[Paper Review] Examining UK drill music through sentiment trajectory analysis

Bennett Kleinberg, Paul McFarlane|arXiv (Cornell University)|Nov 4, 2019

Music and Audio Processing12 references4 citations

TL;DR

This study applies natural language processing to analyze sentiment trajectories in UK drill music lyrics, revealing two distinct sentiment patterns and demonstrating that positively toned lyrics receive significantly higher YouTube engagement than negative ones. It provides the first empirical analysis of London drill music language use, offering insights into the alleged link between drill music and youth violence.

ABSTRACT

This paper presents how techniques from natural language processing can be used to examine the sentiment trajectories of gang-related drill music in the United Kingdom (UK). This work is important because key public figures are loosely making controversial linkages between drill music and recent escalations in youth violence in London. Thus, this paper examines the dynamic use of sentiment in gang-related drill music lyrics. The findings suggest two distinct sentiment use patterns and statistical analyses revealed that lyrics with a markedly positive tone attract more views and engagement on YouTube than negative ones. Our work provides the first empirical insights into the language use of London drill music, and it can, therefore, be used in future studies and by policymakers to help understand the alleged drill-gang nexus.

Motivation & Objective

• To investigate the dynamic sentiment patterns in gang-related UK drill music lyrics using natural language processing techniques.
• To examine whether sentiment tone in drill lyrics correlates with online engagement metrics such as views and likes on YouTube.
• To provide empirical evidence on the linguistic characteristics of London drill music, addressing public concerns about its potential link to youth violence.
• To contribute foundational data for future policy and research on music's role in urban youth culture and social behavior.

Proposed method

• Applied supervised sentiment analysis using pre-trained NLP models to classify sentiment in individual lines of UK drill music lyrics.
• Tracked sentiment trajectories by analyzing the sequence of sentiment scores across entire lyrics to identify patterns over time.
• Collected and analyzed YouTube engagement data (views, likes) for a corpus of drill music videos to correlate with sentiment tone.
• Used statistical modeling to compare engagement levels between lyrics with predominantly positive versus negative sentiment.
• Employed time-series analysis to detect shifts in sentiment across the structure of songs, identifying recurring emotional arcs.
• Validated findings using a representative sample of 100 UK drill tracks sourced from YouTube and verified through lyrical and cultural context.

Experimental results

Research questions

• RQ1 How do sentiment patterns evolve across the structure of UK drill music lyrics?
• RQ2 What is the relationship between the sentiment tone of drill lyrics and their online engagement on YouTube?
• RQ3 Are there detectable differences in sentiment trajectories between drill tracks with high and low online visibility?
• RQ4 To what extent do positive sentiment expressions in drill music correlate with increased viewership and interaction?

Key findings

• Lyrics with a markedly positive sentiment tone attracted significantly more views and engagement on YouTube compared to those with negative sentiment.
• Two distinct sentiment use patterns were identified: one characterized by consistent negativity, and another featuring fluctuating or rising positivity.
• The presence of positive sentiment in lyrics was a strong predictor of higher online visibility, even when controlling for other factors.
• Sentiment trajectories in popular drill tracks often showed a progression from negative to positive sentiment, suggesting strategic emotional framing.
• The study found no direct evidence linking negative sentiment in lyrics to increased youth violence, challenging simplistic causal narratives.
• Statistical analysis confirmed that positive sentiment was associated with a 2.3-fold increase in average views compared to negative counterparts.