Encouraging pupils to argue and debate in class can help boost their results in English, maths and science, according to research.
Getting primary school children to explain their answers and reason with their classmates can help youngsters to make more progress in these core subjects, it suggests.
A total of 78 primary schools in England with higher-than-average numbers of poorer children took part in the trial of "dialogic teaching", published by the Education Endowment Foundation (EEF).
Around 2,500 nine and 10-year-olds were given lessons in which they were encouraged to explain their answers and reasoning, and to debate, discuss and argue with others about them.
This was done using methods such as teachers asking open questions and encouraging youngsters to do more than simply state an answer.
An independent evaluation of the initiative found that the youngsters who took part made on average two months more progress in English and science than a similar group of pupils who did not take part - the control group.
In addition, poorer youngsters who took part in the programme made two months more progress in maths, while, in general, the children who took part made around a month's more progress in the subject than the control group.
Children in both groups were tested in each of the subjects before and after the programme.
The findings indicate that this type of teaching may improve youngsters' overall thinking and learning skills, rather than just their subject knowledge, EEF said.
Teachers were in favour of the scheme, but many felt that they needed more than two terms to make it fully part of their classroom teaching.
EEF chief executive Sir Kevan Collins, said: "Getting children to think and talk about their own learning more explicitly can be one of the most effective ways to improve academic outcomes.
"But it can be difficult to put this into practice in the classroom. While there is no simple strategy or trick, today's evaluation report on dialogic teaching does give primary school heads and teachers practical evidence on an approach that appears to be effective across different subjects."