Seminar I Interpretable Cognitive Linguistic Authorship Analysis Using the Likelihood Ratio Framework
Seminars / Lectures / Workshops
-
Date
12 Feb 2025
-
Organiser
Department of English and Communication
-
Time
17:00 - 18:00
-
Venue
Online via Zoom
Speaker
Dr Andrea Nini
Summary
The talk will introduce the audience to an algorithm for authorship analysis called LambdaG (Nini et al. 2024), a method compatible with Nini’s (2023) Theory of Linguistic Individuality based on Cognitive Linguistics and Cognitive Psychology. This method, which is fundamentally based on calculating a likelihood ratio, exploits early language modelling techniques from the 90s. However, despite the seemingly outdated technology, this algorithm outperforms modern transformer large language models, as well as other traditional machine learning algorithms employed for authorship analysis. The talk will propose that the increase in performance can be explained by the fact that the method is more compatible with the theoretical motivations for individuality in language. The talk will also introduce the audience to a free and open source R package called idiolect (Nini 2024), which can be used to run this analysis and to visualise the results.
References
Nini, Andrea. 2023. A Theory of Linguistic Individuality for Authorship Analysis (Elements in Forensic Linguistics). Cambridge, UK: Cambridge University Press.
Nini, Andrea. 2024. Idiolect: An R package for forensic authorship analysis. https://andreanini.github.io/idiolect/.
Nini, Andrea, Oren Halvani, Lukas Graner, Valerio Gherardi & Shunichi Ishihara. 2024. Authorship Verification based on the Likelihood Ratio of Grammar Models. arXiv. https://doi.org/10.48550/arXiv.2403.08462.
Keynote Speaker
Dr Andrea Nini
University of Manchester, United Kingdom
Andrea Nini is a Senior Lecturer (Associate Professor) in Linguistics and English language at the University of Manchester, UK, where he teaches courses in forensic linguistics, stylistics, and statistics for linguistics. His specialisation is the analysis of linguistic data using computational methods, more specifically for the study of linguistic individuality and authorship analysis. He has published on authorship attribution of historical texts such as the Bixby letter and the Jack the Ripper letters and his PhD thesis was on the authorship profiling of malicious texts in forensic contexts. Andrea has also carried out research and published in other areas of linguistics, such as computational sociolinguistics and register variation. In addition to his academic work, Andrea offers consultancy as a forensic linguist to law firms and law enforcement agencies, predominantly in criminal cases involving documents of disputed authorship.
