September 14, 2003

c-n y-- r--d th-s?

Languagehat has an entry concerning the decipherability of English texts made up of words that have had their letters scrambled (except for the first and last). [via Avva in Russian] I had written something about this phenomenon back in March with a different scrambled text. (I am always amazed how these unattributed texts can spread like folklore across the Web.) It was hard at the time to find a source for the quoted text, but I think I've traced it back to some work that Kourosh Saberi at UC Irvine and David R. Perrott at Cal State Los Angeles have done, mentioned here in an article by D. W. Massaro at UC Santa Cruz. I sent some email to Professor Saberi, but hadn't heard back from him. They wrote up their results in the 29 April 1999 issue of Nature, but I've been unable to find it online. Here's a press release for that article; see also this editorial in Nature Science Update.

[Addendum 09/15/03: The jumbled letters meme continues to spread. In good folkloric fashion, I've seen two variants: the first with "Aoccdrnig to a rscheearch at an Elingsh uinervtisy" and the second with "Aoccdrnig to rscheearch at Cmabrigde Uinervtisy". Here's another reference to the Saberi and Perrott article on the ABCNEWS dot com site.]

[Addendum 09/16/03: Thanks to Ed Snibble who tracked down the Nature article. I've posted a followup entry today, too.]

[Addendum 09/17/03: Thanks to Roger Willcocks who tracked down the source of the first jumbled text. See today's entry above.]

[Addendum 10/07/03: An actual researcher in cognitive neuroscience; computational modelling, and psycholinguistics at Cambridge has assembled a great page on this urban meme. I've added this link to help those folks who're still wondering into this entry long after it feel off the front of the blog.]

Posted by jim at September 14, 2003 06:57 AM | TrackBack
Comments

The hunt begins!

It looks like Saberi and Perrott were dealing with spoken language reversal, not written text. Perception-wise, this is a pretty different domain. So I doubt that their work is the real origin of this meme.

But there's definitely an element of truth to this, in that people do read primarily by looking at the shape of the word rather than the individual letters. A type designer named Hrant Papazian has been doing some work in this area, dealing with the blurry forms that words take when they're in your peripheral vision:

http://www.themicrofoundry.com/ss_read1.html

Thanks for starting this thread. We'll get to the bottom of this yet!

-Cassidy Curtis, alphabet obsessor
http://otherthings.com/uw/alphabet

Posted by: Cassidy Curtis on September 15, 2003 04:06 PM

Thanks, Cassidy. I'll take a look at Papazian's site. I've also left a comment on your entry with some other observations.

Posted by: jim on September 15, 2003 05:04 PM

fukc yuo !

Posted by: mistia on September 15, 2003 06:14 PM

you have got to be kidding me!

Posted by: gs on September 15, 2003 06:30 PM

Woo hoo! Bloggy, bloggy!

Posted by: Al Chumley on September 15, 2003 06:50 PM

Perhaps the important bits of a word are its beginning, end, visual size, and context. Can you read this?


Posted by: Michael on September 15, 2003 07:59 PM

Paris in the
the Spring

Posted by: Nadrian X Underwood on September 15, 2003 08:40 PM

This seems to have some applications for privacy. If the government is trying to filter e-mail or files or newsgroups for certain "bad" words, if you make your posting after putting it through a scrambler that mixes things up like this (except for the first and last letters) it would make it a lot harder to find the "bad" words, but your message could still get through.

Posted by: Matthew Kramer on September 15, 2003 09:17 PM

Hmm, I don't know how good the applications for privacy would be.

To keep in simple (since I somehow stumbled here and have no idea of you peoples expertise in this sort of thing), in cognitive linguistics we talk alot about three issues - semantics, syntax, and orthography.

Semantics is the meaning of a word (i.e. how do you represent "justice?"), syntax is your eighth grade grammar class all over again (e.g. verbs, nouns, etc.), and orthography is just fancy technospeak for privacy.

Wherever this thing came from (and what a meme, it's been sent to me 3 times in the past two days), at a casual glace it falls primarily within syntax and orthography, but the main issue is how do you apply a semantic understanding to words that are all mixed up?

Ok, so circumventing filters with this sort of thing wouldn't work, in my opinion. Why? Well, if it was me, I'd just write a couple lines of code that played with the orthography in a systematic, omitting all the first and last letters until a match was found. Nothing fancy, just brute force.

Also, when you talk about orthography, orthographic neighbors and word frequency come up. I.E., words that are spelled similarly a particular word. Certain words have a higher written word frequency, and have a higher orthographic neighbor frequency, and within a connenctionist model are therefore harder to recognize (I bet you stumbled over a couple of words, at least. If we went back, you'd and checked, you'd probably find that the WWF and ONF are higher for the words that you stumbled on). In short, you could make your filter more elegant by assigning higher probabilities to certain sets of letters with a higher WWF.

One last word or warning about this meme, and about writing this way. Go back and look at the actual text. None of the words are shifted out of their orthographic position more than one or two letters to the right or left. For example, "uinervtisy," for "university."

Try "usitvrenit" instead (take the already scrabled letters and reverse them). Because it isn't as simple as "the first and last letter stay the same, and you read the word as a whole." Uh-uh. Orthography is just as systematic as syntax. If this statement was true, then all you would have to do to work a word scramble problem is place one random letter at the front, another at the back, and PRESTO, you win. But in life, word scramble is rarely that easy.

Hope I provided food for thought.
S.W.

Posted by: S.W.B. on September 15, 2003 10:14 PM

Ak, sorry.

Orthography is a fancy word for spelling.
[geek]
Accessed the wrong area of my lexicon. ;-)
[/geek]
S.W.

Posted by: S.W. on September 15, 2003 10:16 PM

Hebrew is written without vowels thus, YHWH is actually Yahweh- seems that they've been getting along just dandy without a significant portion of the middle of their words. :-)

also, is there perhaps something in our modern (after the advent of instant messengers) mentality/perception that allows out to quickly decipher such "scrambled" words, i.e. in the fashion of many mistyped words that may have crossed out earliest paths in pursuing the task of speed typing to keep up with our thoughts and conversations?

Posted by: SN on September 15, 2003 10:46 PM

Mmm, you argument with Hebrew and speed of reading scrambled words seems to be slightly oxymoronic. Hebrew has been around for awhile, afterall. ;-) But an interesting point, nonetheless.

I don't spend enough time on the net. Just found my argument about how much the words are scrambled over on languagehat. I seem to be too slow in the fast paced world. I hate being original, and then finding out that I was really unoriginal.

S.W.

Posted by: S.W. on September 15, 2003 11:22 PM

I might be getting a little off the subject...
hebrew, as well as other semite languages (arabic) uses (mostly) 3-letter 'roots' for (almost) every word. words with the same root have similar/associative meaning. maybe that eases reading it without vowels.
by the way - although the bible is written in 'missing spelling', it's written with diacritics, so, there is really only one way to pronounce it (with few exceptions).
modern hebrew ('partial spelling') uses vowels here and there, while diacritics are used today mostly in children's books, but also sometimes when there is an ambiguity between 2 words that are spelled the same.

Posted by: boomerang on September 16, 2003 12:46 AM

Try "usitvrenit" instead...

Would be better if written right "usitvreniy"

Posted by: JK on September 16, 2003 12:51 AM

I was thinking that - however he's right, since that still doesn't present 'university' to me.

Posted by: Carl Farrington on September 16, 2003 01:08 AM

Ok, so I look fairly stupid in my details. Apologize to all who find technical faults/misspellings/grammatical errors/general incoherence with my posts. I haven't been to bed for more than a couple of hours in close to four days, and these posts are the bulk of my breaks every few hours.

Ah the life of a student.

Nose back in the books,
S.W.

Posted by: S.W. on September 16, 2003 01:39 AM

I would suggest some sleep.
then some more sleep.

As you reduce the levels of available neurochemicals through use, you lose the ability to concentrate etc.

Eat properly, sleep properly, learn v much faster.

and learn to speed read, it really helps.
You might find this useful, [http://www.winblit.com/], or at least fun.
I can read Alice off gutenburg in 30 minutes ish..

Finally, reviews.
If you review material 5 times over a period with increasing time spans between,
ie: read, 1 day later, 1 week later, 1 month later, 6 months later.
You will have an 80% retention for life.
This is why speed reading is important.

Posted by: ross on September 16, 2003 02:41 AM

Och, I've been telling people something similar for years. Think about it this way, how many people recall auditory learning versus visual learning? The mind itself processes information primarly through the use of it's main "senses". As such, how you learn to use those senses as a baby, plays a major role on how you later interpret all other information. Thus, those people who are more visual stimulated (by behavior, and by genetics), reading becomes less a letter by letter thing of order, and more a picture of whole.

If one were to think of words as a picture, then one letter shifted slightly, doesn't break a the recongized pattern, it just changes it to a slightly different hue (re, rainbows and color depths). Auditory on the other hand, concentrates more on the vowel order, and thus you find those who are more auditory in nature to have more trouble with an out of place vowel, or at the very least having to slow down when reading an out of place letter.

Course, if one where blind, and deaf since birth, you'd end up even being more linear then auditory learners, as letters almost envariably have to be written down (printed in brail) to be understood.

The biggest problem I have, is that up until six months ago, no one quite believed me when I said that I see numbers as colors, and that words were like pictures. The only slightly doubted me less when I showed a 65% comphrension rate speed reading at 2500rpm. Auditory learners just can't do that, even normal visual learners still read patterns, not entire images.

Food for thought.

Posted by: chris on September 16, 2003 04:55 AM

This seems to be the Science piece: http://www.reversespeech.com/images/bakspeech.pdf

Posted by: Ed Snible on September 16, 2003 08:42 AM

This seems to be the Nature piece: http://www.reversespeech.com/images/bakspeech.pdf

Posted by: Ed Snible on September 16, 2003 08:42 AM

chris,

Seeing colored numbers sounds like a form of synesthesia. Interesting recent article in Scientific American can be seen here.

http://www.sciam.com/article.cfm?articleID=0003014B-9D06-1E8F-8EA5809EC5880000&chanID;=sa009

Yes, little sleep is bad. But, no rest for the wicked, and the righteous don't need sleep. Or so they say.

Posted by: S.W. on September 16, 2003 10:57 AM

The original reference is a letter to New Scientist magazine by Graham Rawlinson of Aldershot, Hampshire (vol 162 issue 2188 - 29 May 1999, page 55) titled "Reibadailty".

Rawlinson writes: 'You report that reversing 50-millisecond segments of recorded sound does not greatly affect listeners' ability to understand speech (In Brief, 1 May, p 27).

'This reminds me of my PhD at Nottingham University (1976), which showed that randomising letters in the middle of words had little or no effect on the ability of skilled readers to understand the text. Indeed one rapid reader noticed only four or five errors in an A4 page of muddled text.

'This is easy to denmtrasote. In a puiltacibon of New Scnieitst you could ramdinose all the letetrs, keipeng the first two and last two the same, and reibadailty would hadrly be aftcfeed. My ansaylis did not come to much beucase the thoery at the time was for shape and senqeuce retigcionon. Saberi's work sugsegts we may have some pofrweul palrlael prsooscers at work.

'The resaon for this is suerly that idnetiyfing coentnt by paarllel prseocsing speeds up regnicoiton. We only need the first and last two letetrs to spot chganes in meniang.

'This was not easy to type!'

The letter is in New Scientist's searchable online archive (archive.newscientist.com)

Posted by: Roger on September 17, 2003 05:49 AM

Roger: Thanks! It's nice to know who wrote the original text. I'm not sure we'll know who wrote the second and more recent one.

Posted by: jim on September 17, 2003 07:06 AM

Question for Chris

I hope you will see this...I hope you will answer it.

That you see numbers as colors is FASCINATING. Has it always been like this? So can you deal with mathematical operations like people who see numbers as numbers...i.e. add and subtract. etc?

And do you see a specific shape for each color shade or just a blob of color?

I take it you see letters as letters?

If you choose to respond, it would be helpful if you could email me at wumply@metrocast.net

John

Posted by: John D. Miller on September 17, 2003 11:13 AM

S.W. makes an interesting point - about the distance the shuffling takes. Some time soon, I'll write a program to shuffle text in that way. There are millions of pages on etext, so that's a good source. Minor shuffling - transpositions &c; - amount to simple misspelling.

Would "useinivtry" be as quickly recognized? (Certainly not out of context.) Or even better
"utisreviny" (the middle letters reversed - very close to JK's suggestion). Or even "ueiinrstvy" (the middle letters in alphabetical order). Clearly, there's a range of randomness that can't be exceeded.

In that word, the clues are the syllables, "un, i, ver, si, ty". The scrambled word in the sequence is "u, in, erv, tis, y", which really isn't that far out.

I think the "apochryphal sample" may have been "cooked" so as to avoid unintelligible combinations. I'll find out once I start shuffling etext selections.

Posted by: Mike on September 17, 2003 11:42 AM

Mike: Yes, I think these minor shufflings make this akin to classic dyslexia.

Posted by: jim on September 17, 2003 05:50 PM

Am I missing something? Rearrange the letters of "rscheearch" and you get "research" plus a spare "ch"...

Posted by: Chris on September 19, 2003 09:48 AM

Chris: Yes, "rscheearch" seems to be a misspelled "research" or "researcher". The scrambled important was misspelled, too.

Posted by: jim on September 19, 2003 10:12 AM

Hmmm, does this mean we need an anagrammatic spellchecker, to make sure scrambled words are misspelled correctly?

Posted by: Tim on September 22, 2003 04:49 AM

Here's a Microsoft C program that scrambles the inner letters of words (but not too much):

extern struct{char*a;int b;char*c;int d,e,f,g;char*h;}_iob[]; void main(){int a,b,d,e,f;char
c,g[256];for(f=fgetc(&_iob[srand(time(0)),0]);f!=-1;)
{for(;f!=-1&&!isalpha(f);f=fgetc(&_iob[fputc(f,_iob+1),0]));if(f!=-1)
{for(e=0;(e if(a=(e>3))do for(d=b=e-2;--b;rand()&1||(g[g[b]=g[c=g[a=0,b],d],d]=c),--d);while(a);
fprintf(_iob+1,"%s",g);}}return;}

Posted by: James L. Dean on September 26, 2003 04:59 AM

Interesting thread, Regarding the interpretation of shapes and colors as language, this reminded me of "Shape note singing", a wonderful style of group singing where the shapes and colors, rather than actual notes written on a staff represnt the notes. While it is lead by conductor( for lack of a better word), it is a quick and easy way of getting a large group of people who do not read music to all "sing from the same page", in a very short period of time.

-Kurt

Posted by: Kurt Siegel on October 5, 2003 05:52 AM

Interesting thread, Regarding the interpretation of shapes and colors as language, this reminded me of "Shape note singing", a wonderful style of group singing where the shapes and colors, rather than actual notes written on a staff represnt the notes. While it is lead by conductor( for lack of a better word), it is a quick and easy way of getting a large group of people who do not read music to all "sing from the same page", in a very short period of time.

-Kurt

Posted by: Kurt Siegel on October 5, 2003 05:52 AM

Interesting thread, Regarding the interpretation of shapes and colors as language, this reminded me of "Shape note singing", a wonderful style of group singing where the shapes and colors, rather than actual notes written on a staff represnt the notes. While it is lead by conductor( for lack of a better word), it is a quick and easy way of getting a large group of people who do not read music to all "sing from the same page", in a very short period of time.

-Kurt

Posted by: Kurt Siegel on October 5, 2003 05:52 AM

Sorry about the triple post - slow connection didn't seem to react when I hit the "post" button, and I though my mouse was acting up again.

-Kurt

Posted by: Kurt Siegel on October 5, 2003 05:55 AM
Post a comment