On-line sentence processing, aphasic impairments, computational modelling
Final Report Abstract
A major finding in Experiment 1 was that individuals with aphasia (IWA) have a delayed prediction process, but that they are capable of using case information to build the incremental parse. The control data in experiment, which came from two age groups, also showed that age has significant effects on speed of processing and of prediction. Experiment 2 investigated case ambiguous SVO and OVS sentences in which an agreement cue at the verb disambiguated the sentence. Controls (both young and old) showed a tendency for a subject-first bias, which led to a garden path in OVS sentences; controls were able to use the agreement information quickly to revise their initial parse. This shows that, unlike case-marking information, agreement information is easier to use for revision processes. IWA did not show the early subject-first bias that controls showed; IWA took longer to build up an SVO bias, and were slower than controls in revising their parse once the agreement information appeared. Thus, here too we see evidence for delays in processing in IWA. Experiment 3 investigated case-marked subject and object wh-questions. Here too, IWA showed a delay in parsing. Experiment 4 used number marking to disambiguate subject- and object-wh questions, but this experiment could not disentangle the effect of number marking from the effect of case marking. The computational modeling was highly successful; we demonstrated how impairment in IWA can be explained at the individual level by assuming varying degrees of (a) noisiness in activation of chunks in memory, (b) delays in parsing. This model is unique since it can explain both impaired and unimpaired parsing in online processing. We also developed a powerful new framework for future modeling by seamlessly integrating the parsing system in ACT-R with an eye-movement control model; and we developed several new computational models of underspecification, one of which won the best poster award at the AMLaP conference. All this work will serve as a foundation for future research. In sum, this project was very successful in achieving its stated goals, and went far beyond its original objectives. Despite a slow reviewing process we were able to maintain an excellent publication record.
Publications
-
Advanced review: What eye movements can tell us about sentence comprehension. Wiley Interdisciplinary Reviews: Cognitive Science, 2012
Shravan Vasishth, Titus von der Malsburg, and Felix Engelmann
-
(2013). Sentence comprehension disorders in aphasia: The concept of chance performance revisited. Aphasiology, 27, 112-125
Frank Burchert, Sandra Hanne, & Shravan Vasishth
-
A framework for modeling the interaction of syntactic processing and eye movement control. Topics in Cognitive Science, 5(3):452–474, 2013
Felix Engelmann, Shravan Vasishth, Ralf Engbert, and Reinhold Kliegl
