05-04-2009, 06:55 PM
These are the results:
http://colin.shoddybattle.com:81/reading/report.php
It seems that capital letters provide an appreciable increase in reading speed for test 0, but not for the other tests. As you can see, the populations appear to be normally distributed (A*^2 is the adjusted Anderson-Darling statistic), so confidence intervals and hypothesis tests based on the Student t-distribution are valid.
If you have any other particular tests you'd like to see run along these lines, let me know.
I plan on looking into what could possibly have made test 0 (of 0-3) significantly different in terms of reading speed with regard to capitalization, while the other tests were not. This seems interesting enough to look into further.
I also plan on looking into the idea of "reading fatigue" Jester mentioned. The theory is (just to repeat it), people may be able to keep up an equal reading speed in most cases regardless of capitalization, but they might not be able to do so for long periods of time. In other words, there is more of a mental "drain" caused by this. However, it's uncertain whether this is caused by an inherent difference in capitalized vs. non-capitalized sentences, or merely because people are used to reading capitalized letters. I liken it to the difference between qwerty and dvorak keyboards. It's obviously going to slow down a typist when they first begin using a new keyboard, but typists end up at least as fast on dvorak as they are on qwerty.
However, I know of no way to reliably test this. Any long-term study involving a group of people reading lots of lower-case only vs. proper-case only text for a while could always be answered with "Well, they've spent their whole life with primarily proper-case for long texts, so they're still more used to it.", so the hypothesis that capitalization isn't inherently significant in the difference of reading speed + comprehension isn't falsifiable, which is concerning to me.
I would appreciate any possible additions to the next round of capitalization experiments. Other ideas I have for textual experiments are things like combinations of text color (white on black, black on white, various grays, and even more exotic colors like dark blue, brown, purple, red, etc.).
http://colin.shoddybattle.com:81/reading/report.php
It seems that capital letters provide an appreciable increase in reading speed for test 0, but not for the other tests. As you can see, the populations appear to be normally distributed (A*^2 is the adjusted Anderson-Darling statistic), so confidence intervals and hypothesis tests based on the Student t-distribution are valid.
If you have any other particular tests you'd like to see run along these lines, let me know.
I plan on looking into what could possibly have made test 0 (of 0-3) significantly different in terms of reading speed with regard to capitalization, while the other tests were not. This seems interesting enough to look into further.
I also plan on looking into the idea of "reading fatigue" Jester mentioned. The theory is (just to repeat it), people may be able to keep up an equal reading speed in most cases regardless of capitalization, but they might not be able to do so for long periods of time. In other words, there is more of a mental "drain" caused by this. However, it's uncertain whether this is caused by an inherent difference in capitalized vs. non-capitalized sentences, or merely because people are used to reading capitalized letters. I liken it to the difference between qwerty and dvorak keyboards. It's obviously going to slow down a typist when they first begin using a new keyboard, but typists end up at least as fast on dvorak as they are on qwerty.
However, I know of no way to reliably test this. Any long-term study involving a group of people reading lots of lower-case only vs. proper-case only text for a while could always be answered with "Well, they've spent their whole life with primarily proper-case for long texts, so they're still more used to it.", so the hypothesis that capitalization isn't inherently significant in the difference of reading speed + comprehension isn't falsifiable, which is concerning to me.
I would appreciate any possible additions to the next round of capitalization experiments. Other ideas I have for textual experiments are things like combinations of text color (white on black, black on white, various grays, and even more exotic colors like dark blue, brown, purple, red, etc.).