Agreement resolution in translation determines which person, number, gender or noun-class features an agreeing verb, adjective, participle or pronoun should take when its controller contains more than one conjunct. Coordination creates the classic problem: Lina and Omar are two singular nouns but one plural referent; I and you can require first-person inclusive plural; mixed-gender nouns can trigger default, common or hierarchy-based agreement depending on the language.
People searching for agreement resolution, coordination agreement, closest conjunct agreement, semantic agreement, notional agreement, mixed gender agreement are solving a feature-resolution problem. Current comparative syntax research explicitly treats coordination resolution rules and asymmetry among conjuncts as a major cross-linguistic domain, while work on closest-conjunct agreement shows that some languages license agreement with a structurally or linearly local conjunct rather than with a fully resolved coordination phrase.
This article owns the agreement-resolution precision layer for eduKateSG. It does not replace Grammatical Gender and Agreement, Grammatical Number, Coordination and Parallel Structure or Clusivity. Its narrow task is to preserve how coordinated and conceptually complex controllers determine agreement: resolved plural/person/gender, closest-conjunct agreement, semantic/notional agreement and default forms.
Quick Read
List every conjunct’s PERSON, NUMBER, GENDER/CLASS and ANIMACY. Then classify the coordination: AND, OR, NOR, CORRELATIVE, ADDITIVE or APPARENT COORDINATION. Finally choose the language’s strategy: RESOLUTION, CLOSEST CONJUNCT, FIRST/HIGHEST CONJUNCT, SEMANTIC/NOTIONAL or DEFAULT.
One-Sentence Answer
Translate agreement resolution by computing the feature set of the entire coordinated referent—or the language’s licensed local controller—before generating target agreement.
Why This Precision Layer Matters
Agreement can silently exclude people, switch grammatical person, change gender/class reference or alter whether a group is construed as one unit or many members. English hides many of these choices because adjectives do not agree in gender and noun class, but a rich-agreement target cannot avoid them.
What Agreement Resolution Is
The first task is to compute the features of the coordination. Agreement resolution is the set of rules a language uses when one agreement controller contains more than one conjunct with potentially conflicting person, number, gender or noun-class features. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is choosing agreement from one noun by intuition when the language computes features for the coordination as a whole. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to write the feature bundle for every conjunct before choosing the controller features. For example, a singular masculine noun plus a singular feminine noun can trigger plural agreement with a resolved gender value. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Coordination as an Agreement Controller
This problem becomes visible when conjuncts carry conflicting number, person or gender. A coordinated phrase can function as one syntactic controller even though it contains several nouns. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is agreeing with each conjunct separately and producing duplicated or inconsistent target morphology. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to treat the coordination as one phrase before applying target agreement. For example, ‘Lina and Omar are…’ uses plural agreement in English. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Number Resolution
A precise translation should distinguish grammatical resolution from processing attraction. Two or more singular conjuncts often trigger plural agreement, but language-specific exceptions, semantic factors and closest-conjunct patterns exist. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is assuming singular morphology from the nearest noun or plural in every language. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to calculate conjunction number according to the source system. For example, two singular NPs normally create plural reference in English. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Plural Plus Singular
The target language may resolve a coordination differently from the source. A plural conjunct combined with a singular conjunct can still trigger plural agreement, but some systems show proximity or feature-resolution effects. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is copying the number of the first or last conjunct mechanically. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to compare whole-coordination agreement with closest-conjunct options. For example, ‘the students and the teacher are…’ is plural in English. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Person Resolution
Agreement features belong to a controller relation, not simply the nearest noun. Coordinations with different grammatical persons often use a hierarchy such as first over second over third, though the exact system is language-specific. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is translating all mixed-person subjects as third-person plural because several people are involved. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to identify the language’s person-resolution hierarchy. For example, I + you may trigger first-person plural agreement in many systems. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
First + Third Person
The first task is to compute the features of the coordination. A first-person conjunct combined with a third-person conjunct often yields first-person plural agreement in languages with person resolution. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is keeping third-person agreement because the larger lexical NP is third person. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to prioritise the person hierarchy rather than noun length. For example, I and Lina → we-type agreement. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Second + Third Person
This problem becomes visible when conjuncts carry conflicting number, person or gender. A second-person conjunct plus third-person conjunct can resolve to second-person plural in some systems. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is defaulting to third-person plural because third person seems neutral. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to check the source/target person hierarchy. For example, you and Lina can trigger second-person plural agreement in languages that resolve person hierarchically. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
First + Second Person
A precise translation should distinguish grammatical resolution from processing attraction. When first and second person are coordinated, some systems resolve to first-person plural, often inclusive in interpretation; others have special forms. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is assuming the result is exclusive we or ordinary second plural. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to identify clusivity and person resolution together. For example, I + you naturally includes both speaker and addressee. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Gender Resolution
The target language may resolve a coordination differently from the source. Mixed-gender coordinations can trigger masculine/default, feminine, common, neuter or another resolved form depending on language. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is importing the target language’s gender hierarchy into the source or vice versa. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to list grammatical gender of each conjunct and the language’s resolution rule. For example, masculine + feminine may resolve to masculine plural in one language and common plural in another. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Noun-Class Resolution
Agreement features belong to a controller relation, not simply the nearest noun. Languages with noun classes can have complex rules for resolving class agreement over coordinated nouns. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is treating noun-class values as natural gender or choosing one conjunct’s class without evidence. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to consult the target/source class-resolution paradigm. For example, class A + class B can trigger a default/resolved plural class. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Animacy in Resolution
The first task is to compute the features of the coordination. Animacy can override or influence grammatical gender/class resolution in some systems. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is using biological animacy as a universal resolution rule. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to identify whether human/animate coordinations have special agreement. For example, mixed human conjuncts may trigger human plural agreement despite different noun classes. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Semantic Agreement
This problem becomes visible when conjuncts carry conflicting number, person or gender. Agreement can follow meaning rather than strict morphological form, especially with collective nouns, measure phrases, titles and conceptually plural expressions. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is treating formal singular morphology as the only possible controller. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to compare grammatical number with referential number. For example, ‘the committee are…’ is possible in some English varieties when members are foregrounded. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Notional Agreement
A precise translation should distinguish grammatical resolution from processing attraction. Notional agreement is a familiar term for agreement determined by conceptual number or meaning rather than surface morphology. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is overgeneralising semantic agreement and ignoring formal grammar. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to identify whether the language/variety licenses notional agreement in that construction. For example, British English often permits plural agreement with collective nouns more readily than American English. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Collective Nouns
The target language may resolve a coordination differently from the source. Group nouns can take singular or plural agreement depending on whether the group is construed as a unit or as members, especially in some English varieties. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is treating singular/plural agreement as a factual number difference rather than construal. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to ask whether the source foregrounds unit or individuals. For example, the team is winning vs the team are changing shirts. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Measure Phrases
Agreement features belong to a controller relation, not simply the nearest noun. Amounts of money, time, distance or weight can be morphologically plural but take singular agreement when construed as one quantity. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is forcing plural agreement from the noun ending. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to identify whether the phrase denotes one measure. For example, ten kilometres is a long way. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Titles and Quoted Strings
The first task is to compute the features of the coordination. Book titles, names of organisations or quoted phrases can look plural but behave as singular entities. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is agreeing with internal plurality rather than the titled entity. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to treat the title/name as one referent. For example, ‘Great Expectations’ is a novel. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Closest-Conjunct Agreement
This problem becomes visible when conjuncts carry conflicting number, person or gender. Some languages allow agreement with the conjunct linearly or structurally closest to the agreeing element instead of resolving features for the whole coordination. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is calling closest-conjunct patterns errors or treating them as random attraction. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to identify the language’s licensed CCA environments. For example, an adjective or verb can agree with the nearest singular/feminine conjunct even though the coordination is semantically plural. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
First-Conjunct Agreement
A precise translation should distinguish grammatical resolution from processing attraction. Some systems show agreement with the first/highest conjunct in certain word orders or constructions. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is assuming proximity always means the last conjunct. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to distinguish linear closest from structural highest conjunct effects. For example, preverbal versus postverbal agreement can change which conjunct controls features. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Last-Conjunct Agreement
The target language may resolve a coordination differently from the source. Postposed modifiers or predicates can show agreement with the final/closest conjunct in some languages. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is forcing resolved plural/default agreement because coordination is plural semantically. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to check modifier position and local agreement option. For example, a postnominal adjective can agree with the final conjunct under CCA. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Resolved Agreement vs CCA
Agreement features belong to a controller relation, not simply the nearest noun. A language can permit both resolved agreement and closest-conjunct agreement with different preferences or discourse/register effects. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is choosing one strategy as universally correct. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to identify whether both are grammatical and what conditions the preference. For example, French coordination can show varying preferences in adjective agreement. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Structural vs Linear Proximity
The first task is to compute the features of the coordination. Closest agreement can be sensitive to syntactic structure rather than merely number of words between controller and target. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is using physical distance as the only explanation. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to compare prenominal/postnominal positions and constituent structure. For example, the structurally highest conjunct may control a prenominal target. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Default Agreement
This problem becomes visible when conjuncts carry conflicting number, person or gender. When features cannot be resolved naturally, a language can use a default person/gender/number form. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is interpreting default morphology as a real semantic property of the conjuncts. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to label DEFAULT rather than assigning it to one referent. For example, default neuter or masculine can appear without implying all conjuncts have that gender. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Agreement Attraction
A precise translation should distinguish grammatical resolution from processing attraction. Agreement attraction is an error or processing effect where a nearby noun incorrectly influences agreement, unlike grammatical CCA licensed by a language. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is calling every nearest-noun agreement a grammatical resolution strategy. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to distinguish normative/structural CCA from performance attraction. For example, ‘the key to the cabinets are…’ is nonstandard attraction in English. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Semantic vs Syntactic Coordination
The target language may resolve a coordination differently from the source. And-coordination normally combines referents, but phrases that look coordinated can sometimes express one conceptual unit. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is forcing plural agreement whenever and appears. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to test whether the expression denotes one conventional unit. For example, ‘fish and chips is…’ can be singular when treated as one dish. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Additive Phrases
Agreement features belong to a controller relation, not simply the nearest noun. Along with, together with, as well as and similar phrases can resemble coordination without creating the same agreement controller in English. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is using plural agreement as though every additive phrase were and-coordination. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to identify true coordination versus supplementation. For example, ‘Lina, along with Omar, is…’ keeps Lina as singular controller. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Either…Or
The first task is to compute the features of the coordination. Disjunctive coordination can use different agreement strategies from and-coordination, often influenced by proximity and semantics. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is using plural resolution automatically. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to check target-language rules for disjunction. For example, ‘either the teachers or the student is…’ often follows closest noun in formal English. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Neither…Nor
This problem becomes visible when conjuncts carry conflicting number, person or gender. Negative coordination likewise interacts with proximity, number and style. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is copying and-coordination agreement. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to identify the language’s nor-construction rules. For example, neither the manager nor the assistants are… vs reordered variant. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Correlative Coordination
A precise translation should distinguish grammatical resolution from processing attraction. Both…and, either…or, neither…nor and not only…but also can influence agreement and information structure. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is treating the correlative markers as semantically irrelevant once nouns are identified. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to preserve both coordination type and agreement strategy. For example, both Lina and Omar are… differs from either Lina or Omar is/are…. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Predicative Adjective Agreement
The target language may resolve a coordination differently from the source. Coordinated subjects can control number/gender agreement on predicate adjectives in languages where English adjectives are invariant. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is dropping resolved gender/number distinctions that disambiguate the source. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to compute controller features before translating adjective agreement. For example, mixed-gender plural adjective can show a default resolution form. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Participles
Agreement features belong to a controller relation, not simply the nearest noun. Past/passive participles can agree with coordinated arguments under language-specific resolution or closest-conjunct patterns. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is treating participial morphology as tense only. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to identify agreement features on the participle. For example, a participle can expose gender/number resolution that the finite verb does not. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Determiners and Attributive Adjectives
The first task is to compute the features of the coordination. Coordination inside noun phrases can force agreement choices on determiners/adjectives. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is assuming only verbs participate in resolution. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to map every agreement target in the phrase. For example, a determiner can agree with the whole coordination or closest conjunct depending language. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Pronouns Referring to Coordinations
This problem becomes visible when conjuncts carry conflicting number, person or gender. A later pronoun referring back to coordinated nouns must resolve number, person, gender and sometimes animacy. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is choosing a pronoun from the nearest conjunct. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to build the referent set before pronoun selection. For example, Lina and Omar → they; I and Omar → we. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Clusivity after Coordination
A precise translation should distinguish grammatical resolution from processing attraction. If first and second person are included, a target language with inclusive/exclusive we may require inclusive forms. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is using an exclusive first-person plural and excluding the addressee. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to derive the participant set explicitly. For example, I + you necessarily includes the addressee. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Honorific and Social Features
The target language may resolve a coordination differently from the source. Coordinated referents can differ in honorific status, and target systems may have rules for resolving social agreement or respectful reference. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is flattening all social distinctions to one conjunct. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to use target honorific conventions for mixed groups. For example, a respectful plural may be appropriate for a mixed-status group. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Translation into English
Agreement features belong to a controller relation, not simply the nearest noun. English has relatively limited gender/class agreement but still requires number/person resolution and variety-sensitive notional agreement. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is ignoring source gender/class resolution even when it carries referential information. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to preserve semantic referent set and any discourse effects elsewhere. For example, a source masculine-default plural may simply become they in English. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Translation into Rich Agreement Languages
The first task is to compute the features of the coordination. A target with gender/noun-class/person agreement may require choices not overt in English source coordination. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is guessing target resolution from English semantics alone. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to construct the participant feature set and apply target rules. For example, Lina and Omar can require a resolved plural gender/class form. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Legal and Administrative Coordination
This problem becomes visible when conjuncts carry conflicting number, person or gender. Lists of parties, offices and entities require reliable agreement and pronoun reference. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is letting closest noun change which parties a duty applies to. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to treat the coordination as a legal referent set before grammar. For example, the company and its directors shall… refers to both. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
AI Nearest-Noun Bias
A precise translation should distinguish grammatical resolution from processing attraction. Models can choose agreement from the nearest conjunct even in languages/contexts requiring resolved agreement. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is mistaking processing proximity for grammar. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to apply the target language’s resolution rules explicitly. For example, a plural coordination should not become singular merely because the nearest noun is singular. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
AI Default-Gender Bias
The target language may resolve a coordination differently from the source. Models can overuse masculine/default agreement without checking target resolution, animacy or inclusivity. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is erasing feminine/common/human agreement possibilities. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to compute the target’s actual resolution features. For example, mixed groups can resolve differently across languages. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Back-Translation Resolution Check
Agreement features belong to a controller relation, not simply the nearest noun. A final audit should recover the full coordination referent and the rule that generated agreement. Agreement is cross-linguistically diverse: some languages resolve features for the whole coordination, others allow closest-conjunct or position-sensitive agreement, and some alternate between strategies.
The main translation risk is checking only whether the verb sounds grammatical. Because a target form can be grammatical under a different controller, an error may look like a harmless agreement choice while actually changing which participant set is represented.
A practical diagnostic is to write CONJUNCT FEATURES + RESOLUTION STRATEGY + TARGET FEATURES. For example, the target succeeds when no conjunct is accidentally excluded or privileged. Once the strategy is identified, generate all agreement targets—verb, adjective, determiner, participle and pronoun—from the same referent set unless the language licenses asymmetry.
Worked Examples
Worked Example 1: Mixed Person Coordination
Source problem: Source/target referent is I + you.
Diagnosis: The participant set contains first and second person; many systems resolve to first-person inclusive plural.
Target strategy: Use the target’s appropriate inclusive first-person plural if available.
Verification: Do not use exclusive we or third-person plural.
Worked Example 2: Closest Conjunct Agreement
Source problem: A language permits a postnominal adjective to agree with the closest conjunct rather than resolved plural/gender.
Diagnosis: This is a grammatical CCA strategy, not an error.
Target strategy: Preserve the licensed target/source agreement while keeping the whole coordination semantically plural.
Verification: Do not back-translate the adjective features as excluding other conjuncts.
Worked Example 3: Collective Noun
Source problem: Source uses plural agreement with ‘team’ to foreground individual members.
Diagnosis: This is semantic/notional agreement rather than morphological plural noun form.
Target strategy: Choose target agreement according to its collective-noun conventions and the intended construal.
Verification: Unit vs members reading should remain recoverable.
Worked Example 4: Along With
Source problem: Source: ‘Lina, along with Omar, is responsible.’
Diagnosis: The additive phrase is not full and-coordination for English agreement.
Target strategy: Keep Lina as singular controller while preserving Omar’s additional involvement.
Verification: Do not switch to plural solely because two people are mentioned.
Practical Diagnostic Workflow
Create a COORDINATION AGREEMENT GRID: CONJUNCT 1 FEATURES; CONJUNCT 2 FEATURES; ADDITIONAL CONJUNCTS; COORDINATOR TYPE; WORD ORDER; AGREEMENT TARGET POSITION; LANGUAGE STRATEGY. Then compute RESOLVED FEATURES and, separately, CLOSEST/HIGHEST CONJUNCT FEATURES.
Run the reorder test. Swap conjunct order without changing meaning. If agreement changes, the system probably has proximity/structural-conjunct effects. If agreement stays the same, resolution or default agreement is more likely.
Verification Method
Back-read every agreeing element and ask which referent set it encodes. Check that no conjunct has been accidentally excluded, that person hierarchy includes the right speaker/addressee set, and that gender/class resolution follows the target language rather than source habits.
Recovery Method
If agreement was copied from the nearest noun incorrectly, rebuild the coordination feature set and reapply the target rule. If resolved/default agreement erased a meaningful local-conjunct effect, restore the licensed CCA pattern. If notional agreement drifted, decide whether the source views the group as unit or members.
Quality-Assurance Checklist
- List features for every conjunct.
- Classify coordination type.
- Resolve number.
- Resolve person.
- Resolve gender/noun class.
- Check animacy/humanness rules.
- Check clusivity for 1+2 combinations.
- Distinguish CCA from agreement attraction.
- Check first/highest vs last/closest conjunct effects.
- Check semantic/notional agreement.
- Check collective nouns and measure phrases.
- Distinguish and from along with/as well as.
- Check either/or and neither/nor separately.
- Audit AI nearest-noun bias.
- Audit AI default-gender bias.
Frequently Asked Questions
What is agreement resolution?
The grammatical process that determines agreement features when a controller contains multiple conjuncts with potentially different features.
Why does Lina and Omar take plural agreement?
The coordination denotes more than one participant even though each conjunct is singular.
What happens with I and you?
Many languages resolve to a first-person plural form that includes the addressee; a language with clusivity may require inclusive we.
What is closest-conjunct agreement?
A grammatical strategy where an agreement target takes features from the nearest or structurally closest conjunct rather than the resolved coordination.
Is closest-conjunct agreement an error?
Not necessarily. It is grammatical in many languages and constructions; English-style attraction errors are a different phenomenon.
What is notional agreement?
Agreement based on conceptual number or meaning rather than surface morphology, as with some collective nouns.
Do all mixed-gender groups take masculine agreement?
No. Resolution rules vary widely: masculine, common, human, neuter or other defaults are possible.
Why does AI get coordination agreement wrong?
Models often rely on proximity or a default gender/person strategy instead of computing the full coordinated referent.
Where This Fits in the eduKate Translation Architecture
Route broad agreement morphology to the existing Grammatical Gender and Agreement owner; number systems to the Grammatical Number owner; coordination logic to Lists, Coordination and Parallel Structure; inclusive/exclusive participant sets to the live Inclusive and Exclusive We owner.
Final Principle
Agreement with coordination is not a nearest-noun guessing game. Build the participant set, identify the language’s resolution strategy, then let every target agreement marker refer to the correct whole—or to the correct licensed conjunct.