Teaching children how to talk about science

Schoolchildren and teacher in science classOne theory in science education claims that scientific concepts and ways of reasoning are learned through engagement in practical inquiry and social interaction as well as individual study. Adherents believe that students learn and understand science better if they interact productively with other students during scientific investigations. A carefully designed experimental teaching program based on this theory sought to improve children’s ability to use language as a tool for reasoning and improving their achievement in science and math classes. Neil Mercer, Lyn Dawes, Rupert Wegerif and Claire Sams, The Open University, Bedford, England, report on the results involving elementary science classes.

These researchers report that observational research in British schools has shown that “talk which takes place when children are asked to work together is often uncooperative, off-task, inequitable and ultimately unproductive.” The poor quality of much collaborative talk may be explained by the lack of training children receive in how to work together and communicate effectively. Mercer et al. contend that children rarely have a clear conception of what they are expected to do or what constitutes a good, effective discussion. They have no understanding of how to talk together or for what purpose. In science classes children need to learn to describe observations clearly, reason about causes and effects, pose precise questions, formulate hypotheses, critically examine competing explanations and summarize results. They also need to gain relevant knowledge of natural phenomena, investigative procedures, and scientific concepts and terms. To do this, they need to learn how to use language to inquire, reason, and consider information together, to share and negotiate ideas, and to make joint decisions.

The Intervention Program: Thinking Together

The goals of the program were to raise children’s awareness of the use of spoken language as a means for thinking together; to enable them to develop these language abilities as a tool for thinking, both collectively and alone; and to apply these skills effectively to the study of science and math. The goal of this “exploratory talk” was to ensure that:

  • All relevant information is shared
  • All members of the group are invited to contribute to the discussion
  • Opinions and ideas are respected and considered
  • Everyone is asked to make their reasons clear
  • Challenges and alternatives are made explicit and are negotiated
  • The group seeks to reach agreement before making a decision or acting.

Each teacher in the study was provided with 12 detailed lesson plans. Each lesson involved a teacher-led introduction, a group discussion activity and a final sharing session. Lessons were focused on teaching and learning explicit skills such as critical questioning, sharing information or negotiating a decision. The first five lessons were aimed at helping students understand how talk could be used for working together and to set the ground rules for discussion that would generate exploratory talk. Each lesson applied a specific skill and targeted a specific concept in science. Some lessons involved computer-based activities. Granada’s Science Explorer II software was used.

Teacher modeling is necessary component

A necessary component of the intervention was teachers’ modeling and guiding the development of children’s language skills. Each participating teacher received training in the Thinking Together approach based on videotaped examples and activities from earlier research projects. Once the intervention began, this training was reinforced through regular visits to schools by a researcher who demonstrated activities or teaching techniques as needed.

The effects of the study were measured through observation (videotaping of class activities) and formal assessments (Raven’s Progressive Matrices, to measure nonverbal reasoning, and SATs, a set of standardized assessments measuring science knowledge and understanding for Year 5 British students). Children in experimental and matched control classes had pre- and postintervention tests. One hundred and nine students completed the experimental program and were compared with 121 students in control classes at the end of the study. Control classes did not participate in the Thinking Together program, but they followed the same science curriculum and also used the Science Explorer software, though not necessarily as the basis for regular group activities. The intervention continued for 23 weeks.

The videotapes of classrooms’ science activities were evaluated for the appropriateness and effectiveness of students’ language use as a tool for thinking together. To check for biased judgments, a blind assessment was carried out using transcripts from science classes as children worked on an assignment novel to both control and experimental groups. Professors not involved in the project were asked to assess these transcripts. Evaluators also measured science knowledge and understanding and administered standardized tests covering the curriculum studied during the experiment. Also, students were asked to construct a concept map of scientific terms and their relationships.


Research revealed significant differences between experimental and control classes. These researchers believed that if teachers taught students joint of the science curriculum would improve. This proved to be true. An effect size of .29 showed a small but significant achievement advantage for students in the experimental classes. Evaluations of class transcripts showed that students in control classes did not share knowledge, build on each other’s suggestions or provide reasons for their proposals. Verbal exchanges tended to be brief.

In the experimental classes, students made effective use of the strategies they were taught. They asked each other for information and opinions, asked for and provided reasons, shared thoughts and evaluated proposals. Challenges were constructive, and all members of the group worked toward a joint decision. Opinions were treated with respect and each speaker had the opportunity to develop ideas as he spoke. The blind assessment corroborated researchers’ evaluations of classroom activities.

Children in experimental classes solved the nonverbal reasoning problems of the Raven’s Progressive Matrices more effectively. The Raven’s test was used because it correlates highly with other tests of reasoning and with measures of academic achievement. An effect size of .55 supports the hypothesis that engaging in this kind of exploratory talk with peers improves children’s reasoning ability on unrelated tasks.

The concept-mapping exercise was used to test the extent to which children in the experimental classes became, over the six-month intervention, relatively more able to perceive relationships between different scientific concepts and terms. Although the intervention did not involve any specific teaching about these relationships, the children’s superior postintervention performance in producing concept maps (effect size of .74) was attributed to the improved quality of their collective reasoning about science.

Overall, the results show that the children in the experimental classes gained significantly better scores in science than those in control classes, demonstrating the effectiveness of teaching exploratory talk in elementary science classes.


Although the differences between classes on standardized achievement tests were not large after taking into account their pre-intervention performance levels, the experimental classes did demonstrate educationally significant improvement over the control classes. Encouraging the use of exploratory talk has specific, positive effects on the quality of children’s reasoning as measured by a test that is not similar in content or format to any activities in the intervention program. In these researchers’ opinion, therefore, the discursive and reasoning skills gained by the students in experimental classes represented a transferable skill. These children gained important, generalizable communication and thinking skills.

Long-term effects were studied through spotcheck video recordings of activities one year after the intervention ended. These showed that individuals and groups were still able to recall the ground rules and to use exploratory talk appropriately when carrying out problem-solving activities. This research demonstrates that an experimental teaching program enabled elementary students to work together more effectively, improve their language and reasoning skills, and reach higher achievement in science.

Since the intervention program was designed to include both teacher-led and peer-group activities, it is not possible to determine the effects of either factor separately. However, the complementary effect of both types of interaction for science education has been well demonstrated by previous research. Mercer et al. conclude that their findings add to the evidence that the development of scientific understanding is best assisted by a careful combination of peer-group interaction and expert guidance, and they provide an example of how that combination can be achieved.

So that the results of this project can inform educational practice, a set of lesson plans called “Thinking Together in Maths and Science” has been created. When published, it will incorporate guidelines for the effective use of existing educational software as a basis for group activity. It illustrates the structure of the program and the teaching strategies involved in implementing it, using video sequences of classroom interaction to illustrate ways of modeling and scaffolding children’s questioning and reasoning. The authors can be contacted by e-mail:


“Reasoning as a Scientist: Ways of Helping Children to Use Language to Learn Science”, British Educational Research Journal, Volume 30, Number 3, June 2004, pp. 359-377

Published in ERN October 2004 Volume 17 Number 7

Leave a Reply

  • (will not be published)