zompist bboard

Just wondering if anyone could explain them to me. As I've heard that ERG-ABS langs tend to lack passive voice, and that confuses me. Thanks in advance!

What confuses you?

In the passive voice in nominative-accusative languages, you have:
Instransitive: nominative subject
Transitive: nominative agent, accusative patient
Passivized (intransitivized) transitive: nominative patient, oblique agent (if allowed at all)

Now, for a purely ergative language, you have:
Intransitive: absolutive subject
Transitive: ergative agent, absolutive patient
An passive doesn't really have anything to "promote," as intransitives take absolutive marking, which the patient is already taking. Instead, there is often an antipassive that deleted the direct object and "promotes" the agent to absolutive, and the patient can be reintroduced with an oblique.
Antipassivized transitive: absolutive agent, oblique patient
It might also help to understand that ergatives often come from oblique markers themselves*, which would be:
Intransitive: nominative subject
Transitive: oblique agent, nominative patient
Passivized transitive: nominative patient, oblique agent
(And antipassive: nominative agent, oblique patient)
There's nothing to do, as the transitive already has the marking that the passive would have.

However, languages generally aren't completely ergative. It'll be things like the nouns take erg-abs case marking, but the verbs agree in a nom-acc system, or the verbs agree in erg-abs but the nouns are unmarked. Ergative syntax, where for example word order is absolutive-first in both transitives and intransitives rather than the much more common erg-abs-verb, or maybe only the absolutive is available for relativization (rather than the grammatical subject/nominative), is particularly rare. Such mixed-ergativity languages often have room for passives, and sometimes have both passives (that promote transitive object to intransitive subject) and antipassives (that demote the transitive object to intransitive oblique).

*The path of grammaticalization has to do with passives, actually: the structure of nominative patient + passive verb + oblique agent is reanalyzed to absolutive patient + transitive verb + ergative agent.

EDIT: Though I should add, there's plenty of nom-acc languages that don't have passives either. They either use other methods such as topicalizing the object, allowing the agent to simply be omitted, or using "empty" pronouns or words or 3rd person conjugations to leave the agent unspecified.

Salmoneus wrote:What confuses you?

Just that I was wondering how ERG-ABS langs could lack passives.

edit ninja'd.

Think about the alignment; think about the voice.

Alignment. Speaking in general terms and using nominate accusative for comparison --

Nominative-accusative treats the Agent (A) of a transitive clause similarly to a Subject (S) in an intransitive clause. The patient or object (P) is treated differently from A & S.

On the other hand, Ergative-absolutive treats the P of a transitive clause similarly to a S in an intransitive clause. The A is treated differently from P & S.

Voice. Comparing passive and anti-passive --

Passive voice is marked on the verb and derives an intransitive clause from a transitive one. Relative to the default voice, the passive voice removes or demotes A to being an adjunct (for example, appears in a non-core case or is unavailable for poly-personal verb marking). The former P is then treated like an S (or also like an A in nom-acc).

Similarly, anti-passive voice is marked on the verb and also derive an intransitive clause from a transitive one. Relative to the default voice, the anti-passive voice removes or demotes P to an being an adjunct. The former A is then treated like an S (or also like an P in erg-abs).

Synthesis In essence, both derived voices take a more marked argument and promote it to become a less marked argument. If it remains at all, the demoted argument is treated as an adjunct.

In nom-acc, P is the most marked core argument. Since a passive voice removes P, it will be apparent (useful?) when it removes the most marked core argument in a nom-acc language.

Similarly in erg-abs, A is the most marked core argument. Only an anti-passive voice will be apparent when it removes the most marked argument.

If you want to dig deeper, these voices can also free up arguments for constructions where they could not normally appear, syntactic pivots, etc. For example, if a language does not allow an inanimate in a transitive verb’s A, it might be able to appear as a derived S in a passive construction.

Moving beyond what both voices do, why does there appear to be a connection between anti-passive and erg-abs? Looking above, you could argue that useful voices appear in useful places. On the other hand, some nom-acc languages have anti-passives and visa-versa. What would that do? Shift information structure. Consider a nom-acc language with an anti-passive voice that dropped P or shifted it to an oblique – the key being that it is a voice (meaning the voice shift is marked on the verb) rather than a functionally similar ambi-transitive from transitive to intransitive (no shift in non-agreement verb marking).

Taking a different angle, WALS chapter 108 section 3 argues in part that the “connection” of erg-abs and anti-passive is merely skewed data caused by the ease of spotting changes in the marking of the remaining A. Presumably, this logical also applies to the marking of P and why passives seem to “appear” more often with nom-acc languages. Is that a great argument? Meh.

Anyway -- TL;DR: WALS chapter 108 and 98.

What is WALS (Sorry if it's a noobish question, I'm kinda a noob.)

SolarisZBB wrote:What is WALS (Sorry if it's a noobish question, I'm kinda a noob.)

World Atlas of Language Structures

SolarisZBB wrote:

Salmoneus wrote:What confuses you?

Just that I was wondering how ERG-ABS langs could lack passives.

Yes; but what I was getting at was that "how is this possible?" is never a useful question; it cannot be answered, by itself, usefully. It's basically asking us to tell you absolutely everything that we know in case part of what we know might be what you want.

Instead ask yourself "why wouldn't this be possible?" - and when you answer that, with "it wouldn't be possible because X is the case", then ask us "why isn't X the case?"

How can they lack passives? They lack passives by not having passives (although many ergative languages do actually have passives). What is the problem that you perceive here exactly?

You do not "need" passives in ergative languages, because of ambitransitivity. I.e. if with an ergative (i.e. ambitransitive) verb, one can specify it with both an agent and a patient, or just with a patient, by simply specifying or omitting an ergative argument. However this does not mean that one cannot have a passive in an ergative language, e.g. if one forbids ambitransitivity, then a passive would be necessary to carry out the valency reduction to eliminate the agent argument.

Travis B. wrote:You do not "need" passives in ergative languages, because of ambitransitivity. I.e. if with an ergative (i.e. ambitransitive) verb, one can specify it with both an agent and a patient, or just with a patient, by simply specifying or omitting an ergative argument. However this does not mean that one cannot have a passive in an ergative language, e.g. if one forbids ambitransitivity, then a passive would be necessary to carry out the valency reduction to eliminate the agent argument.

Not all "ergative" languages exhibit amibtransitivity. In some Australian languages it is quite normal for common nouns to be inflected on an ERG/ABS basis, but verbs have strict valency rules, with "subject" and "object" slots. The arguments are required, but can appear as S/O clitics or affixes rather than full NPs. In fact, NPs are dropped quite freely and word order is completely free. So you don't rely on ambitransitive verbs, but you can easily focus on the patient if you want.

Speaking cross-linguistically, many if not most languages that exhibit ergative case marking on NPs are not syntactically ergative; they only have ergative morphology.

The difference is in pivots in sentences like:

I saw Carly and _ screamed.

A morphologically ergative language will "pivot" on the subject of I saw Carly (so that I am doing the screaming), in the same way that an accusative language would. It is only in the case of syntatically ergative language that the omitted argument in screamed would be taken to be Carly. To make it me, you would detransitive the first verb - this is the antipassive.

So Haleza Grise wrote:

Travis B. wrote:You do not "need" passives in ergative languages, because of ambitransitivity. I.e. if with an ergative (i.e. ambitransitive) verb, one can specify it with both an agent and a patient, or just with a patient, by simply specifying or omitting an ergative argument. However this does not mean that one cannot have a passive in an ergative language, e.g. if one forbids ambitransitivity, then a passive would be necessary to carry out the valency reduction to eliminate the agent argument.

Not all "ergative" languages exhibit amibtransitivity. In some Australian languages it is quite normal for common nouns to be inflected on an ERG/ABS basis, but verbs have strict valency rules, with "subject" and "object" slots. The arguments are required, but can appear as S/O clitics or affixes rather than full NPs. In fact, NPs are dropped quite freely and word order is completely free. So you don't rely on ambitransitive verbs, but you can easily focus on the patient if you want.

Speaking cross-linguistically, many if not most languages that exhibit ergative case marking on NPs are not syntactically ergative; they only have ergative morphology.

The difference is in pivots in sentences like:

I saw Carly and _ screamed.

A morphologically ergative language will "pivot" on the subject of I saw Carly (so that I am doing the screaming), in the same way that an accusative language would. It is only in the case of syntatically ergative language that the omitted argument in screamed would be taken to be Carly. To make it me, you would detransitive the first verb - this is the antipassive.

I don't know if this is the best example sentence for these kind of pivot effects, since I think in most languages first-person pronominal subjects like this are obligatorily marked on the verb anyway, so that would disambiguate. At least, that's the case for most languages in WALS, although it's not a representative sample of the world's languages.

Sumelic wrote:

So Haleza Grise wrote:

Travis B. wrote:You do not "need" passives in ergative languages, because of ambitransitivity. I.e. if with an ergative (i.e. ambitransitive) verb, one can specify it with both an agent and a patient, or just with a patient, by simply specifying or omitting an ergative argument. However this does not mean that one cannot have a passive in an ergative language, e.g. if one forbids ambitransitivity, then a passive would be necessary to carry out the valency reduction to eliminate the agent argument.

Not all "ergative" languages exhibit amibtransitivity. In some Australian languages it is quite normal for common nouns to be inflected on an ERG/ABS basis, but verbs have strict valency rules, with "subject" and "object" slots. The arguments are required, but can appear as S/O clitics or affixes rather than full NPs. In fact, NPs are dropped quite freely and word order is completely free. So you don't rely on ambitransitive verbs, but you can easily focus on the patient if you want.

Speaking cross-linguistically, many if not most languages that exhibit ergative case marking on NPs are not syntactically ergative; they only have ergative morphology.

The difference is in pivots in sentences like:

I saw Carly and _ screamed.

A morphologically ergative language will "pivot" on the subject of I saw Carly (so that I am doing the screaming), in the same way that an accusative language would. It is only in the case of syntatically ergative language that the omitted argument in screamed would be taken to be Carly. To make it me, you would detransitive the first verb - this is the antipassive.

I don't know if this is the best example sentence for these kind of pivot effects, since I think in most languages first-person pronominal subjects like this are obligatorily marked on the verb anyway, so that would disambiguate. At least, that's the case for most languages in WALS, although it's not a representative sample of the world's languages.

I don't know if it's true about "most" languages marking the 1st singular on the verb, but you're right, I should have for the sake of simplicity chosen something other than a pronoun.