Can a phrasal verb be separated by an object?

Some English constructions allow this, as in the worked example. The behavior must be checked for the particular construction.

Is its meaning always completely unpredictable?

No. Degree of compositionality varies; the important point is that interpreting the construction can require more than independent word meanings.

Does detecting one prove an event happened?

No. Negation, questions, instructions, and hypothetical contexts can contain the construction without asserting the event as a fact.

AI Glossary / Natural language and speech / Beginner

Phrasal verb

Definition

A phrasal verb is a construction in which a verb combines with a particle to function as a lexical unit, often with a meaning that cannot be predicted simply by combining the ordinary meanings of the individual words. English examples include expressions such as “turn down” in the sense of rejecting. In natural language processing, recognizing the whole construction can be essential to interpreting the sentence correctly.

What it means

The term's boundaries vary across grammatical descriptions. Some broader teaching traditions group several kinds of multiword verbs together, while more specific analyses distinguish verb-particle constructions from prepositional verbs. This entry concentrates on the verb-plus-particle sense used in the source textbook.

A phrasal verb can contain common words but express a specialized or idiomatic meaning. Word order can also vary in some constructions, so the relevant verb and particle are not always adjacent. These properties make a simple word-by-word interpretation or fixed neighboring-word search unreliable as a complete solution.

How it works

A language system identifies a candidate verb and an associated particle, then interprets their construction in context. Grammatical relationships help determine whether the words belong together and which participants or objects are involved.

For a separable construction, an intervening object does not necessarily break the relationship. A parser or construction-aware matcher can connect the verb with its particle across that object. It must also distinguish idiomatic uses from literal spatial uses that happen to contain similar words.

After recognition, the system can propose a suitable interpretation or paraphrase, while preserving tense, negation, modality, and participants. It should not replace a whole sentence with an event label that loses whether the action was completed, requested, denied, or merely possible. The construction is one component of interpretation, not a substitute for the rest of the sentence.

Three practical examples

1. Customer-message analysis: A service tool distinguishes a request to cancel an arrangement from an unrelated literal use of the same common verb.

2. Translation support: A system selects a target-language expression appropriate to the complete construction instead of translating its verb and particle independently.

3. Search expansion: A document service considers a relevant single-verb paraphrase when retrieving passages that use an equivalent phrasal expression, with context checks to limit false matches.

Think of it this way

Think of a phrasal verb as a familiar pair of words that sometimes acts like one entry in a phrasebook. Knowing each word separately does not guarantee that the pair's intended meaning is clear. The analogy is limited because the pair can be separated in a sentence, can admit modification, and still requires grammatical and contextual analysis.

Worked example

Consider two original sentences: “Mina called off the test” and “Mina called the test off.” In the intended reading, both describe Mina canceling the test. Each sentence has five words, and each contains one occurrence of the verb-particle construction “call off.”

A fixed adjacent-word matcher for “called off” recognizes the first sentence but misses the second. On this deliberately tiny two-example set, its recall for the target construction is 1/2=50%. A matcher that connects the verb and particle through an allowed object position could recognize both, although its false matches would still need separate evaluation.

Now add “Mina did not call off the test.” The construction is present, but the sentence does not positively assert cancellation. There are three construction mentions across the three examples and only two positive cancellation assertions. These counts are independently checked teaching cases, not the measured performance of a parser or a claim that every phrasal verb permits the same word-order variation.

Illustrative case study

Fictional case: Cedar Scheduling analyzes meeting notes for canceled activities. Its first matcher misses notes where the object appears between a verb and its particle. Reviewers add structural examples, including negated and hypothetical statements, before revising the extraction rule.

The team evaluates whether the output preserves who canceled what and whether cancellation was actually asserted. It retains the original sentence beside proposed records. This scenario demonstrates a possible language-processing issue and review method, without claiming an actual deployment or business improvement.

Common misconceptions and mistakes

Do not assume every verb followed by a short spatial word is an idiomatic phrasal verb. Literal movement expressions and other grammatical constructions require their own analysis. Nor are all phrasal verbs freely separable or interchangeable with one simple verb in every context.

Avoid treating a particle's absence from an adjacent word window as proof that the construction is absent. Also distinguish grammatical terminology from computational labels: an item described as a verbal particle may receive an ADP or ADV category rather than PART in a particular annotation scheme.

Business use

Phrasal-verb recognition matters in informal correspondence, support conversations, and instructions, where common verbs often occur in multiword expressions. It can reduce obvious failures caused by translating or matching each word in isolation.

Build evaluation examples around the constructions relevant to the workflow. Include alternate permitted word orders, literal lookalikes, and statements with negation or modality. Monitor complete extracted meanings rather than only whether two words were found, because a correctly detected construction can still lead to a wrong factual or operational conclusion.

Check your understanding

Why does “Mina called the test off” challenge an adjacent matcher for “called off”?

Answer: The object “the test” intervenes between the verb and particle, while their constructional relationship and intended cancellation meaning remain.

Related terms

Particle
Multiword expressions
Negation
multiword expression
idiom
lexical semantics
dependency parsing
construction
paraphrase

Learn next

Compare a phrasal verb with a literal verb-plus-spatial expression. Then test a separable construction in several word orders and explain which grammatical information a matcher must preserve to avoid misleading event records.

Sources

Sources: Jurafsky and Martin, Speech and Language Processing, Chapter 18, particles and phrasal verbs: https://web.stanford.edu/~jurafsky/slp3/18.pdf . Universal Dependencies compound:prt relation and literal-use distinction: https://universaldependencies.org/u/dep/compound-prt.html . Universal Dependencies PART scope: https://universaldependencies.org/u/pos/PART.html . Relevant primary passages were read. This article uses the narrower verb-particle sense; all cancellation examples, counts, and the fictional case are original.

Previous: Phrasal template Next: Phrase search