What I’ve always wondered about this is if there’s a better way to present this: pronunciation in English is sort of path-dependent and I wonder if a two-phase approach of recognizing the etymological groups (Germanic-origin, French-origin, Spanish-origin, etc.) and then applying the rules for that group would make the pronunciation appear more regular.

I can immediately think of a counter example: corps vs corpse. Literally from the same word origin, wildly different pronunciation.

That’s not a counter-example: the Latin root “corpus” was borrowed through French twice: once in the 16th century (corps) and once in Middle English (first corse, then corpse). This is a case of path-dependence: the newer borrowing is pronounced more like French while the older one has been regularized.