I'm doing more conlang research! Ten years ago I wrote my dissertation on madeup languages, and now I'm updating it for public release. Please fill out this survey to tell me what you know about conlangs and enter to win a 50% off discount code for quindriepress.com!
Even if you have no idea what a conlang even IS, that's valuable information to have! If you are reading this, you gotta fill out the survey, sorry, them's the rules.
anyways ive been thinking about changing sorkish latin orthography a little bit to get rid of the horrid e̋ that i always have to copy and paste everywhere
im thinking ë > ’ and e̋ > ë and since glottal stop is also marked with ’ i would just leave it unwritten, it barely appears in the language anyway
another thing that kinda bothers me are the digraphs but that would probably require new unicode abominations so i might leave them in. at the end of the day i love gh
The snow having fallen much today, it is necessary to place boots on myself. It is difficult to walk, for the snow is deep. I have arrived at the house very tired because of walking.
Everyone in Russia, Ukraine, and Belarus is supposed to have a tripartite name [consisting of a name, patronym and surname]. Single mothers may give their children any patronym[.] [...] In modern Russia, there are cases when [single mothers] give the child their own name instead of a patronymic. This practice is not recognized by law, but the civil registry offices may meet such wishes. A common loophole is when mother's name is a feminine form of a masculine given name, even an obscure one.
so apparently, patronyms are mandated by law in russia—yet russia doesn't recognise matronyms (for some reason. do east slavs not have matronyms? south slavs do), so the legally preferred option is for single mothers to.. lie? and assign the child a meaningless patronym? if this system is actually followed in practice, a substantial minority of russians have patronyms that are completely unrelated to their actual patrilineal ancestry. and like, okay, i imagine a society of insane misogynists would probably not be keen on recognising matrilineality—but did they not consider the option of having a single, stock patronym reserved for fatherless children, if exempting them from possessing a patronym is (for whatever reason) not an option? is everyone in that country retarded?
but over the years i've become more and more convinced that starting a conlang's sound system with a phoneme inventory is the wrong approach. there's too much emphasis on some kind of systematic approach to language phonemes.
that's the hot take: don't do this. phonemes aren't so immediately important when it comes to designing a language's sounds. start off with your phonotactic rules. abstract the "exact" value of the phoneme away (and the word "exact" is incoherent when it comes to phonemic analysis) and instead focus on dealing with broad classes of sounds. it's these features that you're actually trying to express anyway when you make a sound system, right?
there's definitely a strain of conlangery where you want to have a specific phoneme or two in your language and that's cool, honestly. flair is important. but often i see people trying to go for a feel with how their language sounds. and when you want to design a feel, it's best to start with the broad abstractions and then zero in on the surface realisation from there, imo.
how i go about things
start with syllable structure ofc. consonants and vowels. your classic CVC stuff. Or maybe CV. or maybe you want to start off with insane clusters off the bat - don't worry about it, you can always change it later via sound shifts or other tinkering.
also important here is word stress. this is personal feelings advice but unless you're planning to do smth specific with stress then you don't need to go out of your way to define "how" stress works. just put it on the same syllable of every word for now if you wanna. maybe say some vowels are strong/weak and steal or can't take stress or smth.
you don't need to specify the exact vowel right now, that's the whole point of this post. you can abstract your features until you've built your foundation to that point.
for consonants it's about the same. be as broad and vague as possible before zeroing in on closer values. don't choose consonants like they're discrete phonemes that go in certain places on a chart. treat the chart like one big consonant space that you're going to chop up into distinct segments.
this is the true value of the IPA, in that it can tell you the ways that real humans like to distinguish things. but the IPA is designed to make as many cuts as possible, to highlight every single distinction that every known language makes at the same time. no language makes that much distinction.
like, for example: english doesn't care about the difference between the alveolar and retroflex places of articulation, not really. if you hit someone with a retroflex stop in english, the language and its speakers will interpret that as its "front area of the mouth stop," which generally defaults to alveolar and/or dental because those are the ones that require the least effort.
this isn't really "grouping together" different phonemes, this is not drawing a border between two phonetic qualities. a slightly different thing. at least to start, your language has one consonant (C) and one vowel (V). zero borders whatsoever. start drawing distinctions, your phonemes will naturally emerge like a sculpture cut from stone. if you'd really like a retroflex series - make a cut in that "front area of the mouth" ! add a distinction between "tongue straight" and "tongue curled" or smth. and that's that.
once you've made your cuts and have the vowel and consonant charts diced up into distinguishing sections, you can pick a phoneme from each one to "represent" the space. maybe the most common one, maybe something else if you wanna have a little fun. and then you can add that phoneme or two you really personally wanted into the thing.
whoever might have told you that phonemic inventories "needed" to be as balanced as possible has no idea what they're on about. natlangs only loosely adhere to that kind of "balance," and whatever is there is an emergent consequence of the fact that the cuts you made among various phonetic features are sort of naturally straight lines. if you're distinguishing the fricative manner from the approximant manner, of course you're going to have a "fricative series" and an "approximant" series.
and all of these are just approximations. sounds can enter and leave for any reason at any time. a bit of chaos is always good.
moving back to the phonotactics - remember how you did this before starting on your phonemic inventory? that's 'cause you can update this to be more specific as you get more specific with your conlang's sounds. maybe you've made a cut in vowel space and you want it to correspond to vowel harmony, maybe you've made a weird consonant cut (glides/nasals/approximants are famous for this) might have weird effects/placements on the syllabic level.
ultimately, phonotactics probably has a bigger effect on how your language sounds than the exact phoneme classes you carve out. my last bit of advice is to take biblaridion's advice and start with a proto-lang. for phonotactics, what that means is: if you want a bunch of consonant clusters, then start with a ton of small open CV syllables and begin the vowel loss. If you want the opposite, start with a bunch of consonant clusters and then reduce the ever-living daylights out of them.
it'll take like 10-20 sound changes at the absolute most to end up with a sound system that was the opposite of how you started. make your sound changes broad and abstract. again, think in groups, distinguishable by a common set of cuts, not individual phonemes. do this well, and you'll find that along the way you've ended up with a gloriously chaotic and slightly funky system that's lost the sense of human artifice that a fiat-declared sound system has.
this is all phonetics advice and i'm not even an ijo phonetics. but that last bit applies very broadly. if you work on your language in stages and evolve it over in-universe time, it'll build up a whole host of fun naturalistic irregularities. protos are good. but in fairness that's advice for "naturalistic" conlanging, another incoherent term. natlangs do whatever they want, there's no one single vibe.
even if you're not doing that kind of artlang - i feel like the steps outlined here are a slightly better picture of what languages actually do, at least compared to the mad-libs sound inventory nonsense that the IPA might dupe you into thinking is accurate.
or, rather, mad-libs sound inventory nonsense is not what the IPA is for. the IPA is for quickly notating as many sound distinctions as possible. even the close/"phonetic" [t] is just a broad representation of a common phonological distinction. IPA is a messy system.
a "phoneme" is just a collection of distinctive traits being pronounced "at once." a "syllable" is a slightly bigger and often more abstract collection of distinctive traits. so start with the abstract. work your way down into the concrete. all art starts with a feeling that you want to nail down; make sure you're not nailing at random out the gate.
Okay, so this glyph is elilla "honey", and I guess I'll have to talk about this. There's this joke on The Simpsons where Homer is writing up invitations to his barbecue. They look like this:
The image is an outline of a pig, and the text is "Come to Homer's BBBQ. The extra B is for BYOBB." Lisa reads this aloud then asks homer, "What's that extra B for?" Homer's response: "That was a typo."
Back to Valyrian, elilla derives from the word for "bee" ēs. Ēs is irregular, as its stem is actually el-, so, for example, the plural of ēs "bee" is elossa "'bees". This is a common enough phenomenon in High Valyrian. Essentially, you had a stage where it was *els in the nominative singular, and the cluster simplified, resulting in a long vowel. (If anything, it's odd that "bee" is ēs rather than ēz, but we'll set the aside for the moment.) Elilla is a kind of derivation of ēs. The stem el- is extended, giving us elilla. Why -illa? No reason. It's a kind of reduplication. That extension, though, was reanalyzed as a suffix, and that suffix became rather productive, resulting in resultatives ending in -illa, hence hūra "moon" and hūrilla "moonlight".
Now, the trick here is all these words are aquatic. Aquatic words usually end in -r and have their own declension patterns. This one ends in -a. Fair enough. You might expect, though, that if it was aquatic, it would follow the declension patternj of aquatic nouns. But no. It doesn't. These -illa words end in -a, like class I lunar nouns, and they also decline like class I lunar nouns. The only thing that tells you they're aquatic is they trigger aquatic agreement.
Okay, so why are these -illa nouns aquatic? Because elilla is aquatic.
But why is elilla aquatic?
Truthfully, elilla is aquatic because it represents honey, which is classified as a liquid. That's the only reason it's aquatic. I also thought there would be a number of these nouns that looked like one class but triggered the agreement of another class.
But then I didn't add more.
So it's just these. It's the noun elilla and every other noun ending in -illa that looks like one gender/declension class but declines like a different one.
Twelve years on, I don't hate it, exactly, but I also hate it. And we're stuck with it.
Rough map of the far eastern subcontinent, the contemporary spread of its Major language families, and the (VERY, VERY rough strokes of the) three human migrations that formed the majority of this landmass' population (and initiated the spread of these languages).
There were three major major human settlement events in this area, two of which occurred during the paleolithic period. These dispersals occurred over a scale of centuries/millennia, and not all at the same time.
Peoples derived from the same settlement group share at least some degree of 'recent' (on the scale of millennia) common ancestry, and descendants of separate settlement groups have at times merged (IE the North Wardi are of mixed proto-Finnic (northwest settlement) and proto-Wardi (south seaway settlement) descent on a population-wide scale (reflecting a complete merger of previously separate populations, rather than mixed ethnicity on individual or sub-population scales)).
-SOUTH SEAWAY SETTLEMENT: Likely the first human settlement event here. Emerged across former land bridge that fully connected this landmass to the rest of the supercontinent. Initiated the spread of the North-center seaway, Dain, and Viper language families.
-NORTHWEST SETTLEMENT: Emerged from across the far northwest Inner Seaway via sea travel. Initiated the spread of the northwest-seaway and north sea language families.
-SOUTH SEA SETTLEMENT: The most recent settlement event. This originated with a group dispersing via sea travel over a 50-100 year period, fleeing rising sea levels in a formerly extant large island chain to the southeast. This is the only settlement event that has any form of historical documentation, via the Migration Cycle body of stories preserved among the Yanti people. Initiated the spread of the South Sea language family.
---
A language family refers to groups of languages descended from a common ancestor, not mutual intelligibility (IE: The Highlands, Finnic, and Keppeji languages are all from the same language family. A Highlands speaker might recognize cognates in the Finnic language, but could not comprehend a Finnic speaker, LET ALONE a Keppeji speaker). Each can be further subdivided into narrower language families (IE: Wardi/Wogan languages form their own family with closer common ancestry than to other Viper language family groups)
The language map should not be taken dead literally and shows approximate areas in which each family is spoken Predominantly, with no nuance for areas of linguistic convergence or the influence of outside language families. It also does not include comparatively minor language families developed in isolation, or major language families with very few representatives on this landmass (ie: Jazaiti language is of the major White Sea language family, not shown on this map).
Here's a (by NO means exhaustive) list of established peoples who speak languages from each family predominantly:
Northwest Seaway language family: Keppej, Korya, Moorlanders, Finns, Hill Tribes, and North Wardi.
Dain language family: Royal Dains, Dainach, Tho-Tykoso, 'Sea Dains', The Floating House, Hrolje.
North Sea language family: Varkhata-Byla, Rodi-Byla, Uswa-Byla, Urswali, Ursvali.
Viper language family: Wardi, Wogan, Cholemdinae, Ubiyan, Askosi, and Uboe.
South Sea language family: Yuroma, Highland Yuroma, Yanti, Losurrwe, Ur-Yamse.
North-center seaway language family: Buweni, Thingarri.
Der Minnesang war eine Kunstform, bei der Ritter ihre Kompetenz als Mitglieder der höfischen Kultur bewiesen, indem sie, in hoch ritualisierter Form, Liebeslyrik an eine hochgestellte Frau verfassten und vortrugen. Sein Ursprung ist auf die südfranzösische Troubadourdichtung zurückzuführen. Die Praxis entwickelte sich in Italien weiter, wo man der Angebeteten, statt ihr Gedichte zu schreiben, Suppe kochte: die Minnestrone.
Some Russian neuter nouns, by virtue of referring to animate beings, have a distinct accusative in the plural like masculine animates do (that is, the same form as the genitive rather than the nominative). This is the case for:
насекомое "insect"
существо "being"
животное "animal"
чудовище "monster"
дети "children" (only plural)
And some archaic words. Note that in the singular, the accusative is still the same as the nominative, like regular neuters and masculine inanimates. I don't know the reason for this number-based split.
Do I have to read some Chomsky to be good at conlanging/linguistics? If so, which readings do you recommend? I’ve found the phonology part of conlangs the hardest to understand, I need all the help I can get. Thanks in advance
For conlanging specifically, I will recommend you do not read Chomsky.
Much of Chomsky’s work doesn’t lend itself to conlanging. He focuses on English examples almost exclusively and so much of his work is abstract mathematical models of UG. You’d be better off with many other authors: Haspelmath, Aikhenvald, Dixon, Hyman…
On phonology specifically, sure, Chomsky and Halle’s The Sound Pattern of English is influential, but it’s old and most of the valuable parts are baked into general linguistics courses, while feature theory has continued to march on. I can recommend authors based on specifics (if you touch tone or pitch accent you must read “How (not) to do Phonological Typology: The Case of Pitch-Accent” by Larry Hyman).
Some general things I recommend for any conlanger once you’re got some basics down:
Texts about describing language (Paynes Describing Morphosyntax, Dixon’s Basic Linguistic Theory)
Anything typology. Typology helps you define your options and typology papers are usually full of examples from a variety of languages.
Any descriptive grammar you can find. Just trawl the Language Science Press website and download any Grammar of … that catches your eye.
Let me know if there’s something specific you have issues with in phonology. I have some background there, but it’s one of those areas where I’m too far in to know what general recommendation to make.