Valdyas is a fictional country in a fictional world. Ilaini is an invented language. I invented it for my own enjoyment and that of anybody else who enjoys it. It has no scholarly pretensions. Please don't make a fool of yourself by using my language pages for your linguistics case-work.

We'll start with an unmarked instance of a prince turning into a frog:

val.nutea        custen     lea disut
king.child-acc-s frog-ill-s IMP change-PRT-3s

"The prince turned into a frog"

This implies that the prince was turned into a frog by person or persons unknown, or that his ranification took place spontaneously, without an apparent outside agency. We can also translate it as "The prince was turned into a frog", as long as we don't assume a specific agent. Note that valnutea is in the accusative, that is, the case used for the direct object.

Let's look at some possible agents. For one thing, the prince could cast the spell on himself.

val.nute         custen     le  disut
king.child-nom-s frog-ill-s RFL change-PRT-3s

"The prince turned himself into a frog".

Here, le is the object of disut "changed", referring back to the subject of the main clause, "the prince". Nothing impersonal about this - it's clear who is turning whom into a frog, namely the prince himself, who is in the nominative (subject case). We can give it some emphasis:

val.nute         le.arne  custen     le  disut
king.child-nom-s him.self frog-ill-s RFL change-PRT-3s

"The prince turned himself into a frog".

This can also mean (and usually does mean) "The prince turned into a frog all by himself", making learne a modifier of valnute.

We need le as well as learne, because the verb doesn't get any less reflexive when we use the emphatic form, and learne, though it's an object pronoun, can't take the place of the reflexive pronoun. The reflexive pronoun is, after all, primarily a marker for "this object is the same as the subject of the main clause".

For non-third-person subjects, we use only arne:

arne custen     le  disun
self frog-ill-s RFL change-PRT-1s

"I turned myself into a frog (on my own)".

Or more simply, of course:

custen     le  disun
frog-ill-s RFL change-PRT-1s

"I turned myself into a frog"

or even:

i   custen     lea disut
1sO frog-ill-s IMP change-PRT-3s

"I turned into a frog", "I was turned into a frog"

Now, let's see what happens when other people turn princes into frogs. First, a specific person:

orie           val.nutea        custen     disut
magician-nom-s king.child-acc-s frog-ill-s change-PRT-3s

"The sorcerer turned the prince into a frog"

This is perfectly straightforward: a specified animate subject. Let's have an unspecified animate subject and see if it's any different:

salea     val.nutea        custen     disut
someone-s king.child-acc-s frog-ill-s change-PRT-3s

"Somebody turned the prince into a frog"

Clearly, the subject doesn't have to be specific. If there's an inanimate subject where one would expect an animate one, however, the impersonal construction comes back with the logical subject as an object:

or.sean           val.nutea        custen     lea disut
magic.thing-acc-s king.child-acc-s frog-ill-s IMP change-PRT-3s

"The spell turned the prince into a frog"

In this case, both the spell and the prince are in the accusative (direct object) case. This sentence is perhaps better translated by "The prince was turned into a frog by a spell" making it clearer that the spell was only the means. The spell as grammatical subject would imply that it somehow cast itself:

? or.sen            val.nutea        custen     disut
  magic.thing-nom-s king.child-acc-s frog-ill-s change-PRT-3s

"The spell turned the prince into a frog"

It's more usual to do this with the instrumental:

or.sien           val.nutea        custen     lea disut
magic.thing-ins-s king.child-acc-s frog-ill-s IMP change-PRT-3s

"The prince was turned into a frog with a spell"

which can also have a real subject:

orie           or.sien           val.nutea        custen     disut
magician-nom-s magic.thing-ins-s king.child-acc-s frog-ill-s change-PRT-3s

"The sorcerer turned the prince into a frog with a spell"

An unspecified inanimate subject also causes the impersonal construction:

sali      val.nutea        custen     lea disut
something king.child-acc-s frog-ill-s IMP change-PRT-3s

"Something turned the prince into a frog"

We can't tell from the form of sali whether it's in the object case, but we can safely assume that it is, because a specified inanimate subject (such as "the spell") behaves the same.

Now, why don't we just call the pseudopassive impersonal construction "passive" and get it over with? For one thing, it requires the pronoun lea. This might just be a dummy, if it didn't always take a singular verb, even if any or all of the objects are plural:

val.nutena       custenen   lea disut
king.child-acc-p frog-ill-p IMP change-PRT-3s

"The princes turned into frogs"

or.sean           val.nutena       custenen   lea disut
magic.thing-acc-s king.child-acc-p frog-ill-p IMP change-PRT-3s

"The spell turned the princes into frogs"

or.senan          val.nutena       custenen   lea disut
magic.thing-acc-p king.child-acc-p frog-ill-p IMP change-PRT-3s

"The spells turned the princes into frogs"

or.senan          val.nutea        custen     lea disut
magic.thing-acc-p king.child-acc-s frog-ill-s IMP change-PRT-3s

"The spells turned the prince into a frog"

Now that we have at least seventeen princes and three people who may or may not be princes, all turned into frogs, let's see if we can get them back...

custenan   a.lea nusen rastenen     lea disat
frog-acc-p all   back  person-ill-p IMP change-PRS-3s

"All frogs turn back into people"

or, perhaps:

custenan   a.lea nusen rastenen     disayn
frog-acc-p all   back  person-ill-p change-PRS-1p

"We turn all frogs back into people"