I am reading R. Dixon's work on ergativity. He employs three basic syntactic relations:
- S for a single argument of a verb
- A for one argument of a verb
- O for the other argument of a verb
In an ergative language like Basque a transitive verb can be identified by checking whether there is Ergative in a sentence. The following examples are extracted from a machine translator:
Johnek hil zaitu (John killed you)
John erori egin da (John fell)
John Moskura joan zen (John went to Moscow)
However, when it comes to accusative languages, how can one know if a given verb is transitive or not? One possible evidence comes from passivisation but it is now always reliable (cf. to have in English, знать, целовать in Russian, etc.)
I looked at Swahili which has polypersonal agreement: object agreement does not operate exclusively on what can be traditionally called direct object (wa- and tu- are object markers for you and we respectively):
Yohana akamwambia (John said it to you)
Yohana alituambia (John said it to us)
but Yohana alikwenda kwetu (John went to us)
One may define O to mean that an entity is affected by an action so that the addressee and the killed are both taken to be O. But then the verb to laugh can also be said to be transitive as one of its agruments gets affected by someone doing the laughing. What I mean is that we can find verbs one of whose arguments can be interpreted as O but is marked by an adposition or different case (e.g., помогать кому-то (to help someone-DAT) in Russian) unlike other insances of the realisation of O.
My question is: for an accusative language, are there any criteria that allow to distinguish between intransitive and transitive verbs?
Thank you!
