Sound changes

Posted by Gustavo S. Pereira on 16:35 2/7/02

In reply to: (none)

Hi there!

Well, I am attempting to devise a set of sound changes to make a new language from another one, but it's quite difficult to start that list!! I know the types of sound changes I can include, but what changes to include first?? (like the transition to a medieval language, than to a modern one). Could you give us some tips? You've made such a wonderful work on the Cadhinor > Verdurian, Cadhinor > Ismaîn and Cadhinor > Barakhinei sets of sound changes (specially the first one) that I would like to hear some words from you about starting and making a sound changes list.

Gustavo Pereira

Mark responds:

Hmm... it's hard to describe, being more of an art than a science.

You have to play around a lot. Make a smallish wordlist (say, 100 words), run the program, and edit the sound changes. Keep doing this for a long time.

The order comes into play only when one change feeds another. E.g. if you want umlaut and final-syllable truncation, make sure you do it in that order, or the truncation rule will eliminate the input to the umlaut rule. Trial and error is enough to find such problems.

A language does not know, of course, whether it's modern or medieval, and there is no sound change that's likelier at one time than another. In effect, then, you can divide up your medieval and modern rules arbitrarily. The only guideline I'd suggest is that the rule lists be of comparable size (so the medieval language isn't too close to either end) and be fairly diverse (so you don't have, say, just vowel changes leading to the medieval language, and just consonant changes leading to the modern language).

Maybe more tricky is how many rules to include. You can use my grammars as a guideline; or just examine your results. If the wordlist looks too much like the input, add more rules.

The Verdurian list is so long because I was working with existing Cadhinor and Verdurian lexicons. So, I often had to tailor rules to produce the words I knew I wanted. In that way it was more like working with real languages: you examine a bunch of words and try to figure out what sound change could possibly account for the end results!

To make a reply, or see replies, see the index page.