Exploring the cognitive mechanism of irrelevant speech effect in Chinese reading: Evidence from eye movements

A wealth of research shows that irrelevant background speech can interfere with reading behavior. This effect is often described as the irrelevant speech effect (ISE). Two key theories have been proposed to account for this effect; namely, the Phonological-Interference Hypothesis and the Semantic-Interference Hypothesis. Few studies have investigated the irrelevant speech effect in Chinese reading. Moreover, the underlying mechanisms for the effect also remain unclear. Accordingly, with the present research we examined the irrelevant speech effect in Chinese using eye movement measures. Three experiments were conducted to explore the effects of different kinds of background speech. Experiment 1 used simple sentences, Experiment 2 used complex sentence, and Experiment 3 used paragraphs. The participants in each experiment were skilled readers who were undergraduate recruited from the university, who read the sentence while their eye movements were recorded using an EyeLink 1000 eye-tracker (SR Research inc.). The three experiments used the same background speech conditions. In an unintelligible background speech condition, participants heard irrelevant speech in Spanish (which none of the participants could understand), while in an intelligible background speech condition, they heard irrelevant speech in Chinese. Finally, in third condition, the participants read in silence, with no background speech present. The results showed no significant difference in key eye movement measures (total reading time, average fixation duration, number of fixations, number of regressions, total fixation time, and regression path reading time) for the silent compared to the unintelligible background speech condition across all three experiments. In Experiment 1, which used simple sentences as stimuli, there was also no significant difference between the silent and intelligible background speech condition. However, in Experiment 2, which used more complex sentences, normal reading was disrupted in the intelligible background speech condition compared to silence, revealing an ISE for these more difficult sentences. Compared with the silent condition, the intelligible background speech produced longer reading times and average fixation duration, more numbers of fixations and regressions, longer regression path reading time and longer total fixation times. Finally, Experiment 3 also produced evidence for an ISE, with longer total reading times, more fixations, and longer regression path reading times and total reading times in the intelligible background speech condition compared with silence. To sum up, the results of the current three experiments suggest that: (1) unintelligible speech does not disrupt normal reading significantly, contrary to the Phonological-Interference Hypothesis; (2) intelligible background speech can disrupt the reading of complex (but not simpler) sentences and also paragraph reading, supporting the Semantic-Interference Hypothesis. Such findings suggest that irrelevant speech might disrupt later stages of lexical processing and semantic integration in reading, and that this effect is modulated by the difficulty of the reading task.