Full dictation is one of the hardest beginner tasks: perception, segmentation, lexical access, prediction, spelling, handwriting and monitoring all at once. Scaffold it in a progression and keep learning-dictation separate from testing-dictation.
The nine processes (a pedagogical model, not a literal account)
Acoustic input → phonological perception → segmentation → lexical access → grammatical/contextual prediction → orthographic retrieval → grapho-motor execution → working memory → monitoring.
Progression (Language Gym Dictation Trainer)
- Multiple choice (discrimination, no writing)
- Missing detail (one element)
- Sentence puzzle (reorder chunks)
- Break the flow (mark word boundaries; segmentation)
- Spot the word
- Missing vowels / missing consonants
- Guided dictation (support fades ~70% → 50% → 20% → 0)
- Full dictation
Principles: familiar language, short sentences, strategic repetition, progressive difficulty, reflective questions after errors. “Don’t repeatedly test the complex performance you eventually want.”
Limits & critique
- No empirical evidence cited for the sequence or the fade percentages.
- Promotes his own product (conflict of interest).
Apply in my context (Claude)
- “Break the flow” targets exactly the segmentation problem found in Bottom-up listening instruction alone is not enough. Pair the two notes.
- Movers/Flyers Listening (spelling of names, numbers) and KET Listening gap-fills reward steps 2, 5 and 6. Build 5-minute warmers from them.
Connections
- Bottom-up listening instruction alone is not enough · MARS-EARS - the EPI instructional sequence · Pitch tasks inside the ZPD and hand over support (the fading logic is scaffolding)