What does it mean for a language or verb to be one or the other of these typologies (examples would help)? Can it be more than one at once?

  • 2
    I think you may be mixing your terminology up here. (I don't blame you; it's confusing terminology.) There are "accusative languages" and "ergative languages," but there's no such thing as an "unaccusative language" or an "unergative language." "Unaccusative" and "unergative" are categories of verb. Commented Oct 22, 2011 at 23:26
  • Thank you for the clarification. Could someone explain the difference between unaccusative and unergative verbs then? And whether, say, an unergative verb can occur in a nominative/accusative language.
    – user325
    Commented Oct 23, 2011 at 0:36
  • 1
    Unergative verbs are verbs whose subjects are agents. In other words, they describe voluntary activities (like "run" or "talk"). Unaccusative verbs are ones with non-agent subjects; so they describe involuntary human activities (like "die" or "hear") or events affecting inanimate objects (like "melt" or "explode"). As far as I know, every language has both kinds of verb; so, yeah, nominative/accusative languages have unergative verbs, and ergative/absolutive languages have unaccusative verbs. Commented Oct 23, 2011 at 0:56

2 Answers 2


Linguistics is all about patterns.

An ergative pattern is one that treats subjects of transitive verbs differently than subjects of intransitives and objects of transitives, the latter of which pattern together. In the example below, assuming subjects come first, the pattern is ergative because the subject of the transitive verb has a different form than the other two argument positions. In this situation, people tend to call the subject of transitive verb the "ergative" argument.



While the relevant pattern is usually case-marking or agreement, any pattern that treats subjects of transitive verbs differently than objects of transitive and subjects of intransitives could be called an ergative pattern. Langauges are called ergative as they exhibit more and more ergative patterns, though case/agreement are the canonical patterns people care about here.

In contrast, accusative patterns treat objects of transitive verbs differently than subjects of transitive verbs and subjects of intransitive verbs, like below.



What is unergative and unaccusative then? Well, mostly it's bad terminology. Notice that when we talk about ergative and accusative patterns, we're talking about a contrast between transitive and intransitive verbs. When we talk about unaccusative and unergative patterns, we're talking about patterns that treat subjects of one class of intransitive predicates differently than the subjects of another class of intransitives predicates. The pattern is like below:



For example, some languages use special auxiliary verbs with the subjects of one class of intransitives and a different set of auxiliary verbs with another. Usually when people dig deeper they find that the difference hinges on whether the subject of the relevant intransitive is canonically more "agentive" or more like a "patient". This is a very complex question in itself, but the important point is that subjects of transitive verbs tend to be more agentive, so folks have decided to call patterns that treat "agentive" intransitive subjects special "unergative" because ergative patterns treat subjects of transitive verbs special. A similar line of reasoning gives the source of the term unaccusative.

Getting to your question, it is possible for languages to exhibit mixes of these patterns. Things get hairy here, but basically, we reserve special terminology for when a language's case-marking/agreement system exhibits multiple patterns. For example, if a language sometimes has an ergative case marking system and sometimes an accusative case marking system, then it is called split-ergative. A fourth type of case system is found in languages that ignore the accusative/ergative opposition and mark the case of intransitives along a unergative/unaccusative line (this is a gross simplification). This pattern is often called a Split-S. Back to your question, though, it very often the case that languages exhibit different mixes of these patterns in different domains, in which case, there is usually not a name for it and people just say things like: Phenomenon X targets unergatives/unaccusatives, in which case they mean: Phenomenon X treats different intransitive verbs differently, in particular, it treats intransitive subjects that seem more like transitive subjects/objects in a special way.

  • I think your first line needs fixing. The line '... an ergative pattern is a pattern that treats subjects of transitive verbs differently than objects of transitive verbs' should read '... an ergative pattern is a pattern that treats subjects of transitive verbs differently than subjects of intransitive verbs'. Nominative-accusative systems also treat tr subj differently from tr obj. A key point of ergativity is that tr obj and intr subj have the same marking, which is different to that of tr subj. Commented Oct 28, 2011 at 12:41
  • 1
    Thanks for the catch. It was just a typo, but what you say is not exactly right. It isn't just that ergative systems treat subjects of transitive verbs differently than subjects of intransitives, but it is also necessary that the objects of transitives and the subjects of intransitives pattern together. In the characterization you give, a language that treat all three argument positions differently would be called ergative, but this isn't right. These are usually called tripartite systems.
    – rmh
    Commented Oct 28, 2011 at 22:44
  • Isn't that what I said in my comment about 'a key point of ergativity'? Commented Oct 28, 2011 at 23:06
  • 1
    Excellent theoretical explanation...would you mind giving some example languages for ergative and non-ergative? How do non-inflected languages fit in this scheme if at all?
    – Mitch
    Commented Oct 29, 2011 at 1:38

Intransitive sentences can be either unaccusative or unergative based on the the subject's thematic role. If the subject is the agent, the resulting construction is unergative. (e.g. Jane sneezed or John danced).

If the subject is not an agent, the sentence is unaccusative. (e.g. Jane arrived or The snow melted)

Ergative and accusative refer to cases (noun inflections). In languages that have it, accusative marks the objects of transitive verbs. Ergative case marks the subject of transitive verbs.

Languages are often divided into Nominative-Accusative and Ergative-Absolutive, but this is an over-simplification. The Wikipedia article on the subject is pretty good.

I'm sure there is some arcane connection between ergative/accusative and unergative/unaccusative, but in practice they aren't related.

  • Did you mean to use the opposite examples? I.e. the subject of sneeze seems like less of an agent than the subject of arrive.
    – user325
    Commented Oct 23, 2011 at 9:01
  • 1
    It seems the best test for unaccusativity in English is the resultative adjunct test--Jane melted into water, exploded to bits, arrived dissheveled. Vs. (*) Jane sneezed apart. Commented Oct 23, 2011 at 14:30
  • 1
    @Knitter The examples are right but maybe agency isn't the clearest criterion. Semantics was never my strong suit, and the distinction between thematic roles (Agent, patient, theme, etc) has always eluded me. The actual definition is that the subject of unergative verbs behave like external arguments whereas for unaccusative verbs, it behaves as internal objects.
    – Dan Milway
    Commented Oct 23, 2011 at 15:27
  • 1
    @DanielBriggs I'm not familiar with the resultative adjunct test, although it seems like it would work. I use the cognate object test: Jane sneezed (a great sneeze) vs Jane arrived (*a great arrival).
    – Dan Milway
    Commented Oct 23, 2011 at 15:27
  • 3
    Actually, this is a matter of discussion in the semantics literature. One way to think of the relevant distinction isn't between AGENT and themes -- that is, volitional entities performing an action -- but rather, ORIGINATORS and themes (originator being a term from van Voorst 1988), that is to the say, the originator of the event. Then, you can see why "John sneezed" would be unergative, as "The diamond glowed", which also seems to be unergative more often than not. Commented Oct 25, 2011 at 15:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.