Using handwriting to infer a writer’s country of origin for forensic intelligence purposes

Forensic science has traditionally focused the majority of its resources and objectives towards addressing Court-related questions. However, this view restricts the contribution of forensic science to one process and results in a loss of information as the investigative and intelligence roles are largely neglected. A forensic science discipline suffering from this imbalance is handwriting examination, which may be characterised as a time consuming and subjective process that is mostly carried out towards the end of the investigation for the purpose of judicial proceedings. Individual and habitual characteristics are the major handwriting features exploited, however alternate information concerning the author’s native language could potentially be used as a key element in an intelligence framework.

This research focussed on the detection of characteristics that differentiate Vietnamese and English Australian writers based on their English handwriting. The study began with the extraction of handwriting characteristics from the writing of people from the two populations. The data was analysed using a logistic regression model and a classification and regression tree (CRT). Each recognised four class characteristics that were capable of distinguishing between the two nationalities. The logistic regression and CRT models were both capable of correctly predicting 93% of cases. Their predictive capabilities were then tested and supported using blind exemplars in order to mirror casework settings. It appeared that when using their respective class characteristics, the two models were capable of differentiating English Australians from Vietnamese in the data set. This proof of concept research demonstrated the plausibility of exploiting this additional information from a handwriting trace and taking advantage of it in an intelligence-led framework.