In Mandarin, each character has one reading (plus variations sometimes, as with 不 or 一, but these can always be put down to tone sandhi).
In Japanese, a character may have many readings but a particular word will always be read in a particular way (except that かざぐるま and フーシャ are both written as 風車.)
And if you look up a chinese character in Unihan, for Cantonese you might get more than one reading. See this example
http://www.unicode.org/cgi-bin/GetUniha ... eutf8=true
The readings for 聿 are jyut6, leot6 and wat6. Why is this?
Is this situation more like the Mandarin or more like the Japanese?