First, a caveat: it was brought to my attention in my last post (‘Weird things that languages do #1: Ergativity‘) that the phenomenon in question wasn’t exactly ‘weird’ for people who speak ergative languages. This is true! What I mean by ‘weird’, of course, is more that the phenomena I describe are surprising for people who are only familiar with English (or perhaps Indo-European languages). This is quite true of today’s ‘weird thing’: the Antipassive construction.
It is a verb voice that in some languages allows you to delete the patient. Essentially, the Antipassive is the opposite of the passive, which allows you to delete the agent. To understand this better, we’d better remind ourselves of what the passive is, and have a think about what agents and patients are.*
(1) Jane kicked the ball.
(2) The ball was kicked (by Jane).
In (1), Jane is the subject, the ball is the object. In (2), the ball is the subject, and Jane is a non-essential oblique (i.e. you can miss ‘by Jane’ out, hence the parentheses). This is easy enough to understand: the object ‘the ball’ in (1) has been promoted to the subject of (2). Importantly, the semantic roles have not changed. In both (1) and (2), Jane is the agent (i.e. she is deliberately performing the action). The ball is the theme (i.e. it undergoes the action but does not change its state). So what is happening is that the theme in (1), ‘the ball’ is an object, but in (2) it has been promoted to the subject. The agent is no longer necessary in (2), i.e. it is deletable.
If English had a less impoverished case system, you’d find nominative marking on the subjects, and accusative marking on the object. The oblique might be marked by some other case (here, ‘X’):
(1′) Jane.NOM kicked the-ball.ACC.
(2′) The-ball.NOM was kicked (by-Jane.X)
This is because subjects are (usually) always marked nominative in languages like English. (Try replacing the nouns in (1) and (2) with pronouns, which do still show case marking in English.)
There has also been some change to the verb: ‘kicked’ in (1) has become ‘was kicked’ in (2). Some languages (e.g. Latin) show that a verb is passivized by verb morphology, rather than adding a new word (in English: an auxiliary). The word dicit in Latin means ‘he/she/it says’. Dicitur means ‘it is said’. In this case, the suffix -tur signals 3rd person present passive, whereas the suffix -t in dicit signals 3rd person present active.
So, let’s get to the Antipassive construction: the sentence in (1) in an ergative language might look like this:
(3) Kicked the-ball.ABS Jane.ERG
‘Jane kicked the ball’
So, Jane, marked ergative, is the agent, and the ball, marked absolutive, is the patient. You will recall that this is the normal state of affairs in transitive sentences in ergative languages. In the corresponding Antipassive construction, it is the agent that is marked absolutive, and the patient is marked otherwise (sometimes ergative, sometimes in some other way, represented below by ‘X’), and the verb carries antipassive morphology:
(3) kicked.ANTIPASS Jane.ABS (the-ball.X)
In (3), the agent (Jane) has been promted from being marked ergative to being marked absolutive (just like the theme ‘the ball’ in the passive construction was promoted from being marked accusative to being marked nominative) and the theme (‘the ball’) can optionally be deleted (just like the agent, Jane, could be deleted in the passive construction). So it is like the passive but the opposite!
(3) is hard for us to translate into English, as we don’t have an Antipassive construction. Since the semantic roles are the same, the (truth-conditional) meaning is probably the same as the normal transitive sentence ‘Jane kicked the ball.’ Wacky, eh?
* (In order to understand this construction, one has to understand the concept of ergativity (see my previous post.)