The kanji 山 is often read yama when it appears on its own, but in words such as 火山 it is read zan, and in 山脈 it is read san. The reason the same kanji can have different readings is that Japanese kanji often have both on-yomi and kun-yomi readings.
On-yomi are readings that developed from Chinese pronunciations of kanji and were adapted to fit the sound system of Japanese. They are often used in words made up of two or more kanji, such as 山脈 (sanmyaku, mountain range), 火山 (kazan, volcano), and 学校 (gakko, school).
Kun-yomi, by contrast, are readings created by linking the meaning of a kanji with a Japanese word that already existed before kanji were introduced. Reading 山 as yama and 川 as kawa are simple examples. Kun-yomi are also common in words with okurigana, such as 食べる (taberu, to eat) and 高い (takai, high or expensive).
But it is not as simple as saying, “Two kanji together means on-yomi, and one kanji means kun-yomi.” Some words mix the two. 場所 (basho, place), for example, combines the kun-yomi ba with the on-yomi sho. Personal names and place names can be even harder to predict.
This becomes clearer when you look at how Japanese people actually read kanji. A Japanese reader does not normally see 火山 and think, “火 uses the on-yomi ka, and 山 uses the on-yomi zan, so the word must be kazan.” The word 火山 and its reading kazan are usually learned together as a single piece of vocabulary.
That is why even Japanese people can hesitate over unfamiliar place names. In Osaka, for example, there is a place written 放出 and pronounced Hanaten. The same two kanji in the ordinary word 放出 are read hoshutsu, so many Japanese people would not guess Hanaten correctly the first time they saw the place name.
Understanding on-yomi and kun-yomi gives learners an important clue for working out kanji readings. But it cannot predict everything. Japanese people themselves do not read every kanji by calculating the rules one character at a time. In practice, many words are learned together with their readings.
