I thought about this for a bit a month or two back. I decided that I was consistent about my spelling matching my pronunciation (though a -s’ one still entails a very subtle adjustment of the final phoneme compared to its -s origin), but I couldn’t find any consistent rule I was following about the pronunciation. Number of syllables was readily falsified; concluding phoneme (/s/ like Chris or /z/ like James) was readily falsified; I didn’t delve far enough into the concluding syllable (e.g. “Jesus’” already ends in /zəz/, ending in /zəzəz/ just makes me think of Sméagol and his fisheses), so there may still be a pattern, but for now I think it may just be irregular.

Pronunciation-dependent spelling is not novel in English: “a” and “an” are the same word, but the choice depends on the succeeding phoneme. The King James Version of the Bible uses the phrase “an horse for an hundred pieces of silver”, showing that the dominant accent at the time used a silent h in both words: “an ’orse for an ’undred pieces of silver”. I contemplated the matter a few years ago and decided that “a horse for a hundred pieces of silver” was a more reasonable reading of the phrase in Australian English (though most will diligently read the clumsy “an horse, &c.”).

This is one of the subtleties across accents. An American might write “an herb” because they’ll pronounce it “an ’erb”, whereas an Australian would write “a herb” because we sound the h.