A Computational Linguistics Approach to Analyzing Haiku

By Jay Friedenberg and Colleen Farrelly Haiku have been studied qualitatively for decades but there are few statistical approaches to understanding its properties. In this presentation we have analyzed the winning poems from all of the HSA contests since inception and compared them to a representative and random sample of haiku and senryu from every published issue of Frogpond, the society's journal. We note the similarities and differences between exceptional and "average" haiku in terms of word type and syntactical usage.