Interference from Adjacent Words in Visual Word Recognition

Overview

This study investigates the mechanisms underlying reduced lexical activation observed when words appear in their natural contextual state—surrounded by adjacent words. The research tests the hypothesis that such interference arises from parallel processing of flanking words rather than sequential processing constraints. Using a masked visual word paradigm combined with eye-tracking methodology, the investigation examines the spatial parameters governing this interference effect and identifies asymmetries in how adjacent words are processed during visual word recognition.

Methods and approach

Three experiments employed a flanked masked visual word adaptation of the visual world paradigm with adult skilled readers. Target words were presented at fixation for 75 milliseconds, flanked by other words, then masked. Participants' gaze shifts to corresponding images from a four-image display were recorded as the dependent measure. Spatial proximity of flankers was systematically manipulated across conditions. In the third experiment, an image corresponding to one flanker word was included to provide direct evidence of flanker word activation. This design allowed simultaneous measurement of gaze patterns and explicit evidence of lexical engagement with flanking stimuli.

Results

Interference magnitude increased as a function of flanker proximity to the target word. A pronounced asymmetry emerged, with right-side flankers producing substantially stronger interference than left-side flankers. Direct lexical activation was confirmed for right-side flanker words, as evidenced by gaze shifts to corresponding images. Critically, the asymmetry in lexical activation of flankers was substantially more pronounced than the asymmetry observed in overall interference effects, indicating differential involvement of distinct processing levels—lexical and non-lexical—in inter-word interactions.

Implications

The findings constitute evidence for distributed attention mechanisms centered around fixation that differentially modulate various stages of word processing. The dissociation between asymmetric lexical activation and weaker asymmetric overall interference suggests that models positing independent, serial processing of individual words require substantial revision. Word recognition operates within a system of online interactions among adjacent words, with attention dynamically allocated across multiple representational levels. The masked flanked visual word visual world paradigm demonstrates utility as a methodological tool for investigating attentional allocation dynamics during reading processes. Future models of visual word recognition must accommodate parallel processing of multiple words and the asymmetric spatial characteristics of these interactions. The right-hemisphere bias in flanker interference aligns with observations from other laterality studies in reading but requires further investigation regarding its neural and cognitive mechanisms.