Ergativity is a grammatical pattern found in hundreds of the world’s languages, in which the way a language marks “who did what to whom” follows a different logic than what speakers of English or most European languages would expect. In English, “she” stays “she” whether she runs or she kicks someone. Ergative languages flip that: the subject of “she runs” gets treated like the object of “she kicks him,” and it is the subject of the transitive sentence that gets singled out for special marking. The result is a system where the doer of an action on someone else is the odd one out grammatically, rather than the one who has the action done to them.
How Ergativity Differs from the Familiar Pattern
Most readers are accustomed to what linguists call nominative-accusative alignment without ever knowing the term. In English, the pronoun “he” stays in the same form whether you say “he sleeps” or “he sees her.” It is the object, “her,” that changes form. The subject gets consistent treatment no matter the verb. In an ergative-absolutive language, the logic runs differently. The person who sleeps and the person who gets seen share the same grammatical form (called the absolutive). It is the person who actively does something to someone else, the agent in a transitive sentence, who takes a distinct marker (called the ergative). Think of it this way: English treats “the one doing” as the default and singles out “the one receiving.” Ergative languages treat “the one receiving” as the default and single out “the one doing to someone else.”
This is not a quirk of a handful of obscure languages. Basque, in southwestern Europe, is ergative. So are many languages of the Caucasus, Aboriginal Australia, the Tibetan plateau, the Pacific Islands, and vast stretches of the Americas, including Mayan languages and Inuit. The pattern shows up across unrelated language families on every inhabited continent, which tells us it is a deeply natural way for human grammar to organize itself, even if it feels exotic to speakers of English, Spanish, or Mandarin.
How Common Are Ergative Languages
Despite ergativity’s wide geographic spread, nominative-accusative alignment is more common overall. A large-scale typological study found that nominative-accusative alignment is more prevalent than ergative-absolutive alignment across the world’s documented languages.1PLoS ONE. Language Structure Is Partly Determined by Social Structure That said, “less common” does not mean “rare.” Estimates vary depending on how you count, partly because many languages blend both systems, but ergative features appear in a substantial share of the world’s roughly 7,000 languages. The pattern is especially concentrated in certain regions: nearly all Australian Aboriginal languages have some degree of ergativity, and ergative case marking is the norm across the Tibeto-Burman family and in many Austronesian and Mayan languages.
The geographic clustering raises interesting questions about whether ergativity spreads through contact between neighboring communities or arises independently. Both appear to happen. In Australia, the sheer number of unrelated language families showing ergativity suggests the trait is either very old in the region or repeatedly develops on its own. In other areas, like the Caucasus, the picture is murkier, with some researchers arguing for deep shared ancestry and others for long-term contact effects.
Split Ergativity and Why Languages Mix Both Systems
One of the most revealing facts about ergativity is that very few languages are purely ergative in every corner of their grammar. Most ergative languages are split: they behave ergatively in some grammatical contexts and accusatively in others. These splits tend to follow predictable patterns.
The best-known split involves the kind of noun doing the action. A landmark analysis by Michael Silverstein identified what is sometimes called the animacy hierarchy: first-person and second-person pronouns (“I” and “you”) are more likely to take accusative-style marking, while nouns lower on the hierarchy, like names of inanimate objects, are more likely to take ergative marking.2Lingua. Split ergativity based on nominal type The intuition behind this is that “I” and “you” are prototypical agents, so it is the thing being acted upon that needs flagging. Inanimate nouns are prototypical patients, so when they do show up as agents, that role needs special marking. In a sense, ergative marking gravitates toward contexts where agency is surprising or noteworthy.
Other splits are driven by tense and aspect. Many ergative languages switch to accusative alignment in certain tenses. Hindi, for example, uses ergative marking only in the perfective aspect (roughly, completed actions). In the imperfective (ongoing or habitual actions), Hindi behaves like a straightforward nominative-accusative language. The result is that a Hindi speaker uses ergative case for “the boy ate the apple” but not for “the boy eats apples.” Splits like these make ergativity feel less like a single grammatical switch and more like a contextual strategy that languages deploy where it is most useful.
Morphological Versus Syntactic Ergativity
A distinction that matters for understanding how deep ergativity runs in a language is whether the ergative pattern is purely morphological or also syntactic. Morphological ergativity means the language puts different endings, particles, or markers on nouns depending on whether they are agents in transitive clauses. This is the visible, surface-level marker that is relatively easy to spot. Syntactic ergativity goes further: it means the language’s grammatical rules for things like relative clauses, question formation, and coordination also treat the absolutive argument (the intransitive subject and the transitive object) as the “default” for these operations, rather than the agent.
Some languages that mark ergativity on their nouns do also show ergative patterns in core syntactic operations.3Annual Review of Linguistics. Syntactic Ergativity: Analysis and Identification Dyirbal, an Aboriginal Australian language made famous by linguist R.M.W. Dixon, is a textbook case: its rules for combining clauses treat the absolutive as the pivot, not the agent. But many languages that are morphologically ergative remain syntactically accusative. Basque, for instance, marks ergativity clearly on nouns and verbs, yet much of its syntax privileges the agent in ways that look nominative-accusative. This mismatch is one reason linguists are cautious about grouping all “ergative languages” together as though they work identically.
The Antipassive and What It Tells Us
Ergative languages often have a grammatical construction called the antipassive, which is roughly the mirror image of the passive voice in English. In a passive sentence in English, you demote the agent (“the ball was kicked” instead of “she kicked the ball”). In an antipassive, you demote the patient: the object either disappears or gets pushed into an optional side role, and the verb takes special marking to signal that the transitive action is being presented with the agent in focus. The antipassive effectively converts a transitive clause into an intransitive one, which means the agent gets absolutive case instead of ergative.
Cross-linguistic research has found that antipassives marked by special verbal morphology are more widespread than previously thought, appearing even in languages that are not clearly ergative. Their functions vary widely too, and do not necessarily correlate with whether a language has morphological or syntactic ergativity.4Annual Review of Linguistics. Antipassives in Crosslinguistic Perspective In some languages, the antipassive serves a discourse function, allowing the speaker to shift emphasis from the action’s effect to the agent’s activity. In others, it is required in certain grammatical environments. The existence and range of antipassives underlines that ergativity is not just about sticking a case marker on a noun; it is part of a broader system of choices speakers make to package information about events.
When Ergative Marking Signals Meaning
In some languages, the ergative marker is not simply a grammatical requirement but carries a shade of meaning. Certain languages use ergative case in a fluid way, applying or withholding it based on whether the speaker wants to emphasize that the agent acted intentionally. Research on these fluid systems has proposed that there is a correspondence between volitional meaning and ergative case: when the agent is seen as deliberately bringing about an outcome, the ergative marker is more likely to appear.5Lingua. On fluid differential case marking: A bidirectional OT approach This is sometimes called “fluid-S” marking, because the intransitive subject can appear either with or without the ergative marker, and the choice reflects the speaker’s construal of the event.
Consider a language where “the child fell” can be said two ways: one with the ergative marker on “child” (implying the child jumped on purpose) and one without it (implying the fall was accidental). For speakers of these languages, the ergative marker does communicative work beyond just tagging grammatical roles. This phenomenon complicates any simple definition of ergativity as purely structural, because it shows that the same marker can serve both grammatical and pragmatic purposes, sometimes simultaneously.
How Children Learn Ergative Systems
A natural question is whether ergative case marking is harder for children to learn than the more globally common accusative pattern. Research on Basque-speaking children provides an interesting picture. Children acquiring Basque as a first language tend to master ergative marking in the verbal domain, the agreement suffixes on verbs that track who did what, relatively early, around two and a half years of age. But consistent use of the ergative case marker on nouns lags behind by several months.6Lingua. The (in)consistent ergative marking in early Basque: L1 vs. child L2 Children who acquire Basque as a second language (for instance, in immersion schooling) show the same asymmetry but with greater variability.
This gap between verbal agreement and nominal case marking suggests that children are not simply learning “ergativity” as a single package. They pick up the verb-internal cues first, possibly because verb agreement is more regular and frequent in the input they hear, and the overt case particle on nouns takes longer to consolidate. The finding challenges any notion that ergativity is inherently unnatural or harder to acquire; children manage it on a schedule that looks comparable to the acquisition of complex case systems elsewhere in the world, just with a predictable internal ordering.
How the Brain Processes Ergative Case
Neurolinguistic research has begun to examine whether the brain handles ergative and accusative case differently at a neural level. Studies using electroencephalography (EEG) in Hindi, a language with split ergativity, have found that violating ergative and nominative case expectations produces qualitatively different brain responses. When Hindi speakers encounter a sentence where an ergative-marked subject is followed by an intransitive verb, making the ergative case wrong, their brains produce a negative wave peaking around 400 milliseconds. When they encounter the reverse mismatch, a nominative-marked subject paired with a transitive verb that should have triggered ergative case, the brain responds with a later positive wave around 750 to 950 milliseconds.7PubMed Central. Processing ergativity in compound light verb constructions: electrophysiological evidence from Hindi
A related study confirmed this dissociation with a slightly different experimental setup, finding a positive brain response for ergative violations and a two-phase response for nominative violations.8PubMed Central. Structure Influences Case Processing: Electrophysiological Insights from Hindi Light Verb Constructions The details of which brain wave appears in which condition are still being debated, but the consistent finding across both studies is that the brain does not treat ergative and nominative case as mirror images of each other. It appears to recruit partly different cognitive processes for each. This is a relatively new area of research, and most work so far comes from Hindi, so how far these findings generalize to ergative languages with different structures, like Basque or Mayan languages, remains an open question.
Ergativity Under Pressure in Bilingual Communities
In communities where an ergative language is in contact with a dominant accusative language, ergative case marking sometimes shows signs of erosion. Basque provides a well-studied example. So-called “new speakers” of Basque, people who learned Basque through schooling or later in life rather than at home, tend to omit the ergative marker more often than native speakers. But research on these speakers reveals something interesting: even when their omission rates are higher, they still follow the same internal grammatical constraints that native speakers do. They are sensitive to verb type and to the sound environment around the case marker, and their deviations from native norms are structured and systematic, not random.9International Journal of Bilingualism. The role of frequency in the acquisition of structured variation: The case of Basque ergativity
These new speakers also sometimes extend ergative marking into contexts where native speakers would not use it, such as certain focused or topicalized intransitive subjects. And frequency plays a role: verbs and constructions that are more common in the input seem to be better learned, consistent with the idea that sheer exposure helps lock in a system as complex as split ergativity. The pattern suggests that ergative marking is resilient at a structural level even when its surface frequency drops under bilingual pressure. The grammar is being maintained; the surface rate of overt marking is what shifts.
Why Computers Struggle with Ergativity
Modern natural language processing systems, the kind that power translation tools and chatbots, are overwhelmingly trained on accusative languages like English. When researchers have tested how well these systems handle ergative agreement, the results are humbling. An experiment using Basque found that neural sequence models performed substantially worse on agreement prediction tasks in Basque than comparable tasks in English, despite the models’ general success with English grammar.10ACL Anthology. Can LSTM Learn to Capture Agreement? The Case of Basque Analysis suggested the models were relying on local shortcuts rather than learning the hierarchical structure of Basque sentences, where verb agreement tracks both the subject and the object simultaneously.
This matters practically because machine translation between ergative and accusative languages remains notably worse than translation between two accusative languages. If you have ever noticed that Google Translate handles Basque or Georgian oddly, the alignment mismatch is a significant part of the reason. The challenge is not just vocabulary but the fundamental mapping of who does what to whom. Researchers have proposed ergative agreement tasks as a benchmark for evaluating whether future models can truly learn the structural diversity of human language rather than pattern-matching their way through it. For now, ergativity remains one of the clearest demonstrations that not all grammars work on the same blueprint, and that tools built for one blueprint struggle when handed another.
Ergativity and What It Reveals About Language Design
One persistent misconception is that ergative languages are “primitive” or represent an earlier stage of language evolution. There is no evidence for this. Ergativity shows up in languages with elaborate literary traditions (Tibetan), in languages spoken by millions in modern urban settings (Hindi, Georgian), and in languages with rich morphological complexity (Basque). Historical linguists have documented cases of languages shifting from accusative to ergative alignment and vice versa, with neither direction being “progress.” Some Indo-Iranian languages, like Hindi, developed ergative features from older accusative constructions over centuries of natural change. Others have shed ergativity under contact pressure. The shifts are driven by the same forces that drive all language change: frequency, contact, analogy, and reanalysis of ambiguous structures.
What makes ergativity genuinely fascinating is that it provides a window into the range of options available to human cognition for packaging the relationship between an action and its participants. Accusative systems and ergative systems both accomplish the same communicative goal, distinguishing agents from patients, but they start from different baselines. Accusative systems treat the agent as the default role and flag patients. Ergative systems treat the patient as the default and flag agents. The fact that both solutions are stable across millennia, learnable by toddlers, and productive enough to handle anything from poetry to legal contracts tells us something about the flexibility of the human language faculty that you would miss if you only studied languages that work like English.