Fig 1.
Experiment 1 condition examples.
The upper three sentences show what a stimulus could look like in three conditions before the eyes cross the boundary (vertical line). We used an identical condition (with target ‘jumps’ already visible prior to its fixation), a condition with an incongruent preview (‘table’) and a condition with a congruent preview (‘waved’). As soon as the eyes move beyond the boundary, the preview changed into the target.
Table 1.
Pre-target means.
Table 2.
Pre-target duration measures analyses.
Table 3.
Pre-target probability measures analyses.
Table 4.
Target means.
Table 5.
Target duration measures analyses.
Table 6.
Target probability measures analyses.
Table 7.
Experiment 2 condition examples.
Table 8.
Experiment 2 mean RTs (ms) and error rates.
Table 9.
Analyses of RTs and error rates: congruent vs. incongruent flankers.
Table 10.
Analyses of RTs and error rates: correct vs. incorrect sentence flankers.
Fig 2.
Our conceptualization of the reading system.
Sub-lexical orthographic information is gathered across multiple words, with stronger activation of letters in the fovea (here ‘cat’) than letters in the parafovea. Sub-lexical information activates word representations and, importantly, parafoveal information may help to activate the word representation belonging to the fovea if there is orthographic overlap, accounting for the orthographic parafoveal-on-foveal effects reported in the literature. Activated word representations are projected onto a plausible location in a spatiotopic representation, based on visual features such as word length and shape. From here, recognized words append to a sentence-level representation that follows syntactic rules: for instance, if word n is recognized as an article, word n+1 is expected to be a noun or adjective (in English). Feedback from the syntactic level to the individual word positions constrains the recognition process while allowing for the simultaneous recognition of multiple words.