The first question to address: how does people's use of specific chat affordances effect others' perception of them?
In other words, knowing nothing about someone else except seeing how they converse through the interface, how do people form judgements about others?
Of course the content will play a role. So they key is to separate that as a confounding variable in the experimentation.
The overall goal here is to find a way to create a conversational space that allows people to be expressive and clear without significantly warping others' perception of them as a result of the affordances they choose to use. It's our job to figure out what affordances to provide in order to create this space. So the question to start with is if there are affordances that do have this effect, then work backwards to find a more essential version without the effect.