Semi-Daily Journal

Cleverly noted. I thought at a glance either I could not read this or would not care to waste the time, but I can actually read it at a glance.

Notice of this fact has been spreading like wildfire in the blogspace. The weird thing is that this phenomenon was blundered into by folks who had heard about a 1999 Nature paper (Saberi & Perrot) where they inverted every 100 msec segment of connected speech and found it intelligible... The "extension" of this to text is (I think?) pretty novel. As it turns out, a fair number of cognitive psychologists have had some sense that something lke this could work, but as far as I know, few had the strength of their convictions to see how far it would go (quite far, as it turns out).

That said, the original does say some stuff that ain't so:

> Tihs is bcuseae we do not raed ervey lteter by it slef but
> the wrod as a wlohe.

Atclulay, the hmuan mnid deos not raed the wrod as a wohle if waht you maen is mtahcnig a tmepltae of a wolhe wrod. If it did, tehn tihs knid of wnitrig wulod be pterty hlpsoses.

Pyalnig aournd wtih tihs smoe mroe, hwoveer, it smems taht inletlgiilbitiy is imvpoerd if you keep ltteres whtiin the "bdnouraeis" of the mrophmee tehy benolg to. Tehre is smoe recserah sgugetsing taht eye fxitaonis can tgreat mrohpmees usnig non-psotiional sttasitcis.

OK, and it turns out that there are some interesting things happening with backwards text as well:

An ldditionaa gindinf is that eeoplp nac dear text eherw lnitai dna linaf setterl era tusj dwitches; yhet khint eht text si sackward hlthouga ti yeallr tsn'i.

And one variant that *has* gotten some study:

Wealsoknowthatafterabriefadjustmentperiod,peoplecanread
textwithoutanyspacesinitandshownormaleyemovement
patterns.Thattellsusthattheinformationcontainedinspacesis
highlyredundant.

But there are limits:

Itislsescaelrtometahtyoucancmboinetsehetwowcffetswhtouit
cuainsgmroedamrtiacsowlniginrdaineg,bsaecueteh"nosepacsecfeft"
uesslteter-letteertsirnaioitalpboarlbieitilstofnidwrodbindoraeusaswlel.

So that last paragraph was probably painful, but not nearly as painful as you might have been expecting. In other words, it isn't impossible when by rights it should be.

The atuhor mlepsilss "rsceheearr"

Got it. Suddenly remembered that this was long ago discussed in a series of brilliant lectures at Princeton by Ernst Gombrich.

Ernst Gombrich was among the finest art historians and critics, and philosophers, of the century.

See: Art and Illusion - simply brilliant

http://www.guardian.co.uk/obituaries/story/0,3604,587634,00.html

Gombrich engaged for 50 years in a polemic against invoking the collective mind - whether of an age or a nation or a class - as explanatory of changes in art or politics. He did so because he saw such explanations as not only circular but as failing to recognise the essentially rational nature of the way artists experimented and learned from each other.

The work in which he set out to replace the formalisms of the turn of the century was Art and Illusion, first published in 1960 and based on his Mellon lectures given in Washington in 1956. It presented an account of the psychological factors which made it possible for us to see a three-dimensional moving subject - such as people in action - on a flat, still surface. The painter learned to do this by trial and error, checking whether his marks elicited recognition of his subject.

This led Gombrich to argue that the major factors in changes in pictorial style were the result of rational activities rather than mysteriously changing expressions of the age. He was deeply opposed to any account of artistic creativeness which was couched in terms of a collective psyche rather than by reference to individual invention and discoveries which others could then adopt....

Woha, ddue.

Woha, ddue.

I consider this a long overdue response by the NEA, AFT, and other teachers' unions and accreditation organizations to rebut conservatives' effort to bolster support for imposing "phonics" over the more modern "whole language" (look-say, sight-reading, etc) methods of instruction. Ovbioulsy, if teh odrer of letrtes mkaes so litlte dfiferecne taht a gitfed raeder can disecrn maennig, tehes are the tehcniuqes taht shulod be taguht to beniging raedres as wlel: geussnig-and-chceking, conxettualiznig, shpae-recogntiion -- the enitre tooklit of waht mnay porfsresoianl eudactors clal the "psyhco-lnigiusitc geusnsig-gmae".

Is't vtial taht tsehe eivl-deors be defaeted on the phnoics frnot, or esle thye'll tkae oevr the txetoboks on sujbects scuh as evloutoin, hisotry,
cvics, and enconmoics as wlel. Tehy seek nohting lses tahn a retrun to the benghited era of chruch-run shcools of the eihgteenth- and ninteneeth- centruy wehn litreacy was merley nintey precent and pulbic discuorse otfen inculded refernece to
daed-wihte-mlae-donimated cutluarl peices such as
Aeosp, "Hloy Scirptrue", and Skahespaere.

A few general comments:
1) My reading speed drops dramatically on mangled paragraphs. Just because we can read it, that does not mean we should have to.
2) Has anybody tested comprehension of mangled text? Does it improve or degrade it?
3) On the Phonics vs Whole Language debate: Those are schools of thought on how best to TEACH the language. How well highly literate adults read mangled text is questionably relevant to teaching children. Having the extra redundancy probably helps kids learn.

rbb

Not as simple as it looks:

http://www.crookedtimber.org/archives/000525.html

The two best things we can can do for America's children are to a) teach phonics and b) make it a prison offense to hire a member of a teachers' union to teach in the public schools.

Language emerged as speech, and understanding speech requires sophisticated signal processing to break up the sound stream into words, deal with ambiguities, and filter out noise. These are seriously difficult problems, which is why voice recognition software has taken so long to develop. Why should anyone be surprised that we ca also deal quite efficiently with ambiguity, blur and noise in the artificial scheme we call reading? But since writing allows complete clarity at the symbolic level, let's have it.

One of the milestones in writing was the empty space - Greco-Roman insriptions are continuous; Carolingian manuscripts are beautifully separatd. Any advance on crediting Charlemagne's chancellor, Alcuin of York?

It works even when English is not your mother language although the more difficult parts are very hard then.
Some 25 years ago in psychology-classes we learnt about something similar. In an experiment some part of every single letter was cut away. Some 40 or 50% could be left out. Best performances came from individuals that used softdrugs before the experiment!

Mobius Klein writes

"Having the extra redundancy probably helps kids learn"

The same also goes for speech. Languages have redundant sounds built into them. We can understand speech without them. As languages develop these redundant sounds are progressively dropped BUT new ones are simultaenously added changing words and meanings. (It is one of the ways new languages develop out of a common root).