With your Chinese experiment it would only be accurate if it were double blind, i.e. the speaker themself did not know the meaning of the words. There is a lot

With your Chinese experiment it would only be accurate if it were double blind, i.e. the speaker themself did not know the meaning of the words. There is a lot more communication going on with speech than just the words themselves. Regarding word generation, I grew up in poor urban areas where slang was a huge part of language, and new words were invented and spread almost daily. The sound of a word definitely played a part, but it was mostly about context - when it was used, the tone of the speech, the pitch, what older words it might have been derived from, etc. Like math operations and concepts, there might have only needed to be a few words at the start to build a foundation, a seed from which language growth evolved.