I've recently started reading more about ergative languages, such as Basque. I understand that cases in ergative languages differ from nominal-accusative languages. For example, a sentence like "I go" would be "Me go", because the subject is the patient. Unfortunately, I fail to understand certain things I read on the wikipedia page on ergativity (https://en.wikipedia.org/wiki/Ergative%E2%80%93absolutive_alignment).
The following sentence is given as an example for a possible sentence in an ergative language (with English lexicon of course):

Him smiles. (He smiles)

I was wondering: Is the subject of this sentence really a patient? Or is the absolutive case just assigned to the subject of intransive verbs in general?
I was taught in a course that verbs with a patient subject are telic, but "to smile" is not really telic, in my opinion. From my understanding, not all subjects of intransitive verbs are patients. For example, in Italian a verb like "lavorare" (to work) is indeed intransitive, but the subject is not considered to be a patient but an agent.

Does the subject of ergative languages undergo V to T movement? I was taught that the subject in Italian or English moves to SpecT in order to receive the nominative case and share its phi-features with the verb. It seems to me that the very existence of ergative languages is enough to challenge this kind of idea. And can a passive be built out of a ditransitive verb then? I'm thinking about Burzio's generalization:

A verb form that does not express an agent argument in the specifier position cannot check objective case.

I hope this was somewhat understandable. Thanks for anyone's help.

  • 2
    Aren't ergative languages typically lacking the active vs. passive voce distinction? And what's wrong with “Him smiles”? In Russian, which is Nominative-Accusative, “It's funny for me” is Мне смешно (“Me (Dative case) funny”), so it doesn't even need for language to be ergative so as to have an oblique cased subject in such a sentence.
    – Yellow Sky
    Apr 26 at 19:21
  • @YellowSky I'm puzzled at the fact that, on the one hand, it is stated that patients in ergative languages are always expressed in the absolutive case (nothing wrong with that) but, on the other hand, it is assumed that ALL subjects of intransive verbs are patients. Which is not the case, from what I know. For example in a sentence like "He smiles", I wouldn't say the "he" is a patient, although the verb is indeed intransitive. That's where my confusion starts. Maybe someone can correct me.
    – LarenEmpty
    Apr 27 at 0:12
  • 1
    As for Russian, would Мне really be considered the subject in that sentence? I would be tempted to say that there is a covert pro element that is not spelled out but although present as subject. I think it also depends on what you consider to be the subject of a sentence. Is it the semantic element that is being predicated on? In that case Мне might be considered the subject. But if the subject is defined as the constituent agreeing with the verb (in this case covert "be" I think), then I doubt that it is. I hope this was somewhat understandable. Thank you either way!
    – LarenEmpty
    Apr 27 at 0:21
  • 1
    In fact, in the Russian language grammar, Мне смешно is considered to be an impersonal sentence with a zero subject constructed as “[It is] for.me funny”. Мне cannot be considered the subject since it's not in the nominative case, and logically it's rather a patient since the feeling of “funny” is experienced, you don't actually do anything to feel funny.
    – Yellow Sky
    Apr 27 at 11:19

3 Answers 3


In general, it is possible for ergative languages to have passive constructions. WALS allows you to search for combinations of properties, so if you do this: https://wals.info/combinations/107A_98A?iconsize=20&labels=0&v1-1=cF2F3F4&v1-2=c222222&v1-3=cF3C300&v1-4=t875692&v1-5=cF38400&v1-6=cA1CAF1&v2-1=cBE0032&v2-2=cC2B280&v2-3=c848482&v2-4=c008856&v2-6=c0067A5#map-container they show 7 languages in their sample which have it: West Greenlandic, Burushaski, Tukang Besi, Coos, Sanuma, Paumarí, Araona (the last 3 in southern America). Maybe this is not much; the other 25 ergative languages in that sample do not have passives.

Ideally, an ergative system assigns unmarked case (absolutive, i.e. non-ergative) to all intransitive subjects, so you need not wonder about thematic roles there (and English mock examples like "him smiles" are not very close, because the form "him" is not unmarked, but marked as accusative). There are a number of ergative languages, however, in which strongly agentive intransitive subjects get ergative case nevertheless (Georgian, Urdu a.o.). So it's not easy to make quick generalisations.

The same goes for the question of whether "the" subjects of "ergative languages" are moved into the finiteness domain. English does have such raising to nominative, already German does not do it (cf. Haider 2010: The Syntax of German CUP). There's certainly research on this issue with respect to ergative languages, but my hunch would be that they are not all alike either.


In Middle Persian you say things like man kard “of-me made” for “I made”, where kard is historically the perfect passive participle of the verb “to make”, as it is in Old Persian manā kŗtam “my (genitive) made (p.p.p. nominative singular neuter)”. In this sense not only do we have a passive form in an ergative sentence, but ergativity includes passivity by definition. The difference between Old and Middle Persian is that in the former the manā kŗtam construction coexists with the historic past tenses (imperfect, aorist, perfect), while in the latter man kard is the only way to say “I made” and is thus analysed as a ergative construction.


Punjabi and Sindhi both have ergative constructions and passive conjugations for verbs. It is not possible to use an ergative construction with a passive verb though.

The agent in ergative constructions in these languages is always the object however. There is a general preference for constructing sentences with inanimate subjects. For example, conceptually laughing does not involve agency in Punjabi. Since laughter occurs involuntarily, laughter is described as coming to people rather than an action done by people. Languages have their own internal reality, and English allows bending reality in some interesting ways. You can ask someone to sneeze on you in English even if it is not literally possible for people to sneeze on command. In Punjabi using verbs like this for involuntary actions is not possible.

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.