AI Glossary / Natural language and speech / Beginner
Verb
Definition
Definition: A verb is a grammatical class of word that commonly expresses an action, event, process or state and helps organize a clause. Examples include repair, arrive, change and know. Its category depends on grammatical behavior, not simply whether the meaning involves visible movement.
What it means
What it means: The familiar description “doing word” is a useful starting point but incomplete. Know and belong can function as verbs without describing a physical action. Conversely, an action can be named by a noun, as in “the repair took an hour.”
Verbs often indicate or combine with information about tense, aspect, mood and voice. Languages express those properties in different ways, using endings, auxiliaries, particles or context. A verb also affects what other participants a clause can contain, such as someone giving something to someone else. No single English spelling pattern identifies every verb.
How it works
How it works: In language processing, a part-of-speech tagger assigns a category to each occurrence in context. A morphological analyzer can identify forms and features, while a parser relates a verb to its subject, objects and other dependents. A lemmatizer may connect repaired and repairs to the base form repair.
Tagging conventions matter. Universal Dependencies uses VERB for main verbs and a separate AUX category for auxiliaries and certain copulas. Therefore a count of VERB labels is not necessarily a count of every word that a traditional grammar calls a verb. Participles and related forms also require contextual analysis because they can have verbal, adjectival or nominal uses.
Three practical examples
Three practical examples:
1. A maintenance assistant extracts actions from “Inspect the seal, then replace the filter.” Recognising inspect and replace as verbs supports further analysis, but does not establish that the instruction is authorized or physically appropriate.
2. A search system links repaired, repairing and repair where their uses share the relevant lemma. It still checks context because the same spelling can be a noun or a verb.
3. A complaint analyzer distinguishes “The device failed” from “The device may fail.” Identifying the main verb helps, but the auxiliary and surrounding construction are needed to distinguish a reported event from a possibility. A keyword match for fail alone loses that distinction.
Think of it this way
Think of it: A verb can resemble the central event field in a form, with other fields describing participants, timing and circumstances. The analogy helps explain why verbs connect to clause structure. It is limited because clauses can be more complex than forms, not every language requires an overt verb in every predicate, and meaning also depends on context beyond the clause.
Worked example
Worked example: Consider two sentences: “They email the invoice.” “The email arrived.” Ignoring punctuation, the first contains four word occurrences and the second three. In the intended reading, email is a verb in the first sentence and a noun in the second. Arrived is another verb, with lemma arrive.
There are therefore two main-verb occurrences among seven word occurrences: 2/7, approximately 28.6%. There are two occurrences of the spelling email, but only one is a verb. A dictionary rule that always marks email as a verb would produce one false positive in this example.
The grammatical evidence includes the first email's relationship to its subject and object, and the second email's position after a determiner as the subject of arrived. The fraction is only a description of this tiny invented sample; it is not an estimate of verb frequency in business writing.
Illustrative case study
Illustrative case study: Fictional Slate Support wants to classify customer messages by requested action. Its first rule treats every occurrence of refund as an instruction to issue a refund. It misreads “The refund arrived” and “Can you explain the refund policy?”
The team adds contextual grammatical analysis and distinguishes requests from descriptions and questions. It tests “Please refund the charge” separately from noun uses and from negated or quoted requests. Human approval remains attached to any actual account action. This fictional case shows why verb recognition can support interpretation while falling short of complete intent detection or business authorization.
Common misconceptions and mistakes
Common misconceptions and mistakes: A verb is not always an action word, and a word naming an action is not always a verb. Ending in -ing does not settle a word's role without context. A lemma is not a tense label. An auxiliary may receive a different computational tag from a main verb. The subject of a passive clause need not be the person performing the action. Finally, finding a verb does not prove that the event happened: negation, modality, questions and reported speech can alter the interpretation.
Business use
Business use: Verbs are useful for extracting events, organizing task descriptions and improving search, but reliable applications combine grammatical labels with context. Keep the annotation scheme explicit, especially when comparing outputs from different tools. Evaluate confusing noun/verb spellings, auxiliaries, negation and domain terminology. Distinguish a linguistic observation from an operational decision: the sentence “delete the record” can occur in a warning, quotation or hypothetical example. A text-processing label should not itself trigger an irreversible business action.
People also ask
People also ask: Can one spelling be a noun and a verb?
Yes. Email, plan and repair can have different roles depending on their sentence context.
Do all verbs show tense with an ending?
No. Languages and constructions differ, and even English expresses many distinctions through auxiliaries or other context.
Why does a tagger label has as AUX instead of VERB?
In a construction such as “has arrived,” its annotation scheme may distinguish the auxiliary from the main verb. Check the tagset and the particular use.
Check your understanding
Check your understanding: A system sees “They plan the launch” and “The plan changed.” Should every occurrence of plan receive the same part-of-speech label?
Answer: No. Plan is used as a verb in the first sentence and a noun in the second. Context determines the relevant role; a fixed spelling lookup alone cannot resolve it.
Related terms
- Noun
- Part of speech
- Lemma
- Aspect
- Predicate (logic)
- Verb phrase
- auxiliary verb
- tense
- predicate
- argument structure
Learn next
collect a noun/verb pair with identical spelling and explain its role in each sentence. Then add an auxiliary, negation or passive construction. Track which words carry the main event and which change its timing, possibility or relationship to participants.
More in Natural language and speech
- Variable-length encoding - Definition: Variable-length encoding represents different items using code sequences of different lengths.
- Vauquois triangle - Definition: The Vauquois triangle is a conceptual diagram of machine-translation approaches.
- Velar sound - Definition: A velar sound is a speech sound whose primary constriction involves the back of the tongue and the velum, or soft palate.
- Verb alternations - Definition: Verb alternations are related ways in which a verb's participants can be expressed in sentence structure.
- Viterbi algorithm - Definition: The Viterbi algorithm finds the highest-scoring complete sequence of hidden states in a model whose score decomposes into suitable local...
- Vocabulary mismatch - Vocabulary mismatch occurs when people or systems use different words for the same or closely related meaning, preventing a useful match.
Sources
Sources: Universal Dependencies, VERB definition and distinction from AUX, including context-dependent participle and gerund classifications: https://universaldependencies.org/u/pos/VERB.html . Jurafsky and Martin, Speech and Language Processing, Chapter 18, word classes and contextual tagging: https://web.stanford.edu/~jurafsky/slp3/18.pdf . The email calculation and Slate Support case are original illustrations. English examples illustrate rather than define the grammatical possibilities of every language.
Previous: Venn diagrams Next: Verb alternations