Making the optimum language?

Sounds fun.  I would start by identifying the number of plausible one syllable roots we want.  From their I would pick a stripped down version of Spanish. Remover imperfect tpast and make haber and estar  standard verb helpers for the participles. White stripping Spanish, stick with the AR pattern for tense.

Now, of the 1500 basic roots in you collection, the 100 most vocally comfortable are allocated to the oral tradition, music, and every day life. The oral tradition is allow to make the upper left portion of the conjugation chart irregular with some wide latitude. Make voice comfortable.

Then with the next best 800 root construct the middle class work life. Then with another 200 as prependers that define mood, method, order, frequency, make a bunch of compound roots used for literature only. The language of art, science never designed for the vocal; always computed on paper.


For the computer bots tings change, quite a bit, and I am uncertain or unwilling to say.  My experience so far is the natural human language for the computer is an eight byte, or sixteen byte ascii sequence. And order is about the only grammar, order being mostly approximate. The web, or the singularity will live well with very large look up tables, at the expense of complex grammar. With a shared key word, human and bot can easily align on which dictionary is appropriate.