Experiment testing for mirror reading
June 3, 2008 9:34 AM   Subscribe

Experiment testing for mirror reading
A high school student i work with is trying to gather data about the prevalence and distribution of mirror reading (the ability to read text that is flipped left-to-right). He's put up a site that performs a brief test (a couple of minutes max), and asks for some simple demographic data. If you have a chance, try it out. I've asked him to follow the feedback here (he doesn't have a mefi account yet) if you have comments.
posted by dkg (13 comments total)

My only comment would be that the text seems too long -- just a sentence or two is enough to measure differences on typing/reading mirrored text I would think.
posted by mathowie at 9:35 AM on June 3, 2008


I can sort-of mirror read. I can do it, but I'm a little bit slower. Some things beyond the mirroring that may have affected my typing speed:

First, there were a few minor grammatical and capitalization mistakes that made it a bit harder to follow. Even on the forward version, I had to go back a few times.

It was a bit odd to be reading in one direction while my typing went in the other direction. While mirroring the letters might be tricky, changing the direction of the type (via the "dir" tag or similar) might help.

I was surprised at how much difficulty I had with letters that were correct but different when mirrored. I would read "done" as "bone", for example.

I was picking up spead as I went -- possibly because I was out of practice. I suspect that I'd "pass" if I tried the test a few more times, but I don't want to mess up the results.
posted by brokengoose at 10:05 AM on June 3, 2008


Agreed. I also found that the content of the backwards test was much more challenging in terms of vocabulary... I presume you're trying to prove that someone really can mirror read.
posted by lowlife at 10:22 AM on June 3, 2008


I had to mouse-click into the text box to start typing, which slowed down my input quite a bit and probably made my contribution basically meaningless (I lost several seconds to trying to figure out why what I was typing wasn't appearing - could be a Mac or Firefox issue). The use of Harry Potter text provided way too much context help (to easy to guess what is being said, artificially speeding up input). To really measure people's ability to "mirror read" both the control and the test text should be contextually meaningless so that there is not context help, which is irrelevant to the experiment. Using an extremely popular text like Harry Potter is a particularly poor choice. I also think removing all punctuation and formatting would be preferable. Just a block of basically random words would be a better test basis in my opinion.
posted by nanojath at 12:14 PM on June 3, 2008


"Thank you for participating!

I would ask for your continued participation by asking you to email a link to the experimental page (http://leo.urbanacademy.org)."

I don't know what that means.

I'm not that familiar with Harry Potter, but enought that mirror-recognizing "Hermione" wasn't nearly as difficult as it should have been. Random words probably wouldn't be the best idea (that's not really "reading" either, I don't think, as much as "recognizing") but obscure texts would probaby have been a strong choice.
posted by Shepherd at 4:47 PM on June 3, 2008


Will he be sharing the data he collects?
posted by carmelita at 10:28 PM on June 3, 2008


Always happy to help in research projects. I wonder though, is it appropriate to use such long swaths of text? The above comments have already mentioned this, but the second selection did have more difficult vocabulary than the first, and it would seem that you might benefit from more than one trial of shorter selections.

How are you planning to determine mirror-reading ability? Is it a fluency that's on-par with ordinary reading? I noticed you give unlimited time for the first selection of text, is this a way to determine one's normal typing speed, and then use that to gauge the fluency of one's mirror-reading ability?

I notice your instructions don't say anything about mistakes, how are you planning to take into account typos - especially for someone who has typos on even the ordinary text? Should you disregard them, and, if you do, should you not instruct the subject that errors will be discounted in some manner (or not?).

Finally, in your demographics you didn't ask anything about age - shouldn't this be a key piece of information? (i.e. perhaps it's a generational thing created by different types of education). You also don't ask anything about race/ethnicity or religion, but shouldn't these be crucial to consider?

I know nothing about your design study, so just disregard anything above that's irrelevant.

Good luck!
posted by tybeet at 10:33 AM on June 4, 2008


My comments: cool idea, but I agree that the use of Harry Potter text might be doing a disservice to your data. As someone who has read the books many times, I found it much to easy to pinpoint the exact scenes I was typing from and the context clues I got from my memories of the books gave me a definite unfair advantage.
posted by Zephyrial at 10:52 AM on June 4, 2008


Also: shouldn't you include in your demographics a question that asks reading habits (i.e. how much do you read in a given month/year?) as I would think this might have something to do with it.
posted by tybeet at 10:59 AM on June 4, 2008


...religion? Seriously?

I would think a multi-check box indicating knowledge of other languages, especially r-l languages, as well as a primary language, might be useful. I took some Arabic in school, but I'm also left handed and used to mirror-write as a kid...
posted by cobaltnine at 5:07 PM on June 4, 2008


Correct me if I'm wrong, but the idea here seems to be finding correlates with mirror-reading ability. Religion is certainly a fundamental demographic if the purpose is as such, whether or not you can logically justify it a priori.
posted by tybeet at 8:12 PM on June 4, 2008


nth'ing those above who suggest the choice of text may be a problem here; I zipped through the test quickly, but I actually don't think I can mirror-read very well at all: most of it was picking up words through context and familiarity with the text.

A short typing warmup before the actual test might've also helped; I was making quite a few typos and backtracking the first time through, which probably artificially slowed my non-mirrored score.
posted by ook at 4:13 PM on June 9, 2008


Yeah, I wasn't sure if I was supposed to replicate his typos and spacing inconsistencies or not. I did, which seems like a waste of time as it's not at all about mirror reading but on idiosyncratic mistake replication.

A better test of mirror reading might be to read both passages aloud and compare the speeds.
posted by xo at 4:30 PM on June 9, 2008


« Older Tag the Web...   |   Fractal Fripperies... Newer »


You are not currently logged in. Log in or create a new account to post comments.