Beyond Fact-Checking: Lexical Patterns as Lie Detectors in Donald Trump’s Tweets

Abstract

Journalists often debate whether to call Donald Trump’s falsehoods “lies” or stop short of implying intent. This article proposes an empirical tool to supplement traditional fact-checking methods and address the practical challenge of identifying lies. Analyzing Trump’s tweets with a regression function designed to predict true and false claims based on their language and composition, it finds significant evidence of intent underlying most of Trump’s false claims, and makes the case for calling them lies when that outcome agrees with the results of traditional fact-checking procedures.