Conlang Random Thread

[Ahzoh]

Ultimately I decided against the aforementioned marking, since it's really just nom-acc with two nominative cases, and thus kind of pointless.

Frankly, I think i'll stick to Vrkhazhian being nom-acc and not split ergative since the language should be about doers and not doees. That means revitalization and syntactic pivots should center around the agent of a transitive and not the patient; that one should be able to say "king-NOM [people-ACC united-REL]" and not "people-ABS [king-ERG united-REL]" without any explicit pronouns or valence-changing markers like passives and antipassives.

I even drawn up a list of how relativization would work between the two alignments

Code: Select all

NOM-ACC
SV  = king-NOM be_strong
SV  = king-NOM [be_strong-ATTR]
SOV = king-NOM people-ACC unite-APPL
OV  = king-NOM [people-ACC unite-APPL-ATTR]

SOV = king-NOM [who-NOM people-ACC unite-APPL]
OSV = people-NOM [who-ACC king-NOM unite-APPL]

SV  = people-NOM unite-APPL-PASS
SV  = people-NOM [unite-APPL-PASS-ATTR]
SXV = people-NOM [by-PREP king-INS unite-APPL-PASS-ATTR]

Code: Select all

ERG-ABS
SV  = king-ABS be_strong
SV  = king-ABS [be_strong-ATTR]
SOV = king-ERG people-ABS unite-APPL
SV  = people-ABS [king-ERG unite-APPL-ATTR]

OSV = people-ABS [REL-ACC king-ERG unite-APPL-ATTR]
SOV = king-ABS [REL-NOM people-ABS unite-APPL]

SV  = king-ABS unite-APPL-ANTIP
SV  = king-ABS [unite-APPL-ANTIP-ATTR]
SXV = king-ABS [for-PREP people-ERG unite-APPL-ANTIP-ATTR]

And so concluded that nom-acc best fits with my inflected relativized verbs (in terms of being consistent with who the head of the phrase is). A hard decision since having pronoun be nom-acc and full nouns be erg-abs was quite interesting. But I don't like the asymmetry.

[jal]

Hmm, so causatives are always dynamic...

Obviously, as change is the opposite of statis...


JAL

[Ahzoh]

Besides, even ignoring that, this feels the wrong way round to me. Active/dynamic verbs generally group S with the more agentive argument, which is A, giving nominative–accusative alignment. Similarly stative verbs naturally group S with the less agentive argument, namely O, giving ergative–absolutive alignment.

So it would be absolutive for stative and nominative for dynamic. That would be awkward for a transitive verb.

Hmm, so causatives are always dynamic...

Obviously, as change is the opposite of statis...

I did not consider changes of state being actions in of themselves, especially changes of state not effected by the experiencers but by outside forces, such as the cooking of a casserole.

[bradrn]

Ultimately I decided against the aforementioned marking, since it's really just nom-acc with two nominative cases, and thus kind of pointless.

Not really… differential subject marking is well-attested for ergative–absolutive languages, and this would just be the marked-nominative equivalent. The choice of nominative case would be determined by the agentivity of the noun, which is very reasonable.

(If you want, you could also consider optional nominative marking as another possibility. That really just amounts to making one of the case-markers null. My hunch in this case would be to do that for active verbs, but I’m really not sure.)

Besides, even ignoring that, this feels the wrong way round to me. Active/dynamic verbs generally group S with the more agentive argument, which is A, giving nominative–accusative alignment. Similarly stative verbs naturally group S with the less agentive argument, namely O, giving ergative–absolutive alignment.

So it would be absolutive for stative and nominative for dynamic. That would be awkward for a transitive verb.

Well, most active-stative languages always mark transitive verbs the same way, while distinguishing active intransitive verbs (S=A) from stative ones (S=O). Effectively, if they use dependent-marking — which itself seems fairly rare for active-stative languages — this means they have a nominative case and an absolutive case, with either being possible for the sole intransitive argument.

(The one exception I know of is Mayan languages… but right now I can’t quite recall how their system works. It‘s very late here so I won’t look it up now, but I’ll try to summarise it tomorrow if I can find the description again.)

[WeepingElf]

Well, most active-stative languages always mark transitive verbs the same way, while distinguishing active intransitive verbs (S=A) from stative ones (S=O). Effectively, if they use dependent-marking — which itself seems fairly rare for active-stative languages — this means they have a nominative case and an absolutive case, with either being possible for the sole intransitive argument.

Yes, active-stative case marking is rare (this reminds me of a meaningless beef many years ago between me and another member of the CONLANG list who claimed that a conlang of mine - Nur-ellen - did it wrongly), probably because active-stative alignment is found mostly in indigenous languages of the Americas, which tend to be head-marking. The closest to a language with active-stative alignment expressed by case marking I am aware of is Georgian. Georgian has an aspect-based alignment split: it is nom-acc with imperfective verbs and what is usually described as erg-abs with perfective verbs. But there are apparently some intransitive verbs which take an ergative subject in the perfective aspect, though I don't know the facts well enough to decide whether this actually constitutes an active-stative alignment or is just a bunch of quirky verbs - such irregularities apparently occasionally occur in ergative languages, and I have seen it reported e.g. for Basque. (The person/number marking on the Georgian verb, as far as I have understood it, follows a nominative-accusative pattern.)

This rareness of active-stative case systems is certainly the reason why there are no generally accepted naming conventions for such cases, as I noticed when I worked on Nur-ellen - I had to coin my own terms ("agentive" and "objective") for them, which I still use in my present Albic project (which has grown out of Nur-ellen, but is no longer based on Tolkien's languages).

[Ahzoh]

Thought some more and maybe I will shift the alignment split to animate versus animate.
So:

Code: Select all

Animate
NOM: -m
ACC: -s
INS: -n

Inanimate
ABS: -s
ERG: -n
INS: -n

It just makes sense since I already have a morphological animacy distinction and I can futher emphasize that with a change in alignment and morphological marking.

Though, maybe inanimates should be alignment neutral since I'm thinking of making them morphosyntactically unable to be subject/agents since I'm having troubles coming up with second person and third person neuter markers for verbs.

[jal]

I seem to recall that, with regards to Active-Stative languages, Split-S is most likely to use head marking (i.e. on the verb), while Fluid-S is most likely to use dependent marking (i.e. on the subject). This is because with Split-S, it's inherently the verb that determines the marking of the subject, while with Fluid-S it's the subject that determines it (and so it gets marked). It's a long time ago though I did anything with alignment, so I might recall incorrectly.


JAL

[jal]

I did not consider changes of state being actions in of themselves, especially changes of state not effected by the experiencers but by outside forces, such as the cooking of a casserole.

I see where you're coming from, but an "action" needn't be a purposeful action by some agent. It can be any event that's non-stative, i.e. there's a change in state, no matter what brings it on.

As a side, when you say "state not effected by the experiences but by outside forces", I suppose you mean "not by an force mentioned in the clause", i.e. the force isn't a subject. In that case, you may consider a middle voice that indicates that (a middle voice is in between the active (someone does something to someone else) and passive (something is done to someone, by someone else). Some languages, iirc, also use the middle voice for relexives, e.g. "to wash oneself", without overtly encoding the "oneself" part.


JAL

[WeepingElf]

I seem to recall that, with regards to Active-Stative languages, Split-S is most likely to use head marking (i.e. on the verb), while Fluid-S is most likely to use dependent marking (i.e. on the subject). This is because with Split-S, it's inherently the verb that determines the marking of the subject, while with Fluid-S it's the subject that determines it (and so it gets marked). It's a long time ago though I did anything with alignment, so I might recall incorrectly.

I seem to recall reading that somewhere, too; it sounds familiar to me, though I don't have the slightest idea where to look for the source. Also sounds quite reasonable and logical to me. And Old Albic is indeed a fluid-S language. As for the beef with that other CONLANG list member, it now seems to me as if an Africanist claimed that a conlanger who has a tone system inspired by Chinese "did tones wrong" - it just different kinds of tone systems. The same way, Nur-ellen had (and Old Albic has, or will have) a different kind of active-stative alignment than the indigenous North American languages that guy had in mind. Well, why should it behave like a North American language when it is spoken in Britain?

[Ahzoh]

I see where you're coming from, but an "action" needn't be a purposeful action by some agent. It can be any event that's non-stative, i.e. there's a change in state, no matter what brings it on.

I also thought the difference was what I considered internal vs external; to think versus to run, to see versus to hit. But that's not a perfect distinction either.

Also yes, I was describing the middle voice.

[Ahzoh]

This whole topic just demonstrates my anxiety towards anything involving alignments that aren't Nom-Acc. It requires so much thinking and considerations of minute aspects that I otherwise wouldn't have to think too hard about if I stuck to vanilla Nom-Acc.

But I want to make it work, because I think it can work with Vrkhazhian's existing structures and because it adds further uniquity to Vrkhazhian (without being too kitchensinky).

[bradrn]

The closest to a language with active-stative alignment expressed by case marking I am aware of is Georgian.

Batsbi is straightforwardly active-stative with case-marking. There’s also a bunch of supposedly ‘ergative’ languages which really have some form of active-stative alignment: most prominently Basque and Tibetan. (You could even argue that it’s because those ones use case-marking that they often aren’t called ‘active-stative’.)

This rareness of active-stative case systems is certainly the reason why there are no generally accepted naming conventions for such cases […]

I’d say ‘nominative’ and ‘absolutive’ are quite standard choices. In languages where active situations are considered uncommon, the former has often been called ‘agentive’ (especially in Sino-Tibetan), or even ‘ergative’.

This whole topic just demonstrates my anxiety towards anything involving alignments that aren't Nom-Acc. It requires so much thinking and considerations of minute aspects that I otherwise wouldn't have to think too hard about if I stuck to vanilla Nom-Acc.

Not really, you just have to learn how to reason about them. This is why I wrote a thread about alignment systems — it’s not perfect, but I’ll still recommend reading it.

[Ahzoh]

Not really, you just have to learn how to reason about them. This is why I wrote a thread about alignment systems — it’s not perfect, but I’ll still recommend reading it.

I had been reading it and what I learned there is what is giving me anxiety and doubt about whether I want any split ergativity; I didn't like the asymmetricality of S/O pivots, for example.

Well, like basically, since I speak a Nom-Acc language, it is most familiar to me and so I can pick and choose what morphosyntactic aspects to work on and should I miss something I can simply fall back on the strategies of my native tongue. But I can't do that with alignments that aren't the same or closely similar to my mother tongue's; I might just end up doing something that is unusual, uncharacteristic, or illogical. Perhaps if I spoke some language like Basque as my native tongue, I'd feel the opposite way.

As I have said, however, I had resolved at a compromise: deciding the split alignment on animate versus inanimate grounds rather than pronoun versus full-noun. I want to be able talk about a king who united a people, not a people who a king united. With simple relativization and not any extra morphological shenanigans like antipassives and reintroducing arguments through oblique compliments.

[bradrn]

Not really, you just have to learn how to reason about them. This is why I wrote a thread about alignment systems — it’s not perfect, but I’ll still recommend reading it.

I had been reading it and what I learned there is what is giving me anxiety and doubt about whether I want any split ergativity; I didn't like the asymmetricality of S/O pivots, for example.

I’m not sure S/O pivots are any less symmetrical than S/A ones, though of course it depends what you mean by ‘symmetrical’.

That being said, S/O pivots are only relevant for syntactically ergative languages — actually, not even that, it’s relevant for some syntactically ergative languages. So you can get away with having ergative alignment without an S/O pivot.

As I have said, however, I had resolved at a compromise: deciding the split alignment on animate versus inanimate grounds rather than pronoun versus full-noun.

These are both examples of splits based on the animacy hierarchy. It doesn’t necessarily change how the language behaves otherwise.

I want to be able talk about a king who united a people, not a people who a king united. With simple relativization and not any extra morphological shenanigans like antipassives and reintroducing arguments through oblique compliments.

Well, that’s something different again, because it depends on how your language treats relativisation. It’s true that, if a language only allows relativisation on one argument, that’s usually the nominative argument for accusative languages and the absolutive argument for ergative languages… but then again, most languages allow relativisation on more than one argument, in which case the distinction doesn’t matter.

[Ahzoh]

I’m not sure S/O pivots are any less symmetrical than S/A ones, though of course it depends what you mean by ‘symmetrical’.

I would consider "mother came and [mother] saw father" as symmetrical whereas "mother came and father saw [mother]" as asymmetrical.

Yes, I do seem to conflate morphological ergativity with syntactic ergativity.

These are both examples of splits based on the animacy hierarchy. It doesn’t necessarily change how the language behaves otherwise.

Yes, I was merely specifying extent. And it does change which nouns get relativized a certain way since in Vrkhazhian only one kind of argument can be relativized.

Well, that’s something different again, because it depends on how your language treats relativisation. It’s true that, if a language only allows relativisation on one argument, that’s usually the nominative argument for accusative languages and the absolutive argument for ergative languages… but then again, most languages allow relativisation on more than one argument, in which case the distinction doesn’t matter.

Oh that's right, I can simply change word order and have either the object first or the subject first. Though that can't work in Vrkhazhian because the relativized verb must have concord with the head of the relative clause. There is very little if any distinction between verbs, adjectives, and participles.

[bradrn]

I’m not sure S/O pivots are any less symmetrical than S/A ones, though of course it depends what you mean by ‘symmetrical’.

I would consider "mother came and [mother] saw father" as symmetrical whereas "mother came and father saw [mother]" as asymmetrical.

Well, that’s because you’re using nominative–accusative word order, where both S and A go before the verb. If the latter sentence used ergative word order, it would instead be something like, “came mother and father saw”, which is much more obviously ‘symmetrical’.

(Although in practice, most ergative languages seem to be SOV: “mother-ABC came and father-ERG saw”.)

These are both examples of splits based on the animacy hierarchy. It doesn’t necessarily change how the language behaves otherwise.

Yes, I was merely specifying extent. And it does change which nouns get relativized a certain way since in Vrkhazhian only one kind of argument can be relativized.

Why should that change which argument is relativised? Generally speaking, languages with an animacy-based alignment split prefer to relativise the absolutive argument. To my knowledge, this doesn’t change with the precise location of the split. There might be languages with an animacy-based split which prefer to relativise the nominative argument, but I don’t know of them.

Well, that’s something different again, because it depends on how your language treats relativisation. It’s true that, if a language only allows relativisation on one argument, that’s usually the nominative argument for accusative languages and the absolutive argument for ergative languages… but then again, most languages allow relativisation on more than one argument, in which case the distinction doesn’t matter.

Oh that's right, I can simply change word order and have either the object first or the subject first. Though that can't work in Vrkhazhian because the relativized verb must have concord with the head of the relative clause. There is very little if any distinction between verbs, adjectives, and participles.

I really don’t see what word order has to do with relativisation.

[Ahzoh]

Why should that change which argument is relativised? Generally speaking, languages with an animacy-based alignment split prefer to relativise the absolutive argument. To my knowledge, this doesn’t change with the precise location of the split. There might be languages with an animacy-based split which prefer to relativise the nominative argument, but I don’t know of them.

I really don’t see what word order has to do with relativisation.

Where the split happens makes a difference, if I made it so pronouns were nom-acc but full nouns were erg-abs then king-NOM [people-ACC unite-REL] wouldn't be possible.

Code: Select all

NOM-ACC
SV  = king-NOM be_strong              = king is string
SV  = king-NOM [be_strong-REL]        = king who is strong
SOV = king-NOM people-ACC unite       = king united people
OV  = king-NOM [people-ACC unite-REL] = king who united people

SV  = people-NOM unite-PASS
SV  = people-NOM [unite-PASS-REL]
SXV = people-NOM [by-PREP king-INS unite-PASS-REL]

Code: Select all

ERG-ABS
SV  = king-ABS be_strong              = king is strong
SV  = king-ABS [be_strong-REL]        = king who is strong
SOV = king-ERG people-ABS unite       = king united people
SV  = people-ABS [king-ERG unite-REL] = people who king united

SV  = king-ABS unite-ANTIP
SV  = king-ABS [unite-ANTIP-REL]
SXV = king-ABS [for-PREP people-ERG unite-ANTIP-REL]

EDIT: Now, I think of it both arguments could be relativized no matter the alignment since the verb agrees only with the subject/agent, whether it's the extracted argument or not.

[bradrn]

Where the split happens makes a difference, if I made it so pronouns were nom-acc but full nouns were erg-abs then king-NOM [people-ACC unite-REL] wouldn't be possible.

EDIT: Now, I think of it both arguments could be relativized no matter the alignment since the verb agrees only with the subject/agent, whether it's the extracted argument or not.

I honestly have no clue what you’re talking about here. For your first example, yes, if full nouns get absolutive alignment then you can’t mark them with the nominative and accusative case… so what? For the second paragraph, extraction has no connection with verbal agreement that I know of. So at this point I’m very confused.

[Ahzoh]

Where the split happens makes a difference, if I made it so pronouns were nom-acc but full nouns were erg-abs then king-NOM [people-ACC unite-REL] wouldn't be possible.

EDIT: Now, I think of it both arguments could be relativized no matter the alignment since the verb agrees only with the subject/agent, whether it's the extracted argument or not.

I honestly have no clue what you’re talking about here. For your first example, yes, if full nouns get absolutive alignment then you can’t mark them with the nominative and accusative case… so what? For the second paragraph, extraction has no connection with verbal agreement that I know of. So at this point I’m very confused.

I don't know how my examples don't make it very clear that I mark relativization on the verb, turning said verb into a relative clause and that I'm obviously talking about relative clauses and their relations to the head of the clause.

[bradrn]

I don't know how my examples don't make it very clear that I mark relativization on the verb, turning said verb into be a relative clause […]

They do make that clear. What isn’t clear to me is what that has to do with restrictions on which NPs can be extracted out of that relative clause.