On Wed, Feb 22, 2017 at 11:25:43PM -0500, Aidan Aannestad wrote:
> All that to say, this seems to be a case where conlangs could really have spectacular scientific value. If we were to create realistic language families and areas, with as much of reality's complications as we could manage, and then trade them off and reconstruct each other's work sight-unseen, that would be quite a test of the comparative method! I'd be very, very interested to work on a study like this. The one fundamental difficulty with this kind of study would be how to judge the realism of the fictional diachronic processes, as the vast majority of what we believe to be 'realistic' is based in comparative method reconstructions already - rendering a part of the argument circular. There might be a useful way around that, I don't know; it might also be useful enough just to try a study like this, whether or not it's totally perfect.
If this ever gets run, I'd be interested in participating, either by
contributing a language family or trying to reconstruct the ancestor of
someone else's (or both).
- kechpaja -