Skip to main content
eScholarship
Open Access Publications from the University of California

Naturalistic Reading Time Data Support Information Locality

Abstract

Both prediction and working memory constraints have been established as key factors in characterizing incremental sentence processing difficulty. Here we investigate the less explored question: Whether and how predictive expectation and working memory interact with each other using data from naturalistic reading time corpora. We provide broad-coverage evaluations of two hypotheses that make divergent predictions regarding the interaction of expectation and memory constraints: the Information Locality and Prediction Maintenance hypotheses. We first confirmed the predictions of both expectation- and working memory-based theories. Regarding their interactions, we find support the Information Locality hypothesis: Strong mutual predictability can enhance locality effects. We argue that future theory building in sentence processing should therefore take into consideration both prediction and memory constraints, as well as their potential interaction.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View