A Study Of Revision Processing Of Causal Inferences During Chinese Text Reading

Inference,as the core of reading comprehension,has drawn extensive attention from the field of psychology,linguistics,cognitive science and etc.During reading,readers may generate causal inferences,building causal connection between the current sentence and the previous sentence.However,the constructed initial inferences might be inconsistent with the subsequent texts,causing cognitive conflicts.To resolve the cognitive conflicts,readers need to update the previously-constructed situation model and generate revised inferences to maintain a coherent mental representation of discourse.To construct and update the situation model during discourse comprehension,successful causal inference revision is the key to successful comprehension.However,the processing of causal inference revision in Chinese text comprehension is still unknown.Therefore,the current study employed reading time method and eye tracking technique to investigate the mechanism of cognitive processing of causal inference revision in Chinese text comprehension.Four experiments were conducted to examine the cognitive processing of causal inference revision during Chinese discourse reading.Study 1(Experiment la&1b)examined whether readers were able to activate revised causal inferences online.In Experiment 1a,34 university students were recruited and were required to read texts consisting four sentences at their own pace,the inference revision text(semantic inconsistent with triggering inference revision),the non-revised inference text(semantic consistent and supporting initial inferences without triggering inference revision),and the control text(semantically unrelated).After reading,subjects were asked to decide whether a twocharacter word representing the revised causal inference was meaningful or not as accurately and quickly as possible.It was found that participants responded significantly faster to the target word in the inference revision condition than that in the non-revised causal inference and control conditions,and there was no significant difference of the response time of the target word between the non-revised causal inference and control conditions.The results indicated that while reading the inference revision text,readers were able to activate the correct concept related to the revised causal inference.Experiment 1b employed eye tracking technique with high ecological reliability to further examine the cognitive processing of causal inference revision during natural reading.We added a fifth sentence to the material of Experiment 1a,and it was embedded with a target word representing the revised causal inference.Thirty university students participated in the experiment,and they were asked to read the five-sentence texts at a normal reading speed.The results found that there was significantly longer first pass reading time and total reading time in the inference revision text than those in the non-revised causal text.The inference revision text elicited significantly longer regression time before the target word region than the non-revised inference text.However,no significant differences were found between the inference revision text and the non-revised inference text in and after the target word regions.The results suggested that readers could process inference revision successfully before reading the target word in the fifth sentence,which did not affect the subsequent processing of the text.Study 2 combined the eye tracking technique and word-printed visual world paradigm to examine changes of the activation of initial and revised inferences during causal inference revision processing.This paradigm is able to detect the change of semantic activation of initial inferences and revised inferences during discourse comprehension.The experiment recruited 31 university students and they were required to read silently the first three sentences of the causal inference revision text and the non-revised causal inference text.After that,they were asked to listen to the fourth sentence while looking at the words appearing on the screen.Each text was paired with four two-character words,one word representing the initial causal inference related concept,one word representing the revised causal inference related concept,and two other distractors.It was found that,compared with the causal inference revision text,revised causal inference received more fixations while the initial causal inference received less fixations in the non-revised causal inference text during the 1200-1400 ms time window after the discrepancy onset of fourth sentence in both conditions.This result suggested that in order to construct a coherent situation model,the initial causal inference was partially suppressed and the revised causal inference was activated around 400 ms after the discrepancy offset.While the previous two studies focused on causal inference revision processing,Study 3 used the eye tracking technique to examine whether initial inferences inconsistent with the subsequent texts could still influence the processing of subsequent texts after the revision.A total of 36 university students participated in the experiment,and they were required to read inference revision,non-inference(semantic consistent without triggering intial inferences),and control texts.No significant differences were found in the fifth sentence of the inference revision and the non-inference text as well as the initial causal inference of the fifth sentence in the three conditions.The results demonstrated that the initial causal inferences irrelevant to the subsequent texts were likely to be outdated,which might not exert an effect on the subsequent reading processes.The current study employed reading time method and eye tracking technique to systematically investigate the processing of causal inference revision during Chinese text comprehension,and it has theoretical,methodological,and teaching practical implications.Theoretically,the study constructed a four-stage model of causal inference revision of text comprehension to explain the processing of causal inference revision during text comprehension.Methodologically,the study employed the word-printed visual world paradigm to study the processing of causal inference revision,providing an operational technical route for investigating the implicit information processing during Chinese comprehension.In terms of teaching practice,the study offers some guidance for the training of text inference and inference revision ability.