Exploring literacy comprehension and research shifts in undergraduate writing

Generative artificial intelligence (GenAI) is reshaping how students write and conduct research, yet there is little longitudinal, evidence-based research on how undergraduate writing has actually changed. This project investigates how student writing in first-year undergraduate essays has evolved between 2019 and 2025, seeking to determine whether observable shifts in literacy, comprehension, and research practices can be linked to the increasing availability of AI writing tools.

Using a longitudinal dataset of anonymised student essays, the study will assess trends in literacy, comprehension, and argumentation, identifying changes in literary practices over time. It will examine variations in research depth, citation practices, referencing accuracy, and the diversity of sources consulted. By applying data-driven methods, the project will explore patterns consistent with AI-assisted writing at the cohort level, without attributing individual misconduct or breaching confidentiality.

The research has three main objectives. First, to identify longitudinal trends in student writing by examining shifts in argumentation, structure, and comprehension. Second, to analyse transformations in research depth and quality, including referencing integrity, citation accuracy, and source diversity. Third, to explore whether patterns within student writing data suggest an increasing prevalence of AI-assisted authorship at a cohort level – without attributing individual misconduct.

The project will provide systematic analysis of six years of undergraduate writing, highlighting measurable changes in literacy, comprehension, and research depth. Findings will inform academic integrity policies and guide assessment redesign, helping the University and the sector adapt to the realities of AI-assisted writing. The research will generate new knowledge of immediate relevance to teachers, policymakers, and the wider higher education community.

MDAP's expertise is central to achieving these aims. They will design and implement text-analytic workflows suited to large sets of student essays, developing reproducible pipelines for data cleaning, anonymisation, and longitudinal analysis. MDAP will employ natural language processing and large language model-based techniques to identify changes in literacy, argumentation, and referencing patterns over time.

Their involvement ensures the project adheres to best practices in data governance, reproducibility, and ethical AI, particularly important when working with student data and emerging generative AI technologies. This collaborative research partnership expands methodological possibilities, enhances data integrity, and builds institutional capability in the responsible application of AI to higher education research.