De Wahl’s rule

suppose you have a bunch of similar languages. now suppose you’d like to invent a language such that, if someone were to speak any of several of these related languages, they’d have no problem understanding text in the language.

how do you simplify common features in an algorithmic way to create a mutually intelligible product?

in this case, we’re talking about the verb/adjective/noun formation from roots in Western European languages like English, Spanish and French. in wikipedia’s words, not my own:

“The root is obtained by the following way:

1. If, after the removal of -r or -er of the infinitive, the root ends in vocal, the final -t is added or the final y is changed into t: crea/r, crea/t-, crea/t/or; atiny/er, atin/t, atin/t/ion
2. If the root ends in consonants d or r, they are changed into s: decid/er, deci/s-, deci/s/ion
3. In all other cases, with six exceptions, the removal of the ending gives the exact root: duct/er, duct-, duct/ion.

These six exceptions are

1. ced/er, cess-
2. sed/er, sess-
3. mov/er, mot-
4. ten/er, tent-
5. vert/er, vers-
6. veni/r, vent-“

in typical Indo-Euopean fashion, these turn out to be the “three rules of de Wahl + the six exceptions”