Vocabulary knowledge: A significant predictor of silent reading rate

Vocabulary knowledge: A significant predictor of silent reading rate

First Author: Alexandra Spichtig -- Reading Plus, LLC
Additional authors/chairs: 
Jeffrey Pascoe; John Ferrara; Kristin Gehsmann
Keywords: Vocabulary, Reading fluency, Literacy development, Reading Assessment, Reading comprehension
Abstract / Summary: 

Purpose: The lexical quality hypothesis predicts that deeper vocabulary knowledge should facilitate reading efficiency and reading comprehension. Existing literature describes correlations between vocabulary knowledge and oral reading rate, but evidence regarding the relationship between vocabulary knowledge and silent reading rate (SRR) is lacking.

Method: Students who were enrolled in grades 2-12 during the 2018-19 school year (n=157,514) completed an online reading assessment during which 40 vocabulary items were presented within short stems with minimal contextual cues. Students had to identify synonyms for the target words. Vocabulary item difficulty started below a student’s grade level and was adjusted adaptively using IRT methodology. Vocabulary scores from this assessment are highly correlated with results of nationally-normed reading assessments (e.g., Smarter Balanced ELA, r = .80; GRADE, r = .82). A comprehension assessment followed during which students read five leveled passages and answered comprehension questions. Comprehension accuracy and SRR were calculated.

Results: Across grades, the SRRs of students who met a comprehension criterion of >60% were significantly correlated with their vocabulary grade level (r = .38). Hierarchical regression using grade, vocabulary, comprehension, interest, self-efficacy, and reading volume as variables explained 19% of the variance in SRR. Entering vocabulary into the model first accounted for 78% of the explained variance and eliminated the contribution of grade level. When entered last, vocabulary still accounted for 12.4% of the explained variance—the largest unique contribution.

Conclusions: Of the variables considered, the largest amount of variance in SRR was accounted for by vocabulary knowledge. This result supports the lexical quality hypothesis and has implications for reading instruction.