The “Cambridge Effect”

For years a fairly silly hoax has been flying around the internet, passed from one gullible person to another, claiming that the order of the medial letters in words is not important for word identification. As long as the first and final letters in the words are unaltered, the message states, then the medial letters can be scrambled without undermining the “phenomenal power of the human mind” to accurately read the text.

Basically, this is fairly crude parlor trick — one that quite erroneously claims to be based on research conducted at Cambridge University.

It goes like this:

I cdnuolt blveiee taht I cluod aulaclty uesdnatnrd waht I was rdanieg. The phaonmneal pweor of the hmuan mnid. Aoccdrnig to a rscheearch at Cmabrigde Uinervtisy, it deosn’t mttaer in waht oredr the ltteers in a wrod are, the olny iprmoatnt tihng is taht the frist and lsat ltteer be in the rghit pclae. The rset can be a taotl mses and you can sitll raed it wouthit a porbelm. Tihs is bcuseae the huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohe. Amzanig huh? Yaeh and I awlyas thought slpeling was ipmorantt.

In short, here is my response to anybody who believes that serial position of medial letters is not important:

Tihs “Cdirmbgae Eefcft” is srot of ture, but it deos not tlel the wlohe sorty. It is rieealtvly esay to raed sroht wrods, but not so esay wehn ridaneg legonr wdros. Aals, msot cetonnt wdros in Esinglh are sveen leertts lnog or leognr — the mroe leretts terhe are in a wrod, the mroe dulcifift and caipcetmold it bmeecos to cletrorcy infietdy them wehn the ltrtees are ragnearerd. Warhees, more diiivutnme and cmoomn wrdos lkie “blal” and “baer” raimen mltsoy ungnchead and esay to rizocenge, lgneor, less cmoomn wdros, like “pltlooiun” and “soulamitunes” caghne saillbattunsy to the pniot werhe rnciooitegn is srclceay pbsslioe. Fmrrhtrouee, this atiibly smtes form a garet deal of enpicerexe rindaeg cretolcry slelepd wdros — olny plopee who can adrealy raed pelictroinfy can do this tsak. Tihs tirck does not reeavl mcuh aoubt the pscroes of lnnreaig to raed — it only ietaindcs that hhligy slielkd rrdeaes are so pectacred wtih txet tehy can omoercve moinr informieepcts (epshaims on the word “moinr”) in the pnirt to aesccs mnnaeig.

Oh, and by the way, terhe nveer was any phuilbesd rcsaeerh of tihs srot ccountded at Cridbagme or any oehtr usrivetniy. Taht prat is ceepomlty urutne.

This “Cambridge Effect” is sort of true, but it does not tell the whole story. It is relatively easy to read short words, but not so easy when reading longer words. Alas, most content words in English are seven letters long or longer — the more letters there are in a word, the more difficult and complicated it becomes to correctly identify them when the letters are rearranged. Whereas, more diminutive and common words like “ball” and “bear” remain mostly unchanged and easy to recognize, longer, less common words, like “pollution” and “simultaneous” change substantially to the point where recognition is scarcely possible. Furthermore, this ability stems from a great deal of experience reading correctly spelled words — only people who can already read proficiently can do this task. This trick does not reveal much about the process of learning to read — it only indicates that highly skilled readers are so practiced with text they can overcome minor imperfections (emphasis on the word “minor”) in the print to access meaning.

Oh, and by the way, there never was any published research of this sort conducted at Cambridge or any other university. That part is completely untrue.

As you can see, the original author of this hoax avoided longer words because longer words are MUCH harder to read when scrambled. When there are only a few letters in the medial position, they aren’t so much “scrambled” as “interposed” which makes a big difference in word recognition.

When the original author of this hoax did use longer words, it seems she did not so much “scramble” the medial letters, but instead carefully placed the medial letters so as to avoid creating long words that are difficult to read.

Take the “scrambled” word “phaonmneal” for example. A long word that doesn’t really seem all that scrambled (and which, if unscrambled, would have too many letters). It still looks like “phenomenal” because the letters were placed in a fairly non-random sequence. Letting my computer randomize the letters in the longer words from the passage, we have:

PNONEEAMHL
RECASREH
CRGBIDMAE
URTVIIESNY
IPNAROTMT
BUSEACE
AAZNIMG

Even knowing that these words are contained in the passage, and even having just read the words recently, I bet these are not very easy for you to identify correctly. When the medial letters of longer words are randomly scrambled, the word becomes fairly unrecognizable.

The bottom line is that the preponderance of proper research (you know… research that has actually gone through a peer-review process and which has been published in reputable journals) indicates that all of the letters in words are quite important for word identification. Keith Rayner and his colleagues have conducted dozens of studies examining the importance of letter position, and has consistently found that, most of the time, letter position is critically important for efficient and accurate word identification (See for example Rayner, White, Johnson, and Liversedge, 2006).

For readers who would like to read more about this “Cambridge Effect,” I would direct you to Matt Davis’ wonderful website that goes into remarkable detail in examining serial position effects. Dr. Davis is a professor at Cambridge — yes, he is a REAL Cambridge Researcher — and he has a wonderful sense of humor in dealing with this issue.

ADDENDUM — January 5, 2009

An alert reader just wrote me and told me that there are some references that are often provided to substantiate the claim that the “Cambridge Effect” is based on actual research. The two references that supposedly support the idea that medial letter position is not important for efficient and accurate word identification are:

G.E. Rawlinson (1976). “The Significance of Letter Position in Word Recognition.” Unpublished Ph.D. thesis, University of Nottingham, Nottingham, UK.

And

G.E. Rawlinson (1999). Reibadailty. New Scientist, (162):55.

First, and most obviously, the 1976 citation refers to work conducted at the University of Nottingham, which I suspect is in England, but which I don’t think can be easily confused with Cambridge.

Second, this is an UNPUBLISHED thesis — if it was a worthwhile study, why was it never published?

Third, the more recent, 1999 citation does not contain any new data or new information — if you read the article, Rawlinson only makes some passing reference to his own, original (unpublished) dissertation thesis in that article.

Fourth, New Scientist is not a peer-reviewed journal — it is a news magazine that focuses on scientific topics. But it should not be confused with a real scientific, peer-reviewed journal.

So, yeah… I’ll stand by my initial claims — this is a hoax. A rather poor one at that.

References

Rayner, K., White, S.J., Johnson, R.L., & Liversedge, S.P. (2006). Raeding wrods with jubmled lettres: There is a cost. Psychological Science, 17, 192-193.