zompist bboard

So, here's my story.

1. I decide on a beautiful phonetic system for my conlang.
2. I create an amazingly cool grammar (in my opinion, obviously).
3. Profit?
No... not quite.

So, then I create a script that would generate some gibberish sentences in that language abiding by the phonetic rules and by grammar rules. Basically creating pretty much "real" sentences, yet before there are any actual words.
Edit: by script I mean a program that would generate that sentence for me.

And here is the problem. It sounds like shit
Like someone murdering a person with a chainsaw to be specific....

There are no consonant clusters, no diphthongs, etc. Very simple CV[n] syllables. Words are typically 1-2 syllables long + grammar thingies. And still...

This leads me to following solutions:
1. Simplify my phonetic system and make something like Japanese. This way it will sound very good 99% of the time. But it will reduce the possible number of different short words (bad).
2. Ignore "one sound = one meaning" rule which I always try to abide. And this one doesn't excite me as well

So, neither of these solutions are really that good.

Regardless, the only real solution that I see here is to just significantly simplify phonetics and make my words longer to address issue of too little possible combinations.

Or maybe I'm doing all this entirely wrong?

Karutoshika wrote:Or maybe I'm doing all this entirely wrong?

Bingo.

http://badconlangingideas.tumblr.com/ Read through a few dozen of these...if any of them sound familiar, stop.

Also, read these:

http://www.zompist.com/kit.html
http://www.zompist.com/resources/index.html
https://en.wikibooks.org/wiki/Conlang/Beginner/Sounds
http://www.councilofelrond.com/subject/ ... -language/

There are many more...start with these.

Karutoshika wrote: Like someone murdering a person with a chainsaw to be specific....

I see nothing wrong with this. That sounds awesome Can we have a sample?

How can you possibly have a program that generates grammatically correct sentences if your "conlang" has no words?

And I strongly recommend you go and buy Zompist's The Language Construction Kit, Advanced Language Construction and The Conlanger's Lexipedia books, along with grammars of languages you're interested in, be it natlangs or conlangs. Also check out YouTube for conlangers' videos.

Karutoshika wrote:2. Ignore "one sound = one meaning" rule which I always try to abide. And this one doesn't excite me as well

The unworkability of such a principle is presumably one of the big reasons why real-life languages don't have it.

(Of course there's no reason why a made-up language should necessarily be "naturalistic", though most people in this community do tend to aim toward naturalism to a high degree.)

Of course you could also increase the number of one/two syllable words by increasing the permitted complexity of syllables. (Even a small change, such as allowing /m/ as another possible final consonant, or allowing C(y,w) clusters or long vowels, will dramatically increase the number of syllables available.) Or you could adopt a very large inventory of individual sounds - not that this is necessarily a good idea.

@masako, @dyolf, I appreciate your suggestions and links, but I'm not new to this. I read these things many times over the years. And you can see my registration date on this forum as well (2010). Linguistics is my hobby and I spent quite some time researching the subject, even though I don't have formal education in this field. But my main interest is the grammar and how different natural languages work as well as what would happen if you take certain strange elements from different languages and put them together. Or even invent some crap that breaks linguistic universals
Anyway, my point is - while I'm perfectly fine designing grammar, designing actually good phonology is something I'm very bad at.
But I should probably start by learning what sounds shouldn't go goether or even be in the neighboring syllables. If you have links to such information that would be something I would really appreciate.

gufferdk wrote:

Karutoshika wrote: Like someone murdering a person with a chainsaw to be specific....

I see nothing wrong with this. That sounds awesome Can we have a sample?

I would rather not
And to be honest it's nothing really interesting. Just CV[n] syllables, but with tons of different consonants (23 to be specific), including things like sh, zh, ts, dz, ch, j which might be the case it sounds so bad

Karutoshika wrote:@masako, @dyolf, I appreciate your suggestions and links, but I'm not new to this. I read these things many times over the years. And you can see my registration date on this forum as well (2010). Linguistics is my hobby and I spent quite some time researching the subject, even though I don't have formal education in this field. But my main interest is the grammar and how different natural languages work as well as what would happen if you take certain strange elements from different languages and put them together. Or even invent some crap that breaks linguistic universals
Anyway, my point is - while I'm perfectly fine designing grammar, designing actually good phonology is something I'm very bad at.
But I should probably start by learning what sounds shouldn't go goether or even be in the neighboring syllables. If you have links to such information that would be something I would really appreciate.

If you've read Zompist's books many times, then maybe you should revisit the chapters on phonologies and phonetics. But one piece of advice that I can sum up from the books is that "less is more". You don't need to have lots of sounds. If you are bilingual (or have a passion for a particular language) then try writing out the phonemes of each language you're familiar with, then simply remove some. You could throw away entire groups or ones you're not that keen on. For example the conlang I'm working on (very slowly) has it's phonology based on English and Welsh. So it contains many of the things you'd expect to find in English like /p/ /b/ /t/ /d/ /k/ /g/ /f/ /v/ /s/ /h/ /θ/ /ð/ /m/ /n/ /l/ /w/ as well as some Welsh phonemes like /χ/ and /ɬ/ for the consonants (that's not all of them BTW). I have, however thrown away /ʒ/ /z/ /ʃ/ for example. I had originally thrown away /m/ and /j/ but have now included them.

You could also check out these interactive IPA charts so you can hear what some sounds you may be unfamiliar with actually sound like before choosing them.