Linguistic processes intervene much later than visuo-motor processes during an eye fixation: Evidence from Fixation-Related Potentials during reading.
authors
document type
PROCEEDINGSabstract
Eye-movement behavior during reading have been studied for more than a century. Most models share the assumption that readers’ saccades are controlled in a top-down manner by the needs of ongoing word-identification processes. However, linguistic processes are time consuming, and it remains undetermined whether eye fixations during reading (on average 225ms) are long enough to enable predominant top-down control. Here, we investigated the time course of visuo-motor and linguistic processes during reading using Fixation-Related Potentials (FRPs). Twenty participants read 316 sentences from the French Sentence Corpus, as well as 316 z-transformed sentences (all letters replaced by the letter “z”), while their eye movements and their electroencephalogram (EEG) were co-registered. EEG and eye-tracking data were synchronized and integrated using the EYE-EGG toolbox. FRPs were first analyzed using Unfold, an EEG-deconvolution and (non-)linear-modelling toolbox allowing to adjust for overlapping EEG signals. Then, reconstructed FRPs were estimated across participants with Linear Mixed-Effect Models, separately for different time windows. Linguistic content (word vs. z-string), frequency and length of the fixated word/string, and incoming saccade launch site, were entered as predictors. Results showed early effects (about 100/120ms from fixation onset) of visuo-motor factors (word length and launch site). The effects of language-related variables occurred on average later (from about 140ms and 230ms for linguistic content and word frequency, respectively). However, as the saccade was launched from closer to the word’s beginning and/or the fixated word was shorter, these effects emerged earlier on (from about 100ms and 110ms, respectively). Our findings indicate that word-identification processes intervene much later during the time course of an eye fixation than visuo-motor processes, except in the rare instances where words can benefit from parafoveal preview before being fixated. Therefore, given the rather short intersaccadic intervals during reading, language-related processes may not predominate in controlling readers’ eye movements.
