What can we learn about visual attention to multiple words from the word–word interference task?

In this work, we develop an empirically driven model of visual attention to multiple words using the word–word interference (WWI) task. In this task, two words are simultaneously presented visually: a to-be-ignored distractor word at fixation, and a to-be-read-aloud target word above or below the distractor word. Experiment 1 showed that low-frequency distractor words interfere more than high-frequency distractor words. Experiment 2 showed that distractor frequency (high vs. low) and target frequency (high vs. low) exert additive effects. Experiment 3 showed that the effect of the case status of the target (same vs. AlTeRnAtEd) interacts with the type of distractor (word vs. string of # marks). Experiment 4 showed that targets are responded to faster in the presence of semantically related distractors than in presence of unrelated distractors. Our model of visual attention to multiple words borrows two principles governing processing dynamics from the dual-route cascaded model of reading: cascaded interactive activation and lateral inhibition. At the core of the model are three mechanisms aimed at dealing with the distinctive feature of the WWI task, which is that two words are presented simultaneously. These mechanisms are identification, tokenization, and deactivation.