To translate modal verbs accurately, you have to preserve more than dictionary meaning. Words such as can, could, may, might, must, should, will, would and need to tell the reader how certain, possible, permitted, necessary, advisable or expected something is. A translation can be grammatically fluent and still be wrong if it turns a possibility into a fact, advice into an order, permission into ability, or a cautious prediction into a guarantee. That is why accurate translation of modality is one of the clearest tests of whether a translator is moving meaning between languages rather than merely replacing words.
People searching for how to translate can, could, may, might, must and should are often facing exactly this problem: the target language does not divide possibility, obligation, permission and certainty in the same way as English. Some languages prefer modal verbs; others use particles, adverbs, verb endings, auxiliaries, sentence-final forms or context. The practical task is therefore not to find one permanent equivalent for each English modal. It is to identify the force of the source sentence and rebuild that force naturally in the target language.
This guide develops a systematic method for translating certainty, possibility, obligation, permission, advice and probability. It belongs to the wider eduKateSG translation architecture, where the central principle is that meaning must survive movement between languages. Vocabulary matters, grammar matters and natural style matters, but modality is the layer that tells us how strongly the speaker commits to what is being said. If that layer drifts, the entire message can change even when every noun and verb looks correct.
50-second quick read
Modality answers questions such as: Is this certain or only possible? Is the person allowed to do it, able to do it, required to do it, expected to do it or merely advised to do it? Before translating a modal expression, classify its function. Then estimate its strength, identify its source, check the surrounding context and choose a target-language form that carries the same practical force. Never assume that must always equals one fixed word, that could is simply the past tense of can, or that may always expresses permission. Read the proposition first, then the speaker’s stance toward that proposition.
Translation is not only about what happened
Consider two sentences: “The train is delayed” and “The train may be delayed.” The event under discussion is the same: a delayed train. But the speaker’s relationship to the claim is different. In the first sentence, the delay is presented as a fact. In the second, it is presented as a possibility. If a translation removes the equivalent of may, the translator has not made a small stylistic change. The translator has increased certainty.
The same principle appears in instructions. “You should submit the form today” is not automatically the same as “You must submit the form today.” The first normally presents advice, expectation or a strong recommendation; the second commonly presents obligation or necessity. In healthcare, law, school instructions, technical procedures and workplace policies, that difference can alter behaviour. The translator therefore needs to preserve not just the proposition—submit the form today—but the status attached to it.
A useful map of modal meaning
A practical translator does not need to begin with a complicated theoretical vocabulary, but a simple map is extremely useful. Most modal meanings in ordinary texts fall into several recurring families: ability, possibility, probability, permission, obligation, prohibition, advice, willingness, prediction, intention, expectation and hypotheticality. One form can serve several families. Can may describe ability (“She can swim”), possibility (“It can get cold here”), or permission (“You can leave now”). Could may describe past ability, weaker possibility, politeness, suggestion or a hypothetical condition. The form alone is therefore insufficient.
It helps to ask two questions. First: what kind of modal meaning is present? Second: how strong is it? “It is possible” and “It is almost certain” both concern probability, but their strength differs. “You may leave” and “You must leave” both affect an action, but one grants permission and the other imposes necessity. Translation becomes much more reliable when function and strength are separated before wording is chosen.
Step 1: isolate the proposition
Every modal statement contains a basic proposition plus a stance toward it. In “The package might arrive tomorrow,” the proposition is “the package arrives tomorrow.” The modal stance is “this is a possibility, but not a certainty.” In “Visitors must wear a badge,” the proposition is “visitors wear a badge.” The modal stance is “this is required.” Separating these layers prevents a common mistake in which the translator becomes so focused on the modal word that the underlying event is lost, or so focused on the event that the modal force disappears.
This technique also helps with languages that do not use a direct modal auxiliary in the same position. Once the translator understands the proposition and the stance separately, the target sentence can distribute the meaning differently. A target language may express possibility through an adverb equivalent to “perhaps,” obligation through an impersonal construction equivalent to “it is necessary,” or politeness through verb morphology. The surface structure can change while the communicative relationship remains stable.
Step 2: identify the source of the force
Not all necessity is the same. “You must wear a helmet because the law requires it” differs from “You must be exhausted after that journey.” In the first, must expresses obligation. In the second, it expresses a strong inference: the speaker concludes that exhaustion is very likely. A word-for-word system that maps both instances to the same target form may produce a strange or incorrect sentence. The translator should ask where the force comes from: a rule, physical circumstances, social expectations, the speaker’s judgment, available evidence or a hypothetical scenario.
Source matters because target languages frequently distinguish these meanings in different ways. A rule imposed by an institution may call for one construction, while a logical conclusion from evidence calls for another. Even in English, compare “You have to stop at a red light,” “You need to drink water,” and “You must have left your keys at home.” All suggest necessity in some broad sense, but the first points to a rule, the second to a practical requirement and the third to inference. Translating them identically can flatten important distinctions.
Step 3: estimate strength instead of chasing one-to-one equivalents
Think of modal force as a scale rather than a set of boxes. For probability, the scale might run from remote possibility through reasonable possibility to strong probability and near certainty. For obligation, it might run from optional through suggested and recommended to expected, required and absolutely necessary. The exact boundaries vary by context and language. The translator’s job is to place the source expression approximately on the correct scale and then choose a target expression occupying a similar position.
This is particularly important because apparently similar words often differ in strength. English should may be mild advice in one sentence and a strong expectation in another. “You should get some rest” recommends an action. “The parcel should arrive by Friday” expresses expectation. “Applicants should attach evidence” may function as an instruction whose practical force is stronger than casual advice. The surrounding genre tells the translator how strongly to read it.
Can: ability, possibility or permission?
Can is deceptively simple. In “Alicia can solve the problem,” it describes ability. In “Glass can break under thermal stress,” it describes general possibility. In “You can use the library after lunch,” it may grant permission. These meanings can overlap in conversation, but a careful translation distinguishes them. If a target language has separate words for being able to do something and being allowed to do something, choosing the wrong one changes the relationship between the speaker and listener.
A useful diagnostic is to paraphrase. If can means “is capable of,” translate ability. If it means “it is possible for,” translate possibility. If it means “is permitted to,” translate permission. This paraphrase test is more dependable than memorising a bilingual dictionary entry because it forces the translator to interpret the sentence before selecting vocabulary. That is the core habit of meaning-first translation.
Could: past ability, possibility, politeness or condition?
Could is even more context-sensitive. “When she was six, she could read independently” refers to past ability. “It could rain tonight” presents possibility. “Could you open the window?” is a polite request. “We could try a different method” offers a suggestion. “If we had more time, we could test another design” expresses a conditional possibility. A translation that treats could mechanically as past-tense can will fail in most of these cases.
The politeness use is especially important. Many languages do not create politeness by shifting a modal into a seemingly more remote form. They may use honorifics, particles, indirect wording or conventional request patterns. The faithful translation is therefore the form that produces comparable politeness, not necessarily a form that resembles English grammar. Functional equivalence matters more than visible similarity.
May and might: permission and probability
May is famous for its split personality. “You may enter” grants permission, while “The results may change” expresses possibility. Might usually marks possibility, often with a weaker or more tentative tone, although real usage is not a perfectly ordered ladder. Context, dialect and register influence the distinction. A translator should resist the temptation to treat may as always stronger than might or to assign fixed percentages to them.
Instead, read the epistemic environment. Does the writer present several possible outcomes? Is the sentence hedging a scientific claim? Is the speaker cautiously predicting? Is the wording part of a formal notice? “This treatment may reduce symptoms” should not become “This treatment reduces symptoms.” The hedge protects the boundary between evidence and certainty. Removing it turns a qualified claim into an unqualified one.
Must: obligation versus inference
Must can be legally or practically forceful. “Employees must display identification” is an obligation. Yet “They must be at home; the lights are on” is an inference. The translator needs to distinguish deontic force—what someone is required to do—from epistemic force—what the speaker believes is strongly supported by evidence. Conflating them can produce a target sentence that sounds like a command when the source is actually a conclusion.
Negation creates another trap. “You must not enter” normally means entry is prohibited. “You do not have to enter” means entry is not required. Those are nearly opposite instructions. The first removes permission; the second removes obligation. When translating safety notices, school rules, contracts or procedures, this distinction deserves an explicit verification pass because a small grammatical error can reverse the practical instruction.
Should: advice, expectation and normative language
Should often sits in the dangerous middle ground between optional and compulsory. In everyday advice—“You should sleep earlier”—it recommends. In predictions—“The bus should arrive soon”—it signals expectation. In standards and policy documents, should may represent a defined level of recommendation that is deliberately weaker than shall or must. The translator must therefore read the document’s own conventions.
Before translating should, ask whether failure to follow the statement has consequences. Is the sentence offering personal advice, describing a likely outcome, expressing moral judgment, or setting a recommended practice? In professional standards, a terminology note or style guide may define the modal vocabulary explicitly. When such definitions exist, they override casual intuition. Consistency becomes part of accuracy.
Will and would: more than future and past
Learners are often taught that will marks the future and would is its past or conditional partner. Translation quickly exposes how incomplete that rule is. Will can express prediction, willingness, habitual behaviour or insistence. “The device will restart automatically” predicts designed behaviour. “I will help” expresses willingness or commitment. “He will leave the door open” can describe a repeated tendency. The target language may encode these through different resources.
Would likewise appears in conditionals, polite requests, reported future, repeated past behaviour and preferences. “Would you mind waiting?” is not principally about hypothetical time; it is a conventional polite request. A natural translation may use a dedicated politeness construction rather than a conditional marker. Once again, function should lead form.
Need to, have to, be supposed to and be expected to
Modal meaning often lives outside classic modal verbs. “Need to” can express necessity; “have to” often expresses externally imposed obligation; “be supposed to” may express expectation, instruction or an unfulfilled norm; “be expected to” may describe institutional or social expectations. If translation analysis looks only for words like must and may, much of the force in real documents will be missed.
Compare: “You need to back up the file,” “You have to back up the file,” “You are supposed to back up the file,” and “You are expected to back up the file.” In some contexts the behavioural outcome is the same, yet the social source and tone differ. The first may be practical advice, the second a requirement, the third a norm or instruction, and the fourth an expectation tied to role. A good translation preserves whichever distinction is relevant to the document.
Modal adverbs matter too
Certainty is frequently carried by adverbs: perhaps, possibly, probably, likely, apparently, presumably, certainly, definitely, clearly. These words interact with modal verbs and with evidence. “It may possibly fail” is more hedged than “It will definitely fail.” “Apparently” signals that the claim is based on appearance or reported information rather than direct commitment. “Presumably” marks a reasoned assumption. Translators who focus only on verbs can overlook these stance markers.
When the target language prefers adverbs over auxiliaries, the modal burden may shift from one grammatical category to another. That is acceptable. What matters is the reader’s resulting interpretation. The target reader should understand approximately the same level of confidence and the same basis for that confidence as the source reader.
Hedges are information, not clutter
Academic and scientific writing often uses phrases such as “the findings suggest,” “may indicate,” “appears to be associated with,” “is likely to,” and “could be explained by.” These expressions are sometimes mistaken for weak writing. In fact, they often encode the strength of the evidence. Removing them during translation can overclaim. Adding extra hedges can understate the source. Both are forms of semantic drift.
A useful rule is to treat every hedge as a candidate information unit. Ask what it contributes. Does it mark uncertainty, limited evidence, politeness, scope or methodological caution? If yes, preserve that contribution even if the target language expresses it differently. Accurate translation of research depends as much on calibrated confidence as on technical vocabulary.
Modal force in instructions and safety language
Instructions require special discipline because modality drives action. “Disconnect the power before servicing” is a direct imperative. “The power must be disconnected before servicing” expresses mandatory procedure. “The power should be disconnected” may be interpreted as recommendation rather than strict requirement unless the document defines should differently. If the source intentionally distinguishes levels of requirement, the translation should preserve that hierarchy throughout the document.
Create a small requirement glossary before translating a long manual. Record how the project will handle terms equivalent to must, shall, should, may, may not, required, recommended, optional and prohibited. Then test sample sentences. This is not merely terminology management; it is behaviour management. Readers use those words to decide what they are allowed, expected or required to do.
Modal force in law and policy
Legal and policy texts can attach highly specific consequences to modal wording. Different jurisdictions and drafting traditions use terms such as shall, must, may and should differently, so translators should not rely on a universal legal rule. The governing document, institutional style guide and target legal convention matter. When the translation is authoritative or high stakes, terminology should be reviewed by a qualified legal-language specialist.
The broader principle remains simple: do not casually strengthen or weaken legal force. If a provision grants discretion, do not translate it as a duty. If it imposes a duty, do not reduce it to advice. If a prohibition applies only under specified conditions, keep those conditions attached to the prohibition. Modal accuracy is part of legal scope.
Modal force in education
Educational texts are full of modal distinctions. A teacher may say students can use a calculator, should show their working, and must submit the task by a deadline. Those verbs encode permission, recommendation and obligation. A translated classroom notice that collapses all three into the same expression creates confusion about expectations.
For language learners, translation is also a powerful way to notice these distinctions. Instead of memorising isolated equivalents, students can compare small sets of sentences and ask what changes when the modal changes. That connects translation to vocabulary depth, grammar awareness and reading comprehension: learners see that knowing a word means knowing how it changes stance, not simply knowing a dictionary gloss.
Modal force in everyday politeness
Requests often use modality to soften social force. “Open the window” is direct. “Can you open the window?” is conventionally a request even though its literal form asks about ability. “Could you open the window?” is often more deferential. “Would you mind opening the window?” uses yet another indirect strategy. A literal translation that asks only about ability may sound odd because the intended speech act is a request.
Translate the social action, not merely the grammatical shell. Determine who is speaking to whom, how formal the situation is, how much imposition is involved and what politeness resources the target language normally uses. The best equivalent may look structurally different while producing the same interpersonal effect.
The danger of making everything too certain
A frequent translation failure is certainty inflation. The source says may, might, appears, suggests or is likely to, but the target sentence becomes a plain assertion. This can happen because the translator wants smoother prose, because machine translation chooses a stronger form, or because the target language handles hedging differently. The result is readable but epistemically inaccurate.
To detect certainty inflation, highlight every stance marker in the source and compare it with the target. Ask: has a possibility become a fact? Has a probability become a guarantee? Has an inference become direct knowledge? Has advice become a rule? This check can be performed quickly and catches errors that ordinary grammar checking will not find.
The opposite danger: weakening the source
Translators can also weaken statements. A source that says “must” may emerge as “should.” A definite prohibition may become a suggestion. A strong conclusion may be softened into vague possibility. Sometimes this happens because the translator prefers a more polite tone or fears sounding too direct. But translation is not the place to renegotiate the speaker’s level of commitment unless adaptation has been explicitly requested.
Politeness and accuracy must be balanced through target-language convention, not by silently changing force. A command can be expressed politely while remaining mandatory. A firm rule can use respectful grammar without becoming optional. Good translation preserves both the interpersonal register and the underlying requirement.
A five-part modal annotation method
For difficult sentences, annotate five things before translating. First, write the proposition in plain form. Second, label the modal function: ability, possibility, permission, obligation, advice, prediction or another category. Third, rate the approximate strength: weak, medium or strong is often enough. Fourth, identify the source of the stance: rule, evidence, speaker judgment, social convention, physical necessity or hypothetical condition. Fifth, note the register: casual, polite, academic, legal, technical or institutional.
This annotation takes seconds once practised. It transforms a vague intuition into an explicit decision. It also creates a useful record when several translators work on the same document because disagreements become diagnosable. One person may have interpreted should as recommendation while another read it as expectation. The team can then resolve the meaning rather than arguing about vocabulary in isolation.
Worked example: translating a cautious claim
Take the sentence: “The new schedule may reduce waiting time, but delays could still occur during peak periods.” The first clause makes a cautious positive prediction. The second preserves a possibility of failure under a specified condition. A poor translation might strengthen both: “The new schedule will reduce waiting time, but delays will occur during peak periods.” That version changes two probabilities into predictions with much greater certainty.
A meaning-first translation keeps both claims calibrated. The target may use modal auxiliaries, probability adverbs or other grammatical devices, but the reader should still receive two messages: improvement is possible rather than guaranteed, and peak-time delays remain possible rather than inevitable. This is what it means to preserve modal geometry across languages.
Worked example: permission versus ability
Consider: “Students can use a bilingual dictionary during the exercise.” Depending on context, can probably grants permission rather than describes physical ability. A translation equivalent to “students are capable of using a bilingual dictionary” would be semantically possible but pragmatically absurd. The exercise rules are about what students are allowed to do.
The diagnostic question is simple: what real-world uncertainty does the sentence resolve? If the reader is wondering whether the rules permit a dictionary, translate permission. If the reader is discussing skill and the sentence means students know how to use the dictionary effectively, translate ability. Context decides the modal reading.
Worked example: recommendation versus requirement
Now compare “Applicants should provide two references” with “Applicants must provide two references.” The difference may decide whether an incomplete application is rejected. In ordinary English, the first sounds recommended or expected, while the second is mandatory. If the source institution uses these terms formally, the distinction should be maintained in every translated version.
A translator who thinks “two references are obviously important” may be tempted to strengthen should into the target equivalent of must. That is an editorial intervention, not a translation. When the source is ambiguous, flag the ambiguity for clarification instead of silently choosing a stronger rule.
How machine translation handles modality
Modern machine translation can produce impressively fluent modal wording, but fluency does not guarantee calibrated force. Systems may choose a common target equivalent even when context requires a rarer one. They may also smooth away hedges, especially when the target language tends to express stance implicitly. Human review should therefore include a dedicated modality pass for any text where certainty, obligation or permission matters.
A useful workflow is to compare source and machine output sentence by sentence while marking modal triggers. Search for words and phrases associated with possibility, necessity, permission, recommendation, inference and probability. Then test whether the target reader would make the same decision or draw the same conclusion. Translation quality is ultimately behavioural: would the reader understand what is possible, required, allowed and uncertain in the same way?
How to revise modality systematically
During revision, do not rely only on rereading for general smoothness. Run a focused modal audit. First, underline every source modal verb, modal adverb and stance phrase. Second, identify its function and strength. Third, locate the corresponding target expression. Fourth, compare the likely interpretation. Fifth, check consistency across repeated instructions or claims. This targeted review is faster than hoping modal errors will stand out during ordinary proofreading.
Then read the target alone. Does it sound natural? A translation can be semantically calibrated yet stylistically awkward if it copies English modal structures too closely. The final version needs both equivalence and idiomatic target-language expression. Accuracy and naturalness are not enemies; they are two constraints on the same solution.
Build a project-specific certainty ladder
For long projects, create a certainty ladder. List recurring source expressions from weakest to strongest, such as might, may, could, is likely to, probably, should, will, almost certainly, must—but do not assume this order is universal. Record the contextual readings that actually occur in the document. Beside each, note preferred target strategies and examples.
The ladder does two things. It promotes consistency, and it exposes places where one source form has several functions. Instead of forcing one target equivalent everywhere, the translator can record conditional choices: may for permission uses one strategy; may for possibility uses another. The result is controlled flexibility rather than mechanical uniformity.
Teach modality through contrast sets
Students learn modal translation well when sentences are arranged in contrast sets. Start with one proposition: “Kai Kai submit the report today.” Then vary the stance: “Kai Kai can submit,” “may submit,” “might submit,” “should submit,” “must submit,” “will submit,” and “would submit if the data were ready.” Ask learners to describe the change before translating anything.
This makes grammar visible as meaning. It also connects translation to critical reading. Students begin noticing whether a writer is claiming, predicting, recommending, permitting or requiring. That skill transfers beyond language lessons into science, history, mathematics explanations, source evaluation and everyday decision-making.
Common failure 1: one dictionary equivalent per modal
The most basic failure is building a permanent bilingual pair such as can = X, may = Y, must = Z. This feels efficient but breaks as soon as one English form changes function. Dictionary entries are inventories of possibilities, not automatic decisions. The sentence selects the meaning.
A better vocabulary notebook records functions and examples. Under can, separate ability, general possibility and permission. Under must, separate obligation and inference. Under would, separate conditionals, politeness, reported future and repeated past behaviour. This is vocabulary depth applied directly to translation.
Common failure 2: treating politeness as weaker meaning
A polite request can still be a real request, and a respectfully phrased rule can still be mandatory. Translators sometimes confuse interpersonal softness with logical weakness. “Could you please submit the document by Friday?” may be linguistically polite but functionally firm in a workplace setting. The target language should preserve both the courtesy and the practical expectation.
Distinguish two axes: social tone and modal force. Tone asks how respectfully or directly the message is delivered. Force asks how optional, likely or necessary the content is. They interact, but they are not the same thing. Good translation keeps both axes under control.
Common failure 3: ignoring scope
Modal meaning depends on scope: which part of the sentence the modal governs. “You may not enter” can be interpreted as prohibition in many contexts, while “It may not work” usually means there is a possibility that it will fail. The same sequence may not behaves differently because the relation between modality and negation differs.
Before translating a modal plus negative, paraphrase it explicitly: “you are not permitted to enter” versus “it is possible that it will not work.” This simple step prevents one of the most damaging categories of modal error. Scope should also be checked with words such as only, always, necessarily, all and some, because their position can change what is being claimed.
Common failure 4: normalising uncertainty away
Writers sometimes sound repetitive because they repeatedly hedge. A translator may be tempted to vary or remove these markers for elegance. In research, policy analysis and risk communication, that can be dangerous. Repeated uncertainty may reflect repeated limits in the evidence. The translator should not convert epistemic caution into stylistic confidence merely to improve rhythm.
Editorial variation is possible when it preserves the same commitment level. One occurrence of “may” might become a natural target modal, another a probability adverb and another an impersonal possibility construction. The forms may vary while the degree of certainty remains aligned.
A compact decision tree
When you meet a modal expression, ask in order: What is the underlying event or proposition? Is the sentence about ability, permission, possibility, probability, obligation, advice, expectation, willingness, prediction or hypotheticality? How strong is the stance? Where does that force come from? What is the register and relationship between speakers? Does negation change permission or obligation? What target-language construction would make a native reader infer the same thing?
If the answer remains uncertain, do not guess prematurely. Read surrounding sentences, check the document type, consult comparable target-language texts, examine institutional terminology and, when possible, ask the author or subject specialist. Ambiguity is sometimes a property of the source. A translator’s responsibility is not always to eliminate it; sometimes it is to preserve it or flag it.
Quality-control checklist
Before finalising a translation, verify the following: every source modal meaning has a target counterpart; no possibility has become a fact; no recommendation has become a command without justification; no requirement has been weakened into advice; permission is not confused with ability; prohibitions are not confused with absence of obligation; hedges and confidence markers remain calibrated; polite requests still perform the same social action; and repeated requirement levels are consistent across the document.
Then perform the reverse test. Read only the target and ask what a reasonable reader would believe or do. Would the reader treat the action as optional or mandatory? Would the reader understand the result as certain or tentative? Would the reader think permission has been granted? If the behavioural interpretation differs from the source, revise the modal wording.
Frequently asked questions
Does each English modal verb have one correct translation?
No. Most modal verbs are polyfunctional. The correct target expression depends on whether the source expresses ability, permission, possibility, probability, obligation, advice, politeness, inference or another function. Context selects the equivalent.
Is might always less certain than may?
Not in a simple numerical sense. Might can sound more tentative in many contexts, but real usage varies by register, dialect and sentence type. Translate the intended degree of possibility in context rather than assigning fixed percentages.
What is the difference between must not and do not have to?
Must not normally expresses prohibition: the action is not allowed. Do not have to removes necessity: the action is optional. This distinction should receive special attention in instructions and safety texts.
Should a translator preserve hedging?
Yes when the hedge contributes meaning. In academic, scientific, legal, policy and risk texts, hedging often communicates evidence strength, uncertainty or scope. It should not be removed merely to make prose sound more decisive.
The deeper lesson: translate commitment, not just content
A sentence tells us something about the world, but it also tells us how the speaker stands behind that information. Modality is the grammar and vocabulary of that relationship. It marks what is possible, permitted, required, expected, inferred, recommended or uncertain. When a translation preserves the nouns and verbs but changes that relationship, it changes the message.
The practical rule is therefore simple: translate the proposition and the commitment separately, then reunite them naturally in the target language. Do that consistently and modal verbs stop being a list of troublesome grammar items. They become what they really are: a map of certainty, responsibility and choice.