Creating a Conlang with Random Constraints

[Man in Space]

I like to give myself constraints when conlanging--partly out of the challenge, partly out of a sense of it being like real life and having to work with what you have, and partly out of a way to deal with option paralysis. The steps:

1. First, start with gleb and generate a random phonology. The allophonic rules can be ignored, or you can include some or all of them if they're reasonable. Syllable structure can likewise be ignored or amended.
2. Find a list randomizer (I like to use Random.org) and randomize the permissible segments in the syllable--i.e., if your syllable is something like CRVH, randomize the order that the phonemes in the C, R, V, and H slots appear.
3. With these randomized orders, use gen to generate words. What I like to do is have a single, maximal syllable structure and then use zeros to indicate null segments and then a rewrite rule to delete the zeros from the final output, so, say, if my onset is in CR, then I will call a variable R that has a bunch of zeroes at the beginning. gen works on probabilistic selection so a few leading (or trailing) zeros can dramatically affect the output.
4. To generate some grammar (and other features) as a starting point, use Wesley Kuhron Jones' Elided!.

It's perfectly all right to have ideas going in and to edit your results to conform to the same.

[Glass Half Baked]

Why do all this work, when New Guinea already exists?

[Raphael]

Why do all this work, when New Guinea already exists?

By that logic, you might as well ask why anyone should do any conlanging, when New Guinea already exists.

[Glass Half Baked]

More: show

P-TWANG!

[keenir]

Why do all this work, when New Guinea already exists?

Are you paying for our plane fare to and from New Guinea?

[xxx]

no one can afford a ticket to Conlangistan...

[Karavantage]

Here's an idea I had about this very topic:
Use the phonemes in your name!

• this includes all consonants and vowels,
• any diphthongs (not necessarily required but optional),
• if you want added challlenge: stick to your name's syllable structure and phonemic stress,

Maybe a max of two "freebie" phonemes, if your name lacks something you feel a cloņ just can't live without.

[Karavantage]

Use your nickname, full legal name, shorter name (if applicable), whichever you feel gives you the most breathing room.

[Qwynegold]

Here's an idea I had about this very topic:
Use the phonemes in your name!

• this includes all consonants and vowels,
• any diphthongs (not necessarily required but optional),
• if you want added challlenge: stick to your name's syllable structure and phonemic stress,

Maybe a max of two "freebie" phonemes, if your name lacks something you feel a cloņ just can't live without.

Using those one freebie phoneme (because there are too few plosives in my name), my phoneme inventory would be:
/m n/
/b d k/
/f s/
/r/
/i u/
/e:/
/æ A/

I hope scammers can't figure out my full name from this.

[xxx]

There's a little too much randomness in this approach
to be sufficiently serious for my taste...
It's not far off from randomly generating conlangs,
or even paving the way for AI-generated conlangs...

for constraints I prefer to start with limited assumptions
and build the entire language following a certain logic,
so that I feel like I'm discovering it, an observer of its self-organization,
without really making any choices...

The challenge is to be able to say everything,
without adding new elements to these assumptions,
even if it means putting yourself in a particular state of mind
to only reuse the resources of the language,
a bit like what the Sapir-Whorf hypothesis describes...

[Man in Space]

For my name, with a little creative interpretation:

/n/
/b t d c ɟ k g ʔ/
/v s/
/ɾ/
/w l l̴ j/

/ɛ e i ə ɨ a o (u)/
/n̩/

• My freebies are the voiced coronal plosives.
• /e/ and /j/ are separate here because it’s a little more parsimonious.
• /o/ and /w/ are separate for partially similar reasons, though /w/ does also come from my first surname.
• /u/ is marginal as some pronounce my second surname with that vowel instead of reducing it (I do the latter).
• The vowels have been finessed a bit so that there’s a bit more balance to the system (this is also why /u/ is marginal).
• /ʔ ɾ ɨ n̩/ are thanks to phonological processes in my idiolect, though I’ve elected not to explicitly call the /t/ here aspirated, again for symmetry.