부산대학교 도서관

The paper presents a two-part forensic linguistic analysis of an historic collection of abuse letters, sent to individuals in the public eye and individuals' private homes between 2007 and 2009. We employ the technique of structural topic modelling (stm) to identify distinctions in the core topics of the letters, gauging the value of this relatively under-used methodology in forensic linguistics. Four key topics were identified in the letters, 'Politics A' and 'B', 'Healthcare' and 'Immigration', and their coherence, correlation and shifts in topic were evaluated. Following the stm, a qualitative corpus linguistic analysis was undertaken, coding concordance lines according to topic, with the reliability between coders tested. This coding demonstrated that various connected statements within the same topic tend to gain or lose prevalence over time, and ultimately confirmed the consistency of content within the four topics identified through stm throughout the letter series. The discussion and conclusions to the paper reflect on the findings and also consider the utility of these methodologies for linguistics and forensic linguistics in particular. The study demonstrates real value in revisiting a forensic linguistic dataset such as this to test and develop methodologies for the field. [ABSTRACT FROM AUTHOR]