eldin raigmore wrote: ↑05 Oct 2020 02:19
What’s the difference between the Austronesian/Philippine Morphosyntactic Alignment on the one hand, and Split-Ergativity on the other hand?
Unfortunately, both concepts are poorly defined. "Split-Ergative", in particular, is fairly useless as a concept. And all these concepts blur into one another (eg is Direct-Inverse a form of split ergativity or not?). There are similarities between Austronesian and Split-Ergative alignments, sure, but also many differences.
Imagine a split-Ergative language in which the basis of the split is purely pragmatic*;
If the Agent is the Focus, then the transitive clause is Nominative/Accusative;
But if the Patient is the Focus, then the transitive clause is Absolutive Ergative.
*(so the split is not conditioned by the animacy nor definiteness of the participants, nor by the TAM of the verb.)
Is that essentially the same as Austronesian/Philippine Alignment?
OK, so, when we say 'nominative/accusative', we really mean any or all of three things:
a) there is either a nominative or an accusative case. A nominative case is used for the subject of an intransitive, and the agent of a transitive; an accusative is only used for the patient of a transitive. Usually it is the accusative that is marked. (a nominal sense)
b) the verb agrees with the subject of an intransitive or the agent of a transitive. (a verbal sense)
c) the subject of an intransitive or the agent of a transitive within one clause have a special privilege when it comes to co-reference and suchlike between clauses. (an interclausal sense)
And conversely with ergative/absolutive. "Split ergative" just means that a, b and c are sometimes but not always true. But there are many different ways this can be the case.
Usually when we talk about split ergativity, we are talking about ergativity in the nominal sense (having an ergative case), because this is much more common in languages than the verbal sense, which is in turn more common than the interclausal sense. (oh, and there's also a word order sense too).
------------------------
Austronesian systems are not split-ergative in this usual, nominal sense, because they do not necessarily have an ergative case. Consider:
dog frog ate-AV
"the dog ate the frog"
dog frog ate-PV
"the frog ate the dog"
In this Austronesian-style example, there is no ergative case marking. Nor accusative case marking. So we're not dealing with nominal ergativity.
[it is of course possible to have nominal ergative marking (NE) AND austronesian alignment (AA), but one doesn't require the other]
-------------------------------
Are Austronesian systems split-ergative in a verbal sense? Well, if we add number agreement, we can see:
dog frogs ate-AV-3s
dog frogs ate-PV-3s
In this example, the verb does indeed agree with the agent in the first sentence, but the patient in the second, so this is, in the sense defined above, verbally split-ergative.
However, consider a similar language with a different pattern of agreement:
dog frogs ate-AV-3s
dog frogs ate-PV-3p
Here, both sentences have the verb agree in number and person with the agent - so this is not ergative at all. Is this still AA? It's a bit more debateable, because the "voices" are then not entirely symmetrical, but fundamentally, yes, it seems to make sense to still call this AA.
What actually happens in most AA, however, ignoring the case of pronominal arguments (which do in many languages become verb affixes in some voices) is that there is just no agreement on the verb at all - the languages that have developed full agreement tend to have lost AA marking. However, this doesn't appear logically necessary.
Given verbs with no agreement marking, and the theoretical possibility of verbs with nominative-absolutive agreement yet still having AA marking, I think we have to conclude that, no, AA is not split ergativity even in a verbal sense.
-----
Is AA split-ergativity in a word order sense? I don't know. Some languages indicate the "subject" or "focus" purely through word order, so could be called split-ergative in word order. Others use both word order and explicit marking. I don't know if any use marking alone, but I suspect so, and can't see any reason why it wouldn't be possible. So again, I don't think AA is split-ergative in principle, in terms of word order.
-----
What about pivots, switch reference and so forth? I'm afraid I don't know.
--
So the general answer is: some caveats, but broadly, no, AA is not split ergative in any usual sense.
Indeed, AA should be seen as an entirely different sort of thing. Although we talk of AA as an 'alignment', that may just be a coincidence: Earth's AA languages happen to generally lack extensive core case marking or verb agreement. This isn't, however, conceptually necessary. You could have a language with E/A case marking and N/A verb agreement, and STILL have AA markings as well (this would obviously be a bit redundant, but redundancy does happen!). This is probably partly why many people prefer to see AA as having nothing inherently to do with semantic role marking, and instead regard it as being about information structure... although real AA languages do indeed use it for semantic purposes as well.
......
Is Split-Transitive Alignment the same as Austronesian/Philippine Alignment?
I'm not familiar with 'split transitive alignment'. [although I have used the term in the past to describe my own conlang, Rawàng Ata]
.....
What’s the difference between Circumstantial Voice and Applicative Voice?
In the AA context, the big difference is that an applicative is a voice, whereas a circumstantial is a "voice" in the AA sense; it's not clear that it's a voice in the non-AA sense, because it's not clear if AA is best thought of as a voice system.
That said, there are two concrete differences.
First, circumstantials indicate that the primary argument, focus, subject, topic, trigger, whatever you call it (let's call it 'locus' to sidestep this issue) is semantically olique.
By contrast an applicative voice promotes a semantic oblique to the syntactic role of direct object. Notably, this occurs in languages where the direct object is NOT the primary argument.
To put that another way: the circumstantial makes an oblique be, in some way, equivalent to the agent of many transitive verbs (and, of course, also the equivalent of the patients of many transitive verbs...); the applicative does not, it only makes the oblique equivalent to the patient of a transitive verb.
Second, applicative voice is a true voice - it is prototypically valency-altering. It turns a univalent verb into a bivalent verb. It is true that the outcome when applied to a bivalent verb is less clearcut - in some languages, it becomes a trivalent verb, while in others it remains bivalent. However, its prototypical use is valency-increasing.
A circumstantial, however, like other 'focuses' in AA, is symmetrical: it does not increase or decrease the valency of the verb.
------------
.....
Must Applicative Voice always promote an oblique (I.e. non-core) argument to the Object ... slot or position ?
That is indeed what 'applicative voice' means, yes. Of course, some languages may have quirks and exceptions.
Is dative movement a kind of applicative?
That's a question of definitions, I suppose. Dative shift (if that's the movement you're thinking of) does increase valency and promote an oblique to object position. However, it does not remove an existing object from that position, which applicatives generally do, and it may not fully move the applicative into the exact position of a normal direct object (as the existing direct object may hold onto it). That said, we've already pointed out that things get hazy with bivalent verbs. I think it makes sense in general to see dative shift as a sort of applicative, yes, and one way of dealing with the question of what to do when you applicativise a bivalent verb. However, it's possible that a language might have dative shift AND a different sort of applicative, so it might make sense to distinguish these in certain languages.
Can the applicative ever promote a non-argument to Object?
I don't understand what you mean by 'non-argument' - what is less an argument than an oblique? Surely an oblique covers anything that could conceivably be promoted? Can you give an example?
----------------
Must Circumstantial Voice always promote an oblique argument to the Subject position or slot?
"Circumstantial voice" in an AA sense marks a semantic oblique as the locus. Whether you call this "subject position" is a matter of (much disputed) definition. I can certainly imagine a conlang that is AA-like enough to have a circumstantial, yet in some way emphatically distinguishes locus from subject.
Is Passive a kind of Circumstantial Voice?
No, they have nothing in common.
Circumstantial marks an oblique as the locus.
Passive promotes the patient to subjecthood and decreases the valency.
Can the Circumstantial voice ever make a non-argument the Subject?
Again, what did you have in mind?
Is Causativization a kind of Circumstantial Voice?
No.
Circumstantial marks an oblique as the locus. Valency remains the same.
Causatives mark an oblique as the subject. Valency is increased.
I would also note that there is probably some language out there somewhere that has distinct circumstantial (eg instrumental/locative/referential) and causative (/benefactive) triggers.