How to perform in the new G2 SEC Science examination depends on more than knowing Science. The learner must also know what kind of answer the question is asking for. A student can understand the topic, use correct vocabulary and still lose marks by describing when the question asks for an explanation, explaining when only an observation is needed, or giving a prediction without the evidence or principle that supports it.
This twenty-fourth Learner’s Guide focuses on Science command words. The central rule is: the command word defines the job. Before writing any scientific content, identify whether the question asks you to state, name, describe, explain, compare, predict, suggest, calculate, determine, justify, evaluate or conclude. Then build the answer to fit that job.
For 2027, SEAB lists G2 Science as K223 Science (Physics, Chemistry), K224 Science (Physics, Biology) and K225 Science (Chemistry, Biology). The shared G2 Science assessment objectives include knowledge with understanding, handling information and solving problems, and experimental skills and investigations. Use the current SEAB 2027 G2 syllabus page and the learner’s actual syllabus for official details.
Command Words Are Not Decorations
A command word is an instruction about the shape of the answer. It tells the learner whether the examiner wants a fact, pattern, mechanism, choice, calculation, comparison or judgement.
Ignoring the command word is like solving the wrong Mathematics problem correctly. The science may be true, but the response is not aligned to the task.
The First Five Seconds
Before writing, spend a few seconds on three questions:
- What is the command word?
- What evidence or information is given?
- What kind of scientific response would complete this job?
This tiny pause prevents a large number of avoidable errors.
State
“State” usually asks for a concise fact, quantity, relationship or conclusion. Do not turn a state question into a long explanation unless the question has another part that requests reasoning.
A strong state answer is direct and scientifically precise. If the question asks for a unit, include the unit. If it asks for a variable, name the variable.
Name
“Name” is even narrower. Give the required term, structure, process, substance, apparatus or quantity. Additional explanation may be unnecessary and can introduce error.
This is where knowing precise scientific vocabulary matters. If the term is “diffusion”, writing a long description of particles moving may not substitute for the requested name.
Identify
“Identify” asks the learner to select the relevant item, feature or relationship from the information provided. The answer may be a label, value, pattern or component.
Do not confuse identification with explanation. First identify the item. Add reasoning only if another part asks why.
Describe
“Describe” asks what happens, what is observed or what pattern the data shows. A description should stay close to evidence.
- increase, decrease or remain constant;
- rise then plateau;
- change colour;
- form a precipitate;
- move faster or slower;
- have a higher or lower value;
- show an anomaly at a particular point.
A description does not automatically require the cause. If the graph rises with increasing temperature, describing the rise is different from explaining the molecular reason.
Describe a Trend Precisely
Avoid vague language such as “it changes a lot”. State the direction, range or shape. When useful, support the description with data values.
A good trend statement might identify that Y increases as X increases, then levels off after a certain point. The exact wording should match the data rather than a memorised template.
Explain
“Explain” asks why. The answer needs a mechanism or causal relationship. A description alone is incomplete.
Use the structure condition/evidence → scientific concept → mechanism → outcome. The written answer may be one or two sentences, but the reasoning should contain the causal bridge.
The Missing-Middle Problem
Many weak Science answers jump from condition to outcome. “The temperature increases, so the reaction is faster.” This may be directionally correct but can miss the mechanism expected at the level of the question.
The middle is where the scientific explanation lives: particle motion, collision frequency, energy transfer, pressure relationship, enzyme activity, diffusion gradient or another relevant mechanism.
Compare
“Compare” requires at least two sides. A learner who describes only A has not compared A with B.
Use paired language: higher than, lower than, faster than, similar to, greater in A but smaller in B, increases in both but more rapidly in one.
The comparison should use the same feature on both sides. Comparing the colour of A with the mass of B is not a valid comparison unless the question specifically links them.
Contrast
“Contrast” focuses on differences. Like compare, it needs both sides, but the answer should emphasise how they differ.
A precise contrast may be shorter than a full comparison if the shared similarities are irrelevant to the task.
Calculate
“Calculate” requires a valid mathematical relationship, correct substitution, controlled units and an appropriately presented final answer.
Do not give only the calculator output when the working is meaningful. Identify the relationship, show the important substitution and state the unit.
Determine
“Determine” can involve calculation, graph reading, deduction or a combination. The learner must use the provided evidence to obtain the requested value or conclusion.
Read the method of determination carefully. A graph may be the intended source rather than an algebraic formula.
Predict
“Predict” asks what is expected to happen based on a trend, model or scientific principle. A prediction is not a guess.
The learner should identify the relevant pattern or concept, then state the expected outcome. If the question asks for a reason, include the justification; if it asks only for the prediction, keep the answer concise.
Suggest
“Suggest” usually appears when more than one scientifically reasonable answer may be possible. The learner should propose a plausible answer based on the information given and relevant scientific knowledge.
“Suggest” does not mean “guess freely”. The answer still needs a scientific basis. If an unfamiliar organism, apparatus or material is involved, use the evidence in the question to constrain the suggestion.
Why Suggest Questions Feel Difficult
Students often search memory for the exact textbook sentence and become stuck when no sentence matches. Suggest questions reward transfer. Ask what known principle could explain the evidence, then propose the most defensible answer.
The learner should be willing to use first principles rather than wait for recognition of a memorised example.
Give a Reason
“Give a reason” is more direct than “explain”, but it still requires a causal or logical basis. State the reason clearly and connect it to the fact or choice in the question.
Avoid circular reasoning: “It is safer because it is less dangerous.” The reason should add scientific meaning.
Justify
“Justify” asks the learner to defend a choice, conclusion or decision with evidence or scientific reasoning. A justification usually contains a claim plus the reason that makes the claim acceptable.
If selecting apparatus, justify why it is suitable for the required range or resolution. If choosing a conclusion, justify it with the relevant pattern or data. If selecting a material, justify the choice with its required property.
The Claim-Evidence-Reason Pattern
- Claim: state the choice or conclusion.
- Evidence: identify the relevant data, observation or property.
- Reason: explain why that evidence supports the claim.
Not every justify question needs three separate sentences, but the thinking should contain all three.
Evaluate
“Evaluate” asks for judgement. The learner must consider evidence, strengths, limitations or whether a conclusion is well supported.
Evaluation is not a list of random problems. It should judge the quality of the method, data, claim or conclusion against a scientific standard.
Evaluate an Experimental Method
When evaluating a method, consider whether the design tests the intended relationship, whether key variables are controlled, whether measurements are suitable, whether repeats are used where appropriate and whether important sources of error remain.
Then state the significance of the weakness: how might it affect the data or conclusion?
Evaluate a Conclusion
A conclusion may fit the data but still be too broad. Ask whether the evidence covers enough conditions, whether anomalies exist, whether another variable could explain the result and whether correlation has been mistaken for causation.
A strong evaluation can agree with part of a conclusion while limiting its scope.
Conclude
“Conclude” asks the learner to state what the evidence supports. A conclusion should answer the investigation aim or question and stay within the tested conditions.
Do not restate one reading. Identify the relationship or finding that the readings collectively support.
Infer
“Infer” asks for a conclusion that is not stated directly but is supported by evidence. This is similar to inference in English comprehension, but the scientific inference must fit scientific principles and the data.
The learner should make the smallest justified move beyond the observation.
Deduce
“Deduce” also requires reasoning from given information, often through a more explicit logical or mathematical chain. The answer should be traceable from the evidence or relationships supplied.
Outline
“Outline” usually asks for the main features or steps without full detail. Give enough structure to show the process, but do not turn the response into a long explanation unless required.
Define
“Define” asks for a precise scientific meaning. Definitions should use the essential relationship or property that distinguishes the term from nearby concepts.
Do not replace a definition with an example. An example can help during learning but may not answer the examination command.
Distinguish
“Distinguish” asks how two things differ. State a clear difference using the same feature on both sides.
A good distinguishing statement makes it impossible to confuse the two terms after reading the answer.
Relate
“Relate” asks the learner to connect variables, ideas or observations. State how one changes with or influences the other and use data or mechanism when required.
This command often appears in questions involving graphs, rates and biological or physical relationships.
Sketch
“Sketch” asks for the important form rather than exact plotted detail. Axes, labels and the broad shape or key features still matter. The learner should understand what features carry the scientific meaning.
Plot
“Plot” is more exact. Use the correct axes, scale and data points, then draw the required line or curve appropriately. The command is graphical, so a verbal description alone is not enough.
Measure
“Measure” requires reading or obtaining a value using the provided instrument, scale, diagram or method. Unit and sensible precision matter.
Record
“Record” asks for observations or data in the required form. Do not convert an observation into an explanation unless the question asks for interpretation.
Observe
An observation is what is detected or measured: colour change, precipitate, gas, temperature change, movement, reading or another direct result. An inference is what the observation suggests. Keep the two separate.
Observation Versus Explanation
This distinction is one of the most important in Science. “The limewater turned milky” is an observation. “Carbon dioxide was present” is an inference based on the test. The learner should answer the command at the correct level.
Command Words and Data
When data is supplied, the command word tells the learner what to do with it. “Describe” means report the pattern. “Explain” means connect the pattern to mechanism. “Predict” means extend the pattern or principle carefully. “Evaluate” means judge how strongly the data supports a claim.
The same graph can therefore generate very different correct answers depending on the command.
Command Words and Calculations
A numerical question may ask to calculate, determine, compare, estimate or justify. These commands are not interchangeable. A calculated number may be only the first step in a justification or comparison.
Always return to the command after the arithmetic is complete.
The Command-Word Ladder
Level 1 — recognise the command
Underline or mentally repeat the instruction before reading the rest of the question in detail.
Level 2 — identify the answer type
Fact, pattern, mechanism, calculation, prediction, suggestion, comparison or judgement?
Level 3 — select the evidence
Choose the data, observation, principle or property needed for that answer type.
Level 4 — build the response
Use the appropriate structure: direct fact, evidence-plus-meaning, mechanism chain, calculation, comparison pair or claim-evidence-reason.
Level 5 — check job completion
Reread the command word and ask whether the answer actually performs that job.
The Wrong-Job Error Ledger
- described when asked to explain;
- explained when asked only to state;
- predicted without using a trend or principle;
- suggested an answer with no scientific basis;
- compared only one side;
- calculated a value but did not complete the requested judgement;
- justified a choice without evidence;
- evaluated by listing weaknesses but never judging significance;
- concluded beyond the tested data;
- gave an inference when the question asked for an observation.
These errors are powerful because they often occur even when the learner knows the Science. Fixing command-word control can therefore recover marks without learning a new chapter.
The 10-Minute Command-Word Drill
- Take ten Science questions from mixed topics.
- Do not answer them yet.
- Classify each command word.
- Write the expected answer type beside each question.
- Only then answer five of them.
- Check whether each response matches the classified job.
This separates question interpretation from content recall and makes hidden command-word weaknesses visible.
The Same-Data Drill
Use one graph or table and create several questions from it: describe, explain, predict, calculate, justify and evaluate. Answer each one.
This drill shows that the evidence can stay the same while the required thinking changes completely.
The Same-Topic Drill
Take one topic such as diffusion, pressure, reaction rate, respiration or energy transfer. Write a state question, a describe question, an explain question and a predict question on the same concept.
The learner begins to see command words as different operations applied to the same knowledge.
The Compare Drill
Present two systems, graphs, organisms, materials or experimental conditions. Require the learner to write paired sentences using the same feature on both sides. Then remove any sentence that describes only one side.
The Explain Drill
Use short prompts that already contain the observation. The learner must add only the mechanism. This prevents the common habit of repeating the observation in longer words and calling it an explanation.
The Suggest Drill
Use unfamiliar but syllabus-linked contexts. Require the learner to identify one known principle before proposing an answer. This makes the suggestion evidence-based rather than imaginative guessing.
The Justify Drill
Give two possible choices of apparatus, material, method or conclusion. The learner must select one and provide the property, data or relationship that makes it the better choice.
The Evaluate Drill
Give a short experimental method and data set. Ask the learner to identify one strength, one limitation, the consequence of the limitation and one targeted improvement. Then ask whether the final conclusion is supported.
This trains judgement rather than memorised criticism.
A Four-Week Command-Word Build
Week 1 — state, name, identify and describe
Build precision at the factual and observational level. Learners should become comfortable giving short complete answers without unnecessary explanation.
Week 2 — explain, compare and relate
Focus on causal mechanisms and paired relationships. Require clear middle steps rather than jumps from cause to outcome.
Week 3 — predict, suggest and justify
Use unfamiliar contexts and evidence-based transfer. Require the learner to state which principle supports the prediction or suggestion.
Week 4 — evaluate, conclude and mixed timed practice
Use experiments, graphs and structured questions under time. Review wrong-job errors separately from knowledge errors.
Physics Command-Word Control
In Physics, “calculate” and “determine” may involve formulas or graph reading, while “explain” often requires a relationship between physical quantities and a mechanism. “Predict” should follow a model or trend, and “justify” may require a property or numerical comparison.
Because Physics uses many quantities, units and graphs, the learner should let the command word decide whether the final response is numerical, comparative or explanatory.
Chemistry Command-Word Control
In Chemistry, distinguish observations from explanations carefully. “Describe what you observe” asks for colour, precipitate, gas, temperature or another directly detected change. “Explain” asks for the particle, reaction or energy reasoning behind it.
“Suggest” questions may require using unfamiliar evidence to infer a substance, process or reason. Stay within the evidence supplied.
Biology Command-Word Control
In Biology, “describe” may ask for a pattern in a graph or a structural feature, while “explain” requires the process connecting structure to function or condition to biological outcome. “Predict” should follow the known process or data trend.
Long biological vocabulary does not replace the causal chain. The command word still controls the answer.
Command Words Under Time
When the paper is moving quickly, students often skip the command word because the topic looks familiar. Build a habit of reading the verb before the nouns. “Explain osmosis…” and “Describe osmosis…” are different tasks even though the topic is the same.
The first five seconds can save the next two minutes.
Command Words and Examination Recovery
If a Science question feels blocked, return to the command word before searching for more content. Sometimes the learner knows the topic but is trying to answer at the wrong level. Vol 0021 explains how changing the task representation can restore progress.
Command Words and Experimental Questions
Use Vol 0020 for variables, measurement quality and method improvements. Command-word control determines whether you should state a variable, explain why it is controlled, suggest an improvement or evaluate the method.
Command Words and Data
Use Vol 0012 for calculations, graphs and data. This guide tells you what operation to perform on that evidence.
The PSLE Bridge
The PSLE rule Evidence Before Explanation remains central. At G2, the learner adds command control: evidence before explanation when asked to explain, evidence before prediction when asked to predict, and evidence before judgement when asked to justify or evaluate.
Use Examination Craft for Whole-Paper Control
For time management, return decisions and checking routines, continue through the Examination Craft hub. Command-word discipline should remain stable even when the learner is tired.
Final Rule
Before you answer the Science, answer the question about the question: what job is this command word asking me to perform?
State when asked to state. Describe what the evidence shows. Explain the mechanism. Compare both sides. Predict from a trend or principle. Suggest from evidence. Justify with a reason. Evaluate the quality of evidence. Conclude only what the data supports. Knowing Science is the foundation; answering the correct scientific job is the performance.
Command Words Can Change Within One Question
A structured question may move through several command levels: first state an observation, then explain it, then predict what would happen under a new condition, then suggest an improvement to the method. The learner should reset at every subpart rather than carrying the response style from part (a) into part (d).
This is a common advanced error. The topic remains the same, so the learner assumes the answer job also remains the same. It does not.
The Subpart Reset
- Read the new command word.
- Identify whether the evidence has changed.
- Decide whether the answer should be factual, comparative, causal, predictive or evaluative.
- Answer only that new job.
- Check whether information from an earlier subpart is still relevant.
The reset takes seconds and prevents an entire multi-part question from drifting off-task.
Command Words and Marks
Do not create a universal rule such as “explain always needs three sentences” or “state is always one mark”. The paper controls the mark allocation. A command word tells you the type of thinking; the specific question tells you how much of that thinking is required.
Use marks as one clue to expected depth, but never replace careful reading with a words-per-mark formula.
Command Words and Evidence Density
Some questions contain abundant evidence while others contain almost none. A “describe” question with a graph may need values to support the pattern. A “describe” question about a visible laboratory observation may need only the observation itself.
The learner should use the amount and type of evidence relevant to the job, not force data into every answer.
Explain With the Correct Level of Science
An explanation should use the level of mechanism expected by the syllabus and question. Adding concepts from beyond the course can make an answer longer without making it more correct. Use the simplest scientific model that fully explains the evidence.
Advanced performance is not displaying the most terminology. It is selecting the right mechanism at the right depth.
Predict With Boundaries
A prediction should be limited by the available trend or scientific principle. If data covers a narrow range, extending the pattern far beyond that range may be unjustified. If a biological system has an optimum, a simple straight-line extrapolation may fail.
Strong predictions recognise the evidence boundary.
Suggest More Than One Possibility Only When Useful
A suggest question may allow several plausible answers, but the learner does not automatically gain by listing many. Give the strongest scientifically supported suggestion unless the question asks for alternatives or more than one idea.
Several weak guesses can dilute one good answer.
Justify Apparatus Choice
A good apparatus justification connects the measurement job to instrument properties. For example, the chosen device may have a suitable range, finer resolution, lower reaction-time dependence or a form that allows direct measurement of the required quantity.
“It is more accurate” is too vague unless the answer explains what aspect of the apparatus makes the measurement better suited to the task.
Justify a Scientific Conclusion
If asked to justify a conclusion, use the data or observation explicitly. State the conclusion, identify the supporting evidence and explain why that evidence points toward the conclusion rather than a competing interpretation.
Evaluate Without Listing Everything
Evaluation is selective judgement. Identify the limitation that most affects the quality of the evidence or conclusion, then explain its consequence. A long list of minor flaws can be weaker than one well-explained major limitation.
The learner should prioritise significance, not quantity of criticism.
Conclude Without Overclaiming
A conclusion should match the experimental design and tested range. If only two conditions were compared, do not claim a universal law. If the sample was small, avoid broad statements about an entire population. If correlation was observed, do not automatically assert causation.
Scientific restraint is part of a strong conclusion.
Infer From Structure, Not Imagination
Inference questions often include enough clues for a careful learner even when the context is unfamiliar. Look at measurements, direction of change, structure, location, function or comparison. Use the clues to build the smallest defensible inference.
If the inference requires an assumption not supported by the question, label the assumption during practice and reconsider the answer.
Command Words in Physics
- Describe: report the pattern, motion or change in quantities.
- Explain: connect the change to a physical principle or relationship.
- Calculate: show the relevant relationship, substitution and unit.
- Predict: use the model or graph trend to state what should happen.
- Justify: defend a numerical choice, apparatus or interpretation with a property or relationship.
Physics often makes the command-word distinction visible because numbers and mechanisms sit close together.
Command Words in Chemistry
- Observe: colour, precipitate, gas, temperature or another directly detected change.
- Explain: particle, bonding, reaction or energy mechanism.
- Suggest: infer a plausible substance, process or reason from evidence.
- Compare: use the same chemical feature for both conditions.
- Evaluate: judge method quality, evidence strength or limitations.
A Chemistry answer can be wrong even when the chemistry fact is true if the command asks for observation and the learner writes inference instead.
Command Words in Biology
- Describe: state a structural feature or data trend.
- Explain: connect structure, process and outcome.
- Predict: use the biological process or data pattern under a changed condition.
- Suggest: use evidence to propose a reason or adaptation.
- Evaluate: judge evidence from biological investigations where variation may matter.
Biology often tempts students to write everything they know about a process. Command-word control keeps the response narrow enough to score efficiently.
The Wrong-Verb Correction Drill
Take a completed Science paper and ignore the marks initially. For every wrong or weak answer, ask whether the scientific content was wrong or the command word was answered incorrectly. Create two columns: knowledge error and job error.
This prevents unnecessary relearning when the real weakness is response control.
The Command-Word Rewrite Drill
Choose one correct answer and rewrite it for a different command word. Turn a description into an explanation. Turn an explanation into a prediction under a changed condition. Turn a conclusion into a justification.
This teaches the learner how the same science can produce different answer forms.
The Timed Verb-First Drill
During a ten-minute mixed set, the learner must write the command word in the margin before answering. After several sessions, remove the written cue but preserve the mental habit.
The aim is automatic command recognition without wasting examination time.
Advanced Standard: Same Knowledge, Different Jobs
A mature G2 Science learner can take one concept and use it to state a fact, describe a pattern, explain a mechanism, predict an outcome, suggest a reason, justify a decision or evaluate evidence. The knowledge remains stable while the operation changes.
That flexibility is one of the clearest signs that Science has moved from memorised content into transferable examination performance.
A Final Command-Word Checklist
Before leaving a substantial Science response, check the verb again. The learner should be able to answer one question quickly: did I perform the job the command word asked for?
- State/name/identify: is the answer direct and precise?
- Describe: did I report what happens or what the data shows?
- Explain: did I include the mechanism, not only the observation?
- Compare: did I address both sides using the same feature?
- Predict: is the outcome supported by a trend or scientific principle?
- Suggest: is the proposal scientifically plausible and evidence-based?
- Justify: did I support the choice or conclusion with a reason or evidence?
- Evaluate: did I judge significance rather than merely list features?
- Conclude: is the claim no broader than the evidence allows?
This checklist is short enough to use during timed practice and selected high-value examination questions.
Why Command-Word Control Improves Revision
When a learner reviews mistakes by topic only, an answer may be labelled “weak in respiration” or “weak in electricity” even though the scientific knowledge was correct. Separating knowledge errors from command-word errors produces better revision. A knowledge error needs relearning; a job error needs response practice.
This distinction prevents unnecessary chapter revision and directs attention to the actual source of lost marks.
Why Command-Word Control Improves Transfer
Unfamiliar contexts become less intimidating when the learner knows the operation to perform. Even if the apparatus or organism is new, “describe” still means report the evidence, “explain” still means give the mechanism and “suggest” still means propose a defensible answer from known science.
Command words therefore act as anchors when surface details change.
Why Command-Word Control Improves Checking
A final Science check should not ask only whether the facts are correct. It should ask whether the response form matches the instruction. This catches a class of errors that factual checking alone cannot detect.
The learner may know every sentence in the answer and still have answered the wrong job. The verb is the fastest way to detect that problem.
The Final Standard
Advanced G2 Science performance means the learner can move from the same body of knowledge into several response forms without confusion. They can observe without over-explaining, explain without merely repeating, predict without guessing, justify without hand-waving and evaluate without listing random weaknesses.
The content matters. The evidence matters. But the command word decides what the final answer must do.
