Institute of Education, University of Reading, Reading, UK.
Department of Education, University of Oxford, Oxford, UK.
Q J Exp Psychol (Hove). 2021 Jul;74(7):1202-1224. doi: 10.1177/1747021821999007. Epub 2021 Mar 15.
Inference generation and comprehension monitoring are essential elements of successful reading comprehension. While both improve with age and reading development, little is known about when and how children make inferences and monitor their comprehension during the reading process itself. Over two experiments, we monitored the eye movements of two groups of children (age 8-13 years) as they read short passages and answered questions that tapped local (Experiment 1) and global (Experiment 2) inferences. To tap comprehension monitoring, the passages contained target words which were consistent or inconsistent with the context. Comprehension question location was also manipulated with the question appearing before or after the passage. Children made local inferences during reading, but the evidence was less clear for global inferences. Children were sensitive to inconsistencies that relied on the generation of an inference, consistent with successful comprehension monitoring, although this was seen only very late in the eye movement record. Although question location had a large effect on reading times, it had no effect on global comprehension in one experiment and reading the question first had a detrimental effect in the other. We conclude that children appear to prioritise efficiency over completeness when reading, generating inferences spontaneously only when they are necessary for establishing a coherent representation of the text.
推理生成和理解监控是成功阅读理解的重要组成部分。虽然这两者都随着年龄和阅读能力的发展而提高,但对于儿童在阅读过程中何时以及如何进行推理和监控他们的理解,我们知之甚少。在两项实验中,我们监测了两组儿童(8-13 岁)在阅读短文并回答问题时的眼动,这些问题涉及局部(实验 1)和全局(实验 2)推理。为了挖掘理解监控,短文包含了与上下文一致或不一致的目标词。理解问题的位置也通过问题出现在短文之前或之后进行了操作。儿童在阅读过程中会进行局部推理,但对于全局推理,证据就不那么明显了。儿童对依赖推理生成的不一致性很敏感,这与成功的理解监控是一致的,尽管这只在眼动记录的很晚阶段才出现。虽然问题位置对阅读时间有很大影响,但在一个实验中对全局理解没有影响,而在另一个实验中先阅读问题则产生了不利影响。我们的结论是,儿童在阅读时似乎更注重效率而不是完整性,只有在为文本建立连贯的表示形式有必要时,才会自发地进行推理。