Are there other voices apart from Act./Pass./Antipassive?

[ebilein]

I was just wondering wheter there are other voices apart from the probably already by you all well known active, passive and antipassive. As far as I know, these are used to reduce the number of obligatory arguments of a verb, thus from the sentence "He's reading the book" we get the passive "The book is being read" and in Ergative languages (although I would say by no means I'm an expert in those) we get from the sentence "He eats a sausage" the antipassive "He eats".

My question in particular is, are there languages which also used a voice to get rid of the Agens but retain both Indirect and Direct Object? The only good example I can think of would be the sentence "He gives her a book", which, of course, in "normal" passive becomes "A book is given to her".

Are there languages which would have some kind of Indirect-Object-Passive, thus a voice that could produce from the example above the sentence "She receives a book"?

[Xephyr]

Questions like these always confuse me.. questions along the lines of "How many cases are there?" or "What other voices are there?"... Do you mean how many are attested in natlangs, or how many there are possible? The answer to the latter is extremely high, and the answer to the former isn't necessarily that easy to answer either, because the "dative case" or "passive voice" in one language may not be the same in another. For instance, K'iche' Maya has 2 passive voices.

Anyway, voices are usually valency-changing operations: operations that change transitive verbs into intransitive or ditransitive verbs, or operations that rearrange nouns but keep the same valency. Passive, for example, is a valency-reducing operation because it takes a transitive sentence and turns it into an intransitive sentence with the former object now acting as the subject. So, depending on how a language handles verbs and oblique arguments, there could be a very high number of conceivable voices.

The two big voices, that you didn't mention, are applicative and causative. Applicative is a valency-increasing operation: it takes an intransitive sentence with an oblique argument and turns it into a transitive sentence, with the oblique argument now acting as an object. There are a couple English verbs that kinda-sorta do kinda-sorta-this: He sees over the garden -> He oversees the garden. "The garden" was formerly an oblique, adjunct to the preposition "over", but in the second sentence it's an object. There could conceivably be an applicative voice for every preposition in a language: I sing because of Ruth -> *I becausesing Ruth.

Causatives vary wildly across languages, so I'm not sure whether to call them valency-increasing or not. A prototypical causative turns a transitive sentence into a ditransitive sentence: I killed the wabbit. -> Þór made me kill the wabbit. English is a bit wonky here, but the valency shift is easier to see in an hypothetical conlang gloss:

kill-1sgSUBJ-3sgOBJ wabbit-ACC
->
kill-CAUS-3sgSUBJ-1sgOBJ wabbit-OBLIQUE

The former object is now an oblique. But again, languages vary greatly in how exactly they shift the valency and theta rolls of causatives.

[Yng]

My question in particular is, are there languages which also used a voice to get rid of the Agens but retain both Indirect and Direct Object? The only good example I can think of would be the sentence "He gives her a book", which, of course, in "normal" passive becomes "A book is given to her".

Yes... the passive. You used it there. Unless I'm misunderstanding you.

Anyway, as Xephyr says, basically any valency-changing operation can be called a voice. Applicatives are fun.

[Mecislau]

The two big voices, that you didn't mention, are applicative and causative.

Also the middle voice (seen, at the very least, in a number of IE languages), which is a valency-decreasing operation in which the patient is promoted to subject and the agent is completely eliminated, and is present in English to some extent:

I broke the window (active)
The window was broken (passive)
The window broke (middle)

I burned the paper (active)
The paper was burned (passive)
The paper burned (middle)

I cut the bread (active)
The bread was cut (passive)
The bread cuts easily (middle)

[jal]

Also the middle voice

Where, to keep it fun, the name "medio-passive" is also used, depending on the language (and perhaps the morpho-syntax involved).


JAL

[eodrakken]

There's also direct-inverse alignment, also called inverse alignment or inverse voice. In that system, the verb is marked differently depending on the position of the subject in some sort of agent-worthiness hierarchy relative to the object. (I say "some sort of" because the details differ by language.) For example, in Plains Cree, when the subject is first or second person and the object is not, the verb is in the direct voice. When the object is first or second person and the subject is not, the verb is in the inverse voice.

ni-sēkih-ā-nān atim
1-scare-DIRECT-1PL dog
We scare the dog

ni-sēkih-iko-nān atim
1-scare-INVERSE-1PL dog
The dog scares us

As you can see, the direct or inverse marking on the verb is the only way to retrieve the case roles, and the verb's agreement marking does not change (i.e. the verb doesn't agree with "dog" in the second example). At first glance, the second example looks a lot like a passive -- "We are scared by the dog" -- but it isn't actually the same. In direct-inverse the verb is not detransitivized, the "logical subject" is not syntactically demoted. It's a valence rearrangement but not a valence reduction.

When both arguments are at the same position on the agent-worthiness hierarchy, a direct-inverse language needs some way of arbitrarily distinguishing which is to be considered higher for the purposes of the case roles being retrievable. In Ojibwe, one of the arguments is given "obviative" marking, and is considered lower on the hierarchy (just for that sentence).

ikwe o-wābam-ā-n ininiw-an
woman 3-see-DIRECT-OBV man-OBV
The woman sees the man

If the verb had inverse marking, or if "woman" were marked as the obviative instead, the sentence would of course mean "The man sees the woman". (The non-obviative noun, unmarked in Ojibwe, is called "proximate".) I got these examples from Grammatical Voice by M. H. Klaiman, who further notes that you can arrange the words in the Ojibwe sentence in any order without altering the case roles.

I gather that not every linguist analyzes direct-inverse alignment as a voice system, but I think it is. If you want to know about all attested types of valence rearrangement, it should certainly be of interest.

(edit: spelling)

[Whimemsz]

(Just a quick correction: the Ojibwe obviative of "man" is ininiwan, with a short /a/, not a long one).

Also the proximate/obviative assignments aren't always just for the one sentence. Depending on the dialect and on the discourse situation it can last only for the one clause, or it can last for the whole paragraph or discourse. I'm still not clear on the conditions under which the distinction is held or not, though.

Anyway, it's also true that the verb and its arguments can appear in any order, although actual clauses in which both subject, object, and verb appear are rare, because person-marking on the verb takes the place of overt (pro)nominals most of the time.

To expand the example, though, to say "the man (prox) sees the woman (obv)", you could say:

Ikwewan owaabamaan inini or Inini owaabamaan ikwewan or Owaabamaan inini ikwewan etc.

For "the woman (obv) sees the man (prox)" you could say:

Ikwewan owaabamigoon inini or Inini owaabamigoon ikwewan etc. etc. (with the inverse suffix -igw-)

[eodrakken]

Cool, thanks for the additional info. (The typo on ininiwan was Klaiman's, though looking again, he had it right in other examples.)

[ebilein]

Yes... the passive. You used it there. Unless I'm misunderstanding you.

I probably didn't explain that thoroughly. "He gave her the book" has, of course, the passive "The book was given to her", but what I was particularly interested in was whether a language can produce "She received a book", thus one argument less, but with the English indirect object "promoted" to the Subject of the sentence (an idea that came following a lecture about Generative Grammar, where our prof explained English passive).

Edit: Okay, I just realised (or at least I think) that it's possible in English, thus "She was given a book." I was probably just thinking too much in German.

[Yng]

Yes... the passive. You used it there. Unless I'm misunderstanding you.

I probably didn't explain that thoroughly. "He gave her the book" has, of course, the passive "The book was given to her", but what I was particularly interested in was whether a language can produce "She received a book", thus one argument less, but with the English indirect object "promoted" to the Subject of the sentence (an idea that came following a lecture about Generative Grammar, where our prof explained English passive).

Edit: Okay, I just realised (or at least I think) that it's possible in English, thus "She was given a book." I was probably just thinking too much in German.

Ahh, I see. Hmm, yeah, I think that's how you'd have to do it. Of course, it doesn't work with verbs that don't allow that funny dative shift:

I said the words to the man
*The man was said the words.

[TaylorS]

Causatives vary wildly across languages, so I'm not sure whether to call them valency-increasing or not. A prototypical causative turns a transitive sentence into a ditransitive sentence: I killed the wabbit. -> Þór made me kill the wabbit. English is a bit wonky here, but the valency shift is easier to see in an hypothetical conlang gloss:

kill-1sgSUBJ-3sgOBJ wabbit-ACC
->
kill-CAUS-3sgSUBJ-1sgOBJ wabbit-OBLIQUE

The former object is now an oblique. But again, languages vary greatly in how exactly they shift the valency and theta rolls of causatives.

OOH! This is useful! I've been trying to figure out how to get things like causatives and desideritives to work in a polysynthetic verb and this helps a lot.

[TaylorS]

Also the middle voice

Where, to keep it fun, the name "medio-passive" is also used, depending on the language (and perhaps the morpho-syntax involved).


JAL

I my conlang Alpic there is a Mediopassive voice and whether it is semantically middle/reflexive or passive depends on whether the intransitive subject is marked as Agent or Patient.

Kjuspalim "I hurt myself"
hurt-PST.MD-1SG.A

Kjuspale "I got hurt"
hurt-PST.MD-1SG.P

[Jetboy]

Edit: Okay, I just realised (or at least I think) that it's possible in English, thus "She was given a book." I was probably just thinking too much in German.

I think that's called a dechticaetiative construction (dechticaetiative is one of most annoying words to spell). So it is a type of passive, but it involves some sort of departure from English's usual dative morphosyntactic alignment, I think, and is more specifically called a Dative shift.

[Yng]

Edit: Okay, I just realised (or at least I think) that it's possible in English, thus "She was given a book." I was probably just thinking too much in German.

I think that's called a dechticaetiative construction (dechticaetiative is one of most annoying words to spell). So it is a type of passive, but it involves some sort of departure from English's usual dative morphosyntactic alignment, I think, and is more specifically called a Dative shift.

Yes, as I said - it doesn't work with verbs that take indirect objects but don't allow dative-shifting ('say', for example).

[Tropylium⁺]

I once work'd out a bunch of theoretical voices and realized that ditransitivs allow three varieties of most operations. Using a schematic representation A --B--> C with A donor, B theme, C recipient:

• Valence-reducing voices
1) B ---> C
2) A ---> C
3) A ---> B

• Reciprocal voice
1) A --B--> C & A --C--> B
2) A --B--> C & C --B--> A
3) A --B--> C & B --A--> C

• Reflexiv voice
1) A --B--> B
2) A --B--> A
3) B --B--> A

A similar division also emerges when applying valence-increasing voices to a monotransitiv X ---> Y (agent, patient), but as it's not obvious which of theme and recipient is more "patient-like", we get four permutations:
1) Z --X--> Y
2) Z --Y--> X
3) X --Z--> Y
4) X --Y--> Z

1) and 2) are two different types of causativs; 4) is applicativ. 3) might be call'd an "instrumental voice", but I've never come across anything like that (since usually one doesn't add a theme, but just an obliq argument).

The two other types don't seem plausible (they would have receiver patient), but perhaps if the operation didn't create a ditransitiv, but a monotransitiv + obliq… We still only get 4 voices tho, as a voice where the new argument is the obliq actually is the same voice.
1) Z --> Y (X)
2) Z --> X (Y)
3) X --> Z (Y)
4) Y --> Z (X)

[Chuma]

Hm, I thought we had a thread about this recently. Or maybe I'm just dejavuing.

Anyway. In Swedish we have this funny voice called absolute. It basically means that you do something in general. Examples:

"the dog bit him" - active
"the dog was bitten" - passive
"the dog bites (people, in general, so watch out)" - absolute

"he hit him" - active
"he was hit" - passive
"they fight" - absolute

We have morphological voice marking (which is nice) but the voice marking for passive and absolute are usually indistinguishable (which is stupid). They both get a suffix /s/. Sometimes it is different, tho:
"hit" slå /slo:/
"be hit" slås /slo:s/
"fight" slåss /slos/

Can't think of any other examples... maybe "bite" as well, actually, in dialects.

[merijn]

I think I have linked this article before, which is about the Dutch "krijgen"-passive, where the indirect object is promoted instead of the direct object. IIRC this construction also exists in German. A Dutch example is "Zij kregen het boek aangeboden" which means "they were offered the book" lit. "they got the book offered".
Another voice I haven't seen mentioned is one that was called middle by the teacher I had a couple of years ago who was doing her PhD about this voice but which has nothing to do with what traditionally are called middles. You supposedly have them in English as well (at least for some verbs, I don't know exactly if "read" is one of them so I don't know if this examples is valid, but I think you'll get the basic meaning): "These books read easily"

++++++

In Bantu languages voices are expressed by so-called extensions. These extensions are morphemes that directly follow the stem, or another extension. Some extensions are used for verb derivation, and the line is blurry (if there is a difference between inflection and derivation, and to what category voice belongs, but that is a different story). Zulu has the following productive extensions, these examples are mine (so since I am not a native speaker may not be a hundred percent reliable, but I don't have a grammar book with me at the moment, and it costs too much to look up real life examples):
1) passive: -w- or -iw- with palatalization of non-stem-initial bi-labials. the original subject is expressed by the copular particle.
ng-a-cul-a iculo 1sg-pst-sing-pst song "I sang a song" iculo l-a-cul-w-a yi-mi song(nc7) 3sg(nc7)-pst-sing-pass-pst cop-1sg "a song was sung by me"
2)stative: -ek-. This one has a variety of closely related meanings, sometimes it is close to a passive, sometimes it is closer to middles (the traditional kind), and sometimes it means "to be x-able". There are a lot of lexically determined slightly irregular meanings:
-thand- "love" -thandek- "be loved"
-cul- "sing" -culek- "be singable"
-lahl- "throw away" lahlek "get lost"
3) reciprocal -an-: this one means "to X each other", this one has lots of lexically determined meanings as well, as well as verbs where the original verb is lost. When the meaning is irregular and the resulting meaning is transitive the object is introduced with the preposition na- "with"
-shay- beat -shayan- beat each other
-thand- love -thandan- be in love with
-ehluk- be deviant from -ehlukan- differ
-hlangan- assemble, meet, there is no verb -hlang-
4) applicative-el-: this one is complicated with lots of different meanings and all kinds of syntactic stuff going on. They can be used with benefactives (for) and circumstantial (because of). where the benefactive or circumstantial is promoted to the object. It can also be used with locatives, but then nothing is promoted to the object position. Sometimes the only difference in that case is just a matter of emphasis, but sometimes with some verbs of movement the original verb means "from" whereas the applicative means "to"
-cul- sing -culel- sing for
ngi-gijim-a emzini : I run away from the village ngi-gijimel-a emzini I run towards the village
5) causative-is-: this one is usually a simple causative, but it can also mean "help with". There are a few verbs where the meaning is lexically determined.
-cul- sing -culis- make someone sing
-sebenz- work -sebenzis- use

[jal]

I my conlang Alpic there is a Mediopassive voice and whether it is semantically middle/reflexive or passive depends on whether the intransitive subject is marked as Agent or Patient.

Kjuspalim "I hurt myself"
hurt-PST.MD-1SG.A

Kjuspale "I got hurt"
hurt-PST.MD-1SG.P

The thing is though, that the second is more of an example of a middle voice than a passive (as opposed to "I was hurt"). "got" passives are usually more middle-like in English. The best English verb to demonstrate reflexive vs. middle vs. passive is "to drown":

"He drowns himself" - reflexive, suicide
"He drowns the kittens" - active, murder
"He drowned" / ?"He got drowned" - middle, accident [the latter may or may not be acceptable to a/some/all native speakers]

It of course depends on the language if any are lumped together, and in which (semantic) cases.


JAL

[jal]

Anyway. In Swedish we have this funny voice called absolute. It basically means that you do something in general.

Which is why semantically it is more of a habitual aspect, or general mood. Which is not to say it isn't called a voice, and it seems to reduce valency.


JAL

[jal]

"These books read easily"

I would say that's a pretty normal middle voice, cf. "these vases break easily", "these clothes wear comfortably". They translate nicely to "these books are easy to read", "these vases are easy to break", "these clothes are comfortable to wear" etc. There may be a semantic difference between one form or the other, but not a syntactic one. Cf. "he drowns easily", which may both mean "he is prone to drowning" and "it's easy to drown him".


JAL

[Travis B.]

I my conlang Alpic there is a Mediopassive voice and whether it is semantically middle/reflexive or passive depends on whether the intransitive subject is marked as Agent or Patient.

Kjuspalim "I hurt myself"
hurt-PST.MD-1SG.A

Kjuspale "I got hurt"
hurt-PST.MD-1SG.P

The thing is though, that the second is more of an example of a middle voice than a passive (as opposed to "I was hurt"). "got" passives are usually more middle-like in English. The best English verb to demonstrate reflexive vs. middle vs. passive is "to drown":

"He drowns himself" - reflexive, suicide
"He drowns the kittens" - active, murder
"He drowned" / ?"He got drowned" - middle, accident [the latter may or may not be acceptable to a/some/all native speakers]

It of course depends on the language if any are lumped together, and in which (semantic) cases.

I should note here that the difference between be-passives and get-passives in English is that the former are static passives whereas the latter are dynamic passives (register differences aside), not of one having the characteristics of a middle voice. Actually, I would say that get-passives are less middle voice-like than be-passives as they strongly indicate that something was done to something, while be-passives only indicate the state something is in regardless of how it entered that state.

Hence "He got drowned" strongly implies that the individual was actively killed by someone or something, whether intentionally or accidentally. Conversely, "He was drowned" has no such implication, and leaves how the drowning occurred open, with suicide and accidents not involving the actions of anyone or anything else not being excluded. In this way, "He was drowned" is more middle voice-like than "He got drowned", and much closer in meaning to the middle voice "He drowned".

("He was drowned", I should note, is not a very natural form in English, as most English-speakers would normally use "He drowned" in its place.)

[jal]

I should note here that the difference between be-passives and get-passives in English is that the former are static passives whereas the latter are dynamic passives (register differences aside), not of one having the characteristics of a middle voice. Actually, I would say that get-passives are less middle voice-like than be-passives as they strongly indicate that something was done to something, while be-passives only indicate the state something is in regardless of how it entered that state.

Hence "He got drowned" strongly implies that the individual was actively killed by someone or something, whether intentionally or accidentally. Conversely, "He was drowned" has no such implication, and leaves how the drowning occurred open, with suicide and accidents not involving the actions of anyone or anything else not being excluded. In this way, "He was drowned" is more middle voice-like than "He got drowned", and much closer in meaning to the middle voice "He drowned".

("He was drowned", I should note, is not a very natural form in English, as most English-speakers would normally use "He drowned" in its place.)

Interesting observation, but though I'm not a native speaker of English, I'm not sure this goes for all native English speakers. "He was drowned" seems to strongly imply "by someone" and not "by accident" (which "he drowned" strongly does). Perhaps it's because of "got" sentences like "he got lost", which to me strongly imply a middle voice, that I parse "he got hurt" (definitely middle voice for me) and "he got drowned" (less strong) as middle. I think I parse the last example as "he got himself into such a situation that he drowned", or the like, but that may be false inference.


JAL

[Chuma]

Anyway. In Swedish we have this funny voice called absolute. It basically means that you do something in general.

Which is why semantically it is more of a habitual aspect, or general mood. Which is not to say it isn't called a voice, and it seems to reduce valency.

That's certainly true - real languages don't always neatly follow abstractions like voice-tense-aspect-mood. But there is another good reason to call it a voice: It can't be combined with any other voice, but it can be combined with another mood (namely imperative).

[merijn]

"These books read easily"

I would say that's a pretty normal middle voice, cf. "these vases break easily", "these clothes wear comfortably". They translate nicely to "these books are easy to read", "these vases are easy to break", "these clothes are comfortable to wear" etc. There may be a semantic difference between one form or the other, but not a syntactic one. Cf. "he drowns easily", which may both mean "he is prone to drowning" and "it's easy to drown him".


JAL

Voice has both semantic and syntactic sides, so I don't know if saying that the difference is just semantic is a valid reason to say that it is essentially the same voice. Besides, there is a syntactic difference between the traditional middle and this midde; you can say "he drowns" but you cannot say "*these books read" with the intended meaning.

[TaylorS]

I my conlang Alpic there is a Mediopassive voice and whether it is semantically middle/reflexive or passive depends on whether the intransitive subject is marked as Agent or Patient.

Kjuspalim "I hurt myself"
hurt-PST.MD-1SG.A

Kjuspale "I got hurt"
hurt-PST.MD-1SG.P

The thing is though, that the second is more of an example of a middle voice than a passive (as opposed to "I was hurt"). "got" passives are usually more middle-like in English. The best English verb to demonstrate reflexive vs. middle vs. passive is "to drown":

"He drowns himself" - reflexive, suicide
"He drowns the kittens" - active, murder
"He drowned" / ?"He got drowned" - middle, accident [the latter may or may not be acceptable to a/some/all native speakers]

It of course depends on the language if any are lumped together, and in which (semantic) cases.


JAL

Oh, good point, I never noticed that before, I knew there was a difference in meaning between Be Passives and Get Passives, but I could never figure out what.