Noun-verb combinations are the structural backbone of nearly every sentence you produce or hear. When you say “break a promise,” “catch a cold,” or “run a business,” the specific pairing of noun and verb carries meaning that goes beyond the individual words. Your brain processes these pairings faster when they are familiar, assigns them to partially separate neural networks, and even reshapes verb meanings on the fly when a surprising noun shows up. The science behind how nouns and verbs interact touches on everything from reading speed to brain injuries to how modern AI systems understand language.
Why Certain Nouns and Verbs Belong Together
Not every verb can take every noun as its partner. You can “drink water” but not “drink a chair.” Linguists call this phenomenon selectional restriction: verbs impose constraints on the kinds of nouns they can sensibly combine with, based on semantic features. An information-theoretic model proposed in the 1990s formalized this idea by treating selectional constraints as probabilistic patterns rather than rigid rules, computed from how often specific verbs and nouns co-occur in real language use. This approach captured something important: semantic oddness is usually a matter of degree, not a binary on/off switch. “Drink soup” feels slightly unusual; “drink a conversation” feels deeply strange. The model handles that gradient naturally because it works from observed co-occurrence frequencies rather than hard-coded definitions of what each word means.1PubMed. Selectional constraints: an information-theoretic model and its computational realization
The strength of these restrictions varies enormously from verb to verb. Some verbs are promiscuous: they happily combine with a wide range of nouns. Others are highly selective. Research on the Ebughu language documented this variation by mapping eight verb clusters and finding that only two of them had what the researchers called “nuclear” verbs, meaning verbs flexible enough to co-occur with many different noun objects. The nuclear verbs in those clusters were the equivalents of “buy” and “cut,” which can take virtually any noun whose meaning is compatible with being purchased or being severed. The remaining six clusters had no such flexible core verb; their members were each restricted to a narrow set of nouns.2Theory and Practice in Language Studies. Verb-noun (Object) Selectional Restriction in Ebughu
This pattern holds broadly across languages. Verbs describing concrete physical actions (“cut,” “eat,” “throw”) tend to accept a wide range of nouns because many things can be physically manipulated. Verbs describing narrow or specialized actions (“percolate,” “conjugate”) pair with far fewer. Understanding these natural pairings matters for language learners, writers, translators, and anyone building tools that work with language.
Your Brain Reads Familiar Pairings Faster
When you encounter a noun-verb combination you have seen many times before, your eyes move through it more quickly. Eye-tracking research on Chinese readers tested this by comparing how people read verb-noun collocations (frequent pairings) versus novel expressions made of the same individual words rearranged into unfamiliar combinations. The collocations consistently showed a processing advantage. Readers spent less time fixating on them and moved through the text more smoothly. The effect was even stronger for “strong” collocations, meaning pairings that appear together very frequently, compared to “weak” collocations that co-occur less often. Crucially, this speed boost was independent of how predictable the phrase was from surrounding context, suggesting that the brain stores something about the pairing itself, not just the likelihood of one word following another in a particular sentence.3PubMed Central. Is there a processing advantage for verb-noun collocations in Chinese reading? Evidence from eye movements during reading
This advantage is not limited to words sitting right next to each other. A separate study found that collocations are processed faster than matched control phrases even when the noun and verb are separated by other words in the sentence. The facilitation effect was real for these nonadjacent collocations, though it appeared to be somewhat larger when the verb and noun were adjacent.4PubMed. Are nonadjacent collocations processed faster? In practical terms, this means your reading brain is constantly tracking multi-word patterns, pulling them from memory as pre-assembled chunks rather than computing each word’s fit from scratch every time.
For anyone writing professionally, this research carries a real implication. Familiar noun-verb combinations reduce cognitive load on the reader. Using them when clarity matters (instructions, legal documents, health information) speeds comprehension. Breaking them deliberately, as poets and novelists do, creates a productive friction that slows the reader down and forces more active interpretation.
Nouns and Verbs Recruit Different Brain Regions
One of the more striking findings in neuroscience is that producing a noun and producing a verb are not the same task for your brain. Brain imaging studies have shown that when people generate verbs, the left prefrontal cortex and the left superior parietal lobule become particularly active. When people generate nouns, a region in the left inferior temporal lobe lights up instead.5PubMed Central. Cortical signatures of noun and verb production A commentary on this research noted that the left anterior fusiform gyrus, also in the temporal lobe, was activated during noun production, while verb production recruited the left superior temporal gyrus in addition to the frontal and parietal areas.6Current Biology. Functional Anatomy: Nouns, Verbs, and Objects
The proposed explanation is that these brain regions are involved in representing fundamentally different kinds of concepts. Nouns tend to represent objects and entities, which are primarily identified by their sensory properties: what they look like, how they feel, where they are found. The temporal lobe regions activated during noun production are closely tied to visual and sensory processing. Verbs represent actions and events, which require understanding of motion, force, causation, and sequences in time. The frontal and parietal regions activated during verb production are involved in planning, sequencing, and spatial reasoning. In other words, the grammatical split between nouns and verbs is not just a rule humans invented for grammar textbooks; it reflects a genuine organizational divide in how the brain stores and retrieves different kinds of meaning.
What Brain Damage Reveals About Noun-Verb Separation
The strongest evidence that nouns and verbs are handled by partly independent neural systems comes from people who lose the ability to use one category while retaining the other. In aphasia caused by stroke or other brain injuries, some patients develop difficulty naming actions (verbs) while still naming objects (nouns) reasonably well, and a smaller number show the reverse pattern. A study of aphasic patients found that selective difficulty with verbs was more common, affecting about a third of the sample, while selective difficulty with nouns affected about one in ten.7PubMed. Verb-noun double dissociation in aphasic lexical impairments: the role of word frequency and imageability
The anatomy lines up with the imaging studies. Research on 15 patients with predominant verb impairment and 5 with predominant noun impairment found that noun impairment was mostly tied to lesions in the middle and inferior left temporal area. Verb deficits, by contrast, were associated with three different lesion sites: left posterior temporo-parietal lesions, large left fronto-temporal lesions, or deep lesions involving the insula and basal ganglia.8PubMed. Verb-noun double dissociation in aphasia: theoretical and neuroanatomical foundations The fact that verb deficits can arise from multiple different lesion locations hints that verb processing depends on a broader and more distributed network than noun processing does.
Further work has shown that the dissociation can occur at multiple levels of language processing. Some patients struggle to retrieve the right word for an action but can still apply grammatical rules to verbs normally. Others can retrieve verb words but cannot form them into correct grammatical structures. In one study, two patients with poor object naming and two out of ten with poor action naming could still apply identical grammatical transformations to both nouns and verbs without errors, meaning the word-finding problem was separate from the grammar problem.9PubMed. The dissociability of lexical retrieval and morphosyntactic processes for nouns and verbs: A functional and anatomoclinical study This matters because it tells clinicians that a patient who cannot name actions may still understand and produce action-related grammar, opening different rehabilitation strategies for different levels of breakdown.
When Nouns and Verbs Clash, Verbs Give Ground
What happens when you deliberately pair a noun with an odd verb? Consider “The stone walked to the store.” The noun “stone” and the verb “walked” are a strange match, creating what researchers call semantic strain. The interesting finding is that in these strained combinations, verb meanings bend more than noun meanings do. People interpreting a sentence like “The stone walked” tend to preserve the noun’s identity (it is still a stone) and instead shift the verb’s meaning (maybe it rolled, or was carried, or moved slowly). Across three experiments, researchers confirmed this verb mutability effect: under increasing semantic strain, verb meanings changed more while noun meanings remained stable.10PubMed Central. Verb Metaphoric Extension Under Semantic Strain
This asymmetry has far-reaching consequences for how language evolves and how metaphors work. Because verbs are more semantically flexible, they are the primary vehicle for metaphorical extension. When we say “the economy shrank” or “ideas flow,” we are exploiting the verb’s willingness to shift its meaning to fit a new noun. The noun anchors the interpretation, and the verb adjusts. This is one reason noun-verb combinations are so productive for creative language: the verb half of the pair is built to stretch.
Idioms as Frozen Noun-Verb Pairings
Idioms like “kick the bucket” or “spill the beans” look like ordinary noun-verb combinations on the surface but carry meanings that do not follow from the individual words. A long-standing question is whether idioms are completely frozen units stored whole in memory, or whether their internal parts still carry some independent meaning. The answer appears to depend on the specific idiom. Research has found that idioms whose individual components contribute to the overall figurative meaning, like “go out on a limb” (where “limb” suggests precariousness), are judged as less disrupted when you swap in a synonym (“go out on a branch”) compared to idioms where the parts carry no individual figurative weight, like “kick the bucket” changed to “punt the pail.”11PubMed. Speakers’ assumptions about the lexical flexibility of idioms
Corpus analysis of idiom variants drawn from 450 million words of American English has provided additional support for this decomposability idea. The more an idiom’s parts contribute to its meaning, the more morphological flexibility the noun component shows, such as changes in number or the addition of modifiers. Of the flexibility measures tested, number variation in the noun was significantly dependent on how decomposable the idiom was.12Yearbook of Phraseology. Idiom variation and decomposability Part II: Variation in the noun phrase The practical takeaway is that idioms are not a separate species of language entirely walled off from regular noun-verb combinations. They exist on a continuum, with the most transparent ones behaving quite similarly to ordinary collocations and the most opaque ones resisting any modification.
Reading Action Words Activates Your Motor System
When you read a sentence like “she kicked the ball,” your brain does not just decode the symbols. According to a body of research on embodied cognition, understanding action-related noun-verb combinations involves a kind of mental simulation: your motor system partially reactivates, as if you were performing or observing the action described. The Action-Sentence Compatibility Effect shows that people respond physically faster when the action described in a sentence matches the direction of a response they are asked to make. If a sentence describes pushing something away and you need to push a button away from your body, you are quicker than if the sentence described pulling something toward you.13PubMed Central. Spatial and Motor Aspects in the “Action-Sentence Compatibility Effect”
This perspective holds that language comprehension, especially for action verbs, is grounded in the sensorimotor system. The same perception, action, and emotional mechanisms you use when physically interacting with objects are recruited when you simply read about those interactions.14PubMed Central. Sentence processing: linking language to motor chains The noun in the combination specifies what object is involved; the verb specifies what action to simulate. Together, they activate a rich motor and sensory representation that goes well beyond the dictionary definitions of the individual words. This is one reason why vivid, concrete noun-verb combinations (“shattered the glass,” “kneaded the dough”) feel more engaging than abstract ones: they give your motor system something specific to latch onto.
How AI Systems Handle Noun-Verb Knowledge
Modern large language models have absorbed billions of noun-verb combinations from their training data, and they have picked up a lot of the selectional knowledge that comes with those pairings. Testing reveals that these models are quite good at distinguishing possible events from impossible ones. Given a pair like “The teacher bought the laptop” versus “The laptop bought the teacher,” the models almost always assign higher likelihood to the sensible version. But the picture gets murkier with events that are merely unlikely rather than impossible. The distinction between “The nanny tutored the boy” and “The boy tutored the nanny” is harder, because both are technically possible, just differently probable. LLMs show much less consistent preferences for the likely over the unlikely version.15PubMed. Event Knowledge in Large Language Models: The Gap Between the Impossible and the Unlikely
This gap matters for anyone relying on AI-generated text or AI-assisted writing. The models are good at avoiding obviously absurd noun-verb combinations, but they can stumble on the subtler distinctions that humans navigate effortlessly. A sentence that is grammatically correct and semantically possible but contextually odd may sail through an AI system without being flagged, because the model’s knowledge of how nouns and verbs relate is built from surface-level co-occurrence statistics rather than the embodied, experiential understanding that humans bring to language.
On the computational linguistics side, researchers have also worked on fine-grained automatic classification of verb-noun collocations, using semantic typologies to sort pairings into categories based on the relationship between the verb and noun. This work, which has been demonstrated with Spanish data, draws on a framework that identifies different functional types of verb-noun pairings and trains algorithms to recognize them from semantic features.16Natural Language Engineering. Towards automatic fine-grained semantic classification of verb-noun collocations The goal is to move beyond simply detecting that two words co-occur frequently and toward understanding why they co-occur: whether the verb intensifies the noun, whether it initiates it, whether it describes its creation, and so on.
Languages Where the Noun-Verb Line Gets Blurry
English speakers tend to assume that every word falls clearly into a “noun” box or a “verb” box. But cross-linguistic research complicates this picture substantially. There is growing evidence that the distinction between nouns and verbs is scalar rather than absolute across the world’s languages, and that in some languages the boundary may be extremely weak or difficult to identify at all.17Rivista di Linguistica. Verbs and nouns from a cross-linguistic perspective Some languages have what look like minor verb classes (very few dedicated verb words) or minor noun classes, relying instead on context and grammatical markers to do the work that English splits between distinct word categories.
English itself illustrates the fuzziness, because many stems function as both nouns and verbs without any change in form. “Walk,” “love,” “kill,” “run,” “cut” — all of these are used freely as both. One analysis proposes that such words are not secretly nouns being used as verbs (or vice versa) but instead belong to a third category called “flexibles,” which can map to either the predicate function (verb-like) or the argument function (noun-like). Under this account, the most common language type in the world has all three categories: nouns, verbs, and flexibles.18Language Sciences. Nouns, verbs and flexibles: implications for typologies of word classes The existence of flexibles means that many of the noun-verb combinations we use every day are not truly combining two separate categories at all but rather putting a flexible stem into a specific grammatical role.
This cross-linguistic variation matters for understanding the processing research discussed earlier. If nouns and verbs are partly distinguished by how the brain represents objects versus actions, then languages that blur the categories might handle those representations differently. Children acquiring language seem to be sensitive to these patterns from an early age. Research on late language learners in a signing context found that aspects of two-word combinations that are primarily linguistic in nature, like utterance length, tracked language experience, while semantic relations between words tracked cognitive development more broadly.19PubMed Central. The Two-Word Stage: Motivated by Linguistic or Cognitive Constraints? In other words, how nouns and verbs get combined in early speech reflects both the structure of the language being learned and the developing mind of the learner, with those two factors pulling in partly independent directions.
Internet Language and the Creation of New Pairings
Online communication has become a laboratory for novel noun-verb combinations. Internet slang frequently generates new pairings through processes like conversion (turning a noun into a verb: “to Google,” “to Uber”), blending, and semantic shift, where existing verb-noun combinations acquire entirely new connotations in digital contexts. A synthesis of recent work on internet language describes how lexical creativity online relies heavily on these mechanisms, producing combinations that would have been incomprehensible a decade ago but now feel natural to millions of speakers.
What makes online noun-verb creativity different from traditional figurative language is the speed at which new combinations spread and either solidify or disappear. A phrase like “doom scrolling” pairs a noun and a verb (well, a gerund) in a way that was meaningless before smartphones existed. Once it catches on, it follows the same trajectory as any other collocation: repeated exposure builds a processing advantage, and eventually your brain reads the combination as a single chunk rather than two separate words. The verb mutability research discussed earlier helps explain why verbs are so often the words being stretched in these new formations. Verbs are inherently more flexible in meaning, so they accommodate new nouns more readily than nouns accommodate new verbs.
For writers and communicators, this ongoing churn of noun-verb combinations is both an opportunity and a hazard. New pairings can feel vivid and precise to an audience that shares the right cultural context, but opaque or even grating to readers outside that community. The same collocational processing advantage that makes familiar pairings easy to read works against novel combinations, at least until they become familiar enough to earn their own chunk in memory.