Case instead of voice

Museum for the best conlanging and conworldery threads. Ask mods to move threads here.
User avatar
Chuma
Avisaru
Avisaru
Posts: 387
Joined: Sat Oct 28, 2006 9:01 pm
Location: Hyperborea

Case instead of voice

Post by Chuma »

As you probably know, some languages use case to distinguish between voluntary and involuntary action:
he.ERG slide -> he slides (voluntarily, for example when skiing)
he.ABS slide -> he slips (involuntarily)

In my conlang, I've got a different system, and I'm wondering if something similar occurs in any natlang. It seems like a very natural system to me, but as I've studied more linguistics it's occurred to me that maybe no one is actually doing that.
Perhaps an example is the best explanation:
it.ABS melt -> it melts, i.e. goes from solid to liquid
it.ERG melt -> it melts something, i.e. causes something to melt

Similarly, you can leave out either argument of a transitive clause:
he.ERG hit -> he hits (something)
he.ABS hit -> he gets hit (by something)

You can basically treat any verb as being transitive; if there is no obvious agent, you can assume that the person causing the event is the agent. And then you can leave out any argument.
That is, that's one way of seeing it. You could also say that it's plain old ergative alignment, but the object can be left out.

My grammar professor said: "That sounds very logical - I don't know if there is any language that does that, but there ought to be."
What do you think?

User avatar
finlay
Sumerul
Sumerul
Posts: 3600
Joined: Mon Dec 22, 2003 12:35 pm
Location: Tokyo

Re: Case instead of voice

Post by finlay »

I don't honestly see why not. I should note that I haven't included a passive voice in Sentalian; it's nominative/accusative, but you can leave the nominative subject out, which is either pro-drop or the equivalent of a passive sentence. In some ways it also resembles an ergative system in that it's the nominative that's marked explicitly and not the accusative (apparently this is attested, and noted as a special case by WALS), but the subject of an intransitive verb is in the nominative.

To take a simple example from the fluency thread:

jatl ŋèbm
beer.ACC like
I have translated this as I have beer, because the language is pro-drop, but it also means "Beer is liked". To make it explicitly me that likes the beer, one has to say:
gamup jatl ŋèbm
1.NOM beer.ACC like

Also, your second set of examples is suspiciously like the middle voice in English – indeed, you've used the middle voice to translate it (IIRC what the middle voice properly is). It's not an explicit voice in English, but the "it" in your English examples cover different thematic roles.

User avatar
Imralu
Smeric
Smeric
Posts: 1640
Joined: Tue Mar 28, 2006 9:14 pm
Location: Berlin, Germany

Re: Case instead of voice

Post by Imralu »

Chuma wrote: You can basically treat any verb as being transitive; if there is no obvious agent, you can assume that the person causing the event is the agent. And then you can leave out any argument.
That is, that's one way of seeing it. You could also say that it's plain old ergative alignment, but the object can be left out.

My grammar professor said: "That sounds very logical - I don't know if there is any language that does that, but there ought to be."
What do you think?
It's perfectly logical. I don't see any reason to analyse this alignment as erg/abs rather than nom/acc. I've had the same system in conlangs before. Nominative or accusative is marked on the (pro)noun and whatever is unknown or irrelevant is omitted. Verbs have one alignment only. For example, the verb "break" would mean that the subject breaks the object, never that the subject breaks, but this is simply part of the semantics of the verb. Esperanto can be used this way - I've seen it done and done it myself and always been understood.

BALL.nom/erg BREAKS (the ball breaks something) = Pilko rompas (ion)
WINDOW.acc/abs BREAKS (the window breaks, the window is broken (by something)) = Fenestron rompas (io). (a more typical sentence would be "Fenestro rompiĝas" but the accusative suffix conveys the same meaning)
BALL.nom/erg BREAKS WINDOW.acc/abs (the ball breaks the window) = Pilko rompas fenestron.

You could of course simply do it with word order as well, rather than case, although in this case only SVO or OVS really work.

Possible sentences with (S)V(O) order:
BALL BREAKS = the ball breaks something
BREAKS WINDOW = the window breaks, the window is broken (by something)
BALL BREAKS WINDOW = the ball breaks the window, the window is broken by the ball
BREAKS = something breaks something, something is broken
Glossing Abbreviations: COMP = comparative, C = complementiser, ACS / ICS = accessible / inaccessible, GDV = gerundive, SPEC / NSPC = specific / non-specific
________
MY MUSIC

User avatar
Legion
Avisaru
Avisaru
Posts: 522
Joined: Sat Mar 05, 2005 9:56 pm

Re: Case instead of voice

Post by Legion »

That would be pretty much any language which has case but no syntactic pivot, and thus can drop any argument without transformation. South-east asian languages often have such features.

User avatar
Jetboy
Avisaru
Avisaru
Posts: 270
Joined: Sat Apr 17, 2010 6:49 pm

Re: Case instead of voice

Post by Jetboy »

What would you do for actual intransitive verbs, like sleep?
"A positive attitude may not solve all your problems, but it will annoy enough people to make it worth the effort."
–Herm Albright
Even better than a proto-conlang, it's the *kondn̥ǵʰwéh₂s

User avatar
alice
Avisaru
Avisaru
Posts: 707
Joined: Wed Oct 30, 2002 4:43 pm
Location: Three of them

Re: Case instead of voice

Post by alice »

I've done something very similar, if not identical, in some of my conlangs.
Zompist's Markov generator wrote:it was labelled" orange marmalade," but that is unutterably hideous.

User avatar
zelos
Sanci
Sanci
Posts: 58
Joined: Mon Oct 06, 2008 1:02 am

Re: Case instead of voice

Post by zelos »

I thought of this osmetime ago

User avatar
Chuma
Avisaru
Avisaru
Posts: 387
Joined: Sat Oct 28, 2006 9:01 pm
Location: Hyperborea

Re: Case instead of voice

Post by Chuma »

Finlay: Middle voice, I don't think I've heard it referred to that way. You mean this particular category of verbs, like "melt"? Wikipedia apparently calls them "ergative verbs". But it sounds odd to call a category of verbs a voice - or did I misunderstand you?

Imralu: I suppose that's true, it could be called nom/acc instead. I guess I figured the agent is more likely to be left out, so it made more sense this way.

Actually, maybe we can distinguish between nom/acc and erg/abs in this kind of language by looking at those verbs that are generally intransitive, and seeing what kind of extra role we give them to make them transitive. If we take the word "sleep" for example (to answer Jetboy's question), how would we make that transitive? We might say that a nom/acc language would generally add a patient - something like "he slept a good night's sleep", which English does, or maybe "he slept the bed", i.e. slept in/on. Whereas an erg/abs language would be more likely to add an agent - "she slept him", i.e. helped him to sleep. (That's what I would do in my conlang.)

But if you have nom/acc with unmarked acc, as Finlay says, then I suppose it's a pretty small difference between nom/acc and erg/abs.

Just to self-derail a little, Legion's answer makes me think of another thing. If I understood it right, pro-dropping can often imply either that the subject is known from context or that it's generic, right? Much like Finlay's example. Now, when I use pro-drop for things that are known from the previous sentence, the question is, which of the actors from the previous sentence am I referring to? Some examples I can think of are
1) always the thing that was abs
2) always the thing that was erg
3) the thing that has the same case as what I'm dropping
But another possibility would be to use word order for this. My conlang usually has erg-verb-abs word order, but verb first is also common, and the abs can be moved to the front when it is the topic. So you could also say that dropped pronouns refer to
4) the first noun phrase of the previous sentence
5) the first noun phrase of the last sentence where there was a noun phrase before the verb
Would this seem reasonable?

User avatar
finlay
Sumerul
Sumerul
Posts: 3600
Joined: Mon Dec 22, 2003 12:35 pm
Location: Tokyo

Re: Case instead of voice

Post by finlay »

http://en.wikipedia.org/wiki/Middle_voice#Middle
English doesn't really have middle voice because it's not inflected, but I have been taught a form of English grammar that involves it as a category, yes. Syntactically active but semantically passive, as the article says.

Vardelm
Avisaru
Avisaru
Posts: 329
Joined: Sat Sep 20, 2008 2:37 pm
Contact:

Re: Case instead of voice

Post by Vardelm »

Chuma wrote:Finlay: Middle voice, I don't think I've heard it referred to that way. You mean this particular category of verbs, like "melt"? Wikipedia apparently calls them "ergative verbs". But it sounds odd to call a category of verbs a voice - or did I misunderstand you?
I've had to struggle understanding middle voice since it has ended up as the default voice in my lang. FWIW, this thread has bunch of links about middle voice topics.

I'm thinking that "voice" could be considered as changing a verb into one of the other types, I.E. an accusative verb to an ergative verb, etc. I don't know if that's technically correct, but the whole notion of voice and verb semantics WRT participant roles are highly intertwined. Finlay is probably referring to the phrase or sentence as a whole, which makes sense. The verb being accusative/ergative/etc. just plays a part in that.


Chuma wrote:Actually, maybe we can distinguish between nom/acc and erg/abs in this kind of language by looking at those verbs that are generally intransitive, and seeing what kind of extra role we give them to make them transitive. If we take the word "sleep" for example (to answer Jetboy's question), how would we make that transitive? We might say that a nom/acc language would generally add a patient - something like "he slept a good night's sleep", which English does, or maybe "he slept the bed", i.e. slept in/on. Whereas an erg/abs language would be more likely to add an agent - "she slept him", i.e. helped him to sleep. (That's what I would do in my conlang.)
I'm guessing you likely know this, but keep in mind that not all verbs in a lang will be able to be made transitive.

I think what you'll get into here is a hierarchy of thematic relations. The person who sleeps is an experiencer. For adding transitivity, are you adding an agent, patient, theme, instrument, etc.? That will have some effect on the result, and it will also interact with whether the verb is accusative or ergative, and what voice is being used. The whole thing can get pretty complicated, and I'm still working on getting a full handle on all the variables.
Tibetan Dwarvish - My own ergative "dwarf-lang"

Quasi-Khuzdul - An expansion of J.R.R. Tolkien's Dwarvish language from The Lord of the Rings

User avatar
Chuma
Avisaru
Avisaru
Posts: 387
Joined: Sat Oct 28, 2006 9:01 pm
Location: Hyperborea

Re: Case instead of voice

Post by Chuma »

Yeah, I think I get it, with the middle voice, more or less.

True, not all verbs can be made transitive in any useful way. But with the idea of turning "to X" into "to make someone X", you could sort of make all verbs syntactically transitive, even if semantically it's not going to be very useful.
(Technically that wouldn't work for weather verbs, because they would only be made monovalent. But never mind that.)

I've thought of case in a "logical" way, basically letting each case in the conlang correspond to a semantic role. There are after all a few different views on just what semantic roles should be used, so I can happily make up my own variant that fits the language. One quite reasonable design principle is that no verb should have more than one thing in the same role, and if two roles never occur together they might be considered the same role. I put experiencer and recipient as one (which gets the dative case); that way "Bill showed Bob the ball" (where "show" is actually the same verb as "see") has the same role structure as "Bill gave Bob the ball".
This makes the whole idea of transitivity basically a semantic question rather than syntactic.
As for the person who sleeps, I wouldn't consider him an experiencer, but I suppose that works too.

User avatar
Legion
Avisaru
Avisaru
Posts: 522
Joined: Sat Mar 05, 2005 9:56 pm

Re: Case instead of voice

Post by Legion »

Well the thing is if you don't have a syntactic pivot and thus are able to drop any context obvious argument, co-reference becomes ambiguous:

"The man beats the dog and leaves."

In English, this means that the man leaves. In ergative language, this will often (but not necessarily) mean that the dog leaves.

But in a language like Chinese, this sentence would be ambiguous, you couldn't tell if that was the man or the dog which was leaving.

User avatar
Chuma
Avisaru
Avisaru
Posts: 387
Joined: Sat Oct 28, 2006 9:01 pm
Location: Hyperborea

Re: Case instead of voice

Post by Chuma »

I see. Ambiguity, can't have that, now can we? :P

What about languages where there is a topic marker? Is it reasonable to let the next sentence drop the word that has been topic marked?
It sounds okay to me - it's basically just reusing a topic, after all. And marking the topic with word order should also be okay, right?

User avatar
finlay
Sumerul
Sumerul
Posts: 3600
Joined: Mon Dec 22, 2003 12:35 pm
Location: Tokyo

Re: Case instead of voice

Post by finlay »

Vardelm wrote: I'm thinking that "voice" could be considered as changing a verb into one of the other types, I.E. an accusative verb to an ergative verb, etc. I don't know if that's technically correct, but the whole notion of voice and verb semantics WRT participant roles are highly intertwined. Finlay is probably referring to the phrase or sentence as a whole, which makes sense. The verb being accusative/ergative/etc. just plays a part in that.
Something like that; voice tends to be a valence-altering construction, where valency is how many arguments a verb can take (intransitive is 1, transitive is 2, ditransitive is 3, lojban verbs can have 6 or something). Causatives are also valence-altering because you are adding another argument to the verb: the causer.

Vardelm
Avisaru
Avisaru
Posts: 329
Joined: Sat Sep 20, 2008 2:37 pm
Contact:

Re: Case instead of voice

Post by Vardelm »

finlay wrote:Something like that; voice tends to be a valence-altering construction, where valency is how many arguments a verb can take (intransitive is 1, transitive is 2, ditransitive is 3, lojban verbs can have 6 or something). Causatives are also valence-altering because you are adding another argument to the verb: the causer.
Yeah, that's a better way of saying it. :D

If a given verb has a certain valence, doesn't it also place certain "expectations" on what role will appear in each "slot"? And wouldn't then voice be changing what role's will appear where as well as simply changing valence?
Tibetan Dwarvish - My own ergative "dwarf-lang"

Quasi-Khuzdul - An expansion of J.R.R. Tolkien's Dwarvish language from The Lord of the Rings

User avatar
finlay
Sumerul
Sumerul
Posts: 3600
Joined: Mon Dec 22, 2003 12:35 pm
Location: Tokyo

Re: Case instead of voice

Post by finlay »

Vardelm wrote:
finlay wrote:Something like that; voice tends to be a valence-altering construction, where valency is how many arguments a verb can take (intransitive is 1, transitive is 2, ditransitive is 3, lojban verbs can have 6 or something). Causatives are also valence-altering because you are adding another argument to the verb: the causer.
Yeah, that's a better way of saying it. :D

If a given verb has a certain valence, doesn't it also place certain "expectations" on what role will appear in each "slot"? And wouldn't then voice be changing what role's will appear where as well as simply changing valence?
yup

User avatar
James0289
Sanci
Sanci
Posts: 24
Joined: Fri Jul 06, 2007 6:32 pm
Location: The SE of the East Midlands

Re: Case instead of voice

Post by James0289 »

Legion wrote:Well the thing is if you don't have a syntactic pivot and thus are able to drop any context obvious argument, co-reference becomes ambiguous:

"The man beats the dog and leaves."

In English, this means that the man leaves. In ergative language, this will often (but not necessarily) mean that the dog leaves.

But in a language like Chinese, this sentence would be ambiguous, you couldn't tell if that was the man or the dog which was leaving.
I just discovered this on Wikipedia, and it reminded me of this: http://en.wikipedia.org/wiki/Syntactic_pivot

User avatar
Miekko
Avisaru
Avisaru
Posts: 364
Joined: Fri Jun 13, 2003 9:43 am
Location: the turing machine doesn't stop here any more
Contact:

Re: Case instead of voice

Post by Miekko »

Legion wrote:Well the thing is if you don't have a syntactic pivot and thus are able to drop any context obvious argument, co-reference becomes ambiguous:

"The man beats the dog and leaves."

In English, this means that the man leaves. In ergative language, this will often (but not necessarily) mean that the dog leaves.

But in a language like Chinese, this sentence would be ambiguous, you couldn't tell if that was the man or the dog which was leaving.
Actually, what surprises typologists and researchers wrt ergative languages is that in most of them, it's still the man that leaves. The number of ergative languages that have an absolutive pivot are vanishingly few, and most actually have an underlying nominative pivot.
< Cev> My people we use cars. I come from a very proud car culture-- every part of the car is used, nothing goes to waste. When my people first saw the car, generations ago, we called it šuŋka wakaŋ-- meaning "automated mobile".

User avatar
Chuma
Avisaru
Avisaru
Posts: 387
Joined: Sat Oct 28, 2006 9:01 pm
Location: Hyperborea

Re: Case instead of voice

Post by Chuma »

Yup. They're all traitors of the ergative cause!

My professor said that Chinese also has nominative pivot. But he's not really an expert at Chinese, I guess. And anyway, the point still stands.

I wonder how it would work in Japanese tho, with the topic and stuff.

And while we're at it (I've got so many questions today!): Another thing I've been thinking about is having different words for "it" depending on whether it refers to the previous agent or patient (or perhaps a couple of other roles). Example:
"The big dog chases the small dog. It barks."
Using different words for "it", we could distinguish between which dog it was.
Is this weird?

User avatar
Salmoneus
Sanno
Sanno
Posts: 3197
Joined: Thu Jan 15, 2004 5:00 pm
Location: One of the dark places of the world

Re: Case instead of voice

Post by Salmoneus »

You mean fourth-person, or obviate? It's pretty widespread, yes. Particularly in north america.
Blog: [url]http://vacuouswastrel.wordpress.com/[/url]

But the river tripped on her by and by, lapping
as though her heart was brook: Why, why, why! Weh, O weh
I'se so silly to be flowing but I no canna stay!

User avatar
Chuma
Avisaru
Avisaru
Posts: 387
Joined: Sat Oct 28, 2006 9:01 pm
Location: Hyperborea

Re: Case instead of voice

Post by Chuma »

Hm, that seems like a clever idea, from what I can gather. But I don't think that's quite what I meant.

If I understand it right, proximate/obviate third person distinguishes between "that thing we were talking about" and "that thing we just sort of mentioned briefly". (Which, by the way, seems rather confusing, considering that the "proximal demonstrative" is something completely different.) Is that about right?

My idea is that you would distinguish between "that thing we talked about in the ergative case" and "that thing we talked about in the absolutive case". (And I would probably add dative and some sort of oblique, just because I'm a bit of a maximalist. :) )

User avatar
Imralu
Smeric
Smeric
Posts: 1640
Joined: Tue Mar 28, 2006 9:14 pm
Location: Berlin, Germany

Re: Case instead of voice

Post by Imralu »

Miekko wrote:Actually, what surprises typologists and researchers wrt ergative languages is that in most of them, it's still the man that leaves. The number of ergative languages that have an absolutive pivot are vanishingly few, and most actually have an underlying nominative pivot.
Would that mean the second clause would also have the same inferred ergative? For example:

1 [Man-ERG] 2 [dog-ABS] 3 [beat] and 4 [leave]
1 [The man causes that] 2 [the dog] 3 [is beaten] and 4 [leaves]

You can see clearly that the dog is the one leaving, but also, in my English pseudo-equivalent you can see that the man being the cause is carried across as well. Would this be an expected tendency of absolutive-pivot languages, something that simply depends on how each language handles it (neither expected nor unexpected) or am I just being mislead by my English pseudo-equivalent??

EDIT: I just realised that saying "it-ABS" or "the dog-ABS" with "leaves" changes it for me so that I no longer see the man as a "carried-over" ergative.
Glossing Abbreviations: COMP = comparative, C = complementiser, ACS / ICS = accessible / inaccessible, GDV = gerundive, SPEC / NSPC = specific / non-specific
________
MY MUSIC

User avatar
*Ceresz
Lebom
Lebom
Posts: 140
Joined: Tue Nov 16, 2010 10:53 am
Location: Scania

Re: Case instead of voice

Post by *Ceresz »

Chuma wrote: In my conlang, I've got a different system, and I'm wondering if something similar occurs in any natlang. It seems like a very natural system to me, but as I've studied more linguistics it's occurred to me that maybe no one is actually doing that.
Perhaps an example is the best explanation:
it.ABS melt -> it melts, i.e. goes from solid to liquid
it.ERG melt -> it melts something, i.e. causes something to melt
Heh, even though it's not really the same this made me think of the intransitive vs. transitive form of Swedish 'smälta' (to melt).

Smöret smalt - The butter melted
Jag smälte smöret - I melted the butter

I don't make this distinction myself, but it's interesting.

Anyway, I like this idea, and I might use it in a future project.

User avatar
Miekko
Avisaru
Avisaru
Posts: 364
Joined: Fri Jun 13, 2003 9:43 am
Location: the turing machine doesn't stop here any more
Contact:

Re: Case instead of voice

Post by Miekko »

Imralu wrote:
Miekko wrote:Actually, what surprises typologists and researchers wrt ergative languages is that in most of them, it's still the man that leaves. The number of ergative languages that have an absolutive pivot are vanishingly few, and most actually have an underlying nominative pivot.
Would that mean the second clause would also have the same inferred ergative? For example:

1 [Man-ERG] 2 [dog-ABS] 3 [beat] and 4 [leave]
1 [The man causes that] 2 [the dog] 3 [is beaten] and 4 [leaves]

You can see clearly that the dog is the one leaving, but also, in my English pseudo-equivalent you can see that the man being the cause is carried across as well. Would this be an expected tendency of absolutive-pivot languages, something that simply depends on how each language handles it (neither expected nor unexpected) or am I just being mislead by my English pseudo-equivalent??

EDIT: I just realised that saying "it-ABS" or "the dog-ABS" with "leaves" changes it for me so that I no longer see the man as a "carried-over" ergative.
what do you mean by "inferred ergative"? sounds very unclear. Really, in most ergative languages that permit similar gas in coordintion, it's the man that leaves, even though he isn't overtly absolutive anywhere. Since your question is unclearly stated, I cannot really answer whether you're mislead by your English pseudo-equivalent or not.
< Cev> My people we use cars. I come from a very proud car culture-- every part of the car is used, nothing goes to waste. When my people first saw the car, generations ago, we called it šuŋka wakaŋ-- meaning "automated mobile".

User avatar
Imralu
Smeric
Smeric
Posts: 1640
Joined: Tue Mar 28, 2006 9:14 pm
Location: Berlin, Germany

Re: Case instead of voice

Post by Imralu »

Miekko wrote:what do you mean by "inferred ergative"? sounds very unclear. Really, in most ergative languages that permit similar gas in coordintion, it's the man that leaves, even though he isn't overtly absolutive anywhere. Since your question is unclearly stated, I cannot really answer whether you're mislead by your English pseudo-equivalent or not.
Yeah, I was more or less musing through my fingertips. Sorry about that. I'll try for a little bit more clarity this time. In a language in which "dog" could be inferred as filling the absolutive role with "leave" (ie. the dog leaves), do you think "man" could also be inferred simultaneously as filling the ergative role in the same clause (ie. the man makes the dog leave / kicks the dog out)?

For example, with Nom Acc. (brackets stand for non-explicitly stated, "understood" elements)
Man-NOM dog-ACC beats and (man-NOM) leaves.

Three possible examples, as far as I see it, with Erg-Abs:

(1) The one I believe you're saying is most common:
Man-ERG dog-ABS beats and (man-ABS) leaves.

(2) The other possibility that has been mentioned.
Man-ERG dog-ABS beats and (dog-ABS) leaves.

(3) The possibility I'm asking about. Is this attested. Does it seem likely?
Man-ERG dog-ABS beats and (man-ERG) (dog-ABS) kicks-out
Glossing Abbreviations: COMP = comparative, C = complementiser, ACS / ICS = accessible / inaccessible, GDV = gerundive, SPEC / NSPC = specific / non-specific
________
MY MUSIC

Post Reply