Translate Precisely | Ellipsis and Negation Scope — How to Translate Not All, None, Not Both and Not Yet

Direct answer: To translate negation with an omitted verb, recover the full action, identify what not applies to, and compare the source and target with a concrete scenario. Not all does not mean none, and not yet does not mean never.

In this guide

• The shortest answer: decide exactly what the negative denies
• Why a tiny word can change a large claim
• What scope means without formal logic
• A four-file truth test that exposes scope errors
• Why ellipsis makes the negative’s target harder to see
• Not every is not the same as not any
• Not both versus neither: one person may still act
• No longer versus not yet: opposite directions in time
• Never and not always are not equivalent

The shortest answer: decide exactly what the negative denies

When a sentence leaves its verb unspoken, do not let the negative word float free. In “Mina checked every invoice, but Leo did not,” the second clause usually means that Leo did not check every invoice, not necessarily that Leo checked none. The difference is enormous: Leo might have checked seven of eight invoices. Translating “did not” as “checked no invoices” adds a stronger claim. Begin by recovering check every invoice; then place the negative over that recovered content and test what the sentence permits.

This is the central problem in ellipsis and negation scope. Scope asks which piece of meaning is affected by not, no, never, neither, hardly, no longer or a negative auxiliary. Ellipsis means part of a verbal expression is understood rather than spoken. These combine whenever an English sentence says “she didn’t,” “they weren’t,” “none did,” “not all,” “not both” or “not yet” and relies on earlier text to supply the action.

A useful translation has three things: the correct recovered action, the correct extent of denial, and the correct time or person. You may repeat the full predicate in the target language, but you must not strengthen *not every* into *none*, weaken *must not* into *does not have to*, or turn *has not yet* into *never will*.

Why a tiny word can change a large claim

Imagine an internal audit report that says, “The first team reviewed all twelve files, but the second did not.” One interpretation is that the second team did not review all twelve. That is compatible with eleven reviews, one review or no reviews. A target sentence saying “the second team reviewed none” chooses only one possibility and may create a false accusation.

Now replace the sentence with “The second team reviewed none of the twelve files.” This explicitly states zero reviews. Replace it with “The second team has not reviewed the files yet.” The phrase not yet places the report before expected completion, rather than necessarily denying performance forever. Change it again to “The second team could not review the files.” This reports inability, which may explain nonperformance but is not identical to simply saying the team did not.

Negatives interact with auxiliaries, quantifiers and context. A grammatical word-by-word conversion is therefore insufficient. Someone translating a contract, research report or school notice needs to identify what the source commits to, not merely where the visible not sits. Two polished translations can convey different numbers of completed tasks and different judgments about responsibility.

To diagnose the problem, write one question: “What would have to be true for this negative sentence to be accurate?” Test at least two concrete scenarios. If the source is true in a scenario that the target makes impossible, the target has likely overtranslated its negative.

What scope means without formal logic

A quantifier is an expression that tells us how many or which members of a group are involved: all, every, some, most, at least one, no, neither. Negation can operate over a quantified proposition or be built into a negative quantifier. English often supports different readings depending on word order, prosody and discourse. Not all students answered ordinarily says that universal completion is false; no students answered says completion by any student is false. These are not stylistic alternatives.

A small formula is useful as a check, but it need not appear in the final translation. If the class has eight students, “not all answered” means the number who answered is less than eight. It does not tell us whether that number is zero, three or seven. “None answered” sets the number at zero. “Most did not answer” normally says more than half did not, which is different again.

Scope can also affect two actions. “They did not print and distribute the notice” may deny the combined print-and-distribute workflow, without proving that neither printing nor distribution occurred. “They neither printed nor distributed” explicitly denies both. Context may favor a stronger conversational inference, but an inference is not always an entailment.

The first principle is to record a set and an operator separately. The set might be all eight students, the tasks might be printing and distribution, and the operator might be not. Once these are mapped, natural-language translation becomes a constrained decision rather than a guess.

A four-file truth test that exposes scope errors

Use files A, B, C and D as an imaginary review set. Ask whether Jo reviewed every file, no file, or both specifically named files A and B. Different negative claims remain true in different worlds. This helps beginners and expert reviewers see why a phrase that sounds nearly identical can differ materially.

If Jo reviewed A, B and C but missed D, Jo did not review all four is true. Jo reviewed none is false. Jo did not review both A and B is also false, because Jo reviewed both of those named files. If Jo reviewed only A, “not all” and “not both A and B” are true, while “none” is still false. Count-based reasoning is simpler than arguing over translations in the abstract.

A practical quality assurance procedure generates two test cases: one where the claim is obviously satisfied, and one where it clearly fails. Translate the sentence and ask a fluent target-language speaker whether the two cases behave the same way. When they do not, inspect the negative operator and the quantifier. The goal is not to teach readers to run formal logic for every sentence, but to use a small model when the distinction changes a real conclusion.

The table below compares the three claims and makes their different truth conditions visible without requiring mathematical notation.

Jo reviewed	Not all four	None of four	Not both A and B
No files	True	True	True
A only	True	False	True
A, B and C	True	False	False
A, B, C and D	False	False	False

Why ellipsis makes the negative’s target harder to see

With a full predicate, the reader sees the negative beside its content: “Jo did not inspect every unit.” With ellipsis, a later clause may say only “Jo did not.” The reader must recover an antecedent such as “inspect every unit” from an earlier sentence. If the first clause contains more than one action or a complex quantified object, the negative may target a larger proposition than an inattentive translator expects.

For example: “The engineer calibrated every sensor and documented the deviations. The technician did not.” Does the final did not mean no calibration, no documentation, or failure to complete the combined workflow? Context and discourse emphasis can determine the intended reading. Do not automatically pick the last verb simply because documented is nearest. A source author may be contrasting what each employee completed as an entire two-stage assignment.

An editor’s working note should identify possible antecedents separately. Candidate A: calibrate sensors; candidate B: document deviations; candidate C: perform both stages. The text may eliminate two candidates through neighboring sentences. If it does not, a high-stakes translation should flag the ambiguity rather than fabricate a decisive failure.

The crucial order is antecedent first, operator second. Decide which full action the auxiliary refers to, then interpret the negative. If you place not over the wrong action, even a perfectly translated negative particle will produce an inaccurate target.

Not every is not the same as not any

“Not every candidate submitted a portfolio” says at least one candidate did not submit, under the ordinary nonempty-domain reading. It allows others to have submitted. “Not any candidate submitted a portfolio” is an awkward but possible way to express that no candidate submitted; “no candidate submitted” is much clearer. The difference is about how much of the group is excluded, not about the emotional strength of the speaker’s disappointment.

Ellipsis can hide that contrast: “The first cohort completed every section; the second did not.” An expansion “the second did not complete every section” is a reasonable starting interpretation. A translation “the second completed no sections” may turn partial completion into complete nonparticipation. This can misinform parents, course leaders or funders.

Beware of translating a negative plus an indefinite into the target language’s ordinary universal negative without checking the intended quantifier. Languages differ in their use of negative concord and polarity items. Two overt negatives can express one semantic negation in some languages, while in other constructions two negatives may cancel or contribute distinct meanings. Translate the truth conditions, not the number of negative words.

If the source is a statistical statement, keep percentages and denominators exact. “Not every respondent” does not imply the fraction who declined, and it cannot be replaced with a claim like “most respondents did not” without data. A translation should remain as informative—and as limited—as the source.

Not both versus neither: one person may still act

Consider “The printer and scanner did not both work.” That sentence can be true if the printer worked and the scanner failed, if the scanner worked and the printer failed, or if neither worked. A target that says “neither the printer nor the scanner worked” excludes the first two situations. The stronger negative may be wrong.

Now turn the idea into ellipsis: “Both branches were expected to send reports, but they did not both.” The recoverable predicate is send reports, and the negative may deny that both branches did so. It does not identify which branch failed and does not necessarily mean both failed. “Neither branch sent a report” would claim exactly that no branch sent one.

This distinction matters when a sentence refers to people instead of machines. “Both coordinators did not sign” can be ambiguous in informal English between each did not and not both did; many editors rewrite the source to avoid this uncertainty. Translation should not silently claim a unanimous refusal when the document only says both signatures were not obtained.

A source-to-target test uses two checkboxes, one per person or device. With one checked and one unchecked, “not both” is true but “neither” is false. This is a particularly fast way to catch overstrong negatives before a translated report reaches an administrator or customer.

No longer versus not yet: opposite directions in time

“The laboratory has not calibrated the instrument yet” and “The laboratory no longer calibrates the instrument” describe different timelines. Not yet looks forward toward possible future performance; it marks noncompletion up to a reference time. No longer describes cessation of a previously relevant activity or state. An ellipsis can reduce these to “it hasn’t yet” and “it no longer does,” making context especially important.

If a source says “The first team has uploaded the records, but the second hasn’t yet,” a suitable expansion is that the second team has not uploaded the records up to now. Do not write that it abandoned uploads permanently, or that it cannot upload them, unless the source explains that. Pending and impossible are different statuses.

If the source says “The group used to inspect every shipment; it no longer does,” the translation should convey that the practice ended. It does not say the group never conducted inspections. A target that says “the group has not inspected any shipments” without a time range may erase the historical practice.

Temporal negatives are common in status updates. Verify whether the source asserts previous performance, current nonperformance, expected later performance or a permanent prohibition. A one-word adverb can govern those distinctions. Preserve it even when the target language prefers a full sentence rather than an elliptical auxiliary.

Never and not always are not equivalent

“The team did not always check the temperature” means there were exceptions to a relevant practice, not necessarily that checking never occurred. “The team never checked the temperature” denies every instance in the relevant period. Translate the first as an irregular or incomplete practice, not as total neglect.

A compressed report might read: “The morning team checked the temperature at every handover, but the evening team did not always.” The understood action is temperature checking at handovers. The adverb always now sits under negation. A target that says the evening team did not check at any handover creates a more serious and unsupported failure report.

Frequency operators also interact with only, sometimes and rarely. A statement such as “the assistant rarely did” can be true if the assistant performed a task once, while “never did” cannot. The target language may put its frequency term before the verb or use an aspect marker, but the number or distribution of occasions must remain compatible with the source.

One good audit question is “Could this sentence be true if the task occurred once?” With not always and rarely, possibly yes. With never, no. Concrete counterexamples make frequency scope more legible than abstract labels, especially when translating short incident summaries or school attendance reports.

Didn’t either is not the same as didn’t both

“The first editor did not approve the changes. The second editor didn’t either.” Both editors individually did not approve. The particle either adds a parallel negative relation, not a partial-denial reading about the combined group. “Both editors did not approve” may be understood similarly in ordinary conversation, but its scope can be ambiguous and should be handled cautiously in formal records.

Compare “The first editor approved the changes, but the second didn’t.” Only the second is reported as nonapproving; it is wrong to translate this as neither approved. A third pattern, “Not both editors approved the changes,” says the pair did not both approve, leaving open whether one or neither did. These are three different decision records.

If the source uses an elliptical reply after a question, resolve the reply’s polarity. “Did the department and contractor authorize release?” “Not both.” That answer is not equivalent to “No, neither authorized it,” unless context establishes it. A translation that treats all short negative replies as simple no may eliminate important distinctions.

For bilingual minutes or approvals, repeat the participant nouns if the target language’s short form leaves readers unsure. The final text should preserve the decision state for each person where known and openly retain uncertainty where unknown. There is no gain in elegantly concise prose that misreports who consented.

Must not versus does not have to

“You must not publish the list” is a prohibition or strong negative obligation in the ordinary instruction reading. “You do not have to publish the list” removes an obligation; it may leave publishing permitted. They are almost opposites from a compliance perspective. If English first says “The report must be archived” and later says “The draft must not,” the understood act may be archiving the draft, with a prohibition against it. Do not translate this into mere lack of obligation to archive.

The difference also occurs in short answers. “Must we submit the attachments?” “No, you needn’t” means the attachments are not required. “No, you mustn’t” means they are forbidden under the relevant rules. Confusing these can cause readers to violate instructions even if both versions contain a negative and a modal.

Some target languages mark obligation and prohibition through constructions that do not look like English modal-plus-not. Use their ordinary legal or instructional wording while retaining the actual force. Do not write a literal equivalent if native readers would interpret it as advice rather than a prohibition.

Review the larger policy before translating isolated lines, particularly where exceptions and conditions govern the prohibition. The grammar can identify a negative obligation, but the document determines who is bound and what action is prohibited. Never invent a penalty, statutory source or exception solely to make the sentence sound more official.

Cannot versus did not: ability is not performance

“The researcher could not reproduce the result” can report lack of ability or failure under the conditions, while “the researcher did not reproduce the result” reports nonperformance and may leave reason unclear. The distinction matters when the result was tested, when a replication attempt failed, or when resources were unavailable. A translator should not convert did not into could not unless the source establishes impossibility or inability.

Elliptical continuations amplify this risk. “The first lab reproduced the signal; the second didn’t” says the second did not reproduce it. Whether it tried, used another method, or was unable to is left to surrounding evidence. Translating the line as “the second lab was incapable of reproducing it” adds an ability judgment and could misrepresent the science.

Conversely, “The first lab was able to complete the assay, but the second couldn’t” specifically concerns ability or feasibility. A target that says the second team chose not to perform the assay imputes a decision or refusal not in evidence.

A status ledger should distinguish not attempted, attempted but failed, unable, not authorized and successfully completed. These outcomes can have similar-looking negative English auxiliaries in different contexts but are not interchangeable. Source-grounded ambiguity should remain visible where the author has not specified the reason.

Negation inside a relative clause

“The researcher checked every file that the reviewer didn’t.” This sentence may be understood as referring to files that the reviewer did not check, with checked recovered inside the relative clause. The main researcher checked those files. That is not the same as “the researcher didn’t check every file the reviewer checked.” The negative is inside a different clause and applies to a different participant’s action.

An editor should expand both relations. First determine which files belong to the set: files not checked by the reviewer. Then determine what is asserted of them: the researcher checked every file in that set. If the target places the negative over the researcher’s action instead, the truth conditions reverse. An operational report could end up crediting or blaming the wrong person.

Relative-clause attachments can be difficult across languages because the target may place the relative clause before its head noun or use a nominalized expression. The negative must remain attached to the intended checking predicate, not to a higher verb or a noun meaning nonexistent file.

This case connects to Antecedent-Contained Deletion, which examines related relative-clause ellipsis in detail. The practical decision remains who performed the understood action and where the negative operator takes effect.

Ellipsis in comparatives: fewer than versus not as many as

“The first group interviewed twelve people, but the second didn’t interview as many.” The second group interviewed fewer than twelve, assuming the reference comparison is twelve. That is not the same as interviewing nobody. If a translation says “the second interviewed none,” it loses the comparative standard and overstates nonparticipation.

Likewise, “The second group did not interview more people than the first” does not necessarily imply that the groups interviewed equal numbers; it allows the second to have interviewed the same or fewer, depending on how comparative negation is interpreted. This is a numerical relation, not a simple event denial.

Now compress the second predicate: “The first group translated more pages than the second did.” Here did recovers translated pages, and the comparative more determines the relation. A target must not treat did as a new action or change pages into documents, because the measured unit matters. Negation can interact with such comparisons in ways that are easy to misread: “not more than” need not mean “less than,” because equality may be possible.

For formal numerical claims, test equality explicitly. Could both groups have the same count while the source statement remains true? If yes, a target that insists the second group has fewer is probably too strong. This simple equality test catches many mistakes before they appear in dashboards, reports or exams.

Questions and short negative replies

A conversation may run: “Did the managers approve every request?” “Not all.” The short answer refers to requests, not managers, if the discourse makes that grouping clear. It says some requests were not approved, without identifying which or whether a manager approved any. A target that interprets not all as not all managers approved changes the relevant quantifier domain.

Compare “Did both managers approve the request?” “Not both.” This answer concerns the managers. “Did either manager approve it?” “Neither.” This says neither approved. The questions establish what the short answers quantify over. Reading only the replies strips away the semantic information needed for accurate translation.

If subtitles or chat transcripts must stand alone, a translator may expand a reply enough for target readers to identify the correct group. That can preserve meaning when the target language lacks a compact form. However, do not add which manager or which request failed unless the conversation reveals that information.

Speaker labels are another source of error. A negative reply belongs to the person who uttered it; it does not automatically express the entire group’s institutional position. Keep attribution, question scope and reply polarity aligned. A two-word answer can be precise, but only with the right conversational antecedent.

A worked report: eight lab samples and two review teams

Suppose a laboratory has samples S1 through S8. Team A documented all eight. Team B documented S1 through S7 but had not yet documented S8. A source reports: “Team A documented every sample, but Team B did not.” That sentence is true under the ordinary negative-universal reading. A target saying “Team B documented no samples” is false against the actual data.

If the writer instead says “Team B has not documented S8 yet,” the report is more precise about the missing item and time. If the writer says “Team B must not document S8,” it is a prohibition, not a status update. Translators need to preserve which of these meanings the actual source selected, not optimize the report into whatever would help managers most.

A useful review matrix has rows for each sample and columns for documentation by Team A and Team B. The matrix allows an editor to test candidate translations with facts rather than intuition. The problem is especially serious where a summary sentence appears in an audit report without the table beside it.

This example also illustrates why a word like documented is not interchangeable with verified. A sample can be documented but not validated. Negation scope does not erase that vocabulary distinction. The translation must get the action right before it can get the number of actions right.

A worked notice: publication permission and prohibitions

An organization announces: “The approved version may be distributed. The draft must not.” The understood action after must not is be distributed, and the prohibition applies to the draft. The notice distinguishes permission for one version from prohibition for another. A target saying “the draft does not need distribution” misses the prohibition; a target saying “the draft was not distributed” reports past nonperformance instead of a rule.

Now add an exception: “The draft must not be distributed unless Legal approves it.” A translation that drops unless makes the ban unconditional. A translation that says “may be distributed without Legal approval” reverses the constraint. Quantifiers and conditions need to be attached to the right document version.

If the source uses the word approved, keep its specific workflow meaning. Approval for distribution is not proof of scientific validity or public endorsement. Official controlled labels may need to be preserved exactly. The grammar clarifies which action is hidden, but a reviewer must still verify which document status applies.

For high-stakes notices, it is often safer to repeat the action fully: “The draft must not be distributed.” Natural clarity outranks English-style ellipsis. Read the target back as an instruction to a staff member; if that person could distribute a draft believing the rule merely says it is optional, the translation has failed.

A worked education note: homework and partial completion

A teacher writes: “Most students completed the practice set; three did not.” Under the most direct reading, three students did not complete the set. The sentence does not say they attempted zero questions. One could have finished nine of ten questions and still count as not completing. A target saying the three students refused to do homework adds motive; saying they could not do it adds an ability judgment.

If the source says “Not all students completed the practice set,” the count of noncompleters remains unspecified. Do not replace it with “most failed to complete” unless the data supports that majority. Completion is also not correctness: a learner can finish every item incorrectly or accurately, and the sentence does not report performance.

Suppose there are twenty students, seventeen completed the work and three did not. The target should allow exactly that model. A clear classroom example makes it obvious why not all, most and none should never be casually substituted for one another in translated learning reports.

The final educational translation should be easy for parents to understand quickly. Explicit wording about incomplete work may be better than retaining a compressed did not, but its scope must stay the same. It should not shame students or invent explanations, just report what the teacher actually observed.

A worked service log: not resolved versus not closed

A support center note reads: “The morning shift resolved four urgent tickets; the evening shift did not.” The most likely recovery is resolve four urgent tickets, though context may mean it resolved no urgent tickets or fewer than four. The numerical scope matters. A target that says the evening shift resolved zero tickets may be too strong if it resolved two. If the source requires exactly zero, it should say so.

The distinction becomes sharper when tickets have states such as acknowledged, escalated, resolved and closed. “The evening shift did not close the ticket” does not necessarily mean it failed to resolve the underlying problem, because administrative closure may happen later. Conversely, a ticket marked closed is not proof of a satisfactory fix unless the workflow defines it that way.

A bilingual support manager should compare the target with the case log. Does the translated status preserve the source verb? Does any negative refer to all tickets or a named ticket? Is noncompletion temporary, permanent or simply unrecorded? These are operational facts, not stylistic flourishes.

If the source is too compressed to decide the number of resolved tickets, note that in editorial review. Translation is not an invitation to calculate missing statistics from intuition. A transparent query is better than publishing a confident but unsupported performance claim.

An example in a research abstract: uncertainty and nonreplication

“The original experiment detected a weak effect, but the replication did not.” The omitted predicate is detect a weak effect. This does not itself demonstrate that the effect is impossible, that the original study was fraudulent or that the replication was poorly conducted. It says the replication did not detect that effect under its conditions.

A target saying “the replication disproved the effect” is stronger than the source. Failure to detect can arise for multiple reasons, including sample variation, design differences or statistical power, none of which the short statement resolves. Translators should preserve the exact evidential verb and the negative, then leave interpretation to the researchers.

If the original adds “not consistently,” the claim may refer to variation across runs rather than total failure in all attempts. That adverb matters. “Never replicated” has a different temporal and distributional force from “not replicated in this study.” An accurate target avoids expanding a local result into a universal research verdict.

For scientific writing, a qualified domain reviewer should check methodological terminology and any statistical values. The translator can still perform the core meaning audit: action = detect; target object = weak effect; negation = no detection in the replication; wider causal conclusion = not asserted. That is enough to prevent some of the most consequential scope errors.

How to preserve genuine ambiguity

Not every source sentence allows one clear scope reading. “All the students didn’t respond” can be interpreted by some readers as none responded, and by others as not every student responded, especially in certain varieties and contexts. A translator should not claim a unique logical reading without examining discourse, stress and the intended report.

For ordinary prose, the best editorial step may be to suggest a clearer source rewrite before translating: “None responded” or “Some responded, but not all.” But when translating historical quotations, testimony or literary speech, rewriting the source is not always permitted. In those cases, preserve the ambiguity where the target language can do so, or explain the choice in an authorized note.

Do not hide uncertainty by producing a translation that sounds more certain simply because the target language lacks matching word order. Try two fully explicit readings, check which the context rules out, and record unresolved alternatives. Ask the source writer what they meant when possible. If no answer is available and the distinction matters, disclose the limitation rather than silently upgrade an inference into fact.

The goal is faithful communication. A clear sentence can be inaccurate if it decides more than the source did. A carefully framed ambiguity can be the most accurate way to represent a genuinely underdetermined source.

A practical ten-step negative-scope review

1. Identify the exact lexical predicate: reviewed, approved, filed, printed, checked or paid. 2. Recover the material omitted after an auxiliary such as did, will, has or may. 3. Identify the grammatical subject of the second clause. 4. List all relevant quantifiers, such as every, some, neither or both. 5. Identify negative operators and time adverbs.

6. Expand the clause into one or two complete source-language sentences. 7. Test the likely interpretation using a small hypothetical set of people or objects. 8. Translate into natural target-language wording without changing the truth conditions. 9. Check whether the target strengthens a partial denial into a total denial, changes inability to nonperformance or replaces prohibition with no obligation. 10. Ask an independent reviewer to reconstruct the action, affected set and timeline.

This is more reliable than a vague instruction to “watch out for negatives.” Every item has a purpose. The action check prevents predicate substitution. The quantifier check prevents count changes. The temporal check prevents not yet becoming never. The source expansion makes the reviewer see the entire claim before target-language grammar introduces another possible ambiguity.

For a routine chat message, the review may take seconds. For clinical, legal, financial or safety documentation, a qualified reviewer may be required and source clarification should be logged. One shared method can be scaled according to the consequences of a mistranslation.

Practice questions with model answers

1. “The first team completed all six tasks; the second did not.” Does it imply zero tasks completed by Team B? No. 2. “The first team completed none, and the second didn’t either.” Does this report completion by Team B? No; the intended reading is parallel noncompletion, subject to context. 3. “Both reports weren’t submitted.” Is the sentence maximally clear? No; distinguish not both from neither.

4. “The draft must not be released.” Is release optional? No; it is prohibited. 5. “The draft need not be released.” Is release prohibited? Not by that sentence; the duty is absent. 6. “The sample has not been tested yet.” Has testing become impossible? No, not according to the source.

7. “Not every respondent agreed.” Did none agree? Not necessarily. 8. “No respondent agreed.” Did anyone agree? No within the relevant set. 9. “The device could not record the event.” Does this prove it deliberately refused? No; devices do not have such intent in ordinary technical reporting.

10. “She didn’t always verify the readings.” Does it say she never verified them? No. 11. “The editor reviewed the files but did not approve them.” Can the files still have been reviewed? Yes, explicitly. 12. “The assistant did not send more letters than the manager.” Does it guarantee fewer? No; equality can satisfy the negative comparison.

Write each answer in a complete sentence rather than only yes or no. Then translate the full answer and compare it with a second target phrasing. The exercise is complete when both versions allow the same possible worlds and do not attribute an action to the wrong participant.

Frequently asked questions

Does “not all” mean “some”? It usually means at least one member of the relevant group does not satisfy the predicate; it does not specify exactly how many do. In an ordinary nonempty context there may be zero or some positive number satisfying it. Is “not both” equivalent to “neither”? No. One participant may satisfy the predicate when not both is true.

What does “didn’t either” refer to? It ordinarily repeats a negative predicate from preceding discourse and indicates parallel nonperformance for another subject. Why is “not yet” important? It reports noncompletion up to a reference time, often with a possibility of later action; it should not become never.

Can negation move in translation? The target word order can change, but the negative must retain its intended scope. Does “must not” mean “doesn’t have to”? No. One prohibits; the other removes obligation. Does “couldn’t” always mean “didn’t”? No. One can concern ability or feasibility; the other reports nonperformance.

How do I choose the antecedent for an elliptical negative? Read the full discourse and recover the source-supported action before applying the negative. What if more than one reading is available? Preserve the ambiguity or seek clarification where the consequences are material.

Is it safe to let AI rewrite an ambiguous negative? AI can suggest interpretations, but an elegant paraphrase can change a quantifier or scope. Compare concrete truth conditions and source evidence rather than trusting fluency.

Related learning routes and source checks

For the broader method of preserving meaning, return to Master Art of Translation. Antecedent-Contained Deletion explains a relative-clause configuration in which a hidden predicate interacts with a quantified phrase. Strict and Sloppy VP Ellipsis covers who a recovered possessive refers to. Sluicing and Wh-Ellipsis handles missing questions.

For background on scope and inference, see the research discussion of quantifier scope and the Cambridge Grammar overview of modal meanings. These are reference routes, not substitutes for the source sentence, the relevant subject-matter vocabulary or an authorized institutional review.

The final rule is concise: recover the action first, locate the negative second, and test the size and timing of the claim third. If the source says a group failed to complete every task, do not declare that it completed none. If it says an act was not yet done, do not make its impossibility permanent. Fidelity is the accurate boundary between what a sentence denies and what it leaves open.