Reading Comprehension, Bridging Inferences, and their Relation to Working Memory Processes in Children in Grades Three through Eight

Date

2013-01-07

Authors

Johnston, Amber, M.

Journal Title

Journal ISSN

Volume Title

Publisher

University of Guelph

Abstract

Reading comprehension (RC) is a complex, dynamic process by which readers interact with text to construct meaning. It relies on word-level decoding and vocabulary skills, text-level skills such as inference, and general-purpose cognitive processes such as working memory (WM). Bridging inferences, which connect different parts of text to maintain semantic coherence, are necessary for comprehension. WM is thought to provide the mental workspace that allows readers to comprehend text, including making bridging inferences. This dissertation includes two studies that investigate related but unique questions regarding inference-making, WM, and RC in school aged children. The first study demonstrated that bridging inference making increased with age and was affected by text distance such that inferences across larger chunks of text were more difficult to make than those between adjacent sentences. Bridging inferences were also affected by knowledge domain such that affective inferences based on character goals, desires, or motivations were made correctly less often than were concrete inferences based on concrete, causal information. Semantic updating, an aspect of WM that involves efficiently revising the contents of WM, predicted variance in the far-concrete inferencing condition. Semantic reactivation, an aspect of WM that involves bringing previously processed information back into WM, predicted variance in the far inferencing conditions regardless of knowledge domain. The second study investigated the word-level and text-level skills and general purpose cognitive processes that support performance on three different standardized RC measures. Semantic updating and semantic reactivation predicted variance on the RC tool considered to measure mental model building (WJIII-PC, Woodcock-Johnson-III passage comprehension subtest). Semantic reactivation also predicted variance on the RC tool considered to measure the ability to build and then reflect upon the mental model (WIAT-II-RC, Wechsler Individual Test of Achievement reading comprehension subtest). A measure of verbal WM predicted performance on one RC measure in the presence of word-level and text-level skills but only when the hypothesized components of WM (updating and reactivation) were not included in the model. Additionally, bridging inference making uniquely predicted performance on all three RC measures. The findings suggest readers coordinate different cognitive processes depending on the comprehension task.

Description

Keywords

reading comprehension, bridging inferences, working memory, children

Citation