Previous research in the sentimental analysis of lyrics has tended towards overarching research across popular songs without differentiating between artists, genres, or genders. A specifically gender-based comparison is a less well-explored avenue, and this gap is what has motivated our research. 

 

An example of such research is the project conducted by Salim Zubar - ‘Sentiment Analysis of all Billboard Hot 100 Songs Over Time (1958–2019)’. Zubar’s used the LyricsGenius API Client on his representative sample of Billboard Hot 100 songs, finding the overall sentiment was relatively neutral, but modern music has a tendency towards more negative sentiments. He displayed his findings in a single line graph and offered a handful of the most popular lyrics in 1958 and 2019 (Zubair, 2020). Another example is the research by Cristobal Veas, looking specifically into Metallica’s lyrics. Veas offered more details in his displayed results, using word clouds, word frequency tables, and a scatter plot mapping each song’s positive and negative valence (Veas, 2020). His specific selection of Metallica clearly limits the gender and the genre of the lyrics he analysed. 

 

Taking a step back from lyrics to general speech opens the door to a much broader debate about the existence and prevalence of gender-based speech characteristics. It is a controversial topic, with many studies at odds with each other (Stokoe, 2018). It has been theorised that concrete, distinguishing features of speech do exist, and that they are both systematic and socially significant (Gallee, Levitt, 2016). An example of this is the tendency for women to use more rapport-style speech, as opposed to men’s tendency to use report-style speech to establish dominance (Gallee, Levitt, 2016). Whether or not this is true, it remains difficult to establish the extent that such characteristics could be applied to lyrics, as messages are conveyed differently between speech and song. Assumptions are thus made cautiously. Song-writing is incredibly lucrative, and songs are becoming increasingly homogenous (Serrà. Et al, 2012). It is safe to assume that a significant proportion of popular song lyrics are written to appeal to the masses, thus may exhibit different patterns than everyday speech.