Unmasking Fake News: Lexical Patterns in COVID-19 Misinformation

Abstract

The surge of fake news on social media during the COVID-19 pandemic posed serious risks to public health and social stability. While previous studies have largely focused on computational detection of misinformation, less attention has been paid to identifying human-recognisable linguistic features that enable real-time fake news detection. This exploratory study investigated the lexical features that characterise fake COVID-19 news and identified the most distinctive features that differentiate fake news from authentic reporting. Using a qualitative research design, this study analysed 27 misinformation texts to examine a corpus of English-language fake COVID-19 news articles collected from an International Fact-Checking Network (IFCN)-certified fact-checking website: SEBENARNYA.MY (Malaysia). The findings may suggest preliminary evidence of linguistic patterns of recurring lexical features in fake news, including random capitalisation (capitalised words and all-uppercase text), irregular use of lowercase letters, repetition of content words and phrases, multiple types of spelling errors, informal shortened words, inappropriate lexical choices, and non-standard formatting such as excessive bold text and improper word spacing. The results indicate that fake news shows orthographic instability, reduced lexical diversity, and register mismatch, reflecting deceptive intent and cognitive constraints associated with fabricated narratives. This study contributes to fake news research by foregrounding interpretable, surface-level lexical cues that enable readers to identify misinformation without reliance on automated tools. The findings have important implications for media literacy, public health communication, and linguistic approaches to detect misinformation.