Natural Language Processing in Social Media Discourse Analysis: A Scoping Review

Abstract

This scoping review explores how Natural Language Processing (NLP) techniques are being used to analyse social media discourse, with a particular focus on which topics are being studied and how computational tools support research across disciplines. Drawing on 44 studies published over the past decade, the review highlights key trends in the types of online discourse being examined, from political and health communication to identity and environmental issues. It also maps the tools and methods researchers are using—such as sentiment analysis with LIWC, topic modelling with LDA, and more advanced approaches like BERT—to extract insights from digital conversations. This review emphasises how NLP techniques are facilitating interdisciplinary inquiry into linguistic, psychological and social phenomena. It also highlights key methodological challenges, including the limitations of lexicon-based tools in capturing nuance, the narrow focus on certain platforms, and inconsistencies in reporting analytic approaches. This review underscores the growing role of NLP in enabling richer, more systematic analysis of online discourse across diverse fields.

Keywords: scoping review, natural language processing, discourse analysis, Social Media, linguistics