Automated insights: verbal cues to deception in real-life high-stakes lies

Abstract

This study differentiated between the language of deceptive and genuine pleaders who were pleading for the return of a missing loved one during a televised press conference. The Wmatrix linguistic analysis tool was used to examine the language of 78 pleaders. Approximately half (n = 35) of these individuals were deceptive and were responsible for the disappearance. Transcripts of the pleas were analyzed for various linguistic cues, and a separate analysis was conducted across gender. Results revealed that deceptive pleaders used the word ‘they,’ singular indefinite pronouns (e.g., ‘anybody,’ ‘somebody’), and exclusivizers/particularizers (e.g., ‘just’) significantly more than genuine pleaders, while genuine pleaders used more temporal words (e.g., ‘days,’ ‘weeks’), and the word ‘we’ more frequently in their pleas. Specific gender differences were also revealed across credible and deceptive pleaders. Our analysis of linguistic differences across pleader veracity provides an enhanced understanding of the verbal elements of high-stakes deception and what differentiates truths from deceptions in high-stakes cases. It also provides further validation of the use of automated linguistic tools like the Wmatrix in forensic contexts.

Keywords:

• deception
• language
• high-stakes lies
• missing persons
• automated linguistic analysis

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. A breakdown of the number of words in both the deceptive and genuine pleader corpora across whole plea and direct plea and gender are included in Table 1.

2. A careful analysis of the Wmatrix output revealed two word misclassifications. First, the Wmatrix classified the word ‘appeal’ into the physical attributes category when its intended meaning did not fit that category. For example, this word was used in the following context, ‘I would appeal someone to come forward.’ This word was used as frequently in the guilty corpus as it was in the innocent corpus. The second misclassification involved the word ‘just.’ For example, this word was used by a deceptive pleader in the following statement, ‘it’s just unbearable to think about.’ The word was incorrectly tagged as a personality trait. This word was used twice in the genuine pleader corpora and once in the deceptive pleader corpora.