Output details
11 - Computer Science and Informatics
University of Edinburgh
A probabilistic corpus-based model of syntactic parallelism
<22> Originality: The first paper to propose a computational model of parallelism in human sentence processing.
Significance: The parallelism effect has long been an unexplained property of human sentence processing. This paper changed the way we think about the effect by showing that caching mechanisms can give rise to parallelism. This provides a unified computational theory of parallelism and priming, which was tested in an experimental companion paper (Sturt, Dubey, Keller, J Memory and Language 2010).
Rigour: Probabilistic parsers and statistical language models trained on realistic corpora were applied to build a broad-coverage model of human sentence processing.