Features found only in conlangs

[Porphyrogenitos]

I've been thinking of typological features that, while fairly common in conlangs, are unattested (or nearly so) in natlangs - especially features that are seemingly "naturalistic".

The prime example I've thought of is the alignment system known as the conlang trigger system, which is inspired by certain Austronesian alignment systems, but is not actually found in any real language. It works, apparently - it's speakable and human-usable - but it's not really "natural".

Another good example - though somewhat of a border case - is the vowel system /a e i o u y/. It's very common in conlangs, but it only occurred in a particular stage of medieval Greek and I think one or two other small languages.

More features that tend to appear in conlangs - mainly weird auxlangs and engelangs - are antonyms formed by word reversal, and self-segregating morphology. These, of course, are blatantly artificial.

Can anyone think of more examples? I'd be interested to hear them. (I know there must be one about gender and noun class systems, but I'm not sure what it'd be.)

[WeepingElf]

Here are some blatantly artificial traits:

Oligosynthesis: there is only a closed class of lexical roots (from several dozen to a few hundred), from which all other words are formed by compounding.

Taxonomic vocabulary: specific words are derived by adding affixes to more generic ones, such as ka 'animal', kaf 'fish', kafi 'shark', kafir 'hammerhead shark'. Popular in the 17th century, but there are later examples, too.

Arithmographic vocabulary: semantic primes are mapped to prime numbers, more complex concepts are formed by multiplying the primes. E.g., if 'life' is assigned the number 2, all living beings have even numbers, and all inanimate things have odd numbers. The numbers are then converted into speakable words by some kind of pronunciation rule. First envisioned by G. W. Leibniz.

Logical languages: Grammar based on a system of formal logic.

[vokzhen]

Verb-initial languages don't correlate well with robust case systems. Two- or three-case systems aren't too uncommon, but more than four is restricted entirely to a few languages in the Uganda-South Sudan-Lake Turkana region, I believe. There's some other rarities when it comes to V1 languages that I assume a lot of people miss, like a transitive possessive verb (X has Y is almost always rendered as an intransitive, with the possessor as a non-argument) or using copulas in nominal and adjectival predication.

There's a few phonological contrasts that are incredibly rare or non-existent in natlangs. One that stands out is more than one voiced uvular /ɢ~ʁ~ʀ/ which I've only found in Tsakhur-Rutul and Nivkh, the latter of which [ɢ] exists but is very rare and likely an allophone of /q/. Almost every natlang with a full voiced-voiceless stop-fricative system has a gap at /ɢ/. Another I've mentioned before, afaik /tɬ/ exists without [ɬ] in a single dialect of a single language, which the other dialect "corrected" by deaffricating it. I believe /tɬ'/ is also dependent on [ɬ]. /q qχ/ contrasts are non-existent as far as I've found, as are /c cç/ (though a contrast between a non-sibilant and sibilant front palatal /c(ç) tɕ/ does occur extremely rarely, e.g. rGyalrong and Komi).

[Pabappa]

Are you sure about the /qχ/ thing? It could just be considered a normal cluster.

Arithmographic vocabulary: semantic primes are mapped to prime numbers, more complex concepts are formed by multiplying the primes. E.g., if 'life' is assigned the number 2, all living beings have even numbers, and all inanimate things have odd numbers. The numbers are then converted into speakable words by some kind of pronunciation rule. First envisioned by G. W. Leibniz.

Thats very intriguing. I remember in 1994 I created a conlang called Moonshine where every phoneme had a numeric value and you could make new words by adding the numbers together. e.g. išestád "muscular dystrophy" was compounded from be + i + ē + mi + ni + stód, or something. A better example perhaps is just the end of that same word, where i + stód combine into stád. THis was pre-Internet for me, so I didnt realize that this was an unnatural feature. I figured, "What? Hebrew and Greek had num,bers, right?" And I added tone without realizing wehat it was, i trhought it was some sort of vowel heihgt thing. Labials had a value of 0 and were often added to words to break up vowel sequences.

Anyway,

from that I developed a conlang with no name (i call it "Gaze" now but that was actually the religion) where i planned to replace the numbers with prime numbers because it was more beautiful. I also had a plan that the lasnguage would have pictographs, and I dont really know what I was planning to do, or if it was even possible, but I was going to have those pictographs on graph paper with each phoneme corresponding to a part of the image, so .e.g \ would be a pghoneme, — would be a phoneme,| woukld be a pghojneme, etc. I knew that this would create hilaruiously long words, like ḟɣ̟ʰř̥̏ðɣʰř̏þ̫̆ū̆̂ə̰̫̰̆ʷēɣḟ̇ðḟʰḟɣkʲ̥̥̫̇̆ɣʰɣʰkʲṃ̂̇̆̂ʰɣŋṧ̟̫ðṿ for "fish", but it was OK because i figured people would mostly use those words for things they otherwise couldnt drscibre. I kind of missed the point in that I was still using addition, not multiplication,m even though prime numbers are a prime choice for multiplication, but if I had stuck with the lkanguaghe more than a few months Im sure I wouldve figured that out.

/a e i o u y/ is in some dialects of basque, isnt it? ANyway I admit Ive been tempted by orthpgraphy to favor simple phonologies. In Poswa I have {a e i o u y} for the vowel system but the {y} is [ɨ], i.e. the samer as russian Ы, rather than french /y/. I wonder if LACK of features can also be added to thios list. e.g. Almost none of muy conlangs have diphthongs, and i think its because i hgonestly just dont like the messy way they occupy two vowe leltters instead of one. i hate digpraphs in general, and thats why e.gf. Moonshine has so many pohonemes but so many diacritics too because I dont like writing them with dipgraphs.

Anyway my excuse for any impossible idea is ANIMALS. I just say that human languages developed those traits because they live with animals that in some cases are more powerful than the humans, and humanbs have to do those things to make the animals understand them, or in some cases animals actually create4ed the traits firtst and humans followed.

[Nortaneous]

There's some other rarities when it comes to V1 languages that I assume a lot of people miss, like a transitive possessive verb (X has Y is almost always rendered as an intransitive, with the possessor as a non-argument) or using copulas in nominal and adjectival predication.

?

/a e i o u y/ is in some dialects of basque, isnt it?

One dialect of Basque, and "/y/" is really more like /ø/, even though it came from high vowels.

[vokzhen]

Are you sure about the /qχ/ thing? It could just be considered a normal cluster.

Probably close to half of the descriptions of a voiceless uvular occlusive I've run across - that is not just /qχ/ but also /q(ʰ)/ - involve heavy uvular friction. If a language has /qʰ/ or /q/ it's fairly often phonetically [qχ], at least in some positions and often everywhere. A small sample include Nez Perce (Plateau Penutian), Kathlamet (Chinookan), Kaqchikel (Mayan), Totonac, some Quechua dialects, Kabardian (Northwest Caucasian), Tsez, Avar, Archi and Hunzib (Northeast Caucasian), Ket (Yenesian), Lahu (Lolo-Burmese), Wolof and Serer (Atlantic-Congo), Iraq (Cushitic), Aari (Omotic), and Burushaski (isolate). I suppose strictly speaking a few languages have a [q qχ] contrast but it's underlying /q qʰ/, and a number of others have gone farther to /q χ/ or /χ q'/.

There's some other rarities when it comes to V1 languages that I assume a lot of people miss, like a transitive possessive verb (X has Y is almost always rendered as an intransitive, with the possessor as a non-argument) or using copulas in nominal and adjectival predication.

?

What's "?"? Verb-initial languages generally express predicative possession with intransitives + a non-argument possessor (as for me, there exists a book; there exists a book to/at me; my book exists; or I exist and/with a book) and allow predicate adjectives/nouns to inflect directly, juxtapose, or otherwise exist without a copula. There are some exceptions, Kwak'wala has a series of grammaticalized elements that encode 1/2/3prox/3med/3dist and take verbal inflection for equative predication (equative being "I am the doctor," instead of the general "I am a doctor," which just inflects like a verb), and if I'm not remembering the same thing twice I think another language in that area requires a bare pronoun to link the two elements in an equative.

[M Mira]

I wonder if LACK of features can also be added to thios list. e.g. Almost none of muy conlangs have diphthongs, and i think its because i hgonestly just dont like the messy way they occupy two vowe leltters instead of one. i hate digpraphs in general, and thats why e.gf. Moonshine has so many pohonemes but so many diacritics too because I dont like writing them with dipgraphs.

I think Old Japanese was one of them. Before the influx of Chinese loanwords, diphthongs weren't allowed in the language and in combinations where they would occur, a vowel would be deleted or a consonant added in between.

I don't the the prevalence though, and it seems that WALS doesn't have an entry for diphthongs.

[WeepingElf]

My main conlang Old Albic has two features which I haven't yet seen in a natlang, though I don't think they are unnatural, as both are extensions of patterns found in natlangs.

1. Vowels are autosegmental, behaving somewhat like tones in some African languages. Vowel features bind to morphemes, and vowels are inserted into the consonant string by phonotactic rules. This also results in vowel harmony and umlaut phenomena.

2. Degrees of volition - four of them - are marked by means of subject cases. This is an extension of fluid-S morphosyntactic alignment found in languages such as Bats or Guaraní.

[Nortaneous]

I don't know the details of vowels in Old Albic, but you might be interested in natlangs like Moloko: there's one phonemic vowel, /a/, and then an epenthetic schwa, and front/back is a word-level feature that can spread to roots from affixes.

[Porphyrogenitos]

I don't know the details of vowels in Old Albic, but you might be interested in natlangs like Moloko: there's one phonemic vowel, /a/, and then an epenthetic schwa, and front/back is a word-level feature that can spread to roots from affixes.

Wait, only one phonemic vowel?! I thought that the only language that qualified for that was Nuxalk, with the syllabic /w/ /j/ analysis...Is there any literature available on this language?

[WeepingElf]

I don't know the details of vowels in Old Albic, but you might be interested in natlangs like Moloko: there's one phonemic vowel, /a/, and then an epenthetic schwa, and front/back is a word-level feature that can spread to roots from affixes.

The autosegmental behaviour of Old Albic vowels is described here. One dialect, Ivernirin, goes even further: the vowel features affect neighbouring consonants, which therefore fall into three groups: neutral, palatalizing, labializing. The vowel positions are inserted into the consonant string as in Classical Old Albic, but the vowel colours depend on the neighbouring consonants. That dialect thus can be said to have no phonemic vowels. I call such languages where vowels can be predicted from knowing the consonants, vowel generator languages.

BTW, I wonder whether if it wasn't for the Afroasiatic family, triconsonantal roots would be considered "unnatural".

[vokzhen]

BTW, I wonder whether if it wasn't for the Afroasiatic family, triconsonantal roots would be considered "unnatural".

Probably, though upon deeper reflection this might be less profound than it seems. Trigger systems would be "unnatural" were it not for Austronesian. Apart from about eight small languages (Northwest Caucasian, Arrernte and Kaytetye, and Marshallese) vowels shedding their features entirely to surrounding consonants would be crazy. Without Khoisan, strident vowels. Without Semitic, pharyngealization acting as a phonation contrast (and along with Circassian, ejective > pharyngeal). Fused case-number morphemes is all but entirely Indo-Uralic. Tlapanec and verbal case. Bantu's extensive noun classes that's rigorously prefixed on every major word class. Dravidian and Australian for >2 coronal stop POAs.

[Porphyrogenitos]

BTW, I wonder whether if it wasn't for the Afroasiatic family, triconsonantal roots would be considered "unnatural".

Another thought - a lot of conlangers design triconsonantal or biconsonantal languages that really don't resemble Afro-Asiatic at all. I.e. these conlangs posit abstracted roots, à la K-T-B, as real things with an independent existence that vowels are actually "plugged into" - while in natural triconsonantal languages the "roots" are really just a product of the analysis of languages with very thoroughgoing processes of analogy. If I understand correctly.

[KathTheDragon]

BTW, I wonder whether if it wasn't for the Afroasiatic family, triconsonantal roots would be considered "unnatural".

Probably, though upon deeper reflection this might be less profound than it seems. Trigger systems would be "unnatural" were it not for Austronesian. Apart from about eight small languages (Northwest Caucasian, Arrernte and Kaytetye, and Marshallese) vowels shedding their features entirely to surrounding consonants would be crazy. Without Khoisan, strident vowels. Without Semitic, pharyngealization acting as a phonation contrast (and along with Circassian, ejective > pharyngeal). Fused case-number morphemes is all but entirely Indo-Uralic. Tlapanec and verbal case. Bantu's extensive noun classes that's rigorously prefixed on every major word class. Dravidian and Australian for >2 coronal stop POAs.

A total lack of fricatives without Australian

[CatDoom]

Tamil doesn't have any fricatives either, though a few may appear allophonically between vowels and in loan words. This is presumably the case with some other Dravidian languages as well, since Proto-Dravidian is reconstructed with no fricatives, though most seem to have developed at least one through sound changes (as opposed to borrowing). There are also some languages from other parts of the world that have really small consonant inventories which lack fricatives as well.

[Nortaneous]

@purpdick http://www.silcam.org/documents/bow_thesis.pdf, five seconds on Google, also http://www.eva.mpg.de/fileadmin/content ... slides.pdf

it's less one-vowel-like than nuxalk, since there are 'prosodies' that color whole words for front/back, so [e a o] are /aʲ a aʷ/. you could think of it as vowel harmony if you were so inclined. you would know this if you had bothered to spend the five seconds it takes to type 'moloko vowels' into your search engine of choice

[kire kʷuro kəra] /kraʲ kraʷ kra/

[cromulant]

Fricativelessness is not really that uncommon. Australia has nothing like a monopoly.

[Porphyrogenitos]

@purpdick http://www.silcam.org/documents/bow_thesis.pdf, five seconds on Google, also http://www.eva.mpg.de/fileadmin/content ... slides.pdf

it's less one-vowel-like than nuxalk, since there are 'prosodies' that color whole words for front/back, so [e a o] are /aʲ a aʷ/. you could think of it as vowel harmony if you were so inclined.

Thanks.

you would know this if you had bothered to spend the five seconds it takes to type 'moloko vowels' into your search engine of choice

Well, I apologize. I guess this is my "let me google that for you" moment.

[Nortaneous]

does fricativelessness + >15-or-so consonants happen outside australia, most fricativeless languages have like ten consonants don't they

[WeepingElf]

@purpdick http://www.silcam.org/documents/bow_thesis.pdf, five seconds on Google, also http://www.eva.mpg.de/fileadmin/content ... slides.pdf

it's less one-vowel-like than nuxalk, since there are 'prosodies' that color whole words for front/back, so [e a o] are /aʲ a aʷ/. you could think of it as vowel harmony if you were so inclined. you would know this if you had bothered to spend the five seconds it takes to type 'moloko vowels' into your search engine of choice

[kire kʷuro kəra] /kraʲ kraʷ kra/

Interesting, though I haven't grasped the whole picture yet. At any rate, it does look quite similar to what I have going on in Old Albic - the "vowel features that bind to morphemes" can be called "prosodies", I think.

[cromulant]

does fricativelessness + >15-or-so consonants happen outside australia, most fricativeless languages have like ten consonants don't they

Marshallese with 22 consonants and no fricatives. Dinka with 20.

Yele Dnye has only voiced frics according to wikipedia, no frics according to WALS. Possibly the voiced frics are approximants, or there is allophonic ping-pong between approximants and voiced fricatives? And Yele has a monstrously large inventory.

[vokzhen]

does fricativelessness + >15-or-so consonants happen outside australia, most fricativeless languages have like ten consonants don't they

Marshallese with 22 consonants and no fricatives. Dinka with 20.

Yele Dnye has only voiced frics according to wikipedia, no frics according to WALS. Possibly the voiced frics are approximants, or there is allophonic ping-pong between approximants and voiced fricatives? And Yele has a monstrously large inventory.

This is one of those times where I think it's both useful and necessary to distinguish between /phonemic fricatives/ and [phonetic ones]. Marshallese may have no /fricatives/ but [zʲ] is a common sound (/tʲ/ between vowels or sonorants, including across word boundaries). Dinka has [c~cç~cɕ~ç~ɕ] apparently in free variation, and [ɦ] as an allophone of /ɰ/ before front vowels. Yele Dnye, in addition to having a set that's sometimes described as fricatives and sometimes as approximants, has [ɕ ʑ] as allophones of palatalized alveolars. Australian stands out that they tend to have no phonetic fricatives either.

[Nortaneous]

a trill inventory of /r ʀ/ is afaik only found in Kavalan

[Porphyrogenitos]

Hmm, I guess another obvious one I forgot about was part-of-speech marking on every noun, verb, etc (especially if through an invariant morpheme). Mainly found in auxlangs, but also in some naturalistic-ish artlangs I've seen.

[cromulant]

At every POA, voiced and voiceless plosives, affricates and fricatives. No gaps.