Nel Fie wrote: ↑04 Aug 2022 15:03
Thank you both for the in-depth answers. I'm not sure I understand every detail that's been brought up, but I think I get the picture. If I get this correctly, there's a big jumble going on between terms like "Austronesian/Phillipine voice/focus/alignment", "symmetrical voice", "trigger system", etc... that get used synonymically sometimes and sometimes not, and it causes lots of confusion and misunderstandings?
Yes, I believe so.
Specifically, I think there's three things there:
Philippine alignment - a system with more than two symmetrical voices defining the role of the primary argument of a clause. Because symmetrical voices aren't like normal voices, and the 'primary arguments' (my phrase) aren't entirely like conventional subjects (for a start, they're more commonly patients than agents), sometimes people avoid using 'voice' to describe these systems altogether, prefering 'focus' or 'trigger' or a number of other things.
Austronesian alignment - sometimes treated as synonymous with Philippine alignment, even though most Austronesian languages do not have Philippine alignment. Others, if they use the phrase, consider Austronesian alignment a broader case of Philippine alignment, and include Indonesian languages, which traditionally had only two symmetrical voices, rather than the three or four of Philippine alignment. Apparently modern Indonesian itself may no longer have an Indonesian alignment, although in the past it did (the inherited symmetrical passive is the di-passive, the null-passive being I think a later development, as is the use of a preposition in the di-passive). However, proto-Austronesian itself presumably had a Philippine alignment, with the simplification being a later areal development.
Symmetrical voice - one of the defining characteristics of Austronesian/Philippine alignment.
But, as you say, I'm not sure everyone uses these terms in the same way!
So, if I understand you correctly, one needs to distinguish "morphological symmetry" and "syntactic symmetry", with the first being understood in the sense that all the voices are equally marked; whereas the second means that the valency of the verb remains unchanged?
I wouldn't say 'need to', but you can, yes. [I've no idea to what extent these two things go together in practice]
That's a very useful and interesting description.
I'm slightly confused by one aspect of it though: if they can be considered to be nominal constructions, how could they have arguments or be considered to be like univalent verbs at the same time? Or do you mean that they behave like univalent verbs that have been nominalized?
It's common for nominalised verbs to still have an argument, but much more common to have one than to have two. In English, for example, it's very common to have nominalised verbs with an argument: car crash, javelin-throwing, truth-telling, water leak, headlock, asset forfeiture, etc. English does have a gerund form with two arguments - "me eating the ham is not a big deal!" - but they're used comparatively rarely, and in speech they're often now treated as having only one argument, due to influence from the verbal noun: "my eating the ham is not a big deal!" [where the subject arrgument of the gerund, 'me', has been extracted and converted into a marked possessor of the gerund, 'my'].
The alignment of the arguments of nominalised verbs is (not always but) usually ergative-absolutive: the noun can be an agent of an intransitive verb ("water leak"), but where the verb is semantically transitive the retained argument is usually the object (the assets are forfeited, the head is locked, etc). [in English this isn't absolute - there are exceptions, like 'police forfeiture'. In some languages this is stricter; in others weaker; but I believe the general tendency toward erg-abs alignment is if not universal then very widespread].
This is one of the main ways erg-abs alignments in active verbal clauses can develop, and why ergative cases are often identical to genitive cases: the nominalised construction is re-analysed as verbal, the unmarked argument retains erg-abs alignment, and any further ergative is supplied through a possessive, just as in "my eating the ham", "the man's stamp auction". But the univalent tendency of nominalised verbs remains, with that possessive argument being less 'core' than the unmarked erg-abs argument.
Note also however that although the alignment of nominalised verbs is usually erg-abs, it's often possible not only to have an agent argument be the single argument (police forfeiture) but also to have the single argument be an oblique argument (usually a locative or benefactive) that wouldn't normally be allowed as a core argument: a 'car boot sale' is not a sale of, or even by, car boots, but OUT OF car boots. A charity auction is an auction not of charity, but FOR charity.
One way to understand where Philippine alignment comes from, therefore, is to see the 'trigger' (subject, focus, topic, whatever you want to call it) as the argument of a nominalised verb (by default erg-abs, but with the possibility for transitive agents and even benefactive/locatives to be used instead), where the language has reduced ambiguity by marking the role of the argument on the verb itself. The entire construction has then been analysed as a verbal construction, but retaining oddities that develop from the nominalised origin, including:
- the 'subject' is by default the patient (that is, 'passive' is the default voice for transitives)
- the 'voices' don't change the valency
- there are no applicatives or dative shifts and in general the 'object' slot has no particular importance
- 'voices' can promote benefactives and locatives to 'subject'
Synchronically, Philippine clauses probably are not nominalised (although it's harder to be sure than you might think; in general, Austronesian languages have relatively little distinction between nouns and verbs). Diachronically, they probably arose from nominalised constructions, and this may explain some of their oddities. Technically, they probably are not univalent (although, again, that's probably something people have argued), but their nominal origin gives them some properties of that, and it can sometimes be helpful to think of them that way.
That's my (non-native, non-linguist) understanding, at least.