help I made my conlang kind of ergative (plot twist: didn't)

[Torco]

so, I'm working on this new conlang, Bau Danyi, first I've done anything on for a few years. I had forgotten how fun and complicated this thing is. Anyway, I found myself in a strange situation.

The language is meant to be minimalistic with regards to inflections: I'm sure i'll give it some bound morphemes but not so far. originally I was thinking SOV syntax with nominative alignment, nothing fancy, but I decided to give it a twist: the stipulation was only the object phrase is going to carry any articles (there's a bunch of articles). so, normally, you get stuff like this.

I'm eating the soup
oka em hayra sin ta
oka em hayra sin ta
1PER.SING ART.DEF soup to.eat CONTINUOUS.ASPECT

I'm eating
oka sin ta
oka sin ta
1PER.SING to.eat CONTINUOUS.ASPECT

but then I decided that it makes no sense to not be able to express definiteness when, say, answering a question, so this is allowed.

do you know my friend ?
oka dun yorin e madr ?
oka dun yorin e madr
1PER.S POSSESIVE friend 2PER.S.INFORMAL know

the tall one?
em sotok so?
em sotok so
ART.DEF height of
sotok so is a relative phrase, em functions as 'the'.

him
yes

Thing is, that ends up looking a lot like an abs-erg system, if we take the article to be a case marker... which I like the idea of it being a case marker because the language even has a sort of neutral article, which doesn't convey definiteness information but you still use it when, say, a pronoun is in the object.

call him/her
lao ga gja
lao ga gja
call ARTICLE 3P.SING.FORMAL

imperative sentences use SVO/VO syntax, but notice that instead of em here we use ga, which doesn't encode any info about definiteness, and can function as an indefinite article even though in this case the pronoun gja, third person singular formal, is understood to be known by both speakers. so articles are kind of case markers, and it turns out that sometimes the cases get mixed up. Does this kind of thing happen? does it maybe make more sense not to treat these articles as case markers?

(damn, the board ate all my extra spaces in the gloss. also, sorry if the glosses suck, I'm p rusty)

[Vardelm]

so articles are kind of case markers, and it turns out that sometimes the cases get mixed up. Does this kind of thing happen? does it maybe make more sense not to treat these articles as case markers?

Yes, I think it does happen. AFAIK various languages make use of cases and definiteness gets wrapped up in that. Tagalog might do this, at least for the noun that is "in focus" or is the trigger of the verb voice. Turkish has a definite accusative case marker; take a look here for info.

(damn, the board ate all my extra spaces in the gloss. also, sorry if the glosses suck, I'm p rusty)

Try the gloss tags. Works great! For example:

Code: Select all

I'm eating the soup
[b]oka em hayra sin ta[/b]	
[gloss=1PER.SING]oka[/gloss]  [gloss=ART.DEF]em[/gloss] [gloss=soup] hayra [/gloss] [gloss=to.eat]sin[/gloss]  [gloss=CONTINUOUS.ASPECT]ta[/gloss]

I'm eating the soup
oka em hayra sin ta

oka

1PER.SING

em

ART.DEF

hayra

soup

sin

to.eat

ta

CONTINUOUS.ASPECT

[Ares Land]

Hey, nice to see you conlanging again!

That looks really like a good kind of problem to have.
Honestly, this looks good and I hope to see more of it.

I've found that if the explanation you end up for a phenomenon makes sense... it probably happened in a natlang somewhere. So I wouldn't worry about that.

[Torco]

ANADEW, nice. I like the gloss tags, neat addition. I'll check Tagalog as well thanks.

Honestly, this looks good and I hope to see more of it.

oh, there's a bunch more. I'm going about writing the grammar in kind of a meandering way, so expect random snippets.

[Qwynegold]

also, sorry if the glosses suck, I'm p rusty

Your glosses are unnecessarily long. Wikipedia has a good list of glossing abbreviations: https://en.m.wikipedia.org/wiki/List_of ... reviations

[bradrn]

so articles are kind of case markers, and it turns out that sometimes the cases get mixed up.

I’m not quite sure what you mean by the cases getting ‘mixed up’… could you elaborate?

Does this kind of thing happen?

Well, I don’t know of any natlangs which underwent this exact process of developing ergative markers from articles (I wrote up a fairly comprehensive listing earlier and saw nothing quite like this), but it doesn’t seem obviously implausible. After all, it’s no less strange than Tauya getting an ergative marker from its third person object pronoun!

I must say, however, that some parts of this system strike me as rather strange. Most prominently, it marks absolutive arguments but not ergative ones. This is, to say the least, extremely odd. (According to Handshuh’s A typology of marked-S languages, there is exactly one language which does this — the Austronesian language Nias — and even that case is debatable.)

One thing I’m confused about is the exact distribution of the articles (or absolutive case-markers, you could call them either). You state pretty clearly that they aren’t allowed on A arguments — but are articles optional or obligatory on O? And I’m really confused about whether or not they’re allowed on S arguments… I assume they are allowed with S in some situations, otherwise it wouldn’t be an ‘ergative’ system, but you give no examples. (It would actually work out quite nicely if articles are allowed with only a small subset of S NPs… that could well give it enough ‘accusative-ness’, so to speak, that the universal tendency against marked-absolutive alignments no longer applies.)

so articles are kind of case markers, and it turns out that sometimes the cases get mixed up. Does this kind of thing happen? does it maybe make more sense not to treat these articles as case markers?

Yes, I think it does happen. AFAIK various languages make use of cases and definiteness gets wrapped up in that. Tagalog might do this, at least for the noun that is "in focus" or is the trigger of the verb voice. Turkish has a definite accusative case marker; take a look here for info.

Not sure Tagalog is the right comparison here; Greenlandic and Niuean are probably better reference points, given their ergativity.

(Interestingly, those examples indicate that it is the absolutive argument which tends to be intrinsically definite in ergative alignments; accordingly it seems rather strange to me to allow articles on the absolutive and not the ergative. On the other hand, this doesn’t really apply to the same extent in OP’s system, given that that one is marked-absolutive rather than marked-ergative.)

[Travis B.]

Just the idea of marked-absolutive ergativity seems odd to me, since I am used to the ergative being the marked argument just like how the accusative is the marked argument with accusative alignments.

[Vardelm]

Just the idea of marked-absolutive ergativity seems odd to me, since I am used to the ergative being the marked argument just like how the accusative is the marked argument with accusative alignments.

A marked absolutive is odd in that AFAIK it doesn't exist in any natlang, but ergativity isn't common anyway. It's not particularly strange in concept, since it would be similar to marked nominative alignment, which Tolkien used for Adunaic.

I think it becomes easier to accept if you see it tied up with definiteness. From what little I've seen, it seems to be the patientive argument that gets marked as definite. Turkish, as I mentioned, is a good example. Do many languages perhaps assume that the subject is definite, perhaps unless marked as indefininte?

Not sure Tagalog is the right comparison here; Greenlandic and Niuean are probably better reference points, given their ergativity.

I think the issue here is definiteness being tied in with case marking, rather than whether the language is ergative or not. I didn't see anything about definiteness in the Greenlandic example, but did in Niuean. The sample sentence in that Niuean link was interesting because - to me - that looks more like "transitive alignment", where A & O are marked the same, but different from S. Word order is playing some role in that as well, but I'm not quite sure what unless I dive in deeper.

(Interestingly, those examples indicate that it is the absolutive argument which tends to be intrinsically definite in ergative alignments; accordingly it seems rather strange to me to allow articles on the absolutive and not the ergative.

As I mentioned above, if you compare to Turkish where the object (accusative) is also definite, this makes sense.

[bradrn]

Just the idea of marked-absolutive ergativity seems odd to me, since I am used to the ergative being the marked argument just like how the accusative is the marked argument with accusative alignments.

A marked absolutive is odd in that AFAIK it doesn't exist in any natlang, but ergativity isn't common anyway. It's not particularly strange in concept, since it would be similar to marked nominative alignment, which Tolkien used for Adunaic.

Marked-absolutive alignment exists in Nias, as I mentioned above. (This is one of the things I recently discovered I got wrong in my ergativity book.) Here’s an example (from Handschuh 2014):

aukhu n-idanö STAT.hot ABS-water ‘the water(ABS) is hot’
i-fo-houu defau idanö nasi 3s-CAUS-have-rust iron.ABS water.ERG sea.ABS ‘the seawater(ERG) rusted the iron(ABS)’
la-bunu m-baꞵi 3p.RLS-kill ABS-pig ‘they(ERG) killed a pig(ABS)’

I think it becomes easier to accept if you see it tied up with definiteness. From what little I've seen, it seems to be the patientive argument that gets marked as definite. Turkish, as I mentioned, is a good example. Do many languages perhaps assume that the subject is definite, perhaps unless marked as indefininte?
…

(Interestingly, those examples indicate that it is the absolutive argument which tends to be intrinsically definite in ergative alignments; accordingly it seems rather strange to me to allow articles on the absolutive and not the ergative.

As I mentioned above, if you compare to Turkish where the object (accusative) is also definite, this makes sense.

Interesting… I suppose that, since undergoer arguments are naturally less definite, it makes more sense to mark definiteness on them. On the other hand, Turkish is an accusative language; if we go by the dictum that ergative languages work the opposite way to accusative ones (which seems to be the case surprisingly often), that means that ergative languages would mark definiteness on the ergative argument. (And indeed, this is borne out by the observation that the absolutive is assumed to be definite in Greenlandic and Niuean; also consider Adyghe (McGregor 2007), in which the presence of ergative marking is dependent on the definiteness of the NP.) On the third hand, given that marked-nominative systems often show properties similar to ergative systems, we might presume that marked-absolutive systems are more similar to accusative systems, and will mark definiteness on the absolutive… but on the fourth hand, we only have one natlang example of such a system, so it’s hard to know how true this is.

Not sure Tagalog is the right comparison here; Greenlandic and Niuean are probably better reference points, given their ergativity.

I think the issue here is definiteness being tied in with case marking, rather than whether the language is ergative or not. I didn't see anything about definiteness in the Greenlandic example, but did in Niuean. The sample sentence in that Niuean link was interesting because - to me - that looks more like "transitive alignment", where A & O are marked the same, but different from S. Word order is playing some role in that as well, but I'm not quite sure what unless I dive in deeper.

For Greenlandic:

… the absolutive NP has certain special interpretive properties, which much of the traditional literature has interpreted as definiteness …

As for Niuean, I’d be interested why you consider it to have transitive alignment; it doesn’t look like that to me at all. But it is indeed the case that word order is important: Niuean sentences usually have structure ‘V (ERG.marker) A (ABS.marker) O’, but the absolute argument is required to be definite. When it is not, the object must be ‘incorporated’ into the verb (the usual term), involving promotion of A and rearrangement of word order, similar to an antipassive: ‘V O (ABS.marker) A’.

[Torco]

One thing I’m confused about is the exact distribution of the articles (or absolutive case-markers, you could call them either). You state pretty clearly that they aren’t allowed on A arguments — but are articles optional or obligatory on O? And I’m really confused about whether or not they’re allowed on S arguments

If A means the one argument of an intransitive, S the subject of a transitive and O the object of a transitive (bear with me, I'm rusty and was never great at linguistics anyway ), articles are gonna be optional on O, allowed on A and verbotten on S. think of it like this: syntactically, the conlang is a regular old garden variety SOV. word-order wise you can't distinguish the single argument of an intransitive from the subject of a transitive cause they're just coming before the verb, right?

I am a man who likes fish
oka arok yi na e ga tonan atu

oka

I

arok

man

yi

to.be

na

REL.PRONOUN

e

he

ga

ART

tonan

fish

atu

to.like

the crooked cat is dead
kao gan haegda gu gan yi

kao

bend

gan

PARTICIPLE

haegda

cat

gu

dead

gan

PARTICIPLE

yi

to.be

and the other argument of a transitive is going to be the noun or the noun phrase that's coming between the first noun phrase and the verb thingie. So then take that and add a rule that goes "no articles on the subject of a transitive verb". the main way speakers figure out if it's the fish or the man who is doing the liking is still word order, but if you see an article (and there's not a lot, there's like four) you know that you've reached the object part of the sentence. Unless, of course, the article was preceded by nothing, in which case you know you're looking at an intransitive sentence, or maybe just a naked noun phrase, like when someone answers a question or shouts "fuck, the cat!"

gjay, em haegda !

gjay

EXPLETIVE

,

em

ART

haegda

cat

From what little I've seen, it seems to be the patientive argument that gets marked as definite.

Yup.

[Vardelm]

As for Niuean, I’d be interested why you consider it to have transitive alignment

I was looking at this example:

Ne

PAST

inu

drink

e

ERG

Sione

Sione

e

NOM

kofe

coffee

‘Sione drank the coffee.’

Ne

PAST

inu

drink

kofe

coffee

a

ERG

Sione

Sione

‘Sione drank coffee.’


The 2 "e" particles compared to the single "a" particle made me think "a-HAH! Transitive alignment!" Looking at it now, I obviously didn't look close enough since both sentences are transitive and differ only by the definite "coffee". Apparently when I looked at it this morning, my own coffee hadn't quite kicked in yet. 1st day back after the holidays & all that.

[bradrn]

One thing I’m confused about is the exact distribution of the articles (or absolutive case-markers, you could call them either). You state pretty clearly that they aren’t allowed on A arguments — but are articles optional or obligatory on O? And I’m really confused about whether or not they’re allowed on S arguments

If A means the one argument of an intransitive, S the subject of a transitive and O the object of a transitive (bear with me, I'm rusty and was never great at linguistics anyway ), articles are gonna be optional on O, allowed on A and verbotten on S.

Other way round: S is the sole argument of an intransitive verb, while A and O are the agent and object of a transitive verb. (If you want a refresher, I’ve written some articles on this stuff.) So rephrasing your statement using standard terminology: articles are forbidden on A (=transitive subject) but allowed on O and S (=transitive object and intransitive subject). I assume that articles are never obligatory, correct?

think of it like this: syntactically, the conlang is a regular old garden variety SOV. word-order wise you can't distinguish the single argument of an intransitive from the subject of a transitive cause they're just coming before the verb, right? … and the other argument of a transitive is going to be the noun or the noun phrase that's coming between the first noun phrase and the verb thingie. So then take that and add a rule that goes "no articles on the subject of a transitive verb". the main way speakers figure out if it's the fish or the man who is doing the liking is still word order, but if you see an article (and there's not a lot, there's like four) you know that you've reached the object part of the sentence.

OK, this makes sense as a case-marking strategy. But then why are articles needed on S? After all, if a sentence has only one argument, there’s no need to disambiguate it from the other arguments, because there aren’t any! (This is the whole reason why marked-S alignments are so rare.)

Unless, of course, the article was preceded by nothing, in which case you know you're looking at an intransitive sentence, or maybe just a naked noun phrase, like when someone answers a question or shouts "fuck, the cat!"

gjay, em haegda !

gjay

EXPLETIVE

,

em

ART

haegda

cat

Unless I’m getting something wrong here, I’m pretty sure that’s not an intransitive sentence. As far as I’m aware, intransitive sentences are verbal sentences in which the verb takes exactly one argument: I jump, the window breaks, my text editor is not working right now. On the other hand, your sentence is non-verbal, thus is neither transitive nor intransitive. (I’ve seen ‘sentence fragment’ used in traditional grammar for sentences like yours… not sure if the term has any counterpart in modern syntax, but I like it as it clearly distinguishes these sentences from the more usual sort of verb-containing sentences.)

As for Niuean, I’d be interested why you consider it to have transitive alignment

I was looking at this example:

Ne

PAST

inu

drink

e

ERG

Sione

Sione

e

NOM

kofe

coffee

‘Sione drank the coffee.’

Ne

PAST

inu

drink

kofe

coffee

a

ERG

Sione

Sione

‘Sione drank coffee.’


The 2 "e" particles compared to the single "a" particle made me think "a-HAH! Transitive alignment!" Looking at it now, I obviously didn't look close enough since both sentences are transitive and differ only by the definite "coffee". Apparently when I looked at it this morning, my own coffee hadn't quite kicked in yet. 1st day back after the holidays & all that.

Ah, yes, that example. That’s just due to syncretism: if you look at page 494, e is both the ergative marker for proper nouns and the absolutive marker for common nouns. (And syntactically, the second sentence actually is intransitive: that’s why its ‘subject’ takes the absolutive marker a.)

[Torco]

articles aren't obligatory, no.

The concept of a non-verbal sentence... is this different from just a regular old phrase?

I guess that's kind of the thing, here, innit. Cause I'm okay with letting go of articles in intransitive sentences, it's those guys that I'm hesitant to leave without an easy way of definiteness marking. Yeah, you've clarified things: you don't put articles in the S or in the A but you do in Os and also in loose phrases.

false alarm, it's not kind of ergative. And I'll give that thread a look.

[bradrn]

The concept of a non-verbal sentence... is this different from just a regular old phrase?

Yes, they’re different. By ‘non-verbal’, I just meant clauses which don’t have a verb. Which includes plain NPs, interjections and sentence fragments like your example, but also other types of non-verbal clauses — most notably, copula clauses in languages without a verbal copula.

I guess that's kind of the thing, here, innit. Cause I'm okay with letting go of articles in intransitive sentences, it's those guys that I'm hesitant to leave without an easy way of definiteness marking.

It doesn’t really matter: plenty of languages have no dedicated definiteness markers! And even when they do, there are still ways of marking definiteness — most prominently, demonstratives are often used in this function.

Yeah, you've clarified things: you don't put articles in the S or in the A but you do in Os and also in loose phrases.

Yep, that seems to be the case.

(I must say, it does still strike me as a little strange that bare NPs can take what is after all the marked case. But it’s still within the bounds of plausibility.)