Learning and Instruction 19 (2009) 272e286 www.elsevier.


Improving students’ reading comprehension skills: Effects of strategy instruction and reciprocal teaching
Nadine Spo ¨ rer a,*, Joachim C. Brunstein a, Ulf Kieschke b

Department of Psychology, Justus-Liebig University of Giessen, Otto-Behaghel-Strasse 10F, D-35394 Giessen, Germany b Department of Psychology, University of Potsdam, Karl-Liebkecht-Strasse 24/25, D-14476 Potsdam, Germany Received 20 December 2007; revised 18 March 2008; accepted 6 May 2008

Abstract The aim of this study was to investigate the effects of three different forms of strategy instruction on 210 elementary-school students’ reading comprehension. Students were assigned to any one of three intervention conditions or to a traditional instruction condition (control condition). Training students were taught four reading strategies (summarizing, questioning, clarifying, predicting) and practiced these strategies in small groups (reciprocal teaching), pairs, or instructor-guided small groups. At both the post- and follow-up test the intervention students attained higher scores on an experimenter-developed task of reading comprehension and strategy use than the control students who received traditional instruction. Furthermore, students who practiced reciprocal teaching in small groups outperformed students in instructor-guided and traditional instruction groups on a standardized reading comprehension test. Ó 2008 Elsevier Ltd. All rights reserved.
Keywords: Reciprocal teaching; Reading comprehension; Reading strategies; Strategy instruction

1. Introduction A widespread goal of education in the elementary school is reading comprehension for all students because reading comprehension provides the basis for a substantial amount of learning in secondary school (Alvermann & Earle, 2003; Kirsch et al., 2002). In the last 20 years, a major goal of reading comprehension research has been to identify effective reading strategies that increase children’s comprehension (National Reading Panel, 2000). But as Guthrie, Wigfield, Barbosa, et al. (2004) pointed out, the evidence rests primarily on instructional research in which single cognitive strategies are taught in controlled experiments. Relatively little is known about the issue of how multiple strategies can, and should, be combined in comprehension instruction. In multiple strategies programs, strategy practice is often supported by peer-assisted learning arrangements (Brown, Pressley, Van Meter, & Schuder, 1996; Klingner, Vaughn, & Schumm, 1998; Palincsar & Brown, 1984). However, only a few investigations have addressed issues related to the identification of the effective elements inherent in multiple strategies programs. Consequently, the aim of this study was to examine the effect of strategies being taught on reading comprehension and how these strategies are practiced in relevant instruction. 1.1. Reading-comprehension strategies A substantial body of research suggests that reading-comprehension instruction should include explicit cognitive strategy instruction (Guthrie, Wigfield, Barbosa, et al., 2004). The theoretical bases for this suggestion are reading comprehension models,
* Corresponding author. Tel.: þ49 641 9926194; fax: þ49 641 9926199. E-mail address: nadine.spoerer@psychol.uni-giessen.de (N. Spo ¨ rer). 0959-4752/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.learninstruc.2008.05.003

Hacker and Tenent (2002) found that elementary-school teachers made many modifications to adapt RT to the requirements of mainstream classroom instruction. reciprocal teaching (RT) is an instructional procedure developed by Palincsar and Brown (1984) to improve students’ text comprehension skills through scaffolded instruction of four comprehension-fostering and comprehension-monitoring strategies (Palincsar & Brown. Hart & Speece. such as (a) activating background knowledge (Dole. many studies have been conducted to test the effectiveness of reciprocal teaching. Some teachers combined small-group activities with whole-class instruction to make the collaborative learning process easier for students as well as for the teacher. These processes are considered to be cyclic or recursive because each process entails information that can lead to changes in a subsequent step of the cycle. Palincsar. The overall goal is to promote. No study has examined students’ mastery of clarifying strategies. who can be a teacher or a student. 1994). Although there is clear evidence that RT promotes reading comprehension. Other teachers required their students to write down their questions. & Brown. vocabulary. As the students in the group become more familiar with the strategies and the procedure. Meister.N. no study analyzed the separate effects of the various aspects of RT and only a few studies have examined if RT students improved their strategies skills in terms of successfully applying a strategy to a passage. 1984). Rosenshine and Meister (1994) reported a mean effect size of . and (d) predicting what might come next in the text. Brown. Marks et al. answers. and (c) self-assessment of strategy outcome and task performance. Reading comprehension is correlated with a number of cognitive and metacognitive strategies. An underlying assumption of RT is that by applying the strategies in a group process. The dialogue leader. 1996). 1987). / Learning and Instruction 19 (2009) 272e286 273 such as Cromley and Azevedo’s (2007) direct and inferential mediation (DIME) model which in turn is based on Kintsch’s (1988. The model further suggests that the effect of strategies on comprehension is mediated by inference. 1994. Guthrie. 1991). effects on making predictions were assessed in only one of these studies (Dermody. during reciprocal teaching students are engaged in cognitive and metacognitive activities: they alternate between prompting the use of a strategy. 2006). (b) summarizing text (Armbruster. it is thus unclear which strategies of RT significantly contribute to the development of students’ reading comprehension skills. David. Anderson. & Wardrop. So far. Klingner et al. Marks et al. At present. For generating questions. for training college students see also Hart & Speece. & Ostertag. and summaries. Wigfield. In a meta-analysis involving 16 studies.32 for standardized test and .88 for experimenter-developed task favouring RT over control groups. 1990. these processes qualify as self-reflective cognitions in the sense that selfmonitoring of learning activities and associated corrective processes are central features of each step included in the cycle. In a qualitative analysis. 1987). the self-directed and flexible use of the learned strategies. Even though students were taught the entire set of four strategies in 12 out of 16 studies. (1993) observed that teachers sometimes changed RT in a way that elements supposedly playing a critical role in promoting deeper levels of reading comprehension. Besides these open questions regarding the empirical identification of effective strategies involved in RT. and inference that together result in reading comprehension. dialogue leaders fade their involvement and other students take turns as discussion leaders. The DIME model hypothesizes relationships among background knowledge. Hence. 1998. such as students’ . 2004. These four strategies are involved in RT in ongoing dialogues between a dialogue leader and the remaining students of the learning group. Reading vocabulary and background knowledge directly contribute to reading comprehension and also have effects that are mediated by inference. applying the selected strategy. word reading. To sum. through scaffolding instruction and collaboration. and providing scaffold instruction during which teachers gradually fade their modelling of the strategies (Hacker & Tenent. self-regulation procedures as described by Zimmerman (1998) are integral to RT. Spo ¨ rer et al. In addition. For example. Drawing on Zimmerman’s model. & Martin. (c) clarifying word meanings and confusing text passages. and helps students to apply a strategy to a passage. a training effect for summarizing only could be established. some researchers have examined how they work together in more complex strategy packages (Brown et al. age groups. 1984. application of the strategies using rich and meaningful reciprocal dialogues. five out of six studies found no reliable difference between RT and control groups. Since Palincsar and Brown’s (1984) seminal work. 1996. The procedure has been applied to different settings. and populations (Alfassi. There were significant improvements in four out of five studies in which researchers collected summarization probes. (b) selfmonitoring of one’s accuracy in implementing a selected strategy. (b) summarizing parts of the text. reading strategies. models the use of the strategies. 1998). & Vye. it is not clear if all or only one of the taught strategies is effective in fostering students’ reading comprehension (Rosenshine & Meister. and monitoring its accurate implementation. 2002. provides conditional knowledge about strategy use. 2003. & Chapman. Valencia. Souvignier & Mokhlesgerami. especially less able students can learn from their more knowledgeable peers.. Palincsar & Brown. (a) generating one’s own questions.. 1993). 1998. that is. Lysynchuk. Furthermore. Moore. and (c) generating questions to capture the main idea of the passage (Rosenshine. Greer. 1984. the following elements are essential to RT: instruction of the four comprehension-fostering and comprehension-monitoring strategies. although in all six studies RT students significantly improved in their reading comprehension relative to control students. 1998. Palincsar & Brown. Pressley. 1998) constructioneintegration model. & Wilkinson. self-regulation is assumed to be organized within a learning cycle that capitalizes on three types of self-reflective thoughts: (a) goal setting and strategic planning. Palincsar. Although these cognitive and metacognitive strategies have most frequently been investigated in isolation. In this model. 1988).. 1989. Rosenshine & Meister. 2001. Le Fevre. A theoretical basis for suggesting effects of strategy instruction (which strategies are taught) and reciprocal teaching (how are strategies practiced) is Zimmerman’s (1998) self-regulation model. a number of difficulties with implementing and practicing RT have been reported in the literature (Fuchs & Fuchs. & Perencevich.

many teachers are unfamiliar with the procedure of reciprocal teaching. student practice. Fuchs. In the first intervention condition (RT condition). thus. we integrated methods of direct instruction with cognitive modelling and phases of independent reciprocal teaching to help students acquire the four reading strategies of summarizing. summarizing. and a prediction RT ¼ reciprocal teaching. and predicting. ask students to apply a strategy and give feedback about the quality of strategy used. 2001). clarifying. thus. 1984). Compared to RT. One possibility of implementing RT in regular classroom lessons without losing major features of the program is to combine RT with peer-assisted learning arrangements in which students read in pairs (Fuchs & Fuchs. 2006). 2006). & Simmons. Also. and predicting. such as partner reading. is more likely to be accepted by teachers (Fuchs & Fuchs. the instructional technique for helping children to develop responsibility for strategic behaviour is challenging. For young children. 1. and then led to continue practicing these strategies in pairs. we adopted and further advanced the argument that there is a need for identifying effective elements of a multiple strategies program. Fuchs. In a third strategy condition. reading in pairs has several advantages regarding the implementation of strategy instruction in regular classrooms. labelled reciprocal teaching in pairs (RTP condition). þ þ RT þ þ þ RTP þ þ þ þ .274 N. IG ¼ instructor-guided reading. Similar to RT. Since teachers strongly rely on their beliefs and knowledge about instruction when adding new practices to their teaching repertoire (Borko & Putnam. The task of the instructor was to model the four reading strategies. Spo ¨ rer et al. At the beginning. a small group of 4e6 students was guided by a graduate assistant (the instructor) during the course of the intervention. The present study In our study. in their pairs students keep track of their reading activities on score cards that serve as external metacognitive guides and. and students are taught to enact these activities independently. two students are paired to share and practice reading activities (McMaster. Altogether. As Fuchs and Fuchs (2001) stated. partner reading) and. Second. 1996). Fuchs. students were first taught the four reading strategies of summarizing. reading in pairs is more similar to instructional procedures teachers often adopt in reading lessons (e.. Table 1 Characteristics of intervention conditions. RTP ¼ reciprocal teaching in pairs. our first aim was to examine if both strategy instruction and reciprocal teaching contribute to the acquisition of reading strategies and. questioning.g. Peer-assisted learning lessons consist of a set of structured activities. Third. Similar to Palincsar and Brown’s (1984) RT program. Accordingly. Mathes. Hacker and Tenent (2002) argued that researchers have to consider that teachers. First. we examined three intervention conditions and a traditional instruction condition (control condition) in terms of their effectiveness. students who work in pairs have more opportunities to practice the use of reading strategies than students who work together in small groups. 1994). / Learning and Instruction 19 (2009) 272e286 continuous engagement in reciprocal teaching. Later on. RT may be inappropriate for elementary-school children because students of all ability levels are assigned the role of the group leader and thereby have to take on responsibility for the group’s learning. in which students practiced traditional RT (Palincsar & Brown. Our second aim was to examine the effects of a potentially more classroomappropriate intervention and so we created a condition in which RT was practiced in pairs. a summary. there are two main reasons why it may be difficult for teachers to implement RT in naturally constituted classrooms: First. To manipulate the presence versus absence of reciprocal teaching as an integral part of the strategy training we created a second strategy condition labelled instructorguided reading condition (IG condition). & Fuchs. were completely dropped from the instruction. and predicting. questioning. this may result in a cognitive overload (Rosenshine & Meister. 1994). namely RT (Rosenshine & Meister. too. In this condition. to the development of young students’ reading comprehension. Second. students’ roles can be reciprocal so that both students in a pair serve as tutor during each lesson. facilitate structured working in pairs (McMaster et al. clarifying. Instructional elements Conditions IG Strategy instruction Instructors use explicit instruction and cognitive modelling of reading strategies Practice in strategy use Students apply a strategy to a passage Scaffolding instruction and reciprocal teaching Instructors fade their involvement and students take turns as discussion leaders Recording reading activities on worksheets Students write down words to clarify. 2001. these pairs engaged in reciprocal dialogues while they practiced the use of the learned strategies (see also Table 1 for a comparison of the three intervention conditions). 1997). thus. need to take ownership of their learning by constructing their own understanding of new curricula and methods using their prior knowledge.2. questions a teacher might ask. and teacher feedback.. teachers use a set of briefly scripted lessons including teacher presentations.

one or two groups per class) were instructed in the RTP condition. As Rosenshine and Meister (1994) stated. teaching students from Grade 1 to 6. et al. Furthermore. Guthrie.1. Because we were not allowed to collect data about parents’ household income and education level. 1996. we posed the following question: To what extent does RTP differ from RT in influencing reading strategies and reading comprehension? 2.. 1984) and emphasizing the influence of explicit reading strategy instruction on reading comprehension. Another eight groups (42 students. Wigfield. The primary language of the children was German (86%). and (c) the transfer of the learned strategies to experimenter-developed task (near transfer) and standardized reading comprehension test (far transfer). one school was randomly assigned to the traditional instruction condition as control group. Public schools are not stratified at this stage. Second. that is. students were instructed in reading comprehension by their regular teachers in German language lessons. metacognitive practices into the training of cognitive strategies (Elliott-Faust & Pressley. Klingner et al.and follow-up test as well as reading comprehension assessed with near-transfer task.to sixth-graders from two elementary schools serving middle-class neighbourhoods in a mediumsized German town. Complying with earlier research highlighting the potential benefits of incorporating self-reflective. comparing RT and RTP conditions between them. in the IG condition (Hypothesis 2).. posttest. The schools had no obligation to participate in the study. instead of formulating a hypothesis describing differences between the two RT conditions. and the implementation of the reading intervention was completely voluntary. outcome measures included both experimenter-developed task and standardized comprehension test. Follow-up test was conducted 12 weeks after the posttest. questioning. as indicator of socioeconomic status we asked the children how many books their family had at home. Students of the intervention conditions were taught by instructional assistants in groups of 4e6 students.. the total reading instruction time was comparable across conditions. 1998. Third. To control for instructional time. without a special profile. Barbosa. The hypotheses guiding this investigation were as follows: First. as regards far-transfer tests. we expected that the two conditions are similarly effective in fostering comprehension strategies and reading comprehension. (b) the maintenance of strategies across time (use of reading strategies at follow-up test). In the control condition. Hence. drawing on reading comprehension models (Cromley & Azevedo. and follow-up test design. teachers of intervention classes provided no reading instruction during the course of the training. whereas the other school was assigned to the intervention. respectively. 1986. in each intervention class instruction was provided in small groups after regular lessons by graduate students. Method 2. Therefore.N. 2003). text passages are longer and organized in a topic-sentence-and-supporting-detail format.and posttest materials were administered one week before and after the intervention. In the RT condition eight groups (with a total of 42 students. there were no significant differences ( p > . The use of the four strategies: clarifying. one group per class) were instructed. Fuchs et al. Second. that is. Spo ¨ rer et al. Palincsar & Brown. First. So students of all ability levels are instructed. to examine the level of generalization. Finally. Participants e design Participants were 210 third. we expected the combination of strategy instruction with RT to be more successful than the control condition and the condition in which students were taught strategies for reading in the absence of RT. Therefore. experimenter-developed task (Hypothesis 1). According to independent ANOVA and chi-square tests. summarizing. both an experimenter-developed task as indicator of near transfer and a standardized test as indicator of far transfer were administered. In contrast. both RT and RTP conditions should be especially effective in terms of far transfer. So. and predicting was assessed to analyze if differences in reading comprehension could be accounted for by differences in students’ strategy acquisition. 2007) we assumed that RT effects on standardized reading comprehension tasks (for transfer tasks) would be mediated by the extent to which reading strategies were used correctly at the end of the intervention (Hypothesis 3). Students were randomly assigned to the different conditions in two steps. / Learning and Instruction 19 (2009) 272e286 275 We applied three criteria to evaluate the four conditions: (a) the effectiveness of the conditions (use of reading strategies at posttest). depending on class size. drawing on previous reading comprehension research (Brown et al. even though control students received reading instruction during regular lessons. we randomly assigned students of each intervention class to the three different intervention conditions. The study involved a pretest.. experimenter-developed comprehension tasks may be easier to answer because compared to standardized tests. we predicted that compared to the control condition. answering experimenter-developed questions usually requires less background knowledge and searching of the text. . Most of the children (47%) indicated that their family had 26e100 books at home. while intervention students received after their regular lessons. Their ethnic identification was predominantly (97%) Caucasian. Both schools are public half-day schools. 2004. Pre. Finally. the three intervention conditions would be more effective in fostering the acquisition of reading strategies at post.05) between conditions in demographic data (see Table 2). one group per class) were instructed in the IG condition. 14 groups (60 students.

She first read the title of the passage and made a prediction about the content of the text. Initially. concepts. Spo ¨ rer et al. The first lesson served to familiarize students with the instructor. The third lesson was devoted to instruction of question generating and to recapitulate the strategy prediction. Explicit teaching was chosen as instruction form (Palincsar et al. and working out behaviour rules during training lessons. students in the three intervention conditions received the same collaborative. modelling. 1987). In the second phase of the training. or both. In the next two lessons summarization and clarification were practiced.2. In the first six training lessons students were introduced to the four strategies. that may be obscure.2. and questions that require inference. At the end of the explicit teaching phase. students were taught the clarification strategy by identifying either words. she modelled how word meanings or confusing passages could be clarified and how the paragraph could . ambiguous. Reading strategy instruction Reading strategy instruction was delivered in 14 lessons (two lessons per week) each consisting of a 45-min lesson. and (d) invent the main idea when author does not provide it (Kintsch & van Dijk.1. In the fifth lesson. using worksheet activities led by the instructor. (b) combine similar ideas into categories. and instructional materials for the three treatments are available from the first author upon request. This was done by discussing why reading strategies are important and which strategies students already knew. outlining the upcoming training lessons. (Complete reading passages. or hard to understand. Here. students practiced applying strategies to reading passages through different forms of teaching which are described next. Furthermore. interactive. students practiced collaboratively as well as independently the four reading strategies and received a bookmark with the most important information about the four reading strategies. / Learning and Instruction 19 (2009) 272e286 Conditions RT RTP 60 22 38 16 16 18 10 49 11 6 10 23 5 16 IG 42 16 26 11 10 12 9 35 7 2 6 19 9 6 Control 66 31 35 18 18 18 12 59 7 1 7 38 9 11 Total 210 85 125 57 55 57 41 180 30 9 31 99 30 41 Number of participants Sex Male Female Grade 3 4 5 6 Language primarily spoken at home German Other language Books at home 0e10 11e25 26e100 101e200 More than 200 42 16 26 12 11 9 10 37 5 0 8 19 7 8 RT ¼ reciprocal teaching. RTP ¼ reciprocal teaching in pairs.) 2. this strategy was described as a means of determining what might be discussed next by the author in order to help the reader to think about what he or she already knows about a topic in preparation for what might be coming next. IG ¼ instructor-guided reading. Characteristics N. students were introduced to the strategy of making predictions. questions that compare and contrast.. Control ¼ control condition (traditional instruction). (c) state the main idea when the author provides it. Following Hart and Speece (1998). Intervention conditions 2.276 Table 2 Participant characteristics by conditions. Traditional reciprocal teaching During the second part of the training (Lessons 7e14) students in the traditional reciprocal teaching (RT) groups received the intervention described by Palincsar and Brown (1984).2. After reading aloud the first paragraph. questions about the main idea. lesson plan. students were informed about the types of questions teachers might ask: questions about details. and practicing the strategies. questions about cause and effect. one by one. In the first phase of the training. The following rules were used to teach summarization: (a) delete minor and unimportant information. In the second lesson. the instructor demonstrated how the reading strategies were to be applied to a paragraph. In the sixth and concluding lesson of the explicit teaching part. 1978).2. students received folders for storing materials and passages. students had acquired knowledge about the four strategies but still had not applied the strategies to longer reading passages. 2. and scaffolding instruction in three stages of strategy instruction: discussing.

Beebe-Frankenberger.2. although our strategy interventions were time-based. control students were instructed in reading comprehension by their regular teachers in two German language lessons per week with traditional instruction. he or she was asked to identify words the meaning of which was unclear. we informed them that all conditions would be effective in fostering reading comprehension. Then. MacMillan. Students wrote down these words on a worksheet.N. asked other students to help. Instructors were required to model each lesson until they demonstrated a high level of proficiency in modelling strategies. The instructor decided which of the students applied a strategy. 2000). the leader asked the other student to generate a prediction. To facilitate reading comprehension. & Bocian. Strategies such as activating background knowledge. Treatment integrity was assessed over time and by lesson (Gresham. encouraged students to proceed. / Learning and Instruction 19 (2009) 272e286 277 be summarized. The task of the instructor was to monitor the dialogues of the pairs and to provide assistance upon request. tried together to clarify the words’ meanings and. She then formulated 2e3 questions and predicted what might come next in the text. the IG students were never assigned the role of the dialogue leader. In Lesson 9 students started working in pairs and used reciprocal dialogues to practice strategies. Four months before the first training lesson the six assistants met with the first author. the instructor stated that it would now be a student’s task to lead the dialogue. The leader provided feedback and together the two students discussed which question they wanted to write down. The instructors were randomly assigned to student groups. Then. fading instructional support. Instructor-guided reading As in the above-described conditions the instructor modelled how to apply the strategies to a text paragraph.3.2. Students kept on practicing reciprocal dialogues in pairs until the end of Lesson 14. At this stage. it was ensured that at the end of the explicit-teaching phase. Again the leader provided feedback and together the two students formulated the summary and noted it on the worksheet. a summary is shorter than the paragraph. the two students switched their roles. the dialogue leader asked the student to summarize the paragraph. the dialogue leader asked the other student to formulate a question a teacher might ask. 2. Similar to the traditional RT condition the instructor modelled how to apply a strategy to a paragraph. Next. Each instructor received a manual describing in detail the strategies. and praising them for good work. why don’t you first state the main idea of the paragraph?’’). materials. if necessary. In each of the remaining lessons she guided students’ activities by using prompts (‘‘What question did you think a teacher might ask?’’). and modifications (‘‘If you are having a hard time formulating a summary. the instructor explained the function of praise and feedback and helped the dialogue leader to formulate appropriate comments.4. At the beginning of Lesson 9. and instructions to be taught and assigned to students in each lesson included in the respective intervention condition. gave feedback and wrote down what they thought the best prediction would be. Different from the traditional RT condition. Spo ¨ rer et al. clarifying. providing temporary guidance to students. Different from both RT and RTP students. Finally. Next. and gave praise and feedback. all students had completed the assigned activities. Initially. 2. Hence. In their logs. Reciprocal teaching in pairs In the RTP condition students were taught in Lessons 7 and 8 how to apply the four reading strategies. exercises. 2.3. instructors had no guidelines how many paragraphs students should read per lesson. instructors encouraged students to apply the four strategies and discuss each . and predicting were taught implicitly as appropriate to the text. For the second phase of the training. the instructor requested one student of each pair to lead the dialogue in the first paragraph. as students were encouraged to provide instructional support for each other. the instructor asked a student to apply one of the four strategies and provided praise and feedback.4. Traditional instruction consisted of an extensive amount of text interaction with age-appropriate reading materials. instructions (‘‘Remember. instructors checked each step of a lesson as it was completed and jointly discussed intervention progress in weekly staff meetings. In the first six lessons of the three strategy instruction conditions 100% of the steps were completed. Then she asked a student to apply one of the strategies. Instructional assistants and treatment integrity Instructors were six female graduate assistants who had gained in pilot work extensive experience in teaching reading skills to elementary-school children. Every instructor taught at least one group in each of the three intervention conditions. Whole-class reading as well as reading in small groups was used for practicing reading. and praised students for the correct use of a strategy. After the other student of the pair had read aloud the first paragraph of the passage. the instructor demonstrated how to apply the reading strategies to a paragraph. 2. encouraging children. The pair discussed if the prediction came true and started to apply the strategies to the paragraph. To prevent instructors from creating their own expectancies concerning the differential effectiveness of the three intervention procedures. Control condition During the course of the training. asked the instructor for help. During the first phase of the training (Lessons 1e6) estimated time of activities was compared with instructors’ needed time. Students kept on practicing reciprocal dialogues until the end of Lesson 14. pairs recorded their reading activities on worksheets.’’). providing criterion-referenced feedback. For this purpose.

These worksheets were originally developed for English speaking students by Brady (1990) and were translated by Demmrich (2005) for a training study with German elementary-school students. controlled practice for each measure. Training materials For the first phase of the training. Measures 2. respectively. The length of the passages varied between 254 and 481 words in length (Grade 3: M ¼ 255. posttest. the six female graduate assistants. After students had read the passage they were asked to identify words or concepts that might need some further clarification. received the same passages in the same order (one passage per lesson). To avoid overload each student received a reading set consisting of two out of four passages.05). 2. On the second page of the questionnaire the whole passage was printed. None of these passages was assigned to any of the participating students during the instructional period. Grade 6: M ¼ 465. Finally.6. and school. student. A 3-h training session was conducted including the presentation of procedures. Passages for third. For each grade different testing passages were selected from German workbook and magazine sources. and independent scoring of each measure. four passages with 10 reading comprehension questions each were developed and administered to 106 third.7. comprising topics of age-appropriate science and social studies.5. Grade 4: M ¼ 311. Pairwise contrasts among within-subjects means revealed no significant differences between reading comprehension scores ( p > . All passages were visibly divided into paragraphs. with the teachers of the participating classes we discussed each passage and questions and they judged the finally selected passages and questions to be equivalent in interest value. respectively. Then they were requested to generate questions. Spo ¨ rer et al. For each grade. intervention condition. Interrater reliability (Pearson’s coefficient) was computed for each measure and testing period. SD ¼ 1.1. SD ¼ 40. Instructors noted in their logs of Lessons 7 to 14 how many paragraphs were read by groups and pairs. students were asked to write a summary about the passage. Scorers were trained to ensure reliability and accuracy in each measure. SD ¼ 44). which contained the explicit teaching of the four reading strategies. SD ¼ 39). Eight reading passages for third. each questionnaire was assigned a code number so that the scorers. Furthermore.7. / Learning and Instruction 19 (2009) 272e286 paragraph in detail. Students were not permitted to use any external aid throughout the testing sessions. for each grade. To ensure that passages and comprehension questions were equal in difficulty passages were tested in the following way: Four months before the training started. rubrics for scoring each measure. Since the pilot testing ensured equal difficulty across texts. the reading passage sets were presented in the same order for each student across the three measurement points. students were first asked to read a passage. Grade 5: M ¼ 423. Passages for fifth. and follow-up test. different passages were administered at pretest. SD ¼ 5. Students were asked to read the paragraph and to write down how the text might continue. Each expository passage was chosen on the basis of its possible appeal to a diverse student population. On the basis of students’ reading comprehension scores three passages with nine questions each were chosen for the training study’s pretest. Written measures were used to assess students’ acquisition of reading strategies as well as reading comprehension.to sixth-graders from other classes. Strategy acquisition Experimenter-developed assessments of strategy acquisition were modelled after Hart and Speece’s (1998) study in which students were asked to apply all four strategies to a passage.278 N. Scoring Before scoring students’ responses to the open-ended questions.and sixth-grade students. students’ reading comprehension skills were assessed with a standardized scholastic achievement test. Each of the paragraphs consisted of at least three sentences building a meaningful unit of the text to ensure that a summary could be made. Each student received a bookmark that depicted the name and a symbol of each of the four reading strategies.and fourth-grade students and eight passages for fifth. In each class one of the six research assistants collected data in whole-classroom arrangement.and fourth graders and fifth. After a 10-min break. Third. To assess students’ ability to make predictions. 2. 2. For each measure 20% of the assessments were randomly selected for a reliability check and independently rated by a second scorer.and sixth-graders. 2. . were selected for student practice of reading strategies during the second phase of training. worksheets for each strategy were handed to the students. apply all four reading strategies step by step and answer a number of comprehension questions. Procedure Testing sessions lasted 90 min. only the title and the first paragraph of the text were printed on the first page of the questionnaire. Subjects and texts for reading passages were obtained from workbook and magazine sources.1. At each session. would be unaware of the testing session. posttest and follow-up test. SD ¼ 14).and sixth-grade students consisted of between 236 and 368 words in length (M ¼ 330.6.and fourth-grade students consisted of between 179 and 312 words in length (M ¼ 260.

4 (a question based on a main idea using own language). Students were given 16 min to answer 20 (Version A) and 18 (Version B) multiple-choice questions. 2 (a question of detail using own language). / Learning and Instruction 19 (2009) 272e286 279 These assessments were evaluated by a scorer on a 6-point scale adopted from Hart and Speece (1998) which was specific to each strategy (available from the authors). minor details are included). Example item was ‘‘How much fun did you have?’’ Cronbach’s alpha for the internal consistency of the eight items was . Consequently.3. 1 (a question of detail using a sentence from the text). Preliminary analyses and overview of statistical procedures The unit of analysis was each student’s individual score. students responded to eight items reflecting their motivation and involvement in the training as well as their enjoyment of working in a group. These self-constructed tasks were administered as follows: After students read the passage and applied step by step all four reading strategies to the text (see Section 2. Social validity At posttest students were asked questions regarding their own perception of the effectiveness of the intervention. respectively. For each testing time.74 at pretest and . no minor details are included. Consequently. 2 (a response based on an unimportant detail of the paragraph).85. groups of students were not used as unit of analysis. at pretest and follow-up test all students completed Nauck and Otte’s (1980) reading comprehension test from the ‘‘Diagnostischer Test Deutsch’’ [Diagnostic Test German]. 1 (a response but untied to the passage). therefore.68 at follow-up test. posttest. the text was removed. comparison. 2. Performance on this test was in standard scores. we administered the one test form at pretest and the other at the follow-up test to assess transfer effects of strategy instruction. no minor details are included.84. Predictions were evaluated as follows: 0 (no response). Specifically. Results 3. 3 (a difficult word). 2 (a meaningful answer covering the main idea of the text paragraph) and.7. does not quite capture the gist of the passage). When a student indicated that s/he had finished reading. This test was more difficult than the experimenter-developed task because passages were not organized in a topic-sentence-and-supporting-detail format and answering the questions required re-reading the text and combining different sentences. control students did not. Cronbach’s alpha was . 2. the range of responses was 0e15. 2 (inclusion of topic sentences as well as of invented sentences.7. Each test form consists of two texts: Version A comprises a letter from a friend of 146 words and a short story about wasps of 203 words. Reading comprehension Students’ performance in reading comprehension was measured with both experimenter-developed task and standardized test. and follow-up test. evaluative.1). 1 (only topic sentences from the text are used. Clarification was evaluated as follows: 0 (no response). 1 (a meaningful answer covering details of the text paragraph). 4 (a response based on two or more features of the paragraph). Nauck and Otte (1980) reported the correlation of students’ reading comprehension scores in this standardized test inventory with a measure of fluid intelligence to be .7. . They indicated their response to each item on a 1 (no/not at all) to 4 (very much) point scale. with M ¼ 50 and SD ¼ 10.2. some minor details are included.88.1. where difficult meant that a word or concept was neither directly nor indirectly explained in the passage. interscorer reliability was >. 2 (a concept whose meaning is stated in the text). 4 (a difficult concept). 5 (invented sentences are used.35. interscorer reliability was >. Summaries were evaluated as follows: 0 (no response). In addition. 3 (invented sentences are used. Questions were evaluated as follows: 0 (no response). For each testing time. For fourth. Each of the latter questions was scored as follows: 0 (incorrect answer). At pretest. 5 (an inference.to sixth-graders. students’ reading comprehension skills were tested with experimenter-developed tasks. students were asked to respond to three open-ended questions: (a) What have you learned since we started working together? (b) What did you like most in the lessons? (c) What would you change in the lessons? 3. interscorer reliability was >. For each testing time. complex answers. Version B comprises a fable about a farmer of 142 words and a fable about a bishop of 236 words. 3 (a response based on one feature of the paragraph). Furthermore. Spo ¨ rer et al. 5 (a difficult word and a difficult concept). 5 (a response based on two or more features of the paragraph clearly demonstrating the link between the paragraph and what probably would come next). 1 (a word whose meaning is stated in the text). interscorer reliability was >. Three questions were designed to tap single pieces of information about the text (scored with one point for each correct answer) and six were designed to cover main ideas of the text and evoke longer. For each testing time. interscorer reliability was >.N. minor details are included). or cause and effect question). they were requested to indicate that they had finished by raising their hands.91. This standardized reading comprehension test has two parallel forms. does not quite capture the gist of the passage). completely captures the gist of the passage). Students were then presented with nine comprehension questions. 3 (a question based on a main idea using a sentence from the text). For each testing time. Although students in the intervention conditions worked together in fixed groups.85.85. 4 (invented sentences are used.

40 À. We were also interested in examining the extent to which potential differences in the instructional effectiveness of our research assistants might have influenced the outcomes of each of the interventions.44 .03 . Correlations >.18 .08 À. Reading comprehension (ED) Follow-up test 12. we computed partial eta-squared as a measure of the variance accounted for by intervention condition in the dependent variable of interest.06 À. & Sheets. and follow-up test measures were calculated (see Table 3). effect sizes. Cohen. West.34 .30 . Measures Pretest 1.15 .16 À.02 .26 .35 À.05).01 .05 . we conducted the Sobel test (Sobel. a relationship was detected in the bivariate correlations of strategy acquisition scores and reading-comprehension scores. Clarifying 2. for three reasons we used an analysis of covariance (ANCOVA) approach of posttest and follow-up test data: (a) to reduce the probability of a Type II error. 2002).01 . No statistically significant differences were found (each p > .46 . Predicting 11.20 . In keeping with our observations from the treatment fidelity check.32 .15 À.38 À.42 .01 .14 . standard deviations. these variables are not discussed further. Significant ANCOVAs were followed by Bonferroni-adjusted pairwise contrasts among between-subjects means with the appropriate pretest as the covariate.30 .24 . These effects are reported in standard deviation units (Cohen’s d.30 . Bonferroni correction of alpha level was applied and alpha was set at .05).22 À.30 .29 À.28 . Pretest scores did not interact with intervention condition ( p > .23 .23 .19 À.42 .25 .01 .280 N. Reading comprehension (ST) Posttest 7.47 . First.25 . adjusted means.34 . Hoffman.35 . using posttest and follow-up test scores as the dependent variables and the respective pretest as the covariate. Means.01 À. 1988). 1982). For this purpose. Clarifying 8.44 À. Student gender and grade did not modify any of the inferential analyses reported below.06 .36 . Reading comprehension (ED) 6.02 . Summarizing 3.37 .15 À.45 .29 .35 À.04) on RTP students’ reading comprehension score (experimenter-developed task) at posttest.41 À.00625).26 .21 . Summarizing 9. and follow-up test. For each intervention separately. correlations among pretest.01 .25 .25 .26 .10 . Questioning 15.22 .38 . Questioning 4. Summarizing 14. these results suggest that each of the three interventions had been implemented properly with a high degree of homogeneity across instructors. using condition as between-subjects factor. Lockwood.38 .28 .43 .28 .006 (.20 . To calculate effect sizes.42 .19 .10 . To test whether training effects on far transfer (on the standardised reading comprehension test) were mediated by the correct use of reading strategies at the end of the intervention regression analyses were run (Baron & Kenny.11 À.05 . all effects for instructor were nonsignificant ( p > .48 .14 . Therefore.43 .11 .02 . To estimate the practical significance of treatment effects. To further explore treatment effects on students’ reading behaviour. Except for Clarifying.04 . ST ¼ standardized test (far transfer).50 . 2000).26 .02 .13 À.07 .30 . / Learning and Instruction 19 (2009) 272e286 Scores for strategy acquisition and reading comprehension measures were tested for significant differences between conditions using ANOVAs.45 .29 .22 . Except for one instructor effect ( p ¼ .05).25 .10 .05/8 ¼ . we analyzed for each condition separately changes in repeated assessments of the reading strategies and comprehension measures.38 . Predicting 16.42 . Further. and (c) to control for variability in the pretest (Huck.09 .20 .38 . Clarifying 13.31 .29 . Although treatment differences at pretest were statistically nonsignificant. 1986).36 . posttest. a statistically based method by which mediation can be formally assessed (MacKinnon.30 . Testeretest correlations are in bold.24 . Next.05 .10 .24 .19 . Questioning 10. Reading comprehension (ST) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 .35 .28 .08 À.05).01 .05 À. Table 3 Correlations among measures of pretest.34 .26 .03 . one-way ANOVAs were conducted for each measure at pretest to evaluate differences between conditions prior to the instruction.08 .36 .08 . we adopted a procedure that paralleled the ANCOVA approach: We first covaried the pretest from the dependent variable and then used the residualized means and standard deviations to estimate the size of the effect.08 . we adopted a multiple dependentsample t test procedure preceded by a multivariate F test for the Condition  Time of Assessment effect. we thus analyzed the reading-comprehension measures in a series of ANCOVAs.09 À.12 .21 .24 À. .26 .32 .11 . Spo ¨ rer et al. Reading comprehension (ED) 17.51 ED ¼ experimenter-developed task (near transfer).01 . Because of the number of comparisons. posttest.18 . and gain t-values with Bonferroni-adjusted significance levels for each condition by measure are presented in Table 4.28 .42 À.36 .55 .38 .30 À.33 .13 are statistically significant ( p < .09 . Acquisition of predicting 5. (b) to increase power by reducing the error variance.37 .29 .29 . indicating that (a) our data met the ANCOVA assumption of homogeneous regression slopes and (b) pretest levels of reading skills did not moderate treatment effects on the dependent variables.09 .

00* 3. partial h2 ¼ .05) .39 2.70 6.47 2.36* 2.40.32 5. variations in pretest scores were partialled out. At follow-up test IG students failed to create better summaries than control students.91 .53 1.60 1.09 1. 2.86.17 ns 1.01.56 2.16. RTP 2.93 1.22 1.70 .05) 3.89 48.39 Control < RT ( p < .68* 2.32 IG 46. 205) ¼ 4. p < .40 1.80 1.29 1.29 2. standard deviations.05) 6.46 2.48 1.61 ns SD 1. IG 6.98 1.44 1.N. Except for the small effect of RTP students’ . F(3.79 ns 1.24. (c) Predicting: Similar to the acquisition of summarizing skills. F(3.03 1. and for Predicting at follow-up test. for Summarizing at follow-up test. / Learning and Instruction 19 (2009) 272e286 281 Table 4 Means.37 1.16 ns 1.85 1.45 1. RTP.06 1.77 1.52 1. for Questioning at follow-up test.37 2. partial h2 ¼ .90 Control 4. for Summarizing at posttest. partial h2 ¼ .57 . RTP ¼ reciprocal teaching in pairs.44 7. 205) ¼ 3.006 (Gains t Bonferroni-adjusted alpha level ¼ .72 1.76.39 2.05 Control < RT.84 2.09 1.24 .95 1.60 1. but RT and RTP students did.31* 3.70 .25 1.50* 2.05) 3.05 6.45 1.50 ( p < .16 2.48 1. RT and RTP students formulated better questions than control students.47 2. ST ¼ standardized test (far transfer).05 RTP 46.93 9.09. p < . all condition effects turned out to be significant.59 1. 205) ¼ 1. students in each of the three intervention conditions wrote better summaries than control students.46* 1.39 1.46 2.47 Reading comprehension (ED) RT 2.90 1.01 2.22 Adjusted M 1.59 .05) .79 .08* 3.05) .60 1.65 7. Contrast analyses revealed the following results (see Table 4): (a) Summarizing: At posttest.66 2.63 3.38 4. Control ¼ control condition (traditional instruction).46 1.10 2.03* 3.06.67 Gains t 3. for Predicting at posttest.06.37 1.58 Control < RT.21 1. At follow-up test.63 2.80 .11 Adjusted M 1. p > .41 Control < RT.13 1.01.40 1.51 RTP 3.29 1.58 1.31 .24 1.05/8).25 1.75 0.48 1.48 1.29 1. RTP. 205) ¼ 4.25* 6.72 2.90 ns SD 1. 205) ¼ 4.17 4.55 1.54 .20 1.19* 52.68 IG ( p < .63* 6.47 Follow-up test M 1.55 2.2.97 ( p < .06.59 2.00 0.95 6.15 2. IG ¼ instructor-guided reading.15 1. p < . p < .28 .66 1.62 6.25 1.87 d .12 . (b) Questioning: At posttest.01 1.38 2.01.51 1.40 1.03 IG 3.82 6.62 1.51 2. partial h2 ¼ .98 2.95 0.39 6.12 47. F(3.41 1.04 1.26 3.48 9.26 49.55 4.91 1.81 .63 2. Spo ¨ rer et al.80 IG ( p < .13. although the effect sizes were small to moderate: for Clarifying at posttest.26 9. Conditions Pretest M Clarifying RT RTP IG Control Contrasts Summarizing RT RTP IG Control Contrasts Questioning RT RTP IG Control Contrasts Predicting RT RTP IG Control Contrasts 0. and adjusted means (with pretest scores as covariate).05. p < . F(3. Before group contrasts were tested for significance.33 1.69 2. *p < .64 1.40 Contrasts ns 1. 205) ¼ 9.92* 3.48 0. for Questioning at posttest.76* 1.01.36 49. only RT students outperformed control students.79 5.96 1.12 2.63 1.66 6.42 1. Strategy acquisition We analyzed posttest and follow-up test scores of strategy acquisition with four separate one-way ANCOVAs using the pretest as the covariate.25 .90 4.14 3.49* 7. Except for Clarifying at follow up. 205) ¼ 6. F(3.45 3.18* 3.32* 3. partial h2 ¼ .74 .28 1.67 .05. RTP.27 4.54 5.67 2.44 6.92* 1.49 1. RTP. Cohen’s d ¼ effect size in standard deviation units (control vs. 205) ¼ 4.55 2.31 1.36 1.06 2.73* 6. RTP ( p < .05) 2.74 8.58 6. ED ¼ experimenter-developed task (near transfer).82 Control < RT.38 3.05) 6.46 1.77 48.44 2.65 2.87 2.80 47.26 1.66 .83 1.58 2.63 Control 45.00 0. at posttest students in the intervention conditions made better predictions than control students. At follow-up test. F(3. p < . RT students continued to outperform control students.17 7.03* 1.84* 1.18 1. F( 2.51* 1. F(3.40 1.33 3.02 .92 7.48 1.66 .01.47 IG ( p < .38.92 RT ¼ reciprocal teaching.50 1.75 0.87 ns SD 1.39 2.86 5.48 0.29 2.60 1.15 Control < RT.38 3.31 2. p < .13 1. 3.94 0.48 3.07 Control < RT ( p < .10 4.42 2.13 52.29 1.17 Posttest M 1.84 d .62 1.17 2.93 6.73 Control.26 . IG < RT ( p < . 3. partial h2 ¼ . each treatment group).09 1.32* 4.16 .63 1.31 Gains t 2.50* 0.06.19 . effect sizes (Cohen’s d ) and gains t for strategy measures by testing occasion and condition.57 2.05) 2.31 3.05.03* 4.52 2.87* 7.65 .94 Contrasts ns Reading comprehension (ST) RT 47.64 Control < RT.53.31 0.58 1. 6.22 2. partial h2 ¼ .97 2. 410) ¼ 2. p < . although the effect sizes were small to moderate: for Clarifying. Both at post.19. In three separate regression analyses. p < . into one and the same regression equation. Interaction of condition with time of assessment The Condition  Time of Assessment effect on reading comprehension scores was significant for the near transfer test.05. for Summarizing. to 5.98. Predicting: r ¼ . partial h2 ¼ . Table 4 shows the complete results of the within-subjects analyses. 206) ¼ 1. (c) Only RT students improved in their reading comprehension (assessed with the far transfer test) from pretest to follow-up test. 3. in this multiple-mediator analysis. Reading comprehension Both at posttest. p < .2. F(3. with moderate effect size. we examined if variations in far transfer test scores might be accounted for by variations in the respective strategy variables with pretest reading-comprehension scores controlled for.05) correlated with the far transfer test scores (Summarizing: r ¼ . F( Models 2e4).and at follow-up tests. The multiple dependent-sample t tests yielded the following pattern of results: (a) At posttest. all follow-up test scores of strategy acquisition (a) were significantly ( p < . Wilks’s lambda ¼ .and follow-up tests. Finally. p < .42. and control students (Cohen’s d ¼ . the variance accounted for by the dummy-coded treatment factor in students’ reading comprehension declined from 7. F(3.001. At post. we adopted a hierarchical regression approach using a dummy variable to code intervention condition (0 ¼ control and IG. F(3. p < . F(6.4% but still remained significant (see Table 5.09. partial h2 ¼ . Wilks’s lambda ¼ . 205) ¼ 6.3. 1 ¼ RT).92.01. RT students improved in all strategy acquisition measures from pre. The effect of condition on reading comprehension as assessed with the standardized readingcomprehension test (far transfer) at pretest and follow-up test was significant.07.07. Summarizing. partial h ¼ . (b) Intervention and control students displayed a significant change in the near transfer reading comprehension measure from pretest to follow-up test. partial h2 ¼ .20. p < .282 N. In doing so. F(6. Across testing periods. partial h2 ¼ .and post-/follow-up-test scores.to posttest and surpassed their pretest scores at follow-up test. Wilks’s lambda ¼ .01. the effect of condition with the pretest as the covariate on the experimenter-developed task (near transfer) was significant. Questioning: r ¼ . for follow-up test.90. 3. p ¼ . So.57).55). 3. 410) ¼ 3. for Predicting.1.006): Except for Clarifying at follow-up test. for Questioning. This pattern of results met the conditions required by a mediational analysis according to the principles specified by Baron and Kenny (1986).14. The control students did not display a significant change in strategy acquisition neither from pre. excluding RTP students because they had failed to outperform control and IG students on the far transfer test. partial h2 ¼ . Spo ¨ rer et al.4. 2 3. students in each of the three intervention conditions improved their reading comprehension as assessed with the near transfer test compared to the pretest p ¼ .09. 410) ¼ 6.006. but not for the far transfer test. p ¼ . Pairwise contrasts revealed that RT students outperformed IG students (Cohen’s d ¼ . F(3. Notably.82. and Predicting each had a (nearly) significant ( p . and at follow-up test. students in each of the three intervention conditions outperformed control students. adding strategies into the model reduced the variance accounted for by treatment by 29%. 205) ¼ 11.to posttest nor from pretest to follow-up test. / Learning and Instruction 19 (2009) 272e286 questioning skills. all p < . ns. F(6. the magnitude of the effects reflecting the superiority of the intervention conditions relative to the control condition was large. Mediational analyses of far transfer test Except for Clarifying. Summarizing constituted the only strategy that had a significant unique effect on changes in reading comprehension from pretest to follow-up test. The quality of RTP students’ strategy acquisition improved from preto posttest but at follow-up test only the quality of their summaries and predictions was significantly better than at pretest. 410) ¼ 5. 410) ¼ 2.43. Therefore. When the three aforementioned strategies were entered simultaneously.to posttest but relapsed at follow-up test (except for Predicting).32. Wilks’s lambda ¼ . as a set. significant differences between conditions were not found at either posttest or follow-up test.86. Interaction of condition with time of assessment The Condition  Time of Assessment on strategy acquisition scores in the mixed design ANOVA with Condition as between subjects factor and Time of Assessment as within subjects factor were significant in the following cases. we used the formula given by Sobel (1982). p < . all effect sizes were in the medium to upper range. Wilks’s lambda ¼ . mediational effect on the relationship between intervention condition and reading comprehension at follow-up test (see Table 5. Wilks’s lambda ¼ . 205) ¼ 5. IG students’ strategy acquisition improved from pre. Questioning. F(6. For Clarifying.001.001.03. p ¼ . To test the statistical significance of the hypothesized mediational relationships. Model 5). .30) and (b) were reliably predicted by intervention condition. partial h2 ¼ .22. within-subjects analyses yielded the following results (see Table 4 for t-values comparing pre. partial h2 ¼ .

The associated effect sizes were large and exceeded the effect sizes for near transfer measures of reading comprehension reported in the reciprocal teaching literature (Rosenshine & Meister. 146) ¼ 22.36 3.001 (5.5.30 Model 3 b .21** .001 (1.06 (moderate effect size). p < . SD ¼ 0. students in all intervention conditions were asked to evaluate the intervention.67 3.24** .59. Most of them (65%) stated that they had learned reading strategies.32 (3. Furthermore.36.36 2.52) and IG students (Cohen’s d ¼ . Spo ¨ rer et al. Predictors Model 1 b Pretest RC Condition Summarizing Questioning Predicting R2 F . 3. intervention condition.32 2. IG) ANOVA was conducted.58 . Condition dummy was coded (0 ¼ control and instructor-guided reading. 146) ¼ 23.001 DR2 . ***p < . Social validity At posttest. 1994).12 3.08 DF (1. The magnitude of the means reflecting how much students liked the intervention was large for all three conditions (for RT students.47.73 Model 4 b . M ¼ 3. 6% of the students did not like the training at all. SD ¼ 0.N.24** . the atmosphere during the lessons (25%).01. students had to respond to three open-ended questions.44. F(2.97 Model 2 b . First. p < .001 .96 0. questioning.001 . 1 ¼ reciprocal teaching). According to chi-square tests. we asked students what they had learned in the lessons. M ¼ 2. 146) ¼ 6. Clarifying was not added to the regression analysis because it was not significantly correlated with the standardized reading comprehension measure and hence did not meet the conditions required by mediational analysis (Baron & Kenny. students from all intervention conditions found their training useful for improving reading comprehension.44.05.35*** .22. In summary.54.08 . SD ¼ 0.00. for RTP students. Medium to large effect sizes were obtained when RT students were compared with control students. Control students showed very limited improvement in their reading comprehension over the course of this study. p < . relative to control students.99. p < .33*** .07 . There was a statistically significant main effect. 1986). 147) ¼ 30.09 . p < . Only 9% liked nothing at all. 25% answered that they had learned how to understand difficult texts and 10% indicated that they learned nothing at all.29 (2.39*** .28*** T 6. 4. RTP and IG students outperformed control students at posttest but did not maintain their superior performance at the follow-up test. First. / Learning and Instruction 19 (2009) 272e286 283 Table 5 Hierarchical regression analysis with the follow-up standardized reading comprehension scores as dependent variable and the pretest standardized reading comprehension scores (Pretest RC). and predicting when reading a text at the posttest.67. compared with the control students. and strategy acquisition as predictors.53 4.09 2. and cooperative learning and being the tutor (16%). In the final question.80. A single factor (condition: RT.01 (3. Finally. students replied that they liked reading different texts and applying the strategies (50%). . 141) ¼ 4.25. p < .05 (3.19* .24** T 5.37 (3. Second.58 3. students in the three intervention conditions scored higher on the near transfer test of reading comprehension both at the posttest and at the follow-up test. p < .001 T 5. partial h2 ¼ . for IG students.71) when compared with RT students. p < . students were asked to tell what they would change in the lessons. 144) ¼ 6. When asked what they liked most in the lessons. From 144 students 92 responded to these questions.24** . 146) ¼ 18.03 .001 *p < .38 Model 5 b .39*** .40). Discussion The main objectives of our study were (a) to investigate the effects of explicit instruction of reading strategies on third. 144) ¼ 17. 146) ¼ 28.38 T 4. RTP.43 3.60. **p < . p < .30*** T 5. students in the intervention conditions were better able to use the strategies of summarizing.93 1.95.001.to sixth-graders’ strategy acquisition and reading comprehension achievement and (b) to study the differential impact of practicing the strategies in reciprocal small group (RT) and pair (RTP) activities as compared to instructor-guided (IG) activities and traditional instruction (control group). while 33% did not like working together with a particular classmate and 12% did not like to fill out worksheets.44*** . there were no significant differences between groups in students’ responses.05 (1. The following three major results emerged from this study. 29% responded that they would change nothing in the lessons. 20% replied that they did not like a particular strategy or text and thus would leave it out. Of them. Furthermore. M ¼ 3.03 . p < . 146) ¼ 7.01. Pairwise contrasts revealed effects in the medium range for RTP students (Cohen’s d ¼ .92. Our findings indicated that especially RT students who practiced strategies in reciprocal small group activities benefited in the short as well as the long run from training lessons.20** .

In the present study. However. Researchers and practitioners are in agreement that strategic processing of text is critical to reading comprehension (Cromley & Azevedo. students who participated in one of the three intervention conditions showed near transfer in the sense that they reached higher reading comprehension scores as assessed with the experimenter-developed task. however. Compared to IG students. First. to ensure the ecological validity of small group procedures.284 N. 1998). a main aim of our study was to find out which strategies are most effective in improving reading comprehension. Hypothesis 2 was partially confirmed. Third. Fuchs. Hacker and Tenent’s (2002) qualitative research in mainstream classes showed that the observed elementary-school teachers encountered different obstacles while implementing and practicing RT in their classes and therefore made many modifications to strategy use and teaching. Thus. 1984.. but worthwhile. With respect to RT students’ growth in reading comprehension. there was a positive improvement of performance in the far transfer test only for RT students when compared with IG and control students. students from all intervention conditions found their training useful for improving reading comprehension. questioning. working in pairs was more a completing of given tasks than a lively discourse. It turned out that RT but not RTP students improved in the standardized reading comprehension measure from pretest to follow-up test.. 1995). summarizing) and comprehension-monitoring (i. unfamiliar texts. Furthermore. RT students had learned to summarize text paragraphs more concisely and this growth in strategic reading helped them to better understand difficult. research on cognitive strategy instruction suggests that peer support procedures are particularly effective in stimulating metacognitive activities while students work together and share their ideas. why did students who practiced reciprocal teaching in small groups (but not students in pairs) perform better after the training lessons? A possible explanation is that RT students took greater benefit from small group activities because they got more room for discussing a paragraph and exchanging their ideas. Hypothesis 3 was confirmed. These results add to the extant knowledge about strategy instruction and reciprocal teaching (Palincsar & Brown. RT students.1. However. However. RT students had the opportunity to lead the dialogue. Although this far transfer test was administered not immediately but 12 weeks after training. we could show how the RT strategies contribute to reading comprehension. Spo ¨ rer et al. no such relation was found for clarifying. RT groups showed lasting effects of strategy acquisition. the instructional approaches need further refinement so that they can be implemented by teachers in naturally constituted classrooms.e. & Karns. Whereas a strong relationship existed between reading comprehension and the strategies of summarizing. Limitations and implications for future research Before closing. it should be taken into consideration that . For instance. and predicting. the role of the teacher in encouraging students to provide instructional support for each other. Furthermore. Not only in the domain of reading but also in the domains of writing (Harris. reported the lowest social validity scores. communication is naturally limited to two students. 4. Relative to IG and RTP counterparts. we assumed that students taught to monitor and regulate their reading behaviour through reciprocal teaching (RT and RTP) would display superior performance in the standardized reading comprehension measure compared to IG students. the pairs filled out their worksheets systematically. and mathematics (Fuchs. Before arguing that clarifying is a less effective strategy. the obtained effect sizes were larger than the median effect size (. 2006). discussing and writing at the same time might be too difficult to handle. generalization. several limitations of our study should be acknowledged. Graham. & Mason. A mediational analysis revealed that differences in the far transfer test at the follow-up test were accounted for by differences in students’ acquisition of the comprehension-fostering strategies of summarizing. questioning. and follow-up test. In addition. Given that for elementary students writing is very challenging. So.32) reported by Rosenshine and Meister (1994) for reciprocal teaching interventions. such as monitoring and regulating students’ comprehension (Zimmerman. therefore. be combined in comprehension instruction to achieve lasting effects of reading instruction. questioning) strategies and the application of the strategies using reciprocal dialogues in small heterogeneous groups were identified as effective elements of RT. and should. In pairs. The findings corroborate the view that for elementary students traditional RT is challenging.e. 2007). Phillips. Thus. As the research assistants observed. / Learning and Instruction 19 (2009) 272e286 Hence. Hypothesis 1 was confirmed. no significant correlation could be obtained between clarifying and the other three strategies. On the contrary. reading without writing may have made it easier for the dialogue leader in RT groups to give full attention to metacognitive skills. our findings supported the viewpoint that especially summarizing skills play a central role in mediating the effects of the RT method on the improvement of reading comprehension performance. Hamlett. and predicting. only students who practiced reciprocal teaching in small groups showed far transfer in the sense that they got higher reading comprehension scores as assessed with the standardized test. Rosenshine & Meister. Both the instruction of the comprehension-fostering (i. Component analysis of RT and RTP procedures could provide insight into the unique contributions of reciprocal teaching and writing components to learning. is to be emphasised. Compared to IG and control students. 1994) in the sense that the relative advantage of RT indicates how multiple strategies can. To summarize. Second. More research is needed to identify instructional settings and techniques that are both effective and feasible when strategies for fostering reading comprehension are to be integrated by teachers into the daily routine of classroom lessons. the findings of the present study confirmed the efficacy of explicit reading instruction as a feasible tool to enhance students’ reading comprehension.

S. In this case. et al. we assessed the acquisition of clarifying skills by asking students to note words or concepts that needed clarification. L. Alvermann. Accessed 28. Unpublished doctoral dissertation. strategic. Following Hart and Speece (1998)... R. A. & Simmons. Barbosa. The moderator-mediator variable distinction in social psychological research: conceptual. P. M. (2004). L. B.N. (2004) found out. Journal of Educational Psychology. it is important to mention that after a relatively short time of reading comprehension instruction students had become self-regulated readers. et al.and long-term listening comprehension monitoring.02. Testing and refining the direct and inferential mediation model of reading comprehension.. T. an instructional framework combining motivation support and strategy instruction is more successful in increasing students’ reading comprehension than strategy instruction alone. M. Guthrie. see Guthrie. N. W. Metacognitive strategies for development of reading comprehension for younger children. E.. Demmrich.. Owen. Borko.. D. and high school. 24. H. Journal of Educational Psychology... (2005). no response (coded with 0) could represent that a student felt no need to clarify a word or concept and. 51. MacMillan. J.. providing scaffolding for the four strategies. & Earle. Calfee (Eds. Wigfield. M. it is worthwhile because this form of reciprocal teaching was the most effective. Burch. M. Improving reading comprehension by enhancing metacognitive competences: an evaluation of the reciprocal teaching method. K. Hamlett.. Peer-assisted learning strategies: making classrooms more responsive to diversity. A. D. J. Fuchs. Fuchs. Dole. D. D. Perencevich. J. & R.. & C. P.. Dermody. Fuchs.. R. 78. (1995). 15e21. Davis. In A. / Learning and Instruction 19 (2009) 272e286 285 implications might be limited by the measurement of the acquisition of clarifying.g. Berliner.07. et al. (2003). D.. Armbruster. Brady. Mathes. Valencia. School Psychology Review.. (2003). D. 142e159.. 309e332. (1996). & Ostertag.. 35. Hence. P. wrote nothing down. New York: Guilford. . Van Meter. H.. Remedial & Special Education. 26.. New York: Macmillan. it may be that during training lessons the role of the dialogue leader in RT groups was especially motivating for students and therefore increased engaged reading and reading comprehension. 403e423. Rethinking reading comprehension (pp. T. kobv. C. 174e206. & Azevedo. Increasing reading comprehension and engagement through concept-oriented reading instruction.. Although it may be challenging for teachers to practice RT in the traditional way (that is. Prentice. Phillips. A..). (2001). 673e708). Cromley. Even though motivation support was not an explicit part of the training.. G. E. Peer-assisted learning strategies in reading: extensions for kindergarten. 2004) as further potential constituents of a mediation model that could explain in greater detail than we did it here how reading strategies translate into reading comprehension. J. M. & Fuchs. L. 18e37. D. Elliott-Faust. References Alfassi.2. 306e315. A. Louisiana. Pressley. 198e205. M. M.. J. C. K. J. 2.. (2000)..de/ubp/volltexte/2005/524>. Enhancing third-grade students’ mathematical problem solving with self-regulated learning strategies. Brown. & Karns. (1986). B. 1988). J. S.. first grade. P. Fuchs. Comprehension instruction. (1996). Improving the reading comprehension of middle school students through reciprocal teaching and semantic mapping strategies. Handbook of educational psychology (pp. it might be useful to consider motivation variables (e. D. New Orleans. Fuchs.. Hillsdale. (1990). 27e33. L. In D.. C.. Treatment integrity in learning disabilities intervention research: do we really know how treatments are implemented? Learning Disabilities Research and Practice. (2007). C. F.. the results of our study corroborate the view that explicit instruction of multiple reading strategies is a feasible tool to enhance students’ reading comprehension and that third. J. Conclusion Despite these limitations. Above all. a student’s interest in reading.). 12e30). As Guthrie. Does text structure/summarization instruction facilitate learning from expository text? Reading Research Quarterly.. 96. University of Oregon-Eugene. & Putnam. (1987). Statistical power analysis for the behavioral sciences (2nd ed. American Educational Research Journal. Anderson. & Wardrop. Journal of Educational Psychology. Paper presented at the American Association of Colleges for Teacher Education.. & Kenny. L. 99. Additional measures and rating categories are needed to tap more exactly the acquisition of clarifying.. Effects of two types of prereading instruction on the comprehension of narrative and expository text. Beebe-Frankenberger. American Educational Research Journal. R. M. 4.. & Bocian. (1991). (1986). Greer. Snow (Eds. Hamlett. K. L. 22. A. thus. Learning to teach. Journal of Educational Psychology. C. (1988). Spo ¨ rer et al. & Pressley. (1998). (February. S.. C. Barbosa. Sweet.). M. R. Explicit strategy instruction and reciprocal teaching as part of the overall curriculum appear to be a promising procedure to get this process off to a good start. a lack of correlation between reading comprehension and clarifying might have been due to the ambiguity of categories. 1173e1182. Taboada. Acquisition and transfer effects of classwide peer-assisted learning strategies in mathematics for students with varying learning histories. Third. Journal of Personality and Social Psychology. H. 331e346.to sixth-graders benefited most from explicit reading instruction supplemented with practice in small groups’ reciprocal teaching activities. Fuchs. 88. and statistical considerations. M. How to teach comparison processing to increase children’s short. et al. Reading Research Quarterly. Reading for meaning: the efficacy of reciprocal teaching in fostering reading comprehension in high school students in remedial reading classes. Journal of Educational Psychology. D. A quasi-experimental validation of transactional strategies instruction with low achieving secondgrade readers. <http://opus.. NJ: Erlbaum. Wigfield. Fuchs. & Schuder. Cohen. Wigfield. Gresham. 604e620.. K. 34. 22.. R. S. T. Barbosa. applying strategies in small groups while using reciprocal dialogues). (1997). L. 15. P. Baron. A. 311e325.. G. T. L. S.

Psychological Review. Wigfield. L. M. Comprehension: A paradigm for cognition. E. Diagnostischer test Deutsch. S.. K.. Marks.. National Reading Panel.. 479e530. 22.. McMaster. Reciprocal teaching improves standardized reading-comprehension performance in poor comprehenders.. D.. D. Research on peer-assisted learning strategies: the promise and limitations of peer-mediated instruction. & Otte. (1978). J. 90. Graham. Journal of Educational Psychology.. Toward a model of text comprehension and production. 33. L. (1988). S. 99. W... Huck. C. DC: National Institute of Child Health and Human Development. Psychological Methods. Improving the writing. D.. A.. A. J. Teaching students to generate questions: a review of the intervention studies. 95. K.. M. 181e221. 7. S. J. W. Collaborative strategic reading during social studies in heterogeneous fourth-grade classrooms. W. D. / Learning and Instruction 19 (2009) 272e286 Guthrie. Reading statistics and research (3rd ed. 163e182.. de Jong. C. Sobel. Kintsch. & Brown. Paris: Organisation for Economic Co-operation and Development. 43. & Monseur. Fuchs. 73e86. Brown. David. 16.. Review of Educational Research. A. Using reciprocal teaching in the classroom: a guide for teachers. R. Vaughn. L.. J. Cambridge. R. C. Motivating reading comprehension: Concept-oriented reading instruction. Pressley. Educational Psychologist. M. S. Kirsch. A. 94. J. (1987). Hacker. 295e340. & Sheets. S. L. (1998). Pressley. Souvignier. Using self-regulation as a framework for implementing strategy-instruction to foster reading comprehension. 231e253. A. S. In S.. Reciprocal teaching goes to college: effects of postsecondary students at risk for academic failure.. LaFontaine. 85.. (2006). San Francisco: Jossey-Bass. M. D. L. Learning & Instruction. Elementary School Journal.. Mahwah. (1998). R. B. 1. Academic studying and the development of personal skill: a self-regulatory perspective. McQueen. R. American Educational Research Journal. Hoffman. . 22. Spo ¨ rer et al. knowledge. Leinhart (Ed. J. NJ: Erlbaum. (1980). T. & Mason. V. (2003). Palincsar. & van Dijk. Review of Educational Research. Implementing reciprocal teaching in the classroom: overcoming obstacles and making modifications. Kintsch. (Eds... (2006)... A. Gardner. S. (2000). & Vye. M. (1982). Reading for change: Performance and engagement across countries. S. Unpublished manuscript. K. B. (2002). & Tenent. S. Palincsar. 363e394. (2000).. [Diagnostic test German]. & Fuchs. Reciprocal teaching of comprehension-fostering and comprehension-monitoring activities. Goettingen. 57e71.. Rosenshine. 117e175. Sociological methodology (pp. M. M. Reading & Writing Quarterly. (1996). A. L. 37e58. & Chapman. J.286 N. Le Fevre. 73. & Wilkinson. B. Educational Psychologist.. (2006). I. J.. Hart. Harris. C. J. 90. 670e681. West. M. L. Washington. A.. D. Lockwood. W. (1989). P.. Moore. K.. 699e718. W. 66.. and motivation of struggling young writers: effects of self-regulated strategy development with and without peer support. Peer interaction in reading comprehension instruction. Psychological Review. J. W. J. & Perencevich. A. Elementary School Journal.. New York: Longman. 471e484. 94. 3e22. & Meister.). S. 267e283.. Meister. DePinto. Reciprocal teaching: a review of the research. A. The role of knowledge in discourse comprehension: a construction-integration model. (1990). Nauck. Elementary School Journal. Mendelovits. & Mokhlesgerami. (2002).. Rosenshine. S. Craig. D. Lysynchuk. M. J. & Speece. Cognition and Instruction. Klingner. England: Cambridge University Press. 83e104. (2004). (2002). Teaching children to read: An evidence-based assessment of the scientific research literature on reading and its implications for reading instruction. (1998). (1994). W. Palincsar. E.. Coley. 64. J. Kintsch. A comparison of methods to test mediation and other intervening variable effects. & Schumm. T.. N. C.).. & Martin.. Journal of Educational Psychology. 290e312). Three teachers’ adaptations of reciprocal teaching in comparison to traditional reciprocal teaching. (1984). Tape-assisted reciprocal teaching: cognitive bootstrapping for poor decoders. (1993). D. MacKinnon.. Germany: Hogrefe. Asymptotic confidence intervals for indirect effects in structural equation models. G. & Rose. S.). M. L. Y. British Journal of Educational Psychology. & Brown. G. Zimmerman. 5e25. (1998). I.

Sign up to vote on this title
UsefulNot useful

Master Your Semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master Your Semester with a Special Offer from Scribd & The New York Times

Cancel anytime.