Closing the book reveals whether reading has become retrievable English.
Understanding a page while it is open and reconstructing the event sequence in your own English are different performances. These two de-identified cases separate content recall, sequencing, sentence control, vocabulary, and expression using only supported learner evidence.
We publish the learning bottleneck, not a polished success story.
The two learners are not compared with each other. Each score is tied to the evidence available in that learner’s submission and is used to choose the next action, not to promise another learner’s outcome.
Key characters are recalled, but sentences break before the event is complete
Core story elements such as Jack, Annie and a Viking were retrieved, but sentence completion and event linking broke down when a noun or connector could not be recalled.
5 pages · direct identifiers removed
More content is retrieved, but sentence form becomes the bottleneck
Across two presentations, the learner retrieved more information about appearance, action and reaction. Subject-verb agreement, past tense and active/passive form remained unstable.
7 pages · direct identifiers removed
These scores and cases document specific submissions at a specific point in time. They do not guarantee the same score, timeline or outcome for another learner.
Grammar, vocabulary and expression are corrected from the learner’s own wording.
Feedback does not stop at a general comment. Supported issues are converted into sentence forms the learner can reuse in the next retell.
Read less passively; reconstruct more actively.
Break the source into manageable meaning units rather than holding a long section at once.
Recall who, what happened, and what came next without looking at the source.
Use key nouns and verbs to restore the event sequence instead of memorizing long sentences.
Use the actual speaking attempt to locate recall and sentence-generation bottlenecks.
Repair grammar, vocabulary and expression, then retell the same content more briefly and accurately.
We separate research support from first-party learner evidence.
These HAOLLA cases are learner records, not controlled evidence of a universal effect. Retrieval practice and story retelling, however, are independently studied in education research.
IES · Retrieval-oriented learning
An IES research program on actively retrieving previously learned information rather than relying only on passive review.
Source ↗WWC · Story retelling
A What Works Clearinghouse review record for research examining story retelling, comprehension, story structure and oral language.
Source ↗Frequently asked questions
Does repeated reading automatically produce strong retelling?
Not necessarily. Recognizing meaning while the text is visible and reconstructing the event without the text are different performances, so HAOLLA checks them separately.
How are grammar, vocabulary and expression handled?
Only supported learner wording is used. When an issue is observed, the report shows the original form, an improved form, and the reason so the learner can reuse the pattern.
Can CASE A and CASE B scores be compared?
No. The available speech evidence and submission lengths differ. The scores are diagnostic within each learner, not rankings between learners.
Do the public materials contain student identity information?
No. The public PDFs replace names with CASE A and CASE B and do not publish school, contact, face, or other direct identifiers.
One recent retelling is enough to find the first bottleneck.
Send one recent presentation or assignment. We identify whether the first priority is recall, sequencing, grammar, vocabulary or expression and connect it to the next task.