| 其他摘要 | According to the current situation and linguistic information, predicting the upcoming information is an important cognitive intelligence of human beings. The internal working mechanism and neural mechanism of prediction have been the core hot issues in psychology, cognitive science, and cognitive neuroscience. At present, research on the content of prediction reveals that there is a pre-activation of conceptual and syntactic representations during language comprehension. However, the pre-activation of phonological and orthographic features is controversial, and the important reason is that: Existing findings on word form prediction (i.e., phonological and orthographic prediction) come largely from studies conducted with languages with alphabetic scripts, making it difficult to dissociate the effects of phonology and orthography. Therefore, it is difficult to investigate the pre-activation of phonology and orthography independently. Moreover, most of the previous research focused on the mechanism of prediction processing itself, but little was known about the cognitive regulation mechanism of prediction. Clarifying these key problems is a necessary step in the comprehensive analysis of linguistic prediction processing. Therefore, using Chinese as the target language, the present study examined different levels of prediction (semantic, phonology, and orthography) and the modulation of linguistic prediction by working memory in spoken comprehension using eye-tracking technology and the visual world paradigm.
In Study 1, we used the visual world paradigm to investigate whether comprehenders predict semantic, phonological, and orthography information during comprehension of Chinese sentences, where phonology and orthography are largely dissociable. In Experiment 1, participants listened to sentences containing a highly predictable word while viewing a visual display consisting of a critical object and three distractors (target object, a semantic competitor object, a phonological competitor object, or an unrelated object). Results showed participants fixated more on the semantic competitors and phonological competitors than on unrelated objects. More importantly, the fixation preference began before the participants heard the target word. The results showed that the semantic and phonological features of the spoken target word were pre-activated before the onset of the spoken target word. In Experiment 2, we verified the repeatability of the research results by manipulating preview time (2 s in Experiment 1 and 1 s in Experiment 2) and using a new set of stimuli. The results repeated anticipatory fixations on phonological competitor objects. In Experiment 3, we conducted a printed-word vision of the visual world paradigm to further verify the robustness of the phonological prediction and explore the pre-activation of orthographic information. We repeated the phonological prediction effect but did not find anticipatory fixations on orthographic competitor words.
In Study 2, we investigated how working memory capacity (WMC) modulates different kinds of prediction behavior, i.e., semantic prediction and phonological prediction, in the visual world, which provide insights into the role of general cognitive processing in linguistic prediction. All aspects of Experiment 4 were the same as those of Experiment 1, except that we conducted a Chinese version of the Reading Span task to measure the working memory capacity and grouped participants into high and low WMC groups. We found anticipatory fixations to semantic and phonological features in the predictive time window. Moreover, we found that high WMC speakers are more likely and earlier to have anticipatory fixations on semantic and phonological features than low WMC speakers. This study reveals the modulatory mechanism of working memory capacity on semantic and phonological prediction effects.
In summary, we demonstrated that listeners can predict not only semantic features of predictable words but also more specific phonological information during spoken sentence comprehension. We did not find the pre-activation of orthographic features during spoken comprehension. We also found that verbal working memory modulates semantic and phonological prediction in spoken comprehension. Specifically, the higher the working memory capacity, the larger and earlier the semantic and phonological prediction effects. This study deepens the understanding of the prediction content and processing mechanism and provides scientific evidence for the construction of a more complete language understanding model. It also provides important implications for natural language processing and language-like brain intelligence. |
修改评论