Natural semantic metalanguage has an extremely small number of concepts, but is parasitic on natural language grammar, and is not meant for communication but for definitions. So it’s not quite the same thing.
Regrettably the Zipf’s Law topic here doesn’t have any content yet.
Not sure the graphs would look essentially different, whatever the register of language is: the tail drops off pretty steeply anyway, which is the point of it being a logarithmic distribution.
Palatal sounds are notoriously unstable phonologically: once /k/ goes to [c] (as it did in late Latin), it can then move on to any of [tɕ, tʃ, ʃ, s].
As a back consonant, <c> could be used to convey anything velar or palatal, or even palatoalveolar, given the possible targets of phonetic change for a fronted /k/.
- So <ch> could end up being conscripted as something velar—like a velar fricative /x/, in German.
- Or it can be used to mean that the velar is velar and not palatalised, like <chi> in Italian.
- Or it can be used to mean something palatal instead of velar—like the palatal stop /c/ in Old French.
- Or it can be used to represent any phoneme that the unstable /c/ ends up sounding like, including the palatoalveolar /ʃ/ in Modern French, or /tʃ/ in English and Spanish.
Basic English tries with 850 words, and xkcd’s Up Goer Five English that Robert Collins mentions seems to be of the lineage of Basic English.