Julie A. Van Dyke

Education «
Research «
Methodology «
Publications «
Collaborators «
Positions «

Haskins Laboratories
300 George Street, Ste. 900
New Haven, CT 06511

Phone: (203) 865-6163 x214
Fax: (203) 865-8963
Email: jvandyke at haskins dot yale dot edu

Senior Research Scientist
Investigator, Neurobehavioral Mechanisms in Reading Comprehension
Investgator, Memory Mechanisms in Support of Language Comprehension (NYU)

Education

top of page «

Ph.D., Cognitive Psychology, University of Pittsburgh, 2002

M.Sc., Computational Linguistics, Carnegie Mellon University, 1996

B.Sc., Computer Science and Linguistics, University of Delaware, 1987

Research

top of page «

Language comprehension requires the ability to construct relationships between non-adjacent linguistic items. For example, a subject must agree with its verb; however, the two are often separated by several words, phrases, or clauses, as in The athlete(s) in the training program that was designed by an Olympic gold-medal winner run(s) every day. At the same time, research in verbal memory has long recognized that the ability to concurrently attend to and process information is quite limited. This constraint suggests that the language comprehension system must rely on memory mechanisms to store partially processed information and retrieve it later as needed. My research focuses on specifying the control structure through which language processes interact with memory, with special emphasis on understanding the role of retrieval failure as a source of comprehension breakdown. I investigate these issues from three perspectives:

1) Basic mechanisms
Recent memory research suggests that the amount of information that is actively maintained in memory is quite small, with some studies suggesting a capacity of only 1-4 items. My colleagues and I have demonstrated that a cue-based direct access retrieval mechanism can both compensate for this limited capacity during sentence processing, and provide a parsimonious account of why certain grammatical constructions are particularly difficult to understand. A signature of such systems is susceptibility to retrieval interference, which occurs when retrieval cues cannot reliably identify needed constituents because they are associated with other similar items in memory. This research aims to specify what conditions lead to interference in skilled readers, how it affects the amount of information held in active memory, and the speed with which stored representations are retrieved.

2) Reading disability
Studies of reading development point to an association between low working memory capacity and comprehension failure, even among readers with decoding skill. However, if the capacity of active memory is estimated at only 1-4 items, alternative explanations must be found. This research aims to investigate variation in retrieval speed, and also in the type of retrieval mechanism (e.g., serial, parallel, cue-based, etc.) employed to access orthographic, phonological, and semantic information. This is important because efficient comprehension requires that all components of word knowledge become available at the right time for integration into the evolving interpretation. A key component of this research is to relate measures of retrieval speed to a large battery of standardized diagnostic measures with the aim of developing sophisticated techniques for remediating comprehension difficulty.

3) Brain damage
Aphasic patients often perform very poorly on short-term memory tasks, typically showing reduced spans of 1-3 words. Explanations for this deficit most often focus on damage to phonological rehearsal mechanisms or to particular types of encodings. The retrieval approach offers a novel alternative explanation, positing that trauma to specific brain regions can reduce the efficiency of retrieval mechanisms necessary for restoring information into active memory. Moreover, patients with damage to the left inferior frontal gyrus show increased sensitivity to interference, which is consistent with this approach. Studies of individuals with focal lesions provide insight into the biological foundations of the retrieval mechanism, and may lead to innovative methods for remediating language difficulties in aphasic patients.

Methodology

top of page «

Speed Accuracy Tradeoff

The critical dependent measure for distinguishing different types of retrieval mechanisms and evaluating how particular disabilities may affect the time-course of accessing information is retrieval speed. Measures of reaction or reading times are insufficient for this purpose, however, because they may be influenced both by the speed of accessing information and the quality of memory representations. The Speed-Accuracy Tradeoff (SAT) method avoids this problem by tracking changes in the accuracy of a response as information accrues over time.

The advantage of this method can be seen clearly in Figures 1a & 1b (below). Reaction/reading times (RTs) represent only one point along a potential response function, and it is unclear whether this point is associated with the rate of information accrual or terminal (asymptotic) performance. In Fig 1a, two RT data points (circular symbols) represent two conditions that differ in asymptotic accuracy alone. That these functions have the same rate of accrual can be seen from the intersecting lines, which mark the point at which the response functions reach their 63% point. In this panel, this occurs at the same time (~.45ms). In Fig 1b, the two conditions have the same asymptotic performance, but differ in speed alone, as can be seen from the intersecting lines, which again show the point at which the response function is 63% complete. This time, the 63% point occurs at a different time in the overall response curve (~.45ms versus ~.52ms). Thus, the crucial advantage to this method is the ability to distinguish whether conditions differ in speed, which is determined by the type of retrieval mechanism (e.g., serial vs. direct access) used to restore information into active memory, or in the level of accuracy, which is determined by the quality of memory representations.

Figure courtesy of Brian McElree

Eye-tracking

Eye-tracking technology provides a non-invasive online method for investigating eye-movements while participants are reading. Cameras mounted either on a helmet (Figure 2 below) or on a remote desktop provide multi-dimensional data that captures both the location of fixations and their duration, as well as patterns of re-reading that may emerge over time. Figure 3 illustrates the dynamical nature of this data, with colors representing the passage of time during reading. We also employ the method to study oral comprehension, either via the "visual worlds" paradigm, which involves measuring looks to pictures that represent the content of an utterance, or in sentence-picture matching tasks, where participants judge whether a sentence they hear matches a picture they see.

*Figure 3: Courtesy of Dave Braze

Publications

top of page «

Van Dyke, J.A. (2007) Interference effects from grammatically unavailable constituents during sentence processing. Journal of Experimental Psychology: Learning, Memory, and Cognition,33(2), 407-430.

Lewis, R.L., Vasishth, S., & Van Dyke, J.A. (2006) Computational principles of working memory in sentence comphrension. Trends in Cognitive Science,10(10),447-454.

Van Dyke, J.A. & McElree, B. (2006) Retrieval interference in sentence processing. Journal of Memory and Language,55(2), 157-166.

Shankweiler, D., Palumbo, L.C., Ni, W., Mencl, W.E., Fulbright, R., Pugh, K.R., Constable, R.T., Harris, K.S., Kollia, B., Van Dyke, J. (2004). Unexpected recovery of language function after massive left-hemisphere infarct: Coordinated psycholinguistic and neuroimaging studies. Brain and Language, 91(1), 181-182.

Van Dyke, J.A. & Lewis, R.L. (2003). Distinguishing effects of structure and decay on attachment and repair: A retrieval interference theory of recovery from misanalyzed ambiguities. Journal of Memory and Language, 49(3), 285-316.

Perfetti, C.A., Van Dyke, J.A., & Hart, L. (2001). The Psycholinguistics of Basic Literacy. In M. McGroarty (Ed.) Annual Review of Applied Linguistics, 21, 127-149

Voss, J.F. & Van Dyke, J.A. (2001). Narrative Structure, Information Certainty, Emotional Content, and Gender as Factors in A Pseudo Jury Decision Making Task. Discourse Processes, 32(2&3), 215-243.

Voss, J.F. & Van Dyke, J.A. (2001). Argumentation in Psychology. Discourse Processes, 32(2&3), 89-111.

Britt, M.A.; Perfetti, C.A.; Van Dyke, J.A.; & Gabrys, G. (2000). The Sourcer's Apprentice: A Computer Tool for Document-Supported History Instruction. In P. Stearns, P. Seixas, and S. Weinberg (Eds.), Knowing, Teaching, and Learning History: National and International Perspectives (pp. 437-469). New York: New York University Press.

Van Dyke, J. & Lehman, J. (1997). ESL-Soar: An Architectural Account of Errors in Second Language Learning. In M. Shafto and P. Langley (Eds.), Proceedings of the 19th Annual Conference of the Cognitive Science Society (p. 781-786). Mahwah, N.J.: Lawrence Erlbaum.

Lehman, J.F., Van Dyke, J., & Green, N. (1995). Reactive Natural Language Processing: Comprehension and Generation in the Air Combat Domain. Proceedings of the 1995 AAAI Fall Symposium on Embodied Language.

Lehman, J.F., Van Dyke, J., & Rubinoff, R. (1995). Natural Language Processing for IFORs: Comprehension and Generation in the Air Combat Domain. Proceedings of the Fifth Conference on Computer Generated Forces and Behavioral Representation.

Laird, J., Johnson, W.J., Jones, R., Koss, F., Lehman, J.F., Nielsen, P.E., Rosenbloom, P.S., Rubinoff, R., Schwamb, K., Tambe, M., Van Dyke, J., van Lent, M., Wray, R. E. (1995). Simulated intelligent forces for air: The SOAR/IFOR project 1995. Proceedings of the Fifth Conference on Computer Generated Forces and Behavioral Representation.

Van Dyke, J.A. Using Syntactic Knowledge for Word Prediction. Proceedings of the Fifth Biennial Conference of the International Society for Augmentative and Alternative Communication (ISAAC), Philadelphia, PA: August 8, 1992.

Collaborators

top of page «

Positions

top of page «

Internships

Unpaid internship positions are frequently available and I welcome contacts from motivated undergraduates in either the New Haven or New York City areas. These positions offer valuable hands-on research experience for individuals considering graduate school or who are simply interested in learning more about reading, language, or memory research. I am happy to provide reference letters for students who have made positive contributions to our research, and will offer authorship on published scientific papers when warranted.

Graduate Students

Opportunities for graduate students to take leading roles on developing and analyzing grant-supported experiments are currently available.

Postdoctoral Fellows

A two-year postdoctoral position may be available beginning in April 2009. In addition, openings are periodically available on the Haskins NRSA training grant. Interested graduates are invited to send a statement of research interest, CV, and relevant publications for consideration.