Satoshi Tomioka
University of Delaware, DE·Social Science
Languages often let speakers say the same thing in more than one word order, yet those orders are not equally easy to understand. Reading a sentence where a phrase has been shifted out of its usual slot typically costs the reader extra effort, and the question is where that cost comes from.
- Fields
Social Science
Psychology
- Topics
psycholinguistics
sentence processing
word order
- Grants
- 1 known grant, through Jan 2027
What they work on
- Test whether the difficulty of alternative word orders comes from filler-gap dependency formation, the effort of linking a displaced phrase back to the position it moved from, using self-paced reading across several sentence types.
- Test whether that difficulty instead reflects prediction mismatch, measuring what readers expect to come next with sentence completion tasks and comparing those expectations against reading-time slowdowns.
- Separate the contributions of dependency formation and prediction mismatch by combining online and offline measures in high-powered samples with rigorous statistical analysis.
- Release code for experimental task design and data analysis as reusable infrastructure for future sentence-processing research.
More about the research
This doctoral dissertation project separates two candidate sources. The first is filler-gap dependency formation: when a phrase appears earlier than expected, the reader must hold it in mind and later link it to the empty position it came from. The second is prediction mismatch: as people read, they continuously anticipate what word or phrase comes next, and an unexpected order forces a costly revision.
Why it matters
Word-order flexibility is common across the world's languages, and knowing why some orders are harder to read sharpens general theories of how people build sentence meaning in real time. The shared experimental and analysis code lowers the barrier for other researchers studying sentence processing, and the project's training in experimental design and statistics builds skills used in data science work.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Brian Dillon
University of Massachusetts, MA·Psychology
Explain why a language model's perplexity, its skill at predicting the next word, correlates positively with worse fit to human reading times, by comparing model predictability estimates against reading-time data.
Eszter Ronai
Northwestern University, IL·Psychology
Identify the real-time processing signals of focus, the marking of new or emphasized information, in languages with fixed versus flexible word order, using eye-gaze data and statistical modeling.
Tal Linzen
New York University, NY·Computer Science·Early-career
Explain why language models that predict the next word better match human reading times worse, by comparing model predictability estimates with human reading data.
The school
University of Delaware: 22 labs on PI Pages, $466M a year on research, 298 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.