June 2018: Decoding and Language Factors in Predicting Reading Comprehension

Overview

The current article extends existing research on the Simple View of Reading (SVR; Gough & Tunmer, 1986). SVR is a common theory of reading acquisition that has been extensively researched. SVR can be summarized as an equation relating the component skills of decoding and language comprehension to reading comprehension:

Decoding ∗ Language Comprehension = Reading Comprehension

The fact that the model describes reading comprehension as a product rather than a sum has an important implication. If the model used a sum, deficits in either decoding or language comprehension could be overcome by strength in the other component. This is not the case with the product of decoding and language comprehension—a deficit in either will result in low reading comprehension.

In previous research, studies have shown that an additive version of the model better predicts reading comprehension:

Decoding + Language Comprehension = Reading Comprehension

The current study used the additive version of the model.

SVR also provides expectations for the relations between reading comprehension and the component skills, decoding and language comprehension, during development. SVR predicts declining relations between decoding and reading comprehension and increasing relations between language comprehension and reading comprehension. Finally, SVR suggests that the component skills account for the vast majority of the reliable variance in reading comprehension.

This final theoretical suggestion led to the focus of the current paper—identifying the unique and common sources of variability in reading comprehension. Prediction of one variable from another is possible when variability in the predictor is related to variability in the other variable. In the case of multiple predictors, variance in the outcome may be simultaneously related to variance in more than one predictor. This is called common variance. Variance in the outcome that is related to only a single predictor is called unique variance. 

The authors reviewed the existing literature on SVR and found that prior research tended to report only total variance accounted for by the component skills. The current study delineated variance aligned with reading comprehension into unique and common components.

The authors presented the following two hypotheses: 

1. The unique effect of decoding would decrease and the unique effect of language comprehension would increase after the primary grades.
2. The unique effect of language comprehension and the common effect of language comprehension and decoding would account for the majority of variance in predicting reading comprehension after the primary grades.

To test the hypotheses, the authors represented the three constructs with multiple measures in a confirmatory factor analysis for each of grades 1 to 10. The sample for the study included more than 2,900 students in Florida.

Key Findings

The total variance explained by the model was lower in grades 1–3 (68%, 59%, and 78%) and higher in grades 4–10 (at least 97%). As predicted, the unique variance for decoding generally declined as grade increased and the unique variance for language comprehension generally increased as grade increased. The common variance ranged from 19% to 46%. With the exceptions of grades 2 and 4, the common variance tended to be higher in grades 1–7 than in grades 8–10.

Implications

Most of the previous research on total variation has led to suggestions to focus interventions on word recognition skills. Others have suggested the integration of decoding and language comprehension. Based on the large amount of common variance observed in predicting reading comprehension across grades in the current study, the authors also suggest the integration of decoding and language skills.

Reference

Gough, P. B., & Tunmer, W. E. (1986). Decoding, reading, and reading disability. Remedial and Special Education, 7(1), 6–10.