Exploring Deception in Taiwanese Fake News by LIWC
Dr. Chien Huang
FU JEN Catholic University
Much research has been devoted on exploring the differences in linguistic style between truth and deception. It is assumed that deception stems from the creation of imaginary stories, and fabrication would display linguistic features that are different than when speaking the truth. This study uses TC-LIWC as a research tool and fake news as deceptive text. By conducting different research designs, this research explores the linguistic style differences between the deceptive and genuine text in Chinese.
This study is divided into two parts. The first part of the study explores the differences in linguistic style between deceptive and genuine text under experimental manipulation. The second part of the study discusses the differences between the linguistic style of deceptive and genuine text in a real situation.
The results show that there are differences in linguistic styles. Deception is reflected in the use of more function words, and is related to the input of cognitive resources, showing that deception goes through a more massive cognitive load. The deceptive text as a whole presents a linguistic style with negative emotions. The finding is similar to previous researches. We also established a regression model for predicting fake news with an accuracy over 60%, which is better than chance.
This study is divided into two parts. The first part of the study explores the differences in linguistic style between deceptive and genuine text under experimental manipulation. The second part of the study discusses the differences between the linguistic style of deceptive and genuine text in a real situation.
The results show that there are differences in linguistic styles. Deception is reflected in the use of more function words, and is related to the input of cognitive resources, showing that deception goes through a more massive cognitive load. The deceptive text as a whole presents a linguistic style with negative emotions. The finding is similar to previous researches. We also established a regression model for predicting fake news with an accuracy over 60%, which is better than chance.
