Zompist Bboard Again

I didn't include adjectives because I thought they were covered in the general discussion. But yeah, they're not objects in Quechua.

(And again, Quechua is a weird example of polypersonal agreement because it's quite restricted.)

Whimemsz wrote: ↑Sat Mar 09, 2019 10:12 am What does a proximate/obviative contrast have to do with interrogative pronouns here?

In my language, question words are exclusively sentence-initial. However, also in my language, the typical word order is Proximate > Obviate. It causes problem if interrogative pronoun happens to coexist as SAP pronoun. What should be sentence initial, Interrogative pronoun or SAP pronoun?

Whimemsz wrote: ↑Sat Mar 09, 2019 10:12 am (In any case, Algonquian languages have clause-initial interrogative pronouns -- but I don't know if you meant "clause initial" or really intended "word initial", and if the latter, I'm even more confused about what you're asking...)

The former. Sorry, I mistyped.

Whimemsz wrote: ↑Sat Mar 09, 2019 10:12 am And what "conflict" do you mean with regard to SAPs? Obviation is specifically a feature of non-SAPs, so SAPs are irrelevant.

Actually not. Sure that there is no such thing as Obviate SAP, but that's because in all languages, SAP is the most proximate of anything. It would be unnatural if there is any 3rd person argument that is somehow more proximate than SAP.

Akangka wrote: ↑Sun Mar 10, 2019 12:26 am Actually not. Sure that there is no such thing as Obviate SAP, but that's because in all languages, SAP is the most proximate of anything. It would be unnatural if there is any 3rd person argument that is somehow more proximate than SAP.

I think it would be more common to say that there's an animacy hierarchy (or anyway a hierarchy), and that within that hierarchy proximate 3rd person arguments outrank obviative ones (and both are outranked by SAPs).

Akangka wrote: ↑Sun Mar 10, 2019 12:26 am In my language, question words are exclusively sentence-initial. However, also in my language, the typical word order is Proximate > Obviate. It causes problem if interrogative pronoun happens to coexist as SAP pronoun. What should be sentence initial, Interrogative pronoun or SAP pronoun?

I don't see why this would be different from languages with sentence-initial question words and (say) subjects before objects: the question word goes before the first mentioned argument (maybe with exceptions involving topicalisation).

zompist wrote: ↑Sat Mar 09, 2019 11:52 pm (And again, Quechua is a weird example of polypersonal agreement because it's quite restricted.)

How? I always thought Quechua was fairly standard in this area - although admittedly I don't know all that much about the topic either...

bradrn wrote: ↑Sun Mar 10, 2019 1:14 am

zompist wrote: ↑Sat Mar 09, 2019 11:52 pm (And again, Quechua is a weird example of polypersonal agreement because it's quite restricted.)

How? I always thought Quechua was fairly standard in this area - although admittedly I don't know all that much about the topic either...

Because not all the possible combinations have suffixes. In fact, as I mentioned, there aren't any at all for 3s and 3p objects. And in fact, there are other missing suffixes.

zompist wrote: ↑Sat Mar 09, 2019 9:43 pm

Salmoneus wrote: ↑Sat Mar 09, 2019 6:47 pmWell, no words have person, other than pronouns. Personhood is indexical - it's a property not of the categorical reference of the word (I don't know the linguistic name for this - the quiddity of the reference, is what I mean), but of the haecceity of the referent relative to the perspective of the speech act. You can't take two nouns, like "the doctor" and "the footballer", and tell from the noun what 'person' they are - is a doctor inherently third-person, or is a footballer inherently second-person? Of course not - the noun doesn't have person, the referent has person, relative to a speech act. That guy is third person, regardless of what word I use to refer to him, and you are second person, relative to this speech act, regardless of what word I use to refer to you.

I think the linguistic way of saying this is that personhood is pragmatic, not semantic. It relates to the context of the utterance, not to the meaning of any of the words.

I'm obviously not a linguist, but indexicality (deixis, anaphora, etc) isn't pragmatic in the sense I'm familiar with. It's semantic, in that the reference is hard-coded, objective and fully predictable from the speech-act (granted that the definition of the speech act includes its time and place), whereas to me 'pragmatics' primarily concerns the way in which ambiguous, underspecified or superficially incoherent speech is understood through conventions and context. In particular, deixis (referential indexicality) seems beyond what is normally meant by pragmatics, particularly given just how much semantic content is deictic. But yes, indexicality does seem in some respect to lie between prototypical semantics and prototypical pragmatics, so it's only really a matter of convention what you call it, and I did indeed consider using the word 'pragmatics' in that post.

bradrn wrote: ↑Sat Mar 09, 2019 10:32 pm So it would seem that Quechua just dispenses with the object agreement altogether and treats the copula as syntactically transitive but morphologically intransitive; I think I'm going to use this approach.

What makes you think that the Quechua copula is syntactically transitive? What makes the adjective an object, in your classification, given that it doesn't take object agreement?

bradrn wrote: ↑Sat Mar 09, 2019 7:42 pm This transitivity hierarchy sounds interesting. Do you know of any resources where I can learn about it?

Not specifically, you'd have to ask a linguist. But in general, if you read about languages in which transitivity is marked in some way, you may find some discussions of what transitivity is. Transitivity is 'an issue' in lots of austronesian languages, for example.

What I mean by 'transitivity hierarchy' is that "transitivity" covers a broad range of concepts, and you can lay out the various factors that make something transitive, and a language is likely to have a coherent order such that anything 'left' of a point in the list will be transitive and anything 'right' of it will be intransitive (give or take some confusing compromises in the middle sometimes), but not all languages order their list identically, and some put the transitivity line in different places. Modern English puts the transitivity line very far down - basically every bivalent verb now acts like a transitive. To the point where a lot of English speakers, even those who know about languages, forget that 'transitive', 'bivalent' and 'dyadic' refer to three different properties*. Some other European languages, including Old and Middle English, are less transitivity-happy, with intermediate cases marked by anomolous subject or object case assignment. On the other hand, some Austronesian languages put the limit of transitivity very 'high', with everything that's not purely transitive being considered intransitive.

These issues also come up in languages where 'subjecthood' is an issue, because one prototypical property of a subject is being the agent of a transitive. Also sometimes in languages with 'animacy', as animacy can be seen as a lexical indication of likelihood to be the agent of a transitive (and, eg, in some languages inanimates CANNOT be the agent of a transitive, though they can be the subject of other verbs).




*
"Bivalent" is a syntactic property of verbs that in some way "take" two arguments.
"Dyadic" should probably properly be a semantic property of verbs that project two semantic roles. Although I've also seen it seemingly used as synonymous with 'bivalent'. However, in the pure sense you could argue that a passive verb in English is syntactically univalent, but still semantically dyadic.
"Transitives" are a subclass of dyadic verbs - the archetype of a dyadic verb, you might say. They generally have a syntax different from that of monadic verbs, and dyadic intransitives may follow one pattern, the other, or some third intermediate version.

I don't have any definitions in front of me, but generally a verb is most 'transitive' when:

- it refers to an event (rather than a state or a relationship or a property)
- it refers to an actual event (rather than an irrealis one)
- the event is an action (rather than just an occurence)
- the action is performed by an agent
- the agent is acting knowingly, voluntarily, avoidably, and purposefully
- the action is performed on a patient
- the patient has no control over the action
- the action is 'successful' and 'completed'
- the action brings about a change in the condition, and ideally the absolute properties, of the patient ("I break the vase" is very transitive; "I reposition the vase" is less transitive; "I see the vase" is least transitive)

Just to add a bit to what Sal said, a complication that's relevant to polypersonal agreement is that in plenty of languages with object agreement, you don't get agreement with all objects. Typically this is related to the specificity, animacy, singularity, &c, of the object (there are different patterns). Sometimes you'll see objects that don't trigger agreement described as pseudo-incorporated.

Salmoneus wrote: ↑Sun Mar 10, 2019 9:36 am

bradrn wrote: ↑Sat Mar 09, 2019 10:32 pm So it would seem that Quechua just dispenses with the object agreement altogether and treats the copula as syntactically transitive but morphologically intransitive; I think I'm going to use this approach.

What makes you think that the Quechua copula is syntactically transitive? What makes the adjective an object, in your classification, given that it doesn't take object agreement?

Yes, I suppose I meant bivalent rather than transitive. You said it yourself:

Salmoneus wrote: ↑Sun Mar 10, 2019 9:59 am Modern English puts the transitivity line very far down - basically every bivalent verb now acts like a transitive. To the point where a lot of English speakers, even those who know about languages, forget that 'transitive', 'bivalent' and 'dyadic' refer to three different properties*.

akam chinjir wrote: ↑Sun Mar 10, 2019 10:42 am Just to add a bit to what Sal said, a complication that's relevant to polypersonal agreement is that in plenty of languages with object agreement, you don't get agreement with all objects. Typically this is related to the specificity, animacy, singularity, &c, of the object (there are different patterns). Sometimes you'll see objects that don't trigger agreement described as pseudo-incorporated.

Do you have any examples of languages which do this? I haven't come across any so far.

bradrn wrote: ↑Sun Mar 10, 2019 3:26 pm Do you have any examples of languages which do this? I haven't come across any so far.

You could look this paper---it's about a Bantu language named Ruwund, but starts with a bunch of references.

Hungarian is supposed to be another example.

There are also case-marking languages in which direct objects get case-marked only under similar circumstances, Turkish is an example, I think Spanish too. (That pattern is often called differential object marking.)

akam chinjir wrote: ↑Sun Mar 10, 2019 4:16 pm

bradrn wrote: ↑Sun Mar 10, 2019 3:26 pm Do you have any examples of languages which do this? I haven't come across any so far.

You could look this paper---it's about a Bantu language named Ruwund, but starts with a bunch of references.

Hungarian is supposed to be another example.

There are also case-marking languages in which direct objects get case-marked only under similar circumstances, Turkish is an example, I think Spanish too. (That pattern is often called differential object marking.)

Thanks! I'll definitely have to look at this later.

I feel like I'm only poorly impersonating a conlanger because I'm too indecisive and unable to come up with ideas to actually Make A Finished Conlang.

For example, I can't even decide if long vowels in Classical Āirumāli should have an inherent tone, and if so what (the language otherwise has three tones: hígh, medium, and lòw. Currently, long vowels have an inherent falling tone on ConWorkShop, but CA is a register tone language aaaaaaa.)

Halian wrote: ↑Mon Mar 11, 2019 2:11 am I feel like I'm only poorly impersonating a conlanger because I'm too indecisive and unable to come up with ideas to actually Make A Finished Conlang.

For example, I can't even decide if long vowels in Classical Āirumāli should have an inherent tone, and if so what (the language otherwise has three tones: hígh, medium, and lòw. Currently, long vowels have an inherent falling tone on ConWorkShop, but CA is a register tone language aaaaaaa.)

Most language with register tone system actually registers its tone mora-by-mora, not syllable-by syllable. So, it's perfectly possible. Now add with contour restriction (a mora must have a different tone than its adjacent mora), You may posit a language where the default tone for short vowel is H, and the default tone for long vowel is MM > ML.

It is very common for a language to combine applicative and causative into one suffix based on verb. Usually a transitive verb or an unergative verb gets an applicative meaning, while intransitive verb and unaccusative verb gets a causative meaning. How to derive that from a language that has neither?

Akangka wrote: ↑Mon Mar 11, 2019 6:16 am

Halian wrote: ↑Mon Mar 11, 2019 2:11 am I feel like I'm only poorly impersonating a conlanger because I'm too indecisive and unable to come up with ideas to actually Make A Finished Conlang.

For example, I can't even decide if long vowels in Classical Āirumāli should have an inherent tone, and if so what (the language otherwise has three tones: hígh, medium, and lòw. Currently, long vowels have an inherent falling tone on ConWorkShop, but CA is a register tone language aaaaaaa.)

Most language with register tone system actually registers its tone mora-by-mora, not syllable-by syllable. So, it's perfectly possible. Now add with contour restriction (a mora must have a different tone than its adjacent mora), You may posit a language where the default tone for short vowel is H, and the default tone for long vowel is MM > ML.

I'ven't much of a clue how to distinguish morae from syllables; the current phonotactical rules of CA state that the latter is (C)(L)V(C), where L is a liquid and V can be a liquid if L isn't present.

Halian wrote: ↑Mon Mar 11, 2019 10:20 am

Akangka wrote: ↑Mon Mar 11, 2019 6:16 am

Halian wrote: ↑Mon Mar 11, 2019 2:11 am I feel like I'm only poorly impersonating a conlanger because I'm too indecisive and unable to come up with ideas to actually Make A Finished Conlang.

For example, I can't even decide if long vowels in Classical Āirumāli should have an inherent tone, and if so what (the language otherwise has three tones: hígh, medium, and lòw. Currently, long vowels have an inherent falling tone on ConWorkShop, but CA is a register tone language aaaaaaa.)

Most language with register tone system actually registers its tone mora-by-mora, not syllable-by syllable. So, it's perfectly possible. Now add with contour restriction (a mora must have a different tone than its adjacent mora), You may posit a language where the default tone for short vowel is H, and the default tone for long vowel is MM > ML.

I'ven't much of a clue how to distinguish morae from syllables; the current phonotactical rules of CA state that the latter is (C)(L)V(C), where L is a liquid and V can be a liquid if L isn't present.

tbh, it's something that varies language-by-language. In japanese, mora is either CV, N (nasal coda), or Q (sokuon/geminate). However, in Navajo, coda don't make a new mora, or at least not relevant in tone assignment.

Do you know how to create a fusional language? My declension of Proto-Garaudawa looks too agglutinative to me:

Akangka wrote: ↑Wed Mar 13, 2019 7:55 am Do you know how to create a fusional language?

One thing that might not be obvious is that, once you have declensions, you can borrow affixes from common irregular words in some declensions but not others - so if you have a separate (presumably irregular but not too irregular) paradigm for the pronominals, you can borrow pronominal affixes for the first-declension NOM.PL but not the NOM.PLs of other declensions, or whatever