Integrating Linguistic Character Features in Handwriting Analysis for Ransom Note Author Identification
摘要
Authenticity of authorship in the context of forensic cases, especially the ransom note case is a complicated problem with intentional disguise, small sample size as well as inconsistency in writing behaviour individuality. The previous studies indicate that handwriting analysis and forensic linguistics are used separately, rather than as a combined technique. These studies emphasize integrating graphical and linguistic features for more reliable and comprehensive authorship attribution. The proposed study seeks to consider the efficacy of using a combination of forensic linguistic attributes and handwriting features in analysing authorship. A total of 150 samples from 50 individuals were collected, 3 samples from each person (A, B and C), among which dataset A (50 samples) were anonymous and B and C (50 samples each dataset) were from known individuals. The samples were analysed using conventional methods for handwriting and linguistic features. To further validate the results, Pearson correlation and nearest neighbour analysis were used. Among 8 characteristics (5 linguistic and 3 handwriting), vocabulary showed 80–85% writer consistency using conventional analysis. Additionally, the results for statistical validation established that word count (80%), spacing between words (75%), and spacing between lines (70%) were most reliable characteristics for authorship attribution. Hence, the present study shows that a combined approach of forensic linguistics and handwriting can augment the attribution of the authorship in 75% samples. Therefore, it can be concluded that instead of using individual techniques, combining the handwriting and linguistics characteristics for authorship attribution produce more reliable, comprehensive and accurate results.