This thesis aims to clarify the mechanism of durational contrast between short/long vowels and single/geminate consonants in Japanese, which is transitional according to speech rate. In Japanese, quantitative contrast of vowels and consonants plays an important role in the semantic processing. Acquisition of the contrast is one of the biggest obstacles for second language (L2) learners of Japanese, besides prior researches found that improper deliverance of the contrast leads 'unnaturalness' or 'non-native-like' pronunciation in the speech. Researchers from L2 educational point of view have examined L2 learners' production and perception of the contrast extensively, however that of native production and perception nevertheless received little attention. From a standpoint to ultimately contribute to improve L2 pronunciation education, I believe that defining fundamentals of the native production and perception has a crucial importance. In the presentation, I will talk about the characteristics of the production and perception with quotations from experimental results that I conducted for my sub-thesis.