The Fundamentals of the Zeta Test for Forensic Authorship Attribution Studies

Enfoque

Forensic Linguistics is a modern branch of Applied Linguistics that establishes a connection between Linguistics and the Law. In other words, it focuses on all those legal cases and issues that are related to the use of language to some extent. There are three main areas of study in Forensic Linguistics, these being the written language of the law, which is mainly concerned with the adaptation of legal documents to make them more accessible to the average citizen, the spoken language of the law, which analyses every oral interaction involved in a legal process, from the moment of arrest until the trial stage, and the linguist as expert witness. The latter focuses on the evidence and testimony that linguists can provide to determine the course of a legal dispute. Among the many applications of this area of study, authorship attribution study is probably the main one.

Authorship attribution studies are built upon the idea that every native speaker has an individual version of their language, which is their idiolect, and that idiolectal choices are likely to arise in any given written or spoken production. One of the most reliable authorship attribution methods in the field is the Zeta test, which allows for the comparison between a disputed text and two reference corpora in terms of the frequency and use of lexical words (i.e. names, adjectives, verbs and adverbs), which are much more distinctive than function words.

The present paper intends to present the fundamentals of the Zeta test for forensic authorship attribution studies and the manner in which this can be applied using the software ALTXA for text analysis, which has been specifically designed to facilitate the implementation of Forensic Linguistics in educational contexts as a result of its intuitive interface and the wide range of functionalities it offers. The validity of the tool will be proved with its application in a case study involving Elizabethan texts of known authorship that will be treated as anonymous.