Editing Openai/69186d20-2d14-8013-b9c6-c09000f173f6 (section)

===== This is the real key bit: in Irish, “sh vs s” is just slender vs broad s. =====
* Broad s (with a/o/u) → [s̪ˠ] (dental-ish “s”)
* Slender s (with e/i) → typically [ʃ] (or something very close, e.g. [ɕ])

So the quick rule:

: 

====== 1. Before e or i → usually slender → [ʃ]: ======
* sí “she” → /ʃiː/
* seo “this” → /ʃɔ/
* sin “that” → /ʃɪnʲ/
* seacht “seven” → /ʃaxt̪ˠ/ (here the e is more about slender s; the following a makes ch broad)
# Before a, o, u with no slender vowel nearby → broad → [s̪]:
* salach “dirty” → /ˈs̪ˠal̪ˠəx/
* solas “light” → /ˈs̪ˠɔl̪ˠəsˠ/
* suan “slumber” → /ˈs̪ˠuən̪ˠ/
# Clusters with c, t, p:
* Before e/i, the whole cluster tends to be slenderish → [ʃ]-like: - scéal “story” → /ʃeːl̪ˠ/ - spéir “sky” (in dialects) → often /ʃpeːɾʲ/ - stíobhard “steward” → often /ʃtʲuːəɾˠd̪ˠ/
* Before a/o/u, they stay broad: - scornach “throat” → /ˈs̪ˠkɔɾˠn̪ˠəx/ - sparán “purse” → /s̪ˠpəˈɾˠaːn̪ˠ/
# sh (lenited s) is not [ʃ]; it’s /h/ or zero:
* mo shiopa “my shop” → /mə ˈhɪopə/ or /mɪ ˈopə/ depending on dialect.
* So plain s gives you [s]/[ʃ] via broad/slender; sh is a different animal.

====== You asked earlier about “mixed nearest vowels.” sean is a nice one: ======
* sean “old” → usually [ʃan̪ˠ]
* Spelling: s e a n - s sees e (slender) on its right → slender → [ʃ] - n sees a (broad) → broad → [n̪ˠ]

So you can indeed get slender s → [ʃ] adjacent to broad n in the same word. That’s “mixed” at the word level, but each consonant individually still obeys the broad/slender rule.