An Editorial Note for LESLI 1:1
Linguistic Evidence In Security, Law and Intelligence (LESLI)
Accurate and reliable deception detection is a Holy Grail to all kinds of professions that require interviewing or surveillance: law enforcement, private investigation, executive and corporate security, national and business intelligence, clinical psychology and human resources management. In this issue, four research articles tackle deception detection, with reporting of empirical research results. Amela, Valencia-Garcia and Cantos' "Seeing through Deception: A Computational Approach to Deceit
... Detection in Spanish Written Communication" applies a text-analytic approach to deception detection based on Pennebaker's LIWC (Linguistic Inquiry and Word Count) software to a new dataset, Spanish texts. Amela, Valencia-Garcia and Cantos also provide a good overview of this approach that readers unfamiliar with the word-based approach will appreciate. Interestingly, another popular approach to text analysis, known as bag-of-words, did not perform as well as the LIWC approach, suggesting that certain word categories are important to deception detection in Spanish, as it has been shown in English. Replication studies are essential for scientific progress: Amela, Valencia-Garcia and Cantos' article is both replication and a nice contrast between the two competing approaches. Skillicorn and Lamb, who also provide a clear overview of the LIWC-based approach to deception detection, extend this approach in "Extending Textual Models of Deception to Interrogation Settings." Since Skillicorn and Lamb use datasets comprised of question-answer dialog, each speaker's linguistic strategy is determined in response to another person's language, automatically affecting word usage and kinds of responses. Skillicorn and Lamb take linguistic analysis beyond the word level: language is, after all, far more than a list (or bag) of words, demonstrating an interesting link between word choice and conversational structure. Their research shows how the LIWC-based approach can be improved, and they offer empirically-based advice to interrogators on question-formation.