Eye-movement patterns during reading are consistent with proposition-by-proposition models of speech comprehension. (1) Fixation times are least affected by word length at ends of clauses: this confirms the theory based on the study of speech comprehension that, at such points, attention is directed inward to text integration. (2) Fixation time for words at the end of clauses is longer for canonical sequences which conform to the pattern, "N-V-N = actor-action-object"; this confirms the view that complete semantic integration of such sequences is routinely delayed until their completion. A two-variable model based on principles of exhaustive visual search and speech comprehension accounts for 80% of the fixation durations in a sample paragraph from Just and Carpenter (1980) and Thibadeau, just and Carpenter (1982). This comprehension-based model compares favorably with their models constructed within a production-system framework. This result clarifies the relationship between reading and listening.
ASJC Scopus subject areas
- Experimental and Cognitive Psychology
- Cognitive Neuroscience
- Artificial Intelligence