Quantitative and qualitative analysis of a multi-genre heavy metal corpus: an NLP-based approach

Heavy metal is a music genre that originated at the end of the 1960s, rose to prominence in the 1980s, and has evolved into numerous subgenres ever since. Its prolific nature has made it so diverse in terms of musical discourses, social practices, and cultural meanings that it would be more appropriate to define it as a meta-genre. Indeed, owing to its richness and diversity, heavy metal lends itself to being the object of linguistic analysis. The present work aims to investigate the lyrics of 10 subgenres of heavy metal using the tools of Natural Language Processing (NLP) applied to an ad hoc corpus consisting of 1,091,054 tokens. The results confirm that the umbrella term heavy metal encompasses very different styles in terms of lexical richness, themes, and sentiments. It embraces lexically sophisticated subgenres such as black metal and death metal, characterised by extensive use of rare words, as well as mainstream subgenres such as glam metal, metalcore, and nu metal, based upon simpler words and less diverse vocabularies. The themes cover the whole spectrum of human existence, from life and death to love and pain, and are conveyed through a wealth of different narrative styles, ranging from grindcore’s fierce anger to glam metal’s glossy tones. KEYWORDS: Heavy metal. NLP. Sentiment Analysis. LDA.