Object + Infinitive in English: Why “I Believe Him to Be Honest” Gives “Him” Two Different Grammatical Relationships

Consider: > I believe him to be honest. Why: > him and not: > he? The sentence contains an infinitive: > to be honest. Semantically, who is honest? > him. If we turn the embedded idea into a full clause, we get: > he is honest. So the pronoun behaves like the **subject of the infinitive’s meaning**. Yet English marks it with the objective form: > him. This construction sits at an interesting boundary between clauses. In different linguistic traditions, related patterns are discussed as: – **raising-to-object**; – **Exceptional Case Marking (ECM)**; – **accusative with infinitive**. The exact theoretical analysis is debated. The reader-facing fact is stable: > A noun phrase can appear after the higher verb in object form while receiving its central semantic role from the lower infinitive. That is the intellectual job of this article. ## Quick answer: what is happening in “I believe him to be honest”? Compare: > I believe that he is honest. with: > I believe him to be honest. The proposition is similar: > he is honest. In the second form, **him** appears between the matrix verb **believe** and the infinitive **to be honest**. The pronoun has objective case: > him. But the property **honest** is predicated of that same person. So one noun phrase participates in two grammatical relationships: > objective form in the larger construction > + > semantic subject of the infinitive. ## Pronouns make the pattern visible With a noun: > I believe Amir to be honest. Case is invisible because **Amir** has the same form in subject and object positions. Now use a pronoun: > I believe him to be honest. Not: > *I believe he to be honest. The contrast reveals the objective-case requirement. Other examples: > We expect her to arrive early. > They believe us to be ready. > The committee judged them to be suitable. The pronoun form tells the reader that this is not an ordinary finite clause. ## Compare with a that-clause Finite complement: > I believe that he is honest. Infinitival construction: > I believe him to be honest. The first contains **he** as the overt subject of a finite clause. The second contains **him** before a non-finite **to be** clause. The two can often be close in meaning. Their grammatical packaging differs. ## Why this is called “Exceptional Case Marking” In many traditional generative analyses, a noun phrase normally receives case within a local grammatical relationship. Here, **him** belongs semantically with **to be honest** but appears with objective case associated with the higher environment. That cross-boundary relationship motivated the term **Exceptional Case Marking**. Students do not need the historical theory to use the construction. But the name explains why the pronoun looks “unexpected”. ## Why “raising-to-object” is another label Another tradition describes **him** as originating semantically as the subject of the lower clause and appearing in the higher object position. Hence **raising-to-object**. Modern syntactic frameworks disagree about whether literal movement is the best analysis. So a careful public article should not say the noun definitely moves in one universally accepted way. The safe description is: > English has a construction in which the lower predicate’s subject is realised in objective form in relation to the higher clause. The labels describe analytical traditions. The observable pattern is what students need. ## The expletive test is powerful Consider: > I believe there to be a problem. What does **there** refer to? Nothing. It is the existential/dummy **there** found in: > There is a problem. Does **believe** assign a normal semantic role to **there**? No. This is important. A true object-control verb usually needs a meaningful object participant. But a raising/ECM-type verb can appear with expletive **there** because the noun phrase belongs semantically to the lower predication: > there to be a problem. This is one of the classic diagnostics. ## Compare with “persuade” > I persuaded him to leave. Who was persuaded? > him. Who left? > him. The noun phrase receives a semantic role from both relationships: > target of persuasion > + > understood subject of leave. Now try: > *I persuaded there to be a problem. Impossible. **Persuade** requires a genuine participant who can be persuaded. This is **object control**, not raising-to-object. ## Raising versus control eduKateSG already has: [Raising vs Control in English](https://edukatesg.com/2026/08/29/raising-vs-control-seems-tries-english/) Compare: ### Raising/ECM-like > I believe him to be honest. The higher verb does not describe an action performed directly on **him** in the ordinary sense. The belief concerns the proposition: > he is honest. ### Object control > I persuaded him to be honest. Now **him** is directly involved in the persuasion event. The higher verb selects him as a participant. Same surface sequence: > verb + him + to-infinitive. Different underlying relationship. ## “Expect” shows why classroom labels need care Cambridge grammar gives patterns such as: > The company expects her to be early. The noun phrase **her** is interpreted as the subject of: > to be early. But verbs such as **expect**, **want** and related predicates are analysed somewhat differently across syntactic frameworks. For school readers, the useful level is: > object + to-infinitive construction with an embedded subject relationship. Reserve strong theoretical claims for verbs and diagnostics where the distinction is clear. ## Passive voice exposes the structure Active: > People believe him to be honest. Passive: > He is believed to be honest. Notice the case change: > him → he. The noun phrase becomes the subject of the passive higher clause. The infinitive remains: > to be honest. This is common in formal writing: > The species is believed to be extinct. > The policy is expected to reduce delays. > The artefact is thought to date from the twelfth century. These constructions are highly useful beyond theoretical grammar. ## Formal English uses passive raising-like structures frequently Compare: > Researchers believe that the material is stable. with: > The material is believed to be stable. The second places **the material** in subject position. It also removes or backgrounds **researchers**. This can help a paragraph remain focused on the object of study. But the passive can also hide the evidence source. A careful writer may need to ask: > According to which researchers? Grammar and source accountability interact. ## “He is believed to be…” does not mean proven > The structure is believed to be safe. This means: > relevant people hold that belief. It does not automatically entail: > the structure is safe. The higher verb **believe** controls epistemic commitment. A student should not paraphrase: > is believed to be as: > is definitely. This connects naturally to evidence-sensitive reading. ## “Expected to” is similarly careful > The train is expected to arrive at 9 p.m. This describes an expectation. It does not guarantee: > arrival at exactly 9 p.m. Now: > The reaction is expected to increase with temperature. Again: > prediction/expectation not established actual result. Object/raising structures interact with modality and evidential strength. ## Pronoun case is not arbitrary Compare: > I expect him to arrive. Not: > *I expect he to arrive. But: > I expect that he will arrive. The form of the pronoun reveals the clause type. This is useful in editing because learners may blend the two structures: > *I expect that him will arrive. or: > *I expect he to arrive. The repair begins by choosing one construction. ## Full clause or infinitive? ### Finite > We believe that she is capable. ### Infinitival > We believe her to be capable. ### Passive infinitival > She is believed to be capable. All are possible in appropriate registers. The infinitival versions are often more formal, compressed or institutional in tone. The finite version is often more transparent in everyday prose. Good writing chooses clarity before prestige. ## Small clauses are related but different eduKateSG already has: [Small Clauses in English](https://edukatesg.com/2026/08/29/small-clauses-english-consider-her-capable/) Compare: > We consider her capable. No **to be** appears. Now: > We consider her to be capable. The meanings are close. Different syntactic analyses are possible, but the surface contrast is useful: > object + predicate > versus > object + to-infinitive predicate. The pages can therefore cross-link without owning the same search intent. ## Object complements are another nearby structure eduKateSG’s: [Subject and Object Complements in English](https://edukatesg.com/2026/08/29/subject-object-complements-english-grammar/) covers patterns such as: > They elected her president. Here **president** predicates a role of **her**. Now: > They believed her to be the president. The infinitive introduces a different clause-like relationship. The difference becomes clearer when students identify: – higher verb; – lower predicate; – participant role. ## Expletive “it” can also be diagnostic in related structures Raising constructions can tolerate subjects that carry no ordinary semantic role. For example: > It is believed to be raining. The dummy **it** belongs to the weather expression: > it is raining. The higher predicate **is believed** does not assign a person-like role to **it**. This supports the idea that the semantic content belongs lower: > raining. Again, the reader need not choose one formal movement theory to see the relationship. ## A Singapore classroom example Student sentence: > I believe he to be the best candidate. Problem: > **he** is the wrong case for this construction. Option 1: > I believe him to be the best candidate. Option 2: > I believe that he is the best candidate. Both can be correct. The teacher should diagnose: > blended clause pattern rather than merely writing: > pronoun error. ## Situational writing Formal: > The organisers expect all participants to arrive by 8.30 a.m. The participants are: > expected. They are also the understood subjects of: > arrive. This is a productive formal pattern. Now compare: > The organisers instructed all participants to arrive by 8.30 a.m. Here **participants** are genuine recipients of instruction. That is an object-control relationship. Surface similarity hides semantic difference. ## Science writing > Researchers expect the reaction to slow as the solution cools. The noun phrase **the reaction** is the subject of: > slow. Now passive: > The reaction is expected to slow as the solution cools. The construction lets the writer keep **the reaction** as the paragraph topic. But the claim remains predictive. Do not turn **expected to** into **will definitely**. ## Humanities and institutional writing > The reform is believed to have changed the balance of power. This structure can report an interpretation without naming every scholar in the sentence. But rigorous writing should still identify evidence or source where relevant. The grammar is a packaging device. It is not evidence by itself. ## “To have” changes relative time Compare: > He is believed to be in Singapore. Simultaneous/current-state reading. Now: > He is believed to have left Singapore. The perfect infinitive **to have left** places the leaving earlier than the belief/reporting reference point. This shows that non-finite clauses can still encode temporal relationships. Advanced readers should not treat all infinitives as timeless. ## Negation can sit in different places > I believe him not to be responsible. Formal, somewhat marked. Compare: > I do not believe him to be responsible. These are not necessarily identical in pragmatic emphasis. The first places negation inside the lower proposition. The second negates the higher belief claim. In everyday writing, the finite alternative may be clearer: > I do not believe that he is responsible. Scope matters. ## Why the construction can sound formal > I believe him to be honest. is more formal or compressed than: > I believe that he is honest. Likewise: > The artefact is believed to be authentic. is common in institutional, academic and documentary prose. Students should recognise it even if they do not need to use it frequently in personal writing. Reading competence comes before stylistic imitation. ## Diagnosis before prescription ### Gap 1: pronoun-case blending Learner: > *I believe he to be honest. **Repair:** choose either: > I believe him to be honest. or: > I believe that he is honest. ### Gap 2: every object + infinitive called raising Learner treats: > persuade him to leave like: > believe him to be honest. **Repair:** ask whether the higher verb gives the noun phrase its own semantic role. ### Gap 3: expletive blindness Learner cannot see why: > believe there to be a problem is evidence against ordinary object selection. **Repair:** ask what **there** refers to. ### Gap 4: passive loses epistemic strength Learner changes: > is believed to be into: > is. **Repair:** preserve the reporting/knowledge-status verb. ### Gap 5: formality overuse Learner writes repeated: > I believe X to be Y to sound advanced. **Repair:** compare with the simpler **that**-clause and choose the clearer register. ### Gap 6: lower subject not identified Learner thinks **him** is only the object of **believe**. **Repair:** ask: > Who is honest? The answer reveals the lower semantic relation. ## A practical two-question test Sentence: > We believe her to be capable. ### Question 1 Who is capable? > her. So **her** is semantically tied to the lower predicate. ### Question 2 Does **believe** describe an action done directly to her? Not in the same way that: > persuade her or: > instruct her does. This points toward a raising/ECM-like analysis. ## Expletive test Compare: > We believe there to be a mistake. Natural in formal English. Now: > *We persuaded there to be a mistake. Impossible. The contrast is powerful because dummy **there** cannot be persuaded. It can, however, occupy the lower existential structure: > there is a mistake. ## Passive test Active: > People believe him to be honest. Passive: > He is believed to be honest. Now compare object control: > People persuaded him to leave. Passive: > He was persuaded to leave. Both passivise. So passive alone does not distinguish them. The expletive and semantic-role tests must work together. Grammar diagnosis usually needs more than one clue. ## Internal-link opportunities This article can connect to: – [Raising vs Control in English](https://edukatesg.com/2026/08/29/raising-vs-control-seems-tries-english/) – [Small Clauses in English](https://edukatesg.com/2026/08/29/small-clauses-english-consider-her-capable/) – [Subject and Object Complements in English](https://edukatesg.com/2026/08/29/subject-object-complements-english-grammar/) – [Dummy Subjects “It” and “There” in English](https://edukatesg.com/2026/08/29/dummy-subjects-it-there-information-structure/) – [Selectional Restrictions in English](https://edukatesg.com/2026/08/29/selectional-restrictions-semantic-compatibility-vocabulary/) – [Lexical Entailment in English Vocabulary](https://edukatesg.com/2026/08/29/lexical-entailment-snore-sleep-vocabulary/) – [How Language Works](https://edukatesg.com/2026/08/26/how-language-works/) The distinct reader intent is: > objective-case noun + to-infinitive → noun is semantic subject of lower predicate → distinguish raising/ECM-like structure from object control ## Connections eduKateAI can learn **Pronoun case ↔ clause structure:** **him to be** reveals a different grammatical configuration from **that he is**. **Object form ↔ lower subject meaning:** a noun phrase can appear in objective form while receiving its main semantic role from an embedded infinitive. **Raising ↔ semantic role:** raising-like higher verbs do not assign the same kind of participant role that control verbs such as **persuade** do. **Expletives ↔ diagnosis:** dummy **there** can appear in raising/ECM-like constructions because it belongs to the lower existential clause rather than to a meaningful higher-object role. **Passive ↔ information structure:** **He is believed to be…** lets formal prose foreground the embedded subject while backgrounding the believer. **Reading ↔ epistemic status:** **is believed to be** and **is expected to be** report belief or expectation; they do not make the lower proposition automatically true. **Writing ↔ register:** object + infinitive and passive raising forms are compact and formal, but finite **that**-clauses may be clearer in ordinary prose. **Subjects ↔ evidence:** Science and Humanities often use raising-like passives to organise topic flow, but source and uncertainty still need to remain visible. **AI language understanding ↔ cross-clause roles:** robust parsing must distinguish surface object position from the semantic subject relationship inside the infinitive. ## Final checkpoint Why do we say: > I believe him to be honest. rather than: > *I believe he to be honest? Who is honest? Now compare: > I persuaded him to leave. Who was persuaded? Who left? Finally test: > I believe there to be a problem. Why is dummy **there** useful evidence that this is not an ordinary object-control relationship? If the learner can separate case form, semantic role and higher-verb meaning, the object + infinitive construction has become visible. ## Research basis This article is informed by the public research pass, including: – Cambridge, *English Syntax* index and discussion of Exceptional Case Marking: https://assets.cambridge.org/052183/4996/index/0521834996_index.htm – Cambridge Grammar, **Expect, Hope or Wait?**, including **expect + object + to-infinitive**: https://dictionary.cambridge.org/us/grammar/british-grammar/expect-hope-or-wait – Cambridge Core, **Raising: Dutch Between English and German**, including subject-to-object raising / accusative-with-infinitive comparisons: https://www.cambridge.org/core/journals/journal-of-germanic-linguistics/article/raising-dutch-between-english-and-german/B3411EAAA6246219A25346415C478265 – Oxford Handbook, **A-Movement in Language Development**, covering raising-to-object alongside raising-to-subject and passive: https://academic.oup.com/edited-volume/38688/chapter-abstract/335934702 – Cambridge Core, **Expletive NPs in Object-Position**, with *believe there to be* diagnostics: https://www.cambridge.org/core/journals/canadian-journal-of-linguistics-revue-canadienne-de-linguistique/article/expletive-nps-in-objectposition/DB6749F9CF02EF65BA2B273516BE5B37 The article uses **raising-to-object** and **ECM** as established analytical labels while explicitly avoiding the claim that every framework derives the construction through the same movement mechanism.