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: firstname.lastname@example.org (N. Spo ¨ rer). 0959-4752/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.learninstruc.2008.05.003
and (d) predicting what might come next in the text. / Learning and Instruction 19 (2009) 272e286
such as Cromley and Azevedo’s (2007) direct and inferential mediation (DIME) model which in turn is based on Kintsch’s (1988. & Wardrop. Since Palincsar and Brown’s (1984) seminal work. For generating questions. Hence. The overall goal is to promote. 1984. 1987). No study has examined students’ mastery of clarifying strategies. In a qualitative analysis. Spo ¨ rer et al. and monitoring its accurate implementation. (c) clarifying word meanings and confusing text passages. Pressley. (b) summarizing text (Armbruster. 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. As the students in the group become more familiar with the strategies and the procedure. effects on making predictions were assessed in only one of these studies (Dermody. 2004. Brown. the following elements are essential to RT: instruction of the four comprehension-fostering and comprehension-monitoring strategies. the self-directed and ﬂexible use of the learned strategies. models the use of the strategies. Rosenshine and Meister (1994) reported a mean effect size of . Besides these open questions regarding the empirical identiﬁcation of effective strategies involved in RT. it is not clear if all or only one of the taught strategies is effective in fostering students’ reading comprehension (Rosenshine & Meister. & Chapman. Greer. who can be a teacher or a student. Although these cognitive and metacognitive strategies have most frequently been investigated in isolation. applying the selected strategy. 1990. The procedure has been applied to different settings. 1998. & Perencevich. 1984. Souvignier & Mokhlesgerami. 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. and summaries. 1998). a training effect for summarizing only could be established. 1987). ﬁve out of six studies found no reliable difference between RT and control groups. 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. At present. In a meta-analysis involving 16 studies. 1988)... There were signiﬁcant improvements in four out of ﬁve studies in which researchers collected summarization probes. & Vye. 1994). 2006). Hacker and Tenent (2002) found that elementary-school teachers made many modiﬁcations to adapt RT to the requirements of mainstream classroom instruction. and (c) generating questions to capture the main idea of the passage (Rosenshine. and providing scaffold instruction during which teachers gradually fade their modelling of the strategies (Hacker & Tenent. In this model.88 for experimenter-developed task favouring RT over control groups. dialogue leaders fade their involvement and other students take turns as discussion leaders. 1998. 1998. word reading. 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. An underlying assumption of RT is that by applying the strategies in a group process. reading strategies. Valencia. 1984). 1991). Anderson. through scaffolding instruction and collaboration.32 for standardized test and . Palincsar & Brown. Guthrie. age groups. Klingner et al. For example. for training college students see also Hart & Speece. Drawing on Zimmerman’s model. 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. Reading vocabulary and background knowledge directly contribute to reading comprehension and also have effects that are mediated by inference. Lysynchuk. These four strategies are involved in RT in ongoing dialogues between a dialogue leader and the remaining students of the learning group. Palincsar. 1998) constructioneintegration model. Palincsar. 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. The model further suggests that the effect of strategies on comprehension is mediated by inference. provides conditional knowledge about strategy use. it is thus unclear which strategies of RT signiﬁcantly contribute to the development of students’ reading comprehension skills. 1994. In addition. 2002.. & Brown. and (c) self-assessment of strategy outcome and task performance. application of the strategies using rich and meaningful reciprocal dialogues. Meister. So far. Although there is clear evidence that RT promotes reading comprehension. Rosenshine & Meister. a number of difﬁculties with implementing and practicing RT have been reported in the literature (Fuchs & Fuchs. some researchers have examined how they work together in more complex strategy packages (Brown et al. 1993). 1996). Marks et al. 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. Other teachers required their students to write down their questions. although in all six studies RT students signiﬁcantly improved in their reading comprehension relative to control students. The dialogue leader. answers.N. 2001. Moore. & Wilkinson. David. Even though students were taught the entire set of four strategies in 12 out of 16 studies. and populations (Alfassi. Wigﬁeld. that is. self-regulation procedures as described by Zimmerman (1998) are integral to RT. The DIME model hypothesizes relationships among background knowledge. especially less able students can learn from their more knowledgeable peers. & Ostertag. (b) summarizing parts of the text. such as students’
. many studies have been conducted to test the effectiveness of reciprocal teaching. and inference that together result in reading comprehension. Furthermore. To sum. during reciprocal teaching students are engaged in cognitive and metacognitive activities: they alternate between prompting the use of a strategy. (a) generating one’s own questions. Palincsar & Brown. 2003. and helps students to apply a strategy to a passage. & Martin. (1993) observed that teachers sometimes changed RT in a way that elements supposedly playing a critical role in promoting deeper levels of reading comprehension. (b) selfmonitoring of one’s accuracy in implementing a selected strategy. 1989. 1996. Marks et al. Le Fevre. Reading comprehension is correlated with a number of cognitive and metacognitive strategies. vocabulary. such as (a) activating background knowledge (Dole. Hart & Speece.
namely RT (Rosenshine & Meister. Fuchs. this may result in a cognitive overload (Rosenshine & Meister. Similar to RT. partner reading) and. IG ¼ instructor-guided reading. our ﬁrst aim was to examine if both strategy instruction and reciprocal teaching contribute to the acquisition of reading strategies and. thus. were completely dropped from the instruction. 2001. labelled reciprocal teaching in pairs (RTP condition). two students are paired to share and practice reading activities (McMaster. reading in pairs is more similar to instructional procedures teachers often adopt in reading lessons (e. thus.g. we adopted and further advanced the argument that there is a need for identifying effective elements of a multiple strategies program. & Simmons. 1997). Mathes. teachers use a set of brieﬂy scripted lessons including teacher presentations. 2006).2. 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). is more likely to be accepted by teachers (Fuchs & Fuchs. 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. students who work in pairs have more opportunities to practice the use of reading strategies than students who work together in small groups. there are two main reasons why it may be difﬁcult for teachers to implement RT in naturally constituted classrooms: First. The present study In our study. and teacher feedback. & Fuchs. Accordingly. Compared to RT. 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 small group of 4e6 students was guided by a graduate assistant (the instructor) during the course of the intervention. and predicting. a summary. Fuchs. 1994). Second. clarifying.274
N. summarizing. Later on. / Learning and Instruction 19 (2009) 272e286
continuous engagement in reciprocal teaching. and then led to continue practicing these strategies in pairs. 2006). Fuchs. 1984). questioning. 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.
Table 1 Characteristics of intervention conditions. students’ roles can be reciprocal so that both students in a pair serve as tutor during each lesson. Altogether. 2001). 1. questions a teacher might ask. 1994). In this condition. 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. facilitate structured working in pairs (McMaster et al. many teachers are unfamiliar with the procedure of reciprocal teaching. Peer-assisted learning lessons consist of a set of structured activities. 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). and predicting.. thus. At the beginning. Hacker and Tenent (2002) argued that researchers have to consider that teachers. The task of the instructor was to model the four reading strategies. need to take ownership of their learning by constructing their own understanding of new curricula and methods using their prior knowledge. questioning. Third. 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. the instructional technique for helping children to develop responsibility for strategic behaviour is challenging. First. In a third strategy condition. and students are taught to enact these activities independently. þ þ RT þ þ þ RTP þ þ þ þ
. and a prediction RT ¼ reciprocal teaching. RTP ¼ reciprocal teaching in pairs. reading in pairs has several advantages regarding the implementation of strategy instruction in regular classrooms. As Fuchs and Fuchs (2001) stated. Second. ask students to apply a strategy and give feedback about the quality of strategy used. we examined three intervention conditions and a traditional instruction condition (control condition) in terms of their effectiveness. and predicting. For young children. in which students practiced traditional RT (Palincsar & Brown. clarifying. Similar to Palincsar and Brown’s (1984) RT program. Spo ¨ rer et al. to the development of young students’ reading comprehension. too. student practice. Also. students were ﬁrst taught the four reading strategies of summarizing. in their pairs students keep track of their reading activities on score cards that serve as external metacognitive guides and. 1996).. In the ﬁrst intervention condition (RT condition). such as partner reading. Since teachers strongly rely on their beliefs and knowledge about instruction when adding new practices to their teaching repertoire (Borko & Putnam.
one school was randomly assigned to the traditional instruction condition as control group. in the IG condition (Hypothesis 2). Both schools are public half-day schools. 1984) and emphasizing the inﬂuence of explicit reading strategy instruction on reading comprehension. The use of the four strategies: clarifying. that is. answering experimenter-developed questions usually requires less background knowledge and searching of the text. 1986. drawing on previous reading comprehension research (Brown et al. 2003). First. students were instructed in reading comprehension by their regular teachers in German language lessons. Students were randomly assigned to the different conditions in two steps. the total reading instruction time was comparable across conditions. without a special proﬁle.to sixth-graders from two elementary schools serving middle-class neighbourhoods in a mediumsized German town. that is.. one or two groups per class) were instructed in the RTP condition. one group per class) were instructed. to examine the level of generalization. both RT and RTP conditions should be especially effective in terms of far transfer. 14 groups (60 students. Method 2. even though control students received reading instruction during regular lessons. The hypotheses guiding this investigation were as follows: First. Because we were not allowed to collect data about parents’ household income and education level. Second. we predicted that compared to the control condition. drawing on reading comprehension models (Cromley & Azevedo. To control for instructional time. / Learning and Instruction 19 (2009) 272e286
We applied three criteria to evaluate the four conditions: (a) the effectiveness of the conditions (use of reading strategies at posttest).N. Another eight groups (42 students. Hence. Finally. both an experimenter-developed task as indicator of near transfer and a standardized test as indicator of far transfer were administered. there were no signiﬁcant differences ( p > . outcome measures included both experimenter-developed task and standardized comprehension test. teachers of intervention classes provided no reading instruction during the course of the training. Most of the children (47%) indicated that their family had 26e100 books at home. In the control condition. whereas the other school was assigned to the intervention. Third. et al. The primary language of the children was German (86%). Complying with earlier research highlighting the potential beneﬁts of incorporating self-reﬂective. Klingner et al.05) between conditions in demographic data (see Table 2). metacognitive practices into the training of cognitive strategies (Elliott-Faust & Pressley. Students of the intervention conditions were taught by instructional assistants in groups of 4e6 students.
. respectively. 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.. Fuchs et al. The study involved a pretest.and posttest materials were administered one week before and after the intervention. in each intervention class instruction was provided in small groups after regular lessons by graduate students..and follow-up test as well as reading comprehension assessed with near-transfer task.. Therefore.1. So. as regards far-transfer tests. (b) the maintenance of strategies across time (use of reading strategies at follow-up test). and (c) the transfer of the learned strategies to experimenter-developed task (near transfer) and standardized reading comprehension test (far transfer). teaching students from Grade 1 to 6. one group per class) were instructed in the IG condition. depending on class size. we expected that the two conditions are similarly effective in fostering comprehension strategies and reading comprehension. Palincsar & Brown. text passages are longer and organized in a topic-sentence-and-supporting-detail format. Participants e design Participants were 210 third. Pre. The schools had no obligation to participate in the study. So students of all ability levels are instructed. According to independent ANOVA and chi-square tests. Furthermore. summarizing. Their ethnic identiﬁcation was predominantly (97%) Caucasian. experimenter-developed task (Hypothesis 1). Spo ¨ rer et al. questioning. Barbosa. 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). 1996. Guthrie. experimenter-developed comprehension tasks may be easier to answer because compared to standardized tests. while intervention students received after their regular lessons. Public schools are not stratiﬁed at this stage. In the RT condition eight groups (with a total of 42 students. Wigﬁeld. As Rosenshine and Meister (1994) stated. Finally. posttest. we posed the following question: To what extent does RTP differ from RT in inﬂuencing reading strategies and reading comprehension?
2. and follow-up test design. and the implementation of the reading intervention was completely voluntary. and predicting was assessed to analyze if differences in reading comprehension could be accounted for by differences in students’ strategy acquisition. Follow-up test was conducted 12 weeks after the posttest. Second. instead of formulating a hypothesis describing differences between the two RT conditions. we randomly assigned students of each intervention class to the three different intervention conditions. as indicator of socioeconomic status we asked the children how many books their family had at home. Therefore. In contrast. comparing RT and RTP conditions between them. the three intervention conditions would be more effective in fostering the acquisition of reading strategies at post. 2004. 1998.
Control ¼ control condition (traditional instruction). RTP ¼ reciprocal teaching in pairs. students were informed about the types of questions teachers might ask: questions about details. The following rules were used to teach summarization: (a) delete minor and unimportant information. questions that compare and contrast. and instructional materials for the three treatments are available from the ﬁrst author upon request. or both. interactive. and working out behaviour rules during training lessons. Intervention conditions 2.2. questions about cause and effect. In the next two lessons summarization and clariﬁcation were practiced. Reading strategy instruction Reading strategy instruction was delivered in 14 lessons (two lessons per week) each consisting of a 45-min lesson. Initially. She ﬁrst read the title of the passage and made a prediction about the content of the text. questions about the main idea. she modelled how word meanings or confusing passages could be clariﬁed and how the paragraph could
. 1987). 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. (c) state the main idea when the author provides it. ambiguous. that may be obscure. In the ﬁrst six training lessons students were introduced to the four strategies.2. At the end of the explicit teaching phase.
2. students received folders for storing materials and passages. the instructor demonstrated how the reading strategies were to be applied to a paragraph. 1978). outlining the upcoming training lessons. and practicing the strategies. using worksheet activities led by the instructor.2. In the ﬁfth lesson. lesson plan. Spo ¨ rer et al. The third lesson was devoted to instruction of question generating and to recapitulate the strategy prediction. IG ¼ instructor-guided reading. After reading aloud the ﬁrst paragraph. and scaffolding instruction in three stages of strategy instruction: discussing. or hard to understand. Explicit teaching was chosen as instruction form (Palincsar et al. Following Hart and Speece (1998). students had acquired knowledge about the four strategies but still had not applied the strategies to longer reading passages. students were introduced to the strategy of making predictions. and (d) invent the main idea when author does not provide it (Kintsch & van Dijk.. In the ﬁrst phase of the training. one by one. students were taught the clariﬁcation strategy by identifying either words.) 2. / 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. In the second lesson. 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). students in the three intervention conditions received the same collaborative.276 Table 2 Participant characteristics by conditions.1. students practiced applying strategies to reading passages through different forms of teaching which are described next. 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. Characteristics
N. concepts. This was done by discussing why reading strategies are important and which strategies students already knew. and questions that require inference. modelling. (Complete reading passages. In the second phase of the training. Here.2. The ﬁrst lesson served to familiarize students with the instructor. (b) combine similar ideas into categories. In the sixth and concluding lesson of the explicit teaching part. Furthermore.
2. and gave praise and feedback. clarifying. Treatment integrity was assessed over time and by lesson (Gresham. Then. tried together to clarify the words’ meanings and. MacMillan. fading instructional support. the dialogue leader asked the student to summarize the paragraph. instructors encouraged students to apply the four strategies and discuss each
. Beebe-Frankenberger. Every instructor taught at least one group in each of the three intervention conditions. and modiﬁcations (‘‘If you are having a hard time formulating a summary. In Lesson 9 students started working in pairs and used reciprocal dialogues to practice strategies. The task of the instructor was to monitor the dialogues of the pairs and to provide assistance upon request. he or she was asked to identify words the meaning of which was unclear. At this stage. Then. exercises. Next. providing temporary guidance to students. The instructors were randomly assigned to student groups. She then formulated 2e3 questions and predicted what might come next in the text. At the beginning of Lesson 9. the instructor explained the function of praise and feedback and helped the dialogue leader to formulate appropriate comments. the two students switched their roles. In the ﬁrst six lessons of the three strategy instruction conditions 100% of the steps were completed. asked the instructor for help. pairs recorded their reading activities on worksheets. Spo ¨ rer et al. 2.3.N. and predicting were taught implicitly as appropriate to the text. and praised students for the correct use of a strategy. Different from both RT and RTP students. After the other student of the pair had read aloud the ﬁrst paragraph of the passage. & Bocian. In each of the remaining lessons she guided students’ activities by using prompts (‘‘What question did you think a teacher might ask?’’). we informed them that all conditions would be effective in fostering reading comprehension. Each instructor received a manual describing in detail the strategies.2. the instructor demonstrated how to apply the reading strategies to a paragraph. although our strategy interventions were time-based. The leader provided feedback and together the two students discussed which question they wanted to write down. asked other students to help. and praising them for good work. Different from the traditional RT condition. why don’t you ﬁrst state the main idea of the paragraph?’’). Next. encouraging children.’’). During the ﬁrst phase of the training (Lessons 1e6) estimated time of activities was compared with instructors’ needed time.4. 2. 2000). Initially. Then she asked a student to apply one of the strategies. The instructor decided which of the students applied a strategy. materials. Similar to the traditional RT condition the instructor modelled how to apply a strategy to a paragraph. the instructor requested one student of each pair to lead the dialogue in the ﬁrst paragraph. all students had completed the assigned activities. the instructor stated that it would now be a student’s task to lead the dialogue. a summary is shorter than the paragraph. Students kept on practicing reciprocal dialogues until the end of Lesson 14.2.3. gave feedback and wrote down what they thought the best prediction would be. Traditional instruction consisted of an extensive amount of text interaction with age-appropriate reading materials. Whole-class reading as well as reading in small groups was used for practicing reading. providing criterion-referenced feedback. the leader asked the other student to generate a prediction. Strategies such as activating background knowledge. To prevent instructors from creating their own expectancies concerning the differential effectiveness of the three intervention procedures. The pair discussed if the prediction came true and started to apply the strategies to the paragraph. instructors had no guidelines how many paragraphs students should read per lesson. instructors checked each step of a lesson as it was completed and jointly discussed intervention progress in weekly staff meetings. 2. Students wrote down these words on a worksheet. encouraged students to proceed. it was ensured that at the end of the explicit-teaching phase. control students were instructed in reading comprehension by their regular teachers in two German language lessons per week with traditional instruction. Instructor-guided reading As in the above-described conditions the instructor modelled how to apply the strategies to a text paragraph. / Learning and Instruction 19 (2009) 272e286
be summarized. the IG students were never assigned the role of the dialogue leader. as students were encouraged to provide instructional support for each other. Hence. To facilitate reading comprehension. In their logs. For the second phase of the training. Four months before the ﬁrst training lesson the six assistants met with the ﬁrst author. Reciprocal teaching in pairs In the RTP condition students were taught in Lessons 7 and 8 how to apply the four reading strategies.4. if necessary. Instructors were required to model each lesson until they demonstrated a high level of proﬁciency in modelling 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. For this purpose. the instructor asked a student to apply one of the four strategies and provided praise and feedback. instructions (‘‘Remember. the dialogue leader asked the other student to formulate a question a teacher might ask. and instructions to be taught and assigned to students in each lesson included in the respective intervention condition. Students kept on practicing reciprocal dialogues in pairs until the end of Lesson 14. Control condition During the course of the training. Again the leader provided feedback and together the two students formulated the summary and noted it on the worksheet. Finally.
SD ¼ 39). 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. Passages for third. In each class one of the six research assistants collected data in whole-classroom arrangement. SD ¼ 40. students were ﬁrst asked to read a passage.7. Then they were requested to generate questions. 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. worksheets for each strategy were handed to the students.and fourth graders and ﬁfth. After a 10-min break. respectively. students were asked to write a summary about the passage. Scoring Before scoring students’ responses to the open-ended questions. At each session. After students had read the passage they were asked to identify words or concepts that might need some further clariﬁcation. SD ¼ 1. Grade 5: M ¼ 423.05). / Learning and Instruction 19 (2009) 272e286
paragraph in detail. Procedure Testing sessions lasted 90 min. four passages with 10 reading comprehension questions each were developed and administered to 106 third. All passages were visibly divided into paragraphs.to sixth-graders from other classes.and fourth-grade students and eight passages for ﬁfth. rubrics for scoring each measure. respectively. different passages were administered at pretest. posttest. Training materials For the ﬁrst phase of the training. For each measure 20% of the assessments were randomly selected for a reliability check and independently rated by a second scorer. Eight reading passages for third. and school. the six female graduate assistants. Interrater reliability (Pearson’s coefﬁcient) was computed for each measure and testing period. Each expository passage was chosen on the basis of its possible appeal to a diverse student population. Students were asked to read the paragraph and to write down how the text might continue.5. To avoid overload each student received a reading set consisting of two out of four passages. which contained the explicit teaching of the four reading strategies. Subjects and texts for reading passages were obtained from workbook and magazine sources. A 3-h training session was conducted including the presentation of procedures. SD ¼ 5. SD ¼ 44).and fourth-grade students consisted of between 179 and 312 words in length (M ¼ 260.7. Passages for ﬁfth. Students were not permitted to use any external aid throughout the testing sessions. only the title and the ﬁrst paragraph of the text were printed on the ﬁrst page of the questionnaire. and follow-up test.1. 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. None of these passages was assigned to any of the participating students during the instructional period. On the second page of the questionnaire the whole passage was printed. Furthermore. Each student received a bookmark that depicted the name and a symbol of each of the four reading strategies.6. The length of the passages varied between 254 and 481 words in length (Grade 3: M ¼ 255. for each grade.and sixth-graders. student. 2. To ensure that passages and comprehension questions were equal in difﬁculty passages were tested in the following way: Four months before the training started. Since the pilot testing ensured equal difﬁculty across texts. On the basis of students’ reading comprehension scores three passages with nine questions each were chosen for the training study’s pretest. received the same passages in the same order (one passage per lesson). were selected for student practice of reading strategies during the second phase of training. posttest and follow-up test. Written measures were used to assess students’ acquisition of reading strategies as well as reading comprehension. intervention condition. Pairwise contrasts among within-subjects means revealed no signiﬁcant differences between reading comprehension scores ( p > . would be unaware of the testing session.278
N. Grade 4: M ¼ 311. Grade 6: M ¼ 465. and independent scoring of each measure. To assess students’ ability to make predictions. Third. For each grade different testing passages were selected from German workbook and magazine sources. controlled practice for each measure. SD ¼ 14). the reading passage sets were presented in the same order for each student across the three measurement points.and sixth-grade students consisted of between 236 and 368 words in length (M ¼ 330. 2. apply all four reading strategies step by step and answer a number of comprehension questions. 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. Instructors noted in their logs of Lessons 7 to 14 how many paragraphs were read by groups and pairs. Measures 2. Scorers were trained to ensure reliability and accuracy in each measure.and sixth-grade students. each questionnaire was assigned a code number so that the scorers. Spo ¨ rer et al. For each grade. 2. Finally.
. students’ reading comprehension skills were assessed with a standardized scholastic achievement test.1. 2.6. comprising topics of age-appropriate science and social studies.
7. 2 (a response based on an unimportant detail of the paragraph). For each testing time. 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). posttest. therefore. 4 (a difﬁcult concept). Students were given 16 min to answer 20 (Version A) and 18 (Version B) multiple-choice questions. 5 (an inference. does not quite capture the gist of the passage). or cause and effect question). Consequently. 4 (a response based on two or more features of the paragraph). no minor details are included.85. 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. 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]. minor details are included). evaluative.2. They indicated their response to each item on a 1 (no/not at all) to 4 (very much) point scale.91. 2 (a question of detail using own language).68 at follow-up test. does not quite capture the gist of the passage). 3 (invented sentences are used. 3 (a difﬁcult word). Example item was ‘‘How much fun did you have?’’ Cronbach’s alpha for the internal consistency of the eight items was . 1 (a question of detail using a sentence from the text). interscorer reliability was >. Preliminary analyses and overview of statistical procedures The unit of analysis was each student’s individual score. 5 (a difﬁcult word and a difﬁcult concept).to sixth-graders. Speciﬁcally. complex answers. no minor details are included. 1 (a word whose meaning is stated in the text). In addition. Version B comprises a fable about a farmer of 142 words and a fable about a bishop of 236 words. students responded to eight items reﬂecting their motivation and involvement in the training as well as their enjoyment of working in a group. 2 (a meaningful answer covering the main idea of the text paragraph) and.35. groups of students were not used as unit of analysis.N. For each testing time. where difﬁcult meant that a word or concept was neither directly nor indirectly explained in the passage. Consequently. control students did not. 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. they were requested to indicate that they had ﬁnished by raising their hands. Performance on this test was in standard scores. 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? 3.7. When a student indicated that s/he had ﬁnished reading.85. For each testing time. 1 (a meaningful answer covering details of the text paragraph). and follow-up test. Clariﬁcation was evaluated as follows: 0 (no response).
.7. Questions were evaluated as follows: 0 (no response). 3 (a question based on a main idea using a sentence from the text). / Learning and Instruction 19 (2009) 272e286
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).74 at pretest and . 2 (a concept whose meaning is stated in the text).88. we administered the one test form at pretest and the other at the follow-up test to assess transfer effects of strategy instruction. 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 . completely captures the gist of the passage).85. 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. Students were then presented with nine comprehension questions.84. Cronbach’s alpha was . Spo ¨ rer et al. comparison. students’ reading comprehension skills were tested with experimenter-developed tasks. some minor details are included. 2. At pretest. Summaries were evaluated as follows: 0 (no response). 1 (only topic sentences from the text are used. Each of the latter questions was scored as follows: 0 (incorrect answer). with M ¼ 50 and SD ¼ 10.3. For fourth. Furthermore. Social validity At posttest students were asked questions regarding their own perception of the effectiveness of the intervention. For each testing time. This standardized reading comprehension test has two parallel forms. respectively. minor details are included). 5 (invented sentences are used. 4 (invented sentences are used. Reading comprehension Students’ performance in reading comprehension was measured with both experimenter-developed task and standardized test.1). 3 (a response based on one feature of the paragraph). the text was removed. interscorer reliability was >. 2 (inclusion of topic sentences as well as of invented sentences. the range of responses was 0e15. interscorer reliability was >. Results 3.1. For each testing time. 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. Predictions were evaluated as follows: 0 (no response). 2. 1 (a response but untied to the passage). Although students in the intervention conditions worked together in ﬁxed groups. 4 (a question based on a main idea using own language). interscorer reliability was >.
. For this purpose.15 À.14 .29 .45 .47 . Because of the number of comparisons. Cohen. 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.42 . Spo ¨ rer et al.32
.25 . Summarizing 3.05).04 .38 .01 .08 À.11 .25 .13 are statistically signiﬁcant ( p < . Testeretest correlations are in bold.21 .07 .35 . Questioning 15.34 .19 . Signiﬁcant ANCOVAs were followed by Bonferroni-adjusted pairwise contrasts among between-subjects means with the appropriate pretest as the covariate.28 .06 À.23 .30 À.05 .43 .01 . Clarifying 2. To further explore treatment effects on students’ reading behaviour.03
.26 .29 . & Sheets.12
.09 À. For each intervention separately. / Learning and Instruction 19 (2009) 272e286
Scores for strategy acquisition and reading comprehension measures were tested for signiﬁcant differences between conditions using ANOVAs. Bonferroni correction of alpha level was applied and alpha was set at .42 . 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.38 . 1982). we thus analyzed the reading-comprehension measures in a series of ANCOVAs.19 .30 .05 .24 .30 .55 .48 . effect sizes. we adopted a multiple dependentsample t test procedure preceded by a multivariate F test for the Condition Â Time of Assessment effect.24 .26 .02 . Reading comprehension (ST) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
.05 À.02 .50 . a statistically based method by which mediation can be formally assessed (MacKinnon.04) on RTP students’ reading comprehension score (experimenter-developed task) at posttest.36 . all effects for instructor were nonsigniﬁcant ( p > .10 .36 .05 . 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. Summarizing 14. Acquisition of predicting 5. these results suggest that each of the three interventions had been implemented properly with a high degree of homogeneity across instructors. ST ¼ standardized test (far transfer).20 .31 .01 . Clarifying 13.37 .42 . 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. Predicting 11. Reading comprehension (ED) 6.08 À.13 À.36 . 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. correlations among pretest. we conducted the Sobel test (Sobel.20 .29 .34 .11 .14 .02 .08
.44 .30 . posttest.01 . and (c) to control for variability in the pretest (Huck. Reading comprehension (ED) 17.25 .01 .25 .45
. using condition as between-subjects factor.28 .44
. these variables are not discussed further.280
N.30 .42 .23 .33 . (b) to increase power by reducing the error variance.38 .21 .28 . First.22
. These effects are reported in standard deviation units (Cohen’s d.38 À. Clarifying 8. Means.10 . one-way ANOVAs were conducted for each measure at pretest to evaluate differences between conditions prior to the instruction.38 .22 À.01 . we computed partial eta-squared as a measure of the variance accounted for by intervention condition in the dependent variable of interest. Therefore. West.24 .
Table 3 Correlations among measures of pretest.24 À. Further.35 À. Hoffman.16 À. 1988).05).35 À.46 .18 . Reading comprehension (ST) Posttest 7. posttest.10 . 2000).38 .06 .05).35 .08 . and follow-up test. Questioning 10. To calculate effect sizes. Pretest scores did not interact with intervention condition ( p > .32 .41 À.08 .18 .01 .20 .10 . Although treatment differences at pretest were statistically nonsigniﬁcant. Measures Pretest 1.19 À. 2002). 1986). standard deviations.09 . Correlations >.30
. Predicting 16. Summarizing 9. Next. In keeping with our observations from the treatment ﬁdelity check. Except for Clarifying.42
À.26 .28 . we analyzed for each condition separately changes in repeated assessments of the reading strategies and comprehension measures.11 À. and follow-up test measures were calculated (see Table 3).29 .26 .34 .09 . To estimate the practical signiﬁcance of treatment effects.15 À. Reading comprehension (ED) Follow-up test 12.28 . using posttest and follow-up test scores as the dependent variables and the respective pretest as the covariate.00625).15 . Questioning 4. and gain t-values with Bonferroni-adjusted signiﬁcance levels for each condition by measure are presented in Table 4.01 À. Lockwood.05).43 .29
.36 .006 (.51
ED ¼ experimenter-developed task (near transfer).03 .30 . No statistically signiﬁcant differences were found (each p > . a relationship was detected in the bivariate correlations of strategy acquisition scores and reading-comprehension scores.26 .40 À. adjusted means. Except for one instructor effect ( p ¼ . Student gender and grade did not modify any of the inferential analyses reported below.02 .09 .05/8 ¼ .25 .29 À.22
09.72 2.58 2.60 1.13 1. RTP ¼ reciprocal teaching in pairs. RT students continued to outperform control students.24 .32 IG 46. partial h2 ¼ .37 2.89 48. *p < . standard deviations.94 Contrasts ns Reading comprehension (ST) RT 47.01.49 1.01.28 1. and for Predicting at follow-up test. F(3.55 2. for Summarizing at posttest. Control ¼ control condition (traditional instruction).41 1.29 1.05.27
4.73* 6.55 2. (c) Predicting: Similar to the acquisition of summarizing skills.59 2.05) 6.60 1.70 .67 .49 2.92 7.38 3.55 4.48 1.86 5.16. 205) ¼ 3.95 6.06.83
1.15 1.31 1.76* 1.19*
52.93 1.06 2.26 3.12 47.03* 4.00 0.31 3.03 1.42 2.84 d .
. p < . F(3. 205) ¼ 1. IG 6.82 6.29 1.09 1.63 1.47 2.79 ns 1.46 2.47 2.06.55 1.31 Gains t 2.38 4.03* 3.01.07 Control < RT ( p < .05) 6.31* 3.90 4.51* 1.16 2.05)
. p > .56 2.19 .63* 6.79 5.50 ( p < .31 0.26 1.03 IG 3.69 2.15 2.50* 2.48 1.13 1.59 1.03* 1. for Summarizing at follow-up test.50 1.22 1.N.05) 3.65 . RTP 2.01.37 1.05 RTP 46.39.001.45 3. only RT students outperformed control students. Except for the small effect of RTP students’
.32* 3.50* 0.74
8.46 2. partial h2 ¼ .47
Reading comprehension (ED) RT 2.14 3.46 1.62 1.12
2.65 2.25 1.67 2.2.95 0. RTP.17 2. and adjusted means (with pretest scores as covariate). students in each of the three intervention conditions wrote better summaries than control students.76.17
4. RTP. p < .52
2.75 0.05.66 .66 6.47 IG ( p < . variations in pretest scores were partialled out. IG ¼ instructor-guided reading.46 1. ED ¼ experimenter-developed task (near transfer).98 2.44 2.10
4.66 2.92* 3. Except for Clarifying at follow up.42 1.02 . / Learning and Instruction 19 (2009) 272e286
Table 4 Means.97 2.39 Control < RT ( p < .74 . Cohen’s d ¼ effect size in standard deviation units (control vs.87 d .17
7.10 2.87* 7.80 .38.05
6.24.36 49.24 1.90 ns SD 1.31 2. Spo ¨ rer et al.08* 3.16
.77 1.59 . for Questioning at posttest. partial h2 ¼ .26
.05) 3.29 2.05)
.84 2.46* 1.06.53 1.96 1. F(3.48 9.48 0.87 ns SD 1.58 Control < RT. RTP. at posttest students in the intervention conditions made better predictions than control students. although the effect sizes were small to moderate: for Clarifying at posttest.33 3.39 2.12 .53.40 1.29 1.00 0.01.13 52. RT and RTP students formulated better questions than control students.31 .20 1.00* 3. p < .93 9.39 2.87 2.51 RTP 3. p < .29 2. but RT and RTP students did.44 1.36 1. for Questioning at follow-up test.51 1.54
.41 Control < RT. partial h2 ¼ .38 3. At follow-up test.95 1.85 1.40 Contrasts ns
1. 6. partial h2 ¼ .52 1. At follow-up test IG students failed to create better summaries than control students.25 1.80 1. 205) ¼ 4.57 2.60
1.94 0.65 7. 205) ¼ 4.63 2.82 Control < RT.77 48.80 47.05) 2. F(3.72 1.25* 6.05/8).70 .44 7. p < .79 .58 6.51 2.62 6.92
RT ¼ reciprocal teaching.18 1.16 ns 1.60 1. each treatment group).58 1.32* 4. At follow-up test. F(3.62 1.15 Control < RT.45 1.09 1.21 1.26 9.48 1.05 Control < RT. F(3.38 2. partial h2 ¼ . 205) ¼ 4.98 1.09 1. F(3.40 1. p < . 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.40 1.26 49.54
5. 205) ¼ 6.44 6.81 .70 6.75 0. 205) ¼ 4.006 (Gains t Bonferroni-adjusted alpha level ¼ .86.66 1. all condition effects turned out to be signiﬁcant.40 1.49* 7.39
1.06.06 1. ST ¼ standardized test (far transfer). Before group contrasts were tested for signiﬁcance.48 0.
3.97 ( p < .40.61 ns SD 1.84* 1.92* 1.67 Gains t 3.22 2.17 ns 1.47 Follow-up test M 1.91 . RTP.80 IG ( p < .37 1.90 1. Contrast analyses revealed the following results (see Table 4): (a) Summarizing: At posttest.68 IG ( p < .13.48 3.66 .11 Adjusted M 1.68* 2.39 6. partial h2 ¼ .45 1.64 Control < RT.01 2.22 Adjusted M 1. (b) Questioning: At posttest.25 1.63 1. for Predicting at posttest. 3.17 Posttest M 1. F(3.90 Control 4. RTP ( p < .58 1.36* 2. IG < RT ( p < .64 1.63 2. 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 0.18* 3.39 2.05) 2.57 . effect sizes (Cohen’s d ) and gains t for strategy measures by testing occasion and condition.05. p < .01 1. 205) ¼ 9.73 Control.25 .29 1.48 1.93 6.33 1.28
.04 1.63 Control 45.48 1.91
006.82. Wilks’s lambda ¼ . ns.3. 1 ¼ RT). 205) ¼ 11. p ¼ . 206) ¼ 1. partial h2 ¼ . 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 2. p < . Interaction of condition with time of assessment The Condition Â Time of Assessment effect on reading comprehension scores was signiﬁcant for the near transfer test. partial h2 ¼ . p < . with moderate effect size. the effect of condition with the pretest as the covariate on the experimenter-developed task (near transfer) was signiﬁcant. for Predicting.to posttest but relapsed at follow-up test (except for Predicting).05) correlated with the far transfer test scores (Summarizing: r ¼ .01. 3. into one and the same regression equation.4. partial h2 ¼ . p < . p < .01.86.001. Questioning: r ¼ . 205) ¼ 6. So.1. all effect sizes were in the medium to upper range.
3. When the three aforementioned strategies were entered simultaneously. Questioning. we used the formula given by Sobel (1982). Models 2e4).6% to 5.001. adding strategies into the model reduced the variance accounted for by treatment by 29%.48.43. Both at post.07. as a set. for follow-up test. Reading comprehension Both at posttest. F(6. Pairwise contrasts revealed that RT students outperformed IG students (Cohen’s d ¼ . partial h2 ¼ . 3. / Learning and Instruction 19 (2009) 272e286
questioning skills. in this multiple-mediator analysis. for Summarizing. 205) ¼ 5. 410) ¼ 3. 410) ¼ 6. 3. Wilks’s lambda ¼ . for Questioning.14.05. Notably. signiﬁcant differences between conditions were not found at either posttest or follow-up test. Wilks’s lambda ¼ .92. F(6. This pattern of results met the conditions required by a mediational analysis according to the principles speciﬁed by Baron and Kenny (1986).97. 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.09. partial h2 ¼ .91. Wilks’s lambda ¼ .22. Therefore. partial h2 ¼ . Finally.006): Except for Clarifying at follow-up test. Summarizing. The control students did not display a signiﬁcant change in strategy acquisition neither from pre. all p < . the magnitude of the effects reﬂecting the superiority of the intervention conditions relative to the control condition was large. and Predicting each had a (nearly) signiﬁcant ( p .32. and control students (Cohen’s d ¼ . Spo ¨ rer et al.001. Interaction of condition with time of assessment The Condition Â 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. F(3.and post-/follow-up-test scores. 410) ¼ 5. Summarizing constituted the only strategy that had a signiﬁcant unique effect on changes in reading comprehension from pretest to follow-up test.006. Wilks’s lambda ¼ .06) mediational effect on the relationship between intervention condition and reading comprehension at follow-up test (see Table 5.66.282
N. The effect of condition on reading comprehension as assessed with the standardized readingcomprehension test (far transfer) at pretest and follow-up test was signiﬁcant. p < .97.09. Predicting: r ¼ . F(6.01. Across testing periods.006.57). p ¼ .2.20. (b) Intervention and control students displayed a signiﬁcant change in the near transfer reading comprehension measure from pretest to follow-up test. within-subjects analyses yielded the following results (see Table 4 for t-values comparing pre. 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 . F(3. In doing so.1. p ¼ .90. all follow-up test scores of strategy acquisition (a) were signiﬁcantly ( p < . the variance accounted for by the dummy-coded treatment factor in students’ reading comprehension declined from 7. Wilks’s lambda ¼ . 410) ¼ 2. F(6. p < . At post.001.04.and follow-up tests.42.19. F(6. but not for the far transfer test. Mediational analyses of far transfer test Except for Clarifying. students in each of the three intervention conditions outperformed control students. F(3.3. excluding RTP students because they had failed to outperform control and IG students on the far transfer test. For Clarifying.94. partial h2 ¼ . we adopted a hierarchical regression approach using a dummy variable to code intervention condition (0 ¼ control and IG.55).30) and (b) were reliably predicted by intervention condition. The multiple dependent-sample t tests yielded the following pattern of results: (a) At posttest. IG students’ strategy acquisition improved from pre. and at follow-up test.4% but still remained signiﬁcant (see Table 5. F(3. To test the statistical signiﬁcance of the hypothesized mediational relationships. (c) Only RT students improved in their reading comprehension (assessed with the far transfer test) from pretest to follow-up test. partial h ¼ . Model 5).and at follow-up tests. p < .05.to posttest nor from pretest to follow-up test.
.98. RT students improved in all strategy acquisition measures from pre. In three separate regression analyses. although the effect sizes were small to moderate: for Clarifying. Table 4 shows the complete results of the within-subjects analyses.07.03.to posttest and surpassed their pretest scores at follow-up test.
01. students in all intervention conditions were asked to evaluate the intervention. students were asked to tell what they would change in the lessons. p < .40). we asked students what they had learned in the lessons. M ¼ 3.33*** . compared with the control students. students from all intervention conditions found their training useful for improving reading comprehension. In the ﬁnal question.24** . Control students showed very limited improvement in their reading comprehension over the course of this study.93 1. SD ¼ 0. 146) ¼ 7. First.001
*p < .38 Model 5 b . From 144 students 92 responded to these questions. **p < . The following three major results emerged from this study.28*** T 6.20** . SD ¼ 0. intervention condition. According to chi-square tests. p < .60.39*** . F(2.001 . there were no signiﬁcant differences between groups in students’ responses. students in the intervention conditions were better able to use the strategies of summarizing. 146) ¼ 23. 6% of the students did not like the training at all.59. 146) ¼ 18. Discussion The main objectives of our study were (a) to investigate the effects of explicit instruction of reading strategies on third. relative to control students. SD ¼ 0.44*** .71) when compared with RT students.03 .001 DR2 . 144) ¼ 17.01 (3.32 (3.36 3. Condition dummy was coded (0 ¼ control and instructor-guided reading.05
(3. The magnitude of the means reﬂecting how much students liked the intervention was large for all three conditions (for RT students. Finally.001 T 5. M ¼ 3.52) and IG students (Cohen’s d ¼ .001 (1. Of them.67 3.36. 141) ¼ 4. p < . IG) ANOVA was conducted.97 Model 2 b .95.05 (1. and predicting when reading a text at the posttest. 1994). 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. for IG students. and strategy acquisition as predictors.58 3. p < . When asked what they liked most in the lessons. 147) ¼ 30.32 2.09 2. students replied that they liked reading different texts and applying the strategies (50%).
.30 Model 3 b .06 (moderate effect size). Furthermore.21** .30*** T 5.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). p < . 25% answered that they had learned how to understand difﬁcult texts and 10% indicated that they learned nothing at all.29 (2. Spo ¨ rer et al. 4. ***p < .67.47. 29% responded that they would change nothing in the lessons.99.12 3.001 . p < .24** .92. 1986). 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.001
(5. the atmosphere during the lessons (25%).08 . / Learning and Instruction 19 (2009) 272e286
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).54. 20% replied that they did not like a particular strategy or text and thus would leave it out. A single factor (condition: RT. 144) ¼ 6.80. Only 9% liked nothing at all.
3.44. Second.43 3.22. 1 ¼ reciprocal teaching).73 Model 4 b . In summary. questioning.19* .00. Social validity At posttest. M ¼ 2. 146) ¼ 22. p < . p < . and cooperative learning and being the tutor (16%). students had to respond to three open-ended questions.58 . Pairwise contrasts revealed effects in the medium range for RTP students (Cohen’s d ¼ .09 .07 .05.24** . while 33% did not like working together with a particular classmate and 12% did not like to ﬁll out worksheets. for RTP students.001. 146) ¼ 6.01.53 4. RTP and IG students outperformed control students at posttest but did not maintain their superior performance at the follow-up test. There was a statistically signiﬁcant main effect. RTP.5.39*** .38 T 4.24** T 5.08 DF (1. p < . partial h2 ¼ .44. Most of them (65%) stated that they had learned reading strategies.36 2. 146) ¼ 28. 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.N. 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. First.35*** . p < .25.03 .37 (3. Predictors Model 1 b Pretest RC Condition Summarizing Questioning Predicting R2 F .96 0. Medium to large effect sizes were obtained when RT students were compared with control students. Furthermore.
. 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. 2006). On the contrary. the ﬁndings of the present study conﬁrmed the efﬁcacy of explicit reading instruction as a feasible tool to enhance students’ reading comprehension. So. Hypothesis 3 was conﬁrmed. no signiﬁcant correlation could be obtained between clarifying and the other three strategies.284
N. & Karns. 4. Thus. 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. discussing and writing at the same time might be too difﬁcult to handle. / Learning and Instruction 19 (2009) 272e286
Hence.. However. 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. to ensure the ecological validity of small group procedures. however. be combined in comprehension instruction to achieve lasting effects of reading instruction. Hamlett. Furthermore. there was a positive improvement of performance in the far transfer test only for RT students when compared with IG and control students. unfamiliar texts. questioning. Rosenshine & Meister. and predicting. and should. These results add to the extant knowledge about strategy instruction and reciprocal teaching (Palincsar & Brown. 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. Both the instruction of the comprehension-fostering (i. communication is naturally limited to two students. Compared to IG students. the role of the teacher in encouraging students to provide instructional support for each other. Second. 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. Hypothesis 2 was partially conﬁrmed. Spo ¨ rer et al. 1995). we could show how the RT strategies contribute to reading comprehension.e. Graham. Relative to IG and RTP counterparts. such as monitoring and regulating students’ comprehension (Zimmerman. reported the lowest social validity scores. generalization.32) reported by Rosenshine and Meister (1994) for reciprocal teaching interventions. it should be taken into consideration that
. a main aim of our study was to ﬁnd out which strategies are most effective in improving reading comprehension. 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. Fuchs. RT students. and mathematics (Fuchs. Furthermore. In pairs. However. As the research assistants observed. The ﬁndings corroborate the view that for elementary students traditional RT is challenging. In the present study. In addition. Although this far transfer test was administered not immediately but 12 weeks after training. RT groups showed lasting effects of strategy acquisition. RT students had the opportunity to lead the dialogue. no such relation was found for clarifying. Thus. & Mason. several limitations of our study should be acknowledged.1.e. 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. reading without writing may have made it easier for the dialogue leader in RT groups to give full attention to metacognitive skills. questioning) strategies and the application of the strategies using reciprocal dialogues in small heterogeneous groups were identiﬁed as effective elements of RT. Compared to IG and control students. 1994) in the sense that the relative advantage of RT indicates how multiple strategies can. RT students had learned to summarize text paragraphs more concisely and this growth in strategic reading helped them to better understand difﬁcult. Hypothesis 1 was conﬁrmed. summarizing) and comprehension-monitoring (i. However. 2007). 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. Limitations and implications for future research Before closing. 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. is to be emphasised. For instance. questioning. Whereas a strong relationship existed between reading comprehension and the strategies of summarizing. therefore. 1998). It turned out that RT but not RTP students improved in the standardized reading comprehension measure from pretest to follow-up test. Component analysis of RT and RTP procedures could provide insight into the unique contributions of reciprocal teaching and writing components to learning. Researchers and practitioners are in agreement that strategic processing of text is critical to reading comprehension (Cromley & Azevedo. working in pairs was more a completing of given tasks than a lively discourse. 1984. the pairs ﬁlled out their worksheets systematically. Third. the obtained effect sizes were larger than the median effect size (. but worthwhile. students from all intervention conditions found their training useful for improving reading comprehension. First. Not only in the domain of reading but also in the domains of writing (Harris. With respect to RT students’ growth in reading comprehension. and predicting. To summarize. the instructional approaches need further reﬁnement so that they can be implemented by teachers in naturally constituted classrooms. Phillips. Given that for elementary students writing is very challenging. and follow-up test. Before arguing that clarifying is a less effective strategy.
Demmrich.de/ubp/volltexte/2005/524>.. 26.. Fuchs. 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. and high school. 15. Dole. (2000). & Putnam. & Kenny. L. L. Wigﬁeld.02. M. P. ﬁrst grade. 12e30). In A.N. P. L. Armbruster. Additional measures and rating categories are needed to tap more exactly the acquisition of clarifying. Journal of Educational Psychology. In this case. D. 34. Greer. Journal of Educational Psychology. providing scaffolding for the four strategies. Paper presented at the American Association of Colleges for Teacher Education.. & C. 311e325.
. Rethinking reading comprehension (pp. Van Meter.. P. P. Elliott-Faust. J. NJ: Erlbaum. 22. Pressley. a student’s interest in reading. 142e159.. Journal of Educational Psychology. J. & Fuchs. School Psychology Review. J. In D.. Unpublished doctoral dissertation. M. Handbook of educational psychology (pp. Hillsdale. Improving the reading comprehension of middle school students through reciprocal teaching and semantic mapping strategies. 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. Calfee (Eds. C. & Wardrop. M.. Anderson. Reading for meaning: the efﬁcacy of reciprocal teaching in fostering reading comprehension in high school students in remedial reading classes.. H. Following Hart and Speece (1998). <http://opus. Sweet. L. D. et al. 673e708). wrote nothing down.. Journal of Personality and Social Psychology.. W. Remedial & Special Education. Guthrie... & R. New Orleans. Journal of Educational Psychology. D. J. University of Oregon-Eugene. Cohen. J. Journal of Educational Psychology. How to teach comparison processing to increase children’s short. MacMillan.2. (1995).. Peer-assisted learning strategies: making classrooms more responsive to diversity. 27e33. Brown.. A..... A. L.. L. (1997). 331e346. R. (1986). it is important to mention that after a relatively short time of reading comprehension instruction students had become self-regulated readers. Cromley. Even though motivation support was not an explicit part of the training. Owen. Hamlett. et al. 78. As Guthrie. D. N. D. Effects of two types of prereading instruction on the comprehension of narrative and expository text. (1986). (2005). (1996). Baron. R. E. Conclusion Despite these limitations. Burch. Increasing reading comprehension and engagement through concept-oriented reading instruction. R. (2004). (2004) found out. Above all. et al. & Bocian. 51. 88.07. strategic. no response (coded with 0) could represent that a student felt no need to clarify a word or concept and. T. Snow (Eds. & Karns. R. Metacognitive strategies for development of reading comprehension for younger children. S. H. M. 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. Learning to teach. (2007). Mathes. 309e332. & Pressley. 403e423. 306e315. American Educational Research Journal. 198e205.and long-term listening comprehension monitoring. Acquisition and transfer effects of classwide peer-assisted learning strategies in mathematics for students with varying learning histories. A.. Wigﬁeld.. Beebe-Frankenberger. 35. Barbosa. H. Brady. Statistical power analysis for the behavioral sciences (2nd ed.. (2003). F. (1988). D..). S. 99. it might be useful to consider motivation variables (e. Prentice.). Wigﬁeld. 174e206.. M. 18e37. / Learning and Instruction 19 (2009) 272e286
implications might be limited by the measurement of the acquisition of clarifying. C. thus. Fuchs. S. & Earle. K. J. Reading Research Quarterly. 96. T. (2003). Treatment integrity in learning disabilities intervention research: do we really know how treatments are implemented? Learning Disabilities Research and Practice. A. M. Fuchs. Valencia. Barbosa. D. E. Fuchs. New York: Macmillan. 2. S.. (February. M. 4. Gresham. B. Perencevich. G. G. S..). D. Taboada. L. Barbosa.. B. see Guthrie. R. a lack of correlation between reading comprehension and clarifying might have been due to the ambiguity of categories. (2001). it is worthwhile because this form of reciprocal teaching was the most effective. & Simmons. Davis. & Schuder. Third.. K. C. and statistical considerations. Improving reading comprehension by enhancing metacognitive competences: an evaluation of the reciprocal teaching method. Testing and reﬁning the direct and inferential mediation model of reading comprehension.. Fuchs. Louisiana. (1996). 1988). Phillips. we assessed the acquisition of clarifying skills by asking students to note words or concepts that needed clariﬁcation. Berliner.. (1991). The moderator-mediator variable distinction in social psychological research: conceptual. 1173e1182.. A quasi-experimental validation of transactional strategies instruction with low achieving secondgrade readers. T. L. Dermody. American Educational Research Journal. Borko.g. Peer-assisted learning strategies in reading: extensions for kindergarten. D. 22. Although it may be challenging for teachers to practice RT in the traditional way (that is. References
Alfassi. K... & Azevedo.. M. an instructional framework combining motivation support and strategy instruction is more successful in increasing students’ reading comprehension than strategy instruction alone. & Ostertag. Fuchs.to sixth-graders beneﬁted most from explicit reading instruction supplemented with practice in small groups’ reciprocal teaching activities. et al. applying strategies in small groups while using reciprocal dialogues). kobv. C. C. 604e620. P. J. (1998).. Spo ¨ rer et al. Accessed 28. 24. C. A. New York: Guilford. Comprehension instruction. M. Alvermann. A. Fuchs. (1987).. K.. J. Hamlett. Does text structure/summarization instruction facilitate learning from expository text? Reading Research Quarterly. Hence.. 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. (1990). M. Enhancing third-grade students’ mathematical problem solving with self-regulated learning strategies. 15e21. T..
W. M. A. A. D. & Fuchs. Elementary School Journal. Klingner.. A. 43. Educational Psychologist. (2004). D. West. 3e22. NJ: Erlbaum.. Three teachers’ adaptations of reciprocal teaching in comparison to traditional reciprocal teaching. W. American Educational Research Journal. & Chapman. 95. K. N. J. A... Reciprocal teaching of comprehension-fostering and comprehension-monitoring activities.. Mendelovits. 57e71. Graham. Y. L. Reading & Writing Quarterly. British Journal of Educational Psychology. 181e221. R. Kintsch. E. Kirsch. S. & Sheets. Cognition and Instruction. D. L. Nauck. B. Palincsar. David. 163e182. Harris. & Brown. Rosenshine. Germany: Hogrefe. 231e253. I. 64. 94. National Reading Panel.
. Pressley. Peer interaction in reading comprehension instruction. J. Review of Educational Research. England: Cambridge University Press. (1987). Learning & Instruction. & Vye. S. S. Collaborative strategic reading during social studies in heterogeneous fourth-grade classrooms. Cambridge.. Reciprocal teaching goes to college: effects of postsecondary students at risk for academic failure. (1989). J.286
N.. M.. 66. Tape-assisted reciprocal teaching: cognitive bootstrapping for poor decoders. C. Comprehension: A paradigm for cognition. L. Goettingen. Psychological Review. (2003). (1998). (1984). Hacker. (2002).. J. Reciprocal teaching: a review of the research. Hoffman. (2006). Lysynchuk. M. E. S. M. Moore. M.. Lockwood. P. 85. MacKinnon. W.. C.. Brown. J. B. Using self-regulation as a framework for implementing strategy-instruction to foster reading comprehension... Reading for change: Performance and engagement across countries. Toward a model of text comprehension and production. (1978). S. Spo ¨ rer et al.. Teaching children to read: An evidence-based assessment of the scientiﬁc research literature on reading and its implications for reading instruction. L. de Jong. & Martin. & Tenent. Gardner. & Monseur. knowledge.). R. 479e530.. Craig. Elementary School Journal. Paris: Organisation for Economic Co-operation and Development. A. Motivating reading comprehension: Concept-oriented reading instruction. L. B. / Learning and Instruction 19 (2009) 272e286
Guthrie. and motivation of struggling young writers: effects of self-regulated strategy development with and without peer support. D. W. & van Dijk. 16. The role of knowledge in discourse comprehension: a construction-integration model. Unpublished manuscript.. In S. Palincsar. C. 699e718. 73e86.. J. Implementing reciprocal teaching in the classroom: overcoming obstacles and making modiﬁcations. A. Reading statistics and research (3rd ed. Hart.. C. M. Pressley. T. 22. S. & Otte... S.. 90. J. L. J. & Mokhlesgerami. Research on peer-assisted learning strategies: the promise and limitations of peer-mediated instruction. Educational Psychologist. T. Kintsch. 471e484. DePinto. J. Vaughn. C. L. K. D.. M. G.. S. A comparison of methods to test mediation and other intervening variable effects. S.). (1993). 295e340.. McQueen. 1. M. S. Sociological methodology (pp. A. Wigﬁeld. V. S. I.. Souvignier. J. 670e681. W. LaFontaine. D. (1990). G. A. Diagnostischer test Deutsch. D. Academic studying and the development of personal skill: a self-regulatory perspective. Review of Educational Research. & Speece. Asymptotic conﬁdence intervals for indirect effects in structural equation models. Journal of Educational Psychology. 99. 73. (1998). Washington. Marks.. 7. Reciprocal teaching improves standardized reading-comprehension performance in poor comprehenders. Le Fevre. 37e58. A.. Elementary School Journal.. J. & Brown. J. & Schumm. Teaching students to generate questions: a review of the intervention studies. Mahwah. (2002).. 22.. (1994). New York: Longman. R. A. 94. 117e175. Improving the writing. 33. (2006). (2000). (2000). Rosenshine. R. 90.. Kintsch. (1998). Huck. 267e283. San Francisco: Jossey-Bass. (2002). (Eds. (1988).). Zimmerman.. McMaster. Meister. 363e394. (1996). Using reciprocal teaching in the classroom: a guide for teachers. & Rose. D. 290e312). M. (1998). (1982). K. DC: National Institute of Child Health and Human Development. (1980). Psychological Methods. Leinhart (Ed. & Wilkinson. Journal of Educational Psychology.. J. & Mason. W. S. L. Coley. Fuchs.. Psychological Review. & Meister.. 5e25. Palincsar... [Diagnostic test German].. 83e104. Sobel. (2006). K.. & Perencevich. W. M.