Style markers in authorship studies

Abstract

The last issue of Forensic Linguistics presented two articles on stylemarker selection in studies of questioned authorship. The first paper (Chaski 2001) represents an interesting approach to marker selection, but two significant weaknesses detract from its purpose: it rejects virtually all previous work in stylistics, hundreds studies representing more than a century of work, as unscientific and irrelevant present forensic needs; and it is founded on a theoretical position that views linguistic variation as a feature of linguistic performance, thus missing the point of the inherent variability of language. The second paper (Grant and Baker 2001) hits the mark in three significant ways: it is a good review of the style-marker issue; it recognizes that authorship attribution must be based on an aggregate array of markers; and it describes the statistical and linguistic bases of Principal Component Analysis, a promising method for measuring the collective range of variation needed for authorship identification.