The Shared Knowledge between Reading and Writing across Student Profiles
Saved in:
| Title: | The Shared Knowledge between Reading and Writing across Student Profiles |
|---|---|
| Language: | English |
| Authors: | Qun Yu, C. Patrick Proctor, Rebecca D. Siverman, Society for Research on Educational Effectiveness (SREE) |
| Source: | Society for Research on Educational Effectiveness. 2025. |
| Availability: | Society for Research on Educational Effectiveness. 2040 Sheridan Road, Evanston, IL 60208. Tel: 202-495-0920; e-mail: contact@sree.org; Web site: https://www.sree.org/ |
| Peer Reviewed: | Y |
| Publication Date: | 2025 |
| Document Type: | Reports - Research |
| Education Level: | Elementary Education Intermediate Grades |
| Descriptors: | Profiles, Bilingual Students, Multilingualism, Writing Achievement, Reading Achievement, Knowledge Level, Literacy, Intermediate Grades, Elementary School Students, Academic Achievement, Public Schools, Reading Skills, Writing Skills |
| Geographic Terms: | California |
| Abstract: | Background: The Shared Knowledge Theory (Fitzgerald & Shanahan, 2000) posits that reading and writing share key knowledge domains, including word knowledge (e.g., vocabulary, decoding), universal text attributes (e.g., syntax), and procedural knowledge (e.g., text organization). This framework establishes a foundation for examining the co-development of reading and writing. Building upon this, Proctor et al. (2020) explored the shared linguistic knowledge underlying reading and writing in remedial reading classes. Their structural path analysis findings revealed that shared linguistic knowledge predicted both reading comprehension and written expression, with being stronger predictors in reading. Research design of this study set methodological foundations of the present study that both the shared linguistic knowledge and reading and writing relations could be investigated in one comprehensive statistical model through structural paths analysis. Purpose: Given the growing recognition of shared linguistic knowledge between reading and writing--and the limited research on how this interplay unfolds among multilingual students--this study investigates two questions: (1) What distinct reading-writing profiles exist among multilingual students? (2) Do the relationships between shared linguistic knowledge and literacy outcomes differ across these profiles? Latent Profile Analysis (LPA) will identify distinct reading-writing profiles among upper elementary students. Multi-Group Structural Equation Modeling (SEM) will then examine how shared linguistic knowledge predicts reading and writing outcomes across higher- and lower-performing profiles. I hypothesize four profiles: high reading/high writing, low reading/low writing, high reading/low writing, and low reading/high writing. While shared knowledge is expected to predict outcomes across all profiles, its strength may vary--stronger in high-high and low-low groups, and weaker in mixed profiles. Intervention and Research Design: The data for this study come from a larger randomized controlled trial of a small-group literacy intervention designed for multilingual students and their teachers. Grounded in a language-based approach, the intervention explicitly teaches language skills and promotes student discussions to improve reading and writing. A subgroup of students from California public schools were randomly assigned to either the treatment or business-as-usual (BAU) group. Treatment schools received the curriculum and teacher training, with only a subsample of students selected as focal participants for evaluating the intervention's effectiveness. Non-focal students also received the curriculum to minimize bias. Participants: A total of 573 students were selected for the present study, including those in the BAU group (N = 188) and the non-focal student group (N = 388) from treatment schools. Participants spoke diverse home languages, including Spanish (50%), English (32%), Cantonese (8%), and Mandarin (0.9%), with 502 students classified as current or former English Learners (ELs). Both EL status and home Languages variables will be controlled for potential variability. Data Collection: Participants were given a battery of language and literacy assessments in the fall and spring from 2021 to 2024. Data for the current study are drawn from the 2021-2022 school year in grades four and five. A mixed battery of standardized and researcher-developed assessments on language, reading comprehension and writing were administered: (1) "CAPTI Assess/Read Basix (Fall and Spring)": A web-based literacy test measuring receptive vocabulary, word recognition and decoding, reading comprehension (ranging from 0.674 to 0.927; Sabatini et al., 2019); (2) "Core Analytic Language Skills (CALS) (Spring only)": A norm-referenced assessment measuring shared knowledge, including semantics, syntax, and text organization ([alpha] = 0.93, split-half reliability = 0.90; Uccelli et al., 2013, 2015); (3) "Writing Scoring (Fall and Spring)": Essays were evaluated using the NAEP Persuasive Writing Framework rubric (0-6 scale) on Overall Quality, Development of Ideas, Organization, and Language Conventions. Research assistants achieved reliability after scoring 100 essays (pairwise kappa = 0.81, 0.8) and independently scored the remaining essays; and (4) "Writing Coding (Fall and Spring)": Writing productivity was assessed using a researcher-developed coding scheme across six argumentation categories (stance, reasons, evidence, opposing ideas, rebuttals, conclusion). Reliability was established with a kappa of 1.0 for stance, rebuttal, and conclusion, and 0.63-0.78 for other categories. Analytic Plans: We first identified Reading-Writing profiles using LPA with three reading measures (fall receptive vocabulary, word recognition and decoding, and reading comprehension) and two writing measures (fall holistic writing score and productivity) as continuous indicators. Models with increasing profile numbers were compared using fit indices to determine the optimal solution. Next, a one-way ANOVA examined differences in spring writing scores across profiles, followed by Tukey's HSD post-hoc tests. Finally, Multi-Group SEM tested how shared knowledge -- vocabulary, word recognition, and CALS -- predicts reading comprehension and argumentative writing across profiles, analyzing direct effects and potential mediation. Results: Student reading and writing measures collected in the fall and spring were used to identify five distinct literacy profiles (AIC = 4547.36, BIC = 4999.85, SSABIC = 4891.86, Entropy = 0.7), ranging from high- to low-performing readers and writers. Profiles were validated by significant differences in spring writing scores (F = 26.58, p < 0.001), with the largest difference between the highest- and lowest-performing groups ([vertical bar]d[vertical bar] = 1.22, p < 0.001). Profile 1 included the strongest readers and writers, while Profile 2 had the lowest scores in both reading comprehension and writing. A one-way ANOVA confirmed significant variation (F = 26.58, p < 0.001) in spring writing outcomes across profiles. To address the second research question, a multi-group SEM examined predictive relationships as shown in Figure 2. among fall and spring literacy measures across two student profiles. Most path coefficients were significant (p < 0.01); for instance, fall vocabulary consistently predicted fall reading comprehension. Model fit indices (R[superscript 2], AIC, BIC) indicated a good overall fit. However, fall writing quality had low explanatory power, and spring writing quality in Group 1 showed a degenerate fit, possibly due to collinearity. Despite this, the SEM structure was stable across profiles, supporting the generalizability of relationships among literacy measures. Conclusions: This study contributes to the underexplored area of shared knowledge between reading and writing among multilingual students. By identifying distinct literacy profiles, it highlights the heterogeneity in literacy development and offers a framework for understanding multilingual students' literacy growth. The findings aim to refine theoretical models like the Shared Knowledge Theory by extending them to multilingual contexts and to inform more inclusive, targeted instructional strategies for diverse learners. |
| Abstractor: | As Provided |
| Entry Date: | 2026 |
| Access URL: | https://www.sree.org/2025-conference |
| Accession Number: | ED677712 |
| Database: | ERIC |
| Abstract: | Background: The Shared Knowledge Theory (Fitzgerald & Shanahan, 2000) posits that reading and writing share key knowledge domains, including word knowledge (e.g., vocabulary, decoding), universal text attributes (e.g., syntax), and procedural knowledge (e.g., text organization). This framework establishes a foundation for examining the co-development of reading and writing. Building upon this, Proctor et al. (2020) explored the shared linguistic knowledge underlying reading and writing in remedial reading classes. Their structural path analysis findings revealed that shared linguistic knowledge predicted both reading comprehension and written expression, with being stronger predictors in reading. Research design of this study set methodological foundations of the present study that both the shared linguistic knowledge and reading and writing relations could be investigated in one comprehensive statistical model through structural paths analysis. Purpose: Given the growing recognition of shared linguistic knowledge between reading and writing--and the limited research on how this interplay unfolds among multilingual students--this study investigates two questions: (1) What distinct reading-writing profiles exist among multilingual students? (2) Do the relationships between shared linguistic knowledge and literacy outcomes differ across these profiles? Latent Profile Analysis (LPA) will identify distinct reading-writing profiles among upper elementary students. Multi-Group Structural Equation Modeling (SEM) will then examine how shared linguistic knowledge predicts reading and writing outcomes across higher- and lower-performing profiles. I hypothesize four profiles: high reading/high writing, low reading/low writing, high reading/low writing, and low reading/high writing. While shared knowledge is expected to predict outcomes across all profiles, its strength may vary--stronger in high-high and low-low groups, and weaker in mixed profiles. Intervention and Research Design: The data for this study come from a larger randomized controlled trial of a small-group literacy intervention designed for multilingual students and their teachers. Grounded in a language-based approach, the intervention explicitly teaches language skills and promotes student discussions to improve reading and writing. A subgroup of students from California public schools were randomly assigned to either the treatment or business-as-usual (BAU) group. Treatment schools received the curriculum and teacher training, with only a subsample of students selected as focal participants for evaluating the intervention's effectiveness. Non-focal students also received the curriculum to minimize bias. Participants: A total of 573 students were selected for the present study, including those in the BAU group (N = 188) and the non-focal student group (N = 388) from treatment schools. Participants spoke diverse home languages, including Spanish (50%), English (32%), Cantonese (8%), and Mandarin (0.9%), with 502 students classified as current or former English Learners (ELs). Both EL status and home Languages variables will be controlled for potential variability. Data Collection: Participants were given a battery of language and literacy assessments in the fall and spring from 2021 to 2024. Data for the current study are drawn from the 2021-2022 school year in grades four and five. A mixed battery of standardized and researcher-developed assessments on language, reading comprehension and writing were administered: (1) "CAPTI Assess/Read Basix (Fall and Spring)": A web-based literacy test measuring receptive vocabulary, word recognition and decoding, reading comprehension (ranging from 0.674 to 0.927; Sabatini et al., 2019); (2) "Core Analytic Language Skills (CALS) (Spring only)": A norm-referenced assessment measuring shared knowledge, including semantics, syntax, and text organization ([alpha] = 0.93, split-half reliability = 0.90; Uccelli et al., 2013, 2015); (3) "Writing Scoring (Fall and Spring)": Essays were evaluated using the NAEP Persuasive Writing Framework rubric (0-6 scale) on Overall Quality, Development of Ideas, Organization, and Language Conventions. Research assistants achieved reliability after scoring 100 essays (pairwise kappa = 0.81, 0.8) and independently scored the remaining essays; and (4) "Writing Coding (Fall and Spring)": Writing productivity was assessed using a researcher-developed coding scheme across six argumentation categories (stance, reasons, evidence, opposing ideas, rebuttals, conclusion). Reliability was established with a kappa of 1.0 for stance, rebuttal, and conclusion, and 0.63-0.78 for other categories. Analytic Plans: We first identified Reading-Writing profiles using LPA with three reading measures (fall receptive vocabulary, word recognition and decoding, and reading comprehension) and two writing measures (fall holistic writing score and productivity) as continuous indicators. Models with increasing profile numbers were compared using fit indices to determine the optimal solution. Next, a one-way ANOVA examined differences in spring writing scores across profiles, followed by Tukey's HSD post-hoc tests. Finally, Multi-Group SEM tested how shared knowledge -- vocabulary, word recognition, and CALS -- predicts reading comprehension and argumentative writing across profiles, analyzing direct effects and potential mediation. Results: Student reading and writing measures collected in the fall and spring were used to identify five distinct literacy profiles (AIC = 4547.36, BIC = 4999.85, SSABIC = 4891.86, Entropy = 0.7), ranging from high- to low-performing readers and writers. Profiles were validated by significant differences in spring writing scores (F = 26.58, p < 0.001), with the largest difference between the highest- and lowest-performing groups ([vertical bar]d[vertical bar] = 1.22, p < 0.001). Profile 1 included the strongest readers and writers, while Profile 2 had the lowest scores in both reading comprehension and writing. A one-way ANOVA confirmed significant variation (F = 26.58, p < 0.001) in spring writing outcomes across profiles. To address the second research question, a multi-group SEM examined predictive relationships as shown in Figure 2. among fall and spring literacy measures across two student profiles. Most path coefficients were significant (p < 0.01); for instance, fall vocabulary consistently predicted fall reading comprehension. Model fit indices (R[superscript 2], AIC, BIC) indicated a good overall fit. However, fall writing quality had low explanatory power, and spring writing quality in Group 1 showed a degenerate fit, possibly due to collinearity. Despite this, the SEM structure was stable across profiles, supporting the generalizability of relationships among literacy measures. Conclusions: This study contributes to the underexplored area of shared knowledge between reading and writing among multilingual students. By identifying distinct literacy profiles, it highlights the heterogeneity in literacy development and offers a framework for understanding multilingual students' literacy growth. The findings aim to refine theoretical models like the Shared Knowledge Theory by extending them to multilingual contexts and to inform more inclusive, targeted instructional strategies for diverse learners. |
|---|
