How to learn to make a conlang?

Question

No, I'm not asking for resources. I've consulted every resource I can find, but even after more than a decade of research, I still have no conlang to show for it. I started with the Language Construction Kit 1 and 2, learned everything I could about linguistics, consulted every guide I could find, and I still don't know a damned thing. All I have to show for all this work is mountains of repetitive phonologies that occasionally have some grammar notes that never went anywhere due to them turning out to be impractical.

Clearly I'm doing something wrong; what else do I have to do other than research? No amount of research seems to help me. Obviously, I need to do something different, but what can I do? I've tried to make experimental conlangs, but that seems to just mean I accept that all my conlangs are going to hit a brick wall and then be abandoned rather than expecting it to actually go somewhere for once.

edit: What have I tried? I keep trying to use features I don't fully understand. For the longest time, I tried to make a language with fluid s alignment. I later found I didn't understand what that meant. I tried to make a VSO conlang, but also kept running into syntax problems I could not solve until I found out what VSO really was. Long story, time and time again, I can never make use of a feature I don't understand. What does it mean to understand? I can say what ergative alignment is, but I don't know the endless ramifications of having such a system. I only understand systems I've used myself extensively. What that means is I only know the features of English and German. Nothing else. I can define cases that German does not have, but I only know really know how to use the ones German has. I tried to solve this recently by mixing English and German, but I can't find anyway to make such languages function. Essentially, after all my studying, all I really know how to do is make a perfect relex of either English or German. It has to work 100% like one or the other. That is not conlanging; that's making a bad cipher.

I need to know how to actually conlang. How do I fully understand how to use any grammatical feature? The only way for me is to know and use a language that has it, nothing else works. I did contemplate trying to learn odd languages like Hungarian to fix this, but I've never been able to learn a language I had no use for. I've also been told that learning a whole language just to learn to conlang is highly excessive. What other way is there? How do I learn to use features I haven't had years of personal experience with? That is my problem; I don't fully understand anything. I can make a perfect replica of either English's or German's tense systems, but I can't make one that even slightly deviates from one or the other. I can use nominative, accusative, genitive, and dative cases, but forget anything else. I can use SVO order, and German's odd order, but I can't even make an SOV language let alone VSO and anything with an OS order is literally unfathomable to me.

Everything I've studied has not been sufficient; if I want to understand even one thing, such as a lative case just for an example, I would need a thick book that does nothing but explain it. I need to know every way it can be utilized, what it forces upon a language, because only then do I truly understand it. How can I learn to do this?

Answer 1

I usually hang out on Writing SE and see questions from people struggling to write a novel. The problem often seems to be that people are too critical of everything they write, so they can’t move beyond a phrase or a chapter, and the advice that’s usually given is to get the first draft down – even if it’s crap – and then rewrite it. Your issue sounds similar: you’re getting caught up in the details and trying to have it all come together at once, and not accepting that it won’t be perfect from the outset.

I have very little linguistics knowledge yet have managed to make a basic conlang for my novel that I can write simple sentences with. It’s far from finished, but I can see where it’s heading. The following might sound rather naive, since you've done so much research and work already, but perhaps it will help someone who is just starting out. This is what I’ve done so far:

1. I made up a few words for things important to the culture I’m creating. I used google translate for inspiration, looking at phrases in various languages, and realised I wanted lots of vowels and didn’t want certain consonants. The first words were pretty random but I gradually gained a sense of which letter combinations looked and sounded right. I’ll come back to this at a later stage and try to extrapolate some rules (at which point I will probably need to rewrite some words) but for the moment, it’s enough to move on with.
2. I decided on a phrase I wanted to translate and I chose OVS word order. I don’t know why, it just felt right. The good thing about it is it’s not like any of the other languages I know, so it has forced me to think creatively about grammar and not just translate word for word.
3. The phrase also contained an adjective, a plural, and subject and object pronouns so I had to think about how to deal with all of those. The rules I created were arbitrary, but since I don’t know any better, I didn’t get bogged down with fully understanding anything and I have absolutely no idea of the ramifications of any of my decisions.
4. I started translating more phrases, making words and rules up as I went. I'll be at this stage for some time, I expect. When things don’t work, I retcon rules and rewrite the phrases I’ve already done.

Perhaps I will get stuck at a later stage, but being willing to go back and change and rewrite as new requirements come to light seems important. For me, creating the language is like writing the novel: get a draft down on paper, it doesn’t have to be right, it doesn’t matter if it doesn’t all make sense. Then review and rewrite, over and over, amending the bits that don’t work, filling in the gaps and refining the stuff that does work.

In a comment, you asked how can you know if you will even like your language before you have the grammar finished? My answer is to make decisions that you like. I love my language, even though it’s in its infancy, because every decision I make is something that intrigues me or sounds good to me, and makes me want to explore it further. It’s your language, you can do whatever you want with it, so make fun decisions.

Answer 2

While the Language Construction Kit offers one particular way of building a conlang, it's very much not the only one—and it seems that one in particular isn't working for you. You say you've ended up with a big pile of half-baked phonologies and nothing else, which suggests that starting with the phonology doesn't fit your style.

I would say instead, decide on what interests you about this particular language. For me, this usually happens when I read about some interesting linguistic feature in my research. Did you know Kiowa has an "inverse number system"? Every noun in Kiowa has a "natural" number (hands usually come in pairs, so two is the normal number of hands; blades of grass tend to come in large quantities, so "a lot" is the normal number). You then put a marking on a noun if it's not this natural quantity—you would use the marking for "one hand", or "three hands", or any number that's not two. You'd also use the marking for "one blade of grass", because that's not the normal amount of grass.

Or, in Swahili, noun case and noun gender are part of the same system. A noun cannot have both a case and a gender. Or, in Lingála, there's a special verb marking for things that are irreversible and cannot be undone. Things like this. I'll see something cool and think "oh, what can I do with that?"

If you don't spend your days reading about languages (shocking, I know), inspiration might come from somewhere else. When David J Peterson created the High Valyrian language for Game of Thrones, he had basically nothing to go from. But the books included the phrase valar morghulis / valar dohaeris, meaning "all men must die / all men must serve". Apparently this -is ending on verbs means "all ___ must ___". Why would you have a verb marking for this? What other things could it be used for? How would that fit into the system as a whole?

My advice for how to create a language is, pick the aspect that you have some sort of inspiration for, and start from there. If phonology is uninspiring to you, ignore it for now. Take the phonology from some random language on Wikipedia and move on. If morphology is uninspiring, just take the noun cases from German…or say "there's no morphology in this language at all". But find something that specifically interests you, and use that as your starting point.

Also remember the rule of linguistics: universals aren't. Your language doesn't have to work like the textbook on ergativity says. No two languages with ergative alignment work the same. It's fine for yours to be an exception too.

Answer 3

Project management skills are useful for conlanging. I have rather bad ADHD, and that is something tat has prevented me from making much progress even though I've been composing a conlang in my head for the better part of a decade and have very little written materials to show for it.

It may be a good idea to start a thread on a conlanging forum to share tidbits over time: external encouragement and debates/questions are very useful productivity tools!

Answer 4

It looks to me that you're trying to do too much at once.

Maybe start off with 'modifying' an existing language. Leave the phonology as is, and work on new words only. Play around with morphology, and then perhaps syntax. Languages evolve over time, so maybe that would make the task less daunting for you: don't start from scratch.

Existing languages are well-tested for their utility. So you change one aspect, and it will probably work less well. Now see what else needs adjusting to make it work again. For example, you start with English and introduce free word order. Now the subject/object distinction is tricky. So you introduce morphological markers (which could be free morphemes) to indicate what the subject is. Then make the next change, and see how your own language evolves from the one you started with.

Or maybe start off with a smaller, less complex language. Look at toki pona, which is a bit of an extreme here. What do you need your language for? Is it a lingua franca to be used in trade situations? Then ignore vocabulary etc for areas that are irrelevant. Keep your language small and simple, and play around with it. Then, once you have a language you are happy with, create another one which is more complex and general.

And most of all: don't let perceived lack of progress frustrate you. This is supposed to be fun!