Entropy of Sounds: Sonnets to Battle Rap

AbstractPoetry and lyrics across cultures, from Sonnets to Rap, demonstrate an obvious human cognitive capacity for the perception and production of various multi-syllable sound patterns. Here we use entropy to measure discrete serialized representations of phones and to explore the complexity of these sound structures across genres of creative language arts. The present exploratory analysis has two main objectives. First, our aim is to broaden the scope of cognitive processes and data that are considered in statistical learning approaches to phonological learning and language acquisition. Second, we hope to to provide a basis for more targeted computational and phonological investigations of these patterns. We compare the conditional entropy of sequences of phonological patterns in lyrics and find that, in general, Battle Rap and Sonnets maintain noticeably lower entropy than other genres across sequence sizes, while lyrics from Electronic music and Hip-Hop display relatively high entropy.

Return to previous page