Carnegie Mellon University
20 files

Intervention Study: knowledge enrichment, semantic organization and reading comprehension

Background: Prior research has indicated that semantic organization (as measured by tasks such as semantic fluency and categorization tasks), may contribute to reading comprehension abilities (e.g. Nouwens et al., 2017). In this study, we aimed to examine whether knowledge enrichment experiences brought into a classroom context would (1) increase the taxonomic organization of concepts relevant to the enrichment topic and (2) correspondingly improve reading comprehension for passages on related concepts. Such a finding could strengthen the evidence for a causal link between semantic organization and reading comprehension. Prior knowledge enrichment interventions (specifically a summer camp at a zoo) had been shown to increase children’s taxonomic organization (Unger & Fisher, 2019), although not in a classroom context. Semantic organization was assessed using the Spatial Arrangement Method (see Unger et al., 2016) and reading comprehension was assessed using bespoke passages designed by the researchers to maximize overlap with enrichment intervention content.

Design: 50 children were recruited from 3 public charter schools prioritizing underserved minorities in a northeastern American city. Two third-grade classrooms were recruited from each school. One classroom was assigned to the experimental condition, one to the control condition, in each school. All participating children’s working memory and word recognition skills were measured prior to the reading comprehension and semantic organization pretests. During pre-test session 1, they were administered pretest questions for the subsequent session’s reading comprehension (to determine the children’s baseline knowledge of answers to those questions). The Spatial Arrangement Method, a semantic organization assessment, was then administered, for two image sets (food and living things). The order in which the images sets were tested was counterbalanced across subjects. During pre-test session 2, six experimenter-designed reading passages were presented to participants to read aloud, and comprehension questions were asked after each passage was read. Then two days of hour-long enrichment opportunities were hosted by educators from a local botanical conservatory. Enrichment days were always within one week of each other. Finally, post-testing followed the same 2-session structure as pre-testing: pre- reading questions and semantic organization assessed on day 1, reading and post-reading questions on day 2.

Intervention details:

"Bees and Pollination": During this session, educators from a local botanical conservatory conducted games and activities related to bees, their roles within a hive, insects, pollination and flowers. Additionally, the impact of bees on their ecosystems and on humans particularly was discussed in detail.

"Watersheds": During this session, educators from a local botanical conservatory conducted games and activities related to related to watersheds, freshwater and saltwater ecosystems, fish anatomy, and the importance of water habitat conservation. Additionally, the dynamics of food webs and how damage to one species can impact an entire ecosystem was discussed in detail.

Results: The data showed no effect of the enrichment intervention on semantic organization or reading comprehension as measured by the aforementioned assessments. However, within testing sessions there were some significant findings. In the spatial arrangement measurement of semantic organization, there was a significant effect of type of relationship between items on their pairwise distances, with pairs that were both thematically and taxonomically related placed closest, followed by taxonomic, thematic and unrelated pairs, in that order. This pattern of results replicated for both the living things set and the food set. In the reading comprehension task, question type (easy, literal, inference or taxonomic) was a predictor of correctness, with easy questions answered correctly most frequently, followed by literal questions. Inference and taxonomic questions proved the most difficult with about equal correctness. There was also a weak correlation between “thematic score” (the difference in average distance between thematically related items and unrelated items in the spatial arrangement method) and overall reading comprehension.

Acknowledgements: Many thanks to Anna Vande Velde, Kristen Boyle and Melissa Pocsai for their assistance in data collection. Additional thanks to Cecilia Mastrogiacamo, Rea Isaac, Isabel Rozario, Zhuyi (Elaine) Xu, and Graciela Garcia for coding these data from original response sheets and images.







Usage metrics



    Ref. manager