Longitudinal prediction of reading comprehension: from preschool through to adolescence

Longitudinal prediction of reading comprehension: from preschool through to adolescence

First Author: Kate Cain -- Lancaster University
Keywords: Reading comprehension, Word reading, Oral Language, Longitudinal, Reading development
Abstract / Summary: 

This set of papers focuses on the contributions made by word reading and language skills on children’s reading comprehension longitudinally. Paper 1 by Cain and colleagues demonstrates the importance of both lower-level and higher-level language skills in prekindergarten for listening and reading comprehension in grades 1 through 3. Paper 2 by Restrepo and colleagues examines how language and decoding skills in bilingual children predict reading comprehension across the same age range (preK through to grade 3), and whether the relative contributions made by English and Spanish oral language skills change by grade. In paper 3, Florit and colleagues report an investigation into the influence of the home literacy environment on the development of early reading comprehension, showing that it explains additional variance in first grade reading comprehension, beyond decoding and listening comprehension. In papers 4 and 5, we turn to the prediction of growth in both word reading and reading comprehension. Steacey and colleagues (paper 4) identify common covariates of both constructs from grade 1 to 4. Ricketts and colleagues (paper 5) examine a critical language skill - vocabulary – in additional to reading ability. They find high longitudinal stability of these constructs in early adolescence, suggesting little room for reciprocal influences between them in this period of development. Together these papers provide new insights into the development of reading comprehension as children transition from beginner to skilled readers.

Symposium Papers: 

Predictors of reading and listening comprehension from prekindergarten to grade 3

First Author/Chair:Kate Cain -- Lancaster University
Additional authors/chairs: 
Hugh W. Catts

Purpose
Previous research has demonstrated that both lower-level (vocabulary) and higher-level (inference making) language skills predict narrative listening comprehension between 4 to 6 years (Lepola, Lynch, Laakkonen, Silven, & Niemi, 2012). This work did not investigate how these skills predict later reading comprehension. Neither did it include the range of language skills shown to predict reading comprehension across time, namely grammar, comprehension monitoring, and knowledge of text structure, in addition to vocabulary and inference making (Muter, Hulme, Snowling, & Stevenson, 2004; Oakhill & Cain, 2012). Our purpose was to determine the predictive influence of these language skills in preschool on reading comprehension in grade 3.

Method
We measured lower-level (vocabulary, grammar) and higher-level (inference making, comprehension monitoring, knowledge of narrative structure) language skills, and also narrative listening comprehension in pre-kindergarten (N=420, 41% female), as well as later listening comprehension, word reading, and reading comprehension.

Results
In pre-kindergarten, both lower- and higher-level language skills explained a sizeable proportion of variance in narrative listening comprehension. Across time, performance on these assessments in pre-school predicted later listening (in grades 1, 2, and 3). Critically, these pre-kindergarten measures also predicted reading comprehension performance in grade 3, even after controlling for word reading skill.

Conclusions
This study demonstrates the language bases of reading comprehension and, in particular, language skills important for meaning-making are important for later reading comprehension.

The contribution of oral language, higher-level language and decoding across grades reading comprehension in bilingual children

First Author/Chair:Maria Adelaida Restrepo -- Arizona State University
Additional authors/chairs: 
Gloria Yeomans-Maldonado; Carol Mesa; Shelley Gray; Tiffany Hogan

Purpose
This study tests the contribution of English decoding, lower-level and higher-level language in English reading comprehension in first, second, and third grade bilingual children, using a five-year longitudinal sample. Further, we examine the contribution of Spanish oral language skills to English reading comprehension. Few investigations with bilingual students’ reading comprehension examine higher-level language skills. We hypothesize that the magnitude of the contribution of each factor changes as children move up through grades, with English and Spanish lower-level language skills contributing small but significant variance in first grade, and English higher-level language and vocabulary increasing contribution in variance in second and third grade English comprehension.

Method
235 Spanish-speaking children from preschool to third grade completed multiple measures of language proficiency grammar (3), semantics (3), inferencing, comprehension monitoring, decoding (2), and reading comprehension (3) were given at each point, with slight changes per year in the language of assessment (more Spanish in earlier grades). They attended English-only classrooms in kinder through third grade. Three models by grade examining the effect of Spanish language proficiency and vocabulary, and English decoding, oral language, and high-level language in English reading comprehension will be tested.

Results
Data have been collected and are being cleaned for analyses by grade. Latent factors for vocabulary, grammar, listening comprehension, and higher-level language from PreK to 2nd grade will be used to predict reading comprehension by grade.

Conclusions
We will relate the findings from the factor contributions by grade models to extant models of reading comprehension development.

Does home literacy longitudinally account for reading comprehension beyond the simple view of reading?

First Author/Chair:Elena Florit -- University of Padua
Additional authors/chairs: 
Maja Roch; M Chiara Levorato

Purpose
The present longitudinal study focused on reading comprehension (RC) of Italian first graders. The main aim was to analyze whether the informal literacy practices at home (HL; Sénéchal & LeFevre, 2014) longitudinally accounted for RC beyond the components of the Simple View of Reading (i.e., listening comprehension – LC, and decoding – D; Hoover & Gough, 1990).

Method
Participants were 151 Italian children (51% F; 76 months at T1) and their parents (91% mothers) tested at the beginning (T1) and at the end of the first grade (T2). Children were administered tests to assess non–verbal cognitive ability (NCA) and LC at T1, and D and RC at T2. Parents were administered a questionnaire and a checklist on HL and socio–economic information (SES) at T1. Hierarchical multiple regression analysis was carried out with RC as the dependent variable, SES and NCA as control (Step 1), LC and D (Step 2), and questionnaire and checklist on HL (Step 3) as predictors.

Results
The checklist on HL accounted for a significant amount of variance on later RC, over and above the significant contribution of LC and D.

Conclusions
Findings showed that the Simple View of Reading is valid and the HL is a direct longitudinal predictor of RC of young readers. Informal literacy practices, which are carried out spontaneously by parents before formal instruction, support later RC in addition to individual abilities.

Modeling parallel growth between word reading and reading comprehension in children from first through fourth grade

First Author/Chair:Laura M. Steacy -- Florida Center for Reading Research, Florida State University
Additional authors/chairs: 
Hugh W Catts; Donald L Compton

Purpose
Multiple studies have modeled growth and the correlates of change for word reading (WR) and reading comprehension (RC) separately (e.g., Oakhill & Cain, 2012). However, very few studies have simultaneously modeled parallel growth using a single model incorporating important child-level covariates. The purpose of this study was to model WR and RC development in children from first to fourth grade. We were particularly interested in identifying common and unique covariates of growth across WR and RC.

Method
We modeled parallel growth (N= 588) between WR (WJ-III Word Identification) and RC (WJ-III Passage Comprehension) using latent growth modeling. Variance in the growth parameters was modeled using the following covariates: teacher rating of effort, attention, RAN, PA, vocabulary, listening comprehension, and nonword decoding.

Results
The best fitting model consisted of random intercepts, random slopes, and a fixed quadratic term for both WR and RC. We centered the model with the intercept corresponding to the end of second grade assessment. The overall model fit was adequate (χ2 (20) = 147.81, p<.05; RMSEA=.104; CFI=.977; TFI=.968). The correlations between WR and RC growth parameters were substantial (rintercepts =.97; rslopes =.88). For the intercepts, common predictors included teacher rating of effort, attention, RAN, PA, vocabulary, and nonword decoding. Listening comprehension was a unique predictor of RC intercept. For the slopes, common predictors were attention, listening comprehension, and nonword decoding. PA was a unique predictor of RC slope.

Conclusions
Results indicate that during early reading development growth in WR and RC are highly correlated sharing many of the same covariates.

Oral vocabulary, word reading and reading comprehension: What are the relationships in early adolescence?

First Author/Chair:Jessie Ricketts -- Royal Holloway, University of London
Additional authors/chairs: 
Nicola Dawson; Charles Hulme

Purpose
As part of the Vocabulary and Reading in Secondary School (VaRiSS) project, longitudinal data are being collected to probe the relationships between oral vocabulary and reading in early adolescence. In addition to hypothesising that there would be direct reciprocal relationships between oral vocabulary and reading comprehension, we sought to investigate whether oral vocabulary would predict growth in: (i) word reading (Nation & Snowling, 2004; Ricketts, Nation, & Bishop, 2007); and (ii) reading comprehension indirectly via word reading (Tunmer & Chapman, 2012; Wagner, Herrera, Spencer, & Quinn, 2014).

Method
We present data from the first two phases of the VaRiSS project. A battery of tasks assessing oral vocabulary and reading comprehension was administered to an unselected sample of 210 participants at ages 12 (M = 12.01, SD = .33) and 13 (M = 13.07, SD = .34) years.

Results
Autoregressive structural equation models indicated extremely high longitudinal stability in latent variables indexing oral vocabulary, word reading and reading comprehension performance. Beyond this, there was no evidence of cross loadings between constructs.

Conclusions
In early adolescence oral vocabulary, word reading and reading comprehension ability appear to have remarkably high longitudinal stability, with oral vocabulary at 12 years explaining almost all variance in oral vocabulary at 13 years and similarly reading at 12 years (both word reading and reading comprehension) strongly predicting reading at 13 years. The very high longitudinal stability of these constructs suggests there is very little room for reciprocal influences between them in this period of development.