Language Comprehension | Julie A. Van Dyke

Dr. Julie A. Van Dyke

I study the miracle of language comprehension. Language is hugely underspecified and typically comes to the comprehender in disconnected bits and pieces. Related information can be separated by many words, sentences, or even paragraphs-- totally amazing that anyone understands anything!

My research seeks to identify the neurocognitive mechanisms that comprehenders use to create complex meaning representations out of those bits and pieces. A primary focus is on characterizing the retrieval system used to access the previously stored information needed to maintain coherence as new information is encountered. A chief finding is that certain contexts create retrieval interference, resulting in faulty or incomplete meaning. I investigate how retrieval failures impact normative comprehension processes and how they contribute to poor comprehension in certain individuals (i.e., those with language or reading impairments, ADHD, or age-related cognitive decline.) Identifying the neurocognitive sources of sensitivity to interference is crucial for understanding the biological roots of linguistic and attention-based disabilities, as well as understanding how and why comprehension ability changes across the life-span.

Cue-based Retrieval Theory and Retrieval

Interference

Grammatically related items often occur across long distances. We investigate how cues are used to retrieve the distant information needed to create meaning, and how retrieval interference leads to comprehension failure.

Mechanisms underlying Reading Comprehension difficulties

Reading comprehension disability has the same prevalence as

dyslexia, but is much less studied. Our model specifies how under-specified grammatical encodings and shallow lexical representations

affect comprehension.

Diagnosing Direct Access Retrieval via the Speed-Accuracy Tradeoff Method

The response-signal Speed-Accuracy Tradeoff method provides speed measures that are not confounded by response thresholds. These are necessary for identifying the computational properties of retrieval.

control

Eye-movement models traditionally have not considered person-level factors. We provide data showing the importance of individual skills in determining eye-movements and the effects of skilled oculomotor control on fluency.

Machine Learning to identify skill-based classifiers of reading ability

Clinical research involves datasets with many predictors and few observations. Machine learning techniques provide a means of ranking predictor importance while avoiding model overfitting and collinearity.

Fixation-related Brain Imaging of Word-by-word Reading

Eye-movement monitoring provides a realtime index of brain activations during word-by-word reading. We are using this method to identify brain regions associated with word-by-word integrative processing.

Predictive timing as a controller of Fluency Disorders

Fluency requires scheduling motor responses to be in sync with perceptual processes. We examine the association between disordered predictive timing in the motor system, indexed by beta oscillations, and fluency disorders.

Julie A. Van Dyke, Ph.D.