So I applied my Markov generator to them.
But I don't want them too similar to real-world names. So I decided to "mix" the models together.
For each race, I used the following recipes, with words taken from Swadesh lists (it's hard to find longer lists of more obscure words, but I'm continuously looking). I tried to pick interesting combinations, but I think this will need a lot more experimentation.
- Orcish = Arabic + Somali. These might be too similar linguistically to give a good effect, but we will see. Qolim, Ishan, Hailid, Faris, Afalah
- Elfish = Korean + Egyptian. And these might be too dissimilar. I've already noticed that one meme or another tends to dominate the name. Dohyungminseung, Miunghyun, Daehyeon, Petat, Aprieb
- Dwarfish = Hmong + Egyptian. Khapet, Ziamakiuwel, Sualmayakam, Tirdjedjedi, Menkhteshi
- Humanish = Welsh + Polish. Rznardir, Garybatawart, Nosty, Kwieb, Zhciach
- Halflingish = Bokmal + Russian. Smalt, Ideryy, Flyondel, Mamme, Smannyy
Well, they're pronounceable, at least.
The models can be mixed even further, abstracting it into a second-order chain. So if a city is founded in a kingdom of 50% humans and 50% dwarfs, then I can create a combined chain from all the existing city names of those races.
That is a pretty cool idea. Interested to see how this all comes together... especially city generation!
ReplyDelete