Neal J. Pearlmutter, PhD - US grants

The research will examine the nature of the human sentence comprehension system by examining two aspects of syntactic processing which have previously received little direct attention: whether the comprehension system can maintain multiple competing phrase structure representations in parallel or is instead limited to considering only one possibility at a time; and how the comprehension system handles subject verb agreement, which requires the tracking of syntactic features across intervening material. Both of these issues have implications for a variety of theories of sentence processing and ambiguity resolution. Examining subject verb agreement also provides an opportunity to investigate very general feature processing mechanisms required for a variety of syntactic and non syntactic computations in human language comprehension, and permits consideration of a series of novel predictions made by recent constraint based lexicalist theories. These issues will be investigated with a variety of techniques, including on line reading for comprehension, off line sentence and sentence fragment ratings, sentence completion tasks, and analyses of published text corpora. The experiments examining the possibility of parallelism will determine ambiguity resolution preferences for an ambiguity not previously investigated, and then use the ambiguity to consider the parallelism issue, as well as how it interacts with working memory available to the comprehender, which may be a factor in limiting parallelism. The experiments examining subject verb agreement will investigate the mechanism underlying syntactic feature tracking and the degree to which that mechanism is independent of non syntactic (lexical, semantic/conceptual, and discourse) mechanisms. The results of the experiments should lead to more finely specified linguistic, computational, and psycholinguistic models of comprehension, and a better understanding of human sentence comprehension in particular.

DESCRIPTION (provided by applicant): The proposed research will investigate the nature of the human sentence comprehension system by examining the influence of pragmatic plausibility on the computation of syntactic structure and sentence meaning. The experiments will specifically consider the relationship between plausibility computation, syntactic processing, and individual differences in working memory capacity. Basic behavioral measures of reading difficulty (self-paced reading, continuous anomaly judgment) will be used, along with two neurological measures, event-related potential (ERP) recording and event-related functional magnetic resonance imaging (er-fMRI). These measures will be combined to yield more precise data about behavior and about the functional neuroanatomical loci of syntactic and semantic processes than has been possible with any of the measures alone. The results of these experiments have implications for nearly all theories of sentence processing and ambiguity resolution, as well as for the understanding of individual differences in language comprehension ability and of language impairment due to brain injury. The results should lead to more finely-specified psycholinguistic, neurological, and computational models of human language processing.