combinations of concepts: ‘yesterday’ is 昨日 meaning ‘last-day’, ‘today’ is 今日 meaning ‘now-day’, and ‘tomorrow’ is 明日 meaning ‘clean-day’. Thus, the Japanese were in a bind, e.g., 日 was already ‘Sun’ and called ひ, while 今 was already ‘now’ and called いま, so what do they assign きょう to? The solution is to disregard the individual meanings and sounds of the 日 and 今 characters, and simply call きょう as the combination 今日 and give it the meaning ‘today’.
Calling 今日 as きょう is like giving the whole word a kun-yomi. This type of word borrowing in which a meaning and sound is assigned to a combination of characters regardless of the characters themselves is called ‘ortographic borrowing’ or, in Japanese, jukujikun. Japanese has a large number of such borrowings from Chinese and other languages [WP]. English has done some ortographic borrowing too, e.g., pronouncing the ancient Greek ‘X’, an abbreviation of Χριστός, as ‘Christ’ in the word ‘Xmas’, and the latin character ‘&’ as ‘and’ [WP].
The on-yomi of 今日, of course, is another story because here we are explicitly using the Chinese sounds. Hence, the on-yomi of 今日 is こんにち, where こん is an on-yomi of 今, and にち is an on-yomi of 日; the on-yomi こんにち is the base of 今日は or こんにちは, i.e., ‘about now-day’, meaning ‘hello’. Hence, now we have a kun-yomi and an on-yomi, not for a kanji but for a word:
In summary, the reading きょう of 今日 is a kun-yomi for the multi-character word and it tells us absolutely nothing of the kun-yomis or on-yomis of the individual kanjis that compose it. Hence, when we find a multi-kanji word whose kun-yomi doesn’t decompose into the kun-yomis and on-yomis of the individual kanjis of the word, chances are it was orthographically borrowed. Hence, we have no option but to memorize the kun-yomi of the word independently of the kun-yomis and on-yomis of the kanjis that form it.
Although memorizing the pronunciation of a word that has nothing to do with its writing might seem like an unusual difficult task, it turns out that English has a similar situation. The role of the kanjis in 今日 is to merely give us a hint of what the word means, so we can pronounce it correctly, from memory. This is like reading the English word ‘aisle’ and using it as a hint to retrieve all the exceptions that we know apply to it: the ‘a’, ‘s’ and ‘e’ are silent, and the silent ‘e’ forces the long-vowel sound of the ‘i’ so we end up with an ‘ah-eel’ pronunciation that looks nothing like the spelling of the word. Hence,
For the Japanese-speaker learning English, ‘aisle’ is reminder that the word is pronounced ‘ah-eel’, and has little to do with the characters that form the word.
For the English-speaker learning Japanese, 今日 is a reminder that the word is pronounced きょう, and has little to do with the characters that form the word.
Another reason for a change in pronunciation is to let the sound ‘get along’ with the sounds that precede it or follow it; technically this is called euphony.
For example, the kanji 本, hon, means ‘origin’ when we write ‘Japan’; 本 also means ‘book’, which in old times were written in long thin scrolls; hence, 本 is also used as the counter of any other similarly long thin objects, like bottles, cigarettes, or arrows. Depending on the number of items, 本 is pronounced ‘hon’, ‘ppon’, or ‘bon’, e.g., ‘ippon’ (一本) might be one bottle, ‘nihon’ (二本) two bottles, and ‘sanbon’ (三本) three bottles. Notice that now we have two ways to write ‘nihon’: 日本, meaning ‘Japan’, and 二本, meaning ‘two long thin objects’. Hence, we really cannot write the word ‘ni-hon’ using kanjis unless we know the context in which we are writing it.
Euphony sometimes turns voiceless consonants into voiced ones when they start the second part of a compound word, e.g., “h” becomes “b”, “k” becomes “g”, “s” becomes “z”, “t” becomes “d”, “sh” becomes “j”, etc. For example, here is a list of pairs of compound surnames that use a given kanji; the first surname preserves the voiceless sound of the stand-alone kanji, while the second surname uses the voiced sound:
田 (rice field)
田 (rice field)
We can see above that both Nakada and Nakata have the same spelling, 中田, so the writing in kanji doesn’t tell us how the person pronounces their surname.
This also means that sometimes pronouncing a compound word that uses the same kanji twice might not be as simple as pronouncing the same kanji twice, e.g.,
- 人々 = hito-bito (people)
- 花々 = hana-bana (flowers)
- 島々 = shima-jima (islands)
- 木々 = ki-gi (trees)
- 口々 = kuchi-guchi (every mouth)
- 月々 = tsuki-zuki (every month)
- 時々 = toki-doki (sometimes)
The character 々 is called noma because we write it with the katakanas ノ (no) and マ (ma); the function of 々 is to repeat the previous kanji, i.e., 花々 is exactly the same as 花花, and we can write it either way; often writing 々 is convenient because it can replace a many-stroke kanji, though. As we can see in the examples, we sometimes repeat the kanji to pluralize it, e.g., depending on context 花 (hana) could be flower or flowers, but 花々 (hana-bana) is always flowers.