Dilbilimsel Otopsi
Burcu İlkay Karaman · The Bulletin of Legal Medicine · 2019
Objective: Different from authorship attribution, authorship profiling requires more elaborated investigations. This article deals with the authorship profiling criteria involving methods for the prediction of suspects' gender, age group, profession, native language, location, demography, personality traits, and mental health. Materials and Methods: Authorship profiling elements involve the prediction of gender, age group, profession, native language, location, demography, personality traits, mental health, educational level, ethnicity, cultural level, social status, religious belief, political interests, and cultural values of a suspect by a thorough investigation of the text s/he produced. For these elements, which we refer to as authorship profiling criteria, several methods have been developed involving the analysis of the author’s part of speech usage, the suspected text’s readability features, lexical choice, errors, and topic specific features. The aforementioned criteria are of universal value, i.e. can be applied to any language, however, may occasionally vary from language to language and from writing system to writing system, depending on language family and typology. Results and Discussion: The article specifically deals with the investigation of methods, such as the prediction of gender, age group, profession, native language, location, demography, personality traits, and mental health, concerning authorship profiling elements. For each element the accuracy rate of prediction varies 60-80%. In order to increase this rate, forensic linguists need to develop new methods, depending on the advances encountered in each subfield of linguistics. As it is considerably difficult to acquire original case material in Turkey, it is inevitable to do research with fictional data and corpus.