The linguistics of punctuation

If you have linguist friends who says things like "writing is not under the purview of linguistics" or "linguists don't care about performance, but about the capacity of language", 1) why are you friends with Chomskian linguists and 2) this book is the most pointedly interesting counterproof. An analysis of the grammar of punctuation using the methods of linguistics somehow brings up plenty of hidden structure of the kind linguists are used to discover in natural language. How come?

Punctuation is obviously not a part of the human language-acquisition capacity—babies won't acquire punctuation no matter how many semicolons you show them—but contrary to naïve social opinions about "style guides" and schoolroom torture, the real grammar of texts isn't acquired via conscious study, it's acquired by exposure via reading; its most interesting and complex rules are emergent and used without explicit knowledge, just like natural language. This is because writing was developed by building on analogies of how languages work, until it evolved language-like structures of its own. In other words, text-grammar is, like conlangs, an "application of the principles of language"; and, consequently, it can be acquired like an SLA, and can be analysed with the linguist's toolset. Writing stands to the capacity of language as dancing stands to the capacity of moving.