Japanese may be analyzed in terms of the classical Greek and latin unit called a mora. A mora is a unit of timing. Each mora takes about the same length of time to say. The most common type of Japanese mora is formed by a consonant followed by a vowel. Using the automatic segmentation of speech processing method, we analyzed the effect factors of each mora duration in many various ways to find a regularity of mora rhythm patterns. For some effect factors, we transformed the durations in milliseconds into relative values with zero mean and unit variance using z-transform. Consequently, mora lengths shorter than the mean have a negative value, and those longer than the mean have a positive one. This usage was convenient to identify lengthened or shortened.
