A comparison of thematic relation estimations of poor comprehenders and typically developing young readers

A comparison of thematic relation estimations of poor comprehenders and typically developing young readers

First Author: Alexandra Schmitterer -- DIPF | Leibniz Institute for Research and Information in Education
Additional authors/chairs: 
Garvin Brod
Keywords: Comprehension difficulty, implicit knowledge, semantic learning
Abstract / Summary: 

Purpose: Recent evidence suggests that thematic knowledge is organized in a small-world network structure, which means that most words in a semantic network are only loosely connected while few words have strong and numerous connections. Sensitivity to semantic relations can differ between individuals. For example, poor comprehenders and typically developed readers have been found to differ in their sensitivity to different types of semantic relations (i.e., thematic vs. categorical). This study aimed to uncover whether poor comprehenders differ in their ability to estimate thematic relations.

Method: To this end, 3rd grade students performed a category construction task in which different levels of co-occurrence strength between a target and two test words were contrasted. The target word (i.e., lightning) was embedded in a context sentence (i.e., Miriam sees the lightning.) and presented first. Then, two words were presented that varied in relation strength to the target word (i.e., strong–weak: thunder/fire; strong–distant: thunder/letter; weak–distant: fire/letter). Children decided which word would fit better to the sentence, for each of the 45 pairs.

Results: Preliminary analysis with 48 children (28 poor comprehenders) showed that both groups were more likely to pick a more strongly related word but poor comprehenders did so significantly less often. Poor comprehenders were also less likely to decide for a strongly related word than typically developed readers if strongly interconnected items were presented (i.e., strong-distant, strong-weak conditions). This points to processing differences between the groups in high-integrity areas of the thematic network.