Interaction between ergativity, pro-drop, and causatives?

A forum for discussing linguistics or just languages in general.
Post Reply
Ithisa
cuneiform
cuneiform
Posts: 174
Joined: 14 May 2013 13:14

Interaction between ergativity, pro-drop, and causatives?

Post by Ithisa »

This question is really about whether a feature in a conlang I'm making is realistic or not, so sorry if it belongs more in the conlangs board. But it is also about how real-life languages with ergative-absolutive alignment work, and it seems like the conlangs board is just about showing off conlangs rather than asking questions.

In Kuriyet, an ergative-absolutive, very pro-drop conlang I'm making, I've noticed something curious happening in sentences with pro-drop:
Kuma pada monda.
ku-ma pada monda
1-ERG fish.ABS eat
"I eat a fish"

Pada monda.
fish.ABS eat
"[somebody] eats a fish" = "a fish is eaten"
So essentially, removing the ergative argument makes the sentence sound "passive". Now, this is pretty boring, but I then thought, what if I analyze <monda> as an intransitive verb meaning "to be eaten"? Then we have
Kuma pada monda.
ku-ma pada monda
1-ERG fish.ABS is.eaten.CAUS (zero-marked)
"I cause a fish to be eaten."

Pada monda.
fish.ABS is.eaten
"A fish is eaten"
So essentially, it seems like the whole language can be reanalyzed into not having any "real" transitive verbs, and all sentences with an ERG argument simply being causative. So we can also say things like
Laulai molos.
laula-i molos
flower-PLUR.ABS be.many
"Flowers are many".

Ossema laulai molos.
spring-ERG flower-PLUR.ABS be.many
"Spring causes flowers to be many".
Is this sort of reanalysis totally off the rails? It does allow an interesting way of zero-marking causatives of intransitive verbs like "to be many" without introducing any ambiguity. For "higher order" causatives, would the following be crazy?
Pada monda.
fish.ABS eat
"The fish is eaten."

Amma pada monda.
3-ERG fish eat
"He eats the fish" = "He causes the fish to be eaten"

Mokyattem amma pada monda.
tiredness-ERG 3-ERG fish eat.
"Tiredness made him eat the fish." "Tiredness causes he causes the fish to be eaten"
Is this "real" ergative-absolutive alignment anymore, or did I just turn the ergative case into simply an oblique agent marker? How do real-life erg/abs languages that have pro-drop handle stuff like causatives and passives?
Fluent: :chn: :eng:
Intermediate: :jpn:
User avatar
Creyeditor
MVP
MVP
Posts: 5091
Joined: 14 Aug 2012 19:32

Re: Interaction between ergativity, pro-drop, and causatives

Post by Creyeditor »

In the early linguistic descriptions of some ergative languages they were described as having 'passive' sentences as the default and the ergative as reintroducing the agent as an oblique argument. Your argumentation looks similar, though I do not really understand the stuff about causatives.
Creyeditor
"Thoughts are free."
Produce, Analyze, Manipulate
1 :deu: 2 :eng: 3 :idn: 4 :fra: 4 :esp:
:con: Ook & Omlűt & Nautli languages & Sperenjas
[<3] Papuan languages, Morphophonology, Lexical Semantics [<3]
Ithisa
cuneiform
cuneiform
Posts: 174
Joined: 14 May 2013 13:14

Re: Interaction between ergativity, pro-drop, and causatives

Post by Ithisa »

My point with causatives is that "to be eaten" / "to eat" can be reanalyzed as "is eaten" / "cause to be eaten"; so removing an ergative from a transitive sentence to express the passive can be reanalyzed as adding an ergative to *any* intransitive sentence to express the causative.
Fluent: :chn: :eng:
Intermediate: :jpn:
User avatar
Creyeditor
MVP
MVP
Posts: 5091
Joined: 14 Aug 2012 19:32

Re: Interaction between ergativity, pro-drop, and causatives

Post by Creyeditor »

Adding a causer/agent to a passive verb is usually just called 'reintroducing the agent' and the result is sometimes 'agentive passive'.
Agent Patient Verb <- Active voice
Agent Patient Verb-Passive <-Passive voice
Patient Verb-Passive Agent-OBL <-Agentive Passive construction
If you used a causative, you would expect some marking on ther verb. A lot of languages do not allow causatives to apply to passive verbs, but some do, usually this effects the meaning in a subtle way.
Agent Patient Verb <- Active voice
Agent Patient Verb-Passive <-Passive voice
Causer Agent Patient Verb-Passive-Causative <-Causative voice of passive voice
Creyeditor
"Thoughts are free."
Produce, Analyze, Manipulate
1 :deu: 2 :eng: 3 :idn: 4 :fra: 4 :esp:
:con: Ook & Omlűt & Nautli languages & Sperenjas
[<3] Papuan languages, Morphophonology, Lexical Semantics [<3]
Ithisa
cuneiform
cuneiform
Posts: 174
Joined: 14 May 2013 13:14

Re: Interaction between ergativity, pro-drop, and causatives

Post by Ithisa »

Creyeditor wrote:Adding a causer/agent to a passive verb is usually just called 'reintroducing the agent' and the result is sometimes 'agentive passive'.
Agent Patient Verb <- Active voice
Agent Patient Verb-Passive <-Passive voice
Patient Verb-Passive Agent-OBL <-Agentive Passive construction
If you used a causative, you would expect some marking on ther verb. A lot of languages do not allow causatives to apply to passive verbs, but some do, usually this effects the meaning in a subtle way.
Agent Patient Verb <- Active voice
Agent Patient Verb-Passive <-Passive voice
Causer Agent Patient Verb-Passive-Causative <-Causative voice of passive voice
I'm still not sure you quite understand my question, though it could be that I didn't really formulate it well.

I of course know how "normal" passives and causatives work; my question was rather about my conlang's proposed mechanism, where any intransitive verb can be zero-marked causative by simply adding an agent marked in the ergative. Essentially, all verbs in my language would work like "open" in English: "the door opens" / "I open the door" (or rather "door.ABS open" "I-ERG door.ABS open"). Is such a mechanism remotely realistic?
Fluent: :chn: :eng:
Intermediate: :jpn:
User avatar
Adarain
greek
greek
Posts: 511
Joined: 03 Jul 2015 15:36
Location: Switzerland, usually

Re: Interaction between ergativity, pro-drop, and causatives

Post by Adarain »

Generally speaking ergative and passive sentences look more or less the same, the only exception being that a passive would be clearly a derived structure while an ergative sentence is an underlying structure (hence passives being marked with some form of construction).

In most ergative languages, the ergative NP (A) can be much more readily omitted; in accusative languages the same is true of the accusative NP (O). A passive turns the A into a peripheral case, which by definition can be omitted.

Causatives have an universality about them: in every language, the causer will be the A NP (nominative/ergative depending on the language). Causative constructions are also often morphologically identical to more general transitivizers. What you now propose is basically that all verbs that can be intransitive are S=O ambitransitive and that for some verbs this can have a causative interpretation, correct? This seems fine to me. However, if you're saying that these ambitransitives only have a causative interpretation the question arises how you distinguish between "I eat fish" and "I force-feed fish". Your causative construction seems to put the causee into an oblique case (keeping the S/O argument unchanged in the absolutive, a common way of handling it) so those sentences can become quite ambiguous.

Source for any claims is Ergativity (Dixon).
At kveldi skal dag lęyfa,
Konu es bręnnd es,
Mæki es ręyndr es,
Męy es gefin es,
Ís es yfir kømr,
Ǫl es drukkit es.
Ithisa
cuneiform
cuneiform
Posts: 174
Joined: 14 May 2013 13:14

Re: Interaction between ergativity, pro-drop, and causatives

Post by Ithisa »

Adarain wrote:Generally speaking ergative and passive sentences look more or less the same, the only exception being that a passive would be clearly a derived structure while an ergative sentence is an underlying structure (hence passives being marked with some form of construction).

In most ergative languages, the ergative NP (A) can be much more readily omitted; in accusative languages the same is true of the accusative NP (O). A passive turns the A into a peripheral case, which by definition can be omitted.

Causatives have an universality about them: in every language, the causer will be the A NP (nominative/ergative depending on the language). Causative constructions are also often morphologically identical to more general transitivizers. What you now propose is basically that all verbs that can be intransitive are S=O ambitransitive and that for some verbs this can have a causative interpretation, correct? This seems fine to me. However, if you're saying that these ambitransitives only have a causative interpretation the question arises how you distinguish between "I eat fish" and "I force-feed fish". Your causative construction seems to put the causee into an oblique case (keeping the S/O argument unchanged in the absolutive, a common way of handling it) so those sentences can become quite ambiguous.

Source for any claims is Ergativity (Dixon).
"I force-feed fish" would be "1-ERG fish-ERG eat", with a pro-dropped thing that's getting eaten.
Fluent: :chn: :eng:
Intermediate: :jpn:
User avatar
Frislander
mayan
mayan
Posts: 2088
Joined: 14 May 2016 18:47
Location: The North

Re: Interaction between ergativity, pro-drop, and causatives

Post by Frislander »

Ithisa wrote: Pada monda.
fish.ABS eat
"The fish is eaten."

Amma pada monda.
3-ERG fish eat
"He eats the fish" = "He causes the fish to be eaten"

Mokyattem amma pada monda.
tiredness-ERG 3-ERG fish eat.
"Tiredness made him eat the fish." "Tiredness causes he causes the fish to be eaten"
I think the main problem with this is evidenced by the second sentence here; the two translations given are not equivalent. In the second version, "He causes the fish to be eaten", the causer need not be the same as the person actually eating the fish. In that respect you didn't pick the best example verb for this, since the verb "eat" has an agent which is affected by the action in some way, and verbs of consumption are treated differently than normal transitive verbs in some languages. "Break" would have been a better example, since "cause to break" is semantically equivalent to "break", though note that this is partly due to the ambitransitivity of "break" and does not work for similar verbs such as "smash" and "cut".

Which kind of brings me onto my next point: it's a very strange reanalysis to make. The causative is semantically a marked construction relative to the plain transitive: why would the speakers of this language reanalyse their basic transitive sentences in a semantically more complicated manner?

There's also the issue of relative animacy. A good principle to follow is that people care more about animate things (people) than inanimate things (items of food). From that principle stems things like the fact that humans are significantly more likely to be marked for number than non-humans, as well as direct-inverse systems and other animacy hierarchies (for instance in Navajo the "more animate" participant always comes first in the sentence). Thus for that reason the omission of the agent is a semantically (if not formally) marked construction, since the thing being eaten is more important to the speaker than the eater, or that the eater is indefinite/unknown. The reanalysis of the sentence here would entail a shift of relative importance in the speaker's mind from the more animate participant (the eater) to the more inanimate participant (the food), which would violate that principle regarding animacy.

It is probably for these reason that I have never seen a language which works like this, and why I don't think that it is like that one exists. I'll have a look though, just in case.
Sumelic
greek
greek
Posts: 566
Joined: 18 Jun 2013 23:01

Re: Interaction between ergativity, pro-drop, and causatives

Post by Sumelic »

Yes, I was going to say what Frislander says. A morphologically & semantically productive causative that just means something like "contribute to causing X to happen to Y" doesn't generally express the same semantic meaning as a transitive. There are many semantic roles the subject of a transitive verb can play aside from "causative agent".

Many languages with ergative verb structures also have causatives: e.g. Basque; the usage is not identical, although apparently there is some overlap.
Ithisa
cuneiform
cuneiform
Posts: 174
Joined: 14 May 2013 13:14

Re: Interaction between ergativity, pro-drop, and causatives

Post by Ithisa »

Frislander wrote:the causer need not be the same as the person actually eating the fish.
Indeed, it seems like I missed that fact. It does seem that the causative is a lot more general than "agent".
Fluent: :chn: :eng:
Intermediate: :jpn:
User avatar
Adarain
greek
greek
Posts: 511
Joined: 03 Jul 2015 15:36
Location: Switzerland, usually

Re: Interaction between ergativity, pro-drop, and causatives

Post by Adarain »

Ithisa wrote: "I force-feed fish" would be "1-ERG fish-ERG eat", with a pro-dropped thing that's getting eaten.
That bit is problematic, usually in ergative langs the absolutive NP is mandatory, or at least more mandatory than the ergative one. I would expect that to be able to express “I force-feed fish”, you’d have to add an antipassive on top of the causative, i.e.
fish-ABS 1ERG eat “I eat fish”
[causee]-ABS 1ERG fish-ERG eat.CAUS “I make [causee] eat fish”
1ABS fish-ERG eat.CAUS.ATP “I force-feed fish”

Under the assumption that the use of marking the underlying direct object with the ergative in causative sentences is a secondary use of the ergative (and thus unaffected by further valence changes like the antipassive); and also S/O-A-X-V word order.
At kveldi skal dag lęyfa,
Konu es bręnnd es,
Mæki es ręyndr es,
Męy es gefin es,
Ís es yfir kømr,
Ǫl es drukkit es.
Ithisa
cuneiform
cuneiform
Posts: 174
Joined: 14 May 2013 13:14

Re: Interaction between ergativity, pro-drop, and causatives

Post by Ithisa »

Adarain wrote:
Ithisa wrote: "I force-feed fish" would be "1-ERG fish-ERG eat", with a pro-dropped thing that's getting eaten.
That bit is problematic, usually in ergative langs the absolutive NP is mandatory, or at least more mandatory than the ergative one. I would expect that to be able to express “I force-feed fish”, you’d have to add an antipassive on top of the causative, i.e.
fish-ABS 1ERG eat “I eat fish”
[causee]-ABS 1ERG fish-ERG eat.CAUS “I make [causee] eat fish”
1ABS fish-ERG eat.CAUS.ATP “I force-feed fish”

Under the assumption that the use of marking the underlying direct object with the ergative in causative sentences is a secondary use of the ergative (and thus unaffected by further valence changes like the antipassive); and also S/O-A-X-V word order.
My language is highly pro-drop, so basically any argument to the verb can be freely dropped, like in Japanese. Does that somehow clash with ergativeness? For example, "I ate [something]" would be
Kuma mundan.
1-ERG eat-PST
Fluent: :chn: :eng:
Intermediate: :jpn:
User avatar
Adarain
greek
greek
Posts: 511
Joined: 03 Jul 2015 15:36
Location: Switzerland, usually

Re: Interaction between ergativity, pro-drop, and causatives

Post by Adarain »

It doesn’t necessarily clash with it per se, but in many ergative languages, the absolutive argument cannot be dropped, or if it can, still rarely is - while the ergative is very often omitted. Often a sentence will be antipassivized so that the former absolutive argument becomes ergative, to then drop it, if one wants to omit it. This would make your “I eat it” into

Code: Select all

1-ABS eat.ATP-PST
At kveldi skal dag lęyfa,
Konu es bręnnd es,
Mæki es ręyndr es,
Męy es gefin es,
Ís es yfir kømr,
Ǫl es drukkit es.
Sumelic
greek
greek
Posts: 566
Joined: 18 Jun 2013 23:01

Re: Interaction between ergativity, pro-drop, and causatives

Post by Sumelic »

Adarain wrote:Often a sentence will be antipassivized so that the former absolutive argument becomes ergative, to then drop it, if one wants to omit it. This would make your “I eat it” into

Code: Select all

1-ABS eat.ATP-PST
I don't think the antipassive normally turns the former absolutive argument to ergative, does it? My understanding is that it turns the former ergative argument to absolutive, and just deletes the former absolutive argument (it may be possible to use an adjunct instead with similar meaning, but I don't think this type of adjunct would normally take the same marking as the ergative case--maybe this occurs in some languages though).
User avatar
Frislander
mayan
mayan
Posts: 2088
Joined: 14 May 2016 18:47
Location: The North

Re: Interaction between ergativity, pro-drop, and causatives

Post by Frislander »

Sumelic wrote:
Adarain wrote:Often a sentence will be antipassivized so that the former absolutive argument becomes ergative, to then drop it, if one wants to omit it. This would make your “I eat it” into

Code: Select all

1-ABS eat.ATP-PST
I don't think the antipassive normally turns the former absolutive argument to ergative, does it? My understanding is that it turns the former ergative argument to absolutive, and just deletes the former absolutive argument (it may be possible to use an adjunct instead with similar meaning, but I don't think this type of adjunct would normally take the same marking as the ergative case--maybe this occurs in some languages though).
If it turned the former absolutive to ergative, then that would be an inverse voice not an anti-passive. The demoted absolutive is always either dropped or goes into some other local case like the locative or instrumental.
Sumelic
greek
greek
Posts: 566
Joined: 18 Jun 2013 23:01

Re: Interaction between ergativity, pro-drop, and causatives

Post by Sumelic »

Frislander wrote:
Sumelic wrote:
Adarain wrote:Often a sentence will be antipassivized so that the former absolutive argument becomes ergative, to then drop it, if one wants to omit it. This would make your “I eat it” into

Code: Select all

1-ABS eat.ATP-PST
I don't think the antipassive normally turns the former absolutive argument to ergative, does it? My understanding is that it turns the former ergative argument to absolutive, and just deletes the former absolutive argument (it may be possible to use an adjunct instead with similar meaning, but I don't think this type of adjunct would normally take the same marking as the ergative case--maybe this occurs in some languages though).
If it turned the former absolutive to ergative, then that would be an inverse voice not an anti-passive. The demoted absolutive is always either dropped or goes into some other local case like the locative or instrumental.
Syntactically, the former absolutive must be deleted or demoted to an adjunct rather than an argument in an antipassive construction, but I wondered if Adarain was thinking of some language with case syncretism between the ergative and some other case used to indicate the object of an antipassive (in which case the morphological form would be "ergative" in a sense). That doesn't seem totally improbable to me, since it seems like the ergative is somewhat often syncretic with cases like the instrumental or genitive.
User avatar
Adarain
greek
greek
Posts: 511
Joined: 03 Jul 2015 15:36
Location: Switzerland, usually

Re: Interaction between ergativity, pro-drop, and causatives

Post by Adarain »

Sumelic wrote:
Adarain wrote:Often a sentence will be antipassivized so that the former absolutive argument becomes ergative, to then drop it, if one wants to omit it. This would make your “I eat it” into

Code: Select all

1-ABS eat.ATP-PST
I don't think the antipassive normally turns the former absolutive argument to ergative, does it? My understanding is that it turns the former ergative argument to absolutive, and just deletes the former absolutive argument (it may be possible to use an adjunct instead with similar meaning, but I don't think this type of adjunct would normally take the same marking as the ergative case--maybe this occurs in some languages though).
Antipassive turns an underlying sentence of

Code: Select all

NP1    NP2    V
A(erg) O(abs) V(transitive)
to a derived structure of

Code: Select all

NP1    (NP2) V
S(abs) X     V(intransitive)
Where X is a non-core case, for example a dative or some adpositional phrase; specifics vary from language to language.

I did write a wrong thing there. Replace “ergative” with “peripheral case” and it all makes sense. Sorry for the confusion.
At kveldi skal dag lęyfa,
Konu es bręnnd es,
Mæki es ręyndr es,
Męy es gefin es,
Ís es yfir kømr,
Ǫl es drukkit es.
User avatar
Frislander
mayan
mayan
Posts: 2088
Joined: 14 May 2016 18:47
Location: The North

Re: Interaction between ergativity, pro-drop, and causatives

Post by Frislander »

Well looking around Kabardian does sort-of have this, because the ergative is really just a more general oblique case which is used for indirect objects as well as transitive agents.
User avatar
Adarain
greek
greek
Posts: 511
Joined: 03 Jul 2015 15:36
Location: Switzerland, usually

Re: Interaction between ergativity, pro-drop, and causatives

Post by Adarain »

Ye, ergative (and accusative) are commonly used for peripheral functions as well; I would not be surprised to find some languages that demote the O argument to a peripheral NP marked with the ergative in antipassives (which would be more or less an inverse).
At kveldi skal dag lęyfa,
Konu es bręnnd es,
Mæki es ręyndr es,
Męy es gefin es,
Ís es yfir kømr,
Ǫl es drukkit es.
Post Reply