I've read somewhere that all known ergative languages are either verb-initial, or verb-final. I find this surprising, but I don't know of any counter-examples.

I've seen plenty of nominative languages, and it seems like they can have the verb anywhere. Well, I don't know about languages that have an OS (object, then subject order, as in OVS, OSV, and VOS), but I've never seen such languages so I don't know.

Does alignment somehow restrict what kinds of orders a language can use? I find it hard to believe that ergativity would force a language into a verb-peripheral order, considering that the nominative alignment clearly isn't as restricted (assuming it is at all).

I started thinking about after wondering about split-ergative languages, specifically the kind that treat noun phrases either nominatively or ergatively depending on where they fall on the agentivity hierarchy (Dyirbal is the only example I can seem to find, and it has SOV order as far as I can tell, but the only sentences I've been able to find have only pronouns).

edit: By 'agentivity hierarchy' I mean that some split-ergative languages treat some nouns nominatively and some ergatively. For example, Dyribal treats first and second pronouns nominatively, while everything else is treated ergatively. Another language I know of, called Ritharngu (which I can't find much information about) treats all pronouns nominatively, humans and 'higher animals' are tripartite, while everything is treated ergatively.

Page 120 of the Language Construction Kit 2 gives a heirarchy of agentivity, and claims that all languages with split alignment place the split somewhere on the heirarchy. The order is: verbal agreement, first person, second person, third person, proper nouns, humans, animals, and inanimate objects. Basically if two of these categories are given the same alignment, then everything between them will have the same alignment.


1 Answer 1


Kashmiri is split ergative and word order matches ergativity: imperfective/accusative is SOV, perfective/ergative is SVO. Igbo is claimed here to be abstract ergative and SVO, but also acknowledges the general claim.

  • That Kashmiri pattern is interesting; do you have a reference? I tried to look it up, but so far all I’ve found about Kashmiri word order is the Wikipedia article, which says that it has V2 order. Mar 7, 2018 at 1:00
  • Even if that were true, that's its split-ergativity is based off of the aspect of the verb, not which category the word goes into. I'll update the question to better specify what I mean.
    – user19661
    Mar 7, 2018 at 1:04
  • @sumelic Kashmiri word order/ergativity is described in that same paper linked to by user6726. Mar 7, 2018 at 11:33
  • 1
    @Wilson: Oh I see. It looks like the Kashmiri imperfective is actually S AUX O V, so it's only SOV in the way that the German past tense is SOV. According to Atamiri's answer to a related question, "German is neither SOV not SVO but V2 in the standard typology" (even though people who specialize in the study of German, rather than in rough typological comparisons, seem to agree that the SOV order should be considered the more underlying one). Mar 7, 2018 at 18:21
  • I was once an upper intermediate in German myself. I admit, it does kinda use SOV in more circumstances, but in the simplest constructions it uses SVO, so its hard to classify it as either. And only its 'formal' past can be said to be SOV, its simple past is SVO. And even in its OV orders, most of the time the order is actually S AUX O V. Only sub-clauses are truly SOV.
    – user19661
    Mar 8, 2018 at 0:17

Your Answer

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