The scientific study of readability is obscure enough, without delving into its historiography.
In what other subjects is 50-year old research relevant today?

Or is it because the subject is so obscure and thin on the ground that such classics as Tinker are referenced?
Or is it because a history of the subject will help to debunk the old myths?
(The prevailing wisdom on readability amongst working typographers and graphic designers still acknowledges the predominance of word-shape recognition, which present-day researchers have determined is incorrect.)

Nick, maybe it's because evolution is really really slow...

> Does anyone on Typophile have access to the
> Munoz, et. al., paper, or the journal it is in?

Let me check UCLA.

> Apparently it's #396 in the Spencer bibliography.

I had borrowed the book from UCLA, and made it a
point to copy that fabulous bibliography! I'll look it up.

hhp

… evolution is really really slow ...

Incidentally, this relates to what was once a major political controversy in the United States. This dates back to the book "Why Johnny Can't Read".

Essentially, while it is true that competent adult readers read by reading whole words, the older traditional way to teach reading, by phonics, while boring to the smarter kids who already know how to read, having been taught by their parents at home before starting school, at least works for every child - while sight-reading often fails to reach some children.

… while it is true that competent adult readers read by reading whole words …

The myth.
Haven't you been following what Kevin Larson has had to say about this at Typophile?

Nick, please stop.

hhp

@Peter: I have the following article if it is of any help.

Tinker, M. (1932). The influence of form of type on the perception of words. Journal of Applied Psychology, 16(2), 167–174.

Abstract
The perceptibility of words and groups of unrelated letters when printed in lower-case and in all capitals was obtained by means of the distance method. Both capital letters and words in capitals were read at greater distances from the subject than letters or words in lower-case. There was only a small and statistically insignificant difference between distances at which unrelated capital letters and words in capitals were read correctly. With lower-case type, however, the difference between distances for apprehending words and unrelated letters is greater and the direction of difference is very stable statistically. These findings indicate that total word-form is more potent in the perception of words in lower-case than in all capitals, where perception seems to occur largely by letters. This conclusion is supported by the fact that the reading of words in lower-case yielded more misreadings than the words in capitals, and is suggested as an explanation of the faster reading of text in lower-case type.

Hrant, I don't think Nick should stop pointing out that the repeated claim that we ‘read by reading whole words’ has been fatally holed below the waterline by several decades of studies that set out to test that hypothesis. As you know, I'm willing to entertain the possibilities put forward by Peter and others that some recognition of feature patterns above or across individual letters may play a role in reading, but that's something quite different from the claim that we read whole words, i.e. that words are the unit of recognition rather than the outcome of recognition.

I'll be nice and not say what I want Nick to stop doing. It's never worked before anyway. However it's still useful to plant a seed of doubt among the uninitiated.

John, the idea that we read whole words at a time is as blantantly illogical as the idea that we only read letters (sequentially or not). We read as much as we can at a time; that type of efficiency is at the heart of the heuristic, non-algorithmic system that we are. The dynamic, individual unit of reading is the bouma, and it can be anything from a single letter (especially in the fovea where the resolution affords such certainty) to an entire word (hey, maybe even multiple words). DEPENDING. On a dozen things at least, including the meat of our domain, fonts.

Peter's position BTW is so far ahead of the basic realization above that it's actually a disservice to state it. You teach high-schoolers Newton's Laws, not Einstein's theories. When you see an elephant charging towards your village you don't try to teach it English first so you can then explain why it shouldn't be trampling your home; you either divert it, or you shoot it.

hhp

Thanks Christopher. I have that paper too. But I don't think it's the one Spencer was summarizing. Hrant, I'll be much obliged if you could check.

Nick, I'm trying to get the history straight because I want to know how we got to where we are. If historical researchers around Tinker's time had a holistic view in which the information that is used involved a good deal more than the outline form and encompassed internal details, a debunking of holism by attacking just the notion that the form as outline is what is used or decisive is beside the point.

It is possible to hold that word wholes are normally the unit of perception in continuous reading while maintaining that word shape or the total word form — in the sense of the envelope structure of the word — hardly plays a part. Even Denis Pelli finds word shape in almost this restricted sense plays a role.

The debate between Kevin et. al., and myself is: is independent letter recognition the central mediating process for recognition of word wholes, or are role units — units at the granularity of letter parts, the way they are connected snd organized across an uncrowded span — the critical mediating agents in the rapid automatic visual word-form resolution which subtends sense-following. The items that letter nodes project to in the interactive activation scheme are word wholes. So even in this scheme there is a grasping of word wholes.

I'll even grant the Hrantian caveat.

Hrant: We read as much as we can at a time; that type of efficiency is at the heart of the heuristic, non-algorithmic system that we are.

But reading ‘as much as we can at a time’ is not efficient, reading only as much as we need to at a time is efficient, and that fact that we make reverse saccades to correct word recognition indicates that our default efficiency favours occasional errors over exhaustive information intake.

> reading only as much as we need to at a time is efficient

Agreed. But that just means we don't need all the letters.

And reverse saccades (AKA regressions) are indeed a clue
that something beyond letterwise decipherment is happening.

hhp

Yes: a clue that some of the time that we think we don't need all the letters we are wrong. :)

Since it has been determined elsewhere here that the letters most readers see on screen are fuc... er "quality challenged", you all might better hope people read words, eh!?

John, regressions exist because when they don't happen we're going faster because we're skipping letters (or rather, assimilating them into boumas) - it's just that sometimes we're wrong. To me regressions are evidence that we try to be right at the limit.

hhp

Rayner, K. (1998). Eye movements in reading and information processing: 20 Years of Research. Psychological Bulletin, 124(3), 372–422. (pp 375).

Although most saccades in reading English are made from left to right, readers do not relentlessly go forward: About 10-15% of the saccades are regressions (right-to-left movements along the line or movements back to previously read lines). Many regressions tend to be only a few letters long and could be due to the reader making too long of a saccade, in which case a short saccade to the left may be necessary for reading to proceed efficiently. Short within-word regressive saccades may also be due to problems that the reader has processing the currently fixated word. Longer regressions (more than 10 letter spaces back along the line or to another line) occur because the reader did not understand the text. In such cases, good reader are very accurate in sending their eyes to that part of text that caused them difficulty (Frazier & Rayner, 1982; Kennedy, 1983; Kennedy & Murray, 1987a, 1987b; Murray & Kennedy, 1988), whereas poor readers engage in more backtracking through the text (Murray & Kennedy, 1988).

I forgot that I read this some time ago. It goes on for considerably longer and covers a lot of interesting material. Definitely worth a read if you haven’t already. One of the better papers I have come across.

Sorry, my intent wasn’t to generate debate, just to get accurate information about where the passage I quoted in my Typo text from 2003 originated, and what it's actual form was in the source document. Also I wanted to track down what is the actual Miles Tinker text that Herbert Spencer was referring to when he summarized Tinker’s take on how perceptual processing in reading works.

20 years?
Saccades were discovered in the 1880s.
In the popular consciousness, Life magazine did a famous article (infographic) in the '40s/'50s, comparing male and female saccadic movement over the same picture.

**

Peter, I'm very interested in your historical perspective.
Speaking of perspective, it's a lot like reading, a way of using technology to order visual coding into what is assumed to be natural perception.
Erwin Panofsky established linear perspective as a subject of cultural study, Ernst Gombrich popularized it. Hugo Damisch went postmodern.

Drawing a parallel (so to speak) between reading and linear perspective, in recent years we've seen a rise in the use of 3-point perspective, replacing 2-point, in media (especially virtual environments), at the same time as a rise in the use of sans serif.
These are cultural changes which condition so-called "natural" physiological processes.

“The first era of eye movement research extended from Javal’s initial observations concerning the role of eye movements in reading in 1879.” (Rayner, 1998, pp 1)

This is the 4th sentence of the paper. Twenty years refers to the fact that Rayner is simply reviewing the last 20 years of research, not stating that there has only been 20 years of work done. Strangely, Javal does not appear in the references section. I haven’t read much Javal, certainly none in French, but I suspect Rayner is referencing:

Javal, K. E. (1879). Essai sur la physiologie de la lecture. Annales d’Oculistique, 82, 243-253.

Not sure why he is referenced as K. E. I have always known him as Louis Émile Javal. I have noticed the same typo in different papers from different authours I suspect a typo due to the fact that K and L and next to each other on the keyboard and people often copy/paste references when writing papers.

The same spelling exists in other papers. Probably a Can anyone shed light on this?

Peter, no need to apologize for triggering debate! :-) I'll be digging up
those references - it's just that our garage has become quite inhospitable...

> Short within-word regressive saccades may also be due to problems
> that the reader has processing the currently fixated word.

I posit that this happens only when:
- The word is extremely rare and similar in shape to a more common word.
OR
- The font is flawed.

> Saccades were discovered in the 1880s.

Formally, yes. But one has to suspect that observations of saccades might even be as old as the Ancient Greeks, who were superb observers.

> comparing male and female saccadic movement over the same picture.

Although of course there must be a difference between "2D" saccades performed deliberatively versus lateral saccades performed unconsciously.

> These are cultural changes which condition
> so-called "natural" physiological processes.

- There's no need for the "so-called" or the quotes. Those processes are real - more real than many of the other factors we like to dream up to justify our artistic tendencies.
- There is a limit to such conditioning.
- The other way around is more significant.

Christopher, I'm sure you're also aware of Javal's landmark "Physiologie de la lecture et de l'écriture" of 1905.

hhp

Although of course there must be a difference between "2D" saccades performed deliberatively versus lateral saccades performed unconsciously.

The way that men's saccades generally scoped the physical features of women represented in those mid-century "Life" pictures is assumed to be as unconscious (involuntary) as reading saccades.

Isn't it also possible to re-learn how to read?
Typically, the goal is speed.
This could be compared to re-programming a global skill, such as a golf swing or tennis serve.

There's no need for the "so-called" or the quotes. Those processes are real …

No one's saying they're not real.
But they are cultural, not natural.
Are cultures without written language unnatural?

In the same way, one speaks of "natural" media, as if pen and paper is not a technology, but mouse and computer is.

There is no natural way to read.
Some read right to left, others left to right.
Some use phonetic alphabets, others a logographic system.

- There is a limit to such conditioning.
- The other way around is more significant.

Significant to you perhaps, but it is barely recognized by reading research (with, for instance, its un-typographically-informed axioms about which fonts are easy to read), and it is what I find most interesting.

Hugo Damisch went postmodern

Hubert, rather.

> Isn't it also possible to re-learn how to read?

To some extent, sure. But the efficiencies of how we actually read kick in on their own; we're never taught to saccade - certainly not backwards! We're taught to read one word at a time, compiling the letters sequentially; some people, and only after a certain age, are taught to read words at a time, but this is not done explicitly. True immersive reading happens pretty much in spite of how we're taught!

> But they are cultural, not natural.

The composition of the human retina isn't cultural.
And you cannot over-ride its composition via any conditioning.

> There is no natural way to read.

Not absolutely, but some ways are certainly more efficient than others. For example the fact that the retina's acuity is more horizontal than vertical (probably because during evolution dangers came from the side much more than from above or below) means reading vertical text is slower. As physical creatures we are simply more horizontally capable than vertically.

> Significant to you perhaps

Significant to anybody paying attention and caring about users, as opposed to fabricating and promulgating excuses to justify self-expression.

> it is what I find most interesting

To me the shortcomings of scientific testing of reading aren't interesting, they're just sad.

hhp

… fabricating and promulgating excuses to justify self-expression.

It could certainly be argued that the relativism in cultural studies legitimizes self-expression, but is that bad?
A distinction must be made between self-expression and self indulgence, but even this has its benefits, "the road of excess…" as Blake quipped.

However, quite apart from their implications, cultural studies are also a means and an end in their own right, as a form of knowledge and understanding, and that is what I find significant, especially as it plays so small a part in cognitive research, and that, in my opinion, is a sad shortcoming.

As I'm fond of saying: nothing made by a human can avoid self-expression, and nothing made for a human should avoid self-expression. But that doesn't justify averting one's gaze from our physical realities.

hhp

> In what other subjects is 50-year old research relevant today?

One of the great features of scientific experiments is that you expect that a study rerun today would have the same results as 50 or 100 years ago. I expect this to be true in chemistry and I expect this to be true in reading psychology. The interpretations of the experiment findings certainly change, but the data should not. I think all fields, certainly reading psychology cite 50 year old research. The classic papers tend to be older. I have had the great fortune of attending a few typography conferences. I’m quite sure I saw the occasional older piece of work referenced.

> For example the fact that the retina's acuity is more horizontal than vertical (probably because during evolution dangers came from the side much more than from above or below) means reading vertical text is slower.

Hrant, as you know I disagree with many of your assumptions, such as immersive reading being a qualitatively different reading process, but I don’t think I’ve seen this assumption before. Where did you learn that the retina’s acuity is more horizontal than vertical or that reading vertical (presumably Chinese or Japanese) text is slower?

Actually the data should change (and that's a great feature of being human :-) because some aspects of our behavior do change. For example the changes in how young people spend their time certainly affects their reading performance (even though it doesn't change the great relevance of their physical constancy with older generations).

hhp

> Where did you learn that the retina’s acuity is more horizontal than vertical

I don't remember. But is it not true that we see more detail laterally than vertically, deeper into the field of vision? However if I'm wrong on that it still doesn't mean we're ethereal, purely cultural beings. I think we can all agree for example that the human neck allows us to look left/right more than up/down (even though that's not very relevant to reading per se).

> or that reading vertical (presumably Chinese or Japanese) text is slower?

There are many more things than orientation going on when comparing Latin and Chinese typesetting, so arriving at usable conclusions there is even harder than within Latin! What would be useful is to compare horizontal and vertical Chinese reading. But, as always, first one needs to enable immersion in the test subjects - which I maintain nobody has done yet.

hhp

BTW, Chinese typesetting has been moving towards more horizontal setting. Some of that is probably cultural influence from the West. But I suspect some of it comes from an implicit realization that horizontal is read more easily. You don't have to know why something works to implement it. For example how any typographer worth his salt will not set a book in a sans, even if he can't prove it works better. You cannot explain that away by claiming they're simply following convention.

hhp

Backwards saccades do not necessarily indicate skipped information.

The eye naturally has to move back and forth just in order to see - if you stare constantly at one point, you will find the scene you are looking at fading away unless you move your gaze. The nerves behind the retina lead to the eye registering changes and differences in brightness and color more effectively than static levels - even though the pigments directly react to the level of incoming light, saturation issues in the rods and cones themselves are also part of what make saccades useful, IIRC.

@hrant:
BTW, Chinese typesetting has been moving towards more horizontal setting. Some of that is probably cultural influence from the West. But I suspect some of it comes from an implicit realization that horizontal is read more easily.

The fact that the human visual field is wider than it is high largely stems from the fact that we have two eyes. Since Chinese doesn't have hyphenation issues, the fact that the ideal line length is shorter for vertical writing than horizontal is likely not a problem.

I would tend to think that Western cultural influence is the overwhelming cause of the change as far as mainland China is concerned. Remember how the Cultural Revolution sought to purge China of all traces of its traditional Confucian culture. So there was really no time for individual typographers to gradually come to decisions about what works better.

Now, in the case of Japan - and to a lesser extent, Taiwan - the process could have worked as you say.

My suspicion - merely a guess - is that in Taiwan, although it happened later, the switchover was nearly an all-at-once thing when certain governmental or prestigious publishers decided it was time to stop being different for the sake of being different - even if, to this day, the vertical format is used by those who want a traditional look.

In Japan, though, manga to this day are published with vertical text in their word balloons, and thus a right-to-left ordering of the panels. Horizontal text is just easier to do with existing computer systems, and so what seems to be the case is merely that the Japanese still prefer vertical text, but just don't think achieving it is worth the extra trouble to handle it on computers that need to be useful for handling Western languages as well.

So I think that any intrinsic superiority of horizontal text over vertical never really had a chance to have more than a negligible influence on the process - politics in mainland China, and laser printers and computer software in Japan (and probably also Korea) almost completely account for the change.

Wandell’s excellent textbook Foundations of Vision says “Anatomical images of the cone mosaic (see Figure 3.4) show that in the central fovea the cones form a tightly packed triangular lattice, so that we do not expect any large effects at different orientations of the stimulus” (p.62)

I don’t know if there are advantages for either vertical or horizontal layout in Chinese and Japanese typography. I know I wouldn’t assume one way or the other because one form is currently the convention. I also wouldn’t make the same assumption for serif Latin text.

Actually the data should change (and that's a great feature of being human :-) because some aspects of our behavior do change.

Weren't you just arguing that culture has little relevance to our physical realities?

N.S.> Weren't you just arguing that culture has little relevance to our physical realities?

Don't'cha know? People supposedly read from different distances depending on what OS they use, despite matching device resolutions. Sounds to me like computer culture supposedly having spacial relevance in our physical realities...;)

The Tinker source I’m looking for is most likely:
Tinker, Miles A., “How Children and Adults Perceive Words in Reading,” Invitational Addresses, 1965, Tenth Annual Convention, International Reading Association, pp. 75-93. Reprinted as Chapter 11 in: Readings in reading: practice, theory, research. Theodore L. Torgerson. 1968, pp. 75-88;

The “natural form of reading" statement goes back to this:

“Comparan los trazados obtenidos durante la lectura por diferentes sujetos, que revelan claramente que la forma de leer corriente, no es deletreando ni silabeando, sino captando grupos de palabras, lo que constituye la base fisiológica de la nueva metodologia de la lectura.”

From: Munoz, J.M., Odoriz, J.B. and Lavazza, J. “Registro de los movimientos oculares durante la lectura”, Revista de la Sociedad Argentina de Biologia, XX, Abril de 1944 page 284.

Peter, I found my 1963 Tinker book.
Could you remind me what you needed?
And something from Spencer too?

hhp

Thanks Hrant.

I was looking for #396 in the Spencer bibliography.

Apparently it's the 1965 Basis for Effective Reading. I suspect it contains a version of “How Children and Adults Perceive Words in Reading,” which I believe is the source of Spencer’s information about Tinker’s use of the distinction between ‘total word shape,’ the bare outline of a word, and ‘total word structure’ which included internal form elements.

Now I'm looking for critiques of the use Paterson and Tinker made of the Chapman-Cook Speed of Reading Test in their typographical factor's studies. Because it involved crossing out 'rogue' words it seems like an odd choice for gauging the impact of typographical factors on reading speed in continuous reading situations, which was their aim.

So you need "Basis for Effective Reading"?
UCLA doesn't have that, and that's my go-to.

> it involved crossing out 'rogue' words

Sounds like the test itself needs to be crossed out...

hhp

Hrant, thanks for checking, but I don't really need further confirmation of the “How Children and Adults Perceive Words in Reading,” inclusion in that book. Just knowing that the distinction is clearly stated in the paper is enough for me to get a sense of Tinker’s ideas — somewhat coarse, by today’s standards — on what's used by the eye in reading.

I am still curious about the Munoz, J.M., Odoriz, J.B. and Lavazza, J. “Registro de los movimientos oculares durante la lectura”, Revista de la Sociedad Argentina de Biologia, XX, Abril de 1944 page 284.

UCLA does have that journal, but only volumes 41-49.

From before:
> "It was concluded that the natural form of reading is not by spelling
> or syllabizing but on the basis of whole groups of words. That is, word
> forms or configurations constituted the units of perception in reading."

That does sound tantalizing - if to me totally unsurprising.

hhp

> word-forms or configurations constituted the units of perception in reading <
> tantalizing <

As I mentioned above I used that as my springboard in Thessaloniki.

Did you know that since 2006 there’s a Configural Processing Consortium?
http://www.indiana.edu/~psymodel/CPC2010/index.html
They'll be in Toronto in 2013!

Boumas are IMHO ordered configurations of role-units.

A recent (2010) paper by two of the consortium’s key scientists is:
“Information-Processing Alternatives to Holistic Perception: Identifying the
Mechanisms of Secondary-Level Holism Within a Categorization Paradigm” by
Mario Fific, James T. Townsend, available at:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2933083/?tool=pubmed
The domain is faces, but similar issues apply to words.

The stuff is pretty advanced, but highly suggestive.

Peter