Statistical Prosody: Rhyming Pattern Selection in Japanese Short Poetry

Abstract. Rhyme patterns of Japanese short poetry such as HAIKU, SENRYU, SEDOKAs, and TANKAs are analyzed by a statistical approach. Here HAIKU and SENRYU are poems composed of only seventeen syllables, which can be segmented into five, seven, and five syllables. As rhyming both head and end rhymes are considered. Analyses of sampled works of typical poets show that for the end rhyme composers prefere the avoided rhyming, whereas for the head rhyme they compose poems according to the stochastic law. Subsequently the statistical method is applied to a work of SEDOKAs as well as to those of TANKAs being written with three lines. Evaluation of the khi-square statistics shows that for a certain work of TANKAs the feature being identical to that of HAIKU is seen. 1