Are You Lying Now? A Linguistic Examination of Deceptive Utterances in Online Conversation

Abstract

Extensive research has been done to identify linguistic cues to deception, especially in the rapidly growing field of computer-mediated communication. However, most past research contains an important methodological flaw: the failure to break down deceptive and truthful topics into individual utterances. When assigning research participants to a deceptive role (say, asking them to lie to an unknowing receiver) previous studies have generally given the deceivers a topic to lie about and then asked them to go ahead and communicate with their partner. The deceivers were assumed to be lying whenever they were talking about the topic on which they were supposed to deceive their partners. However, in practice not all of the utterances within a deceptive conversation topic are lies. Some are truths used to support the overall lie. Past research has failed to make this distinction, drawing into question previous findings on linguistic cues. This study sought to validate four of the more well-known linguistic cues to deception by examining them at the utterance level. The results reveal that while there often is a distinction between the linguistic cues at the overall topical level (deceptive topics vs. truthful topics), those markers do not always hold true at the utterance level. Even more interestingly, there is often a large difference between truths told in support of an overall deceptive topic and truths told in support of an overall truthful topic. These findings open up new areas of research into the linguistic cues to deception.