Abstract
The analysis of written language behaviour has attracted attention from many areas of science. Recent years have seen the development of forensic linguistics, a discipline concerned with the examination of the units of language, the results of which have been applied to criminal investigations. Such applications have been concerned largely with the attribution of authorship to questioned text (McMenamin, 1993). The underlying assumption of such studies has been that individuals possess a unique and consistent writing style. However it appears problematic to establish a reliable set of linguistic characteristics of writing style that will enable individual differences and consistencies within written language behaviour to be ascertained. This research has enabled these issues to be explored and using anonymous threat letters has addressed theories of how personal characteristics may contribute to an individual’s writing style. This research examined written language behaviour using a multivariate approach to the analysis of style. It was hypothesized that individual differences in written language could be identified employing such an approach. Smallest Space Analysis was first used to identify if an underlying structure existed between fifteen linguistic variables (Study 1). Two facet elements representing an author’s lexical repertoire (Content) and their syntactic constructions (Structure) were identified. These were submitted to a series of Partial Order Scalogram Analyses (Studies 2 and 3), which demonstrated that broad individual consistencies of written language behaviour existed. It was then possible to review both the Content and Structure of anonymous forms of written threat to ascertain if the measures possessed a practical application within the forensic setting. The context sensitivity of the measures was demonstrated (Study 4) and their ability to discriminate between genuine and simulated threat letters was also evident (Study 5). However it was not possible to relate these measures to personality aspects of the authors (Study 6). The conclusion of this research was that using a multivariate analysis of language, unique individual differences could not be reliably identified (despite the fact that individual consistencies can be identified). However the analysis of written threat, employing the two facet elements identified by the Smallest Space Analysis, offered some promising directions for future research.