Grammatical patterns in language use can distinguish between real news and fake news stories – potentially helping those looking to combat disinformation, a new study reveals.
A detailed linguistic analysis of the work of former New York Times journalist Jayson Blair, who left the newspaper after being found to have plagiarised and fabricated some of his stories, has established that there can be clear stylistic differences in the language of real and fake news.
Published by Cambridge University Press as a book titled, The Language of Fake News, the University of Birmingham study by Professor of Corpus Linguistics, Jack Grieve, and PhD researcher Helena Woodfield, reveals that Blair wrote his fake stories in a less informationally dense style, and with less conviction, compared to his real news articles.
Professor Grieve said: “We conducted a detailed analysis of the grammatical features used by Blair in 64 of his articles, 36 of which were fake news, and discovered 28 important differences in the style of his real and fake articles.
“There are clear stylistic differences between the articles in which Blair told the truth and those in which he lied. He made consistent linguistics choices in his fake news resulting in grammatical patterns that distinguish his fake news from his real news.”
In particular, the study finds that Blair’s fake news was less informationally dense: when Blair lied, he tended to write less concisely, packing less information into his articles. The study also found that he wrote with less conviction, adopting a less confident tone, and being less specific about where he obtained the information he was reporting.
The public has perhaps never been more distrustful of the news as we are right now, but we hope that knowing of the existence of these types of linguistic patterns will help people spot lies when reading the news and encourage further research on the language of fake news.Professor Jack Grieve, University of Birmingham
Fake news is a pervasive problem - identifying and fighting it tops the agendas of many social media companies, news outlets, and governments.
Professor Grieve continued: “The stylistic signals we have identified in Blair’s fake news reporting can only be applied directly to Blair, but this study demonstrates that there are linguistic features which can help identify fake news. In the future, we hope to replicate this study on a larger scale so that we can obtain a more general picture of the language of fake news.
“Our research shows that fake news can be associated with a clear and meaningful linguistic signal. This is potentially good news for those looking to identify and remove false stories, and combat disinformation, particularly online.”
While these language patterns are complex and potentially difficult for members of the public to spot while reading their morning news, the study provides a basis for understanding some of the rhetorical strategies which could be present in fake news.
Professor Grieve concludes: “The public has perhaps never been more distrustful of the news as we are right now, but we hope that knowing of the existence of these types of linguistic patterns will help people spot lies when reading the news and encourage further research on the language of fake news.”