Reading skills are basic competencies that are needed by each individual for active citizenship in today’s society. A low level of these essential skills can severely restrict the possibility to acquire new competencies in the lifelong learning perspective and to keep up with the fast-paced changes of the job market. However, the OECD PISA 2018 findings show that a high percentage of students do not reach a minimum level of reading skills in Europe.

Reciprocal Teaching (RT) - a multiple meta-cognitive strategy advanced by Palincsar and Brown (1984) - is one of the strategies that has received more attention in educational research to enhance reading comprehension. It can be considered as a scaffolded and interactive discussion technique consisting of four repeated strategies that a good reader should use to understand the information of a text and summarize the content (Oczkus, 2018): predicting, that requires students to make assumptions about the content that will be presented in the next paragraph of the text; questioning, that requires generating questions on different levels, such as recalling details of the text, making inferences; clarifying, that is the process to highlight words that may be difficult for students to understand such as unknown terms, new concepts; summarizing, that involves students in identifying, paraphrasing and summarizing the most relevant information of a text.

The Reading Comprehension-Reciprocal Teaching (RC-RT) program was designed in the Italian context of dull inclusion (Calvani & Chiappetta Cajola, 2019) and incorporates the four main strategies of the RT with some variations. The most relevant concern: (i) cooperative learning in small groups was replaced with working in pairs as the unit is more suitable for students to interact with each other and to be cognitively active in the learning process; (ii) moments dedicated to the activation of inferential processes (“looking beyond the text”) were included in the program.

The RC-RT program is provided in 25 hours (about 3 months) of activities for fourth-grade class with biweekly lessons of one hour and a half each. Teachers received a half-day (4 hours) training and a guide to support the implementation of the program.

The program consists of two parts based on the delivery model and 34 texts of progressive length and difficulty. The first part of the program begins with a working example by the teacher who after reading the text used thinking aloud strategies to answer questions related to the content. Later, students work in pairs to analyze and summarize texts. Pairs are created by the teacher who can choose to rotate students in the pairs randomly or to compose homogeneous pairs. In the second part of the program, the level of complexity raises with a particular reference to the inferential phase of reading during which questions to "look beyond the text" are asked to students to give them the opportunity to understand what the text does not explicitly say. All sessions are concluded in a 10-minute discussion at the class level and feedback by the teacher.

The aim of this study – carried out by researchers of the SApIE association (Society for Learning and Education informed by evidence) – was to evaluate the impact of the RC-RT intervention on reading comprehension and summarizing skills of fourth graders.

     RQ1. What impact does the RC-RT intervention have on reading comprehension and summarizing skills of students in fourth grade?

     RQ2. Do students' initial vocabulary skills moderate the impact of RC-RT program?

A nonequivalent control group pretest-posttest design (quasi-experimental) was adopted. Students in classes assigned to the intervention received the RC-RT program and students in the control group continued with the typical practice proposed by their teachers. 995 students were recruited in 51 fourth grade classes across 33 primary schools throughout Italy and 29 classes were assigned to the treatment and 22 classes to the control. Although the assignment could not be random, students at the baseline were similar for most of the characteristics considered (e.g., pretest scores, gender). After attrition (14.8% of differential attrition) the analytical sample consisted of 250 students in the control group and 421 students in the treatment group. The program was delivered at the class level by the Italian teacher twice a week across three months (from January 2019 to March 2019) in one hour and half sessions during regular Italian lessons. Two measures developed by the research team were administered at the pretest and posttest, together with a lexical test (Test of Verbal Meaning - TVM) administered only at the pretest. • Summarizing Test (ST), it aims to evaluate students’ reading comprehension and summarizing skills using closed-ended questions. • Summary Qualitative Assessment (SQA), it is a qualitative measure that evaluates students’ skills in writing a summary of a text. • The “Gimmy” test – a standardized measure of semantic inferences (Tressoldi & Zamperlin, 2007) – was administered to a subsample of students (56 subjects from the experimental group and 50 subjects from the control group). Students are asked to read and answer ten multiple-choice questions, which require them to obtain inferential information from a text. The impact analysis estimated the effect of RC-RT against the control condition using ANCOVA with cluster-adjusted standard errors to take into account the cluster assignment of classes. This model included the centered pretest assignment variable as a covariate at the student level to control for baseline differences. Furthermore, one-way ANCOVA was used to test if any statistically significant difference in results exists between students with different levels (high, medium, low) of proficiency in vocabulary in the treatment group on the basis of TVM pretest score. Effect size measure for pretest to posttest control group designs was used in order to evaluate the magnitude of the impact of RC-RT on dependent variables using the dppc2 procedure described by Morris (2008).

The results of baseline equivalence analysis showed no statistically significant differences between students’ scores in the two groups for all the measures: t(669) = .235, p = .814 for ST; t(668) = .856, p = .393 for SQA; t(642) = 1.446, p = .149 for vocabulary. The packages sandwich (Hothron et al., 2015) and lmtest (Zeileis et al., 2020) were used to estimate all models in the R statistical software (R Core Team, 2020). The results of the impact analysis showed a statistically significant difference between the two conditions on the outcome measures. Effect sizes showed that RC-RT outperformed the control group with a larger effect size for SQA than for ST (dppc2 = +0.323, p < .01 for ST; dppc2 = +0.456, p <.001 for SQA). The analysis of the impact of the treatment on different levels of proficiency in vocabulary revealed no statistically significant difference in reading skills by level of proficiency in vocabulary, whilst controlling for pretest scores: F(3, 416) = 1.41, p = .237 for ST measure; F(3, 415) = .964, p = .410). The RC-RT intervention was demonstrated to be effective compared to the conventional teaching to increase reading comprehension and summarizing skills of Italian fourth graders. After this evaluation, the RC-RT program was further reviewed for the purpose of a large-scale application to address the weaknesses achieving a final version for the whole class and an adapted version for students with intellectual disabilities.

Calvani, A., Chiappetta Cajola, L. (Eds.). (2019). Strategie efficaci per la comprensione del testo. Il Reciprocal Teaching. Firenze: SApIE Scientifica. Hothorn, T., Zeileis, A., Farebrother, R. W., Cummins, C., Millo, G., Mitchell, D., & Zeileis, M. A. (2015). Package ‘lmtest’. Testing linear regression models. https://cran. r-project. org/web/packages/lmtest/lmtest. pdf. Menichetti, L., (2018). Valutare la capacità di riassumere. Il Summarizing Test, uno strumento per la scuola primaria, Journal of Educational, Cultural and Psychological Studies (ECPS Journal), 18, 369–396. Menichetti, L., & Bertolini, C. (2019). Prova qualitativa per la valutazione della capacità di riassunto: il Summary Qualitative Assessment (SQA). In A. Calvani & L. Chiappetta Cajola (Eds.), Strategie efficaci per la comprensione del testo. Il Reciprocal Teaching (pp. 431-462), Firenze: SApIE Scientifica. Montesano, L. (2020). Vocabolario e Comprensione del testo. Uno strumento per la valutazione del lessico nella scuola Primaria. Firenze: SApIE Scientifica Oczkus, L. D. (2018). Reciprocal teaching at work: Powerful strategies and lessons for improving reading comprehension. ASCD. OECD (2019). PISA 2018 and the EU - Striving for social fairness through education. Paris: OECD Publishing. https://ec.europa.eu/education/resources-and-tools/document-library/pisa-2018-and-the-eu-striving-for-social-fairness-through-education_en Palinscar, A. S., & Brown, A. L. (1984). Reciprocal teaching of comprehension fostering and comprehension-monitoring activities. Cognition and instruction, 1(2), 117–175. R Core Team (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria. Tressoldi, P. E., & Zamperlin, C. (2007). La valutazione della comprensione del testo: proposta di una batteria di approfondimento. Psicologia clinica dello sviluppo, 11(2), 271–290. Zeileis, A., Lumley, T., Berger, S., Graham, N. (2019) Package ‘sandwich’. Robust Covariance Matrix Estimators. https://cran.r-project.org/web/packages/sandwich/index.html