Are there other voices besides active and passive?

[Khvaragh]

3. I would also throw impersonal verbs, which have no agent or patient.

Isn't that liked the sentence "It rained." where the pronoun "it" merely fills a mandatory dummy slot in English but grammatically the sentence has zero valence?

[TomHChappell]

Btw, I don't think anyone's mentioned the Inverse voice, which is like the passive only "better"-- instead of demoting the agent to an oblique role, it makes it into the object.

That's not precisely how I'd explain what happens in inverse voice.
Inverse voice applies to the Hierarchical morphosyntactic alignment.
In Hierarchical morphosyntactic alignment, the Direct voice -- the unmarked voice -- signals that whichever participant is higher on the hierarchy is the agent, and the lower participant is the patient.
In Hierarchical morphosyntactic alignment, the Inverse voice -- the marked voice -- signals that whichever participant is higher on the hierarchy is the patient, and the lower participant is the agent.
Usually the verb agrees with the higher participant and does not agree with the lower participant.
So what you said is something like what happens; but I think it could be misleading to not state that it couldn't apply to accusative nor to ergative nor to active/stative nor to tripartite morphosyntactic alignment systems.

@Xephyr:
I have been reading Farrell's "Grammatical Relations" (unfortunately I have to return it Monday even though I've just finished Chapter 2).
Apparently there are two good ways to interpret "Inverse Voice"; one is indeed that which I described above, but the other is exactly what you described -- that is, in Inverse Voice the semantic Patient occupies the Subject grammatical relation while the semantic Agent occupies the Object grammatical relation. For his main example language, Farrell actually prefers the second (i.e. your) interpretation.

===============================================

I guess in a tripartite system they'll be three types of passive. But only one would probably be used, as there's little sense putting an intransitive into the ergative or such like. But i could be wrong.

Antipassive is a bit tricky to explain; it's like passive the other way round. In a passive sentence what would be the object (I read the book) becomes the subject (the book was read by me); with antipassive the ergative noun phrase becomes absolutive.
Hmm. This makes me wonder: is there any weird extra voice you can get if your language has a tripartite system?

In a word: no. Remember that the third case is the subject of an intransitive verb, which by definition has no object.

Isn't "antipassive" a term used by early linguists encountering ergative langs to explain away its system?
Or something like that.

Not quite sure what you mean here. There are many ergative languages that don't make use of an antipassive - just as many nominative ones don't have a passive. Those ones that do have the system though, do require a distinct label.
What the passive and the antipassive both have in common is that they're essentially a system of detransitivising. This means in a nominative system, that the subject loses its special status, and in an ergative system the patient loses its special status. Both can only appear, after all, with fully transitive verbs.

@Salmoneus, @Ahribar, and @you other guys:
Farrell gives examples of languages that have both a passive and an anti-passive. I'd never have expected that.

[Whimemsz]

What is Farrell's main example language for inverse voice?

[TomHChappell]

What is Farrell's main example language for inverse voice?

The Chilean language Mapudungun.
He also uses examples from the Algonquian language Plains Cree.
He mentions Algonkin, Ojibwa, Tanoan, Kinyarwanda, and Tzotzil.

But those aren't the languages I was referring to when I said he prefers Xephyr's analysis; that's the Amazonian language Jarawara. (I made a mistake in my memory.) He does discuss Jarawara at some length (pp. 81-83), so it is a main example if not the main example.

[TomHChappell]

3. I would also throw impersonal verbs, which have no agent or patient.

Isn't that liked the sentence "It rained." where the pronoun "it" merely fills a mandatory dummy slot in English but grammatically the sentence has zero valence?

Sort of, yes.
You misused the word "grammatically"; you meant "semantically".
Semantically, the sentence "it rained" has zero valence, but grammatically, it has a single participant, the subject "it", which is a dummy.
In some languages it isn't necessary to provide a dummy subject. These languages allow 0-valent verbs and 0-valent clauses. (Some people use the word "avalent" to apply to these; but maybe some think "avalent" refers to things that have no valency whatsoever, rather than things whose valency is 0.)
(BTW the valency of a verb in the lexicon is the number of participants it "specifically licenses"; whereas the valency of a verb used in a clause is the number of such participants which explicitly appear in that clause. So the valency of a simple clause is at most the valency of its verb, but may be less than that.)
Neek's definition of impersonal verbs appears to be those that have no agent and no patient. The definition I'm used to is those that have no participants -- no arguments -- at all; in other words 0-valent ones. Conceivably a verb could have no agent and no patient, yet have a participant that was neither agent nor patient. If so such a verb wouldn't be impersonal by my definition, but would by Neek's.

[Chuma]

Despite being indo-european, Swedish has got some interesting voice phenomena.

Just like Latin there are deponent verbs (passive in form but active in meaning). Also, there are reciprocal verbs, which if put into passive mode translate as "each other". An example: Att se, to see, in 1st person plural becomes vi ser. However, to say "we'll see each other" it's perfectly legal to say vi ses, which also means "we are being seen".

Third, there are two ways to turn verbs into passive voice. You can do it the English way: "we were being seen", vi blev sedda, or the aforementioned way, vi s?gs. Only the second way can be reciprocal.

Nifty, huh?

I knew it's a strange language I speak, but you make it sound more exotic than I ever imagined...

Let me elaborate on these "reciprocal verbs".

Most verbs in swedish, except a few shorter ones, end on -a in the infinite form.
Example: att köra, "to drive"; att sparka, "to kick"; att se, "to see".
The imperative is often the same, or you remove the -a:
Kör!, Sparka!, Se!
To get the present tense, you add -r:
han kör, "he drives"; han sparkar; han ser.
Older Swedish used to inflect for person and number, but in modern Swedish there is no difference.

Like theforceman said, Swedish has two ways to make things passive. You can use the "real" passive voice, which usually means adding an -s:
att köras, "to be driven"; han körs, "he is driven"; att sparkas, "to be kicked"; han sparkas, "he is (being) kicked".
Or you can do it like in English:
att bli körd, "to be driven"; han blir körd, "he is (being) driven".
(Bli means "become", but is used as an auxilliary here. Incidentally, körd can also mean "stupid".)

Some words also have intransitive / transitive forms, where English uses the same word: brinna and bränna both mean "burn", but the former is intransitive (so brännas means "be burnt", while brinnas is not a word at all).

Then, there's that special form, which in Swedish is called "absolute". It works sort of like transitive, but the object is "general". Unfortunately, this form is ALSO expressed by adding an -s. (Before you complain about what a stupid language it is, take a look at "its" an "it's".)
So:
Hunden biter honom, "the dog bites him".
Han bits av hunden, "he is (being) bitten by the dog".
Hunden bits, "the dog bites (people, in general, if you don't watch out)".
Only with some words is there a difference between passive and absolute. The only example I can come up with now is:
att slå, "to hit"; han slår, "he hits (someone)";
att slås, "to be hit"; han slås, "he is (being) hit";
att slåss, "to fight" ("to hit each other"); de slåss, "they fight".
(The past tense would be slogs, spelled the same way for both, but pronounced differently.)

Similarly, sparkas can also be absolute, meaning something like "kick eachother" or "kick people (in general, as a habit)". Also, brännas can be some sort of absolute. You might say about a stove, or nettles, or something:
den bränns!, "it's really hot!"
Or you can try the word kittla, "tickle", to get some interesting combinations:
han kittlar henne, "he tickles her";
hon kittlas, "she is (being) tickled" (passive);
han kittlas, "he tickles (me, or her, or someone else, whatever)" (absolute);
det kittlas!, "that tickles!" (also absolute).

[dhok]

Yay thread revival!

Anyway, when I look at some languages, it seems like voice and mood seem to blend together.

And, causitivity doesn't have to be a voice. You can have regular active/passive AND causitive/non-causitive.

[TomHChappell]

when I look at some languages, it seems like voice and mood seem to blend together.

Nearly any accident of a verb can blend with nearly any other accident of a verb, in some languages; especially fusional languages. So, just as tense and aspect, tense and mood, and/or aspect and mood can "blend together" (or even "fuse"), so also can voice and tense, voice and aspect, and/or voice and mood.

And, causativity doesn't have to be a voice.

For most languages which have morphological causatives, causativity is just as much of a "voice" as applicativity is (in most languages which have applicatives).

You can have regular active/passive AND causative/non-causative.

Which doesn't prove causative is not a voice, does it?
It may depend on what you mean, which, I'm afraid, isn't perfectly clear to me.
Do you have some examples?
Is there a language with:

• an active causative sentence
• an active non-causative sentence
• a passive causative sentence
• a passive non-causative sentence

all with the same root verb?
If not, why isn't causative always a voice?
If so, why isn't causative a voice in that language?

[dhok]

Parkenese is the only major language of the Micro-Miqobian languages- none of the others have more than 500,000 speakers. They display verb conjugation which looks triconsonental but isn't necessarily. They have active, passive, stative, reciprocal, and reflexive voices-although North Micro-Miqobian languages lost the reciprocal a long time ago, using a particle+ the reflexive. They're all going strong in Parkanese, though.
By the way, how does the cooperative work?

[TomHChappell]

Parkenese is the only major language of the Micro-Miqobian languages- none of the others have more than 500,000 speakers.

Urban Dictionary definition of Parkenese

Tb'elajlaji'n tiipaky' (19)
Kyaqiilqe xjaal tzoqpi'nqe ti'j kyxinb'il b'ix kyyool tu'n ti' kyq'ama'n, aj tokleen lu' milaay chi ook mayeet tu'n tlaaj kyyool, tu'n tjayeet jun xinb'il, tu'n tkaanan tpakb'aal b'ix yool noq alkyeex tu'mal, tuj puqb'il yool, tuj u'j miti' maqob'il tee.

The 19th-century Lomi-Ek poet Baraqat Maqöb --briefly canonized in volumes of the literature of the Soviet peoples, only to be removed in the mid-1930s and forgotten until the 2003 publication by Duquesne University Press of an anthology of Great Nationalist Poets of North Asia, where he is hailed in Rosalind Needleman's introduction as a genre-transcending, playful modernist, remarkably anticipating the European avant garde from his distant colonial outpost--

So;

Is it Micro-Maqobian or Micro-Miqobian?

Any relation to Mam?

Any relation to Lomi-Ek?

Is it Parkanese or Parkenese?

Any relation to the language mentioned in the Urban Dictionary?

[dhok]

Oops. No relation. It comes from parke, people.
It's Micro-Miqobian.

[abeygail]

I do not understand the differnece between a "Dative applicative voice" and a "Dative Case". Is it that Dative case requires a transitive clause while a dative Voice does not?

[dhok]

One's for nouns, and one's for verbs...can't wait for someone to pull out a language with 3 speakers that has dative case on verbs.

[cromulent]

I do not understand the difference between a "Dative applicative voice" and a "Dative Case". Is it that Dative case requires a transitive clause while a dative Voice does not?

The one's a case, the other's a voice. Case is for nouns; voice, for predicates.

Dative case marks the Recipient of ditransitive clauses. It requires a ditransitive verb. (There are quirky uses that don't though).

AFAICT, "dative applicative voice" refers to "dative shifting," which is the voice-like process (in for example English) where the Recipient (semantic role) is mapped onto the accusative rather than the dative syntactic role: "I gave a cake to Walden" becomes "I gave Walden a cake." It's not a true applicative voice because there is no increase in the valency of the verb.

[cromulent]

One's for nouns, and one's for verbs...can't wait for someone to pull out a language with 3 speakers that has dative case on verbs.

You're gonna wanna read about Tlapanec. It marks case on the verb. This includes the elusive pegative case!

A pegative language is somewhere on my backburner of ideas to fuck around with.

[TomHChappell]

I do not understand the differnece between a "Dative applicative voice" and a "Dative Case". Is it that Dative case requires a transitive clause while a dative Voice does not?

"Dative Applicative Voice" is a transformation that promotes some oblique argument into the "Dative" slot. The verb gets marked to show that something has happened, and now looks ditransitive. The formerly-oblique argument gets marked with the Dative case (whether that's actually an inflection of the noun, or some separate word like an adposition). The addressee can tell from the (new) form of the verb that it's "Indirect Object" isn't meant to be a Recipient, but rather is an argument or participant that would have been expressed as an oblique argument if the clause were not "Dative Applicativized".

Some theoretical examples:

Pre-transformation: "I-NOM hit-ACTV the ball-ACC for him".

The phrase "for him" says the Beneficiary is "him". It's not a core argument; it's an oblique argument. It's not a direct participant of the core of the verb "hit" -- not its Agent nor its Patient. In this clause the verb "hit" is a standard monotransitive bivalent verb.

After Dative Applicativization: "I-NOM hit-DAT.APPL the ball-ACC him-DAT"
(or, maybe, "I-NOM hit-DAT.APPL the ball-ACC to him-DAT" since in English the dative is marked by the adposition "to".)

The -DAT.APPL voice of the verb "hit" lets the addressee know that the dative argument, "him-DAT", is not the recipient, but rather is some other argument; in other words, the addressee knows I'm not hitting the ball to him, but rather, at him, for him, in him (highly unlikely, IMO), near him, off him, with him, or something like that.

In about half of languages that have Dative Applicativization, there would be more than one kind of -DAT.APPL inflection that could go on the verb; which one was used could give the addressee some hint as to which kind of oblique argument has been promoted to "Dative" (to "Indirect Object").

AFAICT, "dative applicative voice" refers to "dative shifting," which is the voice-like process (in for example English) where the Recipient (semantic role) is mapped onto the accusative rather than the dative syntactic role: "I gave a cake to Walden" becomes "I gave Walden a cake."

No, but that's close.

"Applicativization" is any transformation that can move an oblique argument into the "Direct Object" or "Primary Object" slot. If performing that transformation requires marking the verb, that marking is called "Applicative Voice".

"Dative shifting" is very much like a kind of Applicative Voice, because it also promotes a "lower" participant into the Direct Object "slot"; in the case of Dative Shifting, it's the Indirect Object that gets so promoted.

If "Dative Shifting" isn't a kind of Applicativization, the only reason it fails to be Applicativization is that the promoted argument wasn't oblique before the transformation, but instead was an Indirect Object or Secondary Object -- in other words, something in the Dative slot and perhaps in the Dative Case, before the transformation.

It's not a true applicative voice because there is no increase in the valency of the verb.

No, again; but again, partly correct.

The reason it's not a "true applicative" -- if it isn't -- is that applicativization is supposed to promote and oblique argument to Direct Object, but Dative Shifting promotes the Indirect Object to Direct Object. In languages that have ditransitive verbs -- most of them, in other words -- Indirect Objects don't count as Oblique (though they're a lot closer to Oblique than either Subjects or Direct Objects).

Applicativization may or may not raise the valency of the verb.

If the verb doesn't already have a direct object until the transformation, applicativization gives it one, thus raising its valency from 1 to 2.

If the verb does already have a direct object before the transformation, applicativization may -- in some languages -- require that that original D.O. be demoted or made implicit, in order to make room for the "new" D.O. (In some languages no clause can have two Direct Objects; if such a language has Applicativization, it would behave as just described.)
Applicativization tends not to occur in languages that actually have three Grammatical Relations; languages with Applicativization tend to be two-GR languages. Likewise, languages with Primary Objects and Dechticaetiative alignment, tend not to be languages with three GRs, but rather with just two GRs; for them, the Secondary Object -- (is that the Dechticaetiative?) -- is a kind of oblique argument.

(But I don't know that there are no languages with three GRs and Dechticaetiative alignment -- I wouldn't be surprised if there are some, I'd just be curious how they work.)

Likewise, Dative Applicativization may or may not raise the valency of the verb.

If the verb doesn't already have an Indirect Object until the transformation, Dative Applicativization gives it one, thus raising its valency from 2 to 3.

If the verb does already have an Indirect Object before the transformation, Dative Applicativization may -- in some languages -- require that that original I.O. be demoted or made implicit, in order to make room for the "new" I.O. (In some languages no clause can have two Indirect Objects; if such a language has Dative Applicativization, it would behave as just described. But languages are much likelier to tolerate two Indirect Objects than to tolerate two Direct Objects.)

Languages with three GRs are much more likely to have Dative Applicativization and not to have Applicativization, than the other way around; and also likelier to have just Dative Applicativization and not "true Applicativization" than to have both.)