According to the misperception hypothesis, a word (ankle) may be misperceived as a high frequency neighbor (angle) during reading, an error that is only noticed and corrected after disambiguation (Slattery, 2009). In contrast to previous studies, we found that sentence fragment contexts biasing towards the target word over its neighbor elicited shorter early fixation times in both standard eye movement measures and an ex-Gaussian analysis, even before disambiguation. We propose that predictable context allows the processor more rapid access and integration of a target word over its lexical neighbors, rather than total misperception of a word as its neighbor.
--------------------------------