Examples of truly unique conlang features?

[Duns Scotus]

That is, features which do not occur in any known natlang.

[R.Rusanov]

The phoneme /zgzdɣʷ/ is, I'm told, rather rare in natural languages, but stands tall as the most common reflex of the rhotic consonant in some conlang I just made up

[Pogostick Man]

WeepingElf commented in my thread on Kgáweq' that he had no knowledge of any natlang analogue to the success affix. A few of the success affixes may be present to some degree (indicating conativity and evidentiality) but others indicate success, failure, or qualifications thereof.

[WeepingElf]

WeepingElf commented in my thread on Kgáweq' that he had no knowledge of any natlang analogue to the success affix. A few of the success affixes may be present to some degree (indicating conativity and evidentiality) but others indicate success, failure, or qualifications thereof.

Of course, I am far from knowing every natlang on this planet, so there may be a natlang precedent hiding somewhere in the mountains of New Guinea, the Amazon jungle or wherever obscure natlangs may hide.

For my own part, I haven't seen any degree-of-volition system as elaborate as Old Albic's, but Central Pomo at least gets close.

[Nortaneous]

I don't know of any natlang that has a voiceless fricated retroflex lateral flap [ɺ̝̠̊], which Hanheliubl does -- it's pretty rare, but it occurs in lhu:pó /ɺ̝̠̊uːˈpo/ 'priest' and ṭɯlhe:rsí /ʈɯɺ̝̠̊eːʁˈçi/ 'lanolin', and as the second element of a cluster in plo /pɺ̝̠̊o/ 'cloud'.

[احمکي ارش-ھجن]

My conlang Arkéan has a Lateral Trill /ɺ͡r/, this sound is not attested in any natural language.

[Nortaneous]

My conlang Arkéan has a Lateral Trill /ɺ͡r/, this sound is not attested in any natural language.

What's that? Can you pronounce it?

In your thread, I thought you meant a retroflex trill like in Toda.

Come to think of it, I'm not sure if the voiced equivalent of Hanheliubl lh is attested in a natlang either -- it's like the Warlpiri retroflex flap, except lateral. Oh wait apparently Pashto and Iwaidja have it. Pashto even has a nasal equivalent, which I guess could be an allophone of the Hanheliubl retroflex nasal.

[Salmoneus]

I try to avoid being 'unique', but a few things do slip in.
My current primary conlang, Rawàng Ata, has a complicated alignment - I'm rejigging the details, but it may well be unseen in natlangs (wouldn't bet on it though). The core of it is a split-AP transitive verb system where the syntactic subjects of some verbs are in the same case as the syntactic objects of other verbs. This is sort of like a Philippine alignment in terms of cases, but the 'voices' are primarily determined lexically, rather than by the speaker, so I think of it as split-AP and philippine languages as fluid-AP, paralleling normal split- and fluid-S.

My at-times-primary-but-currently-virtually-nonexistant conlang (by that I mean that I'm no longer attached to any particular words or phonologies or anything, but I continue to have an idea of what type of language it will be when I get around to revising it), Handorian, has a more definitely unique (though I still wouldn't bet my house on it) feature, whereby each noun has a one-syllable auxiliary, which takes the place of the noun in many constructions (so eg "I ate the chicken" but "I ate AUX on tuesday the chicken") or even takes the place of the verb in some situations. Diachronically, this is because the object was marked by initial reduplication, the reduplicated segment then got broken off and treated as its own word in some circumstances (and thanks to sound changes is no longer always just the same as the first syllable of the noun).

Nort: I once toyed with a language with a lateral trill, although it didn't get far enough that I'd claim it as a real conlang. It seems to me that I can make a sound that to me sounds different from a rhotic trill, and that I perceive subjectively as being 'L-like'. I don't know whether this would be considered phonologically a lateral trill, though. My version of it doesn't involve a central stoppage, but instead a central trill - the difference from the ordinary 'rhotic' trill is that the sides of the tongue are raised and the airflow at the sides seems to be more important to the sound. I can't tell if what's happening is that it's the whole of the tongue trilling (so the same as the normal trill but with a broader point of contact), or if it's that the trill of the tongue tip causes a rapid alternation between lateral fricative/approximant (when the centre is closed) and laterally-restricted vowel (when the centre is open). Since it subjectively sounds sort of like a lateral fricative coarticulated with a trill, I think it's the latter that's happening.

Broader lesson: throwing around words like 'lateral trill' doesn't really mean much unless you define exactly what you mean.

[احمکي ارش-ھجن]

My conlang Arkéan has a Lateral Trill /ɺ͡r/, this sound is not attested in any natural language.

What's that? Can you pronounce it?

In your thread, I thought you meant a retroflex trill like in Toda.

Come to think of it, I'm not sure if the voiced equivalent of Hanheliubl lh is attested in a natlang either -- it's like the Warlpiri retroflex flap, except lateral. Oh wait apparently Pashto and Iwaidja have it. Pashto even has a nasal equivalent, which I guess could be an allophone of the Hanheliubl retroflex nasal.

Wikipedia says one can make a lateral trill, but occurs in no known natural language, supposedly it's the sound you make one makes certain bird calls.
I can make a lateral trill, as the unofficial IPA symbols show, you make a lateral flap and then trill it. Though, for me, I just put my tongue in the "L" position and vibrate it like one does for a normal trill. While it does sound distinctly "L-like" I also find it sounds more rhotic near the end... But I do notice an albeit vague difference between an Alveolar Trill, and an Alveolar Lateral Trill.

I notice the difference when I say "lava" and "rava" and that is when I notice the difference. I also know the diference when I say "blood" and when I say "brood"

it sounds kinda close to this: http://www.youtube.com/watch?v=kr4_zYBRaWY

alas, I have no recording software to record my self.
there's a different IPA symbol for a retroflex trill.

[Duns Scotus]

My at-times-primary-but-currently-virtually-nonexistant conlang (by that I mean that I'm no longer attached to any particular words or phonologies or anything, but I continue to have an idea of what type of language it will be when I get around to revising it), Handorian, has a more definitely unique (though I still wouldn't bet my house on it) feature, whereby each noun has a one-syllable auxiliary, which takes the place of the noun in many constructions (so eg "I ate the chicken" but "I ate AUX on tuesday the chicken") or even takes the place of the verb in some situations. Diachronically, this is because the object was marked by initial reduplication, the reduplicated segment then got broken off and treated as its own word in some circumstances (and thanks to sound changes is no longer always just the same as the first syllable of the noun).

Wow! That's really cool, I like this a whole lot. It kind of reminds me of a more naturalistic version of this.

Basically, if I understand you correctly, every noun basically has its own unique pronoun-like form?

[Curlyjimsam]

My conlang Viksen has an ergative alignment that, due to the absence of case or agreement morphology, is expressed entirely through word order, and furthermore is only apparent in the subjunctive which has an SVO order (i.e. AVP for transitives and VS for intransitives), non-subjunctive sentences being VSO or SOV. I'm now fairly sure this sort of thing probably wouldn't occur in a real life language, though when I created this feature my intention was for it to be exotic but not implausible.

What I think is a unique orthographic feature is found in the writing system of the Kohol Empire, where vowel distinctions are shown by changing the colour in which the symbols are written (e.g. a word ka and a word ke would be written with the same symbol, but with the colour red in one instance and blue in the other).

[Yaali Annar]

My conlang 23454 is nothing but tonemes.

[Drydic]

What I think is a unique orthographic feature is found in the writing system of the Kohol Empire, where vowel distinctions are shown by changing the colour in which the symbols are written (e.g. a word ka and a word ke would be written with the same symbol, but with the colour red in one instance and blue in the other).

Red ink was used to show stressed syllables in Vedic Sanskrit. Not identical, but a similar idea.

[Rhetorica]

I'm planning on putting together an alien language where all nominals are treated like events or processes, and hence are subject to verbal inflections. E.g. to discuss a person you discuss their life, and noun specificity is encoded via the hypothetical case. I'm hoping to go for a language with just one lexical category, though there'll be a lot of gerunds if that falls through.

[Benturi]

I'm planning on putting together an alien language where all nominals are treated like events or processes, and hence are subject to verbal inflections. E.g. to discuss a person you discuss their life, and noun specificity is encoded via the hypothetical case. I'm hoping to go for a language with just one lexical category, though there'll be a lot of gerunds if that falls through.

I have a more or less similar thing in an otherwise vaguely defined conlang: Tense markers are clitics that can be attached to nouns and verbal nouns. In my case, it doesn't imply that all nominals are treated like events or processes. Diachronically those tense markers come from a verb meaning "exist" or "occur". So, for example, house-PRES means "There is a house." (etymologically "a house exists"), while singing-PRES means "Someone is singing." (etymologically "singing occurs"). Basic and verbal nouns are marked for possession, so "I have a house." is expressed as my-house-PRES ("my house exists") and "He has sung." as his-singing-PERF ("his singing has occurred"). I don't know if this occurs in any natural or constructed language, but that doesn't mean much. I'm sure there are several one-verb-conlangs out there (which generally means that there's only one fully conjugable verb and all other verbs have only non-finite forms).

[gach]

It's hard to be original in a realistic way. I have a few ideas that might come close but are directly inspired by features in real world languages.

- Kišta has a category of distanced verbs that appears in marking temporal and pragmatic distance to the action as well as politeness and inferential clauses. This feature was inspired by Golin which can mark its verbs proximal or distal to mark at least temporal and pragmatic distance.

- Kišta also has a focus based alignment in which, on top of a nom-acc pattern, core arguments falling into narrow focus get marked by a separate case with originally a local function. This is ripping off Tundra Yukaghir as well as secondary ergative patterns you see floating around in the world's languages.

- I have a yet unnamed language that has developed a rich noun class system for its object agreement out of incorporated nouns. Coincidentally there are some very specific classes that can function perfectly well without without their corresponding noun as well as lexical items that only exist as independent class morphology. This is based on expanding the ideas of the agreement patterns found in Yimas as well as blurring the line between agreement and lexical affixes.

- The article system of Nooníí kiskn is based on Salish features in that it it has deictic articles and makes the distinction between referential and non-referential. However, it also marks full blown definiteness and I don't know if any language has grammaticalised both referentiality and definiteness marking at the same time.

- Of course, Nooníí kiskn only has eight segmental phonemes and two tones which is less than any language on Earth. I consider this to be a lesser feat since it's only playing with numbers.

[Rhetorica]

I have a more or less similar thing in an otherwise vaguely defined conlang: Tense markers are clitics that can be attached to nouns and verbal nouns. In my case, it doesn't imply that all nominals are treated like events or processes. Diachronically those tense markers come from a verb meaning "exist" or "occur". So, for example, house-PRES means "There is a house." (etymologically "a house exists"), while singing-PRES means "Someone is singing." (etymologically "singing occurs"). Basic and verbal nouns are marked for possession, so "I have a house." is expressed as my-house-PRES ("my house exists") and "He has sung." as his-singing-PERF ("his singing has occurred"). I don't know if this occurs in any natural or constructed language, but that doesn't mean much. I'm sure there are several one-verb-conlangs out there (which generally means that there's only one fully conjugable verb and all other verbs have only non-finite forms).

Conlangery reviewed something called Fairylang a while back that marked the whole sentence's tense on the subject, i.e. "Bob-PAST walks," meaning "Past Bob walked." That, and a thread about specificity, convinced me it would be possible to pare everything down into events and phenomena.

[ol bofosh]

Hnaf's "volition/motivation/aspect" verb system seems unique. Not sure if any language inflects for inspired, on purpose, without thinking, by convention/peer pressure and because nature made me that way.

Also, I haven't heard anything quality-quantity inflection: a dog, some dogs, a type of dog, some types of dog.

[vec]

I think the craziest feature I have in any of my conlangs is Lomanin's word order to indicate tense feature, where VSO is past, SVO is present and SOV is future.

Imuthan's morphysyntactic alignment based entirely on specific theta-role = certain case is not attested in any conlang to the degree that it does in Imuthan. There is no overarching basic case alignment, and while Imuthan may seem roughly nom.-acc. (though with both cases morphologically conflated in the direct case). So if we take the verb krakai "break", Imuthan seems innocent enough:

kraka ê‘vôm ê‘lubêķ
break-PRES.IND DEF=woman-DIR.SG DEF=glass-DIR.PT
"the woman breaks the glass"

However, a verb of emptying such as vimei "give birth" casts the woman giving birth in the theta-role of source which requires the oblique case and the newborn as theme which takes the direct:

vim ê‘vômim to‘diņ
give_birth-PRES.IND DEF=woman-OBL.PL INDF=child-DIR.SG
"the woman gives birth to a child"

And a verb of experience such as kigi "to see" casts the person seeing in the role of experiencer (dative) and the object as theme (direct):

kig-er to‘diņņa
see=1S.DAT INDF=child-DIR.PL
"I see some children"

Icelandic does this to a limited degree with a handful of verbs; i.e. the subject of a few transitive verbs with experiencer subjects rather than agent subjects take accusative or dative on that argument instead of nominative. But other similar verbs do not, taking the nominative. (This is changing in favor of the oblique cases, the nominative subjects losing ground). So maybe in the future, Icelandic will be similar to Imuthan, but it isn't today.

If anyone else knows of languages that do this, I'd be very interested to know about it.

[finlay]

I do this thing in Yaufulti that I think is unrealistic and just slightly beyond the bounds of speedy human cognition, which is that each noun takes a prefix for an article, but a lot of these are perhaps underlyingly just a single consonant, which then need a vowel, so I end up reduplicating the first vowel of the root (I think I did something similar on my old conlang Panceor but with consonants). So far so plausible, but then what I do is that as a kind of redundancy/agreement thing, the whole prefix is repeated as a suffix to the verb, complete with the reduplicated vowel. I've no idea if this happens or can happen in natlangs.

Thus we get:
(1) Susupe tuyesute.
COUNT.INDEF-cat have-AGR.COUNT.INDEF\cat-AGR.1S
I have a cat.

(2) Sitisin tuyesite.
COUNT.INDEF-dog have-AGR.COUNT.INDEF\dog-AGR.1S
I have a dog.

I don't think this is realistic, though, and I'm considering changing it for something consistent, so that instead of the above I'd have:
(1') Susupe tuyesite.
(2') Sitisin tuyesite.

The other thing is that right now, only the order of the two suffixes serves to distinguish object from subject, so (3) and (4) are the same:
(3) Sitisin seten nanakasesi.
C.INDEF-man C.I-dog PFV-bite-AGR\man-AGR\dog
A dog bit a man.

(4) Seten sitisin nanakasesi.
A dog bit a man.
((4) is a marked order, though, and would place some kind of emphasis on the object in this case, I guess.)

However, (5) and (6) would be the opposite:
(5) Sitisin seten nanakasise.
C.INDEF-man C.I-dog PFV-bite-AGR\dog-AGR\man
A man bit a dog.

(6) Seten sitisin nanakasise.
A man bit a dog.
(This time (5) is the marked order because "sitisin"/"a dog" is the object)

It's kinda useful to be able to do this kind of syntactical acrobatics, but actually if I had the consistent suffix on the verb for any indefinite antecedent, it might end up like this:
(7) Sitisin seten nanakasisi.
A dog bit a man.

(8) Seten sitisin nanakasisi.
A man bit a dog.
(except that because this is semantically weird it could instead be interpreted as "a dog bit a man")

The other option is to have a different suffix for objective and subjective antecedents, so that for instance the suffix to agree with an indefinite object is "si" and with an indefinite subject is "sa", hence:
(9) Sitisin seten nanakasisa.
A dog bit a man.

Maybe.

[Imralu]

Ngolu has a closed class of nominals (essentially pronouns which can be used as articles), with the open class of content words being represented by verbs.

It also has a weird gender system which is basically a straightforward animate/inanimate division, but with an extra masculine gender which is reserved for adult men who have passed the initiation ceremony. The use of animate and inanimate genders also depends on the speakers gender, with initiated men referring to slaves and animals (other than pets) in the inanimate gender and people other than initiated men referring to slaves and animals in the animate gender.

Gender roles and taboos are strongly represented in the language, with some roles of certain verbs only being available for people of one grammatical gender or another, and when performed by another person, either the verb is marked for non-agreement or the argument is. The non-agreement marker can either be gu 'be similar to' or hui, which indicates a strong condemnation for breaking gender rules.

My other conlangs Ahu and Arahuan have a lot of these ideas in them too, to greater or lesser degrees.

I have a more or less similar thing in an otherwise vaguely defined conlang: Tense markers are clitics that can be attached to nouns and verbal nouns. In my case, it doesn't imply that all nominals are treated like events or processes. Diachronically those tense markers come from a verb meaning "exist" or "occur". So, for example, house-PRES means "There is a house." (etymologically "a house exists"), while singing-PRES means "Someone is singing." (etymologically "singing occurs"). Basic and verbal nouns are marked for possession, so "I have a house." is expressed as my-house-PRES ("my house exists") and "He has sung." as his-singing-PERF ("his singing has occurred"). I don't know if this occurs in any natural or constructed language, but that doesn't mean much. I'm sure there are several one-verb-conlangs out there (which generally means that there's only one fully conjugable verb and all other verbs have only non-finite forms).

This is very similar to how most of my conlangs work. However, tense has not been an obligatory marking in any of my recent conlangs and it's just a matter of the predicate being indicated.

In Ahu and Arahuan, which have quite similar grammar, content words are essentially nouns and the predicate is overtly expressed with what could be thought of as a copula. In Ahu, this is with i, which I gloss as PRED and in Arahuan it is with a prefix which also indicates grammatical person and gender and definateness, such as ki- (3s.DEF.INAN-) which means 'it is' or i- (3s.INDEF.INAN-) which means 'something is' or 'there is'.

Ahu:

• i teli | i laala
  PRED house | PRED sing.AG
  there's a house | there's a singer / someone is singing
  (Or with a contextually understood subject: it's a house | s/he sings/sang/is a singer)

Arahuan:

• i-muakki | a-haruanna
  3s.INDEF.INAN-house | 3s.INDEF.ANIM-sing.AG
  there's a house | there's a singer / someone is singing

Content words can have an 'oblique role', which is indicated in both languages with a suffix. The oblique role is essentially equivalent to 'of', so with semantically noun-like words, such as house, it indicates a possessor (eg. house of person) and with semantically verb-like words such as 'sing(er)' it indicates roughly an object (ie. singer of song).

Ahu:

• i teli-a-n | i laala-ha liáala
  PRED house-OBL-1s | PRED sing.AG-OBL song
  I have a house | someone is singing a song
  (Or with a contextually understood subject: it's my house | s/he sings/sang a song / is a singer of a song)

Arahuan:

• i-muakki-mu | a-haruanna-ka haaruanna
  3s.INAN-house-1s.MASC | 3s.ANIM-sing.AG-3s.INAN song
  I have a house | someone is singing a song

___________________________________________

In Ngolu, content words are essentially verbs and it is arguments that are explicitly marked rather than the predicate.

• mala
  be.house
  there is a house
  
  waya
  sing
  a man is singing (a different word, laha, is used for a woman or child singing)

An argument can be added. The possessive case is used to mark a legal possessor. The accusative case to mark an object.

• mala's una
  be.house POS.1s.MASC
  I have/own a house
  
  waya zi waja
  sing
  a man is singing a song

Imuthan's morphysyntactic alignment based entirely on specific theta-role = certain case is not attested in any conlang to the degree that it does in Imuthan. There is no overarching basic case alignment, and while Imuthan may seem roughly nom.-acc. (though with both cases morphologically conflated in the direct case).

[...]

And a verb of experience such as kigi "to see" casts the person seeing in the role of experiencer (dative) and the object as theme (direct):

kig-er to‘diņņa
see=1S.DAT INDF=child-DIR.PL
"I see some children"

[...]

If anyone else knows of languages that do this, I'd be very interested to know about it.

Sounds quite a bit like Ngolu. Ngolu verbs, however, always have a possible nominative argument role, although it is not always expressed, but what would generally be thought of as the subject may be in another case entirely. Similarly to Imuthan, experiencers are in the dative case.

• xeva's ene's imu kili
  see DAT.1s.MASC NOM.3p.SPEC.ANIM.REL be.child
  I see some children. (Speaker: Initiated man.)
  
  lai wa's ini
  be.romantically.loved NOM.2s.MASC DAT.1s.ANIM
  I love you. (Addressee: Initiated man. Speaker: Person who is not an initiated man.)
  
  tu ca mahu ene xu tia.
  be.finished be.not be.known DAT.1s.MASC NOM.3s.DEF.INAN be.that2
  I didn't know that (to which you refer). (Speaker: Initiated man.)

And sometimes apparent subjects are expressed by another case, such as a genitive case for verbs where the nominative role equates to a body part and the accusative role to an entity that has some action performed on it by that body part.

• ka'r omo wa's une
  be.going.to be.pair.of.eyes ACC.2s.MASC GEN.1s.MASC
  I'll be watching you. (Speaker: Initiated man. Addressee: Initiated man.)

I do this thing in Yaufulti that I think is unrealistic and just slightly beyond the bounds of speedy human cognition, which is that each noun takes a prefix for an article, but a lot of these are perhaps underlyingly just a single consonant, which then need a vowel, so I end up reduplicating the first vowel of the root (I think I did something similar on my old conlang Panceor but with consonants). So far so plausible, but then what I do is that as a kind of redundancy/agreement thing, the whole prefix is repeated as a suffix to the verb, complete with the reduplicated vowel. I've no idea if this happens or can happen in natlangs.

I have to say, I quite like it. I don't know if it fits with what if any diachronics you've done already, but just as an idea, this could come from a complex gender system like bantu gone mad, with a lot of noun classes. If at some point, nouns tend to get assigned to certain classes based on their initial sound rather than semantic reasons, you could get a system of nominal prefixes not too dissimilar to what you have, with verbal suffixes reflecting those nominal prefixes. Under further analogy, it could expand so that every initial syllable of a noun could be used as a verbal suffix, as though you ended up with a noun class corresponding to every possible initial syllable. Just an idea.

[Rhetorica]

Also, I haven't heard anything quality-quantity inflection: a dog, some dogs, a type of dog, some types of dog.

Lilitika has this:

kano (dog),
mí kano (one dog),
kanoi (multiple dogs),
kanino (the class of all dogs, normally used with the gnomic),
mí kanino (one class of dogs),
kaninoi (multiple classes of dogs).

-in- is pretty flexible; it can be used to indicate a complete group or an entire species depending on the context, but generally implies an "all of". So mí kanino might mean either a breed of dogs, a family of dogs, or all the dogs in a petshop, or perhaps something else entirely.

[finlay]

I do this thing in Yaufulti that I think is unrealistic and just slightly beyond the bounds of speedy human cognition, which is that each noun takes a prefix for an article, but a lot of these are perhaps underlyingly just a single consonant, which then need a vowel, so I end up reduplicating the first vowel of the root (I think I did something similar on my old conlang Panceor but with consonants). So far so plausible, but then what I do is that as a kind of redundancy/agreement thing, the whole prefix is repeated as a suffix to the verb, complete with the reduplicated vowel. I've no idea if this happens or can happen in natlangs.

I have to say, I quite like it. I don't know if it fits with what if any diachronics you've done already, but just as an idea, this could come from a complex gender system like bantu gone mad, with a lot of noun classes. If at some point, nouns tend to get assigned to certain classes based on their initial sound rather than semantic reasons, you could get a system of nominal prefixes not too dissimilar to what you have, with verbal suffixes reflecting those nominal prefixes. Under further analogy, it could expand so that every initial syllable of a noun could be used as a verbal suffix, as though you ended up with a noun class corresponding to every possible initial syllable. Just an idea.

I haven't really done any diachronics, but it doesn't really fit the same mould as a bantu system, because they're not noun classes but determiners. The other thing is that if you compound the noun (which you do every time you add an adjective, as adjectives don't really exist in the language), you end up changing the vowel sometimes.

For example, seten means "a man", tauka means "big", and sataukaten means "a big man" – or suzuyuten means "a crazy man", or sizipiten means "a fast man" – note that the vowel in the prefix sV- changes on a purely phonological basis, just like in English how a or an is chosen solely based on sounds.

I envisaged the diachronics being something along the lines of starting with a consonant-only prefix or preposition , like sten or staukaten, but then the language undergoing some kind of tightening of the phonotactic rules and requiring a vowel after most consonants. Or perhaps səten and sətaukaten before schwa was somehow deleted from the language. But at the same time, I wanted to do this the opposite way and make those two later stages of the language.

[ol bofosh]

Also, I haven't heard anything quality-quantity inflection: a dog, some dogs, a type of dog, some types of dog.

Lilitika has this:

kano (dog),
mí kano (one dog),
kanoi (multiple dogs),
kanino (the class of all dogs, normally used with the gnomic),
mí kanino (one class of dogs),
kaninoi (multiple classes of dogs).

-in- is pretty flexible; it can be used to indicate a complete group or an entire species depending on the context, but generally implies an "all of". So mí kanino might mean either a breed of dogs, a family of dogs, or all the dogs in a petshop, or perhaps something else entirely.

That's cool. Reminds me that I wrote down something about a "group of" suffix, which would turn "dog" into "dog pack" and "tree" into "forest".
And, now you come to mention it, if I suffix it onto "type of Noun/s" it could have a similar effect to yours.

Have you found anything similar in natlangs?

[2+3 clusivity]

A proximal/distal distinction in 2d person pronouns.