Evidence for the Pinocchio Effect: Linguistic Differences Between Lies, Deception by Omissions, and Truths

Abstract

The study used Linguistic Inquiry and Word Count and Coh-Metrix software to examine linguistic differences with deception in an ultimatum game. In the game, the Allocator was given an amount of money to divide with the Receiver. The Receiver did not know the precise amount the Allocator had to divide, and the Allocator could use deception. Allocators either lied, deceived through omission, or were truthful with the Receiver with their allocation decision. Liars used a higher percentage of third-person pronouns, numbers, and profanity than other participants. Participants using deception by omission used fewer words and a lower percentage of causation words than other participants. Support was found for the “Pinocchio effect”: liars generally used more words than other participants, although truthful Allocators with suspicious partners did not significantly differ from liars on word count. Allocators with suspicious partners used more negative emotion words and profanity. Mixed support was found for indexes from Coh-Metrix that measured whether liars had a higher cognitive load. Results are discussed in terms of strategic and non-strategic linguistic cues.

Notes

¹This method is the same method as Van Swol, Malhotra, and Braun (2011).

²There is a desktop version of Coh-Metrix with over 700 linguistic indexes. However, this version is proprietary software that is only available to certain individuals covered under an active grant. We were unable to obtain the desktop version for this article.

³Only two Allocators offered money less than the default amount that the Recipient could receive.

⁴For the stakes variable, there was no difference between the $5 or $30 condition in each type of offer: no deception, χ²(1, N = 102) = 0.01, p = .94; omission, χ²(1, N = 102) = 0.33, p = .57; and lies, χ²(1, N = 102) = 1.26, p = .26. For friends and strangers, there was a difference between whether participants were interacting with a friend or stranger for no deception and omission: no deception, χ²(1, N = 102) = 13.96, p < .0001; omission, χ²(1, N = 102) = 10.87, p < .001; and lies, χ²(1, N = 102) = 1.51, p = .22. Therefore, for each linguistic cue, we checked to see if friend/stranger interacted with type of offer. For first-person pronouns, there was no difference between friends and strangers, F(1, 96) = 0.55, p = .46 (η² = 0.01); nor interaction with type of offer, F(2, 96) = 0.12, p = .89 (η² = 0.00). For third-person pronouns, there was no difference between friends and strangers, F(1, 96) = 1.06, p = .31 (η² = 0.01); nor an interaction with type of offer, F(2, 96) = 0.93, p = .40 (η² = 0.02). For negative emotion words, there was no difference between friends and strangers, F(1, 96) = 0.84, p = .36 (η² = 0.01); but a significant interaction with type of offer, F(2, 96) = 8.27, p < .0001 (η² = 0.15). For profanity, there was a difference between friends and strangers, F(1, 96) = 19.82, p < .0001 (η² = 0.17); and a significant interaction with type of offer, F(2, 96) = 14.43, p < .0001 (η² = 0.23). For negation words, there was no difference between friends and strangers, F(1, 96) = 0.04, p = .83 (η² = 0.00); nor interaction with type of offer, F(2, 96) = 0.97, p = .39 (η² = 0.02). For ratings of concreteness, there was no difference between friends and strangers, F(1, 96) = 0.39, p = .54 (η² = 0.00); nor interaction with type of offer, F(2, 96) = 0.43, p = .65 (η² = 0.01). For mean number of words before the main verb of the main clause, there was no difference between friends and strangers, F(1, 96) = 0.00, p = .96 (η² = 0.00); nor interaction with type of offer, F(2, 96) = 0.85, p = .43 (η² = 0.02). For type-token ratio, there was a significant difference between friends and strangers, F(1, 96) = 4.64, p = .03 (η² = 0.05); but not a significant interaction with type of offer, F(2, 96) = 2.06, p = .13 (η² = 0.04). For connectives, there was no difference between friends and strangers, F(1, 96) = 0.28, p = .60 (η² = 0.00); nor interaction with type of offer, F(2, 96) = 0.76, p = .47 (η² = 0.02). For causation words, there was no difference between friends and strangers, F(1, 96) = 1.90, p = .17 (η² = 0.02); nor interaction with type of offer, F(2, 96) = 0.12, p = .89 (η² = 0.00). For money words, there was no difference between friends and strangers, F(1, 96) = 1.63, p = .21 (η² = 0.02); nor interaction with type of offer, F(2, 96) = 0.77, p = .47 (η² = 0.02). For word count, there was no difference between friends and strangers, F(1, 96) = 0.01, p = .92 (η² = 0.00); nor interaction with type of offer, F(2, 96) = 0.66, p = .52 (η² = 0.01).

^a n = 7.

^b n = 26.

^c n = 69.

^dHigher numbers indicate more concreteness.

*p < .05.

**p < .01.

^a n = 7.

^b n = 26.

^c n = 69.

^a n = 7.

^b n = 26.

^c n = 69.

*p < .05.

**p < .01.

*p < .05.

**p < .01.