Conlang Random Thread

[bradrn]

[...] the issue now is if you have an intransitive verb:

yes-xe

We exist

Is this language nom-acc, since this is the experiencer having the same affix as the agent? Or is it erg-acc, since the experiencer is the same as the patient? What if the rest of the syntax doesn't help e.g. if it's SOV?

I've just realised that I left out the most important part of this: noun cases! If the language has noun cases, then they will tell you whether it's nom-acc or erg-abs. On the other hand, if there are no cases, then the question doesn't even make sense, since (as far as I know) you can't have morphosyntactic alignment without noun cases. Feel free to disregard the above alignment question.

___________________

EDIT: The above reasoning is WRONG. I've just realised you can definitely have alignment without noun cases - it's possible to just do it through syntax. I suppose that it's always possible to figure out some sort of alignment through the sorts of morphological/syntactic features a language has, whether it's noun cases, syntax, presence of passive/antipassive etc. Still, feel free to disregard the above question - I feel that I've answered it well enough in this paragraph.
___________________

Further hypothetical question: Let's arbitrarily say that our language is nom-acc with noun cases, and the system above is used for agreement. Then let's say we have some sound change which applies to the end of words. Now our agent and patient forms are no longer identical: the sound changes applied to the patient but not the agent, because the former was at the end of the word but the latter wasn't. But the sound changes also applied to the experiencer as well (it's also at the end of the word), making it identical to the patient! So now we have the patient/experiencer forms the same, while simultaneously being different to the agent form - in other words, erg-abs alignment in the polypersonal agreement. But we still have nom-acc on the nouns! Now luckily I avoided this situation by having no noun cases, but how could this horrible situation be resolved?

[mèþru]

It doesn't need to be resolved; there are plenty of natlangs that do stuff like that.

[Qwynegold]

So then I thought that maybe I can use relative clauses instead. But the way Liu handles relative clauses is already so complicated, it just fries my brain when I try to think about how to translate "I want to work on Liu again, but I'm not sure what else to do to get the grammar more complete." Do you think it's a good idea to somehow use relative clauses instead of pronouns? Like something like "...that* which do to get the grammar more complete".

I'm not sure I understand your terminology, but it sounds as if you're asking whether to allow headless relative clauses. (Both your examples have relative clauses, and whether to allow headless ones doesn't have anything to do with whether you use relative pronouns to form them, as far as I know.)

Sorry, I'm so confused about this myself, and I get more and more confused the more I think about the issue. Let me ask another question, and maybe I'll be able to work something out myself; is the word what in the sentence "I want to work on Liu again, but I'm not sure what else to do to get the grammar more complete" a relative pronoun or not? Does anyone want to split this sentence into clauses for me?

[mèþru]

http://twitter.com/PlymUni/status/1098292451536658434
Apparently DJP is branching into writing librettos for opera

[bradrn]

It doesn't need to be resolved; there are plenty of natlangs that do stuff like that.

Really? I'm surprised. What languages have different alignment in their noun morphology and verb agreement?

[akam chinjir]

Sorry, I'm so confused about this myself, and I get more and more confused the more I think about the issue. Let me ask another question, and maybe I'll be able to work something out myself; is the word what in the sentence "I want to work on Liu again, but I'm not sure what else to do to get the grammar more complete" a relative pronoun or not? Does anyone want to split this sentence into clauses for me?

I'd split it into clauses like this:

[I want to work on Liu again] [but I'm not sure [what else to do [to get the grammar more complete]]]

Other than that---I made a mistake in my last comment, the "what" clause isn't a relative clause at all, it's an interrogative complement or content clause. And that means that "what" isn't a relative pronoun.

This is a tricky bit of grammar, so I'll go over it a bit, if only to convince myself that I understand it. (I'm drawing a lot on The Cambridge Grammar of the English Language, especially 1068--1073.)

You do get phrases apparently headed by "what" that function pretty much like noun phrases (or DPs, if you prefer) that include a relative clause. These are the constructions that I referred to as headless relative clauses. Here's an example:

I'll eat what you give me

"Eat" can't take a clausal complement; "what you give me" is just a noun phrase. Maybe "what" is a relative pronoun, I'm not sure (the Cambridge Grammar doesn't think so). You can replace it with "that which," suggesting it has the role both of the noun head and of the relative pronoun.

These phrases differ from complement clauses in a few ways. In a complement clause you can't replace the "what" with "that which" or "whatever":

*I'm not sure that which to do
*I'm not sure whatever to do

The "what" can also pied-pipe a preposition in a complement clause but not in a relative clause:

I'm not sure with what to cut the potatoes
*You can use with what you cut the carrots

Occasionally you can get ambiguities:

What you wrote is not clear

(Example stolen from the Cambridge Grammar.) Interpreted with "what you wrote" as a complement clause, this sentence is confessing ignorance; interpreted with "what you wrote" as a relative clause, it's commenting on your writing.

One thing that makes this tricky in English is that we use interrogative pronouns as relative pronouns, and they move to the front in both interrotative and relative clauses (and embedded interrogative clauses don't even require subject inversion and do-support). So the two constructions look very similar.

Making relative clauses this way is pretty rare, though.

One problem I have is that I don't know much about how various languages handle the cases where English uses an embedded interrogative clause---often it would make semantic sense to do somethnig with a relative clause (e.g., I don't know the thing that I should do). Maybe that's common?

[akam chinjir]

Further hypothetical question: Let's arbitrarily say that our language is nom-acc with noun cases, and the system above is used for agreement. Then let's say we have some sound change which applies to the end of words. Now our agent and patient forms are no longer identical: the sound changes applied to the patient but not the agent, because the former was at the end of the word but the latter wasn't. But the sound changes also applied to the experiencer as well (it's also at the end of the word), making it identical to the patient! So now we have the patient/experiencer forms the same, while simultaneously being different to the agent form - in other words, erg-abs alignment in the polypersonal agreement. But we still have nom-acc on the nouns! Now luckily I avoided this situation by having no noun cases, but how could this horrible situation be resolved?

I'll copy a table from Woolford, Ergative agreement systems:

	nom-acc agreement	agr with both erg and abs	agr with abs only	agr with erg only
unequivocal nom-acc case	English	unattested	unattested	unattested
morphologically unmarked case	Swahili	Jacaltec	unattested	unattested
unequivocal ergative case	Warlpiri	Kabardian	Hindi	unattested

So it sounds like the sort of system you're talking about---with nom/acc case-marking but ergative agreement---is unattested. I suspect that part of the reason is that it's rare to have agreement with an argument that's marked for case, so if you've got accusative case on objects, you're unlikely to have agreement with objects.

One other possible issue with your set-up is that the obvious way to get that agreement pattern is with non-case-marked pronominal clitics, when presumably you want to end up with case-marked free pronouns.

...which doesn't sound like it should make it impossible (though it's always possible there are deep reasons why these things can't go together).

[Qwynegold]

Sorry, I'm so confused about this myself, and I get more and more confused the more I think about the issue. Let me ask another question, and maybe I'll be able to work something out myself; is the word what in the sentence "I want to work on Liu again, but I'm not sure what else to do to get the grammar more complete" a relative pronoun or not? Does anyone want to split this sentence into clauses for me?

I'd split it into clauses like this:

[I want to work on Liu again] [but I'm not sure [what else to do [to get the grammar more complete]]]

Other than that---I made a mistake in my last comment, the "what" clause isn't a relative clause at all, it's an interrogative complement or content clause. And that means that "what" isn't a relative pronoun.

I thought that too. >__< This is why it was so hard!

I'll eat what you give me

"Eat" can't take a clausal complement; "what you give me" is just a noun phrase. Maybe "what" is a relative pronoun, I'm not sure (the Cambridge Grammar doesn't think so). You can replace it with "that which," suggesting it has the role both of the noun head and of the relative pronoun.

I think it's interesting that this what is kinda the object of two different verbs at the same.

One problem I have is that I don't know much about how various languages handle the cases where English uses an embedded interrogative clause---often it would make semantic sense to do somethnig with a relative clause (e.g., I don't know the thing that I should do). Maybe that's common?

Yeah, I was trying to compare with other languages, but Swedish, Finnish and Japanese all do the exact same thing as English. Or well, not exactly, but they all use a word for what like this.

Jag vet inte vad jag ska göra.
1SG know NEG what 1SG FUT do

En tiedä mitä tehdä.
e-n tiedä mi-tä tehdä
NEG-1SG know what-PTV do

Nani wo sureba ii ka wakaranai.
nani wo sur-eba ii ka wakar-anai
what OBJ do-HYP good Q* understand-NEG.PRS.FAM
*This is normally a question particle, but it's also used for some reason in certain cases when you use question words in non-questions.

Anyway, thank you. It has become clear how what strategy I should use for translating this sentence. And I will probably use the word for what. The only problem is what do I call this what that is kinda like a relative pronoun but not. So far the rule in my grammar has been that question words are only used in questions, so I should mention something about when that's not actually the case.

[bradrn]

Further hypothetical question: Let's arbitrarily say that our language is nom-acc with noun cases, and the system above is used for agreement. Then let's say we have some sound change which applies to the end of words. Now our agent and patient forms are no longer identical: the sound changes applied to the patient but not the agent, because the former was at the end of the word but the latter wasn't. But the sound changes also applied to the experiencer as well (it's also at the end of the word), making it identical to the patient! So now we have the patient/experiencer forms the same, while simultaneously being different to the agent form - in other words, erg-abs alignment in the polypersonal agreement. But we still have nom-acc on the nouns! Now luckily I avoided this situation by having no noun cases, but how could this horrible situation be resolved?

I'll copy a table from Woolford, Ergative agreement systems:

	nom-acc agreement	agr with both erg and abs	agr with abs only	agr with erg only
unequivocal nom-acc case	English	unattested	unattested	unattested
morphologically unmarked case	Swahili	Jacaltec	unattested	unattested
unequivocal ergative case	Warlpiri	Kabardian	Hindi	unattested

So it sounds like the sort of system you're talking about---with nom/acc case-marking but ergative agreement---is unattested. I suspect that part of the reason is that it's rare to have agreement with an argument that's marked for case, so if you've got accusative case on objects, you're unlikely to have agreement with objects.

Yes, that makes a lot of sense. But what I'm asking is: what would happen if a language did end up in this situation? Would the system shift in some way or would it remain as is?

One other possible issue with your set-up is that the obvious way to get that agreement pattern is with non-case-marked pronominal clitics, when presumably you want to end up with case-marked free pronouns.

...which doesn't sound like it should make it impossible (though it's always possible there are deep reasons why these things can't go together).

Good observation; that would imply that - luckily - this system couldn't even evolve, which would save a lot of trouble.

_______________

I notice that part of my original question still hasn't been answered:

Would it be realistic to have a polypersonal agreement system where the same affixes are used for both agent and patient?

In a way, this was the most important part of my question (even if the alignment discussion is much more interesting), since this is the situation I put into my protolanguage.

[Kuchigakatai]

You do get phrases apparently headed by "what" that function pretty much like noun phrases (or DPs, if you prefer) that include a relative clause. These are the constructions that I referred to as headless relative clauses. Here's an example:

I'll eat what you give me

"Eat" can't take a clausal complement; "what you give me" is just a noun phrase. Maybe "what" is a relative pronoun, I'm not sure (the Cambridge Grammar doesn't think so). You can replace it with "that which," suggesting it has the role both of the noun head and of the relative pronoun.

Why does the Cambridge Grammar suggest that this is not a headless relative clause exactly? Not only can "what" be replaced by "that which" in this case, there's also not much of an interrogative force. For what it's worth, in Spanish we use what is clearly a relative clause here:

Comeré lo que me dés.
eat-1S.FUT ART.NEUT REL 1S.DAT give-2S.PRES.SUBJ

Occasionally you can get ambiguities:

What you wrote is not clear

(Example stolen from the Cambridge Grammar.) Interpreted with "what you wrote" as a complement clause, this sentence is confessing ignorance; interpreted with "what you wrote" as a relative clause, it's commenting on your writing.

I find this example very interesting, because my first impression was that it's the word "clear" that is semantically ambiguous, but if I look at Spanish I do see something syntactic going on. They are different sentences.

Lo que escribiste no está claro.
ART.NEUT REL write-2S.PRET not is-3S.PRES clear-SG.M
No está claro lo que escribiste.
not is-3S.PRES clear-SG.M ART.NEUT REL write-2S.PRET
(comment on your writing; relative clause; subject-verb order preferred)

Qué escribiste no está claro.
what write-2S.PRET not is-3S.PRES clear-SG.M
No está claro qué escribiste.
not is-3S.PRES clear-SG.M what write-2S.PRET
(ignorance of what was written; interrogative complement; verb-subject order preferred)

Also, I feel that in English there would generally be a difference in pronunciation between the two as well. "What" as the interrogative pronoun in an interrogative complement clause would generally be stressed, but as a relative pronoun it would generally be UNstressed.

I think it's interesting that this what is kinda the object of two different verbs at the same.

Yeah. It's worth further noting that in Latin, if the syntactic functions conflict (object of subordinate verb as subject of main verb, or viceversa), the case of the pronoun is resolved as the case of the function taken with the subordinate verb.

Quem vocāvistī nōn manifestum est.
who.ACC call.PRF-2S not obvious-SG.NEUT is-PRES.3S
'Who you called is not obvious.' (It's not easy to tell who you called.)
(ignorance of who was called; object of "to call" in interrogative clausal subject of "to be"; accusative "who")

Quis tē percussit nesciō.
who.NOM 2S.ACC hit.PRF-3S not.know.PRES-1S
'I don't know who hit you.'
(ignorance of who hit you; subject of "to hit" in interrogative clausal object of "to not know"; nominative "who")

[akam chinjir]

I'll eat what you give me

"Eat" can't take a clausal complement; "what you give me" is just a noun phrase. Maybe "what" is a relative pronoun, I'm not sure (the Cambridge Grammar doesn't think so). You can replace it with "that which," suggesting it has the role both of the noun head and of the relative pronoun.

I think it's interesting that this what is kinda the object of two different verbs at the same.

Yeah: it really is doing the work of both "that" and "which" in "that which."

Nani wo sureba ii ka wakaranai.
nani wo sur-eba ii ka wakar-anai
what OBJ do-HYP good Q* understand-NEG.PRS.FAM
*This is normally a question particle, but it's also used for some reason in certain cases when you use question words in non-questions.

It's an embedded question, though, lots of languages will use interrogative constructions in this sort of context. (They might differ from their main-clause versions in some ways---like in English you don't get subject inversion or do-support, or in Mandarin you can't have the question particle ma 嗎).

The only problem is what do I call this what that is kinda like a relative pronoun but not. So far the rule in my grammar has been that question words are only used in questions, so I should mention something about when that's not actually the case.

It's still just the regular question word, not a relative pronoun at all.

You do get phrases apparently headed by "what" that function pretty much like noun phrases (or DPs, if you prefer) that include a relative clause. These are the constructions that I referred to as headless relative clauses. Here's an example:

I'll eat what you give me

"Eat" can't take a clausal complement; "what you give me" is just a noun phrase. Maybe "what" is a relative pronoun, I'm not sure (the Cambridge Grammar doesn't think so). You can replace it with "that which," suggesting it has the role both of the noun head and of the relative pronoun.

Why does the Cambridge Grammar suggest that this is not a headless relative clause exactly? Not only can "what" be replaced by "that which" in this case, there's also not much of an interrogative force.

Oops, what I wrote wasn't clear (har har). They take it to be equivalent to that which you give me, which isn't headless and isn't actually a relative clause (it's a noun phrase that includes a relative clause). (They call these fused relatives, fwiw.) What I meant to say was that they don't call this "what" a relative pronoun because it does the work both of "that" and of "which"; but now I look back I can't find them saying anything about that issue.

[Kuchigakatai]

Oops, what I wrote wasn't clear (har har). They take it to be equivalent to that which you give me, which isn't headless and isn't actually a relative clause (it's a noun phrase that includes a relative clause). (They call these fused relatives, fwiw.) What I meant to say was that they don't call this "what" a relative pronoun because it does the work both of "that" and of "which"; but now I look back I can't find them saying anything about that issue.

Oh, okay. Is there anything the Cambridge Grammar calls a "headless" relative clause?

[akam chinjir]

[Oh, okay. Is there anything the Cambridge Grammar calls a "headless" relative clause?

No. (The only other things you might call that use "whatever" (&c.), and they also call those fused relatives. On page 1036 they have a footnote mentioning that they think the word "headless" is misleading here.)

(Aside: it's quite daunting to poke around in a 2000-page grammar; thank goodness for searchable PDFs.)

[Qwynegold]

Thanks for the replies! I'll read through them soon. I just logged in to post this question:

What are some ways other natlangs (or conlangs) construct the meaning "except for"? Like, in Swedish we say "for-beyond" (förutom) and in Finnish we have an unanalyzable word (paitsi) for that, but you can also phrase it as "other as" (muuta kuin). And does this have any interesting effects on the syntax of the given language?

[malloc]

Still struggling with some sandhi rules in my current project. The highly minimalist phonology often creates situations where combinations of phonemes must change in improbable ways. The language only allows geminates and homorganic nasal-stop clusters, so liquids forced into the coda must change somehow. The language has no hushing sibilants so processes that would normally palatalize consonants must do something else. The current rules I have are difficult to explain from a featural standpoint, requiring spontaneous nasalization and depalatalization. Yet the alternatives mean discarding phonological miminalism and allowing certain things as marginal phonemes or codas.

[Kuchigakatai]

Thanks for the replies! I'll read through them soon. I just logged in to post this question:

What are some ways other natlangs (or conlangs) construct the meaning "except for"? Like, in Swedish we say "for-beyond" (förutom) and in Finnish we have an unanalyzable word (paitsi) for that, but you can also phrase it as "other as" (muuta kuin). And does this have any interesting effects on the syntax of the given language?

I made a separate thread. This is more of a question about natlangs anyway.

http://verduria.org/viewtopic.php?f=4&t=292

[zompist]

Would it be realistic to have a polypersonal agreement system where the same affixes are used for both agent and patient?

Sure, this is what Swahili has. Well, almost has: the prefixes are the same for noun classes and 3 of the 6 pronouns. (The prefixes are distinguished by order.)

[bradrn]

Would it be realistic to have a polypersonal agreement system where the same affixes are used for both agent and patient?

Sure, this is what Swahili has. Well, almost has: the prefixes are the same for noun classes and 3 of the 6 pronouns. (The prefixes are distinguished by order.)

Thanks! (I was wondering since it seems to collapse fairly quickly into an erg-abs agreement system if you have any sound change at all at the end of a word.)

[zompist]

Just to be clear, Swahili is nom/acc. E.g.:

ni-li-wa-sikia
1s-past-3p-hear
I heard them

wa-li-ni-sikia
3p-past-1s-hear
They heard me.

ni-li-jibu
1s-past-answer
I answered.

That is, the prefix order makes it nom/acc.

[akam chinjir]

Thanks! (I was wondering since it seems to collapse fairly quickly into an erg-abs agreement system if you have any sound change at all at the end of a word.)

Thinking about zompist's Swahili example made me remember something else that might be relevant: when there are both subject and object agreement markers, the object markers tend to be closer to the stem (so it'd generally be the subject markers that get affected by word-edge sound changes, resulting in a nom/acc pattern).

(In the paper I linked earlier, she argues that cases of ergative agreement patterns with unmarked case all involve clitics: in an intransitive clause, it'll double the subject, but in a transitive clause, it'll double the object, with the subject getting an agreement marker on the verb. Her explanation is threefold: these languages choose a clitic over an agreement marker when there's a choice; they allow a maximum of one clitic; when there'll be both a clitic and an agreement marker, it's the subject that gets agreement, because object agreement entails subject agreement.)