In the historical studies of Japanese accent, especially in those using the written texts, the pitch accent is often described with the discreet levels. Since this 'level' view is sometimes subject to criticism, I review the discussions concerning the notions such as 'tone-bearing units' or 'compound accent' and examine their validity. The purpose is not to argue against the 'non-level' view, but to develop further the insights obtained both by the level and non-level approaches. The discussion here would contribute toward establishing the method of the historical research of Japanese accent which is more practical both for the text and field researches and which at the same time can approach the true nature of Jananese accent.
