Two years ago, a widely circulated statement on the Internet claimed that resarceh at Cmabrigde Uinervtisy fuond that sentecnes in whcih lettres weer transpsoed (or jubmled up), as in the setnence you are now raeding, were easy to read and that letter position in words was not important to the ability to read successfully. In actuality, the statement was a hoax in that no such research had been conducted at the University of Cambridge (see http://www.mrc-cbu.cam.ac.uk/personal/matt.davis/ Cmabrigde/). We report here results from a study showing that although some variations of sentences with transposed letters are relatively easy to read, other variations are not, and that there is generally always a cost associated with reading words with transposed letters.

We asked 30 college students at the University of Durham, United Kingdom, to read 80 sentences in which letters were transposed. In each sentence, transpositions were consistently located at the beginnings, middles, or ends of words (see Table 1). About 40% of the words in the sentences (all content words longer than four letters) had letter transpositions. In addition, the students read sentences without any transpositions. Eye movements were recorded via a Fourward Technology Dual Purkinje eyetracker; the spatial resolution of this eyetracker is less than 10 min of arc. Comprehension questions were asked after 30% of the sentences. Readers were able to answer the questions with high accuracy, but 50% of them indicated that there were a few words that they did not understand.

Whereas the base reading rate for normal sentences was 255 words per minute (wpm), all of the variations involving letter transpositions resulted in some cost to reading. When internal letters were transposed, the reading rate was 227 wpm (an 11% decrement in reading speed). However, when the transpositions involved the ending letters of words, reading rate was 189 wpm (a 26% decrement), and when the transpositions were at the beginnings of the words, reading rate was 163 wpm (a 36% decrement).1 Readers made more and longer eye fixations (see Table 1) with the more difficult transpositions.

The Internet statement was correct in that some letter transpositions do yield words that are relatively easy to read. However, our results clearly demonstrate that transpositions always carry a cost. Furthermore, our research also shows that transpositions vary in their costliness depending on their location in the word: Transpositions of internal letters are much less costly than transpositions of ending letters, which in turn are less costly than transpositions of beginning letters. These results demonstrate the importance of beginning letters for word recognition (see Rayner & Pollatsek, 1989, for a summary). In other work (Christianson, Johnson, & Rayner, 2005), we have also demonstrated that letter transpositions that cross morpheme boundaries (even with internal letters) are associated with an additional cost. Thus, susnhine is more difficult to read than sunhsine.

Finally, a previous study showed that when letters are substitutedratherthantransposed,readers takemuchlongertoread sentences (Rayner & Kaiser, 1975). In the case of substitutions involving visually similar letters, substitutions for internal letters (e.g., problem printed as pncblem) doubled reading time, as did substitutions for ending letters (e.g., problnc); substitutions for beginningletters (e.g., qroblem) were associated with reading times 2.5 times longer than normal. In the case of substitutions involving visually dissimilar letters, substitutions for internal letters (e.g., prkylem) or final letters (e.g., problky) tripled reading time; substitutions for beginning letters (e.g., fyoblem) quadrupled reading time. In all cases (except when visually similarlettersweresubstitutedforinternalletters),substitutions also reduced comprehension.

1 Thedecrementsreported areundoubtedlyan underestimation of thetruecost AddresscorrespondencetoKeithRayner,DepartmentofPsychology, of reading text with transposed letters because we transposed letters only in University of Massachusetts, Amherst, MA 01003, e-mail: rayner@ content words. One might expect the cost to be even greater if transpositions psych.umass.edu. occur in all the words of a sentence.

192 Copyrigh pss.sagepub.co t r 2006 m A ssociation for Psychological Science Volume 17—Number 3

K. Rayner et al.

TABLE1

Example Sentences and Results for Measures of Fixation

Condition	Example	Number of fixations	Percentage regressions	Average fixation duration (ms)
N	The boy could not solve the problem so he asked for help.	10.4a	15.0a	236a
INTa	The boy cuold not slove the probelm so he aksed for help.	11.4b	17.6b	244b
END	The boy coudl not solev the problme so he askde for help.	12.6c	17.5b	246b
BEG	The boy oculd not oslve the rpoblem so he saked for help.	13.0d	21.5c	259c

Note. N 5 normal text; INT 5 transpositions of internal letters; END 5 transpositions of letters at word endings; BEG 5 transpositions of letters at word beginnings. Within each column of results, entries that do not share a subscript differ at p < .01.

There were two types of internal letter transpositions: Either the beginning internal letters (as in slove) or the ending letters (as in probelm) could be transposed. Because the results showed no difference between beginning-internal and ending-internal transpositions, we have collapsed the results across these two types. Note that the example shown was constructed to illustrate both of

these types and was not used in the experiment itself.

Thefact that text withlettertranspositions is so mucheasier to read than text with letter substitutions demonstrates that the specific letters of a word are critical for identifying what the word is (Grainger & Whitney, 2004) and that readers cannot rely exclusively on context for word recognition. In comparison to letter substitution, letter transposition makes it much easier for readers to recover what the actual form of the word should be. But the main point of our findings is that although it may seem that it is easy to read text with transposed letters, and although some transpositions do not entail as much added difficulty as others, there is always a cost involved in reading such text in comparison to normal text.

Acknowledgments—This research was supported by Grant 12/S19168 from the Biotechnology and Biological Sciences Research Council (United Kingdom) and by Grant HD26765 from the National Institutes of Health (United States).

Volume 17—Number 3 pss.sagepub.com

REFERENCES

Christianson, K., Johnson, R.L., & Rayner, K. (2005). Letter transpositions within and across morphemes. Journal of Experimental Psychology: Learning, Memory, and Cognition, 31, 1327–1339.

Grainger, J., & Whitney, C. (2004). Does the human mind raed wrods as a wlohe? Trends in Cognitive Sciences, 8, 58–59.

Rayner, K., & Kaiser, J.S. (1975). Reading mutilated text. Journal of Educational Psychology, 67, 301–306.

Rayner, K., & Pollatsek, A. (1989). The psychology of reading. Englewood Cliffs, NJ: Prentice Hall.

(RECEIVED 9/30/05; ACCEPTED 10/3/05;

FINAL MATERIALS RECEIVED 10/5/05)

193