GenAIsis
The Generative AI-assisted L2 Writing Corpus: a collaborative project between Radboud University, The Netherlands, and the University of Antwerp, Belgium.
GenAIsis is a corpus of writing by university-level learners of English designed to enable systematic comparisons between unaided and GenAI-assisted writing through a fully replicable methodology. The data for this corpus were collected at the University of Antwerp (BE) in 2026.
Currently, the corpus contains product data comprising approximately 70,000 tokens produced by 81 different learners, as well as process data that allows us to reconstruct their writing processes. Metadata include LexTalproficiency scores for vocabulary knowledge assessment, questionnaire data about Generative AI use, and, for a subset of participants, screen recordings of the writing process.
A second round of collection at the University of Antwerp is planned in Spring 2027.
The tasks:
Each learner completed two tasks (counter-balanced across GenAI and non GenAI conditions):
- A persuasive essay on the topic of English Medium Instruction (EMI) in Flanders (source material provided)
- A narrative essay about one's experience with language
The data at a glance:
81 Learners |
161 Texts |
71,198 Tokens |
87.2 Average LexTale score |
B2-C1 Most frequent self-assessed proficiency level |
133 Inputlog recordings |
20 OBS Screen recordings |
4.78 Average number of interactions with GenAI |
19.69 Average age of participants |
More information about the project can be found in the following conference presentation:
Thewissen, J., Zaffinelli, A., Guo, S., Jadoulle, P., & Vandeweerd, N. (2026, september 19). Introducing a new GenAI-assisted L2 writing corpus: A replicable methodology for collecting L2 writing product and process data Learner Corpus Research Conference 2026, Prague, Czechia.
For citation details, please refer to the section 'How to cite'.
GenAIsis team:
Spokespeople
Other team members
How to cite:
The dataset and associated manuscript are currently in preparation. We will provide the final citation information once they are formally released. In the meantime, if you would like to cite this work, please contact the following team members:
Jennifer Thewissen - lead dataset curator
jennifer.thewissen@uantwerpen.be
Anna Zaffinelli - lead author
anna.zaffinelli@ru.nl
This section will be updated regularly as the manuscript and dataset progress toward formal release.
Updates:
- Corpus presented at LCR conference on September 19th 2026. Slides of LCR2026_GENAISIS
- GenAIsis website launched 17 September 2026
- Corpus presented at LANSPAN research group meeting on May 11th, 2026 Slides of LANSPAN2026
- First round of data collection completed in Antwerp February 26th, 2026
- Project presented at LeLaN research group in Antwerp February 6th, 2026 Slides of LELAN2026
Related Projects:
- AI Authenticity Project: Exploring What Makes L2 Writing Sound Human
- ESL writing in a changing landscape: Examining the impact of GenAI assistance and task type on L2 writing quality.
- Keeping the Human in the Text: Learner Voice and GenAI
- Learner voice in the age of GenAI
Get in touch:
We hope to make the data available in Spring 2027. If you would like to stay up-to-date, please sign up using the form below. You will receive occasional emails regarding major project outcomes, including the corpus release and related publication.
If you no longer wish to receive these emails, please contact us at genaisis.corpus@ru.nl
Funding:
This project was funded by a Starters Grant from the Centre for Language Studies (Radboud University) and a Bijzonder Onderzoeksfonds (BOF) grant from the University of Antwerp.
Ethical approval:
The study received ethical approval from the Ethics Committee for the Social Sciences and Humanities (EASHW) of University of Antwerp (SHW_2025_199).
 |
 |