Improve the Capability, Not the Label
An Achievement Level is an outcome. It does not identify which capability is limiting performance. Two pupils with the same overall result can need completely different repairs.
A useful improvement sequence is:
Baseline → Observe → Locate Bottleneck → Intervene → Retest → Transfer → Maintain.
1. Build a Baseline From Real Output
Use recent schoolwork, writing samples, comprehension responses, oral practice and timed papers. Look for repeatable patterns rather than one bad day.
| Evidence | Question |
|---|---|
| Writing | Where does meaning become unclear or structure collapse? |
| Comprehension | Is the failure vocabulary, inference, evidence selection or answer construction? |
| Grammar | Does the rule fail in isolation or only under writing load? |
| Oral | Is the bottleneck idea generation, language, fluency or interaction? |
| Timed work | What disappears when sustained load is added? |
2. Find the Earliest Bottleneck
A visible error may be downstream from the real problem. Repeated weak vocabulary in composition may be a retrieval problem. A rushed ending may begin with an oversized plan. Grammar errors late in the paper may be load failure rather than missing grammar knowledge.
Repair the earliest layer that explains the most downstream symptoms.
3. Choose One Targeted Intervention
Do not prescribe “more English” when the evidence points to one specific weakness.
- Vocabulary retrieval weak → retrieval and context practice.
- Sentence relationships weak → clause and grammar reconstruction.
- Comprehension evidence weak → evidence-location drills.
- Writing route weak → idea selection and causal skeletons.
- Time failure → checkpoint and drift-control practice.
4. Retest the Same Capability on a Changed Surface
After intervention, use a different passage, prompt, sentence set or oral stimulus that still requires the repaired capability.
If the improvement appears only on the training worksheet, the skill may not yet have transferred.
5. Do Not Map Personality or Study Hours to AL Bands
It is not valid to say that an AL1 pupil necessarily studies a certain number of hours, uses advanced vocabulary, or has one personality profile while an AL6 pupil has another. Achievement Levels are score bands, not learner archetypes.
Use current output and error patterns instead of assigning fixed characteristics to a grade.
6. Improvement Is Not Always Linear
A learner can improve one capability while total scores remain noisy for a while because other bottlenecks still exist. Track local gains as well as whole-paper performance.
Useful measures include:
- fewer repeated error types;
- faster retrieval;
- greater accuracy under time;
- more independent self-correction;
- transfer to unfamiliar tasks.
7. Separate Foundation, Integration and Examination Failure
| Failure layer | Meaning | Repair |
|---|---|---|
| Foundation | The capability is not yet secure. | Teach and practise the component. |
| Integration | It works alone but fails when combined. | Combine gradually and control load. |
| Examination | It works in practice but fails under time/state pressure. | Simulate, monitor and stabilise execution. |
8. Parents Need a Smaller Feedback Language
Replace “Your English is weak” with an observable statement such as “Your evidence is correct, but the answer does not explain the inference” or “Your vocabulary is available in exercises but disappears in composition.”
Smaller diagnosis produces smaller, more actionable repair.
9. The Improvement Check
BASELINE — What can the learner do now?
BOTTLENECK — What earliest weakness explains the failure?
INTERVENTION — What is the smallest targeted repair?
RETEST — Does the repaired skill survive a changed task?
TRANSFER — Does it work under full paper load?
MAINTAIN — Is it still available after delay?
The Exit Condition
Performance is improving when interventions are chosen from evidence, repeated weaknesses shrink, repaired capabilities transfer to unfamiliar tasks and the learner increasingly knows what to fix without needing a generic grade-based programme.