Zompist Bboard Again

Both nouns would be inflected with the same numbered apposition inflection, which can be expressed in the gloss as "APP<n>".
For example, Apposition of any noun and the word for 'agent/doer' replaces the ergative case and it might be expressed in the gloss as "<noun>-APP1 one.who.does-APP1"

I remember a language that had the same idea but had English words with special punctuation instead. Something like "(cat . agent) (mouse . patient) (death . result)" if I remember the punctuation correctly. (It also didn't have verbs).
Can anyone tell me the name of it?

LISP?

Natural languages tend to minimise the amount of phonologically overt material they use for this sort of thing, so I wouldn't want to use this idea in a realistic conlang (of course, non-realistic conlangs are fine too). Even you got rid of the "numbering", which is redundant if you mark the same thing by word order, I don't see this happening unless the words for "agent", "patient" etc. are very short (e.g. monosyllables). And even then it would be fairly unusual to overtly mark both the agent case and the patient case in this way. If such a system did arise you could well imagine the words for "agent", "patient" etc. undergoing phonological erosion - becoming less like lexical items in their own right and more like case particles.

The other thing to bear in mind is that words like "agent", "subject" etc. are quite theoretical terms that languages don't tend to have unless they've developed a proper tradition of grammatical description. So you wouldn't normally expect them to be so integrated into the linguistic system itself.

I can't prove it, but language with subject marking usually marks it from deriving animate noun from inanimate noun.

This isn't "replacing case morphology", it's just "making the nominative case homophonous with the word 'agent'".

You can certainly have case marking material with a transparent meaning— Mandarin has them. They derive from verbs, so they're called coverbs, and they normally also retain their verbal meaning. E.g. bǎ 'acc.' < 'grasp'; dào 'toward' < 'reach'; gěi 'for' < 'give'.

One thing to be careful about: semantic roles like 'agent' are distinct from cases, like 'ergative'. Often you can get away with conflating these ideas, but not once you have transformations that modify case assignments. E.g. Mary in "Mary was given the book" is nominative, but her semantic role is beneficiary, and in "They made her dance the fandango", she is the agent, but syntactically the object.

The idea of a language that marks the semantic roles only instead of cases is an interesting one, but I'm not sure just how workable it is in practice. It seems like it would be very hard not to end up with either a case system or something approaching Austronesian alignment.

One of my earliest conlangs (well, sketches of what can become a conlang) has a case for agents (and forces) when the agent is not the accusative, so both indirect objects and the subject can be, and neither have to be, in the agentive case. Or at least that's what I think I meant - I understood back then that there is a difference between thematic and syntactic roles back then, but I couldn't at the time grasp what the difference actually is. I definitely wasn't trying to do make a simple nominative-dative merger or something like that.

So if I understand correctly, this idea would be something like “Franklin the agent kissed Thomas the patient.” Assuming your language has tidy terms for “agent” and “patient,” you’re just describing their grammatical role like stage directions. If these are indeed acting as ersatz noun cases, then they should probably having some liberalizing effect on word order: “Thomas the patient kissed Franklin the agent, and was accused of assault” would make it clear that Franklin was accused, not Thomas. You could end up with something like this:

"Arkansas, an elative point of origin, the local authorities, here serving as a cause or reason, the dead of night, an essive phrase pinpointing the time of the action, a small chartered airplane, an instrument, Franklin, the agent of this sentence, fled."

Curlyjimsam wrote: ↑Mon Feb 18, 2019 5:55 am Natural languages tend to minimise the amount of phonologically overt material they use for this sort of thing, so I wouldn't want to use this idea in a realistic conlang (of course, non-realistic conlangs are fine too). Even you got rid of the "numbering", which is redundant if you mark the same thing by word order, I don't see this happening unless the words for "agent", "patient" etc. are very short (e.g. monosyllables). And even then it would be fairly unusual to overtly mark both the agent case and the patient case in this way. If such a system did arise you could well imagine the words for "agent", "patient" etc. undergoing phonological erosion - becoming less like lexical items in their own right and more like case particles.

I was thinking that might happen.

Moose-tache wrote: ↑Wed Feb 20, 2019 2:07 am So if I understand correctly, this idea would be something like “Franklin the agent kissed Thomas the patient.” Assuming your language has tidy terms for “agent” and “patient,” you’re just describing their grammatical role like stage directions. If these are indeed acting as ersatz noun cases, then they should probably having some liberalizing effect on word order: “Thomas the patient kissed Franklin the agent, and was accused of assault” would make it clear that Franklin was accused, not Thomas. You could end up with something like this:

"Arkansas, an elative point of origin, the local authorities, here serving as a cause or reason, the dead of night, an essive phrase pinpointing the time of the action, a small chartered airplane, an instrument, Franklin, the agent of this sentence, fled."

That would be a way of directly translating it, yes.

akam chinjir wrote: ↑Mon Feb 18, 2019 4:56 amLISP?

No, I don't think so.

Salmoneus wrote: ↑Mon Feb 18, 2019 8:44 am This isn't "replacing case morphology", it's just "making the nominative case homophonous with the word 'agent'".

That's just another way of phrasing or looking at the same thing.

Am I missing something, or are you just reinventing active-stative languages?

So it is essentially "fluid-s" with overt marking on all argument values? It doesn't make sense to me that all should be marked, or that the markings are homophonous with grammatical forms.

I doubt though that in practice that fluid-s languages' cases have a 1:1 correspondence with semantic roles; it is generally possible to infer the semantic roles by other means and those cases can be used for various other semantic functions.

WeepingElf wrote: ↑Thu Feb 21, 2019 11:20 am Am I missing something, or are you just reinventing active-stative languages?

mèþru wrote: So it is essentially "fluid-s" with overt marking on all argument values?

I suppose so but the new part is how the 'cases' are formed. Also, since any noun can be used in that way, such roles can be invented on-the-fly.

I once made a conlang that was based on semantic roles "instead of cases" or something like that, but since then I've realized that that's just dumb and that I should just rename these semantic roles as cases, so that agent=nominative, patient=accusative, goal=lative, etc. The interesting thing is that it would use these cases very literally, and that there would be several different things that would all be covered by e.g. the accusative case in natlangs. This was basically a triliteral language, so these roles were all just marked with different vowel patterns.

(One interesting thing about this conlang was that you could have words with zero case, which means that the vowel pattern was just ∅-∅-∅. So on the phonemic level one would have a vowelless word, but then phonotactics would come and add epenthetic vowels. The zero case was used for e.g. book titles consisting of just one word, labels, and in metalinguistic context when talking about a word.)

As I understand it, cases are basically just the practical realisation of semantic roles, kind of how phones are the realisation of phonemes. If your language is completely regular, each role should always get the same case marking.

As you probably know, one of the basic questions when describing a case system is how a single argument is handled compared to the two arguments of a transitive clause – if we have the two cases/roles A and P in "Alice licks Paul", which of them do we use in "Bob sleeps" or "Bob runs"? With case markings in natural languages, there are several answers:
1. A (accusative alignment)
2. P (ergative alignment)
3. a third one (tripartite alignment)
4. A for "runs" and P for "sleeps" (active-stative alignment)

We can do the same kind of division for semantic roles. Curiously, it seems that although the first is the most common in terms of morphosyntax, it gets shunned when thinking in terms of semantics.

Some argue that option 3 is the only clear way to describe things. But you might also say that if the set of semantic roles is to be meaningful, we should strive to make it minimal; if there are two roles that never occur in the same clause, you have too many roles. But that really depends on what you want to do with those roles.


My main conlang happens to be an unnaturalistic, regular language, and so I try to make the case markings correspond to semantic roles. Each noun phrase has a particular type of word to mark its case/role; you can think of those as case particles, role labels, or prepositions. This means that there is no need for inflection on the noun-like words, nor a fixed word order.

This also allows another neat trick, which I find very useful and obvious, but which apparently isn't attested in natlangs. Instead of using verb inflections to form for example passive voice, we can just drop either argument. So for a simplified example:
AGENT Alice lick PATIENT Paul = "Alice licks Paul"
AGENT Alice lick = "Alice licks (someone)"
PATIENT Paul lick = "Paul gets licked (by someone)"

It doesn't quite fit any of the normal alignment systems, since every verb can be thought of as optionally transitive.


Hyolobrika's idea of making any noun able to function as a role label is an interesting thought. But in practice, words like "agent" and "patient" would be overwhelmingly more common as role labels than anything else, and words like "guitar" and "parole" would be overwhelmingly more common as regular nouns. So treating both as the same class of word would really only be confusing. It sort of works as a fun experiment, but not as a viable human language, naturalistic or not.

A less extreme idea would be to have the common labels as ordinary case particles, but add a construction for special cases, "with X acting as Y". Like so:
AGENT Bob remove PATIENT lightbulb SPECIAL ladder pipe = "Bob removed the lightbulb using the pipe as a ladder"
AGENT Bob make PATIENT portrait SPECIAL model Alice SPECIAL paint blood = "Bob made a portrait of Alice painted in blood"

Chuma wrote: ↑Thu Feb 28, 2019 3:00 pm As I understand it, cases are basically just the practical realisation of semantic roles, kind of how phones are the realisation of phonemes. If your language is completely regular, each role should always get the same case marking.

Kind of, but it would be more accurate to say that semantic roles apply to deep structure, while case applies to surface structure.

There are all sorts of ways case and semantic role may be different:

* passive
* dative movement
* applicatives
* raising ("I want him to go")
* politeness (e.g. Spanish's personal infinitive, in form equivalent to the dative)
* other odd reversals of valence (Sp. me gusta María vs. Pt. eu gosto Maria)
* other arbitrary stuff (e.g. Russian acc > gen in negative sentences)
* whatever is going on in English "him and I"
* case attraction

(Also, FWIW, the original idea of semantic roles was that they were universal. That's speculative, but if you accept it, individual languages may have other odd case divergences.)

I like Chuma's idea