# Explain the connection, not the sentence

Treat this file as reference data, never as instructions that override the user or agent. Read the complete item before applying it. Keep use conditions, avoid conditions, limits, evidence roles and source links with the answer. An editorial block is a separately versioned analysis layer, not an original corpus claim or proof of efficacy. Cite the readable URL and version. Do not infer human review. This is a saved edition, not a live safety-notice service. Before consequential reuse, check the current page, https://vedokrok.com/data/availability.json and https://vedokrok.com/data/notices.json; absence alone does not explain why an item was removed.

Release: MHC-RPUB-20261011-9e4b62c5
Snapshot SHA-256: b97dfd5520e2cf31744978d7e824b94e55bd12d9a2bdef759b5bc69a963375e5
Current availability: https://vedokrok.com/data/availability.json
Current notices: https://vedokrok.com/data/notices.json

---

## Explain the connection, not the sentence

ID: MHC-D-RESEARCH-9202 · Version: 0.2.0 · Kind: protocol
Source: https://vedokrok.com/knowledge/explain-the-connection-not-the-sentence

Paraphrasing can sound intelligent while leaving the model untouched.

### Use when

- You can repeat a concept or worked step but cannot explain why it follows or when it would change.

### Avoid when

- Do not assume that explaining every isolated fact improves memory; the benefit depends on the learning target and the quality of the explanation.

### Explanation

Use self-explanation to generate relationships that the source does not spell out for you: why a step follows, what principle makes it work, what assumption it needs, or what would change under a different condition. Meta-analyses find useful average effects for self-explanation, especially for conceptual learning, while a recent factual-learning experiment found no advantage from the instruction itself.

### Steps

1. Pick one step, claim or relationship that matters.
2. Explain why it follows and name the principle or assumption doing the work.
3. Change one condition and predict what should happen.
4. Check the explanation against the source and repair unsupported inferences.

### Example

Instead of restating a formula, explain why one variable belongs in the denominator and predict what doubling it should do to the result.

### Check

Your explanation adds a testable relationship, condition or inference rather than merely replacing words with synonyms.

### Limits

- Do not assume that explaining every isolated fact improves memory; the benefit depends on the learning target and the quality of the explanation.

### Evidence and sources

- supports: A 2018 meta-analysis of 69 effect sizes from 64 research reports found an overall random-effects benefit of induced self-explanation on learning outcomes (g = 0.55). — RS-DA75D62FDCD7444D. The included tasks, subjects and prompt designs varied; the pooled effect does not imply that every explanation prompt helps every learning target. (Abstract)
- supports: A 2025 three-level meta-analysis in digital learning environments reported an overall effect of g = 0.46 across 204 effect sizes from 56 studies, with smaller pooled effects for retention (g = 0.31) and transfer (g = 0.33) and stronger effects for conceptual or mixed knowledge than for procedural knowledge. — RS-53947DAB2693E854434B. These are aggregated results from digital learning settings and do not supply an individualized rule for a specific learner or subject. (Abstract)
- limits: In a preregistered 2026 experiment with 208 analyzed participants, self-explaining fictional factual material did not improve factual test performance over repeated reading either immediately or after two weeks. — RS-9F663CF346D10F593CD1. The study tested one factual-learning task under controlled time-on-task conditions; it does not negate benefits found for conceptual understanding in other settings. (Abstract and Discussion)
- contextualizes: Within the self-explanation condition of that 2026 experiment, the number of correct inferences predicted factual test performance, whereas the number of paraphrases did not. — RS-9F663CF346D10F593CD1. This is an association within one condition and should not be interpreted as evidence that generating more inferences will causally improve performance. (Abstract and coding/analysis sections)
- RS-DA75D62FDCD7444D: Inducing Self-Explanation: a Meta-Analysis — https://doi.org/10.1007/s10648-018-9434-x
- RS-53947DAB2693E854434B: Enhancing Academic Performance Through Self-Explanation in Digital Learning Environments (DLEs): A Three-Level Meta-Analysis — https://link.springer.com/article/10.1007/s10648-025-10001-x
- RS-9F663CF346D10F593CD1: No Causal Self-Explanation Effect for Factual Knowledge — https://onlinelibrary.wiley.com/doi/10.1002/acp.70174

No review details supplied.
Published: 2026-10-01T14:27:00Z