Source evaluation

In jurisdictions across the world, different actors evaluate oral statements on a daily basis. For example, oral statements are of great significance in so-called “word against word situations” in cases of suspected rape, and in international jurisdictions statements from insider witnesses are essential for establishing who gave orders to attack a civilian population as part of a crime against humanity. Both during investigate and trial phases, methods are used to analyse whether the statements are reliable, despite known difficulties with distinguishing deceptive, truthful and mistaken testimony.

The Source Evaluation project entails two parts, one part focusing on statements from adults and one part focusing on statements from children and adolescents.

For adult statements, we will examine whether automated language analysis or human evaluators are better at deciding whether someone is telling the truth or lying. This entails Natural Language Processing (NLP) and scenario-based experimens with, for example, police officers, prosecutors and judges. Apart from the comparison, this enables an evaluation of what criteria, if any, really distinguish truthful and deceptive statements, as well as what criteria are used by human evaluators in practice.

Statements from children and adolescents are also commonly occurring in criminal cases, and Courts need to decide whether the statements can be trusted and form the basis for verdicts. To understand how this process happens we will examine real cases including both police interviews and Court assessments. Additionally, we will conduct interviews with police officers and investigate what question types can impact on the content of a statement. Because assessments of statements are brought to its head in cases regarding, for example, rape, we will also examine statements specifically in relation to the new crime “negligent rape” (in Swedish criminal law).