Self-Explanation: Why Explaining Out Loud Makes You Understand More Deeply
By Noam Shemla · The Studies
TL;DR Chi and colleagues had eighth-graders learn how the heart and blood vessels work from a textbook passage. Half were asked to explain each sentence to themselves, out loud, as they read. The other half read the passage twice. The explainers learned more, especially on the hardest questions, and the students who explained the most all ended up with a fully correct picture of the circulatory system.
- Paper: Chi, M. T. H., De Leeuw, N., Chiu, M.-H., & LaVancher, C. (1994). Eliciting self-explanations improves understanding. Cognitive Science, 18(3), 439–477.
- Participants: 24 eighth-grade students from an inner-city school, none of whom had studied biology; 14 prompted, 10 in the control group.
- Material: A 101-sentence textbook passage on the human circulatory system, one sentence per page.
- Prompted group: Asked to explain what each sentence meant, out loud, and at 22 points asked what a component was for.
- Control group: Read the same passage twice, roughly matching the time spent on the text.
- Tests: Before and after: explaining 23 terms, drawing the path of the blood, and four categories of questions, from recall to complex inference.
Most people learn from a text by reading it again. Michelene Chi and colleagues at the University of Pittsburgh tested a different habit: stopping after every sentence to say, out loud, what it means and how it connects to what came before.1
What self-explaining sounds like
A self-explanation, in the paper's definition, is anything a learner says that goes beyond the text: an inference. After reading that the septum divides the heart into two sides, one student reasoned out loud that it must be there so the blood does not get mixed up, and that it is like a wall. Neither idea is in the sentence. Both are what understanding the sentence requires.
What they found
Both groups learned from the passage; the question was how much more the explainers learned. Their scores rose by more, and the difference was largest on the questions that could not be answered by finding a sentence in the text.
Talking more was not the same as explaining more. The high explainers did say more overall, but not significantly more, and a larger share of what they said was self-explanation.
A better picture of how the heart works
The most telling result is not a score but a model. From what each student said, the researchers classified how they pictured the circulatory system, on a six-step ladder from wrong to right.
Every prompted student: how much they explained, and how their model changed
| Student | Self-explanations | Model before | Model after |
|---|---|---|---|
| 1 | 111 | Single loop | Double loop 2 |
| 2 | 87 | Loop with lungs | Double loop 2 |
| 3 | 84 | Single loop | Double loop 2 |
| 4 | 64 | Single loop | Double loop 2 |
| 5 | 47 | Single loop | Loop with lungs |
| 6 | 47 | Single loop | Double loop 2 |
| 7 | 42 | Ebb and flow | Double loop 2 |
| 8 | 40 | Single loop | Double loop 2 |
| 9 | 35 | Loop with lungs | Double loop 2 |
| 10 | 33 | No loop | Loop with lungs |
| 11 | 28 | Single loop | Double loop 1 |
| 12 | 20 | Loop with lungs | Double loop 1 |
Ability did not decide the outcome. The five highest-scoring and five lowest-scoring prompted students on a standard achievement test both gained about 30 points. What differed between students was how much they explained, not how able they were.
Has it held up?
It has held up well. A 2018 meta-analysis of 64 research reports and 69 effect sizes found an average benefit of g = 0.55 for prompting learners to self-explain, across subjects, levels of education and kinds of task, from solving problems to studying text.2
What this study cannot tell you
- It is small. 24 students, and the high-versus-low comparisons rest on four students each.
- High explainers were not assigned. They chose to explain more, so part of their advantage may come from whatever made them do it; the prompted-versus-control comparison is the experimental one.
- The prompted students spent longer with the text, about two hours against one, despite the control group reading it twice.
- It is one topic and one age group, though the later meta-analysis extends the finding widely.
What a speaker can take from it
Saying something out loud in your own words is not only how you show you understand it; it is how you come to understand it. The students did not learn by rehearsing sentences but by explaining the connections between them, and the ones who explained the most built the most accurate picture.
That is the best way to prepare a talk. Rather than reading your notes again, explain your subject aloud, point by point, to an imaginary listener, and notice where the explanation breaks down. Those gaps are where the audience would have got lost too.
This article summarises a published paper. The gains, test statistics, counts and models are the authors’ own, from the results text and Tables 4 and 5. The example of a self-explanation is paraphrased.
Measured
- Test scores before and after reading, on four categories of question
- Self-explanations, counted from recordings of students talking
- Each student’s mental model, classified from what they said
Inferred, not measured
- That explaining caused the deeper understanding: shown experimentally for prompting, correlational for high versus low explainers
References
- Chi, M. T. H., De Leeuw, N., Chiu, M.-H., & LaVancher, C. (1994). Eliciting self-explanations improves understanding. Cognitive Science, 18(3), 439–477. https://doi.org/10.1207/s15516709cog1803_3
- Bisra, K., Liu, Q., Nesbit, J. C., Salimi, F., & Winne, P. H. (2018). Inducing self-explanation: a meta-analysis. Educational Psychology Review, 30(3), 703–725. https://doi.org/10.1007/s10648-018-9434-x