Workshop: Introduction to Eye-Tracking for Language and Education Research

Dr. Elaine Schmidt, Cambridge University Press & Assessment

Senior Researcher

Elaine Schmidt is a Senior Researcher at Cambridge University Press and Assessment. Elaine’s research focusses on cognitive aspects of language processing and learning using eye-tracking and EEG. She obtained her PhD in Linguistics and Language Acquisition from the University of Cambridge. After her PhD she worked on cognitive processes of L1 and L2 speech perception in Sydney, Australia, before she moved back to the Linguistics Department at the University of Cambridge. A few years later Elaine then decided to combine her research with more practical applications and joined Cambridge Assessment where she brings her expertise in speech production and perception, eye-tracking and EEG in second language learning to an assessment context.

Talk: "Cognitive Processes in Eye-tracking and Implications for Assessment Research"

Abstract: Research in language assessment has largely concentrated on analysing output, i.e. learners’ writing or speech, or results of multiple-choice or written answers in listening and reading tasks. While discussions of cognitive validity have become more frequent over the last few years, actual experimental data looking at cognitive processing during language assessment are sparse. Eye tracking can be used to tap into underlying cognitive processes which questionnaires and other written or oral responses cannot capture since they are by necessity influenced by deliberation. However, the few studies which have used eye-tracking tools in assessment contexts often analysed the data in a qualitative way by looking at retrospective think-aloud reports from participants. Thus, despite using a tool to investigate top-down cognitive processing, many studies have focused on using eye tracking to analyse subjective deliberations and attention-driven bottom-up processes. This presentation provides an overview of eye-tracking methodology, highlighting key measures and the types of cognitive processing they can reveal. Drawing on examples from second language research, it demonstrates how eye-tracking can deepen our understanding of language processing during assessment. Finally, it explores the potential of this methodology for language testing, particularly in evaluating processing load and informing the design of more cognitively valid test materials.

Webpage: https://www.cambridgeenglish.org/english-research-group/meet-the-team/elaine-schmidt/

Andreas Säuberli, LMU Munich

PhD Student

Andreas Säuberli is a 2nd-year PhD student at MaiNLP research lab (LMU Munich). His research interests lie at the intersection of NLP and psycholinguistics, focusing on applications in the domain of second language education. His work explores the use of eye-tracking data to improve and evaluate technologies for language learning and assessment.

Talk: "Making Language Models Write Like Humans Read: Gaze-Guided Text Generation"

Abstract: Language models have become highly effective tools for generating grammatically fluent and semantically coherent texts. When they are used to generate educational material for language learning or assessment, controlling the readability or difficulty of the texts is essential. Most often, this is achieved by prompting or fine-tuning the model to produce easier or more difficult texts. In this talk, I will present a new approach that uses cognitive signals from gaze data to guide language models, enabling more cognitively informed and personalized text generation. Our eye-tracking study shows that the approach can successfully manipulate reading ease and perceived difficulty for both L1 and L2 readers. Finally, I will stress the importance of sharing eye-tracking data in an open and reusable way to enable the transfer of research insights to applications like in our study.

Webpage: https://saeub.github.io/

Hongyi Yang, University of Cambridge

PhD Student 

Hongyi Yang is a 2nd-year PhD student in Education at the Faculty of Education, University of Cambridge. His research interests include Language Teacher Education, Assessment Literacy, and AI-assisted Language Assessment. His work seeks to understand how language teachers exercise assessment in real classroom situations, and how AI can be meaningfully integrated into the teaching, learning, and assessment of languages. He is currently a Co-PI of the project "Bridging Human and AI Judgement in Writing Assessment: An Eye-Tracking Study", funded by the Cambridge Language Sciences Incubator Fund and Cambridge University Press & Assessment.

Talk: "Assessing with AI: How Raters Negotiate LLM Suggestions When Scoring Writing"

Abstract: As large language models (LLMs) increasingly supplement human scoring in writing assessment, important questions arise about how raters engage with AI-generated input. Do they defer to these suggestions, critically evaluate them, or disregard them altogether? Which analytic traits are more susceptible to AI influence, and how might such input reshape scoring cognition? This study investigates how experienced raters respond to LLM-suggested analytic scores during essay evaluation, offering insights into real-time human–AI interaction in writing assessment.

A mixed-methods design was adopted, integrating eye-tracking, and stimulated recall and qualitative interviews. Human raters were asked to rate two sets of 15 essays across two 2h-long sessions. Essays were drawn from the ELLIPSE corpus. In the first session, no AI-assistance was used. In the second rating session, rubric cells corresponding to the band scores suggested by a Large Language Model for a given essay were visually highlighted.   

These findings have practical implications for rater training and assessment policy in contexts where human–AI co-scoring may become increasingly common.

Webpage: www.linkedin.com/in/hongyiyangvictor

Prof. Andrea Revesz, University College London

Professor of Second Language Acquisition

Andrea Révész is Professor of Second Language Acquisition and Co‑Director of the Centre for Applied Linguistics. Her research sits at the intersection of second language acquisition, instruction, and assessment, with a particular focus on the neurocognitive processes underlying L2 performance and development. She is Editor of the Annual Review of Applied Linguistics and Co‑Editor of the John Benjamins Task‑Based Language Teaching series. She previously served as Associate Editor of Studies in Second Language Acquisition (2015–2024) and as Vice‑President, and, for one year, President, of the International Association for Task‑Based Language Teaching (2015–2023).

Talk: "Gaze‑contingency as a tool for developing second language knowledge"

Abstract: Eye-tracking methodology is becoming increasingly popular as a tool to capture cognitive processes. Through this technology, however, it is also possible to design systems that can interact with eye movements. This gaze-contingency affordance holds particular promise in fostering L2 acquisition during reading tasks. It is generally assumed that, to optimize L2 development, attention needs to be drawn to linguistic features. Some researchers also argued that this is best achieved during task-based performance in response to learner behaviours. In this presentation, I will describe a series of studies that have begun to explore the extent to which interactive, gaze-contingent eye-tracking can facilitate attention to and development in L2 knowledge in the context of reading. In particular, I will describe and discuss three studies focusing on vocabulary learning of different types of lexical features (single words, collocations), by adults and children, and learners with differential cognitive abilities. I will discuss the implications of the results emerging from these studies with respect to attentional models and optimal conditions for vocabulary learning through reading.

Webpage: https://profiles.ucl.ac.uk/48747-andrea-revesz