Emotion Classification, Topic Modelling, and Discourse Evaluation of Audience Responses to SNL's Fast Fashion Sketch on Social Media : Leveraging RoBERTa, BERTopic and Discourse Analysis

This thesis explores audience responses to a SNL sketch about fast fashion, specifically on YouTube, Instagram, and TikTok, utilising a mixed-methods approach. It employs two computational techniques: sentiment analysis using a RoBERTa-based model, and topic modelling with BERTopic to analyse user comments, uncovering prevalent emotional responses and thematic clusters. The study also incorporates discourse analysis, grounded in Gee's framework (2005), to understand the nuanced, context-dependent meanings within the comments. It aims to understand the overall emotional profiles of comments, how users react to the sketch's context, and how their discussions reflect broader power relations concerning fast fashion, Chinese brands (SHEIN and Temu) vs Western brands (such as H&M and Zara among others), and ethical consumption. Key findings indicate that while "approval" is a frequent sentiment, it sometimes masks sarcasm and nuanced critique, with discussions spanning targeted humour and satire towards SNL and brands, double standards in global brands, challenges in supply chain transparency, labour rights and exploitation, environmental injustice, consumer complicity, and sustainable consumerism.