Purpose. Reading speeds drop to 33% of maximum when letters within the text are randomly oriented |Ortiz etui., ARVO 1996). One explanation is that letter recognition in reading is orientation dependent, but it is also possible that the speed reduction is due lo confusions from a suhset of letters that resemble other letters when rotated. We tested the latter explanation. Method. First, a letter-recognition experiment was conducted to identify which letters were most confusable when orientation was randomized. Confusion matrices were constructed from recognition responses to randomly oriented, individual letters. In a second experiment, two lists of words (equated tor word frequency) were generated: lisl A had words containing only the 10 most confusable letters, and list B had words containing only the 10 least-confusable letters. Reading speeds for random sentences of \vords from lists A and B were compared in two conditions: with letters in their normal orientation, and with randomly oriented letters. Results. When randomly oriented, the 10 most confusable letters were [q d u k p be w 1 a]. Reading speeds for words containing these letters dropped to 19% of maximum. The 10 least confusable letters were [e g s o m I i z x h]. Reading speeds for words containing these letters dropped to 33<7<- of maximum, similar to our previous findings for text composed of all 26 letters. Conclusions. Texts that contain a high density of rotationally confusable letters are harder to read when the letters are rotated. But even when these letters arc entirely excluded, reading speeds show a strong effect of random lelter rotation. These results support the concept of an orientation-specific mechanism for letter recognition in reading.
|Original language||English (US)|
|Journal||Investigative Ophthalmology and Visual Science|
|State||Published - Dec 1 1997|