Random ideas: Morphosyntax

[Omzinesý]

What about objects agreeing with subjects? (Maybe this idea popped up before, if so, sorry). Diachronically, this could come from an auxiliary that was dropped in intransitive clauses but attached to the object in transitive clauses. The following examples are SVO but maybe SOV makes more sense.

We ru ti-zo.
boy admire woman-3SG.M
'The boy admires the woman.'

Pa ru ti-sy.
girl admire woman-3PL.F
'The girls admire the woman'

Ru ti-fi.
admire woman-2SG.F
'You (f.sg) admire the woman'

A reflexive possessive suffix could be another source of an object marker.

woman-his -> woman-ACC
woman-their -> woman-ACC
woman-your -> woman-ACC

[Creyeditor]

Maybe this could yield two different agreement marker classes that are used in different contexts. This could depend on the class of verb used for example. That could be fun: Subject agreement on objects depending on the class of the verb. Everything is connected

[Arayaz]

Klallam, a Straits Salish language, has a cool system that I haven't fully unpacked that yields this (it's V-initial):

1. c̓áʔkʷ cn. "I was/got washed."
2. c̓áʔkʷ-t cn. "I washed it."
3. c̓áʔkʷ-əŋ cn. "I was washed by someone."
4. c̓áʔkʷ-t-əŋ cn. "I washed."

Honestly, I'd appreciate some help understanding this, if someone here has a better grasp of how natural languages do morphosyntax (especially if you have experience with the Salish languages)!

[Salmoneus]

Klallam, a Straits Salish language, has a cool system that I haven't fully unpacked that yields this (it's V-initial):

1. c̓áʔkʷ cn. "I was/got washed."
2. c̓áʔkʷ-t cn. "I washed it."
3. c̓áʔkʷ-əŋ cn. "I was washed by someone."
4. c̓áʔkʷ-t-əŋ cn. "I washed."

Honestly, I'd appreciate some help understanding this, if someone here has a better grasp of how natural languages do morphosyntax (especially if you have experience with the Salish languages)!

Not knowing anything (well, almost nothing) about Salish, the obvious first theory is just that -t is a 1s agent marker, and that -əŋ is a 1s patient marker, with the combination of the two creating a reflexive. The absence of either indicates a univalent verb, which in this case has the experiencer as subject.

However, wikipedia, which by a weird coincidence uses the same examples, says that your translations are wrong. In fact:


c̓áʔkʷ cn < I got washed
c̓áʔkʷt cn < I washed it
c̓áʔkʷəŋ cn < I washed (myself)
c̓áʔkʷtəŋ cn < I was washed by someone

So here -t is an active transitiviser, -əŋ applied to an intransitive is an antipassive, and -əŋ applied to a transitive is a passive.

[Arayaz]

Klallam, a Straits Salish language, has a cool system that I haven't fully unpacked that yields this (it's V-initial):

1. c̓áʔkʷ cn. "I was/got washed."
2. c̓áʔkʷ-t cn. "I washed it."
3. c̓áʔkʷ-əŋ cn. "I was washed by someone."
4. c̓áʔkʷ-t-əŋ cn. "I washed."

Honestly, I'd appreciate some help understanding this, if someone here has a better grasp of how natural languages do morphosyntax (especially if you have experience with the Salish languages)!

Not knowing anything (well, almost nothing) about Salish, the obvious first theory is just that -t is a 1s agent marker, and that -əŋ is a 1s patient marker, with the combination of the two creating a reflexive. The absence of either indicates a univalent verb, which in this case has the experiencer as subject.

However, wikipedia, which by a weird coincidence uses the same examples, says that your translations are wrong. In fact:


c̓áʔkʷ cn < I got washed
c̓áʔkʷt cn < I washed it
c̓áʔkʷəŋ cn < I washed (myself)
c̓áʔkʷtəŋ cn < I was washed by someone

I realize I forgot to say what the affixes meant ─ sorry. -t is a transitiviser and -əŋ is called the passive. I'd guess Wikipedia uses those examples because most of the Klallam Wikipedia page is copied from Timothy Montler's Klallam Grammar, which is where I got those examples. And the reason Wikipedia's and my translations differ is that I reordered them, but forgot to also reorder the English. A goof-up on my part.

So here -t is an active transitiviser, -əŋ applied to an intransitive is an antipassive, and -əŋ applied to a transitive is a passive.

Sorry for my mistake, and yeah, it looks like you were right.

[Salmoneus]

Errr... so if you already knew what the affixes did, and obviously you're not confused about how they are realised because it's plain agglutination, then what did you need help understanding exactly?

[Arayaz]

Errr... so if you already knew what the affixes did, and obviously you're not confused about how they are realised because it's plain agglutination, then what did you need help understanding exactly?

The specific meaning of the passive, because while they call it the passive, by a strict definition you can't passivize an intransitive verb. But you've said that

-əŋ applied to an intransitive is an antipassive, and -əŋ applied to a transitive is a passive

which answers what I was confused about.

[Salmoneus]

Errr... so if you already knew what the affixes did, and obviously you're not confused about how they are realised because it's plain agglutination, then what did you need help understanding exactly?

The specific meaning of the passive, because while they call it the passive, by a strict definition you can't passivize an intransitive verb. But you've said that

-əŋ applied to an intransitive is an antipassive, and -əŋ applied to a transitive is a passive

which answers what I was confused about.

OK, but fwiw you can/can't passivise an intransitive exactly as much as you can/can't antipassivise it!

In a strict sense, both passive and antipassive are valency-reducing, and can't be applied to an already only univalent verb.

In a loose sense, however, we could talk about an antipassive that promotes a semantic agent to subject, and equally about a passive that promotes a semantic patient to subject, by taking these operations as applying to the implied dyadic structure:

- I ate [implied object: pork]
> pseudopassive >
- pork was eaten

and likewise:
- the water boils [implied agent: me]
> pseudoantipassive >
- I boil [sth]

This 'pseudoantipassive' is what's going on in Klallam there: I undergo washing > I wash [sth]. But it would be just as possible to have a pseudopassive: I wash [sth] > something undergoes washing

So I can see why you were confused by the same suffix being both passive and antipassive, but it's not actually because one is any more or less compatible with intransitive verbs per se.

[the next question is: is that suffix REALLY a dedicated 'pseudoantipassive' for intransitive verbs in Klallam, or can it also be a pseudopassive depending upon the semantics of the specific verb? Or is that impossible simply because there are no intransitive verbs in Klallam with agent-like subjects?]

[N.B. I don't know what the 'correct' terms for 'pseudopassive' and 'pseudoantipassive' are...]

[Creyeditor]

This reminds me of the impersonal passive in German.

German
Es wurde getanzt.
it AUX.PASS dance.PTCP
`There was dancing.'

Could one apply other valency reducing to intransitive verbs in the same fashion? How about a reflexive?

Pseudo-German
Er xysfe geschlafen.
he AUX.REFL sleep.PTCP
`He slept himself.'

Maybe this could mean that he caused himself to sleep? Maybe something else? And what about antipassive? Maybe it could do the same as passive in German but for unaccusative verbs?

Pseudo-German
Es zatge gefallen.
it AUX.ANTIP fall.PTCP
`There was falling.'

[Omzinesý]

Polysemy between voices and nominalizations.

Donald mour hom.
D bake cake
'Donald baked a cake.'

Hom ma-mour sa Donald.
cake PASS-bake by D
'The cake was baked by Donald.'

Ma-mour la fotah.
NMZR-bake be delicious
'The baked one is delicious.'

Donald sa-mour
D ANTIP-bake
'Donald baked [something].'

Sa-mour mour fotah hom.
NMZR-bake bake delicious cake
'The baker baked a delicious cake.'

Historically that is easily explained as desubordination: nominalizations develop to voices.

[Ahzoh]

Polysemy between voices and nominalizations.

Donald mour hom.
D bake cake
'Donald baked a cake.'

Hom ma-mour sa Donald.
cake PASS-bake by D
'The cake was baked by Donald.'

Ma-mour la fotah.
NMZR-bake be delicious
'The baked one is delicious.'

Donald sa-mour
D ANTIP-bake
'Donald baked [something].'

Sa-mour mour fotah hom.
NMZR-bake bake delicious cake
'The baker baked a delicious cake.'

Historically that is easily explained as desubordination: nominalizations develop to voices.

Repurposing verbal morphology to be used as nominal morphology makes sense, I can see it now: be_killed > [zero-derivation] > the be_killed (one)

[Omzinesý]

Polysemy between voices and nominalizations.

Donald mour hom.
D bake cake
'Donald baked a cake.'

Hom ma-mour sa Donald.
cake PASS-bake by D
'The cake was baked by Donald.'

Ma-mour la fotah.
NMZR-bake be delicious
'The baked one is delicious.'

Donald sa-mour
D ANTIP-bake
'Donald baked [something].'

Sa-mour mour fotah hom.
NMZR-bake bake delicious cake
'The baker baked a delicious cake.'

Historically that is easily explained as desubordination: nominalizations develop to voices.

Repurposing verbal morphology to be used as nominal morphology makes sense, I can see it now: be_killed > [zero-derivation] > the be_killed (one)

Historically grammaticalization proceeds from derivation, including word class changing derivation like nominalization, towards inflection. It's debated if voice is derivation or inflection but it's closer to inflection than nominalization.
If English was a zero-copula language, 'I am beaten' would be 'I beaten'.

[Porphyrogenitos]

I've been thinking about a generic-specific system like in some Australian languages lately, and also a collective/singulative number system, and just thought of a way to combine the two.

Following Wilkinson's (1991) grammar of Djambarrpuyng (Yolngu) and Gaby's (2006) grammar of Kuuk Thaayorre, I posit a class which I term "nomina" which consists of three subclasses: Generics, specifics, and adjectives.

Each of these may stand on their own, but semantically, generics serve to categorize specifics and adjectives, and adjectives serve to qualify generics and specifics. Syntactically, save for emphatic or other discourse-dependent contexts, the three classes always proceed in the order of generic, specific, adjective, which is the core of the noun phrase. Adjectives may freely combine with generics and specifics, but each specifics are fairly restricted with the generics they can qualify. A common discourse pattern is for a

So far this is basically in line with Kuuk Thaayorre.

Where my version diverges is how each subclass of nomina behaves with regard to number. In addition to serving as standalone nomina, generics also act much like greater Southeast Asian classifiers: They make specifics countable. On their own, generics are countable; their base forms are singular and they inflect for plurality. Specifics, by default, are collective or mass nouns: To be countable, or to refer to an individual, either they must be paired with a specific or they must take a singulative inflection. Adjectives work the same way with regard to number; when used as standalones their unmarked form is collective/mass.

Additionally, each noun phrase may only have one nomen with the feature [+count]. Or, whatever, I don't know if that's an appropriate use of a feature, but I like thinking of it that way. In this respect it is rather like the English determiner category, which may be instantiated by an article or demonstrative or by plural inflection: [+count] may be realized by the presence of a singular or plural generic, or by a singulative-marked specific or adjective. A [+count] adjective may never modify a [-count] specific.

Okay, to provide some examples, using some filler vocab created on-the-fly:

gwak 'ʜᴜᴍᴀɴ' (a generic)
yiw 'ᴀɴɪᴍᴀʟ' (a generic for land animals)
waʈa 'dog' (a specific)
likal 'old' (an adjective)

gwak
ʜᴜᴍᴀɴ
'a/the human/person'

yiw
ᴀɴɪᴍᴀʟ
'a/the animal'

waʈa
dog
'dogs (esp. as a collective)'

likal
old
'the elderly, (some/the) old ones'

gwak likal
ʜᴜᴍᴀɴ old
'a/the old person'

yiw waʈa
ᴀɴɪᴍᴀʟ dog
'a/the dog'

waʈa likal
dog old
'old dogs (esp. as a collective)'

yiw waʈa likal
ᴀɴɪᴍᴀʟ dog old
'a/the old dog'

yiwik
ᴀɴɪᴍᴀʟ-ᴘʟ
'some/the animals'

yiwik waʈa
ᴀɴɪᴍᴀʟ-ᴘʟ dog
'some/the dogs'

waʈara
dog-sᴛᴠ (singulative)
'a/the dog'

likalti
old-sᴛᴠ
'a/the old one'

waʈara likal
dog-sᴛᴠ old
'a/the old dog'

A couple more notes: When numerals are used, they require a [+count] nomen. Also, adjectives, unlike other nomina, can be modified by adverbs of degree. There are also at least some demonstratives or possibly articles, but they also do not fulfill the [+count] feature, though some of them may require it. I am thinking that there will be case marking, but it may be a clitic that follows the last word in the noun phrase.

The words I've given have a vaguely Australian vibe to them but I don't know for certain if I'll go in that direction.

[Arayaz]

Okay, I really like this idea that I had, but I'm not sure how naturalistic it is, or how I can make it more naturalistic, so I'd appreciate some advice.

Basically, in this language, nouns don't exist as a cohesive group. Instead, there's three sets of...classifiers, I guess? One set of words for shapes, one set of words for materials, and one set of words for position ─ one each of the first two, maybe more than one positional is allowed. So for example, "the wooden bowl on the table" might be:

šeki nam azau kead
open.top.vessel wooden resting.on.flat.surface raised.off.ground

But this full complex would only be used the first time it's mentioned; later, only the minimal information necessary is given:

I šeki nam azau kead, tažed šeki ue kia?
open.top.vessel wooden resting.on.flat.surface raised.off.ground / want-2sg open.top.vessel xor closed.top.vessel
There's a wooden bowl on the table, do you want it [the bowl] or the box?

I šeki nam azau kead, tažed nam ue ašea?
open.top.vessel wooden resting.on.flat.surface raised.off.ground / want-2sg wooden xor metal
There's a wooden bowl on the table, do you want it [the wooden one] or the metal one?

I šeki nam azau kead, tažed kead ue uadea?
open.top.vessel wooden resting.on.flat.surface raised.off.ground / want-2sg raised.off.ground xor on.ground
There's a wooden bowl on the table, do you want it [the one raised off the ground] or the one on the floor?

But again, I'm not sure how natural this is.

[Salmoneus]

I've been thinking about a generic-specific system like in some Australian languages lately, and also a collective/singulative number system, and just thought of a way to combine the two.

I'm kind of confused by some of your terminology. Why rename nouns as "nomina", which is just a translation of "nouns"? But more significantly, why call your specific nouns (the ones referring to specific individuals) "generic", and your generic nouns (the ones not referring to specific individuals) "specific"? Isn't that going to be an unnecessary source of confusion?

To be honest, it sort of looks like a Germanicesque language, where the information about definiteness and number have bleached off the noun entirely and are located instead on the article. Where the article is still able to stand by itself (as in old Germanic). Just that instead of having three genders to agree with, the article suppletively agrees with a larger number of noun classes? Which, I guess, does basically make it a classifier/counter system of some sort? It certainly seems realistic.

[Salmoneus]

Okay, I really like this idea that I had, but I'm not sure how naturalistic it is, or how I can make it more naturalistic, so I'd appreciate some advice.

Basically, in this language, nouns don't exist as a cohesive group. Instead, there's three sets of...classifiers, I guess? One set of words for shapes, one set of words for materials, and one set of words for position ─ one each of the first two, maybe more than one positional is allowed. So for example, "the wooden bowl on the table" might be:

šeki nam azau kead
open.top.vessel wooden resting.on.flat.surface raised.off.ground

But this full complex would only be used the first time it's mentioned; later, only the minimal information necessary is given:

I šeki nam azau kead, tažed šeki ue kia?
open.top.vessel wooden resting.on.flat.surface raised.off.ground / want-2sg open.top.vessel xor closed.top.vessel
There's a wooden bowl on the table, do you want it [the bowl] or the box?

I šeki nam azau kead, tažed nam ue ašea?
open.top.vessel wooden resting.on.flat.surface raised.off.ground / want-2sg wooden xor metal
There's a wooden bowl on the table, do you want it [the wooden one] or the metal one?

I šeki nam azau kead, tažed kead ue uadea?
open.top.vessel wooden resting.on.flat.surface raised.off.ground / want-2sg raised.off.ground xor on.ground
There's a wooden bowl on the table, do you want it [the one raised off the ground] or the one on the floor?

But again, I'm not sure how natural this is.

Maybe I'm missing something, but haven't you just glossed "bowl" as "open-top-vessel" and used this to claim your language has no nouns? Because if you just swap out every gloss of "open top vessel" for "bowl", all the rest of the grammar and your description of it seem to be equally applicable?

Is the idea just that the language doesngt distinguish between a bowl, a cup and a cauldron? That doesn't seem entirely implausible, although there must be a practical limit to how small you can make the vocabulary, because having to describe everything with very long sequences of adjectives and relative clauses ("open top vessel that you hold in one hand and drink out of or possible eat soup out of with a spoon" vs "open top vessel that you hold in two hands or leave on the table and almost always eat from with a spoon (or chopsticks, depending on the cuisine) these days although traditionally people also drank out of them", etc - just saying "cup" and "bowl" is a lot easier in practice).

[Arayaz]

Maybe I'm missing something, but haven't you just glossed "bowl" as "open-top-vessel" and used this to claim your language has no nouns? Because if you just swap out every gloss of "open top vessel" for "bowl", all the rest of the grammar and your description of it seem to be equally applicable?

I guess the system could equally be analyzed as nouns and adjectives being the same thing and just having this convention of shape/material/position.

Is the idea just that the language doesngt distinguish between a bowl, a cup and a cauldron? That doesn't seem entirely implausible, although there must be a practical limit to how small you can make the vocabulary, because having to describe everything with very long sequences of adjectives and relative clauses ("open top vessel that you hold in one hand and drink out of or possible eat soup out of with a spoon" vs "open top vessel that you hold in two hands or leave on the table and almost always eat from with a spoon (or chopsticks, depending on the cuisine) these days although traditionally people also drank out of them", etc - just saying "cup" and "bowl" is a lot easier in practice).

Nah, that's not the main idea. "Bowl" was just an example; the main idea is that:

nouns don't exist as a cohesive group. Instead, there's three sets of...classifiers, I guess? One set of words for shapes, one set of words for materials, and one set of words for position

Which, again, is kinda just noun-adjective identicality. It was inspired by some stuff I heard about Tzeltal, where the basic words for nouns usually refer to materials rather than shapes. I thought, why not combine the two systems.

[Salmoneus]

My point is that what I said about "bowl" also applies to all other nouns.

To put simply, you say you have "classifiers" that specify the shape of an object, rather than nouns, but my point is that "nouns" are mostly just things that specify the shape of an object, so this isn't a real difference. [99.9% of the time an "object shaped exactly the same as a cat" is a cat, and any difference can be chalked up to the inherent ambiguity of all words (it's not that the other cat-objects are in some way not really cat-objects, just that they're not the type of cat-object you'd normally assume).]

The real difference between real-world classifiers and real-world nouns is just that the former is always a relatively small and usually closed class, while the latter is a vast and open class. If your classifier class is as large as the class of nouns, then they're just nouns, and if it's significantly smaller then you're going to run into issues of ambiguity and extensive paraphrasis.

I'll also note, incidentally, that "open-topped vessel" isn't purely a description of shape, either. The concept of a "vessel" (something used to contain other things) is a functional and teleological one, not a shape-based one. For that matter, "topped" also encodes orientation. "Concave object" would be the purely shape-based classification.

----------------

So yes, I think this is just having a specific order of adjectives, and having adjectives be able to be used substantively.

In terms of the order of adjectives, it's certainly possible to have a fixed order for material and position adjectives, yes, just as English as fixed adjective orders too. [big red dog, not red big dog, except in unusual cases of disambiguation with non-normal stress to mark the abnormal syntax].

I'm skeptical whether a language would ever have a convention as specific as requiring all nouns to be accompanied by adjectives of material and position in all instances, though.

Imagine: "help, a shark is eating my leg!" - "I'm afraid your sentence is ungrammatical child, and so I do not understand. Please specify, what is the shark made out of and where is it located? Likewise, what is your leg made of and where is it?" - "Just grab the harpoon!" - "I'm sorry, but which harpoon? The steel one, or the bronze one?" - "you only have one fucking spear!!!" - "yes, but grammatically you must specify both the substance and the location!"

I think in practice people would often omit these words. After all, once you've said "the cat is lying on the mat", how often do you need to specify that the cat is a cat made out of cat-material and that the cat is lying ON TOP of the mat (not attached beside it like a clock on the wall) and that the mat itself is positioned horizontally on a surface? Do you need to specify that so often that it's actually work taking the time to always say?

[Ahzoh]

I'm thinking of a system where some verbs have indirect alignment and some verbs have secundative alignment. And there would be a separate instrumental and dative case (rather than a shared marker).

[Creyeditor]

My point is that what I said about "bowl" also applies to all other nouns.

To put simply, you say you have "classifiers" that specify the shape of an object, rather than nouns, but my point is that "nouns" are mostly just things that specify the shape of an object, so this isn't a real difference. [99.9% of the time an "object shaped exactly the same as a cat" is a cat, and any difference can be chalked up to the inherent ambiguity of all words (it's not that the other cat-objects are in some way not really cat-objects, just that they're not the type of cat-object you'd normally assume).]

The real difference between real-world classifiers and real-world nouns is just that the former is always a relatively small and usually closed class, while the latter is a vast and open class. If your classifier class is as large as the class of nouns, then they're just nouns, and if it's significantly smaller then you're going to run into issues of ambiguity and extensive paraphrasis.

I'll also note, incidentally, that "open-topped vessel" isn't purely a description of shape, either. The concept of a "vessel" (something used to contain other things) is a functional and teleological one, not a shape-based one. For that matter, "topped" also encodes orientation. "Concave object" would be the purely shape-based classification.

----------------

So yes, I think this is just having a specific order of adjectives, and having adjectives be able to be used substantively.

In terms of the order of adjectives, it's certainly possible to have a fixed order for material and position adjectives, yes, just as English as fixed adjective orders too. [big red dog, not red big dog, except in unusual cases of disambiguation with non-normal stress to mark the abnormal syntax].

I'm skeptical whether a language would ever have a convention as specific as requiring all nouns to be accompanied by adjectives of material and position in all instances, though.

Imagine: "help, a shark is eating my leg!" - "I'm afraid your sentence is ungrammatical child, and so I do not understand. Please specify, what is the shark made out of and where is it located? Likewise, what is your leg made of and where is it?" - "Just grab the harpoon!" - "I'm sorry, but which harpoon? The steel one, or the bronze one?" - "you only have one fucking spear!!!" - "yes, but grammatically you must specify both the substance and the location!"

I think in practice people would often omit these words. After all, once you've said "the cat is lying on the mat", how often do you need to specify that the cat is a cat made out of cat-material and that the cat is lying ON TOP of the mat (not attached beside it like a clock on the wall) and that the mat itself is positioned horizontally on a surface? Do you need to specify that so often that it's actually work taking the time to always say?

Just two tiny additions:

• Some languages obligatorily require specification of orientation, but only in some contexts. The example I have seen most often is in locative/existential copular clauses but Hoocak extends this to certain TAM contexts, IIRC.
• Some languages obligatorily require marking of shape on all nouns. Most of them are restricted to two or three shapes though, e.g. round vs. stick-shaped vs. disc-shaped.

Interestingly, Khoekhoegowab is a language that does both but IIRC not all combinations are licit.

---

Still, I think having three PoS that are basically nouns and can be used together (but are semantically just that) still means exactly that. You have three subclasses of nouns and frequent compound-like structure with a rigid order based on these semantic subclauses.