How can I avoid "relexing" punctuation?

It's quite correct that punctuation only really became at thing after the printing-press. So it's be perfectly natural to go without it in many fictional settings. Even so, even though it was anachronistic, I worked on this question for my own last conscript because I enjoyed it.

I read what I could to find all the "functions" performed by every punctuation mark I could find. Then I picked which functions I wanted for my language.

I added one: a sentence-opener mark, in place of capital letters. For the river-inspired conscript in question, this mountain mark's slight angling helps orient the reader to the "river's" direction, too.

I remixed some: splitting the functions of semi-colon; combining em-dash, colon, and quotation under the function "hard parenthetical"; ending sentences/rivers with the "hard parenthetical" "cataratct" is an exclamation.

I left one out: questions are indicated by a clitic already, so there's no question mark.

In my opinion, my punctuation system relexes even less because many punctuation functions are done by how you write the thoughts with respect to each other, like rivers' tributaries. Information encoded by the angles of clauses isn't exactly punctuation, but perhaps it will suggest more possibilities for using text without relexing.

I could have relexed even less, I feel, because there are more nuances and "functions" languages could express by punctuation, though I don't know a language that does. Cantonese has grammar particles doing interesting tasks. English uses prosody for many expressions, and we write this with italics and such. I am sure other languages have features like these. Grammatical nuances like these could be done by punctuation in writing. It depends on what the bare sounds themselves leave ambiguous.

Setting other languages aside, have you found some ambiguous sentences in your conlang? Every language's rules leaves ambiguity somewhere; you just choose where. Perhaps your language has ambiguity in novel places, and so in the less expressive writing media would need a novel mark.

You could also have not-exactly-punctuation symbols. Consider where you might have a culturally important mark like $, or a short-hand mark like &. Culturally important information, short-hand: other common marks are surely potential hiding categories of punctuation job in front of your face :) In my conlang, the first time you mention the name of a god in a conversation or chapter or whatever, you have to use a respectful prefix reserved for this purpose. That "piety function" could have been done by a punctuation instead of morphologically. Some cultural quirk like that in your setting could have a punctuation mark in print :)

Hope one of these ideas helps!

Punctuation is not universal. Chinese and Japanese didn't used to have punctuation, nor apparently did Latin and Ancient Greek. So I'd say that's the first thing to consider: your proto-script needn't have had any punctuation at all. But over time, scribes found that it was a pain in the backside to write long continuous strings and just figure out by context where sentences end and begin. How they choose to delimit those sentences* is up to them.

Maybe writers split up sentences using a vertical bar or a slash, or a middle dot like the interpunct used in Japanese and Chinese, or some other kooky character. Maybe sentences are grouped using something similar to brackets at the start and end. Or maybe ancient authors actually drew a circle or line around each sentence to delimit it from the next (it never had to be practical!) Future authors realised that, if all they wanted to do was tell where one sentence ended and another began, all they had to do is draw a circle around every other sentence.

(So you might have a paragraph like this.) Where every other sentence is uncircled. (And the next sentence is circled again.) Imagine that these brackets are circles.

Punctuation within sentences is, again, up to you. It's perfectly possible that your script's writers never decided that they needed to split up clauses using commas, separate lists using colons, or mark speech using quotation marks etc. Maybe they mark speech separately with an underline, or a dotted line. Or maybe they write using different coloured inks depending on type of sentence being written.

Maybe, instead of little marks, your proto-script could have fully-fledged characters without any spoken value, but which are used to delimit clauses in the same way as commas are. Of course, over time these might have reduced and transformed into a small, easy to write character - in much the same way that Simplified Chinese emerged over time because some scholars somewhere were too lazy to write out 麗 every time, and went with 丽 instead.

The point is: free yourself from all constraints about how your writing should be written. Start with a blank canvas if you like, and add in features as you see fit. Think about your proto-script and how the first scribes might have written things down - then think about how later writers would have abbreviated those forms and made them shorter, simpler and easier. As Artifexian says: Iterate, Iterate, Iterate. Write down random scribbles and see how they pair with your script.

These are just a few random ideas. I hope some of them help.

* I don't know how your language works, but you might not even need to separate sentences. Maybe just clauses, paragraphs, or "logical ideas".

Armenian is one of examples of modern language that has its own punctuation which differs quite noticeably from the Western one, which is nowadays widely adopted (of course, with variations).

Here's are some noticeable differences:

• Armenian question mark (՞) is placed on the stressed vowel of the word in question rather than at the end of the sentence, for instance: "Ինչպե՞ս ես:"
• Armenian exclamation mark follows the same pattern.
• There's a separate punctuation mark used to indicate a pause that is longer than that of a comma, but shorter than that of a colon.