11.2 Binding theory and constraints on coreference

Binding Theory and Coreference Constraints

Binding theory explains which noun phrases (NPs) in a sentence can refer to the same entity and which ones can't. It does this through a small set of principles that apply to three types of NPs: anaphors (like himself or each other), pronouns (like he or she), and R-expressions (referential expressions like proper names). These principles are grounded in syntactic structure, so understanding them requires knowing how sentence structure constrains meaning.

Principles of Binding Theory

Binding theory has three core principles, each governing a different type of NP. Before diving in, two key terms:

• Binding occurs when one NP (the antecedent) both c-commands and is coindexed with another NP. Coindexing means the two NPs are interpreted as referring to the same entity.
• Local domain is roughly the minimal clause (or complex NP) that contains the expression in question.

With those definitions in hand:

• Principle A: An anaphor (reflexive or reciprocal) must be bound within its local domain. It needs a c-commanding antecedent in the same clause.
• Principle B: A pronoun must be free (not bound) within its local domain. It can be bound by something outside that domain, but nothing inside it.
• Principle C: An R-expression must be free everywhere. No NP anywhere in the sentence can bind it.

These three principles work together to predict which coreference patterns are grammatical and which aren't.

Application of Binding Principles

Principle A in action (anaphors):

Reflexives like himself and reciprocals like each other need a local, c-commanding antecedent.

• ✅ John$_i$ likes himself$_i$. "John" c-commands "himself" within the same clause, so binding is satisfied.
• ❌ John$_i$'s mother likes himself$_i$. Here "John" is buried inside the subject NP "John's mother." It doesn't c-command "himself," so Principle A is violated.

Principle B in action (pronouns):

Pronouns must be free in their local domain but can pick up a referent from outside it.

• ✅ John$_i$ thinks he$_j$ is smart. "He" refers to someone other than John. The pronoun is free in its local clause (the embedded clause), so this is fine.
• ✅ John$_i$ thinks he$_i$ is smart. "He" refers to John. This still works because the antecedent "John" is outside the local domain of "he" (the embedded clause). Principle B only blocks binding within the local domain.
• ❌ John$_i$ likes him$_i$. If "him" is meant to refer to John, this violates Principle B because "John" c-commands and binds "him" inside the same clause.

Principle C in action (R-expressions):

Names and other referential expressions can't be bound by anything, anywhere in the sentence.

• ✅ He$_i$ thinks John$_j$ is smart. "John" is not bound by "he" (different indices), so Principle C is satisfied.
• ❌ He$_i$ thinks John$_i$ is smart. If "he" and "John" are coindexed and "he" c-commands "John," then "John" is bound, violating Principle C.

C-Command and Syntactic Structure

All three binding principles depend on c-command, a structural relation defined over syntax trees:

In practical terms, c-command captures a "structural superiority" relationship. A subject typically c-commands everything else in its clause, which is why subjects can serve as antecedents for reflexives in object position but not the other way around.

The binding principles use c-command to make their predictions:

• Anaphors must be c-commanded by their antecedent within the local domain (Principle A).
• Pronouns must not be c-commanded by a coindexed NP within the local domain (Principle B).
• R-expressions must not be c-commanded by any coindexed NP at all (Principle C).

This is why the same words can behave differently depending on where they sit in the tree. Syntactic position, not just linear word order, determines whether coreference is possible.

Limitations of Binding Theory

Binding theory captures important structural constraints, but it doesn't explain everything about how we interpret reference. Some notable gaps:

• Cross-sentence coreference: Binding theory operates within single sentences. It says nothing about how pronouns pick up referents across sentence boundaries in discourse (John left. He was tired.).
• Pragmatic and contextual factors: Real-world plausibility and context heavily influence how we resolve pronouns. In The hammer hit the vase and it broke, you interpret it as the vase based on world knowledge, not syntax.
• Null pronouns: Languages like Spanish and Italian allow subjects to be dropped entirely (pro). Binding theory in its standard form doesn't fully explain the distribution of these null elements.
• Discourse prominence: Which referent is most "accessible" or salient in a conversation affects pronoun interpretation in ways that go beyond syntactic structure.
• Morphological features: Gender, number, and animacy help narrow down possible antecedents but aren't part of the binding principles themselves.

Binding theory gives you a solid structural foundation for understanding coreference constraints. Just keep in mind that a full account of how reference works requires pragmatics, discourse structure, and more.