Computational analysis of differences in Indian and American poetry

Abstract

Poetry is a verbal art which is motivating the human race from centuries. Indian authors’ English poetry has its own signature in the world poetry. Though it has great significance when compared with Western poetry, very little work has been done in the areas of authorship affinities, classification, style similarity of poets, and comparative studies. In this work, we investigated style and semantic differences between Indian and Western poetry and also compared the poetry in terms of variation in the usage of words and stylistic features such as orthographic, syntactic, and phonetic features. To capture style and variation differences, we considered 84 style features that cover structural, syntactical, sound devices of poetry and computed TF-IDF values using a bag of word method; then, we computed cumulative TF-IDF value of each word across all poems and arranged the values in decreasing order of their cumulative TF-IDF value. Later, we applied ranks and used PCA, LSA, and Spearman correlation to find variance in usage of words by Indian authors’ English poetry and Western poetry and style differences. We observed 40% semantic difference and 30% style difference between Indian authors’ English poetry and Western poetry. Our comparative analysis says that the features that work well with one country poetry may not be necessary to perform well with other country poetry.