Questions: C-command and Binding Theory

Question 1

In the sentence 'John believes that Mary likes him,' can 'him' refer to John? Why or why not?

• A. No — 'him' is a pronoun and must be free everywhere in the sentence, so it cannot refer to any NP in the sentence
• B. Yes — 'him' is a pronoun (Principle B), which must be free only in its local domain (the embedded clause); John is outside that domain, so co-reference is permitted
• C. No — John c-commands 'him' throughout the entire sentence, blocking any co-reference
• D. Yes — pragmatic plausibility allows 'him' to refer to John regardless of structural constraints

Answer: (B)

Principle B states that a pronoun must be *free* in its local domain — the minimal clause containing it and a potential c-commanding antecedent. Here the pronoun 'him' is in the embedded clause 'Mary likes him.' John sits in the matrix clause, outside this local domain. Since John does not c-command 'him' within the embedded clause, co-reference is grammatically available. Contrast this with *John hurt him* — there, John is in the same clause and c-commands 'him,' blocking co-reference.

Question 2

In 'The manager expects John to promote himself,' why can 'himself' not refer to 'the manager'?

• A. Because 'himself' is not licensed by any semantic content in the sentence
• B. Because Principle A requires 'himself' to be bound locally — its local domain is the embedded infinitival clause where 'John' is the c-commanding NP, not 'the manager'
• C. Because 'the manager' is too far removed from 'himself' for any c-command relationship to apply
• D. Because reflexives can only take sentence-initial NPs as antecedents

Answer: (B)

Principle A requires an anaphor (reflexive) to be *bound* — have a c-commanding antecedent — within its local domain. The local domain for 'himself' is the infinitival clause 'John to promote himself.' Within that domain, 'John' c-commands 'himself,' satisfying Principle A. The manager is outside this local domain and therefore cannot bind 'himself.' This shows that binding is determined by tree geometry (local c-command), not linear proximity or semantic plausibility.

Question 3

An R-expression like 'the senator' cannot co-refer with any noun phrase that c-commands it, regardless of where in the sentence that noun phrase appears.

• T. True
• F. False

Answer: True

Principle C states that R-expressions (full referring noun phrases) must be *free everywhere* — no c-commanding noun phrase anywhere in the sentence can be co-referential with them. This is why sentences like 'He₁ said that the senator₁ was corrupt' are ungrammatical with co-reference: 'he' c-commands 'the senator' from a higher position. The constraint applies without locality restrictions, unlike the local-domain requirements of Principles A and B.

Question 4

C-command is a symmetric relation: if node A c-commands node B, then B necessarily c-commands A as well.

• T. True
• F. False

Answer: False

C-command is explicitly *asymmetric*. A c-commands B if the first branching node dominating A also dominates B, and A does not dominate B. A subject NP c-commands the VP and everything within it, but the VP does not necessarily c-command the subject. This asymmetry is what gives c-command its explanatory power for reference: it captures the directionality of binding — antecedents are structurally higher than the pronouns or reflexives they bind.

Question 5

Why is c-command a *structural* account of pronoun distribution, rather than a linear (left-to-right) one, and why does this distinction matter for linguistic theory?

Model answer: C-command is defined entirely on the hierarchical tree structure — it depends on dominance relations, not on which word comes first in the string. This matters because linear order alone fails to predict reference patterns. In languages with different word orders, or in sentences with extraposition and movement, the same structural c-command relations hold and predict the same reference constraints, even when the surface order differs. A purely linear account would incorrectly predict that pronouns preceding their antecedents are always ungrammatical, but 'Near John₁, he₁ saw the snake' is acceptable. The structural account captures the generalization across all word orders and constructions.

The argument that reference constraints are structural rather than linear is one of the core arguments for syntactic theory over purely sequential models of language. C-command generalizes across languages with radically different surface orders, suggesting that the relevant level of description is hierarchical phrase structure, not the observable string. Binding theory is thus an argument not just about pronouns but about the psychological reality of syntactic trees.