Making up the system was not as hard as I first thought. While doing my research I identified that the main frustration that came up in most dialogue systems was that the player had to compromise between a metadiegetic choice and a diegetic choice.

The best example of this problem was this scene, from Dragon Age Inquisition: If the player wants to gain a companion (which is a huge advantage in terms of gameplay), the game forces a viewpoint on his character: He trusts Cole. If the player does not want his character to trust Cole (the weird-looking mage that speaks strange and could be a demon), the only option is to kill him. The intention of the player is linked to an action, which means that if the player wants to act, the game dictates his intentions behind the action.
To me, the solution was quite obvious: Give the player a choice on intention, then a choice of action. Split the two and allow for player combination.

My first thought was that I should make my prototype in engine (I thought about Unreal Engine 4 at the time), but I quickly realized that I would not have the time to make it playable (keep in mind that this was a school project and that I had 4 other projects to realize in the meantime). For this prototype to be playtested, I needed it to be playable with as little bugs as I could.
I also did not want the subjects of the playtest to be distracted by any issues regarding art (not being an artist, I'm positive that anything I made would have been distracting to my participants).
This led me to choose Twine as my weapon of choice in prototyping.

This made the prototype a bit more "wordy" since I did not make UI arts, but I feel confident about my choices. In about 3 days I had a fully functioning prototype and I integrated a story to go with it.
I went with a Skyrim Questline to save me some time writing the story and some of the branches. It also allowed me to simply take the dialogue from the original game for my "standard" prototype which I ran in comparison to my "new" prototype.

While planning for the test, I realized that I needed to add a feature to my prototype to gather data on which nodes were deemed problematic by the player.

The button simply turns a boolean on or off. At the end of the playtest, I take a screenshot of the last panel indicating me whether a conversation node was visited, and whether or not it was flagged by the player. This was the first phase of playtesting and was meant to check if the phrasing was written awkwardly.

After this, I made a playtesting grid (available to download here) and called out for people at my school to come and playtest for a day.

Once this was done, it was time to gather my data and try to make it as easy to understand as possible. I also tried to identify any type of biases that I let slipped by.

Once all of this work was done, all that was left was to sit down and write up some conclusions I had drawn. I saw a lot of issues with the playtest I did and planned out what I would do in another playtest (like using A/B testing instead of making the participants play both.)

All things considered, I was happy with my system, although it came with its flaws, I think it adds to the discussion about how to make the player feel more empowered during game dialogues.