Leaving Words Out: Gapping and Ellipsis

Ellipsis is not careless incompleteness. It is the systematic omission of words whose structure and meaning the listener can recover. Hebrew uses it constantly: a repeated verb disappears from a coordinated clause, a whole predicate is represented by כן or לא, or only a question word survives after אני לא יודע.

Bare answers and conversational cutoffs are discussed as a discourse genre on fragments. This page asks a narrower syntactic question: what unspoken material is licensed by an overt parallel clause?

Gapping: keep the contrasts, omit the repeated verb

In gapping, two coordinated clauses share a verb. The first clause pronounces it; the second keeps the contrasting subject and object or directional phrase but omits the repeated verb.

The omitted material must be recoverable as the same predicate. The remaining pieces form a parallel contrast: אני versus רונית, לחם versus חלב. This symmetry is why the short second clause is clear. Coordination itself—including punctuation and conjunction choice—belongs to coordination syntax.

One subject across a chain of verbs

A different omission occurs when one subject controls a coordinated sequence. Past and future Hebrew verbs carry person, number and often gender, so a stated subject need not repeat before every verb.

In the past example, קם, התלבש and יצא are all masculine singular; in the future example, the three verbs all mark first-person plural. The omitted subject stays stable. Present-tense participles do not mark person, so a long present-tense chain may need an overt pronoun for clarity. The broader conditions belong to subject dropping.

This construction is not gapping: no second subject contrasts with the first. It is one subject shared across coordinated predicates.

כן and לא as predicate pro-forms

Hebrew can let כן stand for an affirmative version of a previously mentioned predicate and לא for its negative version. English usually needs do/does/did, so the Hebrew bare כן has no one-word structural equivalent.

In the first example, כן recovers רוצה לצאת. It does not merely answer “yes”; it reverses the polarity of the earlier negative predicate. In the second, רציתי selects an understood להתקשר. Hebrew and English both allow that infinitival complement to remain silent.

Here כן appears before an overt verb because it contrasts the positive event with the earlier negative expectation. That is emphasis, not ellipsis. Compare הוא כן, where the predicate itself is silent.

Sluicing: only the question phrase survives

In sluicing, an indefinite in one clause creates an unknown value and a following embedded question keeps only the question phrase. The rest of the question is understood.

The silent material after מי is approximately התקשר. The remnant normally behaves like the corresponding full question: מי התקשר? With a governed preposition, the reliable learner pattern likewise preserves the preposition found in the full question.

Hebrew never strands bare עם after the question word. Retaining the preposition—עם מי—is also the stable norm under sluicing and is obligatory with many governed prepositions, including ל־. There is nevertheless documented speaker variation in prepositionless מי after עם: some speakers accept it, some reject it and others find it marginal. Use עם מי as the dependable model, but do not recast every prepositionless sluice as universally impossible. The categorical ban is on a stranded shape such as מי עם; see no preposition stranding.

Stripping: one contrastive remnant

Stripping leaves a single contrastive constituent, commonly with גם or לא. The previous clause supplies the missing predicate.

גם אני can also be a standalone conversational answer. Its classification then depends on discourse: as a response it is a fragment; after a parallel clause it is the remnant of stripping. The words are identical, but the licensing context differs.

Prepositions survive short answers too

The no-stranding constraint remains visible in the shortest exchanges. A question headed by למי is answered with a matching ל־ phrase; על מה is answered with על.

The short answer may omit an entire clause, but it cannot discard a preposition that belongs to the surviving phrase. English transfer is especially tempting because English permits “Who did you speak to?” and often answers with a bare name.

Six turns, four ellipsis types

The following ordinary exchange combines gapping, stripping, sluicing and a כן/לא predicate pro-form. The labels describe the silent syntax; speakers do not experience the dialogue as a puzzle.

Common Mistakes

1. Stranding a preposition under sluicing. Hebrew keeps it attached to the question word.

2. Translating bare כן as an ordinary “yes” when it contrasts predicates. In this context it means “does/did/will.” Hebrew can either keep the predicate after contrastive כן or let bare כן recover it.

3. Assuming every short second conjunct has one obvious gapping analysis. A sequence such as ורונית אתמול can acquire a recoverable reading in a sufficiently rich context, so it is not categorically ill-formed. When the intended contrasts are not immediately parallel, pronounce the second verb.

4. Condemning a licensed fragment merely because words are absent. The context determines whether recovery works.

Key Takeaways

• Gapping removes a repeated verb from a parallel coordinated clause.
• One subject can span a coordinated verb chain when agreement keeps its reference recoverable.
• Bare כן and לא can stand for positive and negative predicates; English usually needs do/does/did.
• Sluicing leaves a question phrase, and stripping leaves one contrastive remnant.
• The reliable learner pattern preserves governed prepositions under ellipsis: עם מי, למי and על מה. Bare postposed prepositions are impossible; restricted prepositionless sluices vary across speakers.