Antecedent-Contained Deletion

Resolve a verb-phrase ellipsis whose surface host lies inside the constituent that appears to antecede it, forcing a syntax–semantics analysis that breaks apparent self-containment while preserving ellipsis identity and scope readings.

Version

v3 · 2026-09-19 · History

Domain-specific #

1289

Domain group

Humanities

Origin domain

Linguistics & Semiotics

Subdomain

Syntax Semantics Interface → Linguistics & Semiotics

Aliases

ACD, Antecedent-contained ellipsis

Core Idea

Antecedent-Contained Deletion, or ACD, is a configuration of verb-phrase ellipsis in which the ellipsis site appears at surface structure inside a phrase that is itself contained in the apparent antecedent VP. In I read every book that Mary did, the auxiliary did supports an elided expression interpreted roughly as read, yet it occurs inside the relative clause modifying the object every book that Mary did; that object is inside the larger VP read every book that Mary did. Copying the full apparent VP into the ellipsis site would reproduce the containing phrase and trigger infinite regress. Syrett's handbook review identifies this containment puzzle, its quantificational structure, and its experimental scope evidence.

How would you explain it like I'm…

The Gap Inside Itself

When we talk, we often skip words and let the listener fill them in. In 'I read every book that Mary did', the word 'did' means 'read'. But if you tried to fill the gap by copying the whole part 'read every book that Mary did', the gap would still be inside what you copied, so you would have to copy again and again forever. Antecedent-Contained Deletion is the name for this tricky kind of sentence where the skipped part hides inside the part it copies.

The Never-Ending Copy Puzzle

In English we can leave out part of a sentence when it repeats something said earlier, like 'I can swim, and Sam can too', where 'swim' is left out after 'can'. Antecedent-Contained Deletion is a puzzling version of this. In 'I read every book that Mary did', the missing part after 'did' means 'read', but the gap sits inside the very phrase you would copy to fill it in. Copying that whole phrase would put the gap back in, over and over, without end. Language experts study how our minds avoid this loop and still understand the sentence easily.

Self-Containing Verb-Phrase Gap

Antecedent-Contained Deletion (ACD) is a kind of verb-phrase ellipsis, where a verb phrase is left out and understood from an earlier one (its antecedent). In 'I read every book that Mary did,' the word 'did' stands in for a missing verb phrase understood as 'read'. The puzzle is that the gap is inside the phrase 'every book that Mary did', which is itself inside the verb phrase 'read every book that Mary did', the natural antecedent. Copying the whole antecedent into the gap would put the gap back inside itself, an infinite regress. One well-known solution says that, for meaning, the phrase 'every book that Mary did' moves out of the verb phrase (called Quantifier Raising), leaving just 'read' to be copied. Other theories solve it differently, so ACD is really a puzzle that any theory must handle, not proof of one theory. Not every 'did' inside a relative clause counts: there must truly be a gap contained within its own antecedent.

 

Antecedent-Contained Deletion is a configuration of verb-phrase ellipsis in which the ellipsis site is, at surface structure, contained within its apparent antecedent VP. In 'I read every book that Mary did', the elided material is understood roughly as 'read', yet the auxiliary 'did' sits inside the relative clause modifying the object, and the object is inside the VP 'read every book that Mary did'; copying that VP would reproduce the containing phrase and generate infinite regress. The classic analysis invokes Quantifier Raising: at the interpretive level the quantified object DP moves outside the VP, leaving 'read t' as an antecedent that no longer contains the gap, which also links ACD to scope, since the host DP's landing site affects which antecedent and reading are available. QR is influential but not definitional: direct-compositional and variable-free analyses, such as Jacobson's, avoid the regress by treating the missing material as a transitive-verb-level function, and other approaches use flexible constituency, dependency structures, or semantic composition. Diagnosing ACD requires an understood predicate, an independently licensed ellipsis, a candidate antecedent, and a surface containment that creates circularity under ordinary antecedent identity. Judgments are sensitive to discourse, tense, finiteness, parallelism, and processing, and look-alikes include pseudogapping, null complement anaphora, do-so anaphora, and ordinary VP ellipsis without containment.

Scope of Application

ACD is literal when a predicate ellipsis site is surface-contained in the phrase forming part of its apparent antecedent and interpretation requires a principled escape from self-containment.

• Formal syntax. Testing covert movement, constituency, and locality.
• Formal semantics. Building compositional interpretations without circular copying.
• Ellipsis theory. Refining syntactic, semantic, and discourse identity conditions.
• Quantifier scope. Using available readings to constrain QR height or alternatives.
• Relative-clause structure. Locating the ellipsis host inside the quantified DP.
• Language acquisition. Testing children's access to lower and higher readings.
• Sentence processing. Comparing online cost and antecedent preferences.
• Cross-linguistic research. Establishing language-specific ellipsis and quantification prerequisites before analogy.

Clarity

A clear ACD analysis displays the sentence, marks the elided material, brackets the containing DP or phrase, identifies every candidate antecedent, and states the identity and scope assumptions. If QR is proposed, the landing position and lower copy or trace are represented, and the analysis shows exactly why the selected antecedent no longer contains the ellipsis. If a variable-free or direct-compositional account is proposed, the semantic type and composition that avoid self-application are stated.

Manages Complexity

ACD compresses a large theoretical issue into a sharp diagnostic configuration. The apparent cycle forces a theory to coordinate constituency, ellipsis identity, quantifier scope, and semantic composition. It reveals hidden consequences of seemingly simple copying accounts and provides scope evidence where covert structure is otherwise difficult to observe. Complexity returns because judgments vary, multiple analyses can generate the same reading, and experimental processing effects need not directly reveal grammatical representation.

Abstract Reasoning

1. Locate the overt ellipsis licensor and determine the category of missing material.
2. Identify the phrase that syntactically contains the ellipsis site.
3. Identify the VP or predicate constituent proposed as antecedent.
4. Test whether the containing phrase is itself inside that antecedent at surface structure.
5. Demonstrate the regress or identity failure produced by naive full copying.
6. List available lower and higher antecedents and the corresponding interpretations.

Knowledge Transfer

ACD transfers a general self-containment diagnostic: when an item depends on a representation that appears to contain that same item, naive substitution is unstable, so one must change the dependency domain, factor the representation, or alter semantic type. Similar reasoning appears in recursive definitions, macro expansion, cyclic dataflow, and anaphora. The linguistic accent is indispensable: ellipsis licensing, DP structure, quantifier scope, and compositional interpretation decide whether the apparent cycle is real and how it is resolved.

Relationships to Other Abstractions

Current abstraction Antecedent-Contained Deletion Domain-specific

Parents (1) — outgoing typed relations

• Antecedent-Contained Deletion is a kind of Dependency Prime

  Dependency is the strict parent by composition.

Hierarchy path (1) — routes to 1 parentless root

• Antecedent-Contained Deletion → is a kind of Dependency

Neighborhood in Abstraction Space

Antecedent-Contained Deletion sits in a sparse region of the domain-specific corpus (92^nd percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Syntax, Semantics & Grammar Theory (24 abstractions)

Nearest neighbors

• Discourse representation theory — 0.80
• Nota Accusativi — 0.79
• Predicate transfer — 0.79
• Adverbial complement — 0.78
• Discourse grammar — 0.78

Computed from structural-signature embeddings · 2026-10-25