Small clauses in translation are reduced predicative structures containing a subject-like noun phrase and a predicate without the full finite tense/agreement structure of an ordinary clause. Familiar English examples include I consider him intelligent, They found the room empty, We want the door open and They elected Lina president. The key is that him is intelligent, the room is empty, the door is open and Lina is president are genuine lower predication relations even though no finite is appears.
People searching for small clauses, secondary predication, subject predicate without finite verb, consider him intelligent, object complement, resultative predicate are solving a hidden-predication problem. Cambridge work on verbless predicative structures and secondary predication shows why these constructions cannot be reduced to ordinary noun modification: the nonverbal element predicates a property, identity, state or result of a specific argument.
This article owns the small-clause precision layer for eduKateSG. It does not replace ECM, Absolute Constructions, Resultatives/Depictives or general Complement Clauses. Its narrow task is to preserve the local subject–predicate relation, distinguish selected small-clause complements from optional secondary predicates, and stop translation systems from turning predicative adjectives/nouns into mere attributive modifiers.
Quick Read
For every suspected small clause write MATRIX VERB, SMALL-CLAUSE SUBJECT, SMALL-CLAUSE PREDICATE, PREDICATE TYPE and RELATION=JUDGMENT/OBSERVED STATE/DESIRED STATE/RESULT/OTHER. Paraphrase privately as ‘SUBJECT is/becomes PREDICATE.’
One-Sentence Answer
Translate small clauses by preserving the hidden subject–predicate proposition even when the target needs an overt copula, infinitive or finite clause to express it.
Why This Precision Layer Matters
The surface sequence NP + adjective/noun/PP can be misread as one noun phrase or two objects. That changes asserted meaning. ‘Consider him intelligent’ asserts intelligence; ‘consider the intelligent man’ presupposes it. Small-clause recognition protects that distinction.
What a Small Clause Is
The first task is to identify the local subject–predicate relation. A small clause is a reduced predicative structure containing a subject-like expression and a predicate but lacking the full finite tense/agreement structure of an ordinary clause. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is treating the two elements as simple object plus modifier without recognizing a predication relation. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to paraphrase the structure as SUBJECT is PREDICATE. For example, ‘I consider Lina intelligent’ contains the mini-predication Lina is intelligent. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Subject and Predicate
This construction should be paraphrased as a miniature copular clause before translation. The core diagnostic is a predication relation: one expression is interpreted as having the property, identity, location or state denoted by another. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is attaching the predicate to the matrix subject or nearest noun instead of the small-clause subject. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to write SC SUBJECT and SC PREDICATE separately. For example, in ‘They found the room empty,’ the room is empty. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
No Finite Tense
A precise translation separates predicative meaning from attributive modification. Small clauses lack ordinary finite tense morphology even though their interpreted time is usually related to the matrix event. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is inventing an independent tense simply because the target language needs a full clause. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to infer temporal relation from matrix tense/aspect and context. For example, found the room empty locates emptiness at the finding time. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Adjectival Predicates
The target may insert be/to be without adding new lexical meaning. Adjectives are common small-clause predicates: consider him intelligent, find the room empty, paint the wall red in related resultative structures. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is translating the adjective as an attributive modifier of the noun. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to test whether the sentence asserts the property of the NP. For example, ‘consider him intelligent’ means consider [him to be intelligent], not consider [the intelligent him]. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Nominal Predicates
Matrix verb semantics determines whether the lower predication is judged, observed, desired or caused. Noun phrases can serve as predicates in structures like elect Lina president or consider Omar a genius. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is treating both NPs as two objects rather than subject and predicate. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to paraphrase NP1 is NP2. For example, Lina is president / Omar is a genius is the lower predication. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Prepositional Predicates
The first task is to identify the local subject–predicate relation. PPs can function as small-clause predicates, especially with location/state: consider the matter under control, find Lina in the office. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is treating the PP only as an adjunct of the matrix verb. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to ask whether PP predicates a state/location of the NP. For example, Lina is in the office forms the embedded relation. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Adverbial/Particle Predicates
This construction should be paraphrased as a miniature copular clause before translation. Some constructions allow particles or adverb-like elements as predicates: want him out, find the lights off. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is translating out/off as ordinary adverbs of the matrix verb. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to paraphrase subject is out/off. For example, the target may need a finite copular clause. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Matrix Verbs
A precise translation separates predicative meaning from attributive modification. Verbs such as consider, find, believe, want, make, elect, call and deem can select small-clause-like complements, but each has distinct semantics and syntax. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is assuming all NP+predicate strings are the same construction. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to check matrix verb valency and meaning. For example, consider him intelligent differs from make him angry. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Consider-Class Predicates
The target may insert be/to be without adding new lexical meaning. Consider/deem/judge can introduce evaluative small clauses expressing a judgment about the subject. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is translating the NP as direct object only and losing the predicate. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to reconstruct the judgment proposition. For example, deem the plan unsafe = judge [the plan is unsafe]. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Find-Class Predicates
Matrix verb semantics determines whether the lower predication is judged, observed, desired or caused. Find can introduce a discovered state: find the door open, find Lina asleep. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is translating find only as locate and missing the predication. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to distinguish FIND ENTITY from FIND [ENTITY STATE]. For example, found the door open means discovered it in an open state. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Want-Class Predicates
The first task is to identify the local subject–predicate relation. Want can select reduced predicative complements such as want the door open or want him out. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is adding an infinitive actor unnecessarily. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to identify desired state. For example, want the door open = desire [door be open]. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Make-Class Predicates
This construction should be paraphrased as a miniature copular clause before translation. Make can causatively introduce a resulting state: make him angry, make the room quiet. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is treating angry/quiet as simple adverbs. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to map CAUSE + SMALL-CLAUSE RESULT. For example, someone causes [him be angry]. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Elect/Appoint/Name
A precise translation separates predicative meaning from attributive modification. Verbs of naming/appointing often select NP + nominal predicate. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is translating the predicate as a second recipient/object. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to identify role assignment. For example, elect Lina president = cause/declare [Lina be president]. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Call/Name Labels
The target may insert be/to be without adding new lexical meaning. Call him a genius / call the method dangerous contain predicate-like labels or properties. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is treating the predicate noun/adjective as quoted speech necessarily. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to distinguish naming/judging construction from reporting speech. For example, call him a genius can mean characterize him as a genius. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Small Clauses vs ECM
Matrix verb semantics determines whether the lower predication is judged, observed, desired or caused. ECM usually contains an infinitival predicate with to be or another infinitive: believe him to be honest. Small clauses can omit overt to be: consider him honest. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is treating all small clauses as identical to ECM or inserting to be where the source lacks it. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to compare lower predicate form and matrix licensing. For example, consider him honest can paraphrase with to be but has distinct surface syntax. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Small Clauses vs Object Control
The first task is to identify the local subject–predicate relation. Object control has a matrix object that semantically receives a role from the higher verb and controls an infinitive; small clauses establish a direct NP–predicate relation. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is assuming persuade him angry is parallel to consider him angry. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to test matrix semantic selection and lower predication. For example, persuade him to leave selects a persuadee; consider him intelligent judges a property. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Small Clauses vs Resultatives
This construction should be paraphrased as a miniature copular clause before translation. Resultatives describe a state achieved as the result of the matrix action: hammer the metal flat, paint the wall red. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is treating every resultative as a selected small clause or every small clause as resultative. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to ask whether the predicate describes caused endpoint versus contemporaneous state/judgment. For example, paint the wall red results in red; find the wall red merely observes. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Small Clauses vs Depictives
A precise translation separates predicative meaning from attributive modification. Depictive secondary predicates describe a participant’s state during the matrix event: eat the meat raw, arrive tired. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is treating a depictive as an argument complement selected by the matrix verb. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to identify optional secondary predication. For example, raw describes meat during eating but eat does not require raw. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Secondary Predication
The target may insert be/to be without adding new lexical meaning. Small clauses and secondary predicates overlap conceptually because both add a predication beyond the matrix verb, but structural analyses differ by language and construction. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is using one label as a universal theory. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to focus on the subject–predicate relation needed for translation. For example, the target must preserve who has which property regardless of analysis. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Small Clauses vs Absolutes
Matrix verb semantics determines whether the lower predication is judged, observed, desired or caused. Absolute constructions also contain subject–predicate mini-structures, but they function circumstantially and are more syntactically independent. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is translating an absolute as a selected object complement or vice versa. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to check whether the matrix verb selects the mini-clause. For example, with Lina absent modifies the whole clause; consider Lina absent is selected by consider. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Small Clauses vs Copular Clauses
The first task is to identify the local subject–predicate relation. A finite copular clause overtly expresses tense/agreement; a small clause lacks that full finite structure. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is assuming no copula means no predication. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to use the paraphrase subject is predicate only as an analysis tool. For example, target can require an overt copula. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Case on Small-Clause Subjects
This construction should be paraphrased as a miniature copular clause before translation. The small-clause subject can receive objective/accusative case from the matrix environment in languages such as English. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is using nominative pronouns because the NP is semantically a subject. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to apply target case rules after identifying lower subjecthood. For example, consider him intelligent, not *consider he intelligent. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Agreement on Predicates
A precise translation separates predicative meaning from attributive modification. In agreement-rich languages, adjectives/participles inside a small clause can agree with the small-clause subject. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is ignoring agreement and attaching the predicate to the wrong NP. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to match gender/number/case features. For example, predicate agreement can identify its subject. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Predicate Case
The target may insert be/to be without adding new lexical meaning. Nominal/adjectival predicates can take special case marking, depending on language and construction. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is translating case morphology as an independent semantic relation. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to recognize predicative case. For example, instrumental or nominative predicate forms may correspond simply to be/become relation. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Word Order
Matrix verb semantics determines whether the lower predication is judged, observed, desired or caused. Some languages place predicate before subject or use flexible order inside small clauses. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is assuming first NP/predicate position determines subjecthood. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to use agreement/case/semantic predication. For example, predicate-initial order can still mean NP is the bearer of the property. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Negation
The first task is to identify the local subject–predicate relation. Negation can target the small-clause predicate or the matrix verb. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is moving NOT across the boundary. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to bracket matrix vs small-clause negation. For example, I don’t consider him guilty differs from I consider him not guilty. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Modality
This construction should be paraphrased as a miniature copular clause before translation. Matrix verbs such as want, expect, consider and make contribute attitude/causation around the small-clause proposition. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is translating only the lower property and losing matrix stance. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to preserve matrix semantic frame plus predication. For example, want [door open] differs from believe [door open]. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Temporal Interpretation
A precise translation separates predicative meaning from attributive modification. Because small clauses lack independent tense, their temporal relation is often simultaneous with, resultant from or controlled by the matrix event. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is assigning a separate absolute timeline without evidence. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to classify simultaneous/result/desiderative relation. For example, found him asleep = asleep at finding time; painted wall red = red after painting. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Stative Predicates
The target may insert be/to be without adding new lexical meaning. Many small-clause predicates are states or properties, making copular paraphrases natural. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is forcing eventive verbs in target. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to use state predication where appropriate. For example, consider plan impossible = plan is impossible. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Eventive/Result Predicates
Matrix verb semantics determines whether the lower predication is judged, observed, desired or caused. Resultative constructions can introduce change-of-state meaning around an otherwise stative predicate. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is translating flat/red/open as timeless properties. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to include result relation when matrix action creates the state. For example, hammer metal flat = cause metal to become flat. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Existential/Locative Small Clauses
The first task is to identify the local subject–predicate relation. Location predicates can form reduced clauses under verbs of perception, desire or judgment. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is dropping the location relation as adjunct. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to test NP is at/in/on LOCATION. For example, found him at home means he was at home at discovery. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Subject-Oriented Secondary Predicates
This construction should be paraphrased as a miniature copular clause before translation. A secondary predicate can describe the matrix subject: Lina arrived tired. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is attaching tired to another nearby noun. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to identify controller by syntax/agreement. For example, Lina is tired during arrival. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Object-Oriented Secondary Predicates
A precise translation separates predicative meaning from attributive modification. A secondary predicate can describe the object: Lina ate the fish raw. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is attaching raw to Lina because she is matrix subject. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to map predicate to object controller. For example, fish is raw during eating. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Translation into English
The target may insert be/to be without adding new lexical meaning. English often preserves small clauses directly with verbs such as consider/find/want/make/elect, but some source structures require to be or a finite clause. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is forcing a source verbless structure into an unnatural English calque. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to choose matrix-verb-specific English complement. For example, source NP+ADJ can become ‘consider NP ADJ’ or ‘believe NP to be ADJ’. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Translation into Copula-Rich Targets
Matrix verb semantics determines whether the lower predication is judged, observed, desired or caused. A target may require an overt copula or infinitive in contexts where the source uses a small clause. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is treating inserted be as added semantic content. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to use target grammatical support. For example, NP ADJ can become NP to be ADJ. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Translation into Agreement-Rich Targets
The first task is to identify the local subject–predicate relation. Target predicate adjectives/nouns may require case/gender/number agreement. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is copying invariant English adjective behavior. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to generate agreement from the small-clause subject. For example, him intelligent can require masculine singular predicative morphology. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Legal and Administrative Language
This construction should be paraphrased as a miniature copular clause before translation. Deem, consider, declare and appoint structures are common in legal/administrative prose. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is misidentifying the predicate as a second object and changing legal status. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to reconstruct the embedded proposition. For example, deem the permit invalid means the permit has invalid status. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Clinical and Technical Language
A precise translation separates predicative meaning from attributive modification. Find X elevated/open/unstable and keep X sealed/closed are common technical patterns. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is treating property terms as loose modifiers rather than state predicates. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to map state and temporal relation. For example, find pressure elevated describes measured state. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
AI Modifier Bias
The target may insert be/to be without adding new lexical meaning. Models can attach the predicate adjectivally inside the NP: ‘consider the intelligent man’ instead of ‘consider the man intelligent.’ Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is turning asserted property into presupposed noun modification. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to paraphrase NP is ADJ before generation. For example, predicative vs attributive meaning must stay distinct. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
AI Object-Control Confusion
Matrix verb semantics determines whether the lower predication is judged, observed, desired or caused. Models can interpret NP+predicate as a control infinitive or vice versa. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is adding agency or intention not present. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to classify matrix verb and predicate category. For example, consider him intelligent is not persuade him to be intelligent. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
AI Copula Overinsertion
The first task is to identify the local subject–predicate relation. Models can insert finite is/was inside the source structure and accidentally create a full embedded clause with different tense/syntax. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is changing construction while trying to clarify. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to use a copular paraphrase privately, then generate target grammar. For example, added finite tense must not create unsupported temporal contrast. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Back-Translation Small-Clause Check
This construction should be paraphrased as a miniature copular clause before translation. A final audit should recover small-clause subject, predicate, matrix verb and relation type. Small-clause analyses vary theoretically, but translation only needs the observable predication and matrix relation to be secure.
The main translation risk is checking only that both NP and adjective/noun remain. Because no finite verb visibly separates the NP and predicate, models can group them as one constituent or assign the predicate to the wrong participant. Agreement, case and matrix valency are decisive evidence.
A practical diagnostic is to write MATRIX V + SC SUBJECT + SC PREDICATE + RELATION. For example, translation succeeds when the same entity bears the same property/state/result. Once the mini-proposition is stable, choose the target’s natural small clause, infinitive, copular complement or finite clause.
Worked Examples
Worked Example 1: Judgment Small Clause
Source problem: ‘I consider him intelligent.’
Diagnosis: Him is subject of intelligent; consider supplies judgment.
Target strategy: Keep English small clause or use ‘I consider him to be intelligent.’
Verification: Do not turn intelligent into an attributive modifier.
Worked Example 2: Observed State
Source problem: ‘They found the room empty.’
Diagnosis: Room is empty at the time of finding.
Target strategy: Translate as discovery of state.
Verification: Do not reduce to ‘they found the empty room’ unless source presupposes emptiness.
Worked Example 3: Desired State
Source problem: ‘We want the door open.’
Diagnosis: Door is small-clause subject; open is desired predicate.
Target strategy: Target may use ‘We want the door to be open.’
Verification: No door-opening actor is necessarily specified.
Worked Example 4: Role Assignment
Source problem: ‘They elected Lina president.’
Diagnosis: Lina is assigned the predicate/status president.
Target strategy: Translate status relation clearly.
Verification: President is not a second elected object.
Practical Diagnostic Workflow
Create a SMALL-CLAUSE CARD: MATRIX V; NP; NP CASE; PREDICATE CATEGORY=ADJ/NP/PP/PARTICLE; AGREEMENT; TEMPORAL RELATION; SELECTED OR OPTIONAL; RESULTATIVE/DEPICTIVE?; TARGET COPULA NEEDED? This separates complement predication from loose modification.
Run the copular paraphrase test: replace the NP+predicate with ‘NP is/was/becomes PREDICATE.’ If the matrix sentence meaning remains close, a predicative analysis is likely. Then test whether the relation is state, result, judgment or desire.
Verification Method
Back-read the target and ask exactly who/what has the property or status. Check whether the property is asserted versus presupposed, simultaneous versus resultative, and selected by the matrix verb versus optional. The target passes when the same mini-proposition survives.
Recovery Method
If the predicate became attributive, pull it back out of the noun phrase and restore predication. If it attached to the wrong participant, use agreement/case and semantics to restore the subject. If the target needs a copula, insert one without creating unsupported tense.
Quality-Assurance Checklist
- Identify small-clause subject.
- Identify predicate category.
- Paraphrase subject is predicate.
- Separate predicative from attributive meaning.
- Check matrix verb valency.
- Distinguish ECM from small clauses.
- Distinguish object control.
- Distinguish resultatives and depictives.
- Distinguish absolutes.
- Check NP case.
- Check predicate agreement/case.
- Preserve temporal/result relation.
- Insert target copula only as grammar requires.
- Audit AI modifier bias.
- Audit AI object-control and copula-overinsertion errors.
Frequently Asked Questions
What is a small clause?
A reduced clause-like predication containing a subject and a predicate without ordinary finite tense/agreement.
What is the small clause in ‘I consider him intelligent’?
The lower predication is ‘him intelligent’—roughly ‘he is intelligent.’
How is a small clause different from ECM?
ECM usually contains a nonfinite verbal predicate such as ‘him to be intelligent’; small clauses often have a nonverbal predicate with no overt infinitive.
How is it different from an absolute?
An absolute is a circumstantial, relatively independent mini-clause; a small clause is typically selected by the matrix predicate as a complement or secondary predication.
What are secondary predicates?
Additional predicates describing a clause argument, including depictive and resultative patterns; some analyses treat them as small-clause structures.
Can the predicate be a noun or PP?
Yes. ‘elect Lina president’ and ‘find Lina in the office’ involve nominal or prepositional predication.
Why does English use objective case in ‘consider him intelligent’?
The NP occurs in an object-like matrix environment even though it is semantically subject of the small-clause predicate.
Why does AI mistranslate small clauses?
It often treats the predicate as an adjective modifier inside the NP or confuses the construction with control/ECM.
Where This Fits in the eduKate Translation Architecture
Route infinitival raising/case structures to Exceptional Case Marking; circumstantial subject–predicate mini-clauses to the live Absolute Constructions owner; causative/result-state structure to Morphological Causatives; argument selection to Verb Valency and Argument Structure.
Final Principle
A small clause hides a complete predication in a compact structure. Find the local subject, find what is being predicated of it, then preserve that proposition even if the target has to spell out a copula or infinitive.