Radio Inspire

How To Learn Sign Language

My First Conlang – How NOT to Make a Language


By now you may have seen my “How to Make
a Language” Series. The information presented in those videos
represents what took me years to learn through pure trial and error, and it was intended
to give any beginners a head-start by outlining all the steps to take when creating a naturalistic
conlang. However, just as important as discussing the
things you should do when making a conlang is detailing what you shouldn’t do when
making a conlang, as there are a number of common perils that beginning conlangers might
fall victim to. And what better way to teach you how to avoid
these than to let you learn from my mistakes. Today, we’re going to have a look at Thandian,
the first conlang I ever made… and, I’m not going to lie, this is going to hurt. This thing has been collecting dust on my
harddrive for years now, and to this day I still can’t stand to look at it. But now, I’m finally going to put aside
my shame and look back through this ill-begotten mess in the hope that anyone watching this
may avoid suffering the same fate that I did… and before Conlang Critic can somehow get
his hands on it. But before we get to the language itself,
let’s start with a little stalling – I mean, background. I first had the idea to make my own language
about 6 years ago. I was reading the Silmarillion by J.R.R. Tolkien, and the thing that caught my attention
most of all was the portrayal and application of the elvish languages, which were featured
pervasively throughout the text. I had already known that Tolkien had created
his own languages for his legendarium, but this was the first evidence I had that there
was actual structure behind it. It clearly wasn’t random gibberish, there
was very apparent systematicity to it, and it reinforced the feeling that Middle Earth
had an intricately fleshed-out history and background of which the reader was being exposed
to only a small fraction. And so I thought to myself, “wouldn’t
it be cool if I could do something similar?” At the time I had a worldbuilding project
I was working on for an RPG group I was involved with, and I thought I would try my hand at
creating a language for the setting to pull off a similar effect. Now, by this point, the only language that
I understood the grammar of on a technical level was Latin, since I’d studied it in
High school and tried to teach it to myself for a few years afterward. And so when it came to my own language, I
kept all the grammar of Latin intact and just changed the phonological form of the roots
and affixes, producing a one-to-one relex of Latin. The only aspect of grammar that I changed
was reducing Latin’s five declensions and four verb conjugations down to exactly 1 noun
declension and 1 verb conjugation, because why would I possibly want more than one? That would just be confusing. And so I happily went along coining new words
as I needed them and working with a grammar that I was familiar with. But that all changed when I went to Turkey
for the first time in 2013. It suddenly struck me one morning about 3
months before the trip, that I should try learning as much Turkish as I could in the
time I had, just to see how much I could actually accomplish. I had mixed success, but I kept on with it
even after I got back home and I’m still steadily learning it to this day. But in studying Turkish, I was exposed to
linguistic features and structures that I never would have thought possible; things
like agglutination, negative conjugations, question auxiliaries, and evidentiality. This totally revolutionized my view of how
language could work. So I naturally decided to take all the new
features I’d seen in Turkish and include them in my language as well. As I continued learning Turkish, I started
developing interest in other languages, Greek, Finnish, Mandarin, and Classical Nahuatl to
name a few. And all the new grammatical concepts I came
across were copied whole-sale into my burgeoning language. And as I did this, I found more and more satisfaction
in developing the language further, to the point of neglecting all the other elements
of the world I was creating. My language was no longer just a small facet
of the worldbuilding process, it became a hobby independent of the project it was a
part of. I would spend hours ravenously searching through
Wikipedia pages looking for new pieces of grammar, and every time I came across something
I’d never seen before, I said to myself “well, I’ve gotta put this in there too…”
even if it made absolutely no sense with how the language worked already. Oh, what’s this?… Switch reference? That’s neat, I’ll throw that in as well,
even though I already have multiple systems of referent tracking… And what’s that? Some languages distinguish alienable and inalienable
possession? Well I’m going to use that too… even though
I’ve already got three different ways of marking possession… And this went on and on and on, until eventually
the language became so bloated and unwieldy, such a patchwork monstrosity crammed with
as many features as I could possibly squeeze in, most of which I didn’t even fully understand,
that it became impossible to keep track of which strategy to use in which instances,
and even very simple sentences became difficult to translate, until finally, I had to take
a step back and see my beloved creation for what it had become – utter trash. And after investing a year’s work into it,
I was forced to consign Thandian to an isolated corner of my conlanging folder where it has
festered ever since…until today. So lets take a closer look and dissect the
hulking corpse of this linguistic abomination to see what went so horribly wrong. Oh boy… what have I got myself into? Let’s start with the…uh… phonology,
if you could even call it that. This is the consonant inventory of Thandian. Does that seem familiar to you at all? Here, maybe this will help. Yep. Exactly the same as English. And the vowels aren’t much better either,
basically English’s vowel inventory with a bit less variation. This seems to have been one area where I was
apparently reticent to deviate from English. At the time my understanding of phonology
more or less boiled down to “there’s all the sounds we have in English, and then there’s
a couple of other weird ones, but I won’t include any of those because I don’t want
my language to be weird.” And of course, there was no allophonic variation
of any kind, at least none that I defined explicitly, and there were no rules defining
syllable shapes or phonotactics. I didn’t even know how the IPA worked at
the time, so all that was well beyond me. I would have thought coming up with a bad
romanization system for a language that has pretty much exactly the same phonetic inventory
as English would be basically impossible, but somehow I managed it. E, I, and O, for [ɛ], [ɪ] and [ɔ] makes
enough sense, as does using a schwa for /ə/, but the good stuff ends there. [æ] contrasts with [a], so I decided to represent
the former with an A, and the latter with a double A. And if there were ever any doubts that I’m
a native English speaker, I decided to Romanize [u] and [i] with a double O and E respectively. And as if that wasn’t bad enough, I chose
to represent the diphthongs ei, ai, and ou with macrons. My god. But the worst part was actually in the consonants. Clearly, I was influenced by Klingon’s use
of mixed case for its romanization, so I decided to use the same trick for any consonants I
didn’t know how to Romanize, which turned out to consist of exactly one sound: the voiced
dental fricative [ð]. I wanted a way to differentiate it from the
voiceless dental fricative, which English spelling doesn’t, so I decided to Romanize
it… with a capital TH. Yeah. Again, this was the only place where capital
letters were used to distinguish sounds, so when looking at the language written down,
you’re immediately confronted with all the capital THs that stick out like a sore thumb
and it looks atrocious. What was I thinking?! The only part of the language that at least
kind of holds up is the script. It’s got a fairly consistent aesthetic to
it, and I went to the trouble of adding serifs to make it look like it was carved into stone,
which is what I was going for. You can tell I was trying for a sort of featural
script, where sounds with similar phonetic characteristics are represented with similar-looking
characters… except it’s kind of hard to encode phonetic information in symbols when
you have absolutely no idea how phonology works in the first place. So you can see that the sounds are grouped
based on which corner the intersection occurs in, so top left is post-alveolar fricatives
and affricates, top right is dental and alveolar fricatives, bottom left is alveolar and velar
stops, and bottom right are labial stops and fricatives. Of course, I just made these groupings based
on how similar I subjectively felt the sounds to be. I could have easily taken a quick glance at
the IPA to get a clearer idea of how the sounds were related, but for some reason I just didn’t
care to do that. I was far more concerned with expanding the
grammar. And speaking of the grammar, that’s where
things get even worse… After all the innumerable additions I pulled
from every language I could find resources for, this is how the grammar ended up. Yep. A list of stems on one side, and a list of
affixes on the other. Of course, it was all perfectly regular, and
didn’t have any features that I deemed “irrational”, like grammatical gender or any form of agreement,
but damned if it didn’t have everything else under the sun. Thandian has 6(!) different grammatical numbers:
singular, dual, paucal, plural, distributive, and collective, and these were marked on both
nouns and verbs. The verbal affixes include markers for the
subject and the object, and the object markers can also be used for indirect objects and
I guess other roles as well if you wanted. But all those same roles are also marked on
nouns via case marking, so there’s a whole lot of double-marking going on. Nouns are marked for one of 15 cases. I remember seeing Finnish’s grammar for
the first time and being blown away by the sheer number of cases it had. Latin’s case system, the only one I’d
known until that point and the one I had copied one-for-one, was miniscule by comparison,
so, not to be outdone, I decided I needed to have at least as many noun cases as Finnish. It’s a good thing I didn’t know about
Tsez at the time… for the sake of my own sanity. The cases include two separate genitives,
one for alienable possession and one for inalienable possession, one specifically to mark the topic
and comment of a copula, and two opposing trios of cases for discussing position and
motion in space, and the other for position and motion in time. This disentangling of time and space is actually
quite a persistent feature across the whole language. Like, for every adjective to do with dimensionality
there are two separate roots, one for space and one for time, which seemed like a good
idea when I first made it because they’re two different concepts and therefore they
should get two different words. Which would be fine, and in fact even kind
of interesting for an auxlang or an engelang, but not for a naturalistic language; I’m
not sure if a natural language has ever even existed that doesn’t describe temporal relationships
in terms of spatial ones. And there’s a bunch of things like that
that just come across as artificial, like how there’s three different words for good:
one that means morally good, one that means well or favorable, and one that means adept
or skillful. Now this isn’t necessarily unnaturalistic
and could be an interesting feature if there was etymology behind these words and how they
came to coexist, but here it just feels so forced and inorganic. I was clearly putting these differences in
just because I felt obligated to create as many distinctions as possible. Along those same lines, when I first made
the verb system, the past tenses consisted of the perfect, the imperfect, and the pluperfect,
ported over exactly as they were in Latin. But then I realized I didn’t know whether
to use the perfect or imperfect to translate past simple sentences in English, so I made
a separate past simple tense to cope with that. And then I realized that I had no way of handling
past perfect continuous sentences, so I made a new aspect for that, and then I learned
about languages that have tenses that distinguish between the near past and far past, so then
I had to make those tenses as well, and the simple, completed, and continuous aspects
could be applied to all of those too… again, it’s not necessarily unnaturalistic to have
a language with all these different tenses, but I wasn’t thinking about things in terms
of their evolution or usage, I was just shoehorning in as many affixes as I possibly could just
for the sake of it. The derivational affixes are remarkably English-y,
with many of them being labelled in terms of their English equivalents. There’s a nominalizing suffix that’s supposed
to mean the same thing as the ‘-ness’ suffix in English, so, turning adjectives
into nouns, but there’s another suffix that’s supposed to mean the same thing as the ‘-tude’
suffix, and another for the ‘-ity’ suffix, even though all three of these do basically
the same thing in English, and the only reason we have them all is because of various historical
shenanigans. And, looking at these notes now, I have no
idea whether any of these suffixes can be applied to any adjective, or if a given adjective
can only take one of the three. Speaking of adjectives, they might be the
most embarrassing element in all of this. From my countless hours spent pouring over
Wikipedia, the only real thing I learned about adjectives beyond the basics was that they
sometimes agree in case and number with the nouns they modify, which I knew I didn’t
want to do because that was, in my mind, “irrational”. But I felt like I had to have some sort of
special morphology for them, so I did a google search and found a powerpoint from a middle
school English class that claimed there were four types of adjectives: descriptive, resultant,
limiting, or verbal, so I just made an affix for each of those. And likewise, adverbs are classified as either
adverbs of time, space, manner, degree, or frequency, all of which get their own affixes. *sigh*… dear God, that is painful. And by the way, the verbal adjective is apparently
different from the participle, which is also different from the gerundive, and I couldn’t
tell you for the life of me what the difference is supposed to be. All of these decisions were simply motivated
by my insatiable compulsion to throw in as much morphology as I could, the end result
of which was essentially a poor man’s version of ithkuil if it was made by someone who had
no idea how language works who simultaneously insisted that it was all perfectly naturalistic. So, what can we take away from this? Number 1, Always keep your goals in mind. Whenever you set out to make a new conlang,
it’s critically important to establish your goals from the very beginning and let them
dictate every decision you make. The only objective measure of how “good”
or “bad” a conlang is how well it fulfils its stated goals. If your goal is to make a naturalistic language,
like mine supposedly was, then your number one priority should be to ensure that everything
you put in the language could plausibly occur in a natural language. This was my first mistake. Thandian was supposed to be a naturalistic
language, spoken by a fictional population of native speakers, but I didn’t actually
do anything to make the language appear naturalistic, and indeed, did things that made it unnaturalistic. I assumed what I was doing was naturalistic
because I was copying these features from natural languages, but that doesn’t make
a feature naturalistic in and of itself. For example, when I learned that in Latin,
the suffix “-que” conveyed the same meaning as the word ‘and’ in English, that was
all the justification I needed to make a whole series of “conjunctive affixes”, after
all, Latin does it, so why can’t I? But what I didn’t know was that the Latin
‘-que’ suffix was just the result of a separate word being spoken in the same breath
as the noun it goes with, which isn’t too different from what sometimes happens in English. Knowing this, my “conjunctive affixes”
now look entirely nonsensical. As you might expect, the best way to combat
things like this is to learn as much as you can about how natural languages work, not
just what their features are, but how those features come to be. Number 2, in the words of Antoine de Saint-Exupery
“A designer knows he has achieved perfection not when there is nothing left to add, but
when there is nothing left to take away.” This applies especially, though not exclusively,
to naturalistic conlangs. This is actually a common pitfall of many
beginning conlangers, sometimes called a Frankenlang, where one throws in everything they know about
language into one hideous amalgamation, and this is exactly the trap that I fell for. I was under the impression that putting more
structures and features into the language would make it more interesting, but that’s
really not the case. In fact, you can actually make a conlang more
interesting by omission. For instance, what if when making your language,
you decide that verbs aren’t going to have a morphological future tense? How will the language cope with future statements? Maybe it uses an auxiliary verb, maybe just
uses adverbs or time phrases, or maybe some combination of particles. Any of these would be much more interesting
than just making another affix. For every feature you introduce, always consider
how it interacts with what you’ve already created. And when encoding grammar, try seeing if you
can find a way to convey the meaning with what you’ve already got before creating
something brand new. And number 3, Practice makes perfect. Just like learning any new skill, when you
first get into conlanging, you’re more than likely going to struggle, but don’t let
that discourage you. It’s pretty much inevitable that you’ll
eventually have to seriously overhaul or even completely scrap a project you’ve invested
a lot of effort into. But it’s not a waste of time, because you’ll
have learned what to do differently for the next project. After I realized how terrible Thandian was,
I started a new language, one where this time, I was a lot more selective in what I decided
to include. and I knew from the outset that I wanted it to be as naturalistic as possible
It had things like irregularity, multiple paradigms for nouns and verbs, and a gender
system, … and it also sucked, but it was definitely a major improvement, so I made
another language, and then another, and another, and with every new project I got slightly
less terrible. Even today, while I have four or five languages
that I’m pretty satisfied with, I still have to go back and make revisions periodically. It’s a long, long learning process, but
if you stick with it, you’ll ultimately reach a level where all the innumerable facets
and decisions that once seemed so daunting become second nature to you. So, beyond any of the technical details, these
are perhaps the three most important adages for a conlanger to live by. Unfortunately for me, I had to learn them
the hard way, but hopefully you’re now better equipped to do what I could not. So go forth, then, and conlang! Now if you’ll excuse me, I’m going to
do something I should have done a long time ago… Ah… good riddance…

68 Replies to “My First Conlang – How NOT to Make a Language”

  • Video: This is the consonant inventory of Thandian, does this seem familiar you at all?

    Me: Um, nah.

    Video: Here maybe this will help! Yep, its english.

    Me: Still nah.

  • I think you have learned Turkish with wrong examples.
    13:25 we don't write "Ben kopeği görmıştım" nor "Ben kopeği görmelim"
    It has small letter mistakes.
    The correct ones are "Ben köpeği görmüştüm" and "Ben köpeği görmeliyim"

  • The alphabet I made is way different than the normal English one. The way it works is first I try making all the sounds I can possibly make. I just take the one that is easy to pronounce and not weird (I got 32 letters). Next I group those letters that sounds familiar to one another because my motive behind this language is to make one that is efficient and easy to learn. I was so annoyed how sometimes English pronounce words differently then the letters (cough, bough, through, though and thorough is a few examples). After I grouped them (I got 11 groups) I made the symbols of a letter in the same group slightly different than the other letters.

    For pronunciation of the letter d,b,t,p is in the same group. The symbols I came up with is just what I thought would be appropriate for the sound it make. The Dh sounds feels like a triangle so the symbol is just a triangle with a dot in the middle (so people won't confuse it with an actual triangle when doing geometry) then I just slightly change the symbol of the triangle. the Tuh sound is just the same symbol as the D but there is a straight vertical line above the triangle. The Buh sound is a triangle with a verticle line inside it and the puh sound is a triangle with a plus sign in it.

    After all of that is done, I then started coming up with words and the kanji symbol for it. This is heavily inspired by the japanese and chinese language and I wanted to make it efficient so I decided having symbols for a word would save a lot of time. I Also wanted to make the symbol look like the actual thing (kinda like the Chinese language) so the kanji symbol for fire looks a little bit like this (| () |) (I can't really draw it but it looks like fire). To me coming up with the name and kanji symbol for words is the most fun part. Just because I can I decided fire should be Fi+ ruh. Also you can always write it down using my language normal alphabet because these symbols is only meant to make writing something down more easier. The alphabet is meant to use it to write down words in other languages, write down words that doesn't have a kanji, write down how to pronounce kanjis and to write down names.

  • I feel like you were so caught up in making a cool and complex language you lost sight of the point of language, to easily transmit knowledge. Once you hit a certain point it just becomes confusing as hell

  • I literally have no idea what 90% of this video is about…
    Phonetic sounds? What are those?
    I don't even know what anything in any of the tables are… Still a good video though (I guess)

  • My story has 2 languages.
    Cojincan (the language most characters speak, including the main characters and will be in whatever language the story is)
    And zimstran, which i am making up.

  • Never heard of the word conlang before, I guess this was recomended because I saw a video about the origin of the spanish alphabet. This was a fun exposition of that trainweck tho haha

  • I've had a strange idea for a while about creating a language that completely maximises simplicity and intuitiveness. That means nothing irregular, inconsistent, redundant, unnecessary or deviant. So there'll be no synonyms or antonyms of any words at all. Like, there will be no words like "bad", "cold", "small", "empty" or "dark" since you could literally just say "antigood", "antihot", "antilarge", "antifull" and "antilight". And there will be no large sets of variation in rewording, like "small", "little", "tiny", "minuscule" or "mini", and for "tiny", you'll say "greatly antilarge" or something like that. I mean, I haven't even gotten started on the words, so they're going to be a lot smaller and less awkward than they are in English – usually so small in fact that you can expect any string of phonemes in the language to be a real word in it. Also, it'll be phonetically consistent and also have lots of phonemes to maximise the amount of words you can make in limited strings, but also nothing confusing like phonemes way too close to each other. That's my crazy idea and I have no idea why people go the opposite way and shove as many deviations into their conlang as they can, because that just maximises confusion and memory storage. Also there's going to be no pronouns or anything. I want to squeeze the vocabulary down to raw essence.

    Also…the Chinese people are dumb for needing a new symbol for literally every new word. That doesn't help in scientific and mathematical discovery aspects, like finding new chemicals or googologisms. Plus if you read a word you don't know without access to pen and paper, you won't even be able to ask anyone what it says since you can't even tell how it's pronounced.

  • What I've done while I try making the native language in my epics in the background is use real languages and use the character names as both a placeholder and testing ground, as I know what the character names are meant to mean I hit up google translate and name suggestions in varying languages and see if the words come of as natural in conjunction with each other, which basically means I can't use Chinese or Japanese while I managed to get Welsh and Hindi to fit.
    Fun Fact: putting two letter suffixes in English adds enough letters to most other languages for another word. And Romantic languages apparently have the same word for loyalty written in the English way of spelling.

  • Welsh: * Has 2 dental fricatives *

    Welsh: Newn ni alw'r un lleisiol yn 'dd' a'r un arall yn 'th'.

    Thandian: TH

  • Hey, it's still not as bad as Vasudan where your vocabulary changes depending on your age, gender, social standing, social standing of the person you're speaking to, and your location relative to the emperor.

  • No idea what a conlang is but this is interesting. I'm interested in creating a language for magic in a book I'm writing, and will likely use your guide to help me.

  • Guys, I have been trying to create a conlang-a logical one.the problem came when I reflected existing words in own language that I speak and the my conlang to-be.it turned out that the musics of words or accents and intonation was really hard to decide upon.any idea how to deal with it?

  • I may not understand most of what you talk about here, but what I can tell is that the lesson learned from this experience can be applied to ANY creative project.

    Don't cram in features for the sake of it. Have a vision of what is the most quintessential form of what you want to make, and then concentrate on making just that, with only the features that you feel are the most crucial.

    Once you've done that, you can step back, observe what you created, and evaluate whether adding more to it is really necessary, and if yes, then what features are the most needed and/or would genuinely enhance the project (instead of warping it into an unfocused Frankenstein monstrosity).

  • Huh. I have a story I'm working on with multiple languages – including hybrid languages where two races live together – and so far I've just made a conversion system for one of my languages that changes the order of words, changes the letters, and has a few extra rules. It works, but now I'm worried it comes off as lazy, despite the work that went into the conversion system.

    Been distracted making my world map, anyway. Mapping out the world really feels like the first step for me.

  • No idea what you're saying half the time, but my "language" is just English with extra characters. Only written too.

  • Yeah… i didn't really understand much of anything going on here, but i find it interesting nonetheless. This level of confusion and extreme need to understand how language works is why i pretty much abandoned all efforts to create a language for my own uses, fun tho it may have been, it's ultimately is a distraction from actually writing in the universe and whatever story you're going for….

  • literally don't understand a single word of this, he might as well be speaking hindi. so i'ma just leave this before all my time is wasted on something im tooo dumb to understand

  • I actually like the whole <ee> <oo> <aa> thing for /i:/ /u:/ and /a:/* insofar as it reminds me of the earlier modern English romanisations that occurred globally, but then when you mentioned macron-a* for /ei/… OK, I now see the point you're trying to demonstrate 😉

    (plus the whole ee oo aa and aa vs a thing is not very special, of English's orthography-breaking sound changes those are quite tame, so I find them easier to justify)

    *can't use proper IPA and diacritics on this particular keyboard, I mean the open back~central vowel not the open front vowel here.

  • I feel like im the only non-linguist who's stumbled across this video because I don't know more than half of the words used in this video. Dipthong? the fuck

  • To be fair, while it was horrible, it was most certainly extremely valuable to your later stuff. Without Thandian, you wouldnt have had so much hand on experience on so many weird and niche things that, while at many times you may of not of known how worked, still gave you a foundation to build on for your later projects. :3

  • Lol I did this when I was 8/9 I used my native language then added modified English words with extra words with speech which was just modified bangla
    Shittiest "language" I've ever made

  • I almost made a frankenstein conlag but luckly I realized that my lenguage was going in a awful direction and I abort that… Thing…
    Now I use its ashes and scrap to create a new and better lenguage for my goal: made a not to dificult lenguage to learn but easy to use that does not needs to clarify it self

  • Being a native Finnish speaker, i also fell into the trap of adding too many cases in my first conlang. I don't have it on record, but i remember the feeling very clearly.

  • ah yes the time honored conlang tradition of basing your adjectives off of the opinions of some random middle schooler

  • I’m making an engineered language to be as terrible as possible (similar to Kay(f)bop(t)), with the specific intent to anger Jan Misali as much as possible when I ask him to review it. This is a nice guide as to what I should do.

  • That distinction between Space and Time would be really cool in a language for beings that could move through time, though. Like that’s actually really, really cool

  • 8:24 that what Native American languages look like I see the writing every day they must have come up with a writing system later on because it was mostly just spoken

  • my secret language is just the letters shifted to the right and then reversed

    pretty hard to crack, actually, if you don't have too much secret language skills

Leave a Reply

Your email address will not be published. Required fields are marked *