PPN Home > Programs that Work > Reciprocal Teaching

Programs that Work

Reciprocal Teaching

Program Info
Program Overview
Program Participants
Evaluation Methods
Key Evaluation Findings
Probable Implementers
Implementation Detail
Issues to Consider
Example Sites
Contact Information
Available Resources
Last Reviewed


Program Info

Outcome Areas
Children Succeeding in School

Students performing at grade level or meeting state curriculum standards

Topic Areas

     Age of Child
       Middle Childhood (9-12)
       Adolescence (13-18)
     Type of Setting
       Elementary School
       Middle School
       High School
     Type of Service
       Instructional Support
     Type of Outcome Addressed
       Cognitive Development / School Performance

Evidence Level  (What does this mean?)

Back to topTop  

Program Overview

The purpose of Reciprocal Teaching (RT) is to facilitate a group effort between teachers and students by bringing meaning to segments of text. In order to promote understanding, RT is an instructional activity that utilizes dialogue between teachers and students while segments of text are studied. The teacher and students take turns in assuming the role of instructor in leading the two-way dialogue. Four strategies are used to structure the discussion, including: (1) summarizing, which compels students to determine the most important information in the text; (2) question generating, which reinforces the summarizing strategy and increases comprehension; (3) clarifying, which is especially important for students with a history of difficulty with comprehension; and (4) predicting, which increases students’ depth of understanding by hypothesizing what the author will discuss next in the text. These strategies are modeled by the teacher and then practiced by the students in cooperative groups. While the students engage in discussions aimed at constructing meaning to the text, the teacher provides the support necessary for students to learn how to implement RT strategies.

Back to topTop  

Program Participants

Students in grades 3 through 12

Back to topTop  

Evaluation Methods

Miller, Miller, and Rosen (1988) studied the effects of an eight-week modified RT program among 64 seventh-grade students. The RT program was modified in that it added the identification of key words and phrases, and instruction was conducted biweekly as opposed to daily. Students were randomly assigned to one of three classes taught by a single social studies teacher. The three classes were in turn randomly assigned to one of three conditions: modified RT (26 students), Control Group I (20 students), or Control Group II (18 students). The two control groups differed in the type of assessments that were administered to students; participants in Control Group I completed the same ten-item comprehension tests and writing samples as the RT group, while participants in Control Group II did not complete any comprehension tests or writing samples. The three groups of students did not differ significantly from one another on their vocabulary scores from the Metropolitan Achievement Test, which was administered three weeks before the study began. Two outcome measures--a ten-question multiple-choice comprehension test and a three-minute writing sample (scored by number of words written)--were used for the RT group and Control Group I and administered during 11 of the 16 sessions. The third outcome measure--academic grades--was recorded for all three groups at the end of both the first and second quarters of the school year.

The effects of a 13-session RT program were assessed in a sample of 36 fourth-grade and 36 seventh-grade students from London, Ontario (Lysynchuk, Pressley, and Vye, 1989). The fourth-grade students were enrolled in six schools, and the seventh-grade students were enrolled in two schools. All participants were nominated by their teachers as being adequate decoders of text but poor comprehenders (i.e., the students had satisfactory levels of word recognition but below-average levels of reading comprehension), and none of the students were classified as learning disabled or mentally challenged. All participating fourth-grade subjects had pretested below the 50th percentile on the reading comprehension subtest of the Metropolitan Achievement Test (MAT), while all seventh-grade participants had pretested below the 50th percentile on the reading comprehension subtest of the Gates-MacGinitie Reading Test (GMRT). Following the pretest, pairs of participants were identified at each grade level that had similar pretest scores, and one student in each pair was randomly assigned to the RT condition and the other student to the control condition. Outcome measures included (1) MAT reading comprehension (fourth graders), (2) GMRT reading comprehension (seventh graders), (3) vocabulary subtest of the Canadian Tests of Basic Skills (fourth graders), and (4) GMRT vocabulary (seventh graders).

Taylor and Frye (1992) studied a five-month RT program conducted with a sample of 150 students from four fifth-grade and four-sixth grade classrooms in two suburban elementary schools (the state was not specified). Two fifth-grade and two sixth-grade teachers from one school volunteered to have their average or above average reading students serve as the treatment groups. The control group consisted of two fifth-grade and two sixth-grade groups of students with similar reading abilities from a second elementary school. Average reading ability was determined by the teachers, as indicated by students' placement in the basal series, reading achievement scores, and daily class performance. Four two-group comparisons between the RT and control groups were assessed:

  1. Group 1 (RT) consisted of 24 above-average sixth-grade readers, and Group 2 (control) consisted of 18 average sixth-grade readers.
  2. Group 3 (RT) had 22 average sixth-grade readers and Group 4 (control) had 20 average sixth-grade readers.
  3. Group 5 (RT) contained 15 above-average readers and Group 6 (control) contained 15 average fifth-grade readers.
  4. Group 7 (RT) and Group 8 (control) had 20 and 16 average fifth-grade readers, respectively.
Students in Groups 5 and 6 were paired together because they were considered by their teachers to be of somewhat higher ability than students in Groups 7 and 8. Pretest data were collected during three assessments in December, and posttest data (on the same measures) were collected at the end of May. After reading a three- to four-page segment from their social studies book that had not yet been covered, students were asked to (1) summarize the passage in approximately ten sentences, (2) write six important questions on the material, and (3) respond to a set of short-answer questions on the material.

Reciprocal Teaching was also assessed in a study of 46 eighth-grade students from seven classes in Auckland, New Zealand (Westera and Moore, 1995). Participating students were considered poor comprehenders but adequate decoders and were selected after scoring the lowest of 300 students on a standardized test of reading comprehension. Approximately half of the study participants were Maori or Pacific Islanders. Eight RT groups were conducted over five weeks, with four of the groups (20 students) exposed to between 12 and 16 RT sessions and the other four groups (15 students) receiving between six and eight sessions. The remaining 11 students served as a control group. No significant differences were found between the groups on pretest reading comprehension scores. Student outcomes were assessed through scores on the Progressive Achievement Test Reading Comprehension instrument.

Lovett et al. (1996) studied 46 learning-disabled students in grades 7 and 8 in Toronto, Ontario. Unlike the other studies of RT that were conducted in regular classrooms during regular school hours, the researchers implemented RT in special laboratory classrooms at a pediatric teaching hospital and in satellite classrooms at two community schools. Students were randomly assigned to one of three 25-hour instructional programs. Two programs offered training in expository text comprehension -- RT (16 students) and the Text Content and Structure Program (TCS) (16 students) which required students to elaborate and further process new knowledge from the text being read, focusing on both specific informational text and knowledge of text structure conventions. The third program was the control-group program, which utilized the Classroom Survival Skills program (CSS) (14 students) that trained students in organizational strategies, academic problem solving, and self-help techniques but utilized no text-comprehension training. Subjects were randomly assigned to groups, and were also randomly assigned to instructional texts A (instructed or "taught-to" texts) and B (uninstructed or "not taught-to" texts). Three categories of outcome measures were used: (1) text comprehension strategies and operations, (2) comprehension of text content, and (3) ability to analyze text structure. In addition to measures in these three categories, a battery of standardized tests was also administered to all students, including the Wide Range Achievement Test—Revised (WRAT-R) Spelling, Arithmetic, and Reading subtests; the Gilmore Oral Reading Test; the Test of Reading Comprehension (TORC) Paragraph Reading subtest; and the Woodcock Reading Mastery Test—Revised (WRMT-R) Word Identification, Word Comprehension, Word Attack, and Passage Comprehension subtests.

A 20-day RT program was assessed in a sample of 75 freshman high-school students from two high schools in a middle-class, suburban school district who were enrolled in remedial reading classes (Alfassi, 1998). All students were considered to be poor comprehenders but adequate decoders. The RT group consisted of 53 students who were divided into five reading classes, while the control group included 22 students from a neighboring high school located in the same school district. Control group students were divided into three reading classes. The composition of the classes with respect to race and sex was similar across groups, and no significant differences were found between groups on reading assessments at baseline. Outcome measures assessed eight weeks after program completion included (1) Gates-MacGinitie Reading Test vocabulary and comprehension subtests; and (2) daily passage-comprehension assessments, which included assessing the answers to four text-explicit questions (questions based on information that was directly in the text), four text-implicit questions (questions that required students to locate the information and make links and references across information), and two script-implicit questions (questions that required use of the student’s own knowledge base for the answer).

Finally, Johnson-Glenberg (2000) evaluated RT in a sample of 59 third-, fourth-, and fifth-graders from three schools. Students participated for ten weeks in the RT program (22 students) or the visualizing/verbalizing (V/V) program (23 students), or were part of the untreated control group (14 students). The V/V program involved training students to create mental images from the text being read and to discuss these images with other students. All participants were identified by their classroom teacher as having age-appropriate decoding skills but poor comprehension skills. Students were assigned to one of 12 small working groups (with two to five students each), and these groups were in turn assigned to an experimental condition. Ninety-five percent of the students were Caucasian, and less than 10 percent received free/reduced lunch. Outcome measures included (1) a test of oral reading skills, including mistakes, speed, predictions, question generation, recall, open-ended questions, self-reported strategy use, and listening recall; (2) the Detroit Test of Learning Aptitude; and (3) a task involving identification of associated word pairs.

Back to topTop  

Key Evaluation Findings

The Miller, Miller, and Rosen (1988) study of 64 students found the following:

  • RT students scored significantly higher than control group students on the multiple-choice comprehension tests.
  • Students in the RT group wrote a significantly higher average number of words (61.58 versus 58.59) per three-minute writing sample than did students in the control group.
  • First-quarter grades were not significantly different among the three groups. However, an examination of second-quarter grades indicated that RT students performed significantly better than students in Control Group I. The mean grades for the second quarter, using a four-point grading scale (4.0 equaling an A), were 2.17 for Control Group I, 2.28 for Control Group II, and 2.80 for the RT group.
Lysynchuk, Pressley, and Vye’s (1989) study of 72 students in London, Ontario, reported the following:
  • Overall reading-comprehension performance had greater improvement for the RT students than for the control students. The mean pretest-to-posttest gain of 9.97 percentile points in the RT condition was statistically significant, while the increase of 1.63 percentile points in the control condition was not.
  • When each grade level was examined independently, the same results were found. The fourth-grade MAT pretest-to-posttest improvement in the RT condition was statistically significant; likewise, the improvement of seventh-grade GMRT scores was statistically significant. In contrast, neither pretest-to-posttest change in the control condition approached statistically significance.
  • No significant differences in vocabulary tests were found between groups at either grade level.
The study of 150 fourth- and fifth-grade students (Taylor and Frye, 1992) found the following:
  • In three of the four comparisons of summarizing scores, RT students performed better than control subjects, including above-average sixth-graders (Groups 1 and 2), above-average fifth-graders (Groups 5 and 6), and average fifth-graders (Groups 7 and 8). No significant differences were found between groups for average sixth-graders (Groups 3 and 4).
  • No significant differences were found among any of the groups for scores on student-generated questions.
  • Results for short-answer scores were mixed. Average fifth-graders in Group 7 (RT) significantly outscored readers in Group 8 (control group). However, average sixth-graders in Group 4 (control group) scored significantly higher than students in Group 3 (RT). No other significant differences between groups were found.
Westera and Moore’s (1995) study of 46 students in New Zealand reported the following:
  • Pretest-to-posttest reading comprehension scores differed among the three groups, with the extended RT group scoring significantly higher than the control group. Scores for the short RT group and scores for the control group did not differ significantly.
  • These results indicate that the students who received between 12 and 16 RT sessions gained on average more than one age-equivalent year in tested reading comprehension over the five-week period. Ninety-five percent of the extended-RT students showed gains in reading comprehension, compared with 47 percent of students in the short-RT program and 45 percent of students in the control group.
Lovett et al.'s (1996) study of 46 learning-disabled students in Toronto, Ontario, found the following:
  • Significant program effects were observed for the seven comprehension strategies/operations measures. The RT students’ score gains from pretest to posttest were significantly better than the score gains of both the TCS and control group students for the four strategies of predicting, summarizing, questioning, and clarifying. These advantages in outcomes were evidenced for both instructed ("taught-to") and uninstructed ("not taught-to") texts.
  • For the remaining three comprehension strategies--generating titles, rating the importance of specific ideas, and detecting incongruities in text--RT students scored significantly better on instructed-text materials than the control group students, but not significantly better than the TCS group. For uninstructed texts, no differences were observed among the groups.
  • Significant program effects were revealed for the two types of content-comprehension measures. The first measure included elaboration-comprehension questions on specific text content. Analyses indicated that both the RT and the TCS groups improved their scores by a significantly greater amount than did the control group on instructed texts (but not uninstructed texts).
  • A significant program effect was also found for the second content comprehension measure, which involved semantic mapping exercises (in which students were asked to generate headings and subheadings that would help them remember the ideas from a text). Both the RT and TCS groups performed better than the control group on mapping content from instructed texts, while the TCS group scored higher than both the RT and control groups on mapping content from uninstructed texts.
  • A significant program effect was revealed for the structural analysis measure, in which students attempted to recognize and use conventions of text linguistic structure to understand the author’s intended meaning. Comparisons of outcomes for this measure indicated that the TCS group had a significant post-test advantage over both the RT and control groups for both instructed and uninstructed texts.
  • No significant differences among groups were revealed for any of the standardized measures.
Alfassi's (1998) study of 75 freshman high-school students reported the following:
  • A significant difference between the RT and control group for the measure of passage comprehension—the RT group demonstrated a significant improvement in scores from pretest to posttest, while the control group did not.
  • No significant differences were found between groups for either the Gates-MacGinitie comprehension test or the vocabulary test.
Finally, Johnson-Glenberg’s (2000) study of 59 elementary school students reported the following:
  • The RT and V/V groups on 11 of the measures made significant pretest-to-posttest gains. Significant gains were made by the control group on only one measure (DTLA-Following Directions).
  • The experimental groups demonstrated statistically significant greater gains than the control group on four measures, including WRAT word recognition, question generation, answering explicit open-ended questions, and answering visual open-ended questions. Additionally, the RT group demonstrated a marginally significant gain over control group students on answering implicit open-ended questions.
  • When the RT and V/V groups were compared, RT students significantly outperformed V/V students on answering explicit open-ended questions. Conversely, the V/V group performed marginally better than the RT group on DTLA—Following Directions.

Back to topTop  

Probable Implementers

Public and private elementary schools

Back to topTop  


No information available at this time

Back to topTop  

Implementation Detail

Program Design
RT is an instructional approach in which teachers and students engage in dialogue while segments of text are studied. Four strategies are used to structure the discussion, and each strategy attempts to help students construct meaning from text as well as to ensure that they are in fact understanding what they read.

  1. Summarizing: In summarizing, students must identify and integrate the most important information in the text. Students may be asked to summarize across sentences, across paragraphs, or across the passage as a whole, and the difficulty level increases as the task includes a more comprehensive amount of text.
  2. Question generating: To generate questions, students must first identify the kind of information that is significant enough and substantial enough to provide material for the answers as well as the questions. Students may be required to ask questions focusing on detailed information, or to go so far as to infer new information from the text.
  3. Clarifying: When students are asked to clarify, they become aware that there are many reasons why text can be difficult to understand (e.g., new vocabulary, unclear reference words, and unfamiliar or difficult concepts). Students can be taught to recognize the effects of such impediments to comprehension and to take measures to enhance meaning, for example, rereading the segment or asking for help.
  4. Predicting: To successfully predict what the author will discuss next in the text, students must be cognizant of the relevant background knowledge that they already possess regarding the topic. The predicting strategy also makes use of text structure as students learn that headings, subheadings, and questions imbedded in the text are useful means of anticipating what might occur next.
No specific curriculum is required to implement Reciprocal Teaching, because it involves an instructional strategy rather than explicit content.

Training for RT is available; however, any teacher who endorses the theoretical framework for RT can implement the program in his or her classroom.

Back to topTop  

Issues to Consider

This program received a "promising" rating. Four out of the seven studies of RT were conducted using rigorous experimental designs, and five out of the seven studies incorporated baseline measures in order to assess both pre- and post-test student achievement. While many of the evaluations of RT demonstrate that treatment-group students scored significantly better than control-group students on tests of reading comprehension and vocabulary, the results were often mixed. Overall, evaluation findings for this program should be interpreted cautiously due to small sample sizes and an inconsistent pattern of significant findings on outcome measures.

For example, Lovett et al.’s (1996) study reported mixed effects favoring both the RT and the TCS group on non-standardized outcome measures, but no significant RT effects on any of the standardized achievement measures. Similarly, Alfassi (1998) reported positive findings for RT versus control group students on a non-standardized test of reading comprehension, but no significant differences were found between groups on standardized comprehension or vocabulary tests. Other studies also reported mixed findings (e.g., Taylor and Frye, 1992; and Johnson-Glenberg, 2000).

Despite these mixed findings, RT has been successfully replicated in many locations both within and outside the United States (e.g., Canada and New Zealand). Additionally, all of the studies included in this review were conducted by independent evaluators.

The study by Westera and Moore (1995) suggests that the length of RT implementation may be important. The authors found that the reading comprehension scores of extended-RT students (those who received 12 to 16 sessions) were significantly higher than the reading scores of students in a short version of the program. However, none of the evaluations reviewed conducted long-term follow-up of RT participants, e.g., for one year or more, therefore the sustainability of effects cannot be assessed.

Back to topTop  

Example Sites

Auckland, New Zealand
London, Ontario
Toronto, Ontario

Back to topTop  

Contact Information

Annemarie Sullivan Palincsar
School of Education
University of Michigan
610 East University Ave.
Ann Arbor, MI 48109-1259
Tel (734) 647-0622
Fax (734) 936-1606

Back to topTop  

Available Resources

Web-based hypermedia to support teacher learning about RT is currently under development, and should be ready for dissemination in the summer of 2005. The materials include video and teacher interviews regarding RT lessons conducted with a group of fourth graders, which can serve as a model for teachers interested in implementing the program.

Back to topTop  


Alfassi, Miriam, "Reading for Meaning: The Efficacy of Reciprocal Teaching in Fostering Reading Comprehension in High School Students in Remedial Reading Classes,"  American Educational Research Journal,  Vol. 35, No. 2, 1998, p. 309-332. 

Bell, Nancy, "Visualizing and Verbalizing for Language Comprehension and Thinking,"    1986, Paso Robles, California: Academy of Reading Publications. 

Johnson-Glenberg, Mina C., "Training Reading Comprehension in Adequate Decoders/Poor Comprehenders: Verbal Versus Visual Strategies,"  Journal of Educational Psychology,  Vol. 92, No. 4, 2000, p.772-782. 

Lovett, Maureen W., Susan L. Borden, Patricia M. Warren-Chaplin, Lea Lacerenza, Teresa DeLua, and Rosa Giovinazzo, "Text Comprehension Training for Disabled Readers: An Evaluation of Reciprocal Teaching and Text Analysis Training Programs,"  Brain & Language,  Vol. 54, No. 3, 1996, p. 447-480. 

Lysynchuk, Linda, Michael Pressley, and Nancy J. Vye,  Reciprocal Instruction Improves Standardized Reading Comprehension Performance in Poor Grade-School Comprehenders,  paper presented at the Annual Meeting of the American Educational Research Association, San Francisco, CA, March 27-31, 1989. 

Miller, C. Dean, Larry F. Miller, and Lee A. Rosen, "Modified Reciprocal Teaching in a Regular Classroom,"  Journal of Experimental Education,  Vol. 56, No. 4, 1988, pp. 183-186. 

Taylor, Barbara M., and Barbara J. Frye, "Comprehension Strategy Instruction in the Intermediate Grades,"  Reading Research and Instruction,  Vol. 32, No. 1, 1992, pp. 39-48. 

Westera, Julia, and Dennis W. Moore, "Reciprocal Teaching of Reading Comprehension in a New Zealand High School,"  Psychology in the Schools,  Vol. 32, No. 3, 1995, p. 225-232. 

Back to topTop  

Last Reviewed

March 2005

Back to topTop