Tritransitivity?

[sarcasmo]

In my studies of French, I came across the structure "to buy something (for someone)." (Je t'ai acheté un cadeau = I bought you a gift = I bought a gift for you.) French, and English, I think, treat this as an indirect object, IIRC. My question is, does any language, natural or constructed, permit a direct object, an indirect object and this third type of object all on the same verb?

For example, *"Give me him it." As in, "give it to him for me." I googled tritransitivity, but apparently that refers to such sentences as "I bet you ten pounds that they lose." They always use "that," which isn't exactly what I'm looking for.

[Soap]

Basque.

I cant give much for details, I just know that it does have this feature in at least some verbs (but then, there are only a few dozen verbs in Basque anyway). According to my Encyclopedia Britannica entry, dia means "he has it for him" (you'd fill in the he/it/him with nouns).

[Viktor77]

The basic VP in French could be said to have 4 specific slots. To have them all, you need a ditransitive verb, so we will use the verb "donner" (to give), though there are other verbs that accept all for slots. Here is a classical example :

Je te le lui donne.
Je donne pour toi* le cadeau à ton frère. (Sentence with full NPs)
I give it to him for you.
I give him a gift for you.

* You usually do not explicit the benefactive. It's an emotional statement more than a real benefactive.


This is the basic sentence that we will use for our examples.
In blue, you have the subject (SUB). In red, you have the object (OBJ). In green, you have the dative (DAT). In orange, you have the benefactive (BEN).

So, the basic slots would be SUB BEN OBJ DAT then verb. DAT and BEN work together : BEN can only receive 1st and 2nd person pronouns, while DAT can only receive 3rd person pronouns. If your dative would be a second or first, you treat it as a benefactive. (For reasons that I haven't figured out yet, it seems you cannot use a 3rd person benefactive in French, at least not within the four slots.)

[chris_notts]

Basque.

I cant give much for details, I just know that it does have this feature in at least some verbs (but then, there are only a few dozen verbs in Basque anyway). According to my Encyclopedia Britannica entry, dia means "he has it for him" (you'd fill in the he/it/him with nouns).

Basque isn't an example because the "for him" is just indirect object agreement. Finite verbs don't agree with 4 separate NPs in Basque, the maximum is 3 (absolutive, ergative, and recipient/beneficiary). In Basque, like in many other languages, recipient marking is also typically used for beneficiaries, as in your example, although there is also an optional separate case ending -entzat which only covers beneficiaries (Historically, this derives from a genitive + noun construction where the noun meant something like "benefit").

However, there are some complications in the system. For example, in dialects that retain the informal pronoun hi and its associated agreement, verbs often show agreement with the informal 2nd person even when that person didn't even take part in the action being described. This 2nd person agreement normally crops up in the recipient slot, but it may not always do so. For example:

polita da
pretty-DET 3rd-BE
"It's pretty (neutral)"

polita duk
pretty-DET 3rd-HAVE-2nd.MASC
"It's pretty (informally to a masculine addressee)"
(literally "You have it pretty")

Note that here the 2nd person is referenced by replacing "be" with a transitive verb, and the 2nd person agreement occurs in the ergative slot rather than the recipient slot. This ergative 2nd person agreement, instead of recipient agreement, may occur with almost all intransitive verbs when using the allocutive forms, but since they're mostly restricted to rural dialects, and since my Basque is very rusty, I'm not that confident about how to use them.

Furthermore, 2nd person ergative / recipient agreement shows different masculine and feminine forms, which is the only masculine / feminine agreement in Basque. Note that none of the independent pronouns show this distinction, even "hi".

[sarcasmo]

The basic VP in French could be said to have 4 specific slots. To have them all, you need a ditransitive verb, so we will use the verb "donner" (to give), though there are other verbs that accept all for slots. Here is a classical example :

Je te le lui donne.
Je donne pour toi* le cadeau à ton frère. (Sentence with full NPs)
I give it to him for you.
I give him a gift for you.

* You usually do not explicit the benefactive. It's an emotional statement more than a real benefactive.


This is the basic sentence that we will use for our examples.
In blue, you have the subject (SUB). In red, you have the object (OBJ). In green, you have the dative (DAT). In orange, you have the benefactive (BEN).

So, the basic slots would be SUB BEN OBJ DAT then verb. DAT and BEN work together : BEN can only receive 1st and 2nd person pronouns, while DAT can only receive 3rd person pronouns. If your dative would be a second or first, you treat it as a benefactive. (For reasons that I haven't figured out yet, it seems you cannot use a 3rd person benefactive in French, at least not within the four slots.)

This is exactly what I was thinking of. Strange restrictions based on the person, though. I feel like it's related to the order; in ditransitives, "lui" can never come before "te," so I guess by analogy the same thing applies to the four slots, whose SUB BEN OBJ DAT takes precedence, creating the restriction.

[Yng]

Some Bantu languages permit two applicatives, which I assume would allow their usual agreement for subject and object with an additional dative applicative and benefactive applicative, although this may not necessarily be the case (many languages, including English and French to a degree and definitely Latin, treat datives and benefactives the same).

[chris_notts]

Some Bantu languages permit two applicatives, which I assume would allow their usual agreement for subject and object with an additional dative applicative and benefactive applicative, although this may not necessarily be the case (many languages, including English and French to a degree and definitely Latin, treat datives and benefactives the same).

I have some Bantu examples somewhere that I'll dig up later. Similarly, I'm positive I have some examples of Amazonian languages which permit multiple applicatives (which have their own agreement) on the same verb.

Some Caucasian languages also have verbal prefixes that introduce extra agreement slots. It seems possible to combine them, and they cover a wide range of semantic roles, so that you can make the verb agree with almost all of the NPs in the clause if you want to. I think that agreement isn't obligatory in many cases, but I'm not sure what determines whether it occurs or not.

[Radius Solis]

I googled tritransitivity, but apparently that refers to such sentences as "I bet you ten pounds that they lose." They always use "that," which isn't exactly what I'm looking for.

It is what you're looking for, it's just not obvious at first glance. But don't be fooled by the fact the third object is a subclause in the English examples; those subclauses are nevertheless arguments of the verb. Such complement clauses are permitted for other argument types in English, not just that one - they can also be direct objects ("I like that you did that.") and even subjects ("That he said so bothers me a great deal.").

Some verbs allow or require a complement clause instead of a noun phrase for one of their arguments, and this happens to be the case with the English verb "bet" and maybe others. But these constructions are still tritransitive, or as clearly tritransitive as anything is likely to get.

Far less clearly tritransitive in English are where participants appear in prepositional phrases, like "I gave you a book for him". Such prepositional phrases look like adjuncts to me, whereas transitivity is about how many complements there are. But I've nevertheless seen these referred to as tritransitivity before... which I think is silly.

[TomHChappell]

It is what you're looking for, it's just not obvious at first glance. But don't be fooled by the fact the third object is a subclause in the English examples; those subclauses are nevertheless arguments of the verb. Such complement clauses are permitted for other argument types in English, not just that one - they can also be direct objects ("I like that you did that.") and even subjects ("That he said so bothers me a great deal.").

Some verbs allow or require a complement clause instead of a noun phrase for one of their arguments, and this happens to be the case with the English verb "bet" and maybe others. But these constructions are still tritransitive, or as clearly tritransitive as anything is likely to get.

Far less clearly tritransitive in English are where participants appear in prepositional phrases, like "I gave you a book for him". Such prepositional phrases look like adjuncts to me, whereas transitivity is about how many complements there are. But I've nevertheless seen these referred to as tritransitivity before... which I think is silly.

I think there are languages where such sentences can be stated in a single clause using only a single verb. Such sentences are the ones the O.P. is looking for, I believe.
Especially if all three objects occupy grammatical relations; the verb then would have a Subject, a Primary (or Direct) Object, and two Secondary (or Indirect) Objects.
(But maybe he wouldn't need that.)

You can get tetravalent verbs by applying valency-raising operations to ditransitive verbs.
If this is morphological causativization, it could be argued that you get two subjects and two objects rather than one subject and three objects, in which case the resulting verb might not necessarily be considered tritransitive.
But if it's applicativization or "dative applicativization", such as benefactive applicativization or instrumental applicativization, it's more clear that the inflected verb has three objects and is ditransitive.

If the O.P. is looking for verbs which are already tritransitive in root form, without having any valency-raising morphology applied to them, then most languages that have such verb-roots have only a tiny number of them, so odds are they're going to tend to have similar semantics from one language to another.

But if he will accept bivalent verbs to which two valency-raising morphological operations have been applied, or to which one has been applied twice; or to accept ditransitive verbs to which some valency-raising morphology has been applied; there are plenty of examples out there, just not in English.

But as you say a clause's arguments can (and in English often do) include "oblique arguments", that are just as crucial to the meaning and just as required by the verb as the Subject and/or Object and/or Indirect Object. That's why I'm not sure whether or not the O.P. would be happy with examples in which some of the arguments were oblique, or would instead require that all four arguments (Subject, Primary Object, and two Secondary Objects) occupy grammatical relations.

[sarcasmo]

What I had in mind was naturally tritransitive verbs, but I suppose ditransitives with increased valency satisfy my curiousity just as well.

Side note: are these verbs common enough that it would benefit a conlang to have a specific benefactive inflection on pronouns?

[chris_notts]

What I had in mind was naturally tritransitive verbs, but I suppose ditransitives with increased valency satisfy my curiousity just as well.

Side note: are these verbs common enough that it would benefit a conlang to have a specific benefactive inflection on pronouns?

Isn't that a separate question? As I mentioned before, Basque has a separate benefactive case, which can be used with pronouns, although you can also use the dative for this a lot of the time unless the clause already contains a distinct recipient or there would otherwise be confusion.

[Echobeats]

See here for verbs with a valency of 13.

[Chuma]

Echo: Okay, that's pretty awesome.

Correct me if I'm wrong, but I would say that the line can be a little fuzzy between what's a fundamental argument and what's an adjunct. You could say that the fundamental are always present, but there are certainly exceptions to that. In some languages, you can more or less identify the fundamental arguments by the fact that they are marked by word order or inflection, but the others are marked by some sort of extra words or affixes. That tends to limit the valency.

My conlang doesn't really distinguish between "mandatory" cases (such as dative) and others (such as locative), so valency is just a matter of semantics. I'd be very surprised if this didn't occur in natlangs.

[TomHChappell]

Echo: Okay, that's pretty awesome.

Correct me if I'm wrong, but I would say that the line can be a little fuzzy between what's a fundamental argument and what's an adjunct. You could say that the fundamental are always present, but there are certainly exceptions to that. In some languages, you can more or less identify the fundamental arguments by the fact that they are marked by word order or inflection, but the others are marked by some sort of extra words or affixes. That tends to limit the valency.

My conlang doesn't really distinguish between "mandatory" cases (such as dative) and others (such as locative), so valency is just a matter of semantics. I'd be very surprised if this didn't occur in natlangs.

I don't know, you may be right.
Linguists who tend to think that the concept of "valency" isn't much use are the ones who tend to find natlang examples of 7-valent verbs etc.; and maybe, also, who talk most about 0-valent verbs.
Likewise linguists who think the distinction between "argument" and "adjunct" isn't much use are the ones who find clauses with many many arguments and, maybe also, who talk most about clauses with no arguments.
Linguists who tend to think that the concept of "grammatical relation" isn't much use are the ones who tend to report natlangs of with more than 4 or fewer than 1 grammatical relations.

I don't know who's right.

I find that a conlanger is best served by adopting all theories simultaneously and just re-ranking them for each of his/her conlangs. Pick the pieces of each theory you want to use for this particular conlang and have at it. Sometimes you'll find that, in specific details, two (or even more) theories which are "theoretically" irreconcilable, actually work quite well together in particular parts of your conlang.

Some natlangs do have morphological causativization; some do have double morphological causativiziation.

Also, some natlangs do have (morphological) applicativization; some have (morphological) "dative applicativization"; and in some they can promote oblique arguments of many different sorts into the core.
Of course, for that last sentence to have any meaning, you have to accept that there is a difference between core and oblique, and that there is a difference between argument and adjunct.
A "core argument" is one that's in a grammatical relation; other arguments are "oblique"; and noun-phrases or pronoun-phrases that aren't arguments are "adjuncts" and have to be oblique and can't be promoted into the core.
If you don't believe any of that is true of the particular natlang you're studying (or the particular conlang you're creating), then you have to come up with some other explanation for what "applicativization" or "dative applicativization" (as the case may be) is in the 'lang in question.

Some natlangs not only have morphological causativization but also have applicativization (or dative applicativization). Some of them allow both operations to be applied to a verb.

Linguists who take all of that seriously are mostly agreed that all (as far as they know) of those languages that allow such "core-valency-raising" morphological operations to be applied to a verb, have a highest-allowed core valency for the resulting verb. In some you simply can't apply an operation that would raise the core valency above the limit; in some you can, but if you do, you must demote something out of the core into an oblique argument position or render it implicit.

For instance, in some languages that allow double morphological causativization, and have a maximum-allowable core valency of 4, you can causativize ditransitive verbs and doubly-causativize monotransitive and intransitive verbs, but you can't doubly-causativize ditransitive verbs. Or, in some languages that allow both morphological causativization and applicativization, you can apply either operation to trivalent verbs and can apply both operations (simultaneously, or one after the other) to bivalent and monovalent verbs, but you can't apply both operations to any trivalent verb, because that would violate the language's upper limit of core-valency.

-------------------------------------------------------------------------------------------------------------------------------------------------------------

Hindi is one of the languages that allows double morphological causativization. Hindi has two kinds of morphological causativization; one in which the causer more directly makes the causee do what the causee does (e.g. something that means "causer forced causee to hit himself" but has only one clause and only one verb), and another in which the causer more indirectly gets the causee to do what the causee does (e.g. something that would mean "causer persuaded causee to hit himself", but with only one clause and only one verb). Each of these can be applied to a verb only once; but both of them can be applied together ("instigator convinced middle agent to make performer hit himself" or "instigator forced middle agent to convince performer to hit himself").

Hindi is also one of the languages that allows double passivization; one can passivize a verb that's already passive.

Turkish also allows double-passivization.

So you might consider having a valency-reducing operation in your conlang that can be done twice to the same verb; or having two valency-reducing operations that can both be done to the same verb. In Turkish and Hindi, if you start with a bivalent verb and doubly-passivize it you wind up with a 0-valent verb.

-------------------------------------------------------------------------------------------------------------------------------------------------------------

If your 'lang has both valency-raising and valency-reducing morphology, and you can reduce then raise the valency or raise and then reduce the valency, you can probably essentially move the arguments into any arrangement you like.

Some languages with applicativization can promote most any argument into the direct-object slot via applicativization, but can't promote anything into the subject slot except the direct-object (which they do via passivization). But they allow speakers to applicativize and then passivize; thereby nearly any argument can be promoted into the subject position.

Likewise, some languages with dative applicativization can promote most any argument into the indirect-object slot via dative applicativization, but can't promote anything into the direct-object slot except the indirect-object (which they do via dative movement). But they allow speakers to apply dative applicativization and then dative movement; thereby nearly any argument can be promoted into the direct-object position.

In languages whose maximum allowed core-valency is, say, two, it is sometimes allowed to apply a valency-reducing morphology and then follow that up with a valency-raising morphology. By that means the valency-raising morphology can be applied to verbs that start out bivalent; first demote one of their core arguments out of the core by one of the valency-reducing operations, then promote some oblique argument into the core by one of the valency-raising operations. Something like that could also happen if the language's maximum allowed valency were three or one or four; I'm pretty sure I read of some natlang that attests one of those, but I'm not sure which ones.
But I think in most languages with applicativization, if you applicativize a verb that's already transitive, the applicativization operation not only promotes an oblique argument but also demotes the "current direct object" out of the core into an oblique position; so applicativization is valency-raising only on monovalent verbs in those languages. Likewise in most languages with dative applicativization, if you dative-applicativize a verb that's already ditransitive, that operation demotes the "current indirect object" at the same time as it promotes an oblique argument; so "dative applicativization" is valency-raising only on bivalent verbs in those languages. I don't have any references to back up the word "most", but I'm pretty sure if I'd replaced "most" with "some" there are natlang examples.

-----------------------------------------------------------------------

Is any of that useful to anyone?

______________________________________________________________________________________

Side note: are these verbs common enough that it would benefit a conlang to have a specific benefactive inflection on pronouns?

A benefactive case, whether just on pronouns or both on pronouns and nouns or even on pronouns, nouns, and adjectives, would be quite useful, and IMO even necessary if you don't want to show the beneficiary by means of a benefactive adposition (like English's "for") or by some kind of word-order thing (and I can't think how that could work) or by having the verb agree with the beneficiary.

That is, in every natlang I know about, there's some way of saying that the agent does what it does for the benefit of someone or something. That semantics has to be communicated somehow. It may be marked lexically (e.g. with a benefactive adposition) or syntactically (e.g. via word-order) or by dependent-marking morphology (e.g. a benefactive case-marking on the beneficiary's noun or pronoun) or by head-marking morphology (e.g. the verb agrees with the beneficiary in a special way) or some combination of two or more of those; e.g., many languages have both a benefactive adposition and a benefactive case, or a benefactive adposition whose PP (prepositional or postpositional phrase) has to go in a particular order relative to other arguments in the clause, or both verb-agreement and benefactive case.

What we were talking about before your last post was what can happen to verbs in natlangs. Now you're talking about what can happen to pronouns in natlangs. If you inflect both the verb and the pronoun to show who's the beneficiary, that's double-marking. Some languages do have double-marking; IMO consistent double-marking allows for a lot of freedom of word-order. If your language is heavily head-marking you're going to want to mark the verbs but not the nouns. If your language is heavily dependent-marking you're going to want to mark not only the pronouns, but also the nouns; but not to mark the verbs.

See WALS.info features 23, 24, and 25 for head-marking vs dependent-marking in the clause, in possessive phrases, and in the whole language.
See feature 50 for different case-systems for pronouns than for nouns.

The phenomenon of case-asymmetry is based on the division of a language’s nominals into subclasses sharing common semantic or functional characteristics (i.e., into non-morphologically based natural classes); these subclasses are called NP-types . Case-asymmetry manifests itself as differences between the case inventories of different NP-types (generally in the form of a deviation from a language’s prevalent paradigm pattern found only in some minoritarian NP-type). The NP-type most frequently affected by asymmetrical case-marking is the personal pronoun, but other pronoun types, as well as semantically defined subclasses of full nouns (e.g. nouns referring to humans, or person names), have also been identified as capable of showing deviant case inventories. Classes of nominals based purely on morphological criteria, such as declension classes lacking a common semantic basis or entirely irregular isolated word-paradigms, do not constitute NP-types, and hence do not manifest asymmetrical case-marking.

-----------------------------------------------------------------------

Is any of that useful to anyone?

__________________________________________________________________________________________

What I had in mind was naturally tritransitive verbs, but I suppose ditransitives with increased valency satisfy my curiousity just as well.

I Dogpiled for "tritransitive".
http://wwwstaff.eva.mpg.de/~cschmidt/SW ... ttilae.pdf You can, probably for a limited time, get this article in PDF form. It may be somewhat shorter than the article that appeared in the journal; or it may be an earlier version of what appeared in the journal.
http://www.reference-global.com/doi/abs ... G.2007.015 might be the same article as below.
http://www.accessmylibrary.com/article- ... types.html is an article about all types of tritransitives, or, at least, all that they knew about at the time.

Look in http://archives.conlang.info/whe/zhunsh ... ghuen.html at the bottom of
"Various weirdnesses in natlangs, plus obconlang questions following them
From: Eldin Raigmore <eldin_raigmore@...>
Date: Friday, May 19, 2006, 0:34"
for a list of web articles about tritransitives.

http://teacherseducation.wordpress.com/ ... nesday-29/ has "trade" and "bet".
http://www.jstor.org/pss/1263529 is about tritransitives in Sierra Popoluca, a Middle-American language.


-----------------------------------------------------------------------------------------------------------------------------------------------------

Can you say "Al convinced Bob to force Charlie to give Dasher to Elizabeth for Francis." with just one clause with just one verb in any natlang or conlang?
Al -- instigator
Bob -- middle agent
Charlie -- performer
Dasher -- theme
Elizabeth -- recipient
Francis -- benificiary.

[TomHChappell]

What I had in mind was naturally tritransitive verbs, but I suppose ditransitives with increased valency satisfy my curiousity just as well.

I Dogpiled for "tritransitive".
http://wwwstaff.eva.mpg.de/~cschmidt/SW ... ttilae.pdf You can, probably for a limited time, get this article in PDF form. It may be somewhat shorter than the article that appeared in the journal; or it may be an earlier version of what appeared in the journal.
http://www.reference-global.com/doi/abs ... G.2007.015 might be the same article as below.
http://www.accessmylibrary.com/article- ... types.html is an article about all types of tritransitives, or, at least, all that they knew about at the time.

Look in http://archives.conlang.info/whe/zhunsh ... ghuen.html at the bottom of
"Various weirdnesses in natlangs, plus obconlang questions following them
From: Eldin Raigmore <eldin_raigmore@...>
Date: Friday, May 19, 2006, 0:34"
for a list of web articles about tritransitives.

http://teacherseducation.wordpress.com/ ... nesday-29/ has "trade" and "bet".
http://www.jstor.org/pss/1263529 is about tritransitives in Sierra Popoluca, a Middle-American language.


-----------------------------------------------------------------------------------------------------------------------------------------------------

Can you say "Al convinced Bob to force Charlie to give Dasher to Elizabeth for Francis." with just one clause with just one verb in any natlang or conlang?
Al -- instigator
Bob -- middle agent
Charlie -- performer
Dasher -- theme
Elizabeth -- recipient
Francis -- benificiary.

[hwhatting]

Can you say "Al convinced Bob to force Charlie to give Dasher to Elizabeth for Francis." with just one clause with just one verb in any natlang or conlang?
Al -- instigator
Bob -- middle agent
Charlie -- performer
Dasher -- theme
Elizabeth -- recipient
Francis -- benificiary.

Don't know about natlangs, but you could easily make up such a conlang...

[Yng]

Can you say "Al convinced Bob to force Charlie to give Dasher to Elizabeth for Francis." with just one clause with just one verb in any natlang or conlang?
Al -- instigator
Bob -- middle agent
Charlie -- performer
Dasher -- theme
Elizabeth -- recipient
Francis -- benificiary.

You could certainly manage 'Al forced Charlie to give Dasher to Elizabeth for Francis' in one clause in any language with causatives:

Bob-u Charlie-a Dasher-a Elizabeth-a Francis-a w-a-k-a-t-a-k-a-lung-un
Bob-NOM Charlie-ACC Dasher-ACC Elizabeth-ACC Francis-ACC CAUS-3ps-DAT-3ps-BEN-3ps-ACC-3ps-give-PERF

[TomHChappell]

Can you say "Al convinced Bob to force Charlie to give Dasher to Elizabeth for Francis." with just one clause with just one verb in any natlang or conlang?
Al -- instigator
Bob -- middle agent
Charlie -- performer
Dasher -- theme
Elizabeth -- recipient
Francis -- benificiary.

You could certainly manage 'Al forced Charlie to give Dasher to Elizabeth for Francis' in one clause in any language with causatives:

Bob-u Charlie-a Dasher-a Elizabeth-a Francis-a w-a-k-a-t-a-k-a-lung-un
Bob-NOM Charlie-ACC Dasher-ACC Elizabeth-ACC Francis-ACC CAUS-3ps-DAT-3ps-BEN-3ps-ACC-3ps-give-PERF

(1) Thanks.
(2) What language is that?
(3) Why all the ACCs?
(3a). Why isn't the performer (Charlie) in a different case than the theme (Dasher)?
(3b). Why isn't the recipient (Elizabeth) in a dative case, different than the theme (Dasher)?
(3c). Why isn't the beneficiary (Francis) in a benefactive case, different than the theme (Dasher)?
(4) Where's Al? Did you forget about Al? Why aren't Al (instigator) and Charlie (performer) in core cases and Bob (middle agent) in an oblique case?
(5) I'd think the verb might agree with Al and Charlie but not with Bob. Double-causatives usually don't agree with the "middle agent".
(6) If the Beneficiary isn't the speaker, and isn't the addressee, and isn't one of the other participants (one of the agents (Al or Bob or Charlie) or the theme or the recipient); then "version" won't work to make the verb show who the Beneficiary is; instead, you'd have to use a Benefactive Applicative. Can the verb really agree with all three of the Theme (Dasher), the Recipient (Elizabeth), and the Beneficiary (Frances), if it also has to agree with two (or all three) of the Agents (e.g. the Instigator (Al) and the Performer (Charlie))? Wouldn't Applicativization instead have to demote either the Dative or Indirect Object (Elizabeth) or the Accusative or Direct Object (Dasher) to an oblique argument position out of the slot it's in to make room for the Beneficiary as a "new" core argument?
(7) Suppose double-causativization demotes the middle agent to the Dative or Indirect Object position, provided there's room for it there (that is, provided it's not already occupied). Then, if the Beneficiary has to be put in that slot by Dative Applicativization, wouldn't that knock the middle agent out of that slot? Or, if the language allows two Indirect or Secondary Objects, isn't the competition between three arguments, namely the Recipient and the Beneficiary and the Middle Agent, for those two Indirect Object slots, likely to leave one of them without a chair to sit in, that is, forcing it to be demoted to an oblique argument position?

But: Yeah, I think that in a language with double morphological causatives and benevactive applicatives, it probably would be possible to say "Al convinced Bob to fource Charlie to give Dasher to Elizabeth for Frances" in just one clause with just one verb. I just don't think the verb would be likely agree with all six arguments. Do you have a counterexample? I'd love to see it; preferably with a named language and a source as well; nice if the source is on-line.

[Yng]

(2) What language is that?

I made it up based on Sumerian but without, y'know, any of the effort involved in trawling through my grammar of Sumerian. Here's an equivalent in Sumerian, though:

Bub-e Daša-ø Kali-r Lisbe-r Franši-ra ì-na-ni-in-ba-Ø
Bob-ERG Dasher-ABS Charlie-DAT Elizabeth-DAT Francis-DAT CONJ-3ps.ANIM.TERM-3ps.ANIM.DAT-3ps.ANIM.ERG-give.PERF-3ps.ABS
(ignore the random faux-sumericisations, they're just to make it slightly less aesthetically vom-worthy)

The verb agrees with the two agents, the theme and the recipient. It's possible that in later forms of the language (where there were no first-language speakers and all sorts of weird stuff got written) it could've agreed with both datives simply by the addition of an extra dative prefix but I don't know. A simplified, relevant gloss is here:

ìna-ni-in-ba-Ø
DAT-TERM-ERG-give-ABS

The locative-terminative is to a degree interchangeable with the dative, and here the locative-terminative prefix refers back to a noun marked with the dative - 'middle agents' are marked with the dative. My grammar, written by a Professor Foxvog, proposes that the loc-term prefixes were originally a second ergative marked in the verbal chain.

(3) Why all the ACCs?

Oops, left them in. They should've been in their respective cases. However, a language with applicatives (which Sumerian doesn't really have, incidentally) could conceivably treat them as accusatives, right?

(4) Where's Al? Did you forget about Al?

I left Al out because I couldn't accomplish 'persuaded to', but I could demonstrate a tritransitive verb. I thought Sumerian had a distinct benefactive case but it transpires it marks it with the dative after all.

But: Yeah, I think that in a language with double morphological causatives and benevactive applicatives, it probably would be possible to say "Al convinced Bob to force Charlie to give Dasher to Elizabeth for Frances" in just one clause with just one verb. I just don't think the verb would be likely agree with all six arguments. Do you have a counterexample? I'd love to see it; preferably with a named language and a source as well; nice if the source is on-line.

Here is Foxvog's grammar of Sumerian.

[TomHChappell]

Thanks!
Does/did Sumerian have double morphological causatives?
It looks to me like the answer is "no", based on the source you cited.

So, with no double-causativization morphology and no benefactive-applicativization morphology, Sumerian could say
"Bob made Charlie give Dasher to Elizabeth"
in just one clause with just one verb (which would, natch, be tetravalent), but not
"Al convinced Bob to make Charlie give Dasher to Elizabeth for Francis".
Unless I've mis-read.

[Astraios]

but not
"Al convinced Bob to make Charlie give Dasher to Elizabeth for Francis".
Unless I've mis-read.

Well, there are two verbs there in most languages - to "convince" someone to act doesn't strike me as something that would be marked on a verb in any language. Maybe "Al made Bob make Charlie give Dasher to Elizabeth for Francis" could be a single clause, but I don't think "convince" would work.

[Yng]

Yeah, that's what I meant when I said before that Sumerian couldn't manage 'convinced' but could manage high-valency sentences with verb agreement.

but not
"Al convinced Bob to make Charlie give Dasher to Elizabeth for Francis".
Unless I've mis-read.

Well, there are two verbs there in most languages - to "convince" someone to act doesn't strike me as something that would be marked on a verb in any language. Maybe "Al made Bob make Charlie give Dasher to Elizabeth for Francis" could be a single clause, but I don't think "convince" would work.

Yeah I'd agree. Although there may be languages (and I actually wouldn't be surprised if this popped up in post-fluency Sumerian either) if a language allowed causatives ad infinitum, so 'Al made Bob make...'

[ná'oolkiłí]

I found this this recently. It might be of interest.

[TomHChappell]

I found this this recently. It might be of interest.

Want to attend? Let's all go!

Well, there are two verbs there in most languages - to "convince" someone to act doesn't strike me as something that would be marked on a verb in any language. Maybe "Al made Bob make Charlie give Dasher to Elizabeth for Francis" could be a single clause, but I don't think "convince" would work.

Hindi, which allows double morphological causatives, is supposed to have two different morphological causativization processes. The one for the instigator indirectly getting the performer to perform ("convince", "persuade", etc.) is supposed to be a different one from the one for the instigator more directly making the performer act as directed ('make", "force", etc.).
Hindi's double morphological causativization can be doubled only if the two different types of causativization are both involved. Thus "Al convinced Bob to make Charlie ..." or "Al made Bob convince Charlie to ...".
There are also other languages with two different kinds of morphological causativization, one for indirect and a different one for direct.
Such languages, then, do have "convince" marked on the verb, in the sense you meant.
Also, for some languages, direct causativization can be morphological but indirect causativization has to be shown non-morphologically, e.g. syntactically and/or lexically. Those languages don't have "convince" marked on the verb.

English, and probably many other languages as well, has more than one verb for indirect causativization; convince, persuade, induce, seduce, etc. (Of course for English causativization, whether direct or indirect, nearly has to be shown via a two-clause construction.) The fact that most (? is that true of "most"? it's true of English; FAIK it's true of Sumerian (is it?)) languages have a different verb for "A got B to do something" than for "A made B do something" is probably not bull's-eye relevant to the specific question I asked, though it's clearly interesting to the general topic. There are different verbs for direct ("made", "forced") causativization, and for indirect ("got", "persuaded", "tricked", "led") causativization, in English; the mere fact of "different verbs" doesn't seem to be the most important thing.

----------------------------------------------------------------------------

But, anyway;

Thanks for the answer!
Especially, thanks for the Sumerian example!

----------------------------------------------------------------------------

What about Kabardian, Tessarabian (did I spell that right?), and some other languages with, apparently, four grammatical relations? Don't several of them have "version"? Don't several of them have morphological causatives? Do some of them have double-morphological-causativization? Can you morphologcially causativize ditransitives in some of them? Can you morphologically doubly-causativize transitives in some of them? Do any of them have applicativization or "dative applicativization"? If so what happens when you simultaneously causativize and applicativize one and the same verb (especially if the verb is transitive) in some of them?

[ná'oolkiłí]

What about Kabardian, Tessarabian (did I spell that right?), and some other languages with, apparently, four grammatical relations? Don't several of them have "version"? Don't several of them have morphological causatives? Do some of them have double-morphological-causativization? Can you morphologcially causativize ditransitives in some of them? Can you morphologically doubly-causativize transitives in some of them? Do any of them have applicativization or "dative applicativization"? If so what happens when you simultaneously causativize and applicativize one and the same verb (especially if the verb is transitive) in some of them?

I now have the resources to at least attempt answering these questions. All examples are Kabardian, from The Indigenous Languages of the Caucasus, Vol 2: NWC, ed. George Hewitt (the chapter on Kabardian is by John Colarusso); Colarusso's Kabardian; and Matasović's A Short Grammar of East Circassian (Kabardian). Excuse the discrepancies in the glosses; Matasović and Colarusso have somewhat different approaches.

Polypersonal Inflection and Transitivity Overview
One-argument intransitive:
/tɬ'ə-ha-r ma-a-kʷ'a-ha/
man-PL-ABS 3.INTRANS.PRES-PRES-come.at-PL
"The men are coming"

Two-argument intransitive (intransitive with indirect object):
/tɬ'əhar sa Ø-q'ə-sa-a-ptɬ+a-ha/
— 1S 3.ABS-HORIZON.OF.INTEREST-1S.DAT-PRES-look+at-PL
"The men are looking at me"
(that "horizon of interest" morpheme basically doesn't mean anything, btw)

Two-argument transitive:
/ɬ'ə-ha-m sa sə-q'a-j-ha-ə-ɬaɣʷə-ɣa-ɕ/
man-PL-ERG 1S 1S.ABS-HOR-3.ERG-PL-NON.PRES-see-PAST-AFFIRMATIVE
"The men saw me"

Three-argument causative of an intransitive:
/sa wa a-bə wə-ja-s-ɣa-pɬə+a-ɣa-ɕ/
— 2S 3-DAT 2S.ABS-3.DAT-1.ERG-CAUSATIVE-look+at-PAST-AFF
"I made you look at it ~ I showed it to you"

Three-argument ditransitive:
/abə wa mə-txʲə-ha-r Ø-q'-w-aj-ə-tə-ɣa-ha-ɕ/
— — this-book-PL-ABS 3.ABS-HOR-2S.DAT-3.ERG-NON.PR-givePAST-PL-AFF
"He gave these books to you"

Three-argument causative of a transitive:
/sa abə wa wə-q'-ja-s-ga-a-ɬaɣʷə-ɣa-ɕ/
— — — 2S.ABS-HOR-3.DAT-1S.ERG-CAUS-CONN-see-PAST-AFF
"I made him see you ~ I showed you to him"

Four-argument causative of a ditransitive:
/sa a-bə-ha-m wa a-r Ø-q'-wa-j-ha-a-s-ɣa-a-tə-ɣa-ɕ/
— 3-DAT-PL-DAT — 3-ABS 3.ABS-HOR-2S.DAT-3-PL-DAT-1S.ERG-CAUS-CONN-givePAST-AFF

Version and Applicatives
Kabardian does have version, but not quite in the same way as South Caucasian languages, as you might have in mind; it only expresses benefactive, adversitive (curiously with the same morphemes), and malefactive arguments. The version morphemes are placed right after the personal prefix for whose benefit/detriment the action is performed.

/p-xʷa-s-tx-aː-ɕ/
2S-BEN.VERSION-1S-writePRETERITE-AFF
"I wrote for you"

/s-Ø-ha-xʷa-psaɬa-aɣ-ɕ/
1S.ABS-3-PL-ADV.VER-speak-PAST-AFF

/w-aː-f'ə-da-kʷ'-aː-ɕ/
2S-3P-MAL.VER-COMITATIVE.APPLICATIVEgo-PRET-AFF
"You went with them against their will"

Matasović distinguishes benefactive/malefactive operations ("version) from comitative ones ("applicatives"), but then also refers to all of them as applicatives too. In any case, here are some examples of comitative applicatives:

/sə-Ø-da-kʷ'-aː-ɕ/
1S.ABS-3.DAT-COM.AP-goPRET-AFF
"I went with him"
cf
/səkʷ'aːɕ/
"I went"

This works with transitives too:
/Ø-b-də-z-aw-ʃx/
3S.ABS-2S.DAT-COM.APL-1S.ERG-PRES-eat
"I am eating this with you"

Causatives
All Kabardian verbs can be causitive. See above for a few simple examples.

Double causatives:
/jə-r-jə-ɣa-ɣaː-va-r-jə/
3.ABS-3.DAT-3.ERG-CAUS-CAUS-boil-PRES-and
"and he causes her to cause it to boil ~ and he causes her to cook it"
Where the root van "boil" is causitivized to ɣaːvan "cause to boil ~ cook" and causitivized again. Compare ʑan "burn (intrans)" → ɣaːʑan "burn (trans)" → ɣaɣaːʑan "cause to burn".


Applicatives and causitives can be combined. I can't find an example with a causative transitive verb and an applicative, but I imagine it's much the same:
/sjə dəɕa-r Ø-q'ə-p-xʷ-aw-ɣaː-na/
my gold-NOM 3.ABS-HOR-2S.DAT-BEN.VER-PRES-CAUS-remain
"I am making my cold remain for you ~ I am leaving you my gold"

Other fun stuff
NWC verbs are absurd (probably invented by sadistic Vasco-Caucasian conlangers many thousands of years ago). Here are some relavant tricks they can do.

Valence reduction:
Often you can simply omit person morphemes to reduce valence. For example, here we reduce a trivalent verb to monovalence (with a suppleting (?) stem).
/sə-tə+a-ʒə-aɣ-ɕ/
1S.ABS-give+INTRANS-back-PAST-AFF
"I give again"

Incorporation:
I'm not sure how productive this is, or how it affects valency.
/Ø-w-ʔa-ɕ'ə-s-ɬ+ħa-ɣa-ɕ/
3.ABS-your-hand-INST-1S-lie+enter(=active)-PAST-AFF
"I laid it in your hand"

Positional Prefixes:
Some of these seem a lot like applicatives.
/sə-q'a-p-taj-fa/
1S.ABS-HOR-2S.DAT(?)-on-fall
"I fall on you"

/wə-q'a-t-taj-sə-xəʒ/
2S.ABS-HOR-1P.DAT-"on"-1S.ERG-lift
"I lift you from us"

Pp 39-45ish in the above PDF gives a nice description of valence/transitivity in general.