A child can answer literal, vocabulary or inference questions well in isolation and still lose accuracy when those question types are mixed. Mixed comprehension adds a selection problem: the learner must identify what each question wants before choosing the right reading process.
This page uses Jiemin Primary School only as the school-family context named in the preserved 2019 URL. Its reader job is when comprehension accuracy drops as question types mix.
The old page contained historic Yishun address, tutor, telephone and generic PSLE sales claims. Those are not carried forward. eduKate Singapore is an independent tuition provider and is not affiliated with Jiemin Primary School, MOE or SEAB.
Mixed Questions Remove the Topic Label
In isolated practice, the learner already knows the job: “today is inference”. In a real passage, the next question may be literal, then vocabulary, then evidence, then comparison.
| Question type | Primary reading job |
|---|---|
| Who/what/where | Literal localisation |
| Why/how | Cause, motive or inference |
| What does X mean? | Vocabulary in context |
| What showed…? | Evidence selection |
| How are A and B different? | Comparison across two regions |
The First Repair Is Question-Type Recognition
Before searching the passage, the child should say what the question requires.
READ QUESTION → NAME JOB → CHOOSE STRATEGY → SEARCH → ANSWER.
Literal Questions
Literal questions usually require direct information. The student should resist adding an inference when the answer is already stated.
Inference Questions
Inference requires combining clues into a bounded conclusion. The child should find evidence first and avoid treating plausible guesses as supported meaning.
Vocabulary-in-Context Questions
Students should use the local sentence, grammar, collocation and wider passage meaning. The first memorised dictionary meaning may not fit.
Evidence Questions
Evidence questions ask the learner to support a claim. The strongest answer uses the nearest sufficient detail rather than copying a whole paragraph.
Comparison Questions
Comparison requires both sides at the same level. One detailed side and one vague side produces an incomplete answer.
Why Accuracy Drops in Mixed Sets
- student applies the previous strategy to the next question;
- question verbs are not read carefully;
- learner searches by keywords instead of relationship;
- working memory is overloaded by switching;
- one weak question type disrupts confidence for later items.
Use Contrast Pairs Before Full Mixing
Mix two question types first:
- literal vs inference;
- inference vs evidence;
- vocabulary vs reference;
- cause vs consequence.
Ask the student to state the discriminating feature before answering. Then add more types.
The Mixed-Question Check
- What does the question ask me to do?
- Where should I begin searching?
- Do I need literal information or an inference?
- How much evidence is enough?
- Does my answer satisfy every required part?
A Mixed-Accuracy Diagnostic
| Pattern | Likely issue |
|---|---|
| Strong isolated, weak mixed | Strategy selection |
| Literal answers become over-inferred | Overgeneralising inference strategy |
| Evidence question answered with claim only | Answer contract |
| Accuracy falls after one hard question | Recovery/performance control |
How a 3-Pax Class Trains Mixed Selection
In a maximum three-student group, students can classify the same set of questions before answering. The tutor can see whether the error is in question recognition, evidence search or written response.
Current MOE English Context
The MOE Primary English Language Syllabus 2020 develops students’ ability to use reading strategies flexibly across texts and purposes. Mixed comprehension is valuable when it tests strategy selection rather than random guessing.
School Context Boundary
This article uses Jiemin Primary School only as the school-family context named in the historical URL. It does not imply endorsement, partnership or affiliation.
Signs Mixed Accuracy Is Improving
- Question types are recognised faster.
- Strategies switch cleanly.
- Literal questions stay literal.
- Inference becomes better supported.
- Comparison answers contain both sides.
- One hard item causes less downstream accuracy loss.
Mixed Question Types: Almost-Code Summary
QUESTION:
identify_type()
identify_answer_contract()
SELECT_STRATEGY:
literal
vocabulary
inference
evidence
comparison
SEARCH:
appropriate_evidence_radius()
ANSWER:
complete_and_calibrated()
OUTPUT:
stronger_mixed_comprehension
better_strategy_selection