Agent-based grammatical gender?

[Ashtagon]

This is more of a random musing and wondering if it could work.

I recall reading somewhere that many palaeo-languages had noun classes not based on gender (as in many modern Indo-European languages), but on level of agency. That is, "gods" would be the first noun class, adult humans (or possibly just men, depending on society customs) second, discrete animals and children third, dynamic weather patterns next, then plants, and so on down the line, with the lowest being simple mass nouns such as "wood" or "food".

Under this idea, a lower-order noun can't be the subject of a sentence that has a higher-order noun. That is, it would be ungrammatical (or at least, considered a really weird sentence construction) to say "the child spoke to the man"; instead, you'd have to switch to passive voice, "the man was spoken to by the child", or even anti-passive, "it was the man to whom the child spoke".

Has anyone tried anything like this? How well did it work?

[Travis B.]

In direct-inverse languages, core arguments of transitive verbs that have higher personhood/animacy/topicality than the other core arguments are by default the agents, and to specify the opposite one must specify that the verb is inverse explicitly. Also, IIRC, in some languages, inanimate nouns cannot be agents, and to make them agents one must instead use the passive, marking them with an oblique case. It would not be a stretch to say that the higher personhood/animacy/topicality core argument of a transitive verb is always the agent, and to say otherwise one must use the passive; while I cannot name a particular language here, I would not be surprised if there were languages like this (and even if there were not, it would not be something unrealistic to put in one's conlang).

[Ares Land]

This is more of a random musing and wondering if it could work.

I recall reading somewhere that many palaeo-languages had noun classes not based on gender (as in many modern Indo-European languages), but on level of agency.

Not quite as such, but there are examples that look a bit like that.

It's assumed very early PIE had two classes: animate or inanimate. Inanimates had no distinct accusatives. There were different words for certain concepts, such as water and fire, depending on whether they were conceived as animate or inanimate. (And as it happens, neuter nouns ended up with identical nom/acc cases in descendant languages.)

The Algonquian languages also distinguish animate and inanimate language. There's an animacy hierarchy, in which inanimates come last (the rest of the hierarchy isn't based on status, but on grammatical person and obviation). The verb takes different markings when a less inanimate participant acts on a more animate one.

As for conlangs, you can check out Axunashin which has a hierarchy-based case systems that's very similar to the one you describe. So it's definitely doable!

[Travis B.]

This is more of a random musing and wondering if it could work.

I recall reading somewhere that many palaeo-languages had noun classes not based on gender (as in many modern Indo-European languages), but on level of agency.

Not quite as such, but there are examples that look a bit like that.

It's assumed very early PIE had two classes: animate or inanimate. Inanimates had no distinct accusatives. There were different words for certain concepts, such as water and fire, depending on whether they were conceived as animate or inanimate. (And as it happens, neuter nouns ended up with identical nom/acc cases in descendant languages.)

The Algonquian languages also distinguish animate and inanimate language. There's an animacy hierarchy, in which inanimates come last (the rest of the hierarchy isn't based on status, but on grammatical person and obviation). The verb takes different markings when a less inanimate participant acts on a more animate one.

As for conlangs, you can check out Axunashin which has a hierarchy-based case systems that's very similar to the one you describe. So it's definitely doable!

I assume you mean "more inanimate" or "less animate" there.

I should note that direct-inverse need not require morphological obviation of nouns, which is really a specifically Algonquian thing.

[Ares Land]

I assume you mean "more inanimate" or "less animate" there.

Oh, yes, that's what I meant

[bradrn]

Under this idea, a lower-order noun can't be the subject of a sentence that has a higher-order noun. That is, it would be ungrammatical (or at least, considered a really weird sentence construction) to say "the child spoke to the man"; instead, you'd have to switch to passive voice, "the man was spoken to by the child", or even anti-passive, "it was the man to whom the child spoke".

From a quick skim through my usual sources¹ on direct-inverse systems, it would appear that Tanoan languages, especially Tiwa languages, have something extremely similar, if not identical, to what you describe (Zúñiga 2006): direct sentences are in the active voice, whereas the arguments of inverse sentences need to be switched around using something which looks very much like a passive voice. For instance, from Southern Tiwa:

Seuan-ide

man-SG

ti-mų-ban.

(1SG→3)-see-PST

I saw the man

Seuan-ide-be

man-SG-OBL

ti-mų-che-ban.

1SG-see-PASS-PST

The man saw me

(Oh, and “it was the man to whom the child spoke” is not an antipassive; it’s a cleft construction. An antipassive would be something more like “the child was speaking (to the man)” — the transitive sentence becomes an intransitive sentence where the former transitive subject is the new intransitive subject, a distinction which for obvious reasons is hard to render in English.)

_______________
¹ Principally Zúñiga’s Deixis and Alignment: Inverse systems in indegenous languages of the Americas (2006), ed. Gildea & Queixalós’s Ergativity in Amazonia (2010), and Jacques & Antolov’s Direct/Inverse Systems (2014)