Deception detection from written accounts

Abstract

Most research into deception detection in written accounts has been conducted on transcripts instead of written messages, and has focused on identifying valid verbal deception correlates instead of also examining untrained readers' spontaneous lie-detection attempts (accuracy rates, the cues they use, and so on). Also, the question of whether good liars are also good detectors has not been examined using written accounts. In Study 1, 78 participants handwrote a story and then judged the veracity of another participant's story. Accuracy was at chance level. Good liars were not better detectors than poor liars, but participants who thought they were good liars also thought they were good detectors. The higher the participants' fluidity scores on a standardized test, the poorer liars they were and the better liars they thought they were. The cues participants said they used were related to their judgments but unrelated to actual veracity. In Study 2, some Linguistic Inquiry and Word Count (LIWC) categories (with the Spanish-language dictionary) permitted a 68% classification rate of the written accounts of Study 1.

The Role of Account Length in Detecting Deception in Written and Orally Produced Autobiographical Accounts using Reality Monitoring

“…This suggests that the RM criteria may potentially outperform word count per se in predicting veracity. It can also be noted that, notwithstanding the finding that, other things being equal, truthful accounts tend to be shorter than deceptive ones (Zhou, Burgoon, Nunamaker, & Twitchell, ), a recent meta‐analysis of linguistic cues accessed by computer programs has questioned whether word count per se can generally be considered a reliable cue to deceptive behaviour (Hauch, Blandón‐Gitlin, Masip, & Sporer, ).This finding is in line with research using the Linguistic Inquiry Word Count (LIWC) computer software, which also shows that length per se is not a reliable cue to deception (Masip, Bethencourt, Lucas, Sánchez‐San Segundo, & Herrero, ; Williams, Talwar, Lindsay, Bala, & Lee, ). In other words, to predict veracity with any degree of accuracy, the variable of length needs to be considered in conjunction with other cues and with reference to the conditions under which the study has been conducted.…”

‘Lyin' Ted’, ‘Crooked Hillary’, and ‘Deceptive Donald’: Language of Lies in the 2016 US Presidential Debates

“…The relatively few LIWC‐based deception detection studies using the RM framework have been situated in disparate contexts. Whereas some studies found no support using LIWC for RM classification of message veracity (Ali & Levine, ; Masip, Bethencourt, Lucas, Sánchez‐San Segundo, & Herrero, ), others found otherwise (Bond & Lee, ). As Bond and Lee () indicated, the ‘jury is still out’ regarding the use of LIWC as an RM coding system, but more research is needed, including this research, to determine whether LIWC will capture differences between truth and deception (and debate statements) in the political arena.…”

Markers of Deception in Italian Speech

“…Evaluating the utility of other markers of lying, that can be measured independent of the judgment of a trained observer, is a worthwhile endeavor. To this end, computerized text analysis programs, such as Linguistic Inquiry Word Count (LIWC; Pennebaker et al, 2007) have been applied to the identification of deceptive text and transcribed verbal utterances in languages other than English, including Spanish, Dutch, Italian, and German (e.g., Schelleman-Offermans and Merckelbach, 2010; Fornaciari and Poesio, 2011; Almela et al, 2012; Hauch et al, 2012; Masip et al, 2012; Sporer, 2012). …”