Logo Logo
Help
Contact
Switch language to German
A comparative analysis of metal subgenres in terms of lexical richness and keyness
A comparative analysis of metal subgenres in terms of lexical richness and keyness
Metal music is realized under a vast variety of subgenres all of which have their unique (or shared) characteristics not only in sound but also in their lyrics. Much research has been done to distinguish or classify subgenres but little has addressed the linguistic differences across them. This study seeks to find out the lexical richness and keyness levels of heavy metal, thrash metal and death metal using a corpus of 200 songs from each subgenre with a total of 600 songs. The selection of the bands and songs was carried out finding references in the metal literature. The metal literature in the present study takes into account the academic books and articles on metal as well as noteworthy media productions, websites and metal blogs such as Metal Evolution and Encyclopaedia Metallum. The song lyrics were manually processed and meta-data, mark-ups and repeats have been removed so that the differences in repeat lengths do not affect the comparisons. Furthermore, the analyses used in the study are sensitive to repeats as they measure the frequencies and repeat ratios of the words. The song lengths – after the processing – were limited to lower and upper thresholds of 100 and 400 words. The songs were analyzed for their lexical richness levels in three aspects: 1) lexical variation, 2) lexical sophistication and 3) lexical density. Lexical variation was operationalized as TTR, Guiraud, Uber and HD-D. Lexical sophistication was measured using lexical frequency profile with two different frequency lists – the GSL and the BNC/COCA – by looking at the ratios of tokens and types which fell beyond the most frequent two thousand words (Laufer 1995). Another sophistication measure – P_Lex – which also runs on GSL, was applied. Lexical density analysis was based on the ratio of content words to all tokens in the texts. In order to complement this quantitative and data-driven approach, a keyness analysis was administered to add a qualitative dimension to the research. All lexical richness analyses pointed out to statistically significant differences between all subgenres, marking heavy metal as the least and death metal as the most lexically rich one. Keyness analysis indicated differences among all three subgenres as well. Heavy metal key words tended to be Dionysian whereas thrash and death metal keywords were more Chaotic as proposed by Weinstein (2000). Finally, a correlation analysis showed that all lexical richness measures were statistically significantly correlated to each other. Based on the findings, it could be claimed that 1) these three subgenres differ from each other not only in terms of music but also of lexical richness levels and key words and 2) lexical richness analyses, coupled with keyness, are capable of reflecting the genre differences in song lyrics. However, as a result of a discriminant analysis of the present corpus, a reverse approach whereby genres are attempted to be classified based on lexical features does not provide a pattern which fully corresponds to the existing classifications.
metal, heavy metal, thrash metal, death metal, genre, subgenre, lexical richness, lexical sophistication, lexical density, lexical variation, keyness
Kahraman, Volkan
2020
English
Universitätsbibliothek der Ludwig-Maximilians-Universität München
Kahraman, Volkan (2020): A comparative analysis of metal subgenres in terms of lexical richness and keyness. Dissertation, LMU München: Faculty for Languages and Literatures
[img]
Preview
PDF
Kahraman_Volkan.pdf

6MB

Abstract

Metal music is realized under a vast variety of subgenres all of which have their unique (or shared) characteristics not only in sound but also in their lyrics. Much research has been done to distinguish or classify subgenres but little has addressed the linguistic differences across them. This study seeks to find out the lexical richness and keyness levels of heavy metal, thrash metal and death metal using a corpus of 200 songs from each subgenre with a total of 600 songs. The selection of the bands and songs was carried out finding references in the metal literature. The metal literature in the present study takes into account the academic books and articles on metal as well as noteworthy media productions, websites and metal blogs such as Metal Evolution and Encyclopaedia Metallum. The song lyrics were manually processed and meta-data, mark-ups and repeats have been removed so that the differences in repeat lengths do not affect the comparisons. Furthermore, the analyses used in the study are sensitive to repeats as they measure the frequencies and repeat ratios of the words. The song lengths – after the processing – were limited to lower and upper thresholds of 100 and 400 words. The songs were analyzed for their lexical richness levels in three aspects: 1) lexical variation, 2) lexical sophistication and 3) lexical density. Lexical variation was operationalized as TTR, Guiraud, Uber and HD-D. Lexical sophistication was measured using lexical frequency profile with two different frequency lists – the GSL and the BNC/COCA – by looking at the ratios of tokens and types which fell beyond the most frequent two thousand words (Laufer 1995). Another sophistication measure – P_Lex – which also runs on GSL, was applied. Lexical density analysis was based on the ratio of content words to all tokens in the texts. In order to complement this quantitative and data-driven approach, a keyness analysis was administered to add a qualitative dimension to the research. All lexical richness analyses pointed out to statistically significant differences between all subgenres, marking heavy metal as the least and death metal as the most lexically rich one. Keyness analysis indicated differences among all three subgenres as well. Heavy metal key words tended to be Dionysian whereas thrash and death metal keywords were more Chaotic as proposed by Weinstein (2000). Finally, a correlation analysis showed that all lexical richness measures were statistically significantly correlated to each other. Based on the findings, it could be claimed that 1) these three subgenres differ from each other not only in terms of music but also of lexical richness levels and key words and 2) lexical richness analyses, coupled with keyness, are capable of reflecting the genre differences in song lyrics. However, as a result of a discriminant analysis of the present corpus, a reverse approach whereby genres are attempted to be classified based on lexical features does not provide a pattern which fully corresponds to the existing classifications.