So, discourse particles are a thing that have decidedly been given a stepmotherly treatment in English. A similar disdainful view does exist in Swedish, even though it'd seem Swedish does have more dedicated discourse markers than English does. By 'dedicated discourse marker' I mean a word that cannot also be used in other ways.
As an example of the disdain I am talking of, when the Swedish pop-sci linguistics magazine Språket ('Language') published an article about discourse particles, the discussion in a variety of online language groups was decidedly hostile, people saying this was the final drop re: that magazine, people thinking it dumb that someone defend such frivolous words, etc.
I believe this disdain for them, this view of them as something to be avoided and even scorned, as a sign of low intelligence or lack of education is something that may affect the willingness conlangers have to use them in interesting ways.
I am not saying we share the prejudice, I am saying the prejudice just subtly steers us away from thinking about them, in part because there's less material about them.
So, how about changing their word classes to something more respectable?
Pretty much what it says on the tin. However, one can imagine some further twists: maybe they also entirely replace some verbs, such as copulas. They might "cut across" verbs depending on a variety of factors.
Imagine, for instance, a verb "hæm" that replaces "have" in the case of first person subjects with an NP after, but 'are, is' in the case of other persons:
I hæm a solution -> oh, I have a solution
you hæm a teacher -> oh, you are a teacher
I hæm eaten already -> oh, I have already eaten
you hæm eaten already -> oh, you have already eaten
These could also be secondary-rank auxiliaries or primary-rank ones - depending on how they act in combination with other auxiliaries.
2. Anti-auxiliaries and second-rate auxiliaries
An anti-auxiliary would force any other verb to behave like its auxiliary, and then trigger word order such as that of aux + subordinated verb.
A discourse particle acting as a subject will of course push every other argument down, possibly such that subjects become indirect objects, objects maybe stay as such (or can be demoted to obliques), and indirect objects either are demoted to direct objects or obliques.
4. Object, indirect objects
Similar to the previous one. In the case of (di)transitive verbs, you get some kind of demotion of the actual (indirect) object.
6. Some kind of sliding NP
These would pick the first free slot - subject, object, indirect object, (or maybe io and o switch places in the hierarchy), or some kind of olbique. These could have some kind of "stopping point" which they won't be demoted past, and from that point on they cause the effects described above.
An adjectival discourse particle would mark congruence with nouns, or at the very least have the syntactical properties of an adjective. Maybe they could even be used as complements of verbs, i.e. an adjective that means 'yeah, sure' and behaves adjectivally:
The yeah-sure.neut car.neut was on fire
yeah, sure, the car was on fire
the car.neut that burned down was yeah-sure.net
yeah, sure, it was the car that burned down!
I am just mentioning clitics to exhaust the low-hanging fruit.