Six structure, referent, insertion, and order errors accounted for 16.2 points lost.
HIGH-SCHOOL EXAM · CUMULATIVE ERROR INTELLIGENCE
Do not stop at redoing mistakes. Turn recurring errors into the priority of the next test.
HAOLLA tracks recurring errors in question polarity, referents, sentence function, condition checks, and surface-keyword traps. Recovered weaknesses move to maintenance; recurring ones receive more weight in the next review and the next mock exam. This is a real Grade 10 English record with student and school identifiers removed.
Published2026.10.06
Anonymized caseGrade 10 · English
PrivacyIdentifiers removed
MethodEvidence + limits
Efficiency improves when the next 20 minutes are chosen by current evidence, not by a generic “study more” plan.
HAOLLA stores an error as error type → faulty decision process → missing evidence → recurrence → next verification method. Recovered patterns receive less repetition; persistent patterns shape the next material and mock exam.
LONGITUDINAL PATTERN
The most useful trend is how the error mix changes.
The mock exams were not equated for difficulty, so scores are not treated as a linear growth measure.
→
The only error was pronoun/referent tracking; the prior cluster had largely disappeared.
→
No insertion/order/referent errors. New errors appeared in negative stems, concept definition, sentence/discourse function, and explicit conditions.
What changed
The system shifts attention when an old bottleneck improves and a new one becomes visible.
CUMULATIVE STUDENT ERROR LOOP
Evidence → Classify → Accumulate → Prioritize → Rebuild → Retest
Each new result can change the weight of the next review, assignment, and mock exam.
1
Evidence
Keep the first choice, evidence, trap, and lost-point type.
2
Classify
Convert a miss into a reusable error class.
3
Accumulate
Label repeat, improvement, or relapse across rounds.
4
Prioritize
Move current high-loss errors to the top of study time.
5
Rebuild
Adjust the next material around the current pattern.
6
Retest
Verify recovery and update the record again.
MINIMUM-TIME PRIORITY
The final 25 minutes target only the places where points are currently leaking.
The allocation changes when the student's pattern changes.
6 minConcept recallSeparate similar concepts by definition, purpose, timing, and language signal.
8 minThree hard reading itemsExplain contrast, main claim, and discourse function before options.
4 minNegative stemsPractice NOT / least consistent / cannot be answered with an X-finding routine.
4 minCondition auditBreak notices into Who / What / When / Format / How.
3 minReferent safety checkSubstitute referents for it / they / this / which to prevent relapse.
5-LOCK
Protect a correct answer
- 1Question polarity
- 2Subject-action / claim-result skeleton
- 3One exact evidence pinpoint
- 4One decisive flaw in the best distractor
- 5Reset between questions sharing the same passage
OPEN EVIDENCE
The analysis structure is public; the student and school are not.
The 5-page Korean PDF contains the cumulative pattern, 5-Lock, linked-passage reset, time management, and the final 25-minute routine. Direct student and school identifiers were removed.
EVIDENCE & LIMITS
We publish the limits with the numbers.
72.3 / 97.3 / 85.7 are different HAOLLA mock exams; difficulty was not equated.
The structure/referent cluster moved from six items to one to zero.
The public page and PDF omit student and school identifiers.
Outcomes vary by starting point, time, scope, and individual performance.
The evidence is from English, but the cumulative-error structure can be adapted to other school subjects.
The unit of error must change by subject: concepts and solution steps in math; concept relations, units, graphs, and experiment conditions in science; scope, causality, sources, and chronology in social studies and Korean history.
AEO / DIRECT ANSWERS
FAQ
Is this just an error notebook?
No. The reusable unit is the recurring decision error, not only a wrong question number.
Does this guarantee less study time and higher scores?
No. The aim is to reduce unnecessary repetition and focus time on current point leakage.
Why did a stored score move from 97.3 to 85.7?
The exams were not equated for difficulty. The more useful observation is that older structure/referent errors disappeared while new higher-order errors surfaced.
Are real student and school details public?
No. Direct identifiers were removed from the public case and PDF.
Can this work in other subjects?
The cumulative loop can be reused, but each subject and confirmed exam structure requires its own error taxonomy.
START FROM ONE REAL TEST
One recent test is enough to start mapping recurring point leakage.
Use one recent exam, mock, or assignment to identify the current bottleneck and the first action for the next week.
This is an independent HAOLLA educational case, not an official school resource and not a claim of access to future exams.