Summary: Researchers used eye-tracking technology to evaluate how well people learn English as a second language by analyzing their eye movements while reading.
Source: MIT.
MIT study shows eye movement patterns reveal English language proficiency
Researchers at MIT have demonstrated a novel method for assessing English language proficiency by analyzing eye movements during reading. Using camera-based eye-tracking data, the team found that specific patterns—especially how long readers fixate on individual words—correlate strongly with standardized measures of English ability.
“To a large extent, eye movement captures linguistic proficiency, as we can measure it against benchmarks from standardized tests,” says Yevgeni Berzak, a postdoctoral researcher in MIT’s Department of Brain and Cognitive Sciences (BCS) and co-author of the paper describing the research. “The signal produced by eye movement during reading is rich and highly informative.”
Co-author Roger Levy, an associate professor in BCS, adds that the approach has practical potential as an assessment tool: “It has real potential applications.” The study, titled “Assessing Language Proficiency from Eye Movements in Reading,” appears in the Proceedings of the 16th Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. The authors include Berzak; Boris Katz, principal research scientist and head of the InfoLab Group at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL); and Levy, who directs the Computational Psycholinguistics Lab in BCS.
The illusion of continuous reading
The study examines an often-unnoticed aspect of reading: our eyes do not sweep smoothly across text. Instead, readers make brief fixations on particular words—typically lasting 200 to 250 milliseconds—interspersed with rapid saccadic movements that jump from one word to another in roughly 1/20 of a second. Although readers perceive a seamless flow, the eyes actually move in discrete steps, forward and sometimes backward, while the brain constructs the impression of continuity.
When someone is learning a new language, fixations on certain words tend to be longer as the reader works to understand meaning. These fixation patterns, when analyzed in context, can therefore reveal important information about comprehension and overall language skill.
The research team analyzed a dataset of eye-tracking records collected by Berzak. The dataset included 145 learners of English as a second language, with native languages evenly represented among Chinese, Japanese, Portuguese, and Spanish speakers, and it also included 37 native English readers. Participants read 156 sentences; half of those sentences formed a fixed test set that every reader encountered. Video recordings of the eyes enabled precise measurement of fixation durations on specific words.
From these metrics the team derived what they call an “EyeScore,” a composite set of features based on eye-movement behavior. When they compared EyeScore results to established standardized exams such as the Michigan English Test (MET) and the Test of English as a Foreign Language (TOEFL), the researchers found that EyeScore produced competitive results. In their paper they conclude that the approach “further strengthens the evidence for the ability of our approach to capture language proficiency.”
The authors describe the work as the first proof of concept for a system that uses eye tracking to measure linguistic ability in reading, suggesting a new direction for language assessment that relies on unobtrusive behavioral signals rather than only on written or spoken tests.
Sentence-level assessment and future directions
Experts in the field responded positively to the study. Erik Reichle, head of the Department of Psychology at Macquarie University and an experienced researcher in eye-tracking experiments, called the method “very innovative” and noted its potential to expand the use of eye-tracking technology in second-language learning research and related fields.
The MIT researchers emphasize that this study represents an initial step in a broader investigation into the links between language and cognitive processes. Katz highlights the larger scientific question: how language learning and use reshape the brain. Given the relatively recent human history of reading—only a few thousand years—our capacity to process written language demonstrates considerable neural plasticity. Continued work with eye-tracking data may help answer deeper questions about how the brain supports language and reading.
Levy suggests a practical avenue for refinement: making eye-based assessments more granular. Instead of evaluating comprehension across a set of 156 sentences, future studies might determine how reliably eye movements can indicate understanding on a sentence-by-sentence basis. “One thing we hope to do in the future is ask, for each sentence, to what extent we can tell how well it was understood from the eye movements made during reading,” he says. That question remains open and will be a focus of further research.
Funding and publication details
Funding: The study received support in part from MIT’s Center for Brains, Minds, and Machines through a National Science Foundation grant.
Source: Peter Dizikes – MIT
Publisher: Organized by NeuroscienceNews.com.
Image credit: Christine Daniloff/MIT.
Original research presented at the Proceedings of the 16th Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies.
How to cite this article
MLA: MIT. “Gauging Language Proficiency Through Eye Movement.” NeuroscienceNews, 23 May 2018.
APA: MIT (2018, May 23). Gauging Language Proficiency Through Eye Movement. NeuroscienceNews. Retrieved May 23, 2018.
Chicago: MIT. “Gauging Language Proficiency Through Eye Movement.” NeuroscienceNews, accessed May 23, 2018.