Alex Fink, On 16/09/2014 22:49:
> So, how about a relay where what's passed along is a conscript?  You
> receive a script, which you have to adapt to a conlang of yours (or a
> natlang? or a sketch made up on the spur of the moment?) the way
> bilinguals only literate in your predecessor's language+script might
> realistically do.  You are then free to make evolutionary changes --
> graphic changes, disambiguations, sound changes, reshaping to better
> fit your (morpho)phonology in the large or the small, etc. -- after
> which you pass the new script along to the next player.  This should
> allow for lots of interesting quirks to build up.

An adaptation of a script depends on the adapter's understanding of the source. On the signans side, there is potentially the distinction between the graphemes and the allographs, and on the signatum side there is the question of how to characterize the signatum -- phoneme? allophone? something else?

But in the relay, the adapter's understanding of the source is going to be determined by the sophistication of the info they receive from the previous. This means a main determinant of the course the relay takes is the sophistication of the description sent on. I wonder if this dimension of variation could be reduced by standardizing what is sent on -- but I can't yet think of what would be a viable minimum amount of info. How about, say, we start with an IPA transcription of a text in random language X in conscript Y, and then the next turn takes this as a basis for adapting the script to their conlang, and adapts the text to the phonology of their conlang, writes out the text in the new version of the script, and sends it and an IPA transcription of the new version of the text onwards? I suppose a snag with that is that a text that is not onerously long will likely not include all relevant ingredients of the script.

Unlike ordinary relays, this looks like each stage would take a while to mull over. It's not something can be done mechanically.