Multi-Language Statistical Classification of Natural and Computer-Generated Texts Kingston University in United Kingdom

This PhD project at Kingston University investigates the statistical properties of written documents, drawing analogies to other type-token systems such as city populations and personal incomes. Large-scale studies on the Project Gutenberg corpus have revealed significant statistical variations between texts, with some documents fitting established models and others presenting anomalies that merit deeper exploration. The research aims to analyze differences between languages—such as Finnish and English—using statistical methods, with the potential to classify languages based solely on their statistical characteristics.

A key aspect of the project is the comparison between AI-generated and natural texts across multiple languages. The study will compile a large, balanced corpus representing significant language groups, identify and separate statistical outliers, and generate a corresponding set of AI-generated texts. Each corpus will be characterized using existing type-token models, and these models will be ranked by their effectiveness in distinguishing between languages and between AI and natural texts. The outcomes could lead to the development of a robust language classifier or inform the creation of improved statistical models for document analysis.

The project is supervised by Dr M Tunnicliffe and is part of the Graduate School studentships competition for October 2026 entry. Funding may be available, covering tuition and stipend, subject to the outcome of the competition. Applicants should have a strong background in mathematics, statistics, computer science, or linguistics, with excellent quantitative and analytical skills. Experience in statistical modeling, programming, or corpus linguistics is advantageous. English language proficiency must meet Kingston University standards.

To apply, candidates should consult the Graduate School Studentships information and the Faculty of Engineering, Computing and the Environment research page at Kingston University. The application deadline is March 4, 2026. This opportunity is ideal for students interested in statistical modeling, natural language processing, and the intersection of AI and linguistics.