This looks like a good candidate for timed task tests. You can either do these observationally (with a group of test participants) or naturally (using playback recordings in a live environment, if you have the ability to do a series of AB tests and record them with something like Hotjar).
You're probably looking at answering questions like "Is it better to display paragraph formatting before letter formatting?" rather than "Is it better to show bold before italic?" because users expect bold before italic in most of the software that they use. So, you could run an experiment by showing paragraph formatting before letter formatting to half your test group, and letter before paragraph to the other half, and finding out which is accessed quicker.
You would record the tasks and play back the recordings for the timing data (telling users they're being timed might make them nervous). You can get some self-reported ease-of-use metrics and attitudinal feedback as well by talking to users after each task ("On a scale of 1 to 10, how easy was this?" "How would you make this better, if you could?"). You could also dig into struggles if you see them. This can be done moderated (in person over Zoom) or unmoderated (using a tool like TryMyUI or Usertesting.com).
Pros: You could get some qualitative feedback along with quantitative timings. It might be easy to piggyback this on to another user test in the same system, since formatting text is something most people do when using computers.
Cons: You might need to incentivize the test, which could be expensive. And that might limit how many data points you can collect.
User Session Playback Recording Data
If you're looking at playback recordings, the playback software could help you find people who clicked these particular buttons, and you could look at the timings between their paragraph/word selection and clicking the buttons.
Pros: You can get many data points without having to pay anyone for their time
Cons: You can't get qualitative feedback or ask users their thoughts