International Linguistic Association
Co-Sponsored by the
Department of English
John Jay College of Criminal Justice

Eileen Fitzpatrick
Montclair State University

Detecting Deception in Civil and Criminal Narratives*

The talk will consider the reliability of using linguistic cues to identify deceptive and non-deceptive regions in "real world" narratives — criminal statements, police interrogations and legal testimony. To test the accuracy of these cues in predicting deception, we tagged the narratives for 12 language-based deception cues as well as for the truth value of all propositions that could be externally verified as true or false. A measure of the density of cues was then calculated, with high cue density taken to identify a passage as deceptive. This method was 74.9% accurate in predicting True/False on the externally verified propositions, as compared to the baselines that range from 50-57%. This preliminary result suggests that linguistic cues can provide a reasonable guide to the sectioning of narratives into deceptive and non-deceptive statements.

*Joint work with the LinguisTech Consortium, Oxford NJ.

Saturday, February 9, 2013
11 AM - 12 PM
John Jay College of Criminal Justice
Room: L2.82
524 West 59th Street (between 10th and 11th Avenues)
New York, NY 10019

Contact: Dr. Effie Cochran,
ecochran@jjay.cuny.edu
www.ilaword.org