Linguistic Cues to Deception Assessed by Computer Programs: A Meta-Analysis

Research syntheses suggest that verbal cues are more diagnostic of deception than other cues. Recently, to avoid human judgmental biases, researchers have sought to find faster and more reliable methods to perform automatic content analyses of statements. However, diversity of methods and inconsistent findings do not present a clear picture of effectiveness. We integrate and statistically synthesize this literature. Our meta-analyses revealed small, but significant effect-sizes on some linguistic categories. Liars use fewer exclusive words, selfand other-references, fewer time-related, but more space-related, negative and positive emotion words, and more motion verbs or negations than truth-tellers.