I started out the day by setting right some of the unpleasant code I wrote yesterday. We had to get the app running quickly so we could test it, and so I made some pretty ugly code to get switch the text input source to the source needed for the first study (i.e. have the text reset after every run through, but without prompting for an email and by transferring its data set back to itself when it was recalled. Long story short, I simplified a lot of what I did quickly yesterday, and made everything tied to a single constant variable, named FIRSTSTUDY in the options view controller, but this option is not displayed on the screen, and is thus only a behind the scenes feature. I was thinking about adding it as a dedicated, but defaulted, switch, but in the end it wouldn’t really make a difference, so I opted not to. After that, I couldn’t think of many more improvements to make to the app, besides adding a little more commenting in places. I did notice a slight anomaly in my string extension which counts the number of words before a character index. Essentially, the function will always return the correct amount, unless the final index is a whitespace character between two words, in which case it will return a value which is one word to little. I think this is a fairly easy fix, and will apply it tomorrow.
After noticing that, I took some time to read through some of Keith Rayner’s papers as well as a few others I found around. As Vidushi has already gone through Rayner’s papers in extensive detail I limited myself to looking at his data types. A lot of his work is focused on saccades, which I worry are not relevant to our work, simply because of the way the text scrolls, meaning fixation on text is impossible, and instead the user, from my own self-testing, focuses on the non text area and uses peripheral vision to read. This might mean that our signal will not be correlated to Rayner’s but we will see. The other paper’s I read were about the user friendliness of scrolling text in various contexts, and it was found that labels such as ours significantly lower reading speed if people actually want to fixate, but improve comprehension.