Split Antecedents in VP Ellipsis | Translation Examples

How do you translate split antecedents in verb-phrase ellipsis? Begin with two people who intend different things: “Ava hopes to hike the coastal trail, and Ben hopes to climb the mountain. If the weather improves, they will.” In an appropriate context, the short phrase “they will” can mean Ava will hike the trail and Ben will climb the mountain. It does not necessarily mean both will hike and both will climb. The missing verbal content is recovered from more than one earlier expression, and the people are paired with the actions previously associated with them.

This is the core of split-antecedent VP ellipsis, a linguistics topic that matters to editors, translators and readers of compact prose. People searching for split antecedents examples, ellipsis with two antecedents, how to translate “they will”, VP ellipsis across clauses, or how to preserve respectively in translation need a reliable rule: list the participants, list the earlier actions, and verify the intended pairing before filling the gap. A single English auxiliary can point to a combined interpretation that no one earlier phrase alone expresses.

This guide explains when that reading is available, why it depends on discourse coherence, and how to distinguish it from ordinary VP ellipsis, strict/sloppy pronouns and shared-event anaphora. It works through travel plans, research tasks, publishing, education, service requests and organizational workflows. It also shows when a target language requires repeating both actions and how to preserve uncertainty where more than one recovery is plausible. The goal is accurate information about who is supposed to do what, not the most elegant repetition-free sentence.

A one-minute example with two people and two goals

Consider: “Mara intends to paint the mural; Leo intends to design the exhibition cards. They should if time permits.” In a supportive context, the final clause may refer distributively to each person carrying out the already assigned task: Mara should paint, and Leo should design. The verb should is a modal that marks advice or an expected course of action, not completed performance.

An inaccurate target could say “They should paint the mural and design the cards together.” This adds collaboration and assigns both tasks to both people. Another might say “Mara should design the cards and Leo should paint the mural,” reversing the pairing. The opposite error would be to choose only the nearest antecedent and translate the ending as “They should design the cards,” dropping Mara’s activity entirely.

Make a two-row pairing grid: Mara → paint mural; Leo → design cards. Then interpret the final subject they as the group Mara plus Leo. The ellipsis can distribute the relevant task to each person, rather than copying the same verb phrase identically for both. The word distributively is helpful: it describes a pairwise reading where each member is connected to the action appropriate to that member.

The target may state both pairs explicitly. A translation that says “Mara should paint the mural, and Leo should design the cards, if time permits” can preserve the intended meaning in a language that lacks the source’s neat “they should.” This is more faithful than a shorter phrase that obscures the pairings. Explicitness is acceptable when it comes from source-grounded recovery rather than invented tasks.

What researchers mean by split antecedents

In ordinary VP ellipsis, one salient earlier verb phrase may supply the missing predicate: “Ava will translate the notice, and Ben will too.” The understood predicate for Ben is translate the notice. In split antecedent cases, the intended recovery can depend on multiple different earlier predicates. One person plans one event, another person plans a different event, and a subsequent auxiliary or short clause is interpreted in relation to both.

Adam C. Baker’s paper Discourse Coherence and VP Ellipsis with Split Antecedents discusses how readers recover such cases through discourse relations. It builds on classic examples in which two people have different ambitions and a later plural subject refers to their respective plans. The analysis emphasizes the role of coherence rather than a simplistic rule that the most recent verb automatically supplies every missing word.

Later work by Lyn Frazier and John Duff, Repair or Accommodation? Split Antecedent Ellipsis and the Limits of Repair, examines processing and proposes that readers may accommodate an antecedent appropriate to the discourse rather than merely copy and edit one syntax tree. The detailed theoretical accounts differ, and experimental participants do not necessarily supply identical paraphrases for every case.

For translation, this scholarship supports a cautious practical conclusion: when one short continuation refers to two previously described plans, the mapping of participants to plans is part of the meaning. That map may be inferable from source discourse rather than from a verb phrase adjacent to the ellipsis site. Do not translate the shortest grammatical fragment until you have reconstructed the contextual event assignments.

Why it is not enough to copy the nearest verb

“Rina wants to catalogue the photographs, while Paolo wants to restore the damaged frames. If funding arrives, they can.” A nearest-verb strategy might make both people able to restore frames, because restore is closer to can. That erases Rina’s cataloguing plan. A generic strategy might make both people undertake both activities. That creates a project structure absent from the source.

A distributive interpretation instead pairs Rina with cataloguing and Paolo with restoration. Such a reading is supported by the two separate wants clauses and the plural they in the later sentence. However, the phrasing alone does not tell us whether funding would pay for both activities equally or whether the same organization would oversee them. Those are unrelated questions.

This is the heart of split antecedence: the antecedent for the ellipsis can be a structured combination, not one continuous, fully pronounced verb phrase. An editor can represent it as an ordered set of tasks attached to people. “What can they do?” becomes “What can each of them do, according to the earlier clauses?”

The risk is visible in business and education records, where the final word they may gather people into one grammatical subject without giving them one identical role. A translator who prefers neat group verbs may write “they will jointly complete the project,” making a false claim of shared execution. The better translation may be two clauses or a parallel list. Preserve task ownership before optimizing rhythm.

The role of discourse coherence

Readers make sense of short continuations because the text has an organizing relationship. Two earlier clauses may each introduce a plan, wish, commitment or possibility. A later sentence can summarize the status of those distinct plans as a coordinated question under discussion: will each plan go ahead, can the participants afford it, or should they proceed? The coherence of that larger discourse licenses the distributive reading.

Imagine a paragraph about a festival with a costume maker and a stage designer. Each is given one job; the final sentence states that both will if the budget clears. Readers naturally connect each worker with the job already introduced because the paragraph is structured as two parallel tasks. The interpretation is harder when the prior clauses discuss unrelated incidental actions or when the plural subject refers to a different group.

That context dependence is why split-antecedent analysis cannot be reduced to a punctuation rule. The mere presence of two verbs followed by they will does not guarantee the respective-pairing reading. A final sentence may instead refer to only one event that is relevant to both people, or may be ambiguous. The source’s discourse structure and the author’s communicative goal must support the reconstruction.

For translation, write down the question that the continuation answers. In the festival example, it may be “will the two assigned jobs be undertaken?” That makes a two-task paraphrase natural. If the relevant question is “will both workers attend the meeting?”, the earlier jobs may not be the right antecedents. Textual proximity helps, but coherent event structure is the stronger test.

A pairing table: subjects, actions and conditions

Use a small matrix with four headings: person, intended action, condition and event status. In “Noor wants to index the files; Malik wants to digitize the audio; they will once authorization comes through,” the rows are Noor / index files / after authorization / future; Malik / digitize audio / after authorization / future. The plural final subject refers to the pair of people; the implied predicate has two components.

Do not add the idea that Noor will digitize audio or Malik will index files. That may be a sensible cross-training plan, but it is not supplied by the source. Do not change authorization comes through into authorization has already been granted. The condition remains unsatisfied or at least is not asserted to be satisfied in the clause as written.

A similar example with could reports ability or possibility, not future certainty: “they could if the studio opened.” A version with should recommends the activities; a version with did after a clear past narrative may report actual performance. The same event pairing can appear under different modal and temporal operators.

Translators often pay attention to nouns and verbs while treating modals as stylistic auxiliaries. That is dangerous here because the whole visible content of the final clause may be one modal. If you get will, could, should, must or cannot wrong, the translation can change two people’s obligations or commitments at once. A good pairing grid therefore includes the modal’s force as a separate field.

Split antecedents versus ordinary plural action

“Mina and Kaito will sign the form” contains one action associated with a plural subject. It may involve each signing the same form or a collective action depending on context. By contrast, “Mina will sign the form and Kaito will translate the guide. They will by Friday” invites a reading in which they undertake distinct activities. Translators should not flatten this into “Mina and Kaito will sign and translate the form” without evidence that both activities apply to both people.

The source’s earlier clauses provide the task division. It is not enough to notice that they has two human antecedents; you must also notice that the missing predicate can have two event antecedents. The plural subject and composite verbal content work together.

A different sentence creates a true collective task: “Mina designed the cover, Kaito formatted the interior, and together they will present the book.” Here the last present is overt and describes a joint event. The earlier distinct tasks do not license translating present as two individual presentations. The writer has explicitly marked together and introduced a new action.

This contrast makes a useful editing rule: distinguish joint action, parallel individual actions and an unspecified plural activity. They are three possibilities. Split antecedent VPE can encode parallel individual actions in a compact ending, but other discourse signals may call for joint action or another interpretation. Always check which one the source supports.

Split antecedents versus strict and sloppy readings

Strict and sloppy VP ellipsis typically asks whose person or object is involved when a repeated predicate includes a pronoun. “Lea called her adviser and May did too” can refer to Lea’s adviser or May’s own adviser, depending on reading and context. Split antecedents instead concern how a missing predicate can draw on more than one earlier action.

A sentence may combine both problems. “Lea plans to revise her chapter and May plans to update her slides; they will.” Under a natural split-antecedent interpretation, Lea will revise the chapter and May will update the slides. But ownership of chapter and slides may itself be ambiguous in unusual contexts. A translation that resolves the tasks correctly could still misidentify which person owns a document.

Keep the layers separate. First establish whether the final clause combines two events. Then establish who performs each one. Then resolve pronoun and possessive reference within each event. Finally check time, modality and conditions. Starting with a preferred target pronoun can cause the translator to overlook the wider two-event structure.

The practical rule is not to memorize more names for ellipsis. It is to ask more precise questions. How many distinct actions are intended? How many agents perform them? Does a pronoun refer to the first person’s object or each agent’s own object? What is asserted about actual performance? Each question targets a different potential meaning change. The linked strict and sloppy reading guide covers the possessive dimension in detail.

A worked example: two travellers with different plans

Source: “Lena wants to visit the glacier, and Omar wants to see the lighthouse. They will if the ferry service runs.” The obvious split reading is that Lena will visit the glacier and Omar will see the lighthouse, subject to ferry service. It is not a promise that both people visit both attractions or travel together on one ferry.

In travel writing, the distinction may affect itinerary planning. If a translation says “they will visit the glacier and lighthouse,” a reader may infer one joint route with both stops. The source introduced separate goals. A target language may need an explicit “respectively” construction, though a pair of full clauses is often more natural.

The condition also requires care. The ferry service runs could mean it operates at the relevant time, not that the travellers have booked seats or obtained tickets. The target should not add a departure schedule or transfer route. The two activities remain conditional; nothing in the source says they were completed.

For a review test, imagine a world where the ferry service operates and Lena visits the glacier while Omar visits the lighthouse. The source can be satisfied under the split reading. Now imagine both travellers visit only the lighthouse. A mistranslation that assigns the nearest action to the group may be true while the original reading is false. Concrete scenarios reveal what a pretty but overcompressed translation has changed.

A worked example: two teachers and different lesson plans

“The English teacher hopes to run a debate, and the science teacher hopes to conduct a practical. They can once the timetable is finalized.” In an appropriate context, they can means the English teacher can run the debate and the science teacher can conduct the practical. The timetable is a common condition; the lesson formats and subjects remain different.

A target that says both teachers can conduct practical activities might make the English debate vanish. Another that says they will both teach the debate turns a pair of proposals into one team-teaching plan. A third that says they already held the sessions confuses can with completed action. None is supported.

An educational editor can expand the source into two simple planning statements before translating. Then check whether the target language permits a concise plural modal without making readers associate both educators with the same activity. If not, repeat the two actions. There is nothing stylistically inferior about a longer sentence when the source’s compressed form would mislead.

This also shows why split antecedents often matter in reports. Timetable constraints, funding and facilities may apply to several separate tasks. A compact English continuation can summarize shared feasibility while relying on previous role assignments. A translation should retain that pattern of one common condition and several distinct intended events.

A worked example: research collaborators and separate methods

Source: “Dr Rao intends to sequence the samples, while Dr Lim intends to model the resulting data. If funding is approved, both will.” The intended event pairing is Rao → sequencing; Lim → modeling. The wording does not say Rao will model the data or Lim will perform sequencing. Those contributions are scientifically distinct and may require different expertise.

The temporal relationship between the two tasks is contextually important. Modeling resulting data suggests dependency on some data, but the sentence does not supply the detailed pipeline, instrument, sample count or timeline. A translation must not invent the number of runs or guarantee that sequencing will produce analyzable results. Scientific plausibility is not textual evidence.

The condition “if funding is approved” applies to the future activities as expressed. A target that replaces will with did states the work occurred. A target that changes approval into submission of a funding application weakens or alters the condition. Grant approval and application submission are different states.

In a contribution statement, subject alignment is part of research integrity. A bilingual reviewer should be able to identify who planned which procedure and under what condition. If the target can be read as joint responsibility for both methods, the journal might misattribute work or skills. Splitting the final sentence into two explicit clauses is preferable to an ambiguous translated both will.

A worked example: publishing tasks and production handoffs

An editorial update says, “Anika wants to revise the captions, and Tomas wants to rebuild the index. They will after the author approves the changes.” The compressed ending refers to two separate production tasks, with author approval functioning as a condition. The translator should retain the distinction between caption revision and index rebuilding.

A target that writes “they will revise and rebuild the manuscript” could imply both people will rewrite the entire manuscript. That broadens the objects and hides the division of labor. A target that says “they have revised the captions and index” changes the event status from future conditional to completed. Even when the source sounds routine, those differences can affect production tracking.

The author’s approval is not necessarily publication approval, and it is not proof that either contributor has finished. Some workflows approve revisions in stages. Unless the source defines which changes, the translator should avoid adding edition numbers, dates or a quality-assurance result. Focus on the two known tasks.

A useful comparison table separates person, task, object, dependency and deliverable state. It may contain Anika / revise / captions / author approval / intended; Tomas / rebuild / index / author approval / intended. Compare the translation with those rows. No target style advantage compensates for swapping people or attaching both tasks to the same person.

A worked example: a museum exhibition and two service providers

Source: “The lighting team intends to adjust the gallery illumination, and the conservator intends to stabilize the paintings. Both can once the rooms are closed to visitors.” The plural both refers to two parties with separate plans. The lighting team adjusts illumination; the conservator stabilizes paintings. The room closure is the shared condition enabling those different activities.

Do not translate illumination adjustment into painting restoration. They may occur in the same gallery, but one targets a lighting environment and the other targets physical artworks. Do not say the conservator will operate lighting equipment unless the source establishes that work assignment. The point of this example is not complicated syntax but the ease with which a compressed ending can conceal a division of responsibility.

“Rooms are closed to visitors” could be an operational condition rather than evidence of a permanent closure. Translating it as “the museum is shut down” would exaggerate. Likewise, can may express feasibility or permission; if the site’s safety policy makes the distinction important, consult the context rather than assuming the same modal meaning everywhere.

A natural translation may say: “When the galleries close to visitors, the lighting team can adjust the lighting, and the conservator can stabilize the paintings.” This expansion repeats can and names each task. It is longer but often more accessible to staff and visitors. It makes implicit assignment explicit without inventing joint performance.

A worked example: a service department with two queues

A helpdesk note says, “The first team wants to clear overdue billing tickets; the second wants to resolve access requests. They will when staffing improves.” The final they will can naturally refer to the first team clearing its overdue billing tickets and the second resolving access requests. It does not state that both teams can handle both queues.

A careless translation might say “they will clear all overdue tickets.” That merges billing and access categories into one set and implies a universal completion claim. Another could write “the second team will clear billing tickets,” which reverses a task assignment. Either could mislead a manager deciding where more support is needed.

The staffing condition is important. Improved staffing is not a calendar date or a confirmed hire. A target that says next Monday or after recruitment is complete would invent specifics. If the source simply uses staffing improves, keep that open condition or request clarification if a service-level commitment requires a precise threshold.

The safest expansion retains the pairings and the condition: “When staffing improves, the first team will work on the overdue billing tickets and the second team will resolve access requests,” subject to whether the source’s verbs clear and resolve are being represented exactly. Do not weaken clear into merely work on if completion is genuinely intended. Correct translation tracks the strength of each predicate.

A worked example: applications and independent requirements

An administrative notice states, “New applicants intend to complete the profile, while existing applicants intend to update their documents. They may after the portal reopens.” The continuation refers to two different user groups and actions: new applicants completing profiles; existing applicants updating documents. The portal condition applies to both under the reading encouraged by the parallel source clauses.

A target that instructs existing applicants to complete new profiles could create unnecessary duplicate records. A target that tells new applicants merely to update documents could omit a required starting step. The grammar issue therefore affects user behavior and service delivery, even though the sentence contains ordinary words.

May might mean permission or future possibility. If the portal is temporarily closed, possibility is natural; if the notice communicates authorization, permission may be intended. Do not assume that the English may gives the same force as must or will. The source may describe permitted actions, not guaranteed completions.

For user-facing text, use separate subheadings or sentences if they improve accessibility. State the distinct process for each applicant group. Preserve official labels and do not invent buttons or document lists absent from the portal instructions. A split-antecedent reading can be made explicit without requiring readers to know a linguistic term.

A worked example: audit tasks and separate evidence trails

A project record reads, “The internal auditor plans to review purchase orders, and the external auditor plans to inspect supporting receipts. Both should before the report closes.” The elliptical both should refers to the separately planned review and inspection activities. It does not mean both auditors should review purchase orders and inspect all receipts unless the surrounding instructions establish joint duties.

A target that collapses the tasks into “both should audit the finances” is too broad and obscures which evidence trail each person handles. A target saying “both did” would also falsely mark the tasks as complete. Should is not a completion report; before the report closes describes an intended timing boundary.

The noun report may denote an audit report, closing report or another document, depending on the workflow. Do not invent an official reporting deadline from the mere verb closes. If the organization uses defined terms for review and inspection, keep them distinct from approval, certification and reconciliation.

A strong translation can say, “Before the report is closed, the internal auditor should review purchase orders and the external auditor should inspect supporting receipts.” The distinct obligations are visible. This expanded version does not assign identical tasks or introduce a named deadline. It illustrates how recovering two antecedents improves compliance clarity even when the target is not obliged to preserve the English ellipsis.

A worked example: parents and learners with different responsibilities

“The parent hopes to set a study timetable, and the student hopes to follow it. They can after the examination schedule is known.” The tasks are asymmetrical. A parent sets a timetable; the student follows it. The later they can may refer to the two roles working as planned under a shared condition. It should not imply that both parent and child must sit for the examinations or that the parent will complete the student’s work.

The pronoun it in follow it refers to the timetable, a separate anaphoric link. The final plural they refers to the participants, while the missing predicate combines their respective actions. The translator must track both reference relations at once. An otherwise readable target could make the student set the timetable or the parent follow a schedule intended for a different person.

A useful teaching exercise asks learners to draw three arrows: parent → creates → timetable; student → follows → timetable; schedule known → condition for implementation. The arrows expose the logic without technical labels. Then translate the source into two clear clauses and verify that the child’s action has not become the parent’s.

This scenario also demonstrates that split antecedents do not license claims about results. The student’s following a timetable does not ensure higher marks or completed mastery. Translate what is planned and who will do it; leave academic outcomes to later evidence.

A worked example: conservation and document digitization

A preservation programme says, “The conservator hopes to stabilize the fragile papers; the digital archivist hopes to scan them. They will if the storage space is prepared.” The plural future statement can recover both tasks in their respective roles. Stabilizing physical papers and scanning them are related but not interchangeable. The object papers is shared, while the activity differs by participant.

A target that says “both will scan and stabilize the papers” invents cross-functional work. A target that treats scanning as conservation success upgrades one digital operation into a broader preservation result. Such errors matter in archives because physical stabilization, digitization, descriptive metadata and access management are separate parts of the process.

The storage-space condition should remain source-grounded. Prepared could mean cleared, made safe, equipped or otherwise suitable; the original may not say which. Translate the general preparatory condition unless an attached protocol defines its requirements. Do not insert humidity settings or scanning technology that the source does not supply.

A transparent rendering is often worth its length: “If the storage space is prepared, the conservator will stabilize the papers and the digital archivist will scan them.” The person–task pairs remain intact. An editor can verify each verb and compare the shared conditional dependency without needing to model complex ellipsis in the final target text.

The challenge of and, or, neither and distributive scope

Split-antecedent interpretations become more delicate when the source changes the connective or polarity. “A wants to translate the notice, and B wants to check the figures. They will” invites a positive pairwise reading. “Neither of them can” denies each person’s ability to perform their respective plan under one common interpretation. That does not automatically mean each person is incapable of performing both plans.

Compare “they cannot both” with “neither can.” The first may deny simultaneous or joint feasibility while leaving room for one person or one plan to proceed; the second can deny each relevant ability. Their exact scope depends on how the predicate is recovered. A translator who replaces one with the other can change a project’s resource limitation.

Disjunction introduces another complication. If one person wants either of two choices, a later can may recover alternatives rather than a conjunction of tasks. “Either A or B” and “both A and B” have different truth conditions. Avoid mechanically collecting every earlier verb into a target sentence with and.

A simple check uses four imagined outcomes: both plans succeed, only the first succeeds, only the second succeeds, neither succeeds. Ask which worlds make the source sentence true. Then test the target. If a world ruled out by the source remains possible in the translation, the logical connection may have changed. The exercise is especially useful for bids, budgets and policies, where may, must and cannot have practical consequences.

Why the split reading is not always the best reading

Not every plural auxiliary after two different predicates has a clean distributive interpretation. Suppose the source says, “Jules wants to build a website, and Dana wants to write a handbook. They will next month.” The split reading is available in an appropriate context, but the statement may be more natural to some readers if the previous discourse explicitly frames both as parallel planned deliverables. Without that framing, a reader might wonder whether they will do both tasks jointly.

Prosody, punctuation and the broader question under discussion can affect how readily the construction is accepted. A translator who treats split antecedence as a mandatory decoding rule may overinterpret a source whose writer intended a shared project rather than two separate actions. Such cases need context or author clarification.

The goal is to preserve warranted meanings, not maximize exotic grammar analysis. If the source explicitly says they will do both projects together, follow that. If it explicitly assigns the work separately and asks whether each plan is feasible, use the split mapping. If it is truly ambiguous, retain a neutral target formulation if possible or query the author.

A concise editorial question can resolve the matter: “Does the final ‘they will’ mean each person will do the task assigned earlier, or both will undertake both tasks?” Asking this question is more useful than asking the writer to identify an ellipsis type. Linguistic theory provides the diagnostic; the source writer provides the communicative choice.

What does “respectively” add, and when to avoid it?

Respectively can make a pairing explicit: “Mara and Leo will paint the mural and design the cards, respectively.” This usually associates Mara with painting and Leo with designing. If the source’s split reading is certain, a target-language equivalent of respectively can preserve the mapping efficiently.

But respectively has limits. It can sound technical or stiff in public-facing instructions, and some languages handle pairwise lists more naturally through parallel clauses. Worse, a mistranslator may assign the pairs in the wrong order. If the source lists Leo before Mara in one part and the tasks in another order, blindly appending respectively could reverse the assignments.

For documents with three or four participants, pairwise order becomes even more fragile. A careful editor should construct the explicit role table before employing a compressed list. A sentence that is shorter but forces readers to count list positions may be less accessible than a few clear sentences.

Do not insert respectively when the source intends both people to perform both tasks. That would split a genuinely collective obligation. Nor should the word be used to disguise uncertainty. It signals a strong one-to-one pairing and therefore requires evidence. Use it when the source supports separate mappings; otherwise preserve the more open structure or request clarification.

Tense, aspect, certainty and status in the final clause

In a split-antecedent sentence, the final auxiliary may contain nearly all the explicit time or modality information about future events. “They wanted to, but neither could” differs from “they wanted to, but neither did.” Could indicates ability or feasibility; did concerns actual performance. A translation that substitutes failure to act for inability can misreport why plans were abandoned.

Similarly, “they will” projects future activity; “they have” often reports completed activities with present relevance when followed by an appropriate ellipsis; “they might” expresses uncertainty. If there are two previously introduced tasks, the modality may distribute across the corresponding two events. You must check each one.

The antecedent clauses can carry their own temporal information as well. “A plans to visit tomorrow and B plans to report today; they will” is potentially awkward because a single unqualified future continuation may not preserve the distinct dates clearly. A translator should not force an artificial single timestamp. If the source requires different times, the target may need explicit clauses.

A safe event ledger lists for each participant: task, intended time, certainty level and known completion status. If the source reports plans, do not mark them completed. If it reports completed work, do not reduce it to mere desire. The elegance of “they did” or “they will” is irrelevant if the final target loses the timeline or inflates certainty.

How to translate the construction naturally across languages

Some languages allow a short plural proform that readers can interpret through context; others prefer verb repetition or explicit topic–comment structures. The choice depends on the target grammar and the reading burden. It is not necessary to reproduce one English auxiliary when the intended content is a pair of different predicates.

Start by writing the full expansion: “Rina will catalogue the files and Paolo will restore the frames if funding arrives.” Check the names, objects and condition. Then choose whether to keep one sentence, use two clauses, add an equivalent of respectively, or structure the statement as a conditional paragraph. The target should not make both people agents of both events unless the source says so.

Avoid a literal translation in which a plural verb form requires one shared lexical predicate and thereby selects only the nearest antecedent. If that happens, repeat both verbs. In institutional writing, the result can be clearer and safer. In literary dialogue, a compressed target may preserve pace if readers can follow the assignments.

A bilingual reviewer should read only the target and identify the person-to-action map. If they produce a different map or assume joint performance, the translation needs revision. A semantic checklist matters more than symmetrical word counts. The correct target may be twice as long as the source because it makes the source’s elliptical structure explicit without adding meaning.

A practical AI-assisted QA workflow

A language model can expand split antecedents but may choose the wrong arrangement. It might read “the editor wants to revise the text and the designer wants to adjust the figures; they will” as “they will both revise the text and adjust the figures.” That is a plausible collaborative workflow, but the source may intend a division of labor. The model’s fluency can conceal that difference.

Ask an AI or human checker to make a two-column mapping before generating target prose. Require each participant to be connected to the task explicitly assigned in the antecedent. Then ask whether the final modal distributes over those separate tasks or refers to a new shared action. Review the response against source context rather than accepting a confident explanation.

A second pass should test modality and conditions. Did will become would? Did can become did? Did the target add funding approval, a deadline or a completed outcome? A third pass should examine pronouns and definite nouns. Does each person still own or handle the correct document or object?

For consequential communications, keep the source sentence, proposed expansion and reviewer decision in the project record. If the source genuinely underdetermines the pairing, note the alternatives and seek clarification. Models and translators should not invent a collective obligation just to make a short auxiliary easy to translate. Explicit uncertainty is preferable to polished but false assignment.

Ten practice examples and their intended interpretations

1. “Ella plans to edit the essay and Mark plans to illustrate the cover. They will after exams.” Interpret Ella → edit essay; Mark → illustrate cover. 2. “The first lab intends to test salinity and the second intends to measure acidity. They can when samples arrive.” Interpret two separate measurements; arrival is a condition.

3. “Jia wants to photograph the exhibition and Noel wants to write the catalogue. They will once access is approved.” Keep separate roles. 4. “The accountant plans to reconcile the ledger and the clerk plans to file the receipts. Neither can until the system is restored.” Keep two blocked tasks, not a claim that both people cannot perform any accounting.

5. “One parent wants to organize transport and another wants to arrange meals. They may after permission is granted.” Modal may is not proof of completion. 6. “The first translator wants to prepare a glossary and the second wants to review the captions. Both should before release.” Recover distinct recommended tasks under the release timeline.

7. “The curator plans to label the artifacts and the conservator plans to stabilize them. They will after the inspection.” The same objects can receive different actions by different people. 8. “The researcher hopes to collect interviews and the analyst hopes to code transcripts. Both might if the ethics approval arrives.” Keep tentative modality and domain-specific roles.

9. “The programmer intends to repair the interface and the tester intends to check accessibility. They will next week.” Keep repair distinct from testing; do not invent successful compliance. 10. “The coach wants to review game footage and the learner wants to practise passing drills. Neither can today.” Keep two unavailable activities without saying neither person can perform the other’s activity.

For each example, write two full independent clauses with the right subjects and predicates. Then translate them naturally. Ask whether the finished target still has exactly the expected tasks, conditions and degrees of certainty. If a plural phrase makes the two people appear jointly responsible for both actions, it has lost the intended split reading.

Common errors and how to catch them

Nearest-verb substitution copies only the last antecedent and drops earlier tasks. All-to-all assignment connects every person with every activity. Role reversal assigns the first task to the second person. Object drift replaces the files in one task with the frames or another nearby noun. Collective-action inflation inserts jointly, together or as a team without evidence.

Modal strengthening turns might or could into will or must. Completion inflation turns planned actions into completed work. Negative-scope drift changes neither can into not both can or the reverse. Condition loss drops if funding arrives, after approval or before opening. Overconfident respectively chooses a particular pairing without source support.

There is also genre drift. A staff memo may use an ellipsis as shorthand because everyone understands the assignments. A public notice may require explicit tasks for accessibility. Translators who preserve the abbreviated shape at all costs can reduce usability; translators who overexplain by adding invented deadlines or implementation details can alter the source.

An efficient QA method checks the highest-risk dimensions first: the participants, task distribution, final modal and common conditions. Only after those pass should an editor polish style and shorten sentences. Accuracy is not a result of repeating every source word; it is a result of preserving the source’s relational structure.

Frequently asked questions

What are split antecedents in VP ellipsis? Cases where a later silent predicate is interpreted using information from more than one earlier predicate, often with a plural subject mapping people to their distinct tasks. Does the final “they will” mean both people do both actions? Not automatically. In a split reading, each person may perform the task assigned in the earlier clause.

Is split-antecedent ellipsis grammatical? It is a recognized research topic with attested judgments and processing studies, but acceptability and the preferred interpretation depend on discourse context and particular examples. Is this ordinary VP ellipsis? It is part of the wider study of VP ellipsis but challenges analyses requiring one identical antecedent phrase.

Is this the same as strict and sloppy reference? No. Strict/sloppy interpretations mainly concern how pronoun reference behaves within reconstructed predicates. Split antecedence concerns multiple source events. Should I always use respectively in translation? No. Use it only when a specific pairwise mapping is supported and natural in the target language.

Does a common condition apply to both tasks? Often yes under a relevant parallel reading, but verify scope and avoid adding conditions absent from the source. What about neither can? It can distribute inability across the respective intended actions, not necessarily assert inability to do all possible tasks.

Can AI translate this reliably? It can produce a fluent expansion but may choose a wrong mapping. Verify explicitly. What if the source is ambiguous? Preserve ambiguity when possible or query the writer when the distinction matters.

Translate Precisely | Split Antecedents in VP Ellipsis — Recover Two Different Actions Without Assigning Both to Everyone