After listening to a short English text, you could transcribe every word. Or you could keep only the key words and rebuild the message using your own language resources. The second activity is Dictogloss, sometimes called grammar dictation.
It connects listening, note-taking, speaking, writing and feedback. You do not have to memorise the original, but you must understand the content, choose language and make the sentences cohere.
How it differs from traditional dictation
Ruth Wajnryb's classroom procedure reads a short, dense text twice at natural speed. Learners note familiar words and phrases, pool their notes in small groups and reconstruct a version. Different versions are then compared, analysed and refined.
The goal is not a word-for-word replica. It is a coherent, content-equivalent text with language as accurate and appropriate as the learners can make it. That makes the activity more than a listening test: it creates opportunities to discuss and test language choices.
Masahiro Maeda's 2024 article for Japan's ELEC describes adaptations for EFL classrooms: listen for the whole message first, take notes on later hearings, then listen again after reconstruction to notice gaps. It also recommends individual listening and reading aloud as follow-up work, which translates naturally into solo study.
A 15-minute solo version
1. Choose a coherent 60–90-word text
Most vocabulary should be familiar, and the story or explanation should be easy to understand in your first language. Start with a reliable transcript and audio, not an unfamiliar news report.
2. Listen once for meaning, then take notes
After the first hearing, answer: who was where, what problem appeared and how did it end? On the second hearing, note content words, numbers and connectors such as *flight / delayed / hotel / refund / finally*. Do not attempt complete sentences.
3. Reconstruct aloud without the original
Use only your keywords to speak for 60–90 seconds and record it. Your wording may differ from the source, but preserve the causes, sequence and outcome. Use a simpler word and continue if the ideal expression is missing.
4. Compare for three kinds of gap
- Content: Did you omit a decision, contrast or outcome?
- Cohesion: Does everything depend on *and*, or are time and cause clear?
- Language: Which of your sentences is inaccurate or awkward?
Select at most two priorities. A different version is not automatically wrong if it is clear, accurate and appropriate.
5. Hear the model, then retell from the same keywords
Listen again with attention to missing links and expressions. Close the transcript and reconstruct once more. Finally, change the speaker or situation, perhaps turning someone else's experience into your own, to test transfer.
A partner or small-group version
Take notes independently before showing them. Reconstruct sections orally in turn, while others add content and ask about unclear links. Agree on a shared version and let everyone deliver it before comparing with the original.
Do not let the strongest English user simply write the answer for everyone; the others would lose the retrieval and organisation work. CARLA's teaching material also warns that a group can retain an incorrect solution it constructed together. End with a trustworthy source or informed feedback.
Why this helps when ideas will not become sentences
Dictogloss supplies content support without supplying the complete wording. You are not inventing a story from nothing, but you still have to select words, form sentences and connect them.
It also makes the gap between comprehension and production specific. You may understand everything yet lack connectors such as *because*, *although* and *after that*. Or you may retain every keyword but struggle with verb forms and combinations. The next input should address that particular gap rather than another vague instruction to "listen more".
Do not turn it into a memorisation contest
A long text full of unknown words consumes attention in basic listening. Demanding exact reproduction removes the value of independent reconstruction.
Beginners can use three or four sentences, listen three times or describe a picture before hearing the text. Later, introduce unfamiliar speakers or take fewer notes. The procedure can become harder without becoming longer.
Use MirrorSay for the sentence you cannot reconstruct
Keep your own version before checking the source. When you know the intended meaning but not a reliable expression, say it to MirrorSay, compare the natural version and shadow it. Return to the keyword retelling immediately, so the tool fills a language gap rather than completing the story for you.
To rely less on notes, continue with Retrieval Practice for Speaking. To improve speed across the whole message, use The 4/3/2 Speaking Drill.
References
- Wajnryb, R. (1997). Grammar Dictation: The Dictogloss Procedure as a Means to Explore and Shift Learners' Hypotheses about Language. *The Language Teacher*, 21(9). JALT introduction
- Maeda, M. (2024). The Effectiveness of Dictogloss for English Communication. ELEC Research Institute
- Wajnryb, R. *Dictogloss*. CARLA, University of Minnesota. Teaching steps and research notes (PDF)
*MirrorSay is a free AI speaking-practice app: say a sentence, see a natural way to express it, then shadow the model. No sign-up, and speech is processed on-device whenever possible.*