Give the complete source and preserve the first response before these teaching models. Record prior exposure and actual support. Staff answer keys and reserved R sources remain withheld; public formative resources are not secure tests.
Teacher setup: clean scrap and pencil or usual approved tools; one source/own work. Choose one route. Optional 25-minute planning start:3 encounter,4 model,4 try,10 own work,4 review; split/extend freely. Fresh days use first-response time before models, not a speed score.
Child: “Look at two new pictures. Show what you notice, then tell me what you think.” Fresh E1/E2 are delivered exactly and separately, before today's modelling, examples or routes. Capture literal features, interpretation, comparison and access/support. One unfinished component can guide the next lesson. Teacher feedback and all routes below occur afterwards.
25A · Later feature repair. Compare A and B, naming a feature in each before a possible effect. Teacher asks for a particular edge or position. Notice evidence rather than vague praise. After teaching, use unused R1 at another sitting if useful.
25B · Later description repair. Use an own earlier composition; describe objects/relationships without inferred motives. Teacher asks which words another person could check. Notice a precise representation. Next offer unused R2, keeping its answer private.
25C · Later interpretation choice. Use W10's ordinary bag design and propose a different supported reading. Teacher asks what remains uncertain. Notice an interpretation with a feature. Then offer R1/R2 later, only if their sources remain unused.
Capture only useful evidence: actual learner choice, actual child/adult/tool action, first versus supported later response, and the next useful teaching move. No thirty-route log, photograph or public audience is needed.