Quantifying degrees of randomness in word rhythms of literary works

Abstract

Rhythm is an intrinsic quality in literature. Intuitively, poems are more rhythmic than novels, and novels are more rhythmic than technical papers. But how can rhythm be effectively quantified? A natural approach is to apply word length. We theorize that the various rhythms in literary works can be characterized by applying suitable linear and nonlinear variability measures. These measures are applied to a time series representation of the literary work with the dependent variable being the number of characters in each word of the literary text. Variability measures, such as the fluctuation index and approximate entropy , are shown to be appropriate rhythmic quantifiers, effectively distinguishing the quantity of rhythm in various types of literature. When the number of letters per word is replaced by the amount of time required to speak a word, a high correlation is found between the fluctuation index and approximate entropy. This provides some empirical evidence that word length is an appropriate representation of rhythm, and demonstrates that the fluctuation index and approximate entropy are effective quantifiers of rhythm.